Anyone who has studied "abstract algebra" and linear algebra as an

undergraduate can understand this book. This edition has been com-
pletely revised and reorganized, without however losing any of the
clarity of presentation that was the hallmark of the previous editions.
The first six chapters provide ample material for a first course: begin-
ning with the basic properties of groups and homomorphisms, the
topics covered include Lagrange's theorem, the Noether isomorphism
Joseph J. Ro
theorems, symmetric groups, G-sets, the Sylow theorems, finite abelian
groups, Krull-Schmidt theorem, solvable and nilpotent groups, and the
Jordan-Holder theorem.
The middle portion of the book then uses the Jordan-Holder theorem
to organize the discussion of extensions (automorphism groups, semi-
direct products, the Schur-Zassenhaus lemma, Schur multipliers) and
simple groups (simplicity of projective unimodular groups and, after a
return to G-sets, a construction of the sporadic Mathieu groups).
The book closes with three chapters on: infinite abelian groups, with
emphasis on countable groups; free groups and presentations of
groups, coset enumeration, free products, amalgams, and HNN exten-
sions; and a complete proof of the unsolvability of the word problem
for finitely presented groups, as well as the Higman imbedding theo-
rem and the undecidability of the isomorphism problem for groups.
Fourth Edition

ISBN 0-387-9L285-8
 Plate 1 Plate 2 b

Presentation of the group of the Novikov-Boone theorem Sufficiency of Boone's lemma

Generators:
-----
qi ------
S~ Ti Y t k

Relations:
 Plate 3 Plate 4
Presentation of the group of the Higrnan imbedding theorem Proof of Lemma 12.26; $6 is finitely related

Generators:

Relations:
 Graduate Texts in Mathematics Joseph J. Rotman

TAKEUTI/ZARING. Introduction to Axiomatic 33 HIRSCH. Differential Topology.

Set Theory. 2nd ed. 34 SPITLER. Principles of Random Walk. 2nd ed.
OXTOBY.
SCHAEFFER.
Measure and Category. 2nd ed.
Topological Vector Spaces.
HILTONISTAMMBACH. A Course in
35 WERMER. Banach Algebras and Several ,
Complex Variables. 2nd ed.
36 KELLEYINAMIOKA et al. Linear Topological
An Introduction to
Homological Algebra.
MACLANE.Categories for the Working
Spaces.
37 MONK.Mathematical Logic.
38 GRAUERT/FRITZSCHE. Several Complex
the Theory sf Groups
Mathematician.
HUGHESIPIPER. Projective Planes. Variables.
SERRE.A Course in Arithmetic. 39 ARVESON. An Invitation to (?*-Algebras.
TAKEUT~/ZAR~NG.
HUMPHREYS.
Axiomatic Set Theory.
Introduction to Lie Algebras
40 KEMENY/SNELL/~<NAPP.
Chains. 2nd ed.
Denumerable Markov
Fourth Edition
and Representation Theory. 41 APOSTOL. Modular Functions and Dirichlet
COHEN.A Course in Simple Homotopy Series in Number Theory. 2nd ed.
Theory. 42 SERRE. Linear Representations of Finite
CONWAY. Functions of One Complex Groups. With 37 Illustrations
Variable. 2nd ed. 43 GILLMAN/JERISON. Rings of Continuous
BEALS.Advanced Mathematical Analysis. Functions.
ANDERSONFULLER. Rings and Categories of 44 KENDIG. Elementafy Algebraic Geometry.
Modules. 2nd ed. 45 LOWE. Probability Theory I. 4th ed.
GOLUBITSKY~GUILEMIN. Stable Mappings and 46 LOBVE.Probability Theory 11. 4th ed.
Their Singularities. 47 MOISE.Geometric Topology in Dimensions 2
BERBERIAN. Lectures in Functional Analysis and 3.
and Operator Theory. 48 SACHSIWU. General Relativity for
WINTER. The Structure of Fields. Mathematicians.
ROSENBLATT. Random Processes. 2nd ed. 49 GRUENBERGIWEIR. Linear Geometry. 2nd ed.
HALMOS. Measure Theory. 50 EDWARDS. Fe~lnat'sLast Theorem.
HALM~S. A Hilbert Space Problem Book. 51 I~LINGENBERG. A Course in Differential
2nd ed. Geometry.
HUSEMOLLER. Fibre Bundles. 3rd ed. 52 HARTSHORNE. Algebraic Geometry.
HUMPHREYS. Linear Algebraic Groups. 53 MANIN. A Course in Mathematical Logic.
BARNEs/MACK.An Algebraic Introduction to 54 GRAVERJWATIUNS. Combinatorics with
Mathematical Logic. Emphasis on the Theory of Graphs.
GREUB.Linear Algebra. 4th ed. 55 BROWN/PEARCY. Introduction to Operator
HOLMES. Geometric Functional Analysis and Theory I: Elements of Functional Analysis.
Its Applications. 56 MASSEY. Algebraic Topology: An
~EWI~/STROMB Real
E R and
G . Abstract Introduction.
Analysis. 57 CROWELL/F~X. Introduction to Knot Theory.
MANES.Algebraic Theories. 58 KOBLITZ. p-adic Numbers, p-adic Analysis,
KELCEY. General Topology.
' and Zeta-Functions. 2nd ed.
'+ , -28 ZARSKIISAMGIEL. Commutative Algebra. 59 LANG.Cyclotomic Fields.
. , v0l.l: .:-?i > # ; 60 ARNOLD. Mathematical Methods in Classical
' 29, ZARIS~I/+A~I;IEL. Commutative Algebra. Mechanics. 2nd ed.
$ i. ,>:';.y9!l<-::.::.:< ' 61 WHITEHEAD. Elements of Hornotopy Theory.
' , A \ ,. . ! j X ~ o ~ si&ures
b~. in Abstract Algebra 1. 62 ~ARGAPOLOV/MERLZJAK~V. Fundamentals of
( ,,p -
5 . .-, BasiGConcepts.
the Theory of Groups.
52- JACOBSON.
Lectures in Abstract Algebra TI. 63 B~LLOBAS. Graph Theory.
'Linear Algebra. 64 EDWARDS. Fourier Series. Vol. I. 2nd ed. New York Berlin Heidelberg London Paris
- . . - - ~ ~ - J A G O B SLectures
Q N . ~ in Abstract Algebra 111.
Theory of Fields and Galois Theory.
.. . . ..
continued ajter indew Tokyo Hong Mong Barcelona Budapest
I
Joseph J. Rotman
Department of Mathematics
University of Illinois
at Urbana-Champaign
Urbana, IL 61801
USA

Editorial Board
J.H. Ewing F.W. Gehring P.R. Halmos
Department of Department of Department of
Mathematics Mathematics Mathematics
Indiana University University of Michigan Santa Clara University
Bloomington, IN 47405 Ann Arbor, MI 48109 Santa Clara, CA 95053
USA USA USA

Mathematics Subject Classifications (1991): 20-01

Library of Congress Cataloging-in-Publication Data

Rotman, Joseph J., 1934-
An introduction to the theory ofgroups/Joseph Rotman. - 4th
ed.
p. cm. - (Graduate texts in mathematics)
Includes bibliographical references and index.
ISBN 0-387-94285-8
1. Group theory. I. Title. 11. Series.
QA174.2.R67 1994
512.2-dc20

Printed on acid-free paper.

01995 Springer-Verlag New Yorlc, Inc.

Earlier editions 01984, 1973, and 1965 by Allyn & Bacon.
All rights reserved. This work may not be translated or copied in whole or in part without the
written permission of the publisher (Springer-Verlag New York, Inc., 175 Fifth Avenue, New
Yorlc, NY 10010, USA), except for brief excerpts in connection with reviews or scholarly
analysis. Use in connection with any form of information storage and retrieval, electronic
adaptation, computer software, or by similar or dissimilar methodology now known or hereaf-
ter developed is forbidden.
The use of general descriptive names, trade names, trademarlts, etc., in this publication, even
if the former are not especially identified, is not to be taken as a sign that such names, as
understood by the Trade Marlcs and Merchandise Marks Act, may accordingly be used freely
by anyone.

Production coordinated by Brian Howe and managed by Bill Imbornoni; manufacturing super-
vised by Gail Simon.
Typeset by Asco Trade Typesetting Ltd., Hong Icong.
Printed and bound by R.R. Donnelley & Sons, Harrisonburg, VA.
Printed in the United States of America.

ISBN 0-387-94285-8 Springer-Verlag New York Berlin Heidelberg

ISBN 3-540-94285-8 Springer-Verlag Berlin Heidelberg New York
Preface to the Fourth Edition

Group Theory is a vast subject and, in this Introduction (as well as in the
earlier editions), 1 have tried to select important and representative theorems
and to organize them in a coherent way. Proofs must be clear, and examples
should illustrate theorems and also explain the presence of restrictive hypo-
theses. I also believe that some history should be given so that one can
understand the origin of problems and the context in which the subject
developed.
Just as each of the earlier editions differs from the previous one in a signifi-
cant way, the present (fourth) edition is genuinely different from the third.
Indeed, this is already apparent in the Table of Contents. The boolc now
'begins with the unique factorization of permutations into disjoint cycles and
the parity of permutations; only then is the idea of group introduced. This is
consistent with the history of Group Theory, for these first results on permu-
tations can be found in an 1815 paper by Cauchy, whereas groups of permu-
tations were not'introduced until 1831 (by Galois). But even if history were
otherwise, 1 feel that it is usually good pedagogy to introduce a general
notion only after becoming comfortable with an important special case. 1
have also added several new sections, and I have subtracted the chapter on
Homological Algebra (although the section on Horn functors and character
groups has been retained) and the section on Grothendieck groups.
The format of the boolc has been changed a bit: almost all exercises now
occur at ends of sections, so as not to interrupt the exposition. There are
several notational changes from earlier editions: I now write N 5 G instead
of N c G to denote "H is a subgroup of G"; the dihedral group of order
2n is now denoted by D,,,instead of by D,;the trivial group is denoted by 1
instead of by (1); in the discussion of simple linear groups, I now distinguish
elementary transvections from more general transvections; I speak of the
 ... Preface to the Fourth Edition
VI~I

fitndamental g r o ~ ~ofp an abstract simplicia1 complex instead of its edgepath

group.
Here is a list of some other changes from earlier editions.
Chapter 3. The cycle index of a permutation group is given to facilitate use From Preface to the Third Edition
of Burnside's counting lemma in coloring problems; a brief account of mo-
tions in the plane introduces bilinear forms and symmetry groups; the affine
group is introduced, and it is shown how affine invariants can be used to
prove theorems in plane geometry.
Chapter 4. The number of subgroups of order p q n a finite group is counted
mod p; two proofs of the Sylow theorems are given, one due to Wielandt.
Chapter 5. Assuming Burnside's paqa theorem, we prove P. Hall's theorem
that groups having p-complements are solvable; we give Ornstein's proof
of Schur's theorem that G/Z(G) finite implies G' finite.
Chapter 6. There are several proofs of the basis theorem, one due to
Schenkman; there is a new section on operator groups.
Chapter 7. An explicit formula is given for every outer automorphism of
S6; stabilizers of normal series are shown to be nilpotent; the discussion of
the wreath product has been expanded, and it is motivated by computing the
automorphism group of a certain graph; the theorem of Gaschiitz on comple- Quand j'ai voulu me restreindre, je suis tombk dans l'obscuritk;
ments of normal p-subgroups is proved; a second proof of Schur's theorem j'ai prCfkrC passer pour un peu bavard.
on finiteness of G' is given, using the transfer; there is a section on projective
representations, the Schur multiplier (as a cohomology group), and covers; H. POINCAR$ Analysis situs,
there is a section on derivations and H1, and derivations are used to give Jotrrnnl de lPEcoiePolytechnique, 1895, pp. 1-121.
another proof (due to Gruenberg and Wehrfritz) of the Schur-Zassenhaus
lemma. (Had 1 written a new chapter entitled Cohornology of Groups, 1 4
Although permutations had been studied earlier, the theory of groups really
would have felt obliged to discuss more homological algebra than is appro- began with Galois (1811-1832) who demonstrated that polynomials are best
priate here.) understood by examining certain groups of permutations of their roots. Since
Chapter 8. There is a new section on the classical groups. that time, groups have arisen in almost every branch of mathematics. Even in
Chapter 9. An imbedding of S, into the Mathieu group MI, is used to this introductory text we shall see connections with number theory, conlbina-
construct an outer automorphism of S6. torics, geometry, topology, and logic.
Chapter 10. Finitely generated abelian groups are treated before divisible By the end of the nineteenth century, there were two main streams of group
groups. theory: topological groups (especially Lie groups) and finite groups. In this
Chapter 11. There is a section on coset enumeration; the Schur multiplier century, a third stream has joined the other two: infinite (discrete) groups.
is shown to be a homology group via Hopf's formula; the number of genera- It is customary, nowadays, to approach our subject by two paths: "pure"
tors of the Schur multiplier is bounded in terms of presentations; universal group theory (for want of a better name) and representation theory. This
central extensions of perfect groups are constructed; the proof of Britton's book is an introduction to "pure" (discrete) group theory, both finite and
lemma has been redone, after Schupp, so that it is now derived from the infinite.
normal form theorem for amalgams. We assume that the reader knows the rudiments of modern algebra, by
Chapter 12. Cancellation diagrams are presented before giving the difficult which we mean that matrices and finite-dimensional vector spaces are
portion of the proof of the undecidability of the word problem. friends, while groups, rings, fields, and their homomorphisms are only ac-
In addition to my continuing gratitude to those who helped with the first quaintances. A familiarity with elementary set theory is also assumed, but
three editions, I thank Karl Gruenberg, Bruce Rezniclc, Derek Robinson, some appendices are at the back of the book so that readers may see whether
Paul Schupp, Armond Spencer, John Walter, and Paul Gies for their help on my notation agrees with theirs.
this volume. I am fortunate in having attended lectures on group theory given by I.
Maplansky, S. Mac Lane, and M. Suzulti. Their influence is evident through-
Urbana, Illinois Joseph J. Rotman
1994
 From Preface to the Third Edition

out in many elegant ideas and proofs. I am happy to thank once again those
who helped me (directly and indirectly) with the first two editions: K.I. Appel,
M. Barr, W.W. Boone, J.L. Britton, G. Brown, D. Collins, C. Jockusch,
T. McLaughlin, C.F. Miller, 111. H. Paley, P. Schupp, F.D. Veldkamp, and
Contents
C.R.B. Wright. It is a pleasure to thank the following who helped with the
present edition: K.I. Appel, W.W. Boone, E.C. Dade, F. Haimo, L. McCulloh,
P.M. Neumann, E. Rips, A. Spencer, and J. Walter. I particularly thank
F. Hoffman, who read my manuscript, for his valuable comments and
suggestions.

Preface to the Fourth Edition

From Preface to the Third Edition
To the Reader

CHAPTER 1
Groups and Momomorphisms
Permutations
Cycles 4

Factorization into Disjoint Cycles

Even and Odd Permutations
Semigroups
Groups
Homomorphirims

CHAPTER 2
The Isomorphism Theoreins
Subgroups
Lagrange's Theorem
Cyclic Groups
Normal Subgroups
Quotient Groups
The Isomorphism Theorems
Correspondence Theorem
Direct Products
 ...
Contents Contents Xlll

CHAPTER 3 CHAPTER 8
Symmetric Groups and G-Sets Some Simple Linear Groups
Finite Fields
Conjugates
The General Linear Group
Symmetric Groups
The Simplicity of A,, PSL(2, K)
Some Representation Theorems PSL(tn, K)
Classical Groups
G-Sets
Counting Orbits
Some Geometry CHAPTER 9
Permutations and the Mathieu Groups
CHAPTER 4
The Sylow Theorems Multiple Transitivity
Primitive G-Sets
p-Groups Simplicity Criteria
The Sylow Theorems Affine Geometry
Groups of Small Order Projective Geometry
Sharply 3-Transitive Groups
CHAPTER 5 Mathieu Groups
Normal Series Steiner Systems
Some Galois Theory
The Jordan-Holder Theorem CHAPTER 10
Solvable Groups Abelian Groups
Two Theorems of P. Hall
Central Series and Nilpotent Groups Basics
p-Groups Free Abelian Groups
Finitely Generated Abelian Croups
CHAPTER 6 Divisible and Reduced Groups
Finite Direct Products Torsion Groups
Subgroups of Q
The Basis Theorem Character Groups
The Fundamental Theorem of Finite Abelian Groups 4
Canonical Forms; Existence
Canonical Forms; Uniqueness CHAPTER 11
The ICrull-Schmidt Theorem Free Groups and Free Products
Operator Groups Generators and Relations
Semigroup Interlude
CHAPTER7 Coset Enumeration
Extensions and Cohomology Presentations and the Schur Multiplier
The Extension Problem Fundamental Groups of Complexes
Automorphism Groups Tietze's Theorem
Semidirect Products Covering Complexes
Wreath Products The Nielsen-Schreier Theorem
Factor Sets Free Products
Theorems of Schur-Zassenhaus and Gaschiitz The ICurosh Theorem
Transfer and Burnside's Theorem The van Kampen Theorem
Projective Representations and the Schur Multiplier Amalgams
Derivations HNN Extensions
xiv Contents

CHAPTER 12
The Word Problem
Introduction
Turing Machines
The Marltov-Post Theorem
The Novikov-Boone-Britton Theorem: Sufficiency of Boone's
Lemma
Cancellation Diagrams
The Novikov-Boone-Britton Theorem: Necessity of Boone's
Lemma
The Higman Imbedding Theorem
To the Reader
Some Applications

Epilogue Exercises in a text generally have two functions: to reinforce the reader's
grasp of the material and to provide puzzles whose solutions give a certain
APPENDIXI pleasure. Here, the exercises have a third function: to enable the reader to
Some Major Algebraic Systems discover important facts, examples, and counterexamples. The serious reader
should attempt all the exercises (many are not difficult), for subsequent proofs
APPENDIX I1 may depend on them; the casual reader should regard the exercises as part of
Equivalence Relations and Equivalence Classes the text proper.
APPENDIX I11
Functions
APPENDIX IV
Zorn's Lemma
APPENDIX V
Countability
APPENDIXVI
Commutative Rings
Bibliography
Notation
Index
CHAPTER 1

Groups and Homomorphisms

Generalizations of the quadratic formula for cubic and quartic polynomials

were discovered in the sixteenth century, and one of the major mathematical
problems thereafter was to find analogous formulas for the roots of polyno-
mials of higher degree; all attempts failed. By the middle of the eighteenth
century, it was realized that permutations of the roots of a polynomial f(x)
were important; for example, it was ltnown that the coefficients of f(x) are
"symmetric functions" of its roots. In 1770, J.-L. Lagrange used permutations
to analyze the formulas giving the roots of cubics and quarticsY1but he
could not fully develop this insight because he viewed permutations only as
rearrangements, and not as bijections that can be composed (see below).
Composition of permutations does appear in work of P. Ruffini and of P.
Abbati about 1800; in 1815, A.L. Cauchy established the calculus of permuta-
tions, and this viewpoint was used by N.H. Abel in his proof (1824) that there
exist quintic polynomials for which there is no generalization of the qua-

One says that a poly~lomial(or a rational function) f of p variables is u-unbed if, by permuting
the variables in all possible ways, one obtains exactly r distinct polynomials. For exam-
ple, f(x,, x,, x,) = x, + x, + x, is a I-valued function, while g(x,, x,, x,) = x,x, + x, is a
3-valued function.
T o each polynomial f(x) of degree p, Lagrange associated a polynomial, called its resolvent,
and a rational function of p variables. We quote Wussing (1984, English translation, p. 78): "This
connection between the degree of the resolvent and the number of values of a rational function
leads Lagrange ... to consider the number of values that can be taken on by a rational
function of p variables. His conclusion is that the number in question is always a divisor of
p!. . . . Lagrange saw the 'metaphysics' of the procedures for the solution of algebraic equations
by radicals in this connection between the degree of the resolvent and the valuedness of rational
functions. His discovery was the starting point of the subsequent development due to Rufiini,
Abel, Cauchy, and Galois.. . . It is remarkable to see in Lagrange's work the germ, in admittedly
rudimentary form, of the group concept." (See Examples 3.3 and 3.3' as well as Exercise 3.38.)
4
2 1. Groups and Homomorphisms Cycles 3

dratic formula. In 1830, E. Galois (only 19 years old at the time) invented
groups, associated to each polynomial a group of permutations of its roots,
1.1. The identity function 1, on a set X is a permutation, and we usually denote it by
and proved that there is a formula for the roots if and only if the group of 1. Prove that l a = a = a 1 for every permutation a E SX.
permutations has a special property. In one great theorem, Galois founded
group theory and used it to solve one of the outstanding problems of his day. 1.2. For each a E S,, prove that there is P E S, with a p = 1 = pa (Hint. Let P be the
inverse function of the bijection a).
1.3. For all a, p, y E S,, prove that @(by)= (a&. Indeed, if X , Y, Z, W are sets and
f:X + Y, g: Y -+ Z, and h: Z -+ Ware functions, then h ( g f ) = (hg)f. (Hint:
Recall
Permutations that two functions f, g: A -+ B are equal if and only if, for all a E A, one has
f(a) = s(a1.1
Definition. If X is a nonempty set, a peumastation of X is a bijection a: X -+ X.
We denote the set of all permutations of X by Sx.

In the important special case when X = (1, 2, ... , n), we write St,instead of Cycles
S,. Wote that IS,,I = n!, where I YI denotes the number of elements in a set Y.
In Lagrange's day, a permutation of X = (1,2, . . . , n) was viewed as a The two-rowed notation for permutations is not only cumbersome but, as we
rearrangement; that is, as a list i,, i,, ..., in with no repetitions of all shall see, it also disguises important features of special permutations. There-
the elements of X. Given a rearrangement i,, i,, . .. , i,,, define a function fore, we shall introduce a better notation.
a: X -+ X by a ( j ) = ij for all j E X. This function a is an injection because the
list has no repetitions; it is a surjection because all of the elements of X DeEmitiasrm. If x E X and a E S,, then a Bxes x if a(x) = x and a rnooes x if
appear on the list. Thus, every rearrangement gives a bijection. Conversely, a(x)#x. -
any bijection a can be denoted by two rows:
Definition. Let i,, i,, .. ., i, be distinct integers between 1 and n. If a E St,fixes
the remaining n - r integers and if
a(il) = i,, a(i2) = i3, . . . , a(ir-,) = i,., a(ir) = i,,
and the bottom row is a rearrangement of (1,2,. . ., n). Thus, the two
versions of permutation, rearrangement and bijection, are equivalent. The then a is an r-cycle; one also says that a is a cycle of btzgth r. Denote a by
advantage of the new viewpoint is that two permutations in Sx can be (il i2 .- - ir).

(i::)
"multiplied," for the composite of two bijections is again a bijection. For
example, a = (, ,)
1 2 3
and p = are permutations of (I, 2 , 3 ) The
Every 1-cycle fixes every element of X, and so all 1-cycles are equal i: the
identity. A 2-cycle, which merely interchanges a pair of elements, is called a

product a/i is (:::); we compute this product2 by first applying /3 2nd

trg.6lensposition.
Draw a circle with i,, i, ... , i, arranged at equal distances aiound the
circumference; one may picture the r-cycle a = (i, i, - - i,) as a rotation
then a:
taking i, into i,, i, into i,, etc., and i, into i,. Indeed, this is the origin of the
term cycle, from the Greek word 1c61cLoafor circle; see Figure 1.1.
Here are some examples:

Note that pa = (: ::), so that ap ifi.

We warn the reader that some authors compute this product in the reverse order: first LY and
then p. These authors will write functions on the right: instead of f(x), they write (x)f (see
footnote 4 in this chapter).
 1. Groups and Homomorphisms Cycles 5

OU'UKE FONCTION P E U T A ( : Q U ~ R I RETC.

, .-in Multiplication is easy when one uses the cycle notation. For example, let
us compute y = a@,where a = (1 2) and p = (1 3 4 2 5). Since multiplica-
tion is composition of functions, y(1) = a 0 P(1) = a('(1)) = a(3) = 3; Next,
y(3) = a(P(3)) = a(4) = 4, and y(4) = a(P(4)) = a(2) = 1. Having returned to
1, we now seek y(2), because 2 is the smallest integer for which y has not yet
been evaluated. We end up with
1,'s (IPIIS tcrlncs A,, A, rcnfcrmcnt tlcs indices corrcspo~ttlants c1uit60iv~it
rcq)(ytivrnlrnt itgaus, o n pourra, sans inconv<anient, suppri~nr*rIvs
~~~Plncxs intliccs pour nc conserver q u r c c u s tlcs intlices correspontl;~nts The cycles on the right are disjoint as defined below.
cylli s o ~ l rcspe*ctivcmc~~t
t ini:geos. .4insi, par cscmplc, si I'on fait n = 5 ,
IPS (ICIIS s u b s t i t ~ ~ t i o n s Definition. Two permutations a, P E S, are disjoint if every x moved by one is
fixed by the other. In symbols, if a(x) # x, then P(x) = x and if P(y) # y, then
a(y) = y (of course, it is possible that there is z E X with a(z) = z = P(z)). A
family of permutations a,, a,, .. . , a,, is disjoint if each pair of them is disjoint.
hcront tquivalcntrs cnirc cllcs. Je clirai qu'unc suhstitutio~laura (!ti.
rCduitc sa plus s i ~ n p l ccspression lorsqu'on aura supprinld, clans 1 ~ s
t l c ~ ~tcrmcks,
s tous l ~ intliccs
s corresponela~lts Cgaus.
1.4. Prove that (1 2 ..- r - 1 r) = (2 3 ... r 1) = (3 4 ... 1 2) = - . - =
Soirnt ~ n a i n t c n a n at , P, y, ...,'5, q plusicurs des intlicrs I , 2 , 3, ..., 11
tbll nombrc &gal b p , c t supposons q u d a substitution ::
( ') k111it(\,
(r 1 . - . r - 1). Conclude that there are exactly r such notations for this r-cycle.
1.5. If 1 Ir I n, then there are (llr)[n(n - I). ..(n - r + I)] r-cycles in S,,.
sa pins simple expression prccnnc la formc 1.6. Prove the cancellation law for permutations: if either afi = ay or pa = ya, then
P = Y.
1.7. Let a = (i, i, . . - i,) and /? = (j, j, . - . j,). Prove that a and P are disjoint if and
only if ti,, i,, ..., i,} n {j,, j,, .. .,j,} = @.
1.n sortc quc, pour t1L;tluirc le sccontl tcrmc tlu prernicr, i l suffisc rlo 1.8. If a and b are disjoint permutations, then a/? = pa; that is, a and /3 commute.
r:tngcr en c.erclc, ou plutct en polygone ri!gnlicr, Ics intliccs z, p, y,
1.9. If a, p E S,, are disjoint and ap = 1, then a = 1 = p.
2 , . . ., '5, rj clc la maniibrc suivantc :
1.10. If a, /? E S,, are disjoint, prove that (ap), = akPkfor all k 2 0.1s this true if a and
1are not disjoint? (Define a0 = 1, a' = a, and, if Ic > 2, define a k to be the
composite of a with itself k times.)
1.11. Show that a power of a cycle need not be a cycle.
1.12. (i) Let a = (io i, .. . i,-,) be an r-cycle. For every j, k 2 0, prove that ak(ij)=
tbltic rEtn[)l:tccr c n s ~ ~ i chaquc
tc indicc par colui q u i , prrhlnirr, vit~nl ik+jif subscripts are read modulo r.
~".(~n(lrc!
s : ~plac:c 1orsq11'on fait tourncr tl'orirnl (In occitlr~itIc pol,vgont' (ii) Prove that if a is an r-cycle, then a' = 1, but that a k # 1 for every positive
integer k < r.
A. Cauchy, Mkmoire sur le nombre des valeurs qu'une fonction peut acqukrir, (iii) If a = PIP2... PI, is a product of disjoint ri-cycles Pi, then the smallest posi-
lorsq?'on y permute de toutes les mani6res possibles les quantitks qu'elle renferme, J. tive integer 1 with a' = 1 is the least common multiple of {r,, r,, . .. ,r,}.
de 1'Ecole Poly X V I I Cnhier, tome X (1815),pp. 1-28.
From: Oettvres Completes dlAzigustin Cnzlcliy, 11 Serie, Tome I, Gauthier-Villars, 1.13. (i) A permutation a E St, is regular if either a has no fixed points and it is the
Paris, 1905. product of disjoint cycles of the same length or a = 1. Prove that a is
regular if and only if a is a power of an n-cycle P; that is, a = P'" for
Figure 1.1 some m. (Hint: if a = (a, a, ...a,)(b, b, ...bk)...(z,z, . ..z,), where there are
m letters a, b, ...,z, then let p = (a, b, .. .z,a,b,.. .z,. ..akbk... z,).)
(ii) If a is an n-cycle, then a'' is a product of (n, k) disjoint cycles, each of length
n/(n, 12). (Recall that (11, k) denotes the gcd of n and k.)
(iii) If p is a prime, then every power of a p-cycle is either a p-cycle or 1.
6 1. Groups and Homomorphisms
1.14. (i) Let a = p y in S,,,where P and y are disjoint. If P moves i, then ak(i)= pk(i)
for all k 2 0.
(ii) Let a and /? be cycles in S, (we do not assume that they have the same
length). If there is i, moved by both a and p and if a y i , ) = pk(i,) for all
positive integers k, then a = p.
Factorization into Disjoint Cycles
1 2 3 4 5 6 7 8 9
Let us factor a = into a product of disjoint cycles.
6 4 1 2 5 3 8 9 7
Now a(1) = 6, and'so a begins (1 6; as 4 6 ) = 3, a continues (1 6 3; since
a(3) = 1, the parentheses close, and a begins (1 6 3). The smallest integer not
having appeared is 2; write (1 6 3)(2, and then (1 6 3)(2 4; continuing in
this way, we ultimately arrive at the factorization (which is a product of
disjoint cycles)
a = (1 6 3)(2 4)(5)(7 8 9).
Theorem 1.1. Every permutation a E S,, is either a cycle or a product of disjoint
cycles.
ProoJ The proof is by induction on the number lz of points moved by a. The
base step k = 0 is true, for then a is the identity, which is a 1-cycle. If k > 0,
let i, be a point moved by a. Define i, = a(i,), i, = a(i,), . . ., i,,, = a(ir),
where r is the smallest integer for which i,,, E {i,, i,, i,, . .. , i,.) (the list i,, i,,
i,, .. .,i,, . . . cannot go on forever without a repetition because there are only
n possible values). We claim that a(i,) = i,. Otherwise, a(i,) = ij for some
j 2 2; but a(ij-,) = i,, and this contradicts the hypothesis that a is an injec-
tion. Let a be the r-cycle (i, i, i, ..- i,). If r = n, then a is the cycle a. If
r < n and Y consists of the remaining n - r points, then a(Y) = Y and a fixes
the points in Y. Now a /{i,, i,, . .. , i,) = a /{i,, i, . . ., i,.). If a' is the permuta-
tion with a'/ Y = a (Y and which fixes {i,, i,, . . .,i,), then a and a' are disjoint
and a = aa'. Since a' moves fewer points than does a, the inductive hypothe-
sis shows that a', and hence a, is a product of disjoint cycles. Ell
One often suppresses all 1-cycles, if any, from this factorization of a, for
1-cycles equal the identity permutation. On the other hand, it is sometimes
convenient to display all of them.
Defimition. A complu~efactsriz~&ta'o~~
of a permutation a is a factorization of a
as a product of disjoint cycles which contains one 1-cycle (i) for every i fixed
by a.
In a complete factorization of a permutation a, every i between 1 and n
occurs in exactly one of the cycles.
Even and Odd Permutations 7
Theorem 11.2. Let a E S, and let a = PI .. ./It be a complete factorization into
disjoint cycles. This factorization is unique except for the order in which the
factors occur.
Proof. Disjoint cycles commute, by Exercise 1.8, so that the order of the
factors in a complete factorization is not uniquely determined; however, we
shall see that the factors themselves are uniquely determined. Since there is
exactly one 1-cycle (i) for every i fixed by a, it suffices to prove uniqueness of
the cycles of length at least 2. Suppose a = y, . . .y, is a second complete
factorization into disjoint cycles. If PCmoves i,, then P/(i,) = ak(i,) for all k,
by Exercise 1.14(i).Now some yj must move i,; since disjoint cycles commute,
we may assume that yj = y,. But y,k(i,) = ak(i,) for all k, and so Exercise
1.14(ii) gives P, = y,. The cancellation law, Exercise 1.6, gives P, ...A_, =
y, .. .y,-,, and the proof is completed by an induction on max(s, t).
1.15. Let a be the permutation of {1,2, ..., 9 ) defined by a(i) = 10 - i. Write a as a
product of disjoint cycles.
1.16. Let p be a prime and let a E S,. If aP = 1, then either a = 1, a is a p-cycle, or a is
a product of disjoint p-cycles. In particular, if a2 = 1, then either a = 1, a is a
transposition, or a is a product of disjoint transpositions.
1.17. How many a E Sf,are there with a2 = 1? (Niizt. (i j) = ( j i ) and (i j)(lc I) =
(Ic O(i 8.1
1.18. Give an example of permutations a, p, and y in S, with a commuting with P,
with p commuting with y, but with a not commuting with y.
Even and Odd Permutations
There is another factorization of permutations that is useful.
I Theorem 1.3. Every perrntltation a E S,, is a product of transpositions.
Pro*$ By Theorem 1.1, it suflices to factor cycles, and
I
Every permutation can thus be realized as a sequence of interchanges.
Such a factorization is not as nice as the factorization into disjoint cycles.
First of all, the transpositions occurring need not commute: (1 3)(1 2) =
(1 2 3) and (1 2)(1 3) = (1 3 2); second, neither the factors nor the number

8 1. Groups and Homomorphisms
of factors are uniquely determined; for example,
(1 2 3) = (1 3)(1 2) = (2 3)(1 3)
= (1 3)(4 2)(1 2)(1 4)
= (1 3)(4 2)(1 2)(1 4) (2 3)(2 3).
Is there any uniqueness at all in such a factorization? We now prove that the
parity of the number of factors is the same for all factorizations of a permuta-
tion a: that is, the number of transpositions is always even (as suggested by
the above factorizations of a = (1 2 3)) or is always odd.
DeGmition. A permutation a E S is even if it is a product of an even number of
transpositions; otherwise, a is odd.
It is easy to see that a = (1 2 3) is even, for there is a factorization
a = (1 3)(1 2) into two transpositions. On the other hand, we do not know
whether there are any odd permutations a at all; if a is a product of an
odd number of transpositions, perhaps it also has another factorization as a
product of an even number of transpositions. The definition of odd permuta-
tion a, after all, says that there is no factorization of a into an even number of
transpositions.
Lemma 1.4. If Iz,E 2 0, then
(a b)(a c, ... c, b dl ... d , ) = ( a c, ... c,')(b dl ... dl)
and
(a b)(a c, ... c,)(b dl ... d , ) = ( a c, ... c, b dl ... dl).
Proof. The left side sends a l-,c, H c,; c, t-,ci+,H ci+, if i < k; c , ~t-+ b H a;
b t--+ dl H dl; dj H dj+, H dj+, if j < 1; d, H a H b. Similar evaluation of the
right side shows that both permutations are equal. For the second equation,
just multiply both sides of the first equation by (a b) on the left. Ell
Definaition. If a E St, and a = P I . . . P, is a complete factorization into disjoint
cycles, then signum a is defined by
By Theorem 1.2, sgn is a well defined function (see Appendix 111). If r is a
transposition, then it moves two numbers, say, i and j, and fixes each of the
n - 2 other numbers; therefore, t = (n - 2) + 1 = n - 1, and so
Lemma 1.5. If P E S,, and z is a transposition, then
sgn(rP) = - sgn(P).
Even and Odd Permutations 9
Proof. Let r = (a b) and let P = y, . . .y, be a complete factorization of P into
disjoint cycles (there is one 1-cycle for each i fixed by P, and every number
between 1 and n occurs in a unique y). If a and b occur in the same y, say, in
y,, then y, = (a c, . . . c, b dl ... d,), where k t 0 and 1 t 0. By Lemma 1.4,
zyl = ( a c, .-. c,)(b dl d,),
and so rp = (zy1)y2...y, is a complete factorization with an extra cycle (zy,
splits into two disjoint cycles). Therefore, sgn(rP) = (- l)"-('+') = -sgn(P).
The other possibility is that a and b occur in different cycles, say, y, =
(a c, . . . c,) and y, = (b dl ... dl), where k t 0 and 12 0. But now r p =
(zy,y2)y,. . . y,, and Lemma 1.4 gives
Therefore, the complete factorization of r/3 has one fewer cycle than does P,
and so sgn(t/?) = (- 1)"-('-') = - sgn(8).
Theorem 1.6. For all a, P E Sn,
sgn(aP) = sgn (a) sgn(p).
Proof. Assume that a E S,, is given and that a = r, . . . r, is a factorization of
a into transpositions with m minimal. We prove, by induction on m, that
sgn(ap) = sgn(a) sgn(p) for every p E S,,. The base step is precisely Lemma
1.5. If m z 1, then the factorization r,.. .r, is also minimal: if r2...r, =
G, .. . gq with each oj a transposition and q < m - 1, then the factorization
a = r, G, . . .oq violates the minimality of rn. Therefore,
sgn(ap) = sgn(r ,. ..r,,/3) = -sgn(z2. . .r , P ) (Lemma 1.5)
= -sgn (z2.- . r,,) sgn (P) (by induction)
= sgn(r1. . z,,,) sgn(P) (by Lemma 1.5)
(i) A permutation a E Sn is even if and only if sgn(a) = 1.
(ii) A permutation a is odd if and only if it is a product of an odd number of
transpositions.
P r o d (i) We have seen that sgn(z) = - 1 for every transposition r. Therefore,
if a = r, . ..r, is a factorization of a into transpositions, then Theorem 1.6 gives
sgn(a) = sgn(z,) ... sgn(zq)= (- Thus, sgn(a) = 1 if and only if q is even.
If a is even, then there exists a factorization with q even, and so sgn(a) = 1.
Conversely, if 1 = sgn(a) = (- I)', then q is even and hence a is even.
(ii) If a is odd, then it has no factorization into an even number of transpo-
sitions, and so it must be a product of an odd number of them. Conversely, if
a = z, . . .rq with q odd, then sgn(a) = (- = - 1; by (i), a is not even, and
hence a is odd.
10 1. Groups and Homomorphisms
1.19. Show that an r-cycle is an even permutation if and only if r is odd.
1 2 3 4 5 6 7 8 9
1.20. Compute sgn(a) for a = number of elements of G, must we postulate more intricate associativity
9 8 7 6 5 4 3 2 1 '
1.21. Show that S, has the same number of even permutations as of odd permuta-
If z = (1 2),consider the function f : S, -+ S, defined by f ( a ) = za.)
tions. (Hint.
!, E S,,. If a and ,8 have the same parity, then a,!? is even; if a and ,8 have
1.22. Let a, ?
distinct parity, then a,!?is odd.
We are now going to abstract certain features of S,.
Definition. A (binary) operatior8 on a nonempty set G is a function
p: G x G + G .
An operation p assigns to each ordered pair (a, b) of elements of G a third
element of G, namely, p(a, b). In practice, p is regarded as a "multipli~ation'~
of elements of G, and, instead of p(a, b), more suggestive notations are used,
such as ab, a + bya 0 by or a * b. In this first chapter, we shall use the star
notation a * b.
It is quite possible that a * b and b + a are distinct elements of G. For
example, we have already seen that (1 2)(1 3) # (1 3)(1 2) in G = S,.
The Law ofSdtbstitutiorn (if a = a' and b = b', then a * b = a' * b') is just the
statement that p is a well defined function: since (a, b) = (a', b'), it follows that
p(a, b) = p(al, b'); that is, a * b = a' * b'.
One cannot develop a theory in this rarefied atmosphere; conditions on the
operation are needed to obtain interesting results (and to capture the essence
of composition in S,). How can we multiply three elements of G? Given (not
necessarily distinct) elements a,, a,, a, E G , the expression a, * a, t a, is am-
biguous. Since we can * only two elements of G at a time, there is a choice:
form a, * a, first, and then * this new element of G with a, to get (a, * a,) * a,;
or, form a, t (a, * a,). In general, these two elements of G may be different.
For example, let G = Z, the set of all integers (positive, negative, and zero),
and let the operation be subtraction: a * b = a - b; any choice of integers
a, b, c with c # 0 gives an example with (a - 6) - c # a - (b - c).
Definition. An operation * on a set G is associative if
for every a, byc E G.
Semigroups 11
Exercise 1.3 shows that multiplication in S, is associative. Associativity
allows one to multiply every ordered triple of elements in G unambiguously;
parentheses are unnecessary, and there is no confusion in writing a * b * c. If
we are confronted by four elements of G, or, more generally, by a finite
axioms to avoid parentheses?
Consider the elements of G that can be obtained from an expression
a, *a, * * a,,. Choose two adjacent a's, multiply them, and obtain an ex-
pression with only n - 1 factors in it: the product just formed and n - 2
original factors. In this new expression, choose two adjacent factors (either an
original pair or an original ni adjacent to the new product from the first step)
and multiply them. Repeat this procedure until there is an expression with
only two factors; multiply them and obtain an element of G. Let us illustrate
this process with the expression a * b * c * d. We may first multiply a * by
arriving at (a * b) * c * d, an expression with three factors, namely, a * by c,
and d. Now choose either the pair c, d or the pair a * by c; in either case,
multiply the chosen pair, and obtain the shorter expressions (a * b) * (c * d) or
[(a * b) * c] * d. Each of these last two expressions involves only two factors
which can be multiplied to give an element of G. Other ways to evaluate the
original expression begin by forming b * c or c * d as the first step. It is not
obvious whether all the elements arising from a given expression are equal.
Definition. An expression a, * a, * - .. * a,, needs no parentheses if, no matter
what choices of multiplications of adjacent factors are made, the resulting
elements of G are all equal.
Theorem 1.8 (Generalized Associativity). If * is an associative operation on a
set 6, then every expression a, :I: a, *. - .* a,, needs no parentheses.
PrsoJ The proof is by induction on n 2: 3. The base step n = 3 holds because
* is associative. If n 2 3, consider two elements obtained from an expression
a, * a, * . . . * a,, after two series of choices:
(1) (a, * . .. $:ai) *a,,) and (a, *... * aj) * (aj+, *... *a,)
(the choices yield a sequence of shorter expressions, and the parentheses
indicate ultimate expressions of length 2). We may assume that i 5 j. Since
each of the four expressions in parentheses has fewer than n factors, the
inductive hypothesis says that each of them needs no parentheses. If i = j, it
follows that the two products in (1) are equal. If i < j, then rewrite the first
expression as
(a, * - - - * a i ) * ( [ a i +*, - - - * a j ]*[aj+, *...*a,,])
and rewrite the second expression as
([a, *. .. * a i l * [ai+, * - * aj]) * (aj+, * . * a,).

12 I. Groups and Homomorphisms
By induction, each of the expressions a, *a,, a,+, *-.-*aj,and aj+, * .-.* a,,
* . a .
yield (uniquely defined) elements A, Byand C of G, respectively. Since (2) is
the expression A * (B * C) and (3) is the expression (A * B) * C, associativity
says that both these expressions give the same element of G. !I4
Definition.
A semigroup (G, *) is a nonempty set G equipped with an associa-
tive operation *.
Usually, one says "Let G be a semigroup . . . ," displaying the set G, but
tacitly assuming that the operation * is known. The reader must realize,
however, that there are many possible operations on a set making it a semi-
group. For example, the set of all positive integers is a semigroup under
either of the operations of ordinary addition or ordinary multiplication.
Definition. Let G be a semigroup and let a E G. Define a' =a and, for
n 2 1, define a"'" a a an.
Corollary 1.9. Let G be a semigroup, let a E G, and let m and n be positive
integers. Then an'* a n = am+''= a n* a'" and (an')" = am"= ( a n F
Proof. Both sides of the first (or second) equations arise from an expression
having m + n (or mn) factors all equal to a. But these expressions need no
parentheses, by Theorem 1.8. B
The notation a" obviously comes from the special case when * is multipli-
cation; a" = aa . .. a (n times). When the operation is denoted by +, it is more
natural to denote a * a ~ r . .. . * a = a i-+
a . . . -t- a by nu. In this additive nota-
+
tion, Corollary 1.9 becomes ma + na = (m n)a and (mn)a = m(na).
Groups
The most important semigroups are groups.
Definition. A group is a semigroup G containing an element e such that:
(i) e * a = a = a * e f o r a l l a ~ G ;
(ii) for every a E G, there is an element b E G with
Exercises 1.1, 1.2, and 1.3 show that S, is a group with composition as
operation; it is called the symmetric group on X . When X = ( I , 2, .. ., n),
then S, is denoted by S,, and it is called the symnetric grorrp on n letters.
Groups
Definition. A pair of elements a and b in a semigroup conrnrrrtes if a * b =
b*a. A group (or a semigroup) is abelian if every pair of its elements
commutes.
It is easy to see, for all n 2 3, that Sn is not abelian.
There are many interesting examples of groups; we mention only a few
of them now.
The set H of all integers (positive, negative, and zero) is an abelian group
+
with ordinary addition as operation: a * b = a b; e = 0; -a + a = 0. Some
other additive abelian groups are the rational numbers Q, the real numbers
R,and the complex numbers C. Indeed, every ring is an additive abelian
group (it is only a semigroup with 1 under multiplication).
Recall that if n 2 2 and a and b are integers, then a = b mod n
-
(pronounced: a is congruent to b modulo n) means that n is a divisor of a - b.
Denote the congruence class of an integer a mod n by [a]; that is,
[a] = (b E H: b a mod n)
= (a + kn: ~ E H ) .
The set Z,, of all the congruence classes mod n is called the integers modulo
98; it is an abelian group when equipped with the operation: [a] + [b] =
+
[a b]; here e = [O] and [-a] + [a] = [O] (Z, is even a commutative ring
- -
when "one" is [I] and multiplication is defined by [a] [b] = [ab]). The reader
should prove that these operations are well defined: if [a'] = [a] and [b'] =
[b], that is, if a' a mod n and b' b mod n, then [a' + b'] = [a + b] and
[arb'] = [ab].
If k is a field, then the set of all n x n nonsingular matrices with entries in
k is a group, denoted by GL(n, k), called the general linear group: here the
operation is matrix multiplication, e is the identity matrix E, and if A-I is the
inverse of the matrix A, then AA-' = E = A-'A. If n 2 2, then GL(n, k ) is not
abelian; if n = 1, then GL(1, I<)is abelian: it is the multiplicative group k x of
all the nonzero elements in 1c.
If R is an associative ring (we insist that R has an element I), then an
element u is a unii in R if there exists v E R with uv = 1 = vu. If a is a unit, so
that ab = 1 = ba for some b E R, then it is easy to see that ua is also a unit in
R (with inverse bv) and that U(R), the grorrp ofunits in R, is a multiplicative
group. If R is a field k, then U(lc) = k x . If R is the ring of all n x n matrices
over a field 12, then U(R) = GL(n, 12).
Theorem 1.169. If G is a group, there is a unique element e with e * a = a = a * e
for all a E G. Moreover, for each a E G, there is a unique b E G with a * b = e =
b * a.
Proof. Suppose that e' * a = a = a * e' for all a E G. In particular, if a = e,
then e' * e = e. On the other hand, the defining property of e gives e' * e = e',
and so e' = e.
14 1. Groups and Homomorphisms
Suppose that a * c = e = c * a . Then c = c i e = c * ( a * b ) = ( c * a ) r b=
e * b = b, as desired. El!
As a result of the uniqueness assertions of the theorem, we may now give
names to e and to b. We call e the identity of G and, if a x b = e = b * a, then
we call b the inverse of a and denote it by a-'.
Corollary 1.11. If G is a group and a E G, then
(a-')-I = a.
Proof. By definition, (a-I)-' is that element g E G with a-' x g = e = g * a-I.
But a is such an element, and so the uniqueness gives g = a.
Definition. If G is a group and a E G, define the powers of a as follows: if n is
a positive integer, then a" is defined as in any semigroup; define no = e; define
a-fl = (a-l)".
Even though the list of axioms defining a group is short, it is worthwhile
to make it even shorter so it will be as easy as possible to verify that a
particular example is, in fact, a group.
Thec~rem1.12. If G is a semigroup with an element e such that:
(if) e * a = a for all a E: G; and
(ii') for each a E G there is an element b E 6 with b * a = e, then G is a group.
PPOO$We claim that if x * x = x in G, then x = e. There is an element y E G
with y * x = e , and y * ( x : & x ) = y * x = e . O n the other hand, y * ( x * x ) =
(y * x ) * x = e * x = x. Therefore, x = e.
If b + a = e, let us show that a * b = e. Now (a * b) * (a * b) = a * [(b * a) * b]
= a * [e b] = a * b, and so our claim gives a r b = e. (Observe that we have
used associativity for an expression having four factors.)
l [ f a ~ Gwemust
, show that a * e = a . Choose b ~ G w i t hb * a = e = a * b
(using our just finished calculation). Then a * e = a *; ( b r a) = (a t b ) * a =
e * a = a, as desired. B
EXERCISES
1.23. If G is a group and a,, a,, .. ., a,, E G, then
(al *a,*... *an)-' = a;'*a;?, *--.*a;'.
Conclude that if n 2 0, then (a-I)" = (an)-'.
1.24. Let a,, a,, ... ,a,, be elements of an abelian semigroup. If b,, b,,
rearrangement of the ai, then
Groups 15
1.25. Let a and b lie in a semigroup G. If a and b commute, then (a * b)" = a n* bn for
every iz 2 1; if G is a group, then this equation holds for every n E Z.
1.26. A group in which x2 = e for every x must be abelian.
1.27. (i) Let G be a finite abelian group containing no elements a # e with a2 = e.
Evaluate a, * a, c - ;c a,, where a,, a,, .. . , a,, is a list with no repetitions, of
all the elements of G.
(ii) Prove Wilson's theorertz: If p is prime, then
(p - I)! = -1 modp.
(Hint. The nonzero elements of Z, form a multiplicative group.)
1.28. (i) I f a = (1 2 ... I. - 1 I.), thenol-' = (r I. - 1 ... 2 1).
(ii) Find the inverse of 1 2 3 4 5 6 7 ' 8 9
1.29. Show that a:Z,, -i Z,,, defined by n(x) = 4x2 - 3x7, is a permutation of Z,, ,
and write it as a product of disjoint cycles. What is the parity of a? What is a-I?
1.30. Let G be a group, let a E G, and let m, n E Z be (possibly negative) integers.
Prove that a'"c a'' = anrfn= a n a'" and (a1')" = am"= (all)m.
1.31. Let G be a group, let a E G, and let m and n be relatively prime integers. If
a" = e, show that there exists b E G with a = bn. (Hint. There are integers s and
t with 1 = sin + tn.)
1.32 (CawceSBatiorn Laws). In a group G, either of the equations a * b = a * c and
b a a = c * a implies b = c.
1.33. Let G be a group and let a E G.
(i) For each a E G, prove that the functions La:G -+ G, defined by x c r a * x
(called lvft translation by a), and R,: G -+ G, defined by x cr x * a-I (called
right tra~zslabionby a), are bijections.
(ii) For all a, b E G, prove that Lo,, = L, 0 L, and R,,, = R, 0 R,.
(iii) For all a and b, prove that La o R, = R, 0 La.
1.34. Let G denote the multiplicative group of positive rationals. What is the identity
of G? If a E G, what is its inverse?
1.35. Let n be a positive integer and let G be the multiplicative group of all nth roots
of unity; that is, G consists of all complex numbers of the form e2"ik'n,where
k E Z.What is the identity of G? If a E G, what is its inverse? How many
elements does G have?
1.36. Prove that the following four permutations form a group V (which is called the
4-grol6p):
1; (1 2)(3 4); (1 3)(2 4); (1 4)(2 3).
..., b,, is a 1.37. Let k = IW u (m), and define 110 = ao, 11% = 0, co/m = 1, and 1 - a, =
ao = GO - 1. Show that the six functions W -+ W, given by x, llx, 1 - x,
1/(1 - x), X/(X- I), (X - 1)/x, form a group with composition as operation.
 1. Groups and Homomorphisms
Homomorphisms
Let G be a finite group with n elements a,, a,, . .., an. A ~zuliiplicniiontable
for G is the n x n matrix with i, j entry ai * aj:
Informally, we say that we "know" a finite group G if we can write a multipli-
cation table for it. Notice that we say "a" multiplication table and not "the"
multiplication table, for a table depends on the particular ordering a,, a,, ..
a,, of the elements of G. (One may also speak of multiplication tables
infinite groups, but in this case, of course, the matrices are infinite.) It
customary to list the identity e first so that the first row (and first
display the elements in the order they occur on a chosen list.
Let us now consider two almost trivial examples of groups. Let G be the
group whose elements are the numbers 1 and - 1, with operation multiplica-
tion; let H be the additive group Z,.Compare multiplication tables of these
two groups:
I't is quite clear that G and H are distinct groups; on the other hand, it is
equally clear that there is no significant difference between them. Let us make
this idea precise.
DeEwitioan. Let (G, s) and (H, 0) be group^.^ A function f : G -+ N is a
homomolrj)hls171if, for all a, b E 6 ,
An bomowphisn¶is a homomorphism that is also a bijection. We say that G is
isomorphic to H , denoted by G z H, if there exists an isomorphism f : G -+ H.
The two-element groups G and H, whose multiplication tables are given
above, are isomorphic: define f : G -+H by f(1) = [O] and f (- 1) = [I].
This definition also applies to semigroups.
Homomorphisms 17
Let f : G -+ H be an isomorphism, and let a,, a,, . .. ,a, be a list, with no
repetitions, of all the elements of G. Since f is a bijection, every element of H
occurs exactly once on the list f (a,), f (a,), . . .,f (a,), and so this list can be
used to form a multiplication table for H. That f is a homomorphism, that is,
f (ai * aj) = f (ai) 0 f (aj), says that if we superimpose the multiplication table
of G onto that of H, then the tables "match." In this sense, isomorphic groups
G and H have the "same" multiplication tables. Informally, one regards G
and N as being essentially the same, the only distinction being that G is
written in English and H is written in French; an isomorphism f is a
dictionary which translates one to the other.
Two basic problems occurring in mathematics are: classification of all
systems of a given type (e.g., groups, semigroups, vector spaces, topological
spaces); classification of all the "maps" or transformations from one such
system into another. By a classification of systems, we mean a way to distin-
guish different systems or, what is the same thing, a way to tell when two
systems are essentially the same (isomorphic). For example, finite-dimen-
sional vector spaces over a field k are classified by the theorem that two such
are isomorphic if and only if they have the same dimension. One can even
classify all the maps (linear transformations) between vector spaces; they give
rise to similarity classes of matrices which are classified by canonical forms.
The same two problems arise in Group Theory: when are two groups
isomorphic; describe all the homomorphisms from one group to another.
Both of these problems are impossibly hard, but partial answers are known
and are very useful.
Theo~em1-13.Let f : (6, *) -+ (G', 0) be a homomorphism.
(i) f (e) = e', where e' is the identity in 6'.
(ii) If a E G, then f(a-') = f (a)-'.
(iii) If a E C and n E Z,then f(an) = f(a)".
Prod$ (i) Applying f to the equation e = e J: e gives f (e) = f (e * e) =
f(e) 0 f(e). Now multiply each side of the equation by f (e)-I to obtain e' =
f (el-
(ii) Applying f to the equations a * a-l = e = a-' * a gives f(a) * f(a-') =
e' = f(a-') + f (a). It follows from Theorem 1.10, the uniqueness of the inverse,
that f(n-9 = f (a)-'.
(iii) An easy induction proves f (a") = f (a)" for all n 2 0, and then f (a-") =
f((a-I)") = f(a-I)" = f(a)-". Eli
Here are some examples. Theorem 1.6 shows that sgn: S,, -+ {*I) is a
homomorphism; the function v: Z -+ Z,, defined by v(a) = [a], is a homo-
morphism; if lcx denotes the multiplicative group of nonzero elements of a
field k, then determinant is a homomorpliism det: GL(n, k) k ". -+
 1. Groups and Homomorphisms
1.38. (i) Write a multiplication table for S,.
(ii) Show that S, is isomorphic to the group of Exercise 1.37. (Hint. The ele-
ments in the latter group permute {0, 1, a}.)
1.39. Let f : X 4 Y be a bijection between sets X and Y Show that a w f o aof
is an isomorphism S, -+ S,.
1.40. Isomorphic groups have the same number of elements. Prove that the converse
is false by showing that Z4 is not isomorphic to the 4-group V defined in
Exercise 1.36.
1.41. If isomorphic groups are regarded as being the same, prove, for each positive
integer n, that there are only finitely many distinct groups with exactly n
elements.
1.42. Let G = {x,, ..., x,} be a set equipped with an operation *, let A = [aij] be its
multiplication table (i.e., aij = xi * xj), and assume that G has a (two-sided) iden-
tity e (that is, e a x = x = x a e for all x E G).
(i) Show that * is commutative if and only if A is a symmetric matrix.
(ii) Show that every element x E G has a (two-sided) inverse (i.e., there is x' E G
with x * x' = e = x' * x) if and only if the multiplication table A is a Latin
sqrare; that is, no x E G is repeated in any row or column (equivalently,
every row and every column of A is a permutation of G.)
(iii) Assume that e = x,, so that the first row of A has a,, = xi. Show that the
first column of A has a,, = xi' for all i if and only if a,, = e for all i.
(iv) With the multiplication table as in (iii), show that a is associative if and only
if aijajk= a, for all i, j, Ic.
1.43. (i) I f f : G -+ H and g: H -t K are homomorphisms, then so is the composite
g of: G 4 K .
(ii) If f : G 4 H is an isomorphism, then its inverse f - I : H -+ G is also an
isomorphism.
(iii) If Y' is a class of groups, show that the relation of isomorphism is an equiva-
lence relation on %
1.44. Let G be a group, let X be a set, and let f : G -+X be a bijection. Show that there
is a unique operation on X so that X is a group and f is an isomorphism.
1.45. If lc is a field, denote the columns of the n x n identity matrix E by E , , ..., E,,. A
permutation matrix P over k is a matrix obtained from E by permuting its
columns; that is, the columns of P are g a l , ...,cap,for some a E S,. Prove that the
set of all permutation matrices over k is a group isomorphic to S,,. (Hint. The
inverse of P is its transpose P', which is also a permutation matrix.)
1.46. Let T denote the circle group: the multiplicative group of all complex numbers
of absolute value 1. For a fixed real number y, show that L,: IW -+ T, given
by f,,(x)= eiYx, is a homomorphism. (The functions fy are the only continuous
homomorphisms W -+ T.)
1.47. If a is a fixed element of a group G, define yo: G -+G by y,(x) = a s x a a-' (y, is
called conj16gcatio~zby a).
Homomorphisms
(i) Prove that y, is an isomorphism.
(ii) If a, b E G, prove that yayb =
1.48. If G denotes the multiplicative group of all complex nth roots of unity (see
Exercise 1.35),then G r Z,,.
1.49. Describe all the homomorphisms from Z,, to itself. Which of these are
isomorphisms?
1.50. (i) Prove that a group G is abelian if and only if the function f : G -+ G, defined
by f (a) = a-I, is a homomorphism.
(ii) Let f : G -, G be an isomorphism from a finite group G to itself. Iff has no
nontrivial fixed points (i.e., f(x) = x implies x = e) and iff 0 f is the identity
function, then f(x) = x-' for all x E G and G is abelian. (Hint. Prove that
every element of G has the form x n f(x)-I.)
1.51 (H<aplamsky). An element a in a ring R has a left quasi-inverse if there exists an
element b E R with a + b - ba = 0. Prove that if every element in a ring R
except 1 has a left quasi-inverse, then R is a division ring. (Hint. Show that
R - {I) is a group under the operation a 0 b = a + b - ba.)
\
1.52. (i) If G is the multiplicative group of all positive real numbers, show that
log: G -+ (W, +) is an isomorphism. (Hint: Find a function inverse to log.)
(ii) Let G be the additive group of Z[x] (all polynomials with integer coeffi-
cients) and let H be the multiplicative group of all positive rational numbers.
Prove that G r H. (Hint. Use the Fundamental Theorem of Arithmetic.)
Having solved Exercise 1.52, the reader may wish to reconsider the ques-
tion when one "knows" a group. It may seem reasonable that one knows a
group if one knows its multiplication table. But addition tables of Z[x] and
of H are certainly well known (as are those of the multiplicative group of
positive reals and the additive group of all reals), and it was probably a
surprise that these groups are essentially the same. As an alternative answer
to the question, we suggest that a group G is "known" if it can be determined,
given any other group H, whether or not G and H are isomorphic.
It is easy to see that s3,: G -+ G, defined by 6,(x) = a-' * x s a, is also an isomorphism; however,
6,6, = d,.,. Since we denote the value of a function f by f(x), that is, the symbol f is on the left,
the isomorphisms y, are more natural for us than the 6,. On the other hand, if one denotes 6,(x)
by x", then one has put the function symbol on the right, and the 6, are more convenient:
xu'" = (x"),. Indeed, many group theorists nowadays put all their function symbols on the right!
.

CHAPTER 2
The Isomorphism Theorems
We now drop the * notation for the operation in a group. Henceforth, we
shall write ab instead of a * b, and we shall denote the identity element by 1
instead of by e.
Subgroups
DeBnition. A nonempty subset S of a group G is a srrbgvorap of G if s E G
implies s-' E G and s, t E G imply st E G.
If X is a subset of a group G, we write X c G ; if X is a subgroup of G, we
write % 5 G.
Theorem 2.1. If S IG (i.e., if S is a subgroup of G), then S is a group in its own
right.
Proof. The hypothesis "s, t E S imply st E S" shows that S is equipped with
an operation (if p: G x G 4 G is the given multiplication in G, then its restric-
tion PIS x S has its image contained in S). Since S is nonempty, it contains
an element, say, s, and the definition of subgroup says that s-' E S; hence,
1 = ss-' E S. Finally, the operation on S is associative because a(bc) = (ab)c
for every a, b, c E G implies, in particular, that a(bc) = (ab)c for every
a, b, c E S .
Verifying associativity is the most tedious part of showing that a given set
G equipped with a multiplication is actually a group. Therefore, if G is given
Subgroups
as a subset of a group G*, then it is much simpler to show that G is a
subgroup of G* than to verify all the group axioms for G. For example, the
four permutations of the Cgroup V form a group because they constitute a
subgroup of S,.
Theorem 2.2. A subset S of a group G is a subgroup if and only if 1 E S and
s, t E S imply st-' E S.
Proof. If s E S, then is-' = s-' E S, and if s, t E S, then s(t-')-I = st E S. The
converse is also easy. @I
Definition. If G is a group and a E G, then the cyclic subgroup generated by a,
denoted by (a), is the set of all the powers of a. A group G is called cyclic if
there is a E G with G = (a); that is, G consists of all the powers of a.
It is plain that (a) is, indeed, a subgro~lpof G. Notice that different ele-
ments can generate the same cyclic subgroup. For example, (a) = (aF').
Definition. If G is a group and a E G, then the ordm of a is I (a) I, the number
of elements in (a).
Theorem 2.3. If G is a group and a E G has finite order m, then m is the smallest
positive integer such that an' = 1.
Proof. If a = 1, then m = 1. If a # 1, there is an integer 12 z 1 so that 1, a,
a', . . . , a"' are distinct elements of G while a ' = a' for some i with 0 I i s
k - 1. We claim that a k = 1 = a*. If a k = a' for some i 2 1, then k - i 5
I 12 - 1 and a*-' = 1, contradicting the original list 1, a, a2, ..., ak-' having no
repetitions. It follows that k is the smallest positive integer with a' = 1.
It now suffices to prove that k = m; that is, that (a) = (I, a, a2, .. ., ak-'1.
Clearly (a) 3 ( 1, a, a2, . .. , a'-'). For the reverse inclusion, let a' be a power
+
of a. By the division algorithm, 1 = qk r, where 0 5 r < 12. Hence, a' =
a q ~ c + r = a qk a r - u. r (because a" = I), and so a' = ar E {I, a, a2, . .., ak-'1. O
If a E S , is written as a product of disjoint cycles, say, a = 8, ...P,, where 8,
is an ri-cycle for every i, then Exercise 1.12(iii) shows that the order of a is
lcm{rl, . . .,rt).
Corollary 2.4. If G is a finite group, then a nonempty subset S of G is a
subgroup if and OBEY if S, t E S imply st E S.
Proo$: Necessity is obvious. For sufficiency, we must show that s E S im-
plies s-' E S. It follows easily by induction that S contains all the powers of s.
Since G is finite, s has finite order, say, m. Therefore, 1 = smE S and s-' =
Sm-l E S. @j
22 2. The Isomorphism Theorems
EXAMPLE 2.1. If G is a group, then G itself and {I) are always subgroups (we
shall henceforth denote the subgroup (1) by 1). Any subgroup H other than
G is called proper, and we denote this by H < G; the subgroup 1 is often
called the trivial subgroup.
2.2. Let f : G -+ H be a homomorphism, and define
EXAMPLE
kernel f = (a E G: f(a) = 1)
and
image f = (h E H: h = f (a) for some a E G).
Then K = kernel f is a subgroup of G and image f is a subgroup of H. To see
that K 5 G, note first that f(1) = 1, so that 1 E K. Also, if s, t E I<, then
f(s) = 1 = f (t), and so f(st-') = /(s) f (t)-' = 1; hence st-' E K, and so K is a
subgroup of G. It is equally easy to see that image f is a subgroup of H.
Notation. We usually write kerf instead of kernel f and im f instead of
image f.
We have been using muliiplicative notation, but it is worth writing the
definition of subgroup in additive notation as well. If G is an additive group,
then a nonempty subset S of G is a subgroup of G if s E S implies -s E S and
s, t E S imply s + t E S. Theorem 2.2 says that S is a subgroup if and only if
0 E S and s, t E S imply s - t E S.
Theorem 2.5. The intersection of any family of subgroups of a group G is
again a subgroup of G.
Proof. Let (S,: i E I} be a family of subgroups of 6. Now 1 E Si for every i, and
0 n
so 1 E Si. If a, b E S,, then a, b E Si for every i, and SO ab-' E Si for every
i; hence, ab-' E 0 S,, and Si 5 G. !3
Co~olllary2.6. If X is a subset of a group 6, then there is a smdnllest subgroup
H of G containing X; that is, if X c S and S 5 G, then H _< S.
Pro05 There are subgroups of G containing X; for example, G itself contains
X; define H as the intersection of all the subgroups of 6 which contain X.
Note that H is a subgroup, by Theorem 2.5, and X c H. If S 5 G and X c S,
then S is one of the subgroups of G being intersected to form H; hence,
H 5 S, and so H is the smallest such subgroup. El
Definition. If X is a subset of a group G, then the smallest subgroup of G
containing X, denoted by (X), is called the srsbgro~dpge~~eratehdby X. One
also says that X generates (X).
In particular, if H and K are subgroups of G, then the subgroup (H u I<)
is denoted by H v K.
Subgroups 23
If X consists of a single element a, then (X) = (a), the cyclic subgroup
generated by a. If X is a finite set, say, X = {a,, a,, ..., a,,} then we write
(X) = (a,, a,, . .., a,) instead of (X) = ((a,, a,, .. ., a,,}).
Here is a description of the elements in (X).
Definition. If X is a nonempty subset of a group G, then a word on X is an
element w E G of the form
where xi E X, ei = $1, and n 2 1.
Theorem 2.7. Let X be a subset of a group G. If X = (a, then (X) = 1; if X
is nonempty, then (X) is the set of all the words on X.
Proof: If X = (a, then the subgroup 1 = {I) contains X , and so (X) = 1. If
X is nonempty, let W denote the set of all the words on X. It is easy to see
that W is a subgroup of G containing X: 1 = xi1x1 E W; the inverse of a
word is a word; the product of two words is a word. Since (X) is the smallest
subgroup containing X, we have (X) c W. The reverse inclusion also holds,
for every subgroup H containing X must contain every word on X. There-
fore, W 5 H, and W is the smallest subgroup containing X. @
2.1. Show that A,, the set of all even permutations in S,#, is a subgroup with n!/2
g r o q on n letters.) (Hint. Exercise 1.21.)
elements. (A, is called the adter~tati~~g
2.2. If k is a field, show that SL(n, k), the set of all n ,:n matrices over k having
determinant 1, is a subgroup of GL(n, k). (SL(n,k) is called the special di~~ear
group over k.)
2.3. The set theoretic union of two subgroups is a subgroup if and only if one is
contained in the other. Is this true if we replace "two subgroups" by "three
subgroups"?
2.4. Let S be a proper subgroup of G. If G - S is the complement of S, prove that
(G - S ) = G.
2.5. Let f : G 4 H and g: G --+ H be homomorphisms, and let
I< = (a E G: f(a) = g(a)).
Must I< be a subgroup of G?
2.6. Suppose that X is a nonempty subset of a set Y. Show that S, can be ifnbeddedin
S,; that is, S, is isomorphic to a subgroup of S,.
I
2.7. If n > 2, then A,, is generated by all the 3-cycles. (Hint.(ij)(jk) = (ijk) and (tj)(kI) =
(ijk)(jkl).)
2.8. Imbed Sn as a subgroup of A,+,, but show, for n 2 2, that Sn cannot be imbedded
in Afl+l.
 2. The Isomorphism Theorems
24
2.9. (i) Prove that Sn can be generated by (1 2), (1 3), ...,(1 n).
+
(ii) Prove that Sn can be generated by (1 2), (2 3), ...,(i i 1), .. ., ( n - 1, n).
(iii) Prove that Sn can be generated by the two elements (1 2) and (1 2 .,. rz).
(iv) Prove that S, cannot be generated by (1 3) and (1 2 3 4). (Thus, S, can be
-generated by a transposition and a 4-cycle, but not every choice of transpo-
sition and 4-cycle gives a generating set.)
Lagrange's Theorem
Definition. If S is a subgroup of G and if t E G, then a right coset of S in G is
the subset of G
St = {st: s E S)
(a left coset is tS = {ts: s E S)). One calls t a rppueserrtntive of St (and also
of tS).
EXAMPLE 2.3. Let G be the additive group of the plane R2: the elements of G
are vectors (x, y), and addition is given by the "parallelogram law": (x, y) +
(x', y') = (x + x', y + y'). A line C through the origin is the set of all scalar
multiples of some nonzero vector v = (x,, yo); that is, C = {ru: r E R). It is
easy to see that zf is a subgroup of G. If u = (a, b) is a vector, then the coset
u + C is easily seen to be the line parallel to C which contains u.
-
EXAMPLE 2.4. If G is the additive group Z of all integers, if S is the set of all
multiples of an integer n (S = (n), the cyclic subgroup generated by iz), and if
a E Z, then the coset a + S = {a + qn: q E Z) = {k E Z:li a mod n); that is,
the coset a + {n) is precisely the congruence class [a] of a mod n.
EXAMPLE 2.5. Let G = S, and let H = (r) = {1, r), where r = (1 2). The right
cosets of H in G are
H={l,x); H(l 2 3 ) = { ( l 2 3 ) , ( 2 3 ) ) ;
H(1 3 2) = {(I 3 2), (1 3)).
The left cosets of H in G are
H = 1 ,r (1 2 3)H = ((1 2 3), (1 3));
Notice that distinct right cosets are disjoint (as are distinct left cosets),just as
in the example of parallel lines. Notice also that right cosets and left cosets
can be distinct; for example, (1 2 3)H # H ( l 2 3); indeed, (1 2 3)H is not
equal to any right coset of H in G.
A right coset St has many representatives; every element of the form st
for s E S is a representative of St. The next lemma gives a criterion for
Lagrange's Theorem
25
determining whether two right cosets of S are the same when a representative
of each is known.
Lemma 2.8. If S 5 G, then Sa = Sb if and only if ab-' E S (as = bS if and only
ifb-la E S).
Proof: If Sa = Sb, then a = l a E Sa = Sb, and so there is s E S with a = sb;
hence, ab-' = s E S. Conversely, assume that ab-I = o E S; hence, a = ob.
To prove that Sa = Sb, we prove two inclusions. If x E Sa, then x = sa for
some s E S, and so x = sob E Sb; similarly, if y E Sb, then y = s'b for some
s' E S, and y = s'a-la E Sa. Therefore, Sa = Sb. H
Theorem 2.9. If S 5 G, then any two right (or any two left) cosets of S in G are
either identical or disjoint.
Prooj: We show that if there exists an element x E Sa n Sb, then Sa = Sb.
Such an x has the form sb = x = ta, where s, t E S. Hence, ab-' = t-'s E S,
and so the lemma gives Sa = Sb.
Theorem 2.9 may be paraphrased to say that the right cosets of a subgroup
-
S comprise a partition of G (each such coset is nonempty, and G is their
disjoint union). This being true, there must be an equivalence relation on G
lurking somewhere in the background: it is given, for a, b E G, by a b if
ab-' E S, and its equivalence classes are the right cosets of S.
Theorem 2.10. If S 5 G, then the number of right cosets of S in G is equal to
the number of left cosets of S in 6.
ProoJ: We give a bijection f : W -+9, where 9 is the family of right cosets of
S in G and 2'is the family of left cosets. If Sa E LZ?,your first guess is to define
.f(Sa) = US,but this does not work. Your second guess is to define f(Sa) =
a-IS, and this does work. It must be verified that f is well defined; that is, if
Sa = Sb, then a-'S = b-'S (this is why the first guess is incorrect). It is rou-
tine to prove that f is a bijection.
Definition. If S 5 G, then the i r d a . of S in G, denoted by [G: S], is the
number of right cosets of S in G.
Theorem 2.10 shows that there is no need to define a right index and a left
index, for the number of right cosets is equal to the number of left cosets.
It is a remarkable theorem of P. Hall (1935) that in a finite group G, one
can always (as above) choose a common system ofveprrsenfatiues for the right
and left cosets of a subgroup S; if [G : S ] = n, there exist elements t,, . .., t , E
G so that t,S, . . ., t,,S is the family of all left cosets and St,, . . ., St,, is the
family of all right cosets.
26 2. The Isomorphism Theorems Lagrange's Theorem

Definition. If G is a group, then the order of G, denoted by 1 GI, is the number

of elements in G. 2.10. If G is a finite group and K IH IG, then
[G:K] =[G:H][H:K].
The next theorem was inspired by work of Lagrange (1770), but it was
probably first proved by Galois. 2.11. Let a E G have order rz = mlc, where m, k 2 1. Prove that a k has order rn.
2.12. (i) Prove that every group G of order 4 is isomorphic to either Z, or the
Theorem 2.11 (Lagrange). If G is a finite group and S IG, then IS1 divides I GI 4-group 87.
(ii) If G is a group with \GI I5, then G is abelian.
and [ G : S ] = JGJ/JSI.
2.13. If a E G has order n and k is an integer with a k = 1, then n divides k. Indeed,
(k E Z: a k = 1) consists of all the multiplies of n.
ProoJ By Theorem 2.9, G is partitioned into its right cosets
2.14. If a E G has finite order and f : G --+ H is a homomorphism, then the order of
G = S t l u S t , u .. . u S t , , f(a) divides the order of a.
and so 1GI = x:=, /St,l. But it is easy to see that L: S -t Sti, defined by L(s) =
st,, is a bijection, and so lStil = I SI for all i. Thus / GI = n 1 SI, where n =
2.15. Prove that a group G of even order has an odd number of elements of order 2
(in particular, it has at least one such element). (Hint. If a E G does not have
[G:S]. f2.l order 2, then a # a-I.)
2.16. If N < G has index 2, then a2 E H for every a E G.
Corollary 2.12. If G is a finite group and a E G. Then the order of a divides I GI. 2.17. (i) If a, b E G commute and if an' = 1 = b", then (ab)k= 1, where lc = Icm(m, n ) .
(The order of ab may be smaller than Ic; for example, talce b = a-I.) Con-
Proof. By definition, the order of a is 1 ( a ) 1, and so the result follows at once clude that if a and b have finite order, then ab also has finite order.
from Lagrange's theorem. El (ii) Let G = GL(2, Q) and let A, B E G be given by

Definitioa. A group G has expsaent n if x" = 1 for all x E G. A=[: ] and B = [ -1 -1I].
Show that A4 = E = B3, but that AB has infinite order.
Remark. Some people use the term "exponent" to mean the smallest possible 2.18. Prove that every subgroup of a cyclic group is cyclic. (Hint. Use the division
n such that xn = 1 for all x E G. For us, the 4-group V has exponent 4 as well algorithm.)
as exponent 2. I
i
Lagrange's theorem shows that a finite group G of order n has exponent n. 2.19. Prove that two cyclic groups are isomorphic if and only if they have the same
order.

Corollary 2.13. If p is a prime and IGl = p, then G is a cyclic group. Definido~.The Eakr tp--~ncrfion
is defined as follows:

Proof. Take a E G with a # 1. Then the cyclic subgroup ( a ) has more than
one element (it contains a and I), and its order l(a)l z 1 is a divisor of p. 2.20. If G = (a) is cyclic of order n, then ak is also a generator of G if and only if

- 1
Sincepisprime,J(a)l=p=JG/,andso(a)=G. El (It, n) = 1. Conclude that the number of generators of G is y(n).
2.21. (i) Let G = (a) have order rs, where (r, s) = 1. Show that there are unique
CornMary 2.14 (Fermat). If p is a prirne and a is an integer, then a p a mod p. b, c E G with b of order r, c of order s, and a = bc.
(ii) Use part (i) to prove that if (r, s) = 1, then q(rs) = q(r)y(s).

-
Proof. Let G = U(Zp), the multiplicative group of nonzero elements of Z,;
since p is prime, bpis a field and G is a group of order p - 1.
Recall that for integers a and b, one has a b mod p if and only if [a] =
[b] in Z,. If a E Z and [a] = [O] in Zp, then it is clear that [alp = [0] = [a].
If [a] # CO], then [a] E G and so [a]*-' = [I], by Corollary 2.12; multiplying
I
2.22. (i) If p is prime, then cptpk)= pk - pk-I = p"(1 - l/p).
(ii) If the distinct prime divisors of n are p,, . .., p,, then
cp(tz) = n(1 - l/p,). . .( 1 - l/p,).
2.23 (Euler). If (r, s) = 1, then s'~'') = 1 mod r. (Hint.The order of the group or' units
by [a] now gives the desired result. U(Z,,>is (p(n1.1
 2. The Isomorphism Theorems Normal Subgroups 29

Proof: If I GI = n and if a E G satisfies a d = 1, where dln, then a is a root in F

Cyclic Groups of the polynomial xd - 1 E F[x]. Since a polynomial of degree d over a field
has at most d roots, our observation above shows that the hypothesis of
Lemma 2.15. If G is a cyclic group of order n, then there exists a unique Theorem 2.17 is satisfied. Statement (ii) follows at once from (i). 191
subgroup of order d for every divisor d of n.
When F is finite, the proof does not construct a generator of F " . Indeed,
Proof. If G = (a), then (anId) is a subgroup of order d, by Exercise 2.11.
no algorithm is known which displays a generator of Zb for all primes p.
Assume that S = (b) is a subgroup of order d (S must be cyclic, by Exercise
2.18). Now bd = 1; moreover, b = a mfor some m. By Exercise 2.13, md = nk
Theorem 2.19. Let p be a prime. A group G of order pn is cyclic if and only if it
for some integer k, and b = a"' = (a"ld)k.Therefore, (b) 5 <a"ld), and this
is an abelian group having a unique subgroup of order p.
inclusion is equality because both subgroups have order d.

Theorem 2.16. If n is a positive integer, then
where the sum is over all divisors d of n with 1 < d I n.
Proof. If C is a cyclic subgroup of a group G, let gen(C) denote the set of all
its generators. It is clear that G is the disjoint union
where C ranges over all the cyclic subgroups of G. We have just seen, when
G is cyclic of order n, that there is a unique cyclic subgroup Cd of order d for
every divisor d of n. Therefore, n = \GI = Ed,,Igen(C,)I. In Exercise 2.20,
however, we saw that lgen(Cd)i= q(d); the result follows.
Proof: Necessity follows at once from Lemma 2.15. For the converse, let
a E G have largest order, say pk (it follows that gpk = 1 for all g E G). Of
course, the unique subgroup H of order p is a subgroup of (a). If (a) is a
proper subgroup of G, then there is x E G with x $ (a) but with x p E (a); let
x P = a! If k = 1, then x P = 1 and x E H I (a), a contradiction; we may,
therefore, assume that k > 1. Now
so that 1 = pm for some integer m, by Exercise 2.13. Hence, x p = amp,and so
1 = ~ - ~ a "Since , so x-'a'" E H s (a).
' ~ . G is abelian, x-'arnp = ( ~ - ' a " ) ~and
This gives x E (a), a contradiction. Therefore, G = (a) and hence is cyclic.

We now characterize finite cyclic groups.

2.24. Let G = ( A , B) r GL(2, C), where
Theorem 2.17. A group G of order n is cyclic if and only if, for each divisor d of
n, there is at most one cyclic subgrotrp of G having order d. A= [u :] and [
B = -1 0'1 '

ProoJ If 6 is cyclic, then the r e s ~ ~isl tLemma 2.15. For the converse, recall Show that G is a nonabelian group (so G is not cyclic) of order 8 having a unique
from the previous proof that G is the disjoint union U gen(C), where C subgroup of order 2. (See Theorem 4.22.)
ranges over all the cyclic subgroups of G. Hence, n = /GI = xigen(C)I I
Ed,,q(d) = n, by Theorem 2.16. We conclude that G must have a cyclic sub-
group of order d for every divisor d of n; in particular, G has a cyclic sub- Normal Subgroups
group of order d = n, and so G is cyclic. B
This brief section introduces the fundamental notion of normal subgroups.
Observe that the condition in Theorem 2.17 is satisfied if, for every divisor
We begin with a construction which generalizes that of cosets.
d of n, there are at most d solutions x E G of the equation xd = 1 (two cyclic
subgroups of order d would contain more than d solutions).
Definition. If S and T are nonempty subsets of a group G, then
Theorem 2.13. S T = (st: s E S a n d t~ T).
(i) If F is a field and if G is a finite subgroup of F x , the multiplicative group
If S 5 G, t E G, and T = (t), then S T is the right coset St. Notice that the
of nonzero elements of F, then G is cyclic.
(ii) If F is a finite field, then its multiplicative group F is cyclic. family of all the nonempty subsets of G is a semigroup under this operation:
30 2. The Isomorphism Theorems
if S, T, and U are nonempty subsets of G, then (ST) U = S(TU),for either side
consists of all the elements of G of the form (st)u = s(tu) with s E S, t E and
U E U.
Theorem 2.20 (Product Formnalla). If S and T are subgroups of a finite group
G, then
lSTl IS n TI = IS1 [TI.
Remark. The subset S T need not be a subgroup.
Pro05 Define a function cp: S x T 4 S T by (s, t) I+ st. Since cp is a surjection,
it suffices to show that if x E ST, then Icp-'(x)l = I S n TI. We show that
cpV'(x) = {(sd,d-It): d E S n TI. It is clear that (pp-"(x)contains the right side.
For the reverse inclusion, let (s, t), (a, z) E cp-'(x); that is, s, a E S, t, z E T,
and st = x = ar. Thus, s-la = tz-' E S n T; let d = s-'a = tz-' denote their
common value. Then a = s(s-'a) = sd and d-'t = zt-'t = z, as desired.
There is one kind of subgroup that is especially interesting because it is
intimately related to homomorphisms.
Definaitionn. A subgroup K 5 G is a normal subgrotdp, denoted by K a G, if
gKg-' = IS for every g E G.
If K 5 G and there are inclusions gKg-' 5 K for every g E G, then K a G:
replacing g by g-', we have the inclusion g-'Kg 5 K , and this gives the
reverse inclusion K 5 gKg-'.
The kernel K of a homomorphism f : G 4 H is a normal subgroup: if
a E K, then f (a) = 1; if g E G, then f (gag-') = f (g)f (a)f (g)-' = f ( g )f (g)-' =
1, and so gag-1 E K. Hence, g ~ g - '5 K for all g E G, and so K a G. Con-
versely, we shall see later that every normal subgroup is the kernel of some
homomorphism.
In Example 2.5, we saw that if H is the cyclic subgroup of S , generated by
the transposition z = (1 2), then there are right cosets of H which are not left
cosets. When K is normal, then every left coset of K in G is a right coset.
Indeed, a subgroup I< of G is normal in G if and only if Kg = gK for every
g E G, for associativity of the multiplication of nonempty subsets gives I< =
(Kg)g-I = gKg-'. In terms of elements, this says that there is a partial
commutativity when K a G: if g E G and 1c E K, then there exists k' E I< with
alc = Ic'a. It may not be true that g commutes with every element of K. For
example, the reader should check that the cyclic subgroup K of S, generated
by the 3-cycle (1 2 3) is a normal subgroup. It follows that (1 2)K = K ( l 2)
even though (1 2) does not commute with (1 2 3).
Normal subgroups are also related to conjugations 7,: G + G, where
y,(x) = axa-' (see Exercise 2.34 below).
Normal Subgroups 31
Definition. If x E G, then a conjugate of x in G is an element of the form axa-'
for some a E G; equivalently, x and y are conjugate if y = y,(x) for some
fl E G.
For example, if k is a field, then matrices A and B in GL(n, k) are conjugate
if and only if they are similar.
2.25. If S is a subgroup of G, then SS = S; conversely, show that if S is a finite
nonempty subset of G with SS = S, then S is a subgroup. Give an example to
show that the converse may be false when S is infinite.
2.26. Let {Si: i E I ) be a family of subgroups of a group G, let {Siti: i E I ) be a family
0 n
of right cosets, and let D = Si. Prove that either Siti = or Siti = Dg 0
for some g.
2.27. If S and T are (not necessarily distinct) subgroups of G, then an (S-TI-double
coset is a subset of G of the form SgT, where g E G. Prove that the family of all
(S-T)-double cosets partitions G. (Hint. Define an equivalence relation on G by
a = b i f b = s a t f o r s o m e s ~ : S a n dT
t ~)
2.28. Let S, T _< G, where G is a finite group, and suppose that G is the disjoint union
Prove that [G: TI = xy=I=,
[S: S n g,~g;']. (Note that Lagrange's theorem is
the special case of this when T = 1.)
2.29 (i) (N. B.Mann). Let G be a finite group, and let S and T be (not necessarily
distinct) nonempty subsets. Prove that either G = S T or IG] 2 IS1 ]TI.+
(ii) Prove that every element in a finite field F is a sum of two squares.
2.30. If S _< G and [G : S] = 2, then S a G.
2.31. If G is abelian, then every subgroup of G is normal. The converse is false: show
that the group of order 8 in Exercise 2.24 (the quaternions) is a counterexample.
2.32. If H s G, then H a G if and only if, for all x, y E G, xy E H if and only if yx E H.
2.33. If I< I H I G and I< a G, then I< a H.
2.34. A subgroup S of G is normal if and only if s E S implies that every conjugate of
s is also in S. Conclude that if S _< G, then S a G if and only if y(S) I S for every
conjugation y.
2.35. Prove that SL(rz, k) a GL(n, lc) for every n 2 1 and every field k.
2.36. Prove that A , a S,, for every n.
2.37. (i) The intersection of any family of normal subgroups of a group G is itself a
normal subgroup of G. Conclude that if X is a subset of G, then there is a
smallest normal subgroup of G which contains X; it is called the normzal
sribgrozrp ge~aeuntedby X (or the normal closure of X; it is often denoted by
(X>G).
 2. The Isomorphism Theorems
(ii) If X = @, then ( x ) = a,
~ 1. If X # then (X)G is the set of all words on
the conjugates of elements in X.
(iii) If gxg-' E X for all x E X and g E G, then (X) = (X)G a G.
2.38. If H and K are normal subgroups of G, then H v K a G.
2.39. Prove that if a normal subgroup H of G has index 12, then g" E H for all g E G.
Give an example to show this may be false when H is not normal.
Quotient Groups
The construction of the quotient group (or factor group) GIN in the next
theorem is of fundamental importance.
We have already seen that if X and Y are nonempty subsets of a group G,
then their product
XY = (xy: x E X and y E Y)
defines an associative operation on the family of all nonempty subsets of G.
If H is a subgroup of G, then the family of all right cosets of H in G need not
be closed under this operation. In Example 2.5, we looked at the right cosets
of I-I = ((1 2)) in S,. The product of right cosets
is not a right coset of N, for it has four elements while right cosets of H have
two elements. In the proof of the next theorem, we shall see that if H is a
normal subgroup, then the product of two right cosets of N is also a right
coset of H.
Theorem 2.28. If N a G, then the cosets of N in G fornz a grottp, denoted by
GIN, of order [G : N].
P r o d To define a group, one needs a set and an operation. The set here is
the family of all cosets of N in G (notice that we need not bother with the
adjectives "feft" and "right" because N is a normal subgroup). As operation,
we propose the multiplication of nonempty subsets of G defined earlier. We
have already observed that this operation is associative. Now
NaNb = ~ a ( a - ' ~ a ) b(because N is normal)
= N(aa-')Nab = NNab = Nab (because N I 6).
Thus, NaNb = Nab, and so the product of two cosets is a coset. We let the
reader prove that the identity is the coset N = N 1 and that the inverse of Na
is N(n-I). This group is denoted by GIN, and the definition of index gives
IG/NI = [G:N].
Quotient Groups 33
Corollary 2.22. If N a G, then the natural map (i.e., the function v: G -, GIN
defined by v(a) = Na) is a surjective homomorphism with kernel N.
ProoJ The equation v(a)v(b) = v(ab) is just the formula NaNb = Nab; hence,
v is a homomorphism. If Na E GIN, then Nn = v(a), and so v is surjective.
Finally, v(a) = Nu = N if and only if a E N, by Lemma 2.8, so that N = ker v.
Ed
We have now shown that every normal subgroup is the kernel of some
homomorphism. Different homomorphisms can have the same kernel. For
example, if a = (1 2) and b = (1 3), then y,, y,: S, -,S, are distinct and
ker y, = 1 = ker y,.
The quotient group construction is a generalization of the construction of
Z, from Z. Recall that if n is a fixed integer, then [a], the congruence class of
+
a mod n, is the coset a (n). Now (n) a Z,because Z is abelian, and the
+
quotient group Z/(n) has elements all cosets a (n), where a E Z, and oper-
+ + + + +
ation (a (n)) (b (n)) = a b (n); in congruence class notation,
[a] + +
[b] = [a b]. Therefore, the quotient group Z/(n) is equal to Z,, the
group of integers modulo n. An arbitrary quotient group GIN is often called
G mod N because of this example.
Definitioira. If a, b E G, the csmmu~aaor~
of a and b, denoted by [a, b], is
[a, b] = aba-'b-'.
The counrnl~tntorsubgroa~p(or derived subgroup) of G, denoted by G', is the
subgroup of G generated by all the commutators.
We shall see, in Exercise 2.43 below, that the subset of all commutators
need not be a subgroup (the product of two commutators need not be a
commutator).
Theorem 2.23. The co~nrnutatorsubgroup G' is a normal subgroup of G. More-
over, if N a G, then GIH is abelian if and only if G' < H.
ProoyC I f f : G -+ G is a homomorphism, then f (G') IG' because f([a, b]) =
[fa, fb]. It follows from Exercise 2.34 that G' a 6 .
Let H a G. If GIN is abelian, then HaHb = HbHn for all a, b E G; that is,
Nab = Nba, and so ab(ba)-' = aba-'b-I = [a, b] E H. By Corollary 2.6, G'
N. Conversely, suppose that G' I 19. For every a, b E G, ab(ba)-' = [a, b] E
6' I H, and so Hab = Hba; that is, GIH is abelian. El
Those who write conjugates as b-lab write commutators as a-'b-lab.
34 2. The Isomorphism Theorems
2.40. Let H a G, let v: G -+ GIH be the natural map, and let X c G be a subset such
that v(X) generates GIH. Prove that G = (H u X ) .
2.41. Let G be a finite group of odd order, and let x be the product of all the elements
of G in some order. Prove that x E G'.
2.42 (P. Uff). For any group G, show that G' is the subset of all "long commutators":
G' = {a, a,. . .a,,a;'ay1 . . .a,': ai E G and n 2 2).
(Hint (P.M. Weislhsel).
2.43. The fact that the set of all commutators in a group need not be a subgroup is an
old result; the following example is due to P.J. Cassidy (1979).
(i) Let k[x, y] denote the ring of all polynomials in two variables over a field
k, and let k[x] and k[y] denote the subrings of all polynomials in x and in
y, respectively. Define G to be the set of all matrices of the form
where f(x) E k[x], g(y) E Ic[J~], and h(x, y) E k[x, y]. Prove that G is a
multiplicative group and that G' consists of all those matrices for which
f(x) = 0 = g(y). (Hint. If A is denoted by (f,g, h), then (f,g, h)(f ', g', h') =
+ + + +
(f f', g g', h h' fg'). If h = h(x, y) = ?,xiyj, then
(ii) If (0, 0, h) is a commutator, then there are polynomials f(x), f'(x) E k[x]
and g(y), g'(y) 63 IcL-yl with 4x9 Y ) = f(x)g'(y) - f'(x)g(y).
+ +
(iii) Show that h(x, y) = x2 xy y2 does not possess a decomposition as
in part (ii), and conclude that (0, 0, h) E G' is not a commutator. (Hint. If
C
f(x) = bixi and f '(x) = cixi, then there are equations
Considering k[x, y] as a vector space over k, one obtains the contradiction
that the independent set {1, y, y2) is in the subspace spanned by {g, g').)
Remark. With a little ring theory, one can modify this construction to
give a finite example. If lc = Z,and k[x, y] is replaced by its quotient ring
k[x, y]/I, where I is the ideal generated by {x3,y3, x2y, xy2), then the cor-
responding group G has order P ' ~ Using
. the computer language CAYLEY
(now called MAGMA), I found that the smallest group in which the prod-
uct of two commutators is not a commutator has order 96. There are
exactly two such groups: in CAYLEY notation, they are library g96n197
and library g96n201; in each of these groups, the commutator subgroup has
order 32 while there are only 29 commutators.
The Isomorphism Theorems 35
There is an explicit example in [Carmichael, p. 391 of a group G I S16
(generated by eight permutations) with IGI = 256, IG'l = 16, and with a
specific element of G' which is not a commutator.
The Isomorphism Theorems
There are three theorems, formulated by E. Noether, describing the relation-
ship between quotient groups, normal subgroups, and homomorphisms. A
testimony to the elementary character of these theorems is that analogues of
them are true for most types of algebraic systems, e.g., groups, semigroups,
rings, vector spaces, modules, operator groups.
Theorem 2.24 (First Isomorphism Tlimeorem). Let f : G -+ H be a hornonzor-
phism with lcernel K. Then K is a noruzal subgroup of G and G/K s im f.
Pro05 We have already noted that K a G. Define y: G/K -,H by
cp (Ka) = f (a).
To see that cp is well defined, assume that Ka = Kb; that is, ab-' E K. Then
1 = f (ab-') = f(a)f (b)-', and f (a) = f(b); it follows that y(Ka) = y(Kb), as
desired. Now y is a homonlorphism:
It is plain that im cp = im f. Finally, we show that y is an injection. If cp(Ka) =
y(Kb), then f (a) = f (b); hence f (ab-') = 1, ab-' E K, and Ka = Kb (note
that y being an injection is the converse of q~ being well defined). We have
shown that y is an isomorphism. Dl
It follows that there is no significant difference between a quotient group
and a homomorphic image.
If v : G -t G/K is the natural map, then the following "commutative dia-
gram" (i.e., f = q 0 v) with surjection v and injection y describes the theorem:
It is easy to describe y-l: im f -+ G/IC: if x E im f, then there exists a E G
with f(a) = x, and y-'(x) = Ka. The reader should check that cp-I is well
defined; that is, iff (b) = x, then Ka = Kb.
Given a homomorphism f, one must always salivate, like Pavlov's dog, by
aslting for its Iternel and image; once these are ltnown, there is a normal
subgroup and f can be converted into an isomorphism. Let us illustrate this
36 2. The Isomorphism Theorems
by solving Exercise 2.19: If G = (g) and H = (h) are cyclic groups of order
rz, then G s H. Define a homomorphism f : Z -+ G by f(k) = yk. It is easy to
see that f is a surjection, while Exercise 2.13 shows that kerf = (n). The first
isomorphism theorem gives Z/(n) r G. Similarly, Z/(rz) r H, and so G r H.
Since Z/(n) = Z,,, every cyclic group of order n is isomorphic to Z,,.
Lemma 2.25. If S nrzd T are subgroups of G and if one of them is normal, then
S T = S v T=TS.
Proof. Recall that S T is just the set of all products of the form st, where s E S
and t E T; hence S T and TS are subsets of S v T containing S u T. If S T and
TS are subgroups, then the reverse inclusion will follow from Corollary 2.6.
Assume that T a G. If s, t , and s2t2E ST, then
(s, t1)(s2t,)-l =~ ~ t ~ t 2 ~ ~ ; ~
= s,s,lt,
= (s,s,')t, E ST,
where t, = s2(t,t,')sz1 E T because T a G. Therefore, ST = S v T. A simi-
lar proof shows that TS is a subgroup, and so TS = S v T = ST.
Suppose that S < H 5 G are subgroups with S a G. Then S a H and the
quotient HIS is defined; it is the subgroup of G/S consisting of all those cosets
Sh with h E H. In particular, if S a G and T is any subgroup of G, then
S _< S T 5 G and ST/S is the subgroup of G/S consisting of all those cosets
Sst, where st E ST. Since Sst = St, it folfows that ST/S consists precisely of all
those cosets of S having a representative in T.
Recall the product formula (Theorem 2.20): If S, T I G, then ISTI ISn TI =
IS// TI; equivalently, I TI//S n TI = I STI//SI. This suggests the following
theorem.
Theorem 2.26 (Seco~BgHsomorphism Theorem). Let N nrzd T be subgroups of
G with N normal. Then N n T is normal in T and T/(N n T) r NT/N.
Rernark. The following diagram is a mnemonic for this theorem:
Correspondence Theorem 37
proof. Let v: G -, GIN be the natural map, and let v' = v (T, the restriction of
v to T. Since v' is a homomorphism whose kernel is N n T, Theorem 2.24
gives N n T a T and T/(N n T) E im v'. Our remarks above show that im v'
is just the family of all those cosets of N having a representative in T; that is,
irn v' consists of all the cosets in NTIN. El
Theorem 2.27 (Third l[somorphism Theore~n).Let K 5 H 5 G, where both K
and H are normal subgroups of G. Then HIK is a normal szibgroup of G/K and
PPUooJ Again we let the first isomorphism theorem do the dirty work. Define
f : G/K -+ G/H by f(I<a) = Ha (this "enlargement of coset" map f is well
defined because K I H). The reader may check easily that f is a surjection
with kernel H/K. El
Imagine trying to prove the third isomorphism theorem directly; the ele-
ments of (G/I<)/(H/K) are cosets whose representatives are cosets!
2.44. Prove that a homomorphism f : G -+ H is an injection if and only if ker f = 1.
2.45. (i) Show that the 4-group V is a normal subgroup of S4. (We shall do this more
efficiently in the next chapter.)
(ii) If K = ((1 2)(3 4)), show that I<a V but that K is not a normal subgroup
of S,. Conclude that normality need not be transitive; that is, K a H and
H a G need not imply K a G.
2.46. Let N a G and let f : G -+ H be a homomorphism whose kernel contains N.
Show that f induces a homomorphism A,: GIN -+ H by f,(Na) = f (a).
2.47. If S, T 5 G, then ST is a subgroup of G if and only if ST = TS.
2.48 (Modulrar Law). Let A, B, and C be subgroups of G with A 2 B. If A n C =
B n C and AC = BC (we do not assume that either AC or BC is a subgroup),
then A = B. ,
2.49 (DedeErilmd Law). Let H, K, and L be subgroups of G with H < L. Then
HI< n L = H(I< n L) (we do not assume that either HI< or H(I< n L) is a sub-
group).
2.50. Let f: G -+ G* be a homomorphism and let S* be a subgroup of G*. Then
f -'(S*) = {x E G: f(x) E S * ) is a subgroup of G containing lter f.
Correspondence Theorem
The theorem in this section should be called the fourth isomorphism theo-
rem. Let X and X* be sets. A function f : X -+ X* induces a "forward
motion7' and a "bacltward motion" between subsets of X and subsets of X*.
38 2. The Isomorphism Theorems
The forward motion assigns to each subset S c X the subset f ( S ) =
( f ( s ) :s E S } of X * ; the backward motion assigns to each subset S* of X* the
subset f -'(S*) = { x E X : f ( x ) E S " ) of X . Moreover, iff is a surjection, these
motions define a bijection between all the subsets of X* and certain subsets
of X. The following theorem is the group-theoretic version of this.
Theorem 2.28 (Correspondence Theorem). Let I< a G and let v: G --+ G/K be
the natural map. Then S t-, v(S) = S/I< is a bijection porn the fatnily of all
those subgroups S of G which contain K to the family of all the subgrozrps
of C/K.
Moreover, if we denote S / K by S*, then:
(i) T I S i f and only if T* 5 S*, and then [S : T I = [S" : T * ] ; and
(ii) T a S i f and only i f T * a S*, and then SIT r S*/T".
Remark. A mnemonic diagram for this theorem is:
ProoJ We show first that S t-, S / K is an injection: if S and T are subgroups
containing I<, and if S/I< = T I K , then S = T. To see this, let s E S ; since
S/I< = T/I<, there exists t E T with K s = K t . Hence, s = kt for some 12 E I< <
T and s E T. The reverse inclusion is proved similarly. To see that the corre-
spondence S H S/K is a surjection, we must show that if A < G / K , then
there is a subgroup S of G containing K with A = SIP;(. By Exercise 2.50,
S = v-'(A) is a subgroup of G containing I<; moreover, that v is a surjection2
implies that S/K = v(S) = vv-"A) = A.
It is plain that if K 5 T 5 S, then T / K < S/K. To prove that [S : T I =
[S* : T * ] , it suffices to show that there is a bijection from the family of all
cosets of the form Ts, where s E S, to the family of all cosets T*s*, where
s* E S*. The reader may check that a, defined by a: Ts t-+ T*v(s), is such a
Iff: X -+ X* is a function and A c X*, then ff-'(A) c A; if f is a surjection, then ff-'(A) = A.
Correspondence Theorem 39
bijection. (If G is finite, then we may prove [S : T I = [S" : T * ] as follows:
[S* : T " ] = IS"I/IT*I = IS/KI/IT/KI
If T a S, then the third isomorphism theorem gives T / K a S/K and
(S/I<)/(T/K)r S / K ; that is, T* a S* and S*/T* r SIT. It remains to
show that if T * a S*, then T a S. The reader may verify that T =
lcer pv,: S -* S*/T*, where v, = vlS and p: S* -t S*/T* is the natural map.
B
2.51. If G' < N 5 G, where G' is the commutator subgroup of G, then H a G and
G/H is abelian.
2.52. Give an example to show that if H a G, then G need not contain a subgroup
isomorphic to G/H.
2.53. Prove that the circle group T is isomorphic to R/Z.
2.54. (i) Let H,I< s G.If(IHI,JKI)= 1, t h e n H n K = 1.
(ii) Let G be a finite group, and let H be a normal subgroup with (1H 1, [G :a)
=
1. Prove that H is the unique such subgroup in G. (Hint: If K is another
such subgroup, what happens to K in GIH?)
2.55 (%assenhaus).Let G be a finite group such that, for some fixed integer n > 1,
(xy)" = x"ynfor all x y E G. If G[n] = {z E G: zn = 1) and G" = {xu:x E G), then
both G[n] and G* are normal subgroups of G and IG"l = [G : G[n]].
2.56. A subgroup II 5 G is a ~ n ~ x i ~normal
n a l snbgrollp of G if there is no normal
subgroup N of G with H < N < G. Prove that H is a maximal normal subgroup
of G if and only if G/H has no normal subgroups (other than itself and 1).
DeG~itioaa.A group G # 1 is si'01~pBe
if it has no normal subgroups other than
6; and 1.
We may restate Exercise 2.56: H is a maximal normal subgroup of G if and
only if G/H is simple.
2.57. An abelian group is simple if and only if it is finite and of prime order.
2.58. Let R4 be a maximal subgroup of G; that is, there is no subgroup S with A4 <
S < G. Prove that if A4 a G, then [G : MI is finite and equal to a prime
2.59 (Schur). Let f : G -+ H be a homomorphism that does not send every element of
G into 1. If G is simple, then f must be an injection.

40 2. The Isomorphism Theorems
Direct Products
Definition. If H and K are groups, then their direct product, denoted by
H x K, is the group with elements all ordered pairs (h, k), where h E H and
k E K, and with operation
(h, k)(hl, k') = (hh', kk').
It is easy to check that H x K is a group: the identity is (1, 1); the inverse
(h, 1c)-I is (h-l, lc-l). Notice that neither H nor K is a subgroup of H x K, but
H x K does contain isomorphic replicas of each, namely, H x 1 = {(h, 1):
h E H) and 1 x K = {(1, k): lc E K).
2.60. (i) Show that (h, 1) E H x 1 and (1, k) E 1 x I< commute.
(ii) H x 1 and 1 x K are normal subgroups of H x I(.
(iii) (H x l ) n ( l x K ) = 1 and(H x 1)(1 x K ) = H x K.
2.61. H x K is abelian if and only if both H and I< are abelian.
2.62. (i) Prove that Z , r Z 2 x Z,.
(ii) If (myn) = 1, then Z,,, E Z,, x Z.. (Hint. Use the Chinese Remainder
Theorem).
2.63. If p is a prime, prove that ZI,2 $ Z , x Z,.
2.64. Let p: G x G 4 G be the operation on a group G; that is, p(n, b) = nb. If G x G
is regarded as the direct product, prove that p is a homomorphism if and only
if G is abelian.
2.65. Let A be an abelian group, and let cc: H -+ A and P: I( -+ A be homomorphisms.
Prove that there exists a unique homomorphism y: H x I< -+ A with y(h, 1) =
a(h)for all h E H and y(1, k) = j(k) for all k E K . Show that this may be false if
A is not abelian.
We now take another point of view. It is easy to multiply two polynomials
together; it is harder to factor a given polynomial. We have just seen how to
multiply two groups together; can one factor a given group?
Theorem 2.29. Let G be a group with normal subgroups H and K. If HIC = G
and H n I C = 1, then G E H x I<.
Proof: If a E G, then a = hk for some h E H and k E K (because G = HK). We
claim that h and k are uniquely determined by a. If a = h, k, for h, E H and
k , E K, then hk = h,k, and h-'h, = klc;' E H n K = 1; hence h = h, and
lc = k,.
Define f : G -+H x K by f(a) = (h, k), where a = hk. Is f a homomorphism?
If a = hk and a' = h'k', then aa' = hkh'k' which is not in the proper form
Direct Products 41
for evaluating f. Were it true that kh' = h'k, however, then we could eval-
uate f(aai). Consider the commutator h'khl-'k-'. Now (h'khl-')k-' E K (for
h'khl-' E K because K is normal), and, similarly, hf(kh'-'k-') E H (because H
is normal); therefore, h'khl-'k-' E H n K = 1 and h' and k commute. The
reader can now check that f is a homomorphism and a bijection; that is, f is
an isomorphism.
We pause to give an example showing that all the hypotheses in Theorem
2.29 are necessary. Let G = S3, H = ((1 2 3)), and K = ((1 2)). It is easy to
see that HK = G and H n K = 1; moreover, H CI G but K is not a normal
subgroup. The direct product H x K s B3 x B, is abelian, and so the non-
abelian group G = S, is not isomorphic to H x K.
Theorem 2.30. If A a H and B a K , then A x B a H x K and
Proof: The homomorphism 9:H x K -+ (HIA) x (KIB), defined by q(h, Ic) =
(Ah, Bk), is surjective and ker = A x B. The first isomorphism theorem
now gives the result.
It follows, in particular, that if N a H, then N x 1 a H x K.
Corollary 2.31. If G = H x K, then G/(H x 1) E K.
There are two versions of the direct product H x K: the external version,
whose elements are ordered pairs and which contains isomorphic copies of
H and K (namely, H x 1 and 1 x K); the iaternal version which does contain
H and K as normal subgroups and in which HK = G and H n K = 1. By
Theorem 2.29, the two versions are isomorphic. In the future, we shall not
distinguish between external and internal; in almost all cases, however,
our point of view is internal. For example, we shall write Corollary 2.31 as
(H x K)/H E K.
2.66. Prove that V z Z , x Z2.
2.67. Show that it is possible for a group G to contain three distinct normal sub-
groups N,K, and L such that G = H x L = K x L; that is, H L = G = K L and
H n L = 1 = K n L. (Hint: Try G = V).
2.68. Prove that an abelian group G of order P', where p is a prime, is either cyclic or
isomorphic to 5 x Z,. (We shall see in Corollary 4.5 that every group of order
p2 must be abelian).
2.69. Let G be a group with normal subgroups H and K. Prove that HK = G and
H n K = 1 if and only if each a E G has a unique expression of the form a = hk,
where I? E H and k E K.
42 2. The Isomorphism Theorems

2.70. If N a H x K, then either N is abelian or N intersects one of the factors H or CHAPTER 3

K nontrivially.
2.71. Give an example of an abelian group H x K which contains a nontrivial sub- Symmetric Groups and G-Sets
group N such that N n H = 1 = N n K. Conclude that it is possible that N I
H x K and N # (N n H) x (N n K).
2.72. Let G be a group having a simple subgroup H of index 2. Prove that either H is
the unique proper normal subgroup of G or that G contains a normal subgroup
K of order 2 with G = H x K. (Hint. Use the second isomorphism theorem.)
2.73. Let 0 denote the trivial homomorphism which sends every element to the iden-
tity. Prove that G r H x K if and only if there exist homomorphisms
HAG$K and K ~ G ~ H
with qi = 1, (the identity function on H), pj = I,, pi = 0, qj = 0, and
(i o q)(x)(jo p)(x) = x for all x E G.
2.74. The operation of direct product is commutative and associative in the following
sense: for groups H, K, and L,
HxICrKxH and Hx(KxL)r(HxK)xL.
The definition of group arose from fundamental properties of the symmetric
Conclude that the notations H, x .. x HI,and n;=, Hi are unambiguous. group S,. But there is another important feature of S,: its elements are func-
2.75. Let G be a group having normal subgroups H,, ...,HI,. tions acting on some underlying set, and this aspect is not explicit in our
(i) If G = (U:=, Hi) and, for all j, 1 = Hj n (Ui,j Hi), then G z presentation so far. The notion of G-set is the appropriate abstraction of this
HI x -.. x H,. idea.
(ii) If each a E G has a unique expression of the form a = h, ...h,,, where each
hi E Hi, then G r H, x .. . x H,,.
2.76. Let H,, ..., HI, be normal subgroups of a group G, and define Conjugates
q : G -+ G/H, x ... x G/Hn by q(x) = (Hlx, ..., Hl,x).
(i) Prove that lter q = H, n -.. n H,.
In this section we study conjugates and conjugacy classes for arbitrary groups;
(ii) If each Hi has finite index in G and if (IG/HiJ,IG/HjI) = 1 for all i # j, then q
is a surjection and
in the next section, we consider the special case of symmetric groups.

[G: H, n - .. n HI,] = ,n1

11

r=l
G/HiI. Lemma 3.1. If G is a group, then the relation " y is a eor$udgate of x in G," that
is, y = gxg-' for some g E G, is an equivalence relation.
2.77. Let V be an n-dimensional vector space over a field F. Prove that, as abelian
groups, V r F, x ..- x F,,, where Fir F for all i. Pros$. Routine.
Definition. If p is a prime, then an elementary ahelian p - ~ P Q is
H aUfinite
~ group Defiaitiow. If G is a group, then the equivalence class of a E G under the
G isomorphic to Z p x ... x Z p . of x in G" is called the co~tjugacyclas.~of a; it is
relation "y is a conj~~gate
denoted by aG.
2.78. Prove that every abelian group G of prime exponent p is elementary abelian,
that G is a vector space over Z,,and that every homomorphism q:G -+ G is a
linear transformation. Of course, the conjugacy class a Gis the set of all the conjugates of a in 6.
Exercise 2.4S(i) can be rephrased: a subgroup is normal if and only if it is a
(disjoint) union of conjugacy classes. If a and b are conjugate in G, say, b =
gag-', then there is an isomorphism y: G -+ G, namely, conjugation by 8,
with y (a) = b. It follows that all the elements in the same conjugacy class have
 3. Symmetric Groups and G-Sets
44
the same order. In particular, for any two elements x, y E G, the elements xy
and yx have the same order.
If a E G is the sole resident of its conjugacy class, then a = gag-' for all
g E G; that is, a commutes with every element of G.
Definition. The center of a group G, denoted by Z(G), is the set of all a E G
that commute with every element of G.
It is easy to check that Z(G) is a normal abelian subgroup of G.
The following subgroup is introduced to count the number of elements in
a conjugacy class.
Definition. If a E G, then the centrulizer of a in G, denoted by CG(a),is the set
of all x E G which commute with a.
It is immediate that CG(a)is a subgroup of G.
Theorem 32. If a E G, the number of conjugates of a is equal to the index of
its centralizer:
laGI = CG : C,(a)l,
and this number is a divisor of ]GI when G is finite.
PuooJ Denote the family of all left cosets of C = CG(a)in 6 by G/C, and
define f : a G-+ G/C by f(gag-') = gC. NOWf is well defined: if gag-' = huh-'
for some h E G, then h-'gag-'h = a and h-'g commutes with a; that is, h-'g E
C. and so hC = gC. The function f is an injection: if gC = f(gagdl) . =
- 1
f(knkd') = kC fo; some k E G, then k-'g E C, k-'g commutes with a,
k-'gag-'k = a, and gag-1 = kak-'; the function f is a surjection: if g E G,
then gC = f(gag-I). Therefore, f is a bijection and laG/= IG/CI =
[G : CG(a)]. When G is finite, Lagrange's theorem applies. (3
One may conjugate subgroups as well as elements.
Definition. If H I G and g E G, then the eorrjargate gHg-' is {ghg-': h E H).
The conjugate gHg-' is often denoted by Hg.
The conjugate g ~ g - 'is a subgroup of G isomorphic to H: if y,: G -+ G is
conjugation by 8, then yglH is an isomorphism from H to gHg-'.
Note that a subgroup H is a normal subgroup if and only if it has only one
conjugate.
DeEinition. If H IG, then the normalizm of H in G, denoted by NG(H),is
NG(H)= {a E G: aHa-' = H).
Conjugates 45
It is immediate that NG(H)is a subgroup of G. Notice that H a NG(H);
indeed, NG(H)is the largest subgroup of G in which H is normal.
Theorem 3.3. If H 5 G, then the number c of conjtigates of H in G is equal to
the index of its normalizer: c = [G : NG(H)], and c divides 1 GI when G is finite.
Moreover, aHa-' = bHb-' if and only if b-'a E NG(H).
Proof. Let [HI denote the family of all the conjugates of H, and let GIN
denote the family of all left cosets of N = NG(H)in G. Define f : [HI -+GIN
by f ( a ~ a - ' )= aN. Now f is well defined: if aHa-' = bHb-' for some b E G,
then b-'ahla-'b = H and b-'a normalizes H; that is, b-'a E N , and so
bN = aN. The function f is an injection: if aN = f(aHaP') = f(cHc-') =
cN for some c E G, then c-'a E N, c-'a normalizes H, c-'aHa-'c = H, and
aHa-' = cHc-'; the function f is a surjection: if a E G, then a N = f(aHa-I).
Therefore, f is a bijection and / [HI / = I GIN I = [G : NG(H)]. When G is finite,
Lagrange's theorem applies. Ib3
The strong similarity of Theorems 3.2 and 3.3 will be explained when we
introduce G-sets.
3.1. (i) A group G is centerless if Z(G) = 1. Prove that S,, is centerless if n > 3.
(ii) Prove that A, is centerless.
3.2. If a E S, is an n-cycle, then its centralizer is (a).
3.3. Prove that if G is not abelian, then G/Z(G) is not cyclic.
3.4. (i) A finite group G with exactly two conjugacy classes has order 2.
(ii) Let G be a group containing an element of finite order n > 1 and exactly two
conjugacy classes. Prove that IGI = 2. (Nint. There is a prime p with a n = 1
for all a E G. If p is odd and a E G, then a' = xax-' for some x, and so
aZk= ax-^ for all k 2 1.)
(There are exnmples of infinite groups 6 with no elements of finite order which
do have exactly two conjugacy classes.)
3.5. Prove that Z(G, x - x G,) = Z(G,) x . .. x Z(G,,).
3.6. (i) Prove, for every a, x t G, that CG(axa-l) = aCG(x)a-l.
(ii) Prove that if H 5 G and h E H, then C,(h) = CG(h)n H.
3.7. Let G be a finite group, let H be a normal subgroup of prime index, and let
x E H satisfy C,(x) < CG(x).If y E H is conjugate to x in G, then y is conjugate
to x in H.
3.8. If a,, .. .,a,, is a list of (not necessarily distinct) elements of a group G, then, for
a11 i, a i . . . a , a 1 . ..ai-, is conjugate to a, ... a,,.
3.9. (i) Prove that NG(aHa-') = aNG(H)aW'.
(ii) If H 5 K G, then N K ( H )= NG(H)n I<.
 3. Symmetric Groups and G-Sets
(iii) If H, K 5 G, prove that NG(H)n NG(K)5 NG(Hn K). Give an example in
which the inclusion is proper.
3.10. Iff : G -t H is surjective and A IZ(G), then f(A) 5 Z(H).
3.1 1. If H I G, then NG(H)_< {a E G: @Ha-' 5 H}; when H is finite, then there is
equality. (There are examples of infinite subgroups H 5 G with aHa-l < H for
some a E G).
Definition. An n x n matrix M = [mij] over a field K is monomial if there is
a E St,and (not necessarily distinct) nonzero elements x,, . . . , x,, E K such that
xi if j = a ( i ) ,
mij =
0 otherwise.
Monomial matrices thus have only one nonzero entry in any row or col-
umn. Of course, a monomial matrix in which each xi = 1 is a permutation
matrix over I<. (This definition will be generalized when we discuss wreath
products.)
3.12. (i) Let k be a field. If G = GL(n, Ic) and T is the subgroup of G of all diagonal
matrices, then NG(T)consists of all the monomial matrices over Ic.
(ii) Prove that NG(T)/Tg S,,.
3.13. (i) If H is a proper subgroup of G, then G is not the union of all the conjugates
of H.
(ii) If G is a finite group with conjugacy classes C,, ..., C,,, and if g iE Ci, then
Symmetric Groups
Definition. Two permutations a, P E S, have the same cycle strlrcttwe if their
complete factorizations into disjoint cycles have the same number of r-cycles
for each r.
Lemma 3.4. If a, ,!? E St,, then spa-' is the permutation with the same cycle
stru.ctui.eas P which is obtained by applying a to the synzbols in P.
EXAMPLE 3.1. If ,!?= (1 3)(2 4 7) and a = (2 5 6)(1 4 3), then spa-I =
(a1 a3)(a2 a 4 a7) = (4 1)(5 3 7).
Proof. Let n be the permutation defined in the lemma. If P fixes a symbol i,
then n fixes a(i), for a(i) resides in a 1-cycle; but spa-l(a(i)) = aP(i) = a(i),
and so spa-' fixes a(i) as well. Assume that ,!? moves i; say, P(i) = j. Let the
complete factorization of P be
Symmetric Groups 47
If a(i) = 12 and a ( j ) = 1, then E: k H 1. But apor-l: k H i H j I-+ 1, and so
apaW1(k)= n(k). Therefore, TC and spa-I agree on all symbols of the form
k = a(i); since a is a surjection, it follows that n = spa-l.
Theorem 3.5. Permutaations a, P E S,, are conjugate if and only if they have the
sanze cycle structure.
Proof. The lemma shows that conjugate permutations do have the same cycle
structure. For the converse, define y E Sn as foIlows: place the complete fac-
torization of a over that of p so that cycles of the same length correspond,
and let y be the function sending the top to the bottom. For example, if
then y(i) = k, y(j) = E, etc. Notice that y is a permutation, for every i between
1 and n occurs exactly once in a complete factorization. The lemma gives
yay-1 = P, and so a and /? are conjugate.
1 2 3 4 5 6
then y = (2 4) = (1 2 5)(3 6 4) Notice that y is not unique; for
example, the 3-cycle in a could also be written (1 2 3), and the "downward"
permutation is now y' = (1 5)(2 6 4 3). The multiplicity of choices for y is
explained by Theorem 3.2.
Corollhny 3.6. A subgroup H of S,, is a normal subgroup if and only if, whenever
a E H, then every /3 having the same cycle structure as a also lies in N.
Proof. By Exercize 2.34, H a Sn if and only if H contains every conjugate of
its elements. El
The solution of Exercise 2.45(i), which states that Y a S,, follows from the
fact that V contains all products of disjoint transpositions.
If 1 < r < n, then Exercise 1.5 shows that there are exactly
(l/r) [n(n - 1). . . (n - r + I)] distinct r-cycles in S,,. This formula can be used
to compute the number of permutations having any given cycle structure if
one is careful about factorizations with several factors of the same length. For
example, the number of permutations in S, of the form (a b)(c d) is
the factor 1 occurring so that we do not count (a b)(c d) = (c d)(a b) twice.
 3. Symmetric Croups and G-Sets
S5
Cycle Structure Number Order Parity
s4 1 1 Even
Cycle Structure Number Order Parity 10 = (5 x 4)/2 2 Odd
1 Even 20 = (5 x 4 x 3)/3 3 Even
(1) 1
6 = (4 x 3)/2 2 Odd 30 = (5 x 4 x 3 x 2)/4 4 Odd
(12) 24 = 5!/5
8 = (4 x 3 x 2)/3 3 Even 5 Even
(123)
(1234) 6 = 4!/4 4 Odd

Even
5 ~ 4 x 32 x 1
(123) (45) 20=-x- 6 Odd
3 2

Table 3.1
A5
Cycle Structure Number
Order Parit,v
,.,
Let us now examine S, using Table 3.1. The 12 elements of A , are eight 1 Even
3-cycles, three products of disjoint transpositions, and the identity. These ele- 3 Even
ments are the 4-group V together with 5 Even
2 Even
(1 2 3); (1 3 2); (2 3 4); (2 4 3);

Table 3.2
We can now see that the converse of Lagrange's theorem is false.
SG
Cycle Structure Number Order Parity
Theorem 3.7. A, is a group of order 12 having no subgroup of order 6.
Cl (1) 1 1 Even
Proof (T.-&. Sheu). If such a subgroup H exists, then it has index 2, and so c
2 (12) 15 2 Odd
Exercise 2.16 gives a2 E H for every a E A,. If a is a 3-cycle, however, then c3 (123) 40 3 Even
C4 (1234) 90
~ , this gives 8 elements in H , a contradiction. C3
a = a4 = ( u ~ )and C, (12345) 144
4 Odd
5 Even
CG (123456) 120 6 Odd
c7 (12)(34) 45 2 Even
C8 (12)(345) 120 6 Odd
3.14. (i) If the conjugacy class of x E G is (a,, . .. , a,}, then the conjugacy class of x-' C9 (12) (3456) 90 4 Even
is (a;', ..., a;'}. Cl0 . (12)(34)(56) 15 2 Odd
(ii) If a E S", then a is conjugate to awl. C II (123)(456) 40 3 Even
3.15. A, is the only subgroup of S4 having order 12. 720 = 6!

Definition. If n is a positive integer, then a partition of n is a sequence of A,

Cycle Structure Number Order Parity
integers 1 5 i, I i, I I i, with ij = n.
Even
3.16. Show that the number of conjugacy classes in S,, is the number of partitions of n. Even
3.17. If n < m, then A,, can be imbedded in A,,, (as all even permutations fixing ( n + 1, Even
..., m]). Even
Even
3.18. Verify the entries in Table 3.2. Even
3.19. Verify the entries in Table 3.3.

Table 3.3
50 3. Symmetric Groups and G-Sets
The Simplicity of A,
We are going to prove that A, is simple for all n 2 5. The alternating group
A, is not simple, for it contains a normal subgroup, namely, V.
Lemma 3.8. A, is simple.
Proof. (i) All 3-cycles are conjugate in A,. (We know that this is true in S,, but
now we are allowed to conjugate only by even permutations.)
If, for example, a = (1 2 3), then the odd permutation (4 5) commutes
with a. Since A, has index 2 in S,, it is a normal subgroup of prime index, and
so Exercise 3.7 says that a has the same number of conjugates in A, as it does
in S, because CA5(a)< CsS(cr).
(ii) All products of disjoint transpositions are conjugate in A,.
If, for example, a = (1 2)(3 4), then the odd permutation (1 2) commutes
with a. Since A, has index 2 in S,, Exercise 3.7 says that a has the same
number of conjugates in A, as it does in S,.
(iii) There are two conjugacy classes of 5-cycles in A,, each of which has 12
elements.
In S,, a = (1 2 3 4 5) has 24 conjugates, so that CS5(a)has 5 elements;
these must be the powers of a. By Exercise 3.2, CA5(a)has order 5, hence,
index 60/5 = 12.
We have now surveyed all the conjugacy classes occurring in A,. Since
every normal subgroup H is a union of conjugacy classes, /HI is a sum of 1
and certain of the numbers: 12, 12, 15, and 20. It is easily checked that no
such sum is a proper divisor of 60, so that 1HI = 60 and A, is simple. C3
Lemma 3.9. Let H a A,,, where n 2 5. If M contains a 3-cycle, then N = A,,.
Pro& We show that (1 2 3) and (i j k) are conjugate in A, (and thus that all
3-cycles are conjugate in A,,). If these cycles are not disjoint, then each fixes
all the symbols outside of (1,2, 3, i, j j , say, and the two 3-cycles lie in A*, the
group of all even permutations on these 5 symbols. Of course, A* z A,, and,
as in part (i) of the previous proof, (1 2 3) and (i j k) are conjugate in A*; a
fortiori, they are conjugate in A,,. If the cycles are disjoint, then we have just
seen that (1 2 3) is conjugate to (3 j k) and that (3 j Iz) is conjugate to
(i j Ic), so that (1 2 3) is conjugate to (i j k ) in this case as well.
A normal subgroup H containing a 3-cycle a must contain every conjugate
of a; as all 3-cycles are conjugate, H contains every 3-cycle. But Exercise 2.7
shows that A,, is generated by the 3-cycles, and so H = A,. EB
Some Representation Theorems
Lemma 3.86). A, is simple.
Proo$ Let H # 1 be a normal subgroup of A,, and let a E H be distinct from
1. If a fixes some i, define
F = (/?E A,: P(i) = i).
Now F A, and a E H n F. But H n F Q F, by the second isomorphism
theorem, so that F simple and H n F # 1 give H n F = F; that is, F s H.
Therefore, H contains a 3-cycle, H = A, (by the lemma), and we are done.
We may now assume that no a E H with a # 1 fixes any i, for 1 i 2 6. A
glance at Table 3.3 shows that the cycle structure of a is either (1 2)(3 4 5 6)
or (1 2 3)(4 5 6). In the first case, a2 E H, a2 # 1, and a2 fixes 1 (and 2), a
contradiction. In the second case, H contains a(pa-'p-'), where P = (2 3 4),
and it is easily checked that this element is not the identity and it fixes 1, a
contradiction. Therefore, no such normal subgroup H can exist.
Theorem 3.111. A, is simple for all n 2 5.
Proo/. Let n 2 5 and let H # 1 be a normal subgroup of A,. If /3 E H and
P # 1, then there is an i with P(i) =j # i. If a is a 3-cycle fixing i and moving
j, then a and p do not commute: Pa(i) = P(i) = j and a/3(i) = a ( j ) # j; there-
fore, their commutator is not the identity. Furthermore, a(/?a-'P-') lies in the
normal subgroup H, and, by Lemma 3.4, it is a product of two 3-cycles
(apc?-')P-'; thus it moves at most 6 symbols, say, i,, ..., i,. If F = ( y E A,:
y fixes the other symbols), then F 2 A, and spa-'/?-' E H n F Q F. Since A,
is simple, H n F = F and F 5 H. Therefore H contains a 3-cycle, N = A, (by
Lemma 3.9), and the proof is complete.
3.20. Show that A,, a group of order 60, has no subgroup of order 30.
3.21. If n # 4, prove that An is the only proper nontrivial normal subgroup of S,,.
3.22. If G 5 S. contains an odd permutation, then IGI is even and exactly half the
elements of G are odd permutations.
3.23. If X = {l,2, ...} is the set of all positive integers, then the infinite alderntatilsg
group A, is the subgroup of S, generated by all the 3-cycles. Prove that A, is
an infinite simple group. (Hint. Adapt the proof of Theorem 3.11.)
Some Representation Theorems
A valuable technique in studying a group is to represent it in terms of some-
thing familiar and concrete. After all, an abstract group is a cloud; it is a
capital letter G. If the elements of G happen to be permutations or matrices,
however, we may be able to obtain results by using this extra information. In
52 3. Symmetric Groups and G-Sets
this section we give some elementary theorems on representations; that is, on
homomorphisms into familiar groups.
The first such theorem was proved by Cayley; it shows that the study of
subgroups of symmetric groups is no less general than the study of all groups.
Theorem 3.12 (Cayby, 1878). Every group G can be imbedded as a subgroup of
SG.In particular, if I GI = n, then G can be imbedded in S,,.
Pro@$ Recall Exercise 1.33: for each a E G, left translation La: G -, G, defined
by x I-, ax, is a bijection; that is, La E S,. The theorem is proved if the func-
tion L: G -,S,, given by a t-, La, is an injection and a homomorphism, for
then G r im L. If a # by then La(l) = a f b = Lb(l),and so La # Lb.Finally,
we show that Lab= La 0 Lb. If x E G, then Lab(x)= (ab)x, while (La o L,) (x) =
L,(L,(x)) = La(bx) = a(bx); associativity shows that these are the same.
Definition. The homomorphism L: G -,S,, given by a t-, La, is called the bfi
regafar represent~tionof G.
The reason for this name is that each La is a regular permutation, as we
shall see in Exercise 3.29 below.
Corollary 3.13. If k is a field and G is a finite group of order n, then G can be
imbedded in GL(n, k).
Proop4: The group P(n, k) of all n x n permutation matrices is a subgroup of
CL(n, k) that is isomorphic to St, (see Exercise 1.45). Now apply Cayley's
theorem to imbed G into P(n, k).
The left regular representation gives another way to view associativity.
Assume that G is a set equipped with an operation * such that there is an
identity e (that is, e * a = a = a * e for all a E G) and each element a E G has a
(two-sided) inverse a' (i.e., a *a' = e = a' * a). Then, for each a E G, the func-
tion La: G --+ G, defined by La(x) = a * x, is a permutation of G with inverse
La,.If * is associative, then 6 is a group and Cayley's theorem shows that the
function L: 6; -,S,, defined by a I-+ La, is a homomorphism; hence, im L is a
subgroup of S,. Conversely, if im L is a subgroup of S,, then * is associative.
For if La o L, E im L, there is c E G with La 0 L, = LC.Thus, La o L,(x) =
Lc(x) for all x E G; that is, a * (b * x) = c * x for all x E G. But if x = e, then
a * b = c, and so c * x = (a * b) * x. This observation can be used as follows.
Assume that G = (x,, ...,x , ) is a set equipped with an operation *, and
assume that its multiplication table [aij] is a Latin square, where aij = xi * xi.
Each row of the table is a permutation of G, and so * is associative (and G is
a group) if the composite of every two rows of the table is again a row of the
table. This test for associativity, however, is roughly as complicated as that in
Exercise 1.42; both require about n3 computations.
Some Representation Theorems
We now generalize Cayley's theorem.
Theorem 3.114. If H < G and [G : H I = n, then there is a homomorphism
p: G -+ S,, with ker p (: H.
ProoJ If a E G and X is the family of all the left cosets of H in G, define a
function pa: X -+ X by gH t-+ agH for all g E G. It is easy to check that each
pa is a permutation of X (its inverse is pa-,) and that a I-, pa is a homomor-
phism p: G --+ S, z S,,. If a E ker p, then agH = gH for all g E G; in particular,
aH = H , and so a E H; therefore, ker p I H. It4
Defiaitioaa. The homomorphism p in Theorem 3.14 is called the re~resmta-
tion of G on the cosets of H.
When H = 1, Theorem 3.14 specializes to Cayley's theorem.
Coroiil2ry 3.15. A simple group G which contains a subgroup H of index n can
be imbedded in S,,.
Prot.$ There is a homomorphism p: G -,St,with lcer p < H < G. Since G is
simple, lcer p = 1, and so p is an injection.
Corollary 3.16. An infinite simple grozrp G has no proper subgroups of finite
index.
Corollary 3.15 provides a substantial improvement over Cayley's theorem,
at least for simple groups. For example, if G z As, then Cayley's theorem
imbeds G in S,,. But G has a subgroup H r A, of order 12 and index
60112 = 5, and so Corollary 3.15 says that G can be imbedded in S,.
Theorrem 3.27. L.et H 5 and let X be the family of all the conjugates of H in
6. There is a homomorphism 111: G -+ S, with lcer $ 5 NG(H).
Proop$.If a E G, define $a: X -+ X by $ a ( g ~ g - l =
) a g ~ g - ' a - l . If b E G, then
$at//,(gHg-l) = tja(bgNg-'b-') = abg~g-lb-'a-l - $a,(g&?-l).
We conclude that t,ba has inverse II/,-~,so that $a E S, and $: G -+ S, is a
homomorphism.
If a E ker t,h, then agPig-la-l = gHg-' for all g E G. In particular, aHa-' =
H, and so a E NG(II);hence lcer $ 5 NG(H). !El
Defiwitio~.The homomorphism 11of Theorem 3.17 is called the represeafa-
$ion of d;: on the conjd~gatesof H.
54 3. Symmetric Groups and G-Set;
3.24. Let a E G, where G is finite. If a n has m conjugates and a has k conjugates, then
mlk. (Hint. CG(a)5 CG(an).)
3.25. Show that S,, has no subgroup of index t for 2 < t < rz.
3.26. (i) If p is the representation of a group G on the cosets of a subgroup H, then
ker p = xHx-'. Conclude that if H a G, then ker p = H.
(ii) If J, is the representation of a group G on the conjugates of a subgroup H,
then ker $ = n,, XN,(H)X-'.
3.27. The right reglrlar representation of a group G is the function R: G -+ SGdefined
by a H R,, where Ra(x) = xa-'.
(i) Show that R is an injective homomorphism. (Hint. See Exercises 1.33 and
1.47.) (This exercise is the reason why R, is defined as right multiplication by
a-' and not by a.)
(ii) If L and R are, respectively, the left and right regular representations of S,,
prove that im L and im R are conjugate subgroups of S,.
3.28. If p is prime and T,a E S, are a transposition and a p-cycle, respectively, show
that S, = (T, a). (See Exercise 2.9(iii).)
3.29. If G is a finite group and a E G, then La is a regular permutation of G. (Hint. If
La = p, .. .Pi is the complete factorization of La and if g is a symbol occurring
in some pi, then the set of all symbols in Pi is the right coset (a)g.)
3.30. (i) Let G be a group of order 2"'k, where k is odd. Prove that if G contains an
element of order 21n, then the set of all elements of odd order in G is a
(normal) subgroup of G. (Hint. Consider G as permutations via Cayley's
theorem, and show that it contains an odd permutation.)
(ii) Show that a finite simple group of even order must have order divisible
by 4.
3.31 (i) (Poinearb). If H and K are subgroups of G having finite index, then H n K
also has finite index in G. (Hint. Show that [G: N n K ] < [G : HI [G : K].)
(ii) If H has finite index in G, then the intersection of all the conjugates of H is
a normal subgroup of G having finite index.
(iii) If ([G : HI, [G : I<]) = 1, then [G: H n I<] = [G :HI [G :K ] .
3.32. Prove that A, has no subgroup of prime index.
3.33. Let G be a finite group containing a subgroup H of index p, where p is the
smallest prime divisor of I GI. Prove that H is a normal subgroup of G.
3.34. Let G be an infinite simple group.
(i) Every x E G with x # 1 has infinitely many conjugates.
(ii) Every proper subgroup H # 1 has infinitely many conjugates.
3.35 (Eilenberg-Moore) (i) If H < G, then there exists a group L and distinct homo-
morphisms f, g: G -+ L with f 1 H # glH. (Hiat. Let L = s,, where X denotes the
family of all the left cosets of H in G together with an additional element oo. If
a E G, define fa E S, by f,(m) = m and f,(bH) = abH; define g: G -+ S, by g =
y of, where y: S, -+ S, is conjugation by the transposition which interchanges
H and oo.)
(ii) If A and G are groups, then a homomorphism h: A -+ G is a surjection if and
only if it is right cancellable: for every group L and every pair of homomor-
phisms I, g: G -r L, the equation f o h = g o h implies f = g.
The elements of symmetric groups are functions; here is the appropriate
abstraction of this property.
Definition. If X is a set and G is a group, then X is a 6-set if there is a function
a: G x X -+ X (called an action),' denoted by a: (g, x) r gx, such that:
(i) l x = x for all x E X; and
(ii) g(hx) = (gh)x for all g, h E G and x E X.
One also says that G acts on X. If 1x1= n, then n is called the degree of the
G-set X.
For example, if G 5 Sx, then a: G x X -t X is evaluation: a(o, x) = (~(x);
using the notation of the definition, one often writes ox instead of o(x).
The first result is that G-sets are just another way of looking at permuta-
tion representations.
Theorem 3.18. If X is a G-set with action a, then there is a homomorphism
B: G -+ Sx given by E(g): x c gx = a(g, x). Conversely, every homomorphism
rp: G -+ Sx defines an action, namely, gx = q(g)x, which makes X into a G-set.
Proof. If X is a G-set, g E G, and x E X, then
E(g-l)E(g): x r E(g-')(gx) = g-'(gx) = (g-'g)x = lx = x;
it follows that each B(g) is a permutation of X with inverse E(g-I). That B is
a homomorphism is immediate from (ii) of the definition of G-set. The con-
verse is also routine. I3
The first mathematicians who studied group-theoretic problems, e.g.,
Lagrange, were concerned with the question: What happens to the poly-
nomial g(x,, ..., xl,) if one permutes the variables? More precisely, if o E S,,,
In this definition, the elements of G act on the left. There is a "right" version of G-set that is
sometimes convenient. Define a right action a': G x X --r X, denoted by (g, x) t-+ xg, to be a
function such that:
(i) xl = x for all x E X; and
(ii) x(gh) = (xg)h for all g, h E G and x E X.
It is easy to see that every right G-set gives rise to a (left) G-set if one defines a: G x X + X by
a(g, x) = xg-' = ayg-', x).
56 3. Symmetric Groups and G-Sets

define B(g) is the set of distinct polynomials of the form go, and G, = Y(g) =
gU(xl,.. -,xn)= g(xol, . . - ,xu,); (o E G = S,,: go = 8).
Given g, Lagrange defined
given g, how many distinct polynomials go are there? If go = g for all o E S,,,
then g is called a symmetric function. If a polynomial f (x) =
roots r,, . .., r,, then each of the coefficients ai of f(x) = a,
zy=,
n~=, nixi has
(x - ri) is a
g*(x1, Y XU) = n
oE S,
(x-s"(x1, ..- xn));
Y

symmetric function of r,, . . . , I;,. Other interesting functions of the roots may
not be symmetric. For example, the discriminant of f(x) is defined to be the
number d2, where d = ni<j +
nix,
(ri - r,). If D(xl, . . . , x,) = (xi - xj), then it
is easy to see, for every o E St,, that Do = D (for all i < j, either xi - xj or
xj - xi = -(xi - xj) occurs as a factor of Do). Indeed, D is an anernating
function of the roots: Do = D if and only if o E An. This suggests a slight
change in viewpoint. Given g(x,, . . . , x,), find
this is precisely what Lagrange did (see Examples 3.3 and 3.3' below). It is
easy to see that Y(g) 5 S,; moreover, g is symmetric if and only if 9 ( g ) = Sn,
while Y(D) = An.Modern mathematicians are concerned with this same type
of problem. If X is a G-set, then the set of all f : X + X such that f (ox) = f(x)
for all x E X and all a E G is usually valuable in analyzing X.
EXAMPLE3.3. If k is a field, then Sn acts on k[x,,. .., x,] by crg = gaywhere
ga(xl, x,,) = g(~o1,..., xan).
EXAMPLE
3.4. Every group G acts on itself by conjugation.
EXAMPLE 3.5. Every group G acts on the family of all its subgroups by conju-
gation.
There are two fundamental aspects of a G-set.
he then defined the resolvent A(g) of g to be the polynomial obtained from g*
by removing redundant factors. If r is the degree of A(g), then Lagrange
claimed that r = n!/(Y(g)((Abbati (1803) proved this claim). This formula is
the reason Lagrange's theorem is so-called; Lagrange's theorem for sub-
groups of arbitrary finite groups was probably first proved by Galois.
EXAMPLE 3.4'. If G acts on itself by conjugation and x E G, then B(x) is the
conjugacy class of x and G, = CG(x).
EXAMPLE 3.5'. If G acts by conjugation on the family of all its subgroups and
it H IG, then B(H) = {all the conjugates of H} and G, = NG(H).
Theorem 3.19. If X is a G-set and x E X, then
Proof. If x E X, let GIG, denote the family of all left cosets of G, in G. Define
f : B(x) -+GIG, by f(ax) = aG,. Now f is well defined: if ax = bx for some
b E G, then b-'ax = x, b-'a E G,, and aG, = bG,. The function f is an injec-
tion: if aG, = f(ax) = f(cx) = cG, for some c E G, then c-'a E G,, c-'ax = x,
and ax = cx; the function f is a surjection: if a E G, then aG, = f(ax). There-
fore, f is a bijection and lB(x)l = IG/G,l = [G : G,]. R!
Corollary 3.20. If a finite group G acts on a set X, then the number of elements
in any orbit is a divisor of I GI.

Definition. If X is a G-set and x E X, then the G-orbit of x is

(i) If G is a finite group and x E G, then the number of conjugates of x in G is
[G : C,(x)l.
(ii) If G is a finite group and H 5 G, then the number of conjugates of H in G
One often denotes the orbit 6(x) by Gx. Usually, we will say orbit instead is [G : NG(H)].
of G-orbit. The orbits of X form a partition; indeed, the relation x = y de-
fined by "y = gx for some g E G" is an equivalence relation whose equiva- fiooJ Use Examples 3.4' and 3.5'. El
lence classes are the orbits.
We have now explained the similarity of the proofs of Theorems 3.2 and
3.3.
DeGnition. If X is a G-set and x E X, then the stabilizer of x, denoted by G,,
is the subgroup
G, = {g E G: gx = x) < G.

Let us see the orbits and stabilizers in the G-sets above.

3.36. If D(x,, ..., x,,) = ni
<, (xi - x,), prove that Y(D) = {o E S,,: Du = Dj = An.
3.37. Let X be a G-set, let x, y E X, and let y = gx for some g 6 G. Prove that G, =
EXAMPLE
3.3'. Let X = k[xl, ..., x,,] and G = St,. If g E k[x,, ..., x,,], then LJG,~-'; conclude that I G,I = I G,I.
58 3. Symmetric Croups and G-Sets Counting Orbits

3.38 (Abbati). If k is a field, g E k[x,,. .., x,], and o E S,, write ga(xl, ..., x,) = of G-orbits of X, then
g(xal, .. ., x,,), as in Example 3.3. Show that, for any given g, the number of N = (1IIGI) C F(4,-
distinct polynomials of the form go is a divisor of n!. (Hint. Theorem 3.19.) tcG

where, for z E G, F(z) is the number of x E X fixed by z.

3.39. If G IS,, then G acts on X = (1, ..., n). In particular, (a) acts on X for every
a E S,. If the complete factorization of a into disjoint cycles is a = P I . . .P, and if
i is a symbol appearing in /Ij, then B(i) = (cxk(i):k E I!) consists of all the symbols
appearing in /Ij. (Compare Exercise 3.29.)
3.40. Cayley's theorem shows that every group G acts on itself via left translations.
Show that there is just one orbit (if x E G, then G = (gx: g E G)) and that G, = 1
for every x E G.
Proof. In the sum EreF(i), each x E X is counted I Gxl times (for G, consists
of all those z E G which fix x). If x and y lie in the same orbit, then Exercise
3.37 gives (G,I = (G,I, and so the [G : G,] elements constituting the orbit of
x are, in the above sum, collectively counted [G : G,] I G,I = IGI times. Each
orbit thus contributes /GI to the sum, and so zteG
F(r) = NIGI.

Definition. A G-set X is tvmsiti~eif it has only one orbit; that is, for every
Corollary 3.23. If X is a finite transitive G-set with 1x1> 1, then there exists
z E G having no fixed points.
x, y E X, there exists o E G with y = ox.
Proof. Since X is transitive, the number N of orbits of X is 1, and so Burnside's
3.41. If X is a G-set, then each of its orbits is a transitive G-set.
lemma gives
3.42. If H 5 G, then G acts transitively on the set of all left cosets of H (Theorem 3.14)
and G acts transitively on the set of all conjugates of H (Theorem 3.17).
Now F(1) = 1x1> 1; if F(z) > 0 for every z E G, then the right hand side is
3.43. (i) If X = {x,, ..., x,) is a transitive G-set and H = GXt,then there are elements
g l , ..., g, in G with gixl = xi such that glH, ..., g,,H are the distinct left
too large. El
cosets of H in G.
(ii) The stabilizer H acts on X, and the number of II-orbits of X is the number Burnside's lemma is quite useful in solving certain combinatorial problems.
of (H-H)-double cosets in G. Given q distinct colors, how many striped flags are there having n stripes (of
equal width)? Clearly the two flags below are the same Gust turn over the top
3.44. Let X be a G-set with action a: G x X -+ X, and let 6: G -+S, send g E G into
flag and put its right end at the left).
the permutation x t-, gx.
(i) If K = ker h, then X is a (G/IC)-set if one defines
(gK)x = gx.

(ii) If X is a transitive G-set, then X is a transitive (G/K)-set.

(iii) If X is a transitive G-set, then llter 21 I IGI/IXI. (Hint. If x E X, then
(O(x)l = [G : G,] < [G :lter cx"].)
1 2 ... n - 1 n
Let z E S,, be the permutation
n n - 1 ... 2 1
. If %'" is the set of all n-
Counting Orbits tuples c = (c,, .. ., c,,), where each ci is any of the q colors, then the cyclic group
G = (z) acts on Wzif we define zc = z(c,, . . . , c,,) = (c,,, . . . , c,). Since both c
Let us call a G-set XJinite if both X and G are finite. and zc give the same flag, a flag corresponds to a G-orbit, and so the number
of flags is the number N of orbits. By Burnside's lemma, it suffices to compute
F(l) and F(z). Now F(1) = IF1/ = q". An n-tuple (cl , . .., c,) is fixed by z if
Theorem 3.222 (Burmside's Lemma2). If X is a finite G-set and N is the number
and only if it is a "palindrome": c, = c,; c2 = c,,-,; etc. If n = 2k, then z =
What is nowadays called Burnside's lemma was proved by Frobenius (1887), as Burnside (1 n)(2 n - I)...(]< k + 1 ) ; i f n = 2 k + l , t h e n z = ( l n)(2 n - 1)...(k k+2).
himself wrote in the first edition (1897) of his book. This is another example (we have already It follows that F(z) = qr("+1)121, +
where [(n 1)/2] denotes the greatest integer
mentioned Lagrange's theorem) of a name of a theorem that is only a name; usually "Smith's in (n + 1)/2. The number of flags is thus
theorem" was discovered by Smith, but this is not always the case. It is futile to try to set things
right, however, for trying to change common usage would be as successful as spelling reform.
 I

3. Symmetric Groups and G-Sets Counting Orbits

60

other hand,
Let us make the notion of coloring more precise.

Definition. If G t; S,, where X = (1, .. ., n), and if V is a set of colors, then Vn

is a G-set if we define ~ ( c , ., ..,c.) = (c,,, ...,ern) for all r E G. If J ' &=
l q, then
an orbit of Vn is called a (q, GI-coloring of X .
and so
Lemma 3.24. Let V be a set of q colors, and let G 5 S, g S,. If r E G, then
F(z) = qt('), where t(r) is the number of cycles occurring in the complete factori-
zation of z.

Proof. Since ~ ( c , .,.., c,,) = (c,,, ...,c,) = (c,, ..., cn),we see that cri = ci for
all i, and so zi has the same color as i. It follows that zxi has the same color
as i, for all k; that is, all i in the (7)-orbit of X have the same color. But
Exercise 3.39 shows that if the complete factorization of r is z = . . .Ptcrl,
and if i occurs in pj, then the set of symbols occurring in Pj is the (r)-orbit
containing i. Since there are t(r) orbits and q colors, there are qt(') n-tuples
fixed by r in its action on V". fi4
Definition. If the complete factorization of r E S,, has e,(r) 2 0 r-cycles, then
the index of z is
ind(z) = E'(')~:(') ...x2(r).
If G 5 S,, then the cycle i n k of G is the polynomial
P,(x,,. . ., x.) = (l/IGI) C ind(4 E QCx,, . - -%,I.
reG
,
In 1937, PolyB pushed this technique further. Burnside's lemma allows one
to compute the number of blue and white flags having nine stripes; there are
264 of them. How many of these flags have four blue stripes and five white
stripes?
Theorem (Polyii, 193737).Let G 5 S,, where 1x1= n, let 1V1 = q, and, for each
i 2 1, define oi = c: + +
- - - cd. Then the number of (q, G)-colorings of X
with f, elements of color c,, for every r, is the coefficient of c:lc?. ..c$ in
pG(g1, ..- ,@I,).
The proof of Polyh's theorem can be found in combinatorics books (e.g.,
see Biggs (1989), Discrete Mathematics). Let us solve the flag problem posed
above; we seek the coefficient of b4w5in

A short exercise with the binomial theorem shows that the coefficient of b4w5
For example, let us consider all possible blue and white flags having nine is 66.
stripes. Here 1x1= 9 and G = (z) t; Sg, where i = (1 9)(2 8)(3 7)(4 6) (5).
Now, ind(1) = x:, ind(r) = x,x$ and the cycle index of G = (z) = {I, z) is

3.45. If G is a finite group and c is the number of conjugacy classes in G, then

Corollary 3.25. If 1x1= n and G 5 S,,, then the number of (q, G)-colorings of
X is PG(q,...,q).
3.46. (i) Let p be a prime and let X be a finite G-set, where IG( = pn and 1 4is not
divisible by p. Prove that there exists x E X with zx = x for all z E G.
ProoJ By Burnside's lemma for the G-set V, the number of (q, G)-colorings (ii) Let V be a d-dimensional vector space over ifp, and let G 2 GL(d, iZ,) have
of X is order p". Prove that there is a nonzero vector v E V with zv = v for all z E G.
(l/lGl) F(7). C
raG 3.47. If there are q colors available, prove that there are

By Lemma 3.24, this number is l(q112 + 2qI(nZ+3)/41 + q I ( ~ ~+1)/21

2
1
distinct rz x n colored chessboards. (Hint. The set X consists of all n x n arrays,
and the group G is a cyclic group {z), where z is a rotation by 90".Show that z
is product of disjoint 4-cycles.)
where t(r) is the number of cycles in the complete factorization of 7. On the
 i
I
Some Geometry
3. Symmetric Groups and G-Sets

Some Geometry
The familiar euclidean n-space is the vector space R" consisting of all n-tuples
of real numbers together with an inner (or dot) product. If gi is the vector
having ith coordinate 1 and all other coordinates 0, then the startdard basis is
(E,, . . .,E.). A vector x = (x,, . . . ,x,) has the unique expression x =
and if y = (y,, . .., y,,), then the h e r p r o h c t (x, y) is defined to be the num-
xi
ber xi xiyi.
A subset {u,, . . . , u,,) of R" is called an orthonormal basis if (ui, uj) = dij;
that is, (ui, ui) = 1 for all i and (u,, uj) = 0 when i # j (it is easy to see that an
orthonormal basis is a basis, for it is a linearly independent subset of R"
having n elements). The standard basis is an orthonormal basis. If {ui, .. .,un)
is an orthonormal basis and if x = xi
xiui, then (x, x) = ( x i xiue zj
xjuj) =
Figure 3.1
xi, xixj(ui,uj) = xi
x:. If x = (x, . .. x,!) E R", define llxll =
I I

~ l x l=J ~(x, x)), and define the distance between x and y to be /Ix - yll.
Jm(thus,

Definition. A motiosr is a distance-preserving function T: R" --+ W";that is,

3.48. If there are q colors available, prove that there are
x
(114 din cp(n/d)qd
colored roulette wheels having n congruent compartments, each a circular sec-
tor (in the formula, cp is the Euler q-function and the summation ranges over all
divisors d of n with 1 < d < n). (Hint. The group G = (i) acts on n-tuples,
where i(c,, c,, ..., en) = (c,,,c,, c,, ..., cap,).Use Corollary 3.25 and Theorem
2.15.)
11 Tx - Tyll = Ilx - yll for all x, y E Rn.
It is plain that if w E R", then the function q,,:
R" --+ Rn,defined by T,(x) =
x + w for all x E R", is a motion (T, is called t~anslaiionby w). Of course,
I;,(O) = w, so that T,, is not a linear transformation if w # 0.
Definition. A linear transformation S: Rn-+R" is orlhogonal if 11 Sxll = llx 11
for all x E R".

Lemma 3.26.
(i) A linear transformation S: Rn-t R" is orthogonal iJ and only if (SE,, ...,
SE.) is an orthonormal basis (where (E,, . . .,E,) is the standard basis).
(ii) Every orthogonal transformation S is a motion.

ProoJ (i) Assume that S is orthogonal. If x = zixi&,and y xiyiei, then

=

Ilx + Y 11 - Ilxll - ll Y 112 = 2 Ci-' XiYi = 2(x, Y).

In particular,
lISx + Syl12 - IlSxll2 - IISylI2 = 2(Sx,Sy).
Since jlx + yl/ = /IS(x + y)ll = llSx + Syll 2, we have (Sx, Sy) = (x, y) for all
x, y. In particular,
aij = ( E ~E, ~=) (SE~,SE~),
SO that (SE,, .. ., SE,,)is an orthonormal basis.
Figure 3.2
64 3. Symmetric Groups and G-Sets Some Geometry

Conversely, if x = xixisi,then Sx xixis&,,

= if it preserves inner products:

llSxll = (SX,SX)=
i, j
xixj(Ssi, Ss,) = x xi
I
= \\x/(~, (Tx, Ty) = (x, y) for all x, y E R".

and S is orthogonal.
(ii) If x = xi xisi and y = zi
yisi, then /iSx - Syl/ = //S(x- y)lj2 =
Ilx + y1I2 = tx + YYx + Y)= llxl12 + 2(x, Y)+ llyIl2.
IICi (xi - yi)S&iI12= Ci (xi - yi12 = IIx - .YI12. Similarly, since T is linear (by Lemma 3.27),
'
Notice that every orthogonal transformation (as is every motion) is an
injection, for if x # y, then 0 # llx - yll = l\Sx - Syll and Sx # Sy; since S is
a linear transformation on Rn with nullity 0, it follows that S is invertible. It
is easy to see that if S is orthogonal, then so is S-'.

Lemma 3.27. Every motion S: R" -+ Rnfixing the origin is a linear transforma-
tion, hence is orthogonal.
Proof. We begin by showing that a motion T fixing 0 and each of the elements
in the standard basis must be the identity. If x = (x,, . . .,x,,), denote Tx by
(y,, ..., y.). Since TO = 0, IITxl/ = (ITx - TO11 = llx - 011 = /Ixli gives
By hypothesis, 111: + yIl2 = IlT(x + y)l12, Ilxll2 = llTxl12, and Ilyll2 = I l T ~ l l ~ ,
so that 2(x, y) = 2(Tx, Ty) and (x, y) = (Tx, Ty).
Conversely, if T preserves inner products, then
/Ix - yIl2 = (X - y, x - y) = (Tx - Ty, Tx - Ty) = IITx - Ty/12
for all x,y E R". Therefore, T preserves distance, hence is a motion. M

The geometric interpretation of this theorem is that every motion fixing

But also 11 Tx - E , ~ I = 11 Tx - TE,11 = (Ix - E, 11 gives the origin preserves angles, for (x, y) = JIxJJ Jl yJl cos 0, where 0 is the angle
(y, - + Y; + . .. + Y; = (x, - + x; + .. + x;. between x and y. Of course, all motions preserve lines and planes; they are,
after all, linear. For example, given a line L' = {y + rx: r E R) (where x and y
Subtracting gives 2y, - 1 = 2x, - 1, and y, = x,. A similar argument gives are fixed vectors) and a motion T,,S (where T, is translation by w and S is
y, = xi for all i, so that Tx = x. + +
orthogonal), then T,S(/) = ( Tv(Sy rSx): r E R) = {(w Sy) rSx: r E R) +
Assume now that Tci = u ifor i = 1, . .., n, and let S: R" 4 R" be the linear is also a line.
transformation with SE,= ui for all i. NOWTS-I is a motion (being the com-
posite of t y o motions) that fixes the standard basis and 0, and so T = S. L3 Defiaitiom. A matrix A E GL(n, W) is csrtbaogonal if AAt = E, where A' denotes
\ the transpose of A.
Theorem 3.28.
(i) The set O(n, W) of all motions S: 08" -+ R" fixing the origin is a Denote the ith row of A by a,. Since the i, j entry of AAt is (a,, a,), it follows
CL(n, R) (culled the real ort~~ogonwlgro~ap). that (a,, . . . , a,) is.an orthonormal basis of Rn.If T is an orthogonal transfor-
(ii) Every motion T: R" 4 IWn is the composite of a trnnslation and an orbhogonal mation with T E= ~ ui for all i, then the matrix of T relative to the standard
transformation, and the set M(n, R) of all motions is n group (called the real basis is an orthogonal matrix. It follows that O(n, W) is isomorphic to the
9 ~ 0 a pof motions). multiplicative group of all n x n orthogonal matrices.
Since det At = det A, it follows that if A is orthogonal, then (det A ) = ~ I,
Pros$ (i) Routine, using the lemma. and so det A = $- 1.
(ii) Let T be a motion, and let T(0) = w. If S is translation by - w (i.e.,
Sx = x - w for all x), then S T is a motion fixing 0, hence is orthogonal, hence Definition. A motion T fixing the origin is called a rotation (or is orkntation-
is a bijection; therefore, T = Y1(ST) is a bijection. The reader may now preseroi~zg)if det T = 1. The set of all rotations form a subgroup SO(n, R) I
show that the inverse of a motion is a motion, and that the composite of O(n, W), called the rotation group. A motion fixing the origin is called ovienta-
motions is a motion. B l tion-reversing if det T = - 1.

Theorem 3.29. A function T: R" -+ Rnfixing the origin is n motion if and only Of course, [O(n, R) : SO(n, R)] = 2.
66 3. Symmetric Groups and G-Sets Some Geometry 67

Here are some examples of orientation-reversing motions. It is a standard Let us now consider the case n = 2. If we identify IW2 with the complex
result of linear algebra that if W is any subspace of R" and WL = {v E V: numbers C,then perpendicular unit vectors u, and u, have the form u, = eiB
(v, w) = 0 for all w E W), then dim W' = n - dim W. A liypevplane H in iW" is and u, = eim,where 9 = 8 & n/2. It follows that if
+
a translate of a subspace W of dimension n - 1: H = W v, for some vector
v,. If H is a hyperplane through the origin (that is, H = W is a subspace of
dimension n - I), then dim HL = 1, and so there is a nonzero vector a with
(a, h) = 0 for all h E H; multiplying by a scalar if necessary, we may assume is an orthogonal matrix, then its columns are the real and imaginary parts of
that a is a unit vector. Acl and A&,. Therefore, either
If I is a line in the plane, then the reflection in ( is the motion p: R2 -+ R2
which fixes every point on I and which interchanges all points x and x' cos 8 -sin(@ + n/2) cos 8 -sin 8
equidistant from L (as illustrated in Figure 3.3; thus, I behaves as a mirror). A = [ sin 8 cos(0 + n/2J = [sin H cos 0
More generally, define the vrflection in a hyperplane H as the motion that and det A = 1 (so that A corresponds to rotation about the origin by the
fixes every point of H and that interchanges points equidistant from H. If p angle 8) or
is to be a linear transformation, then H must be a line through the origin, for cos 8 -sin(@ - n/2) cos 0 sin O
the only points fixed by p lie on H.
A = [ sin 8 cos(8 - n/2J = [sin o c o s o
and det A = - 1 (so that A corresponds to reflection in the line I through the
origin having slope tan 0). In particular, the matrix

corresponds to the motion (x, y) r-, (x, -y) which is the reflection in the
x-axis.
With this background, we now pose the following problem. Let A be a
figure in the plane having its center of gravity at the origin. Define
Y(A) = {S E O(2, a):S(A) = A).
Of course, S(A) = {x E IW2: x = S(y) for some y E A). If A is a triangle with
Figure 3.3 vertices a, b, and c and if S is a motion, then S(A) is also a triangle, say, with
vertices Sa, Sb, and Sc; if S E Y(A),then S permutes X = (a, byc). It follows
\ \ %

that Y(A) acts on X: there is a homomorp1~ism$: Y(A) -+ S,, namely, S H

Theorem 3.30. Every reflection p in a hyperplane H through the origin is SIX, the restriction of S to X. Now is an injection, for a linear transforma-
tion on R2is determined by its values on an independent set of two vectors. It
orientation-reversing.
follows that Y(A) is a finite group; indeed, Y(A) is isomorphic to a subgroup
of S,. If A is an equilateral triangle, then Y(A) E S, (see Exercise 3.58 below);
Proof. Choose a unit vector a E Rn with (h, a) = 0 for all h E H. Define
if A is only an isosceles triangle, then Y(A) z H,; if A is not even isosceles,
p': R" -+ W" by pl(x) = x - 2(x, a)a for all x E IWn. If x E H , then (x, a) = 0,
then Y(A) = 1. The group 9(A) thus "measures" the amount of symmetry
x - 2(x, a)a = x, and p' fixes x; if x # H, then x = h + ra, where h E H and
present in A: bigger groups arise from "more symmetric" triangles. A circle A
r E R. Now (x, a) = (h + m, a) = r and x - 2(x, a)a = h - m; hence,
with center at the origin is very symmetric, for Y(A) is an infinite group (for
pl(h + ra) = h - ra, so that p' interchanges pairs of vectors equidistant from
every 8, it contains rotation about the origin by the angle 8). One calls Y(A)
H and fixes H pointwise. Hence, p' = p.
the sy~nmletryg r s q of the figure A.
If (h,, . .., h,,-,) is a basis of H, then {h,, . . . , h,,-,, a) is a basis of Rn.
Relative to the latter basis, the matrix of p is diagonal with diagonal entries
1, 1, . .. , 1, - 1; therefore, det p = - 1 and p is orientation-reversing. B3 Theorem 3.31. If A is a regular polygon with n vertices, then Y(A) is a group of
68 3. Symmetric Groups and G-Sets
order 2n which is generated by two elements S and T such that
Sn=l, ~ ~ = 1 and
, TST=S-'.
Proof. We may assume that the origin is the center of gravity of A and that
one vertex of A lies on the x-axis. Observe that Y(A) is finite: as in the
example of triangles, it can be imbedded in the group of all permutatons of
the vertices. Now each S E Y(A) also permutes the edges of A; indeed, regu-
larity (that is, all edges having the same length) implies that Y(A) acts transi-
tively on the n edges. Since the stabilizer of an edge has order 2 (the endpoints
can be interchanged), Theorem 3.19 gives IY(A)I = 2n. If S is rotation by
(360/n)" and T is reflection in the x-axis, then it is easy to check that Y(A) =
(S, T) and that S and T satisfy the displayed relations. Gi
Definition. The rlihedralgrordP3D,,,, for 2n 2 4, is a group of order 2n which
is generated by two elements s and t such that
s" = 1, t2 = 1, and tst = s-'.
Note that D,, is not abelian for all n 2 3, while D, is the 4-group V.
The next result explains the ubiquity of dihedral groups.
Theorem 3.32. If G is a finite group and if a, b E G have order 2, then (a, b) z
DZllfor some n.
ProoJ Since G is finite, the element ah has finite order n, say. If s = ab, then
asa = a(nb)a = ba = (ab)-' = s-', because both a and b have order 2.
It remains to show that 1 (a, b) 1 = 2n (of course, I (a, b) 1 = 2m for some m,
but it is not obvious that nz = n). We daim that asi # 1 for all i 2 0. Other-
wise, choose i 2 0 minimal with asi = 1. Now i # 0 (for a # 1) and i
1 = as = aab = b), so that i 2 2. But 1 = asi = aabsi-' = bsi-';
by b gives 1 = si-'b = ~ ' - ~ n b=b si-,a, and conjugating by
is a subgroup. Using Corollary 2.4, one need check only four cases: asinsx =
s-islc = s ~ - i W ;sisk = si+X E H ; s'ask = a(asia)sk= s-'sk =
H ; asisk = asi+k
S k d i H.
Elements of order 2 arise often enough to merit a name; they are called
inu6~lartons.
In earlier editions, I denoted D,,, by D,,.
Some Geometry 69
Let Ll be a figure in iW3 with its center of gravity at the origin, and define
If Ll = Zn is the regular solid having n (congruent) faces, each of which has k
edges, then Y(Zn)acts transitively on the set of n faces of En(this is essentially
the definition of regularity), while the stabilizer of a face f (which is a regular
k-gon) consists of the k rotations o f f about its center. By Theorem 3.19,
Y(C,) has order nk.
It is a classical result that there are only five regular solids: the tetrahedron
Z, with 4 triangular faces; the cube Z, with 6 square faces; the octahedron Z,
with 8 triangular faces; the dodecahedron Z,, with 12 pentagonal faces; the
icosahedron Z2, with 20 triangular faces. The rotation groups of these solids
thus have orders 12,24,24, 60, and 60, respectively.
These considerations suggest investigation of the finite subgroups of the
orthogonal groups O(n, R). It can be shown that the finite subgroups of
0(2, W) are isomorphic to either D,,, or Z,, and that the finite subgroups of
0(3, W) are isomorphic to either D2,1,Z,l, A,, S,, or A,.
3.49. Prove that D, r V and D, E S,.
3.50. Prove that Dl, r S3 x Z,.
3.51. Let G be a transitive subgroup of S,.
(i) If m = [G: G n V], then m16.
(ii) If m = 6, then G = S,; if 1n = 3, then G = A,; if m = 1, then G = V;if rn = 2,
then either G r Z, or G E D,.
3.52. (i) (vcbn Dysk (11882)). Prove that 9(C,) z A, and that Y(C,) E S, E Y(C,).
(Hint. A, = (s, t), where s2 = t 3 = = 1; S4 = (s, t), where s2 = t3 =
(st), = 1.)
(ii) (Hamiltow (1856)). Prove that 9'(Cl2) E A , z Y(C2,). (Hint. A , = (s, t),
where s2 = t3 = (st)5 = 1.)
(Because of this exercise, A, is also called the tetraltedralg~o~p, S, is also called
the oct~IterBYalg~o~~p,and A, is also called the icosahedralgro~tp.)
3.53. Let Tr(n, W) denote the set of all the translations of Wn. Show that Tr(n, R) is an
abelian normal subgroup of the group of motions M(n, R), and M(n, R)/Tr(n, W)
z O(n, W).
3.54. It can be shown that every S E SO(3, W) has 1 as an eigenvalue (there is thus a
nonzero vector v with Sv = v). Using this, show that the matrix of S (relative to
a suitable basis of R3) is
0
0 sin O cos 0
3.55. Prove that the circle group T is isomorphic to SO(2, R).
 3. Symmetric Groups and G-Sets Some Geometry
70 71

be a primitive nth root of unity. Show that the matrices

3.56. Let o = e2"iJ'1 linear transformation A: R2 -+ R2 and a vector z E R2 such that, for all v E R2,

q(v) = A(v) z. +
The set of all affine maps under composition, denoted by Aff(2, R), is called
generate a subgroup of GL(2, C) isomorphic to D,.. the affime group ofthe plane.
3.57. What is the center Z(D,.)? (Hint.Every element of D has a factorization sit'.) Lemma 3.34. Let q be an affine map.
3.58. If A is an equilateral triangle in R' with its center of gravity at the origin, show (i) rp preserves all convex combinations.
that Y(A)is generated by
(ii) q preserves line segments: for all u, v, rp([u, v]) = [qu, qv].
+
(iii) The point tu (1 - t)v, for 0 I t I 1, is called the t-point of [u, v]. If z is
the t-point of [u, v], where 0 It 5 1, then qz is the t-point of [ q y qv]. In
3.59. How many bracelets are there having n beads each of which can be painted any ...
particular, preserves midpoints of line segments.
one of q colors? (Hint.Use Corollary 3.25; D,,,is the group that is acting.) (iv) q preserves triangles: if A = [u, v, w] is a triangle, then q(A) is the triangle
C(%Ju,V ~ 9Y~ 1 .
We now show how one can use groups to prove geometric theorems.
+
Proof. (i) Let rpx = Ax z, where 2 is a nonsingular linear transformation
Recall that if u, v E R2, then the line segment with endpoints u and v, de-
+
noted by [u, v], consists of all vectors tu (1 - t)v, where 0 2 t 5 1. If u, v, w and z E R2. If zitivi is a convex combination, then
are the vertices of a triangle A, then we will denote A by [u, v, w ] .

Definition. If v,, . . ., v , E R2, then a convex comhirtatiorr of v,, . . .,v,, is a linear

combination tivi, where all ti 2 0 and ti = 1. x
ILemrna 3.33. If A = [v,, v,, v,] is a triangle, then A consists of all the convex
combinations of v,, v,, v,.

Proof. Denote the set of all convex combinations of v,, v,, v, by C. We first
+ +
show that C c A. Let c = tlvi t2v2 t,v, belong to C. If t, = 1, then
+
c = v, E A. If t3 # I, then q = t,/(l - t3)v1 t2/(l - t3)v2is a convex com- (ii) Immediate from (i), for [u, v] is the set of all convex combinations of u
and v.
bination of v, and v,, hence lies on the line segment [v,, v2] c A. Finally,
c = (1 - t,)q + t3v3 E A, for it is a convex combination of q and v,, and (iii) Immediate from (i).
hence it lies on the line segment joining these two points (which is /wholly- (iv) Immediate from (i) and Lemma 3.33.
inside of A). /
For the reverse inclusion, take 6 E A. It is clear that C contains t i e perime- Lemma 3.35. Points u, v, w in R2 are collinear if and only if ( u - w, v - w) is
ter of A (sucl~points lie on line segments, which consist of convex com- a linearly dependent set.
binations of two vertices). If 6 is an interior point, then it lies on a line
segment [u, w], where u and w lie on the perimeter (indeed, it' lies on many Proof. Suppose that u, v, w lie on a line 8, and let 8 consist of all vectors of the
such segments). Thus, 6 = tu + (1 - t)w for some 0 t 5 1. Write u = +
form ry z, where r E W and y, z E R2. There are thus numbers ri with u =
t,v, + + +
t2v, t3v3 and w = s,vl s2v2+ s,v,, where ti 2 0, si 2 0 and + +
r, y z, y = r2y z, and w = r,y + z. Therefore, u - w = (r, - r,)y and
x:=, ti = 1 = x:=,
si. It suffices to show that 6 = t ( x tivi) + (1 - t ) ( x sivi)=
[ttivi+ (1 - t)si]vi is a convex combination of v,, v,, 0,. But tti + (1 - t)si2
v - w = (r, - r3)y form a linearly dependent set.
Conversely, suppose that u - w = r(v - w), where r E R. It is easily seen
0, because each of its terms is nonnegative, while +
[tti (1 - t)si] = that u, v, w all lie on the line G consisting of all vectors of the form t(v- w) + w,
+
t(X ti) + (1 - t)(Csi) = t (1 - t) = 1. where t E R. El

DeBnition. A function q : R2 -+R2 is an n j p ~ ~ e if there is a nonsingular

map Lemma 3.36. If A = [u, v, w] and A' = [u', v', w'], are triangles, then there is
72 3. Symmetric Groups and G-Sets

an affine map q with qu = u', qv = v', and qw = w'. Thus, Aff(2, R) acts CHAPTER 4
transitively on the family of all triangles in W2.
The Sylow Theorems
PvooJ The vertices u, v, w of a triangle are not collinear, so that the vectors
u - w and v - w form a linearly independent set, hence comprise a basis of
R2; similarly, (u' - w', v' - w') is also a basis of W2. There thus exists a
nonsingular linear transformation A with A(u - w) = u' - w' and A(v - w) =
v' - w'. If z = w' - R(w), then define q by

It is easy to see that rp is an affine map which carries u, v, w to u', v', w',
respectively. It follows from Lemma 3.33 that (p(A) = A'. FZ

Theorem 3.37. For every triangle A, the medians meet in a cornrnon point which
is a +-point on each of the medians. I
I

Proof. It is easy to see that the theorem is true in the special case of an
equilateral triangle. E. By Lemma 3.36, there exists an affine map q with
q ( E ) = A. By Lemma 3.34, preserves collinearity, medians and f-points.
The order of a group G has consequences for its structure. A rough rule of
thumb is that the more complicated the prime factorization of /GI, the more
The reader is invited to prove other geometric theorems in this spirit, using
complicated the group. In particular, the fewer the number of distinct prime
the (easily established) fact that affine maps preserve parallel lines as well as
factors in /GI, the more tractible it is. We now study the "local" case when
conic sections (in particular, every ellipse is of the form q(A), where A is the
only one prime divides 1 GI.
unit circle, for these are the only bounded (compact) conic sections).
F. Klein's Erla~lgenPrograsfn(1872) uses groups to classify different geome-
tries on the plane (or more general spaces). If G 2 SR2)then a property P of a Definition. If p is a prime, then a p-group is a group in which every element
figure A in W2 is an invariant of G if q(A) has property P for all a, E G. For has order a power of p.
example, invariants of the group M(2, R) of all motions include collinearity,
length, angle, and area; the corresponding geometry is the usual geometry of Corollary 4.3 below gives a simple characterization of finite p-groups.
Euclid. Invariants of Aff(2, W) include collinearity, triangles, line segments,
and t-points of line segments, parallelism, conic sections; the corresponding Lemma 4.1. If G is a finite abelian group whose order is divisible by a prime p,
geometry is called affine geometry. Other groups may give other eometries. then G contains an element of order p.
invariants include connectedness, compactness, and dimensionality;
responding geometry is called topology.
/
#
-'L
For example, if G is the group of all homeomorphisms of the plane, then
the cor-
Proof: Write IGJ= pm, where m 2 1. We proceed by induction on m after
noting that the base step is clearly true. For the inductive step, choose x E G
of order t > 1. If pit, then Exercise 2.11 shows that xCtphas order p, and the
lemma is proved. We may, therefore, assume that the order of x is not divisi-
ble by p. Since G is abelian, (x) is a normal subgroup of G, and G/(x) is an
abelian group of order /Gilt = pmlt. Since plt, we must have m/t < m an
integer. By induction, G/(x) contains an element y* of order p. But the
natural map v: G -+ G/(x) is a surjection, and so there is y E G with v(y) =
y2*.By Exercise 2.14, the order of y is a multiple of p, and we have returned to
the first case. H

74 4. The Sylow Theorems
We now remove the hypothesis that G is abelian.
Theorem 4.2 (Cauehy, 1845). If G is a finite group whose order is divisible by
a prime p, then G contains an element of order p.
Proof. Recall Theorem 3.2. If x E G, then the number of conjugates of x is
[G : C,(x)], where C,(x) is the centralizer of x in G. If x # Z(G), then its
conjugacy class has more than one element, and so / C,(x)l < I GI. If pl lC,(x)/
for such a noncentral x, we are done, by induction. Therefore, we may assume
that pylC,(x)/ for a11 noncentral x in G. Better, since IGI = [G : C,(x)] lCG(x)J,
we may assume that pl[G : C,(x)] (using Euclid's lemma, which applies be-
cause p is prime).
Partition G into its conjugacy classes and count (recall that Z(G) consists
of all the elements of G whose conjugacy class has just one element):
where one xi is selected from each conjugacy class with more than one ele-
ment. Since IG( and all [G : CG(xi)lare divisible by p, it follows that IZ(G)\ is
divisible by p. But Z(G) is abelian, and so it contains an element of order p,
by the lemma.
DePinition. Equation (*) above is called the elms eqnrafion of the finite
group G.
Here is a second proof of Cauchy's theorem, due to J.H. McICay, which
avoids the class equation. Assume that p is a prime and that G is a finite
group. Define
X = ((a,, ..., a,) E G x -.. x G: ala2 ... a, = 1).
Note that 1x1= (GI,-', for having chosen the first p - 1 coordinates arbi-
trarily, we must set a, = (ala, .. .a,-,)-'. Now X is a Z,-set, where g E Z,
acts by cyclically permuting the coordinates (since ai . . .a,a, . . .a,-, is a con-
jugate of a, a,. . . a,, the product of the permuted coordinates is also equal to-
1). By Corollary 3.21, each orbit of X has either 1 or p elements. An / orbit with
just one element is a p-tuple having all its coordinates equal, say,;ai = a for
all i; in other words, such orbits correspond to elements a E G with a p = 1.
Clearly (1, .. . , 1) is such an orbit; were this the only such orbit, then we
would have
1x1= IGIp-' = 1 + kp
for some integer k > 0; that is, IGIp-' = 1 mod p. If p divides IG/, however,
this is a contradiction, and so we conclude that G must have an element of
order p. (As A. Mann remarked to me, if I GI is not divisible by p, then we have
proved Fermat's theorem.)
i
i3 p-Groups
75
I
Corollary 4.3. A finite group G is a p-group if and only if / GI is a power of p.
/
, Proof. If /GI = pn', then Lagrange's theorem shows that G is a p-group. Con-
versely, assume that there is a prime q # p which divides /GI. By Cauchy's
theorem, G contains an element of order q, and this contradicts G being a
p-group. El
I Theorem 4.4. If G # 1 is a finite p-group, then its center Z(G) # 1.
ProoJ Consider the class equation
I GI = l-qG)l + C CG : C,(~i)l.
Each CG(xi)is a proper subgroup of G, for xi $Z(G). By Corollary 4.3,
[G : CG(xi)] is a power of p (since /GI is). Thus, p divides each [G : CG(xi)],
and so p divides IZ(G)I.
If G is a finite simple p-group, then G = Z(G) and G is abelian; therefore, G
must be cyclic of order p. Theorem 4.4 is false for infinite p-groups.
\
Corollary 4.5. If p is a prime, then every group G of order p2 is abelian.
Proof. If G is not abelian, then Z(G) < G; since 1 # Z(G), we must have
IZ(G)/ = p. The quotient group G/Z(G) is defined, since Z(G) 4 G, and it is
cyclic, because JG/Z(G)I = p; this contradicts Exercise 3.3. Q
Theorem 4.6. Let G be a finite p-group.
(i) If H is a proper subgroup of G, then H < NG(H).
(ii) Every nzaximal subgroup of G is normal and has index p.
I
Proof: (i) If H a G, then NG(H)= G and the theorem is true. If X is the set of
all t v conjugates of H, then we may assume that 1x1= [C : NG(H)] # 1.
%o%flacts on X by conjugation and, since G is a p-group, every orbit of X
has size a power of p. As (H) is an orbit of size 1, there must be at least p - 1
other orbits of size 1. Thus there is at least one conjugate gHg-' # H with
igHg-'} also an orbit of size 1. Now agHg-la-' = gHg-' for all a E H, and
so g-lag E NG(H)for all a E H. But gHgF1 # H gives at least one a E H with
g-lag $ N,and so H < NG(H).
(ii) If H is a maximal subgroup of G, then H < NG(H)implies that NG(H)=
G; that is, H 4 G. By Exercise 2.58, [G : HI = p.
Lemma 4.7. If G is a finite p-group and r, is the number of subgroups of G
having order p, then r, = 1 mod p.
Proof: Let us first count the number of elements of order p. Since Z(G) is
 4. The Sylow Theorems
--Since ai = 1 mod p for all i and bj = 1 mod p for all j, it follows that rs E
abelian, all its elements of order p together with 1 form a subgroup H whose rs+, mod p. Lemma 4.7 now gives the result, for r, = 1 mod p.
order is a power of p; hence the number of central elements of order p is &1
\HI - 1 E -1 mod p. If x E G is of order p and not central, then its con- Each term in the class equation of a finite group G is a divisor of ]GI, so
jugacy class xGconsists of several elements of order p; that is, lxG1> 1 is an
"honest" power of p, by Theorem 3.2. It follows that the number of elements
that multiplying by ]GI-' gives an equation of the form 1 = xj
(115) with
each i, a positive integer; moreover, JGl is the largest i, occurring in this
in G of order p is congruent to - 1 mod p; say, there are mp - 1 such ele- expression.
ments. Since the intersection of any distinct pair of subgroups of order p is
trivial, the number of elements of order p is r ,(p - 1). But r, (p - 1) = mp - 1 Lemma 4.9 (Landau, 1903). Given n > 0 and q E Q, there are only finitely
implies r, = 1 mod p. BY many n-tuples (i,, . .., in)of positive integers such that q = xy=,
(I/$),

-
Theorem 4.8. If G is a finite p-group and rs is the number of subgroups of G
having order ps, then rs 1 mod p.
Proof. We do an induction on n; the base step n = 1 is obviously true. Since
there are only n! permutations of n objects, it suffices to prove that there are
only finitely many n-tuples (i, ,.. ., in)with i, I
equation q = &,
i, I . .. < i, which satisfy the
(lli,). For any such n-tuple, we have i, 5 n/q, for
Proof. Let H be a subgroup of order ps, and let K,, ..., Kabe the subgroups
of G of order pStl which contain it; we claim that a E 1 mod p. Every sub-
group of G which normalizes H is contained in N = NG(H); in particular, But for each positive integer k 5 n/q, induction gives only finitely many
each Kj lies in N , for Lemma 4.6(ii) shows that H a Kj for all j. By the (n - 1)-typles (i,, . .. , i,,) of positive integers with q - (l/lz) = x'=, (l/i,). This
Correspondence Theorem, the number of subgroups of order p in NJH is completes the proof, for there are only finitely many such k.
eaual to the number of subgroups of N containing H which have order pstl.
1

By the lemma, a 1 mod p. Theorem 410. For every n 2 1, there are only finitely many finite groups
Now let K be a subgroup of order pstl, and let H,, . . . ,H, be its subgroups having exactly n conjugacy classes.
of order ps; we claim that b E 1 mod p. By the lemma, Hi a I< for all i. Since,
H,H, = K (for the Hi are maximal subgroups of K), the product formula Pro@$ Assume that G is a finite group having exactly n conjugacy classes. If
(Theorem 2.20) gives ID( = ps-', where D = H, n H,, and [K :D] = p2. By lZ(G)( = m, then the class equation is
Corollary 4.5, the group KID is abelian; moreover, KID is generated by two
subgroups of order p, namely, HiJD for i = 1,2, and so it is not cyclic. By
Exercise 2.68, KID z Z,x Z,.Therefore, K/D has p2 - 1 elements of order
p and hence has p + 1 = (p2 - 1)/(p - 1) subgroups of order p. The Corre- If i, = IGJ for 1 5 j 5 7 n and i, = IGI/[G: CG(x,)] = /CG(x,)l for m + 1 ij I
spondence Theorem gives p + 1 subgroups of K of order ps containing D.
Suppose there is some Hj with D $ Hj. Let E = H, n Hj; as above, there is a
n, then 1 = z;=, (I/$). By the lemma, there are only finitely many such n-
tuples, and so there is a maximum value for d l possible 4's occurring therein,
new list of p + 1 subgroups of X of order ps containing E, one of which is H,. say, M. It follows that a finite group G having exactly n conjugacy classes has
Indeed, H, = ED is the only subgroup on both lists. Therefore, there are p order at most M. But Exercise 1.41 shows that there are only finitely many
new subgroups and 1 + 2p subgroups counted so far. If some H, has not yet (nonisomorphic) groups of any given order.
been listed, repeat this procedure beginning with H, n H, to obtain p new
- - Eventually, all the Hi will be listed, and so the number of them is
subgroups.
b = 1 + mp for some m. Hence, b = 1 mod p.
Let H,, ..., firsbe all the subgroups of G of order ps, and-let K,, ...,K,,,, 4.1. Let H a G. If both H and G/H are p-groups, then G is a p-group.
be all the subgroups of order P"". For each Hi, let there be ;;;'Sobgroups of 4.2. If JGJ= pH,where p is prime, and if 0 5 k _< n, then G contains a normal sub-
order containing Hi; for each K,, let there be bj subgroups of'order p" group of order pk.
\
contained in Kj.
Mow 4.3. Let G be a finite p-group, and let H be a nontrivial normal subgroup of G. Prove
s
' rs+1 that H n Z(G) # 1.
C ai = 1 bj,
i=, j=, 4.4. Let G be a finite p-group; show that if H is a normal subgroup of G having order
p, then H s Z(G).
for either sum counts each Kj with multiplicity the number of H's it contains.
 i
I
4. The Sylow Theorems
--,
4.5. Let H be a proper subgroup of a finite p-group G. If IHI = ps, then there is a
subgroup of order ps+' containing H.
4.6. Let g be a prime, let G be a finite group whose order is divisible by p, and assume
that P 5 G is a maxin~alp-subgroup (if Q 5 G is a p-subgroup and P 5 Q, then
-
P = Q).
(i) Every conjugate of P is also a maximal p-subgroup.
(ii) If P is the only maximal p-subgroup of G, then P a G.
4.7. If p is a prime and G is a nonabelian group of order p3, then /Z(G)I = p, G/Z(G)
Z, x Z,, and Z(G) = G', the commutator subgroup.
4.8. Prove that the number of normal subgroups of order ps of a finite p-group G is
congruent to 1 mod p.
The Sylow Theorems
The main results of this section are fundamental for understanding the struc-
ture of a finite group. If pe is the largest power of p dividing (GI,then we shall
see that G contains a subgroup of order pe. Any two such subgroups are
isomorphic (indeed, they are conjugate), and the number of them can be
counted within a congruence.
Definition. If p is a prime, then a Sylow p-subgroup P of a group G is a
maximal p-subgroup.
Observe that every p-subgroup of 6 is contained in some Sylow p-sub-
group; this is obvious when G is finite, and it follows from Zorn's lemma
when G is infinite. (Although we have allowed infinite groups G, the most
important groups in this context are finite.)
Lemma 4.11. Let P be a Sylow p-subgroup of a finite group G.
(i) ING(P)/PIis prime to p.
(ii) If a E G has order some power of p and aPa" = P, then a E P.
Proof. (i) If p divides ING(P)/PI,then Cauchy's theorem (Theorem 4.2) shows
that NG(P)/P contains some element P a of order p; hence, S* = (Pa) has
order p. By the Correspondence Theorem, there is a subgroup s 5 NG(P)I
G containing P with S/P z S*. Since both P and S* are p-groups, Exercise 4.1
shows that S is a p-group, contradicting the maximality of P.
(ii) Replacing a by a suitable power of a if necessary, we may assume that
a has order p. Since a normalizes P, we have a E NG(P).If a $ P, then the coset
Pa E NG(P)/Phas order p, and this contradicts (i).
The observation suggesting the coming proof is that every conjugate of a
Sylow p-subgroup is itself a Sylow p-subgroup.
1
i The Sylow Theorems
I
Theorem 4.12 (Sylow, B 872).
I
of G are conjugate to P.
-
(i) If P is a Sylow p-subgroup of a finite group 6 , then all Sylow p-subgroups
(ii) If there are r Sylow p-subgroups, then r is a divisor of I GI and r 1 mod p.'
Proof. Let X = {PI, ..., P,) be the family of all the Sylow p-subgroups of G,
where we have denoted P by P I . In Theorem 3.17, we saw that G acts on X
by conjugation: there is a homomorphism J,: G -+ S, sending a I--+ where
$a(Pi) = aPia-'. Let Q be a Sylow p-subgroup of G. Restricting J, to Q shows
that Q acts on X ; by Corollary 3.21, every orbit of X under this action has
size dividing lQl; that is, every orbit has size some power of p. What does it
mean to say that one of these orbits has size I? There would be an i with
$a(Pi) = Pi for all a E Q; that is, ,Pi,-' = Pi for all a E Q. By Lemma 4.1 1(ii),
if a E Q, then a E Pi; that is, Q 5 Pi; since Q is a Sylow p-subgroup, Q = Pi.If
Q = P = PI, we conclude that every P-orbit of X has size an "honest" power
of p save {PI1 which has size 1. Therefore, 1x1= r r 1 mod p.
Suppose there were a Sylow p-subgroup Q that is not a conjugate of P; that
-
is, Q 4 X . If (Pi) is a Q-orbit of size 1, then we have seen that Q = Pi, contra-
dicting Q $ X. Thus, every Q-orbit of X has size an honest power of p, and so
p divides 1x1; that is, r 0 mod p. The previous congruence is contradicted,
and so no such subgroup Q exists. Therefore, every Sylow p-subgroup Q is
conjugate to P.
\
Finally, the number r of conjugates of P is the index of its normalizer, and
so it is a divisor of (GI. El
For example, IS4/ = 24 = 2). 3, and so a Sylow 2-subgroup of S4 has order
8. It is easily seen that D, I S4 if one recalls the symmetries of a square. The
Sylow theorem says that all the subgroups of S, of order 8 are conjugate
(hence isomorphic) and that the number r of them is an odd divisor of 24.
Since r. # 1, there are 3 such subgroups.
We have seen, in Exercise 4.6, that if a finite group G has only one Sylow
p-subgroup P, fo; some prime p, then P a G. We can now see that the con-
verse is also true.
Corollary 4.13. A finite group 6 has a unique Sylow p-subgroup P, for some
prime p, $ and only $ P a G.
Proof: If G has only one Sylow p-subgroup P, then P 4 G, for any conjugate
of is also a Sylow p-subgroup. Conversely, if P is a normal Sylow p-
subgroup of G, then it is unique, for all Sylow p-subgroups of G are conju-
gate. f34
I '. Since
-
all Sylow p-subgroups have the same order, this congruence also follows from Theorem
80 4. The Sylow Theorems
Theorem 4.14. If G is a finite group of order pem, where (p, m) = 1, then every
Sylow p-subgroup P of G has order pe.
ProoJ We claim that [G : P I is prime to p. Now [G : P I = [G : N] [N : PI,
where N = NG(P),and so it sufices to prove that each of the factors is prime to
p. But [G : N] = r, the number of conjugates of p, so that [G : N] = 1 mod p,
while [N : P I = I NIP1 is prime to p, by Lemma 4.1 1(i).
By Lagrange's theorem, IPI = pk, where k 5 e, and so [G : P I = / G / / / P =
J Now p cannot divide the size of every orbit of X lest 1x1; therefore, there is
Since [G : P I is prime to p, however, we must have 12 = e. i@l
Corollary 4.15. Let G be a finite group and let p be a prime. If pk divides \GI,
then G contains a subgroup of order pk.
Proof. If P is a Sylow p-subgroup of G, then pk divides I PI, and the result now
follows from Exercise 4.2. El
We have now seen how much of the converse of Lagrange's theorem (if a
divides 1 GI, then G has a subgroup of order m) can be salvaged. If m is a prime
power, then G contains a subgroup of order m; if m has two distinct prime
factors, however, we have already seen an example (Theorem 3.7) in which G
has no subgroup of order m (namely, m = 6 and G = A,, a group of order 12).
Since, for each prime p, any two Sylow p-subgroups of a finite group G are
isomorphic (they are even conjugate), we may list the Sylow subgroups of G,
one for each prime. It is plain that isomorphic groups G give the same list,
but the converse is false. For example, both S, and Z,give the same list.
Here is another proof of Sylow's theorem, due to Wielandt, that does not
use Cauchy's theorem (and so it gives a third proof of Cauchy7stheorem).
Lemma 4.16. If p is a prime not dividing an integer m, then for all n > 1, the
binomial coefficient ti:) is not divisibte by p.
Proof. Write the binomial coefficient as follows:
Since p is prime, each factor equal to p of the numerator (or of the denomina-
tor) arises from a factor of pnm - i (or of p" - i). If i = 0, then the multiplicity
of p in p"m and in pn are the same because pjm. If 1 5 i 5 pn, then i = pl'j,
where 0 5 12 < n and pjj. Now pk is the highest power of p dividing p" - i, for
P" - i = - pl'j = Pk(P"-k- j) and pXpl'-' - j (because n - k > 0). A simi-
lar argument shows that the highest power of p dividing pnm - i is also pk.
Therefore, every factor of p upstairs is canceled by a factor of p downstairs,
and hence the binomial coefficient has no factor equal to p. Eil
1 The Sylow Theorems 81
Theorem 4.17 (Wielandt's Proof). If G is a finite group of order pnm, where
(p, m) = 1, then G has a subgroup of order p".
I
Proof. If X is the family of all subsets of G of cardinal p", then 1x1 is the
binomial coefficient in the lemma, and so pj'IXI. Let G act on X by left
translation: if B is a subset of G with p" elements, then for each g E G, define
gB = (gb: b E B).
some B E X with /B(B)Inot divisible by p, where B(B) is the orbit of B. If GB
is the stabilizer of B, then lGl/lGBI = [G : G,] = IB(B)I is prime to p. Hence,
IGBJ= pnm' 2 p" (for some m' dividing m). On the other hand, if b, E B and
g E GB, then gb, E gB = B (definition of stabilizer); moreover, if g and h are
distinct elements of GB, then gb, and hb, are distinct elements of B. There-
fore, JGBJr JBI = pn, and so JG,j = p". El
The next technical result is useful.
Theorem 4.18 (Frattini Argument). Let I< be a normal subgroup of a finite
group G. If P is a Sylow p-subgroup of K (for some prime p), then
Proof. If g E G, then gPg-I I gKg-' = K, because K a G. If follows that
g ~ g - 'is a Sylow p-subgroup of K, and so there exists k E K with kPk-' =
LJP~-'. Hence, P = (k-'g)~(k-'~)-', so that k-'g E NG(P).The required fac-
torization is thus g = k(k-'8). E3
4.9. (i) Let X be a finite G-set, and let H < G act transitively on X. Then G = IIG,
for each x E X.
(ii) Show that the Frattini argument follows from (i).
4.10. Let {Pi: i E I ) be a set of Sylow subgroups of a finite group G, one for each prime
divisor of jGj. Show that G is generated by U Pi.
4.11. Let P 5 G be a Sylow subgroup. If N,(P) <H 2 G, then H is equal to its own
normalizer; that is H = NG(H).
4.12. If a finite group G has a unique Sylow p-subgroup for each prime divisor p of
[GI,then G is the direct product of its Sylow subgroups.
4.13. (i) Let G be a finite group and let P 5 G be a Sylow subgroup. If H G, then
H n P is a Sylow subgroup of H and HPIH is a Sylow subgroup of GIH.
(Hint. Compare orders.)
(ii) Let G be a finite group and let P 5 G be a Sylow subgroup. Give an example
of a (necessarily non-normal) subgroup H of G with H n P not a Sylow
subgroup of II.

82 4. The Sylow Theorems
4.14. Prove that a Sylow 2-subgroup of A , has exactly 5 conjugates.
\
4.15. (i) Prove that a Sylow 2-subgroup of S, is isomorphic to D,.
(ii) Prove that D, x H2 is isomorphic to a Sylow 2-subgroup of S,.
4.16. If Q is a normal p-subgroup of a finite group G, then Q 5 P for every Sylow
p-subgroup P.
Definition. An n x n matrix A over a commutative ring R is unitriarrgrrlar if it
has 0's below the diagonal and 1's on the diagonal. The set of all uni-
triangular 3 x 3 matrices over B, is denoted by UT(3, a,).
4.17. (i) Show that IGL(3,EP)I = (p3 - l)(p3 - p)(p3- p2).
(ii) If p is a prime, then UT(3, E,) is a Sylow p-subgroup of GL(3, a,).
(iii) Show that the center of UT(3, E,) consists of all matrices of the form
4.18. Show that a finite group G can have three Sylow p-subgroups A, B, and C such
that A n B = 1 and A n C # 1. (Hint. Take G = S3 x S,.)
4.19. Let IG] = pnm, where pJm. If s 2 n and rs is the number of subgroups of G of
order ps, then r, = 1 mod p.
4.20. (i) Let a = (1 2 3 4 5), let P = (rr) < S,, and let N = Ns,(P). Show that
JNI = 20 and N = (0, a), where a = (2 3 5 4).
(ii) If A is the group (under composition),
A={q:Z,-Z,:q(x)=ax+P,a,P~~~,a#O),
then Ns,(P) E A. (Hint. Show that A = {s, t ) , where s: x H x
t: x I-+ 2x.)
Groups of Small Order
We illustrate the power of the Sylow theorems by classifying the groups of
small order.
Theorem 4.19. If p is a prime, then every group G of order 2p is either cyclic or
dihedral.
ProoJ If p = 2, then IGl = 4, and the result is Exercise 2.12. If p is an odd
prime, then Cauchy's theorem shows that G contains an element s of order p
-
and an element t of order 2. If H = (s), then H has index 2 in G, and so H d
G. Therefore, tst = si for some i. Now s = t2st2 = t(tst)t = tsit = si2; hence,
i2 1 mod p and, because p is prime, Euclid's lemma gives i E & 1 mod p.
Groups of Small Order
83
Thus, either tst = s or tst = s-'. In the first case, s and t commute, G is
\--
abelian, and Exercises 4.12 and 2.62(ii) give G E Z, x Z2 z B,,; in the sec-
ond case, G E D,,.
We now generalize this result by replacing 2 by q.
Theorem 4.20. Let (GI = pq, where p > q are primes. Then either G is cyclic or
G = (a, b), where
and mq = 1 mod p but rn f 1 mod p. If q/p - 1, then the second case cannot
occur.
Proof: By Cauchy's theorem, G contains an element b of order p; let S = <b).
Since S has order p, it has index q. It follows from Exercise 3.33 that S a G.
Cauchy's theorem shows that G contains an element a of order q; let
T = (a). Now T is a Sylow q-subgroup of G, so that the number c of its
+
conjugates is 1 kq for some k 2 0. As above, either c = 1 or c = p. If c = 1,
then T a G and G s S x T (by Exercise 4.12), and so G z Z, x Z,E B,,, by
+
Exercise 2.62(ii). In case c = kq 1 = p, then qlp - I, and T is not a normal
subgroup of G. Since S CI G, aba-' = b" for some m; furthermore, we may
+
assume that rn 1 mod p lest we return to the abelian case. The reader may
prove, by induction on j, that ajba-j = b"'. In particular, if j = q, then mq I
1 modp. H
Corollary 4.21. If p > q are primes, then every group G of order pq contains a
t
normal subgroup of order p. Moreover, if q does not divide p - 1, then G must
be cyclic.
+ 1, and
For example, the composite numbers n 5 100 for which every group of
order n is cyclic are:
Definition. The quaternions is a group Q = (a, b) of order 8 with a4 = 1,
b2 = a2, and bab-' = a-I.
We continue describing groups of small order.
Theorem 8 2 2 . Q and D, are the only nonabelian groups of order 8.
ProoJ A nonabelian group G of order 8 has no element of order 8 (lest it be
cyclic), and not every nonidentity element has order 2 (Exercise 1.26); thus, G
has an element a of order 4. Now (a) a G, for it has index 2, and G/(a) E B2.
If b E G and b # (a), then b2 E (a) (Exercise 2.16). If b2 = a or b2 = a3, then
 -
4. The Sylow Theorems Groups of Small Order 85
84

b has order 8, a contradiction; therefore, either
,
b2=a2 or b2=1. -'-,
Furthermore, bab-' E (a), because (a) is normal, so that
bab-I = a or bab-' = a3
(these are the only possibilities because a and bab-' have the same order).
The first case is ruled out, for G = (a, b) and G is abelian if a and b commute.
The following case remain:
(i) a4 = 1, b2 = a', and bab-' = a3; and
(ii) a4 = 1, b2 = 1, and bab-I = a3.
Since a3 = a-', (i) describes Q and (ii) describes D,.
Lemma 4.23. If G has order 12 and G $ A,, then G contains an element of
order 6; moreover, G has a normal Sylow 3-subgroup, hence has exactly two
eEements of order 3.
Pro06 If P is a Sylow 3-subgroup of G, then I Pl = 3 and so P = (b) for some
b of order 3. Since [G : P] = 4, there is a homomorphism $: G --+ S, whose
kernel K is a subgroup of P; as I PI = 3, either K = 1 or K = P. If K = 1, then
$ is an injection and G is isomorphic to a subgroup of S, of order 12; by
Exercise 3.15, G EJ A,. Therefore, K = P and so P a G; it follows that P is
the unique Sylow 3-subgroup, and so the only elements in G of order 3 are b
and b2. Now [G : CG(b)]is the number of conjugates of b. Since every conju-
gate of b has order 3, [G : C,(b)] 5 2 and ICG(b)l= 6 or 12; in either case,
CG(b)contains an element a of order 2. But n commutes with byand so ab has
order 6. ED
We now define a new group.
P has order 4). Notice that G = KP, for P is a maximal subgroup because it
has prime index. Now either P V or P z Z,.
In the first case, P = {1, x, y, z}, where x, y, and z are involutions. Not
every element of P commutes with k lest G be abelian; therefore, there exists
an involution in P, say x, with xkx # k. But xkx is a conjugate of k, hence has
order 3; as G has only two elements of order 3, xkx = k-', and (x, k) D6 "
( E JS,). We claim that either y or z commutes with k. If y does not commute
with k, then, as above, yky = k-', so that xkx = yky and z = xy commutes
with k. If a = zk, then G = (a, x), n has order 6, and xax = xzkx = zxkx =
212-' = a-', so that G E Dl,.
In the second case, P = (x) z Z,.Now x and k do not commute, because
G = (x, 12) is not abelian, and so xkx-' = k-'. But x2 does commute with k,
for x2kx-' = ~(xkx-')x-' = xlz-'x-' = k; hence, a = x212 has order 6. Since
x2 and 12 commute, a3 = ( ~ ~=kx6k3 ) ~= x2. Finally, (ax)2 = axax =
1zx3kx3= (kx-')kx-' = ( ~ - ' k - ' ) l ~ x -(because
~ xkx-' = 12-') = x W 2= x2. We
have shown that G s T in this case.
One can prove this last result in the style of the proof of Theorem 4.22,
beginning by choosing an element of order 6. The reader may see that the
proof just given is more efficient.
Notice that T is almost a direct product: it contains subgroups P = (x)
and K = (k) with PK = G, P n K = 1, but only P is normal.
If n 2 1, define G(n) to be the number of nonisomorphic groups of order n.
No one knows a formula for 6(n), although it has been computed for n 100
and beyond. Here is the classification of all groups of order < 15 followed by
a table of values of O(n) for small n. Of course, 6(n) = 1 whenever n is prime.
Table of Groups of Small Order
Order n 5(n) Groups2

Definition. T is a group of order 12 which is generated by two elements a and

b such that a6 = 1 and b2 = a3 = ( L X ~ ) ~ .

It is not obvious that such a group T exists; we shall construct such a

group later, in Example 7.14.

Theorem 4.24. Every nonabelian group G of order 12 is isomorphic to either

A4, Dl 2 , Or T.
Some Other Values of B(n)
Remark. Exercise 3.50 shows that S3 x Z2 z Dl,. n 16 18 20 21 22 24 25 26 27 28 30 32 64
Proof. It suffices to show that if G is a nonabelian group of order 12 which is 14 5 5 2
Q(n) 2 15 2 2 5 4 4 51 267
not isomorphic to A,, then either G 2 Dl, or G z T.
Let K be a Sylow 3-subgroup of G (so that K = (k) is cyclic oforder 3 The abelian groups in the table will be discussed in Chapter 6.
and, by Lemma 4.23, K a G), and let P be a Sylow 2-subgroup of G (so that
 4. The Sylow Theorems
86
R. James, M.F. Newman, and E.A.O'Brien (1990) have shown, using com-
puters, that G(128) = 2328; E.A.O'Brien (1991) has shown that 6(256) =
56,092, and that 6(512) > 8,000,000 (counting only those groups of "class-T).
G. Higman (1960) and C. Sims (1965) showed that 6 ( p n )is about p2'""7. If
the prime factorization of an integer n is p;lp;'. ..,p:, define p = p(n) to be
the largest e,. Using the fact that there are at most two nonisomorphic simple
groups of the same finite order (which follows from the classification of the
finite simple groups), A. McIver and P.M. Neumann (1987) have shown that
G(n) 5 n;'+pi2.
Here is another type of application of the Sylow theorems.
EXAMPLE
4.1. There is no simple group of order 30.
Such a group would have r Sylow 5-subgroups, where r E 1 mod 5 and
r130, Now r # 1, lest G have a normal subgroup, and so r = 6. Aside from the
identity, this accounts for 24 elements (for distinct subgroups of order 5
intersect in 1).Similarly, there must be 10 Sylow 3-subgroups, accounting for
20 elements, and we have exceeded 30.
EXAMPLE
4.2. There is no simple group of order 36.
A Sylow 3-subgroup P of such a group G has 4 conjugates, and so
[G: PI = 4. Representing G on the cosets of P gives a homomorphism
$: G 4 S4 with ker $ 2 P; since G is simple, $ must be an injection and G is
imbedded in S4. This contradicts 36 > 24, and so no such group G can exist.
EXERCISES
4.21. A, is a group of order 60 containing no subgroups of order 15 or of order 30.
4.22. If G is the subgroup of GL(2, C) generated by
A=[: b] and B = [ -1 0 ' '1
then G z Q. (See Exercise 2.24.)
4.23. The division ring W of real quaternions is a four-dimensional vector space over
R having a basis {l, i, j, k) with multiplication satisfying
ij = k, jk = i, ki = j , ji = -k, kj = -i, ik = -j
(these equations eight determine the multiplication on all of W.) Show that the
eight elements { & 1, & i, +j, ik) form a multiplicative group isomorphic to Q.
4.24. Show that Q has a unique element of order 2, and that this element generates
Z(Q)-
Groups of Small Order
4.25. Prove that D, $ Q.
4.26. Prove that Q contains no subgroup isomorphic to Q/Z(Q).
4.27. Prove that every subgroup of Q is normal.
4.28. Let G = Q x A x Bywhere A is a (necessarily abelian) group of exponent 2 and
B is an abelian group in which every element has odd order. Prove that every
subgroup of G is normal. (Dedekind proved the converse: if G is a finite
- -------
group in which every subgroup is normal, then G = Q x A x B as above.)
Groups with this property are called iiarniltonian, after W.R. Hamilton who
discovered Q.
4.29. Let SL(2, 5) denote the multiplicative group of all 2 x 2 matrices over k, which
have determinant 1.
(i) Show that (SL(2,5)l = 120.
(ii) Show that Q is isomorphic to a Sylow 2-subgroup of SL(2, 5). (Hint. Show
that SL(2,5) has a unique involution.)
(iii) Show that a Sylow 2-subgroup of S, is D,, and conclude that SL(2, 5) 7;. S,.
(iv) Show that A , cannot be imbedded in SL(2,5).
4.30. For every divisor d of 24, show that there is a subgroup of S4 of order d.
Moreover, if d # 4, then any two subgroups of order d are isomorphic.
4.31. Exhibit all the subgroups of S4; aside from S, and 1, there are 26 of them.
4.32. (i) Let p be a prime. By Exercise 4.17, P = UT(3, k,) is a Sylow p-subgroup of
GL(3, k,). If p is odd, prove that P is a nonabelian group of order p3 of
exponent p. (Compare Exercise 1.26.) If p = 2, show that UT(3, k,) 1Q.
(ii) Let p be an odd prime. Prove that there are at most two nonabelian groups
of order p3. One is given in part (i) and has exponent p; the other has
generators a and b satisfying the relations up2 = 1, bp = 1, and bab-' = al+p.
(This group will be shown to exist in Example 7.16.)
4.33. Give an example of two nonisomorphic groups G and H such that, for each
positive integer d, the number of elements in G of order d is equal to the number
of elements in H of order d.
4.34. If G is a group of order 8 having only one involution, then either G E Z,or
GEQ. ,
4.35. If p and q are primes, then there is no simple group of order p2q.
4.36. Prove that there is no nonabelian simple group of order less than 60.
4.37. Prove that any simple group G of order 60 is isomorphic to A,. (Hint. If P and
Q are distinct Sylow 2-subgroups having a nontrivial element x in their intersec-
tion, then C,(x) has index 5; otherwise, every two Sylow 2-subgroups intersect
trivially and NG(P)has index 5.)
4.38. In Corollary 7.55, we shall prove that a group of squarefree order cannot be
simple. Use this result to prove that if IGI = p, .. .p,, where p, < p, < - .. < p,
are primes, then G contains a normal Sylow p,-subgroup.
Definition. If n 2 3, a generalized quafernion group (or dicyclic group) is a
group Q , of order 2" generated by elements a and b such that
a 2 n - l = 1, bab-l = a-l, and b2 = a2r'-'.
 4. The Sylow Theorems

4.39. Any two groups of order 2" generated by a pair of elements a and b for which CHAPTER 5
= b2 and aba = b are isomorphic.

4.40. If G is a group of order 2" which is generated by elements a and b such that
Normal Series
= b2 = (&)*,
a2n-2

.
t -

then G r Q,.
4.41. Prove that Q,, has a unique involution z, and that Z(Q,) = ( 2 ) .
4.42. Prove that Q , , / Z ( Q , )r D z n - l .
4.43. Let G IGL(2, C) be generated by

A =La
-
o w
o] and [
I? = -1 0l'] '

where a is a primitive 2 l l - l th root of unity for n 2 3. Prove that G E Q , , . Con-

clude that Q, exists.

We begin this chapter with a brief history of the study of roots of polyno-
mials. Mathematicians of the Middle Ages, and probably those in Babylonia,
knew the quadratic fovrnrrla giving the roots of a quadratic polynomial
+ +
f(X) = x2 bX C. Setting X = x - $b transforms f(X) into a polynomial
g(x) with no x term:
g(x) = x2 + c - ib2.
Note that a number a is a root of g(x) if and only if a - $b is a root of
f(X). The roots of g(x) are f$JG, and so the roots of f(X) are
$(-b +d m ) .
Here is a derivation of the cubic formula1 (due to Scipione del Ferro,

' Negative coeflicients (and negative roots) were not accepted by mathematicians of the early
1500s. There are several types of cubics if one allows only positive coefiicients. About 1515,
Scipione del Ferro solved some instances of x3 + px = q, but he kept his solution secret. In 1535,
Tartaglia (Niccolb Fontana) was challenged by one of Scipione's students, and he rediscovered
the solution of x 3 $ px = q as well as the solution of x3 = px + q. Eventually Tartaglia told his
solutions to Cardano, and Cardano completed the remaining cases (this was no small task, for
all of this occurred before the invention of notation for variables, exponents, +, --, x, /, ,/-,
and =). A more complete account can be found in J.-P. Tignol, Galois' Theory of Equations.
We should not underestimate the importance of Cardano's formula. First of all, the physicist
R.P. Feynmann suggested that its arising at the beginning of the Renaissance and Reformation
must have contributed to the development of modern science. After all, it is a genuine example
of a solution unknown to the Greeks, and one component contributing to the Dark Ages was
the belief that contemporary man was less able than his classical Greek and Roman ancestors.
Second, it forced mathematicialls to take the complex numbers seriously. While complex roots
of quadratics were just ignored, the casus irreducibilis, arising from one of Cardano's cases,
describes real roots of cubics with imaginary numbers (see the next footnote). Complex numbers
could no longer be dismissed, and thus the cubic formula had great impact on the development
of mathematics.
 Some Galois Theory
90 5. Normal Series 91

Factor g(x) into quadratics:

Tartaglia (Niccoli, Fontana), and Cardano). A cubic f(X) = X3 + a x 2 +
bX + c can be transformed, by setting X = x - ?a, into a polynomial g(x)
with no x2 term: (the coefficient of x in the second factor must be - k because there is no cubic
g(x) = x3 + qx r, + term in g(x)). If k, I, and m can be found, then the roots of g(x) can be found
and a number a is a root of g(x) if and only if a - $a is a root of f(X). If a is by the quadratic formula. Expanding the right side and equating coefficients
a root of g(x), write a = ,8 + y, where p and y are to be found. Now of like terms gives
a3 = (p + y)3 = p3 + y3 + 3(P2y + Py2) l+m-k2=q,

lm = s.
and so evaluating g(a) gives
Rewrite the first two equations as
(1) + +
p3 y3 (3py q)a r = 0. + +
Impose the condition that py = -q/3 (forcing the middle term of (1) to van-
ish). Thus,
p3 + y 3 = -7.. Adding and subtracting these equations gives
We also know that
pSY3= -q3/27,
and we proceed to find P3 and y3. Substituting,
These two equations show that we are done if k can be found. But
p3 - q3/27p3 = --I, + + +
(k2 q - r/1z)(k2 q r/k) = 41m = 4s gives
and so the quadratic formula yields
p3 = $[-r i.(r2 + 4q3/27)'I2]. a cubic in 1c2. The cubic formula allows one to solve for k2, and it is now easy
to determine I, m, and the roots of g(x).
Similarly,
Y3 = $[--I (r2 + 4q3/27)'I2].

If w = eZnil"s a primitive cube root of unity, there are now six cube roots
available: P, wp, w2p, y, wy, w2y; these may be paired to give product - q/3: Some Galois Theory
- 413 = PY = (UP)(w2y) = (w2B)(my).
Let us discuss the elements of Galois Theory, the cradle of group theory. We
+ +
'It follows that the roots of g(x) are P y, wp cozy, and w2P + my; this is are going to assume in this exposition that every field F is a subfield of an
the cuhicf i ~ n a u l a . ~ algebraically closed field C. What this means in practice is this. Iff (x) E F[x],
The quavticformula was discovered by Lodovici Ferrari, about 1545; we the ring of all polynomials with coefficients in F, and if f(x) has degree n 2 1,
present the derivation of this formula due to Descartes in 1637. A quartic then there are (not necessarily distinct) elements a,, . .. ,an in C (the roo0 of
f i x ) = x4+ aX3 + bX2 + CX + d can be transformed, by setting X =
J \ I
f(x)) and nonzero a E F so that
x - $a, into a polynomial g(x) with no xJ term:
g(x) = x4 + qx2 + rx + S; in C[x]. The intersection of any family of subfields of a field is itself a subfield;
moreover, a number a is a root of g(x) if and only if a - i a is a root of f(X). define the smaIkest srih$ieId of C containing a given subset X as the intersec-
tion of all those subfields of C containing X. For example, if a E C, the smallest
The roots of f ( x ) = ( x - l ) ( x - 2 ) ( x + 3) = x3 - 7x + 6 are obviously 1,2, and -3. However, subfield of C containing X = F u {a) is
the cubic formula gives
P+u=1$(-6+ ,/367%)+1$(-6-,/=).
92 5. Normal Series Some Galois Theory 93

F(a) is called the subfield obtained from F by adjoinirtg a. Similarly, one can If CT is a bijection, then a is an isornorpirisrn; an isomorphism a: E -+ E is
define F(a,, . . . , a,,), the subfield obtained from F by adjoining a,, . ..,a,,. In called an raartomorplrism.
particular, if f (x) E F [XIand f (x) = (x - a,)(x - a,). ..(x - a,) E C [x], then
F(a,, . . .,a,), the subfield obtained from F by adjoining all the roots of f(x), Lemma 5.1. Let f (x) E F[x], let E be its splitting field over F, and let 0: E -+ E
is called the splittingfield off (x) over F. Notice that the splitting field off&) be an automorphismfixing F (i.e., a(a) = a for all a E F). If a E E is a root of
does depend on F. For example, if f(x) = x2 + 1 E Q[x], then the splitting f (x), then a(a) is also a root off (x).
field of f(x) over Q is Q(i); on the other hand, if we regard f(x) E ~[x];then
its splitting field off (x) over IW is C. ProoJ Iff (x) =I x =C a(ai)a(a,i
aixi, then 0 = a (f (a))= ~ ( aiai) =I~ ~ a ( a ) ~ ,
It is now possible to give the precise definition we need. and so @(a)is a root of f(x).

Definition. Let f (x) E F [XIhave splitting field E over F. Then f (x) is solvable Lemma 5.2. Let F be a subfield of K, let (a,, . ..,a,,) c K, and let E =
by radicals if there is a chain of subfields F(a,, . . . ,a,,). If I<' is a field containing F as a subfield, and if 0: E -+ K' is a
homomorphism fixing F with a(ai) = ai for all i, then a is the identity.

in which E c K , and each K,,, is obtained from K iby adjoining a root of an Pro@$ The proof is by induction on n > 1. If n = 1, then E consists of all
element of K,; that is, K,,, = K,(P,+,),where Pi+, E Ki+, and some power of g(a,)/h(a,), where g(x), h(x) E F [XIand h(a,) # 0; clearly a fixes each such
Pi+, lies in Ki. element. The inductive step is clear once one realizes that F(a,, . .. , a,,) =
F*(aI,),where F* = F(a,, ..., a,-,).
When we say that there is a formula for the roots of f(x), we really mean
that f(x) is solvable by radicals. Let us illustrate this by considering the It is easy to see that if F is a subfield of a field E, then the set of all
quadratic, cubic, and quartic formulas. automorphisms of E which fix F forms a group under composition.
Iff (x) = x2 + bx + C,set F = Q(b, c). If /? = Jn, then P2 E F; define
K, = F(P), and note that K , is the splitting field of f(x) over F. Definition. If F is a subfield of E, then the Galois group, denoted by Gal(E/F),
If f(x) = x3 + qx + I, set F = Q(q, r). Define p, = d m , and de- is the group under composition of all those automorphisms of E which fix F.
fine K , = F (P, ); define P2 = IT, and define K, = K , (P,). Finally, Iff (x) E F [XIand E = F(a,, .. . , a,) is the splitting field off (x) over F, then
set K , = K2(w), where w is a cube root of unity. Notice that the cubic for- the Gwlois groadp off (x) is Gal(E/F).
mula implies that K, contains the splitting field E of f(x). On the other hand,
E need not equal K,; for example, if all the roots of f(x) are real, then E c W Theorem 5.3. Let f (x) E F [x] and let X = {a,, ...,a,,) be the set of its distinct
but I<, $ R. roots (in its splitting field E = &'(a,, . .. ,a,,) over F). Then the function
If f(x) = x4 + qx2 + rx + s, set F = Q(q, r, s). Using the notation in our q: Gal(E/F) + S, s S,, given by q(a) = ojX, is an imbedding; that is, O, is
discussion of the quartic, there is a cubic polynomial having k2 as a root. As completely determined by its action on X .
above, there is a chain of fields F = K O c K , c I<, c K, with k2 E K,. De-
fine I<, = K,(k), K , = &(A), where y = k2 - 41, and KG= I<,($), where Proof. If CJ E Gal(E/F), then Lemma 5.1 shows that o(X) c X; olX is a bijec-
6 = 1c2 - 4m. The discussion of the quartic formula shows that the splitting tion because a is an injection and X is finite. It is easy to see that 9 is a
field of f(x) is contained in KG. homomorphism; it is an injection, by Lemma 5.2.
Conversely, it is plain that iff (x) is solvable by radicals, then every root of
f(x) has some expression in the coefficients of f(x) involving the field opera- Not every permutation of the roots of a polynomial f(x) need arise from
tions and extractions of roots. some a E Gal(E/F). For example, let f(x) = (x2 - 2)(x2 - 3) E Q[x]. Then
Definitic~n.If E and E' are fields, then a hormomorghism is a function a: E -+ E'
E = O($, fi) and there is no CJ E Gal(E/O) with a(&) = fi.
such that, for all a, fi E E,
a(1) = 1, Definition. If F is a subfield of a field E, then E is a vector space over F (if
a E F and a E E, define scalar multiplication to be the given product aa of two
elements of E). The degree of E over F, denoted by [E : F], is the dimension
of E.
94 5. Normal Series
5.1. Show that every homomorphism of fields a: E -,K is an injection,
5.2. Let p(x) E F[x] be an irreducible polynomial of degree n. If a is a root of p(x) (in
a splitting field), prove that (1, a, a', ..., an-'} is a basis of F(a) (viewed as a
vector space over F). Conclude that [F(a) : F] = n. (Hint. The rings F[x]/(p(x))
and F(a) are isomorphic via g(x) + (p(x))H g(a).)
5.3. Let F c E c K be fields, where [K : El and [E : F] are finite. Prove that [K : F] =
[K : El [E : F]. (Hint. If (a,, ..., a,,) is a basis of E over F and if (PI, .. . ,P,} is
a basis of K over E, then the set of rnn elements of the form aiPj is a basis of K
over F.)
5.4. Let E be a splitting field over F of some f (x) E F [x], and let K be a splitting field
over E of some g(x) E E[X]. If CJ E Gal(K/F),then alE E Gal(E/F).(Hint. Lemmas
5.1 and 5.2.)
5.5. Let f(x) = x" - a E F[x], let E be the splitting field of f(x) over F, and let a E E
be an nth root of a. Prove that there are subfields
with = Ki(Bi+'),Pff] E PCi, and p ( i ) prime for all i. (Hint. If n = st,
an= (aS)?.)
Lemma 5.4. Let p(x) E F[x] be irreducible, and let a and P be roots of p(x) in
a splitting field of p(x) over F. Then there exists an isomorphism A*: F(a) -+
F(P) which fixes F and with A*(a) = P.
ProoJ By Exercise 5.2, every element of F(a) has a unique expression of the
-,
form a, + a, a + . . + a,, at'-'. Define A* by
+
It is easy to see that A* is a field homomorphism; it is an isomorphism
because its inverse can be constructed in the same manner. E4
Remarlc. There is a generalization of this lemma having the same proof.
Let A: F -+ F' be an isomorphism of fields, let p(x) = a, + a,x + + a,,xnE
F[x] be an irreducible polynomial, and let pl(x) = A(a,) + A(a,)x + .. . 4-
A(a,,)xnE F'[x]. Finally, let a be a root of p(x) and let fi be a root of pl(x) (in
appropriate splitti~lgfields). Then there is an isomorphism A*:'F(a) + F1(/3)
with A*(a) = /3 and with A*/F = A.
Lemma 5.5. Let f(x) E F[x], and let E be its splitting field over F. If K is an
"intermediate field," that is, F c K c E, and if A: K -+ K is an autornorphism
fixing F, then there is an automorphism A*: E -+E with A"/ K = A.
Proof. The proof is by induction on d = [E : F]. If d = 1, then E = K, every
root a,, . . .,a,, off (x) lies in I<, and we may take A* = A. If d > 1, then E f K
Some Galois Theory 95
and there is some root a of f(x) not lying in K (of course, a E E, by definition
of splitting field). Now a is a root of some irreducible factor p(x) off (x); since
a $ K, degree p(x) = k > 1. By the generalized version of Lemma 5.4, there
is p E E and an isomorphism A,: K(a) -+ K(p) which extends A and with
&(a) = p. By Exercise 5.3, [E : K(a)] = d/k < d. Now E is the splitting field
of f(x) over K(a), for it arises from K(a) by adjoining all the roots of f(x).
Since all the inductive hypotheses have been verified, we conclude that A,,
and hence A, can be extended to an automorphism of E. H
\
Remark. As with the previous lemma, Lemma 5.5 has a more general version
having the same proof. It says that if f(x) E F[x], )then any two (abstract)
splitting fields off (x) over F are isomorphic.
Theorem 5.6. Let p be a prime, let F be a field containing a primitive pth root
of unity, say, o , and let f(x) = x p - a E F[x].
(i) If a is a root off (x) (in some splitting field), then f(x) is irreducible if and
only if a $ Fa
then
(ii) The splitting field E of f(x) over F is F(a).
(iii) Iff (x) is irreducible, then Gal(E/F) r Z,.
Proof. (i) If a E F, then f(x) is not irreducible, for it has x - a as a factor.
Conversely, assume that f(x) = g(x)h(x), where degree g(x) = Ic < p. Since
the roots of f(x) are a, oa, 02a, ...,oP-la, every root of g(x) has the form o i a
for some i. If the constant term of g(x) is c, then c = +wrak for some r (for c
is, to sign, the product of the roots). As both c and o lie in F, it follows that
+
a k E F. But (k, p) = 1, because p is prime, and so 1 = ks tp for some integers
s and t. Thus
a=aks+t~=
(ak )s (ap t E F.
(ii) Immediate from the observation that the roots of f(x) are of the form
LL)ia.
(iii) If a E Gal(E/F), then a(a) = wia for some i, by Lemma 5.1. Define
cp: Gal(E/F) -+ Zp by (p(a) = [i], the congruence class of i mod p. It is easy to
check that cp is a homomorphism; it is an injection, by Lemma 5.2. Since f (x)
is irreducible, by hypothesis, Lemma 5.4 shows that Gal(E/F) # 1. Therefore
cp is a surjection, for Z, has no proper subgroups.
Theorem 5.7. Let f (x) E F [x], let E be the splitting field of f (x) over F, and
assume that f (x) has no repeated roots in E (i.e., f (x) has no factor of the form
(x - a)2 in E[x]). Then f(x) is irreducible if and only if Cal(E/F) acts transi-
tively on the set X of all the roots off (x).
Remark. It can be shown that if F has characteristic 0 or if F is finite, then
every irreducible polynomial in F[x] has no repeated roots.
96 5. Normal Series
Proof. Note first that Lemma 5.1 shows that Gal(E/F) does act on X. If f(x)
is irreducible, then Lemma 5.4 shows that Gal(E/F) acts transitively on X.
Conversely, assume that there is a factorization f(x) = g(x)h(x) in F[x]. In
n
E[x], g(x) = fl (x - a,) and h(x) = (x - P,); since f(x) has no repeated
roots, a, # Pj for all i, j. But Gal(E/F) acts transitively on the roots of f(x), so
there exists a E Gal(E/F) with a(a,) = Ply and this contradicts Lemma 5.1.
It is easy to see that if a , is a root of f(x), then the stabilizer of a, is
Gal(E/F(a,)) 5 Gal(E/F), and Gal(E/F(a,)) is the Galois group of
f (x)/(x - a,) over F(a,). Thus, f (x)/(x - a,) is irreducible (over F(a,)) if and
only if Gal(E/F(a,)) acts transitively on the remaining roots. We shall return
to this observation in a later chapter (Example 9.3) when we discuss multiple
transitivity.
Theorem 5.8. Let F c K c E be fields, where K and E are splitting fields of
polynomials over F. Then Gal(E/K) a Gal(E/F) and
Gal ( E lF)/Gal(Elf<) E Gal (l</.F).
ProaJ The function @: Gal(E/F) -+Gal(K/F), given by @(a)= a]K, is well
defined (by Exercise 5.4, for I<is a splitting field), and it is easily seen to be a
homomorphism. The kernel of CO consists of all those automorphisms which
fix K; that is, ker @ = Gal(E/l<), and so this subgroup is normal. We claim
that 4D is a surjection. If ;l E Gal(K/F), that is, A is an automorphism of I<
which fixes F, then A can be extended to an automorphism A* of E (by
Lemma 5.5, for E is a splitting field). Therefore, A* E Gal(E/F) and @(A*)=
/2*II< = 2. The first isomorphism theorem completes the proof. B
We summarize this discussion in the next theorem.
Theorem 5.9 (GsRois, 1831). Let f (x) E FCx] be a polynomial of degree n, let F
contain all pth roots of unity for every prime p dividing n!, and let E be the
splitting field of f(x) over F. If f(x) is solvable by radicals, then there exist
subgroups Gi 5 G = Cal(E/F) such that:
(i) G = Go > G, > . - . > G , = 1;
(ii) Gi+, a Gi for all i; and
(iii) Gi/Gi+, is cyclic of prirne order for all i.
ProoJ Since f (x) is solvable by radicals, there are subfields F = I<, c K , c
- c Kt with E c Kt and with Ki+, = Ki(Pi+,),where pi+, E Ki+, and some
power of Pi+, lies in Ki. By Exercise 5.5, we may assume that some prime
power of Pi+, is in K,+,. If we define Hi = Gal(K,/K,), then (i) is obvious.
Since F contains roots of unity, Theorem 5.6 shows that Ki+, is a splitting
Some Galois Theory 97
field over K,; moreover, it can be shown that there is such a tower of fields in
which K , is a splitting field of some polynomial over F. Theorem 5.8 now
applies to show, for all i, that Hi+, = Gal(K,/Ki+,) a Gal(K,/K,) = Hi and
Hi+,/Hi r Gal(Ki+,/Ki); this last group is isomorphic to Z,,by Theorem
5.6, E3
Remarks. 1. We have shown only that Gal(I<,/F) satisfies the conclusions
of the theorem. By Theorem 5.8, Gal(K,/E)aGal(K,/F) and
Gal(K,/F)/Gal(I<,/E) r Gal(E/F); that is, Gal(E/F) is a quotient of
Gal(K,/F). Theorem 5.15 will show that if a group satisfies the conclusions of
Theorem 5.9, then so does any quotient group of it.
2. The hypothesis that F contains various roots of unity can be eliminated.
3. If F has characteristic 0, then the converse of this theorem is also true; it,
too, was proved by Galois (1831).
Definition. A not*malseries of a group G is a sequence of subgroups
in which Gi+, a G, for all i. The factor groaps of this normal series are the
groups Gi/Gi+, for i = 0, 1, ..., n - 1; the length of the normal series is the
number of strict inclusions; that is, the length is the number of nontrivial
factor groups.
Note that the factor groups are the only quotient groups that can always
be formed from a normal series, for we saw in Exercise 2.45 that normality
may not be transitive.
D~finition.~A finite group G is soloable if it has a normal series whose factor
groups are cyclic of prime order.
In this terminology, Theorem 5.9 and its converse say that a polynomial is
solvable by radicals if and only if its Galois group is a solvable group. P.
Ruffini (1799) and N.H. Abel (1824) proved the nonexistence of a formula
(analogous to the quadratic, cubic, and quartic formulas) for finding the roots
of an arbitrary quintic, ending nearly three centuries of searching for a gen-
eralization of the work of Scipione, Tartaglia, Cardano, and Lodovici (actu-
ally, neither the proof of Ruffini nor that of Abel is correct in all details, but
Abel's proof was accepted by his contemporaries and Ruffini's was not). In
modern language, they proved that the Galois group of the general quintic is
S, (nowadays, we know that any irreducible quintic having exactly three
real roots [e.g., f(x) = x5 - 4x + 21 will serve); since S, is not a solvable
group, as we shall soon see, f(x) is not solvable by radicals. In 1829, Abel
proved that a polynomial whose Galois group is commutative is solvable by
We shall give another, equivalent, definition of solvable groups later in this chapter.
 5. Normal Series The Jordan-Holder Theorem 99
98

5.7. A normal series is a composition series if and only if its factor groups are either
radicals; this is why abelian groups are so called (of course, Galois groups simple or trivial.
had not yet been invented). Galois7stheorem generalizes Abel's theorem, for
every abelian group is solvable. Expositions of Galois Theory can be found 5.8. Every finite group has a composition series.
in the books of Artin (1955), Birkhoff and Mac Lane (1977), Jacobson (1974), 5.9. (i) An abelian group has a composition series if and only if it is finite.
Kaplansky (1972), Rotman (1990), and van der Waerden (1948) listed in the (ii) Give an example of an infinite group which has a composition series.
Bibliography.
5.10. If G is a finite group having a normal series with factor groups H,,, H,, ...,H,,,
n
then IGl = 1 Hil.

The Jordan-Hiilder Theorem
Not only does Galois Theory enrich the study of polynomials and fields, but
it also contributes a new idea, namely, normal series, to the study of groups.
Let us give a brief review of what we have learned so far. Our first results
arose from examining a single subgroup via Lagrange's theorem. The second,
deeper, results arose from examining properties of a family of subgroups via
the Sylow theorems; this family of subgroups consists of conjugates of a
single subgroup, and so all of them have the same order. Normal series will
give results by allowing us to examine a family of subgroups of different
orders, thus providing an opening wedge for an inductive proof.
Deffwittisn. A normal series
G=Ho>H12...2H,=1
is a rejinemenlt of a normal series
G=Go>Gl~...~Gl,=l
if Go, GI, . .., G, is a subsequence of H,, HI, ... , HI,.
A refinement is thus a normal series containing each of the terms of the
original series.
Exercise 5.10 shows that some information about G can be gleaned from a
normal series. Let us now consider two composition series of G = (x) z il,,
(normality is automatic because G is abelian):
G 2 (x5) 2 (xlO) 2 1,
G > (x2) 2 (x6) )_ 1.
The factor groups of the first normal series are G/(xS) z B5, (x5)/(x10) r
B2, and (xlO)/l 2 (xlO) E Z,; the factor groups of the second normal series
are G/(x2) z b2, (x2)/<x6) Y b3, and (x6) z Z,.In this case, both com-
position series have the same length and the factor groups can be "paired
isomorphically" after rearranging them. We give a name to this phenomenon,
Defia'atnitiosa.Two normal series of a group G are eqldikralent if there is a bijection
between their factor groups such that corresponding factor groups are
isomorphic.
Of course, equivalent normal series have the same length. The two compo-
sition series for Z,, displayed above are equivalent; the amazing fact is that
this is true for every (possibly infinite) group that has a composition series!
The next technical result, a generalization of the second isomorphism theo-
rem, will be used in proving this.

DeGnition. A composifiorrseries is a normal series Lemma 5.10 (Zlpssenbalps Lemma, 1934). Let A a A* and B a B* be four
subgroups of a grozip G. Then
G=Go2G1>.-.>Gl,=l
A(A* n B) a A(A* n B*),
in which, for all i, either G,,, is a maximal normal subgroup of,Gi or Gi+l =
B(B* n A) a B(B* n A*),
and there is an isomorphism
Every refinement of a composition series is also a coinposition series; it can
only repeat some of the original terms. _
A(A* n B*) B(B* n A*)
-
A(A* n 3 ) - B(B* n A) '
EXERCISES
Remark. Notice that the hypothesis and conclusion are unchanged by trans-
5.6. A normal series is a composition series if and only if it has maximal length; that
is, every refinement of it has the same length. posing the symbols A and B.
100 5. Normal Series
Proof. Since A a A*, we have A 4 A* n B* and so A nB* = A n(A* n B*) a
A* n B*; similarly, A* n B a A* n BI. It follows from Lemma 2.25 and
Exercise 2.38 that D = (A* n B)(A n B*) is a normal subgroup of A* n B*.
If x E B(B* n A*), then x = bc for b E B and c E B* n A*. Define
f : B(B* n A) -+ (A* n B*)/D by f ( x )= f (bc) = cD. To see that f is well de-
fined, assume that x = bc = b'c', where b' E B and c' E B* n A; then c'c-' =
b'-'b E (B* n A*) n B = B n A* 5 D. It is routine to check that f is a surjec-
tive homomorphism with kernel B(B* n A). The first isomorphism theorem
gives B(B* n A) a B(B* n A*) and
B(B* n A*) N- B* n A*
B(B* n A) - D '
Transposing the symbols A and B gives A(A* n B) a A(A* n B*) and an
isomorphism of the corresponditlg quotient group to (B* n A*)/D. It follows
that the two quotient groups in the statement are isomorphic.
Theorem 5.11 (Schreier ReBnernent Theorem, 1923). Every two normal series
of an arbitrary group G have refinements that are equivalent.
Pros$ Let
G=GoZG,Z...2G,,=1
and
G = W o Z H 1 2 . . . ~ H , , =1
be normal series. Insert a "copy" of the second series between each Gi and
+,
G, . More precisely, define G,, = Gi+,(Gi n Hj) for all 0 5 j 5 m. Thus
Notice that G,,, = Gi because No = G and that G,,,,, = Gi+, because H , = 1.
Moreover, setting A = Gi+,, A* = Gi, B = Hj+,, and B* = Hj in the Zas-
senhaus lemma shows that G,,j+l a G,,j. It follows that the sequence
is a refinement of the first normal series (with mn terms). Similarly, if Hi, is
defined to be Hj+,(Hjn Gi), then Hi, 2 Hi+,, and
is a refinement of the second normal series (with mn terms). Finally, the
function pairing G,,j/Gi,j+,with Hi,j/Hi+,, is a bijection, and the Zassenhaus
lemma (with A = Gi+,, A* = Gi, B = Hj+,, and B* = H,) shows that corre-
sponding factor groups are isomorphic. Therefore, the two refinements are
equivalent. ! B
Theorem 5.12 (Jordan-Walder). Every two composition series of a grotip G are
equivalent.
! The Jordan-Holder Theorem 101
1 Remark. C. Jordan (1868) proved that the orders of the composition factors
of a finite group depend only on G; 0.Holder (1889) proved that the compo-
sition factors themselves, to isomorphism, depend onIy on G.
PBrooJ Composition series are normal series, so that every two composition
series of G have equivalent refinements. But a composition series is a normal
series of maximal length; a refinement of it merely repeats several of its terms,
and so its new factor groups have order 1. Therefore, two composition series
of G are equivalent. @I
DeEnition. If G has a composition series, then the factor groups of this series
are called the co~~tposigionf~ctors
of G.
One should regard the Jordan-Holder theorem as a unique factorization
theorem.
Coro6[Bary 5.13 (Fa~damental he or em of Arithmetic). The primes and their
multiplicities occurring in the factorization of an integer ;rz > 2 are determined
by n.
PYOOJLet n = pip,. . .p,, where the pi are (not necessarily distinct) primes. If
G = (x) is cyclic of order n, then
> (xP1) ... 2 ( x P ~ - . . P t - l )
i G = (x) 2 (xPlP2) 2
is a normal series. The factor groups have prime orders p,, p,, .. . ,p,, respec-
21
tively, so this is a composition series. The Jordan-Holder theorem now shows
that these numbers depend on n alone.
Recall that a finite group G is solvable if it has a normal series (necessarily
a composition series) with all its factor groups cyclic of prime order. Thus,
one sees that a particular group is not solvable by checking whether every
composition series has all its factor groups cyclic. With Jordan-Holder, one
need look only at a single composition series of G. For the next result, how-
ever, the JordanLHolder theorem is not needed, for we have essentially seen
(in Exercise 3.21) that S, has a unique composition series when n 2 5.
1 Theorem 5.14. If n 2 5, then S, is not solvable.
PmoJ Since A,, is a simple group for n 2 5, the normal series S,, 2 A, 2 1 is
a composition series; its composition factors are Z,and A,,, and hence S,, is
not solvable. El
I 5.1 1. Let G and H be finite groups. If there are normal series of G and of H having the
same set of factor groups, then G and H have the same composition factors.
102 5. Normal Series
5.12. (i) Exhibit a composition series for S,.
(ii) Show that S,, is solvable for n 5 4.
5.13. Assume that G = H, x . -. x H,, = K x , x K,., where each Hiand K j is sirn-
ple. Prove that in = n and there is a permutation n of { 1,2, . .., a) with KKii,1
H ifor all i. (Hint. Construct composition series for G.)
5.14. Prove that the dihedral groups Dz,,
are solvable.
Solvable Groups
Even though solvable groups arose in the context of Galois Theory, they
comprise a class of groups of purely group-theoretic interest as well. Let us
now give another definition of solvability which is more convenient to work
with and which is easily seen to agree with the previous definition for finite
groups.
Definition. A solvable series of a group G is a normal series all of whose factor
groups are abelian. A group G is solvable if it has a solvable series.
We are now going to manufacture solvable groups. Afterwards, we will
give another characterization of solvability which will give new proofs of
these results.
Theorem 5.15. Every strbgroup H of a solvable groltp G is itself solvable.
Proof. If G = Go 2 G, 2 .-. 2 G, = 1 is a solvable series, consider the series
H = Ho 2 (H n G , ) 2 ... 2 (H n G.) = 1. This is a normal series of H,
for the second isomorphism theorem gives H n G,+, = (H n Gi) n Gi+' a
H n Gi for all i. Now (H n Gi)/(H n Gi+,)z G,,,(H n G,)/G,+, 5 Gi/Gi+,;as
G,/G,+, is abelian, so is any of its subgroups. Therefore, H has a solvable
series.
Theorem 5.16. Every quotient of a solvable group is solvable.
Proof. It suffrces to prove that if G is a solvable group and f : G -+ H is a
surjection, then H is a solvable group. If
is a solvable series, then
H = f(Go) 2 f(G1) 2 ... 2 f(Gt) = 1
is a normal series for H: if f (xi+,)E f (Gi+,) and f(xi) E f(Gi), then
f (xi)f (xi+,)f (xi)-' = f ( ~ , x x;'
~ +) E~ f (G,), because G,,, 4 G,, and so
f (G,,,) 4 f (G,) for every i. The map (p: G, -+f (Gi)/f (Gi+,), defined by
Solvable Groups 103
xi H f(xi)f(Gi+,), is a surjection, for it is the composite of the surjections
Gi 4 f (G,) and the natural map f (G,) -+ f (G,)/f (Gii1). Since Gi+, Iker rp, this
map 9 induces a surjection Gi/Gi+, -+ f(Gi)/f(Git1), namely, xiGitl H
f (xi)f (Gi+,). Now f (Gi)/f (Giil) is a quotient of the abelian group Gi/Gi,, ,
and so it is abelian. Therefore, H has a solvable series, and hence it is a
solvable group. D l
In the proof of Theorem 5.9, we showed only that Gal(K,/F) is solvable,
whereas we wanted to prove that Gal(E/F) is solvable. But Gal(E/F) is a
quotient of Gal(K,/F), and so Theorem 5.16 completes the proof of Theorem
5.9.
Theorem 5.17. I f H a G and if both H and G/H are solvable, then G is solv-
able.
be a solvable series. By the correspondence theorem, we can construct a
sequence looking like the beginning of a solvable series from 6 to H; that is,
there are subgroups K i(with H I:K iand Ki/H u K r ) such that
Ki+' a Ki, and Ki/lCi+, ( ~ l ( ~ ~ / l C i ris, ,abelian.
) Since H is solvable, it has a
solvable series; if we splice these two sequences together at H, we obtain a
solvable series for G. !3!
Corollary 5.18. If H and K are solvable groups, then H x K is solvable.
P r o d If G = N x K, then W a G and G/H r K.
Theorem 5.18. Every finite p-group G is solvable.
Proo/. The proof is by induction on IGJ.By Theorem 4.4, IZ(G)I # 1. There-
fore, G/Z(G) is a p-group of order < /GI, and hence it is solvable, by induc-
tion. Since every abelian group is solvabfe, Z ( G ) is solvable. Therefore, G is
solvable, by Theorem 5.17.
Here is an alternative proof. The composition factors of G must be simple
p-groups. But we remarked (after proving that finite p-groups have nontrivial
centers) that there are no finite simple p-groups of order greater than p. It
follows that G is solvable.
Another approach to solvability is with commutator subgroups; that we
are dealing with abelian quotient groups suggests this approach at once.
104 5. Normal Series
1 Solvable Groups
Definition. The higher commutator subyroups of G are defined inductively:
) G.
~ ( 0= @if')
Y
= ~(0';
that is, G"+') is the commutator subgroup of G"). The series
G=~'02 ' . ..
' G'" 2 ~ ' 2 2
is called the derived series of G.
To see that the higher commutator subgroups are normal subgroups of G,
it is convenient to introduce a new kind of subgroup.
Definition. An an~tomorpkismof a group G is an isomorphism cp: G -, G. A
subgroup H of G is called charaeteuistic in G, denoted by H char G, if q(H) =
H for every automorphism q of G.
If q(H) 5 H for every autornorphism p, then H char G: since both q and
q-' are automorphisms of G, one has q(H) 5 H and qV'(H) 5 H; the latter
gives the reverse inclusion H = qcp-'(H) 5 q(H) and so q(H) = H.
For each a E G, conjugation by a (i.e., x I--+ axa-') is an automorphism of
G; it follows at once that every characteristic subgroup is a normal subgroup
(but there are normal subgroups which are not characteristic; see Exercise
5.28 below.)
(i) If H char K and I< char G, then H char G.
(ii) If H char K and K a G, then H a G.
Proof. (i) If q is an automorphism of G, then q(K) = K, and so the restric-
tion ql K: I<+ K is an automorphism of K; since H char K, it follows that
= (qlK)(H) = H-
(ii) Let a E G and let q :G --+ G be conjugation by a. Since K a G, p jK is an
automorphism of K; since H char K , (cplK)(H) 5 H. This says that if h E H,
then aha-' = q(lz) E H. El
Theorem 5.21. For every group G, the higher commutator subgroups are char-
acteristic, hence normal subgroups.
Proo$ The proof is by induction on i 2 1. Recall that the commutator sub-
group G' = G(') is generated by all commutators; that is, by all elements
of the form aba-'b-'. If q is an automorphism of C, then q(aba-'b-') =
q(a)cp(b)cp(a)-'q(b)-' is also a commutator, and so q(G') IG'. For the in-
ductive step, we have just shown that G"") char Gti);since G") char G, by
induction, Lemma 5.20(i) shows that G(i+l)is characteristic in G.
It follows that the derived series of a group G is a normal series if it ends
at 1. The next result shows that if G is solvable, then the derived series
descends faster than any other solvable series.
Lemma 5-22. If G = Go 2 GI 2 2 GI, = 1 is a solvable series, then Gi 2
G(') for all i.
Proof. The proof is by induction on i 2 0. If i = 0, then Go = G = G(O).For
the inductive step, Theorem 2.23 gives Gi+' 2 GI, since Gi/Gi+, is abelian.
The inductive hypothesis gives Gi 2 G"), so that GI 2 G")' = G(itl). There-
fore, Gi+' 2 G('+'), as desired. El
Theorem 5.23. A group G is solvable if and only if G(")= 1 for some n.
Proof. Let G = Go 2 GI 2 . . . 2 GI, = 1 be a solvable series. By the lemma,
Gi 2 G(') for all i. In particular, 1 = G,, 2 G'"), and so G(")= 1.
Conversely, if G") = 1, then the derived series is a normal series; since it
has abelian factor groups, it is a solvable series for G.
Thus, the derived series of G is a normal series if and only if G is a solvable
group.
The following new proofs of Theorems 5.15,5.16, and 5.17 should be com-
pleted by the reader; they are based on the new criterion for solvability just
proved.
Theorem 5.15. If H 5 G, then H") 5 G ( ~ for
) all i ; hence Gtr')= 1 implies
H(")= 1, so that every subgroup of a solvable group is solvable.
Theorem 5.16. Iff: G -+ K is surjective, then f(G")) = f ( ~ ) ( 'for
) all i. There-
fore, G(")= 1 implies 1 = f(G(")) = f(G)("),so that every quotient of a solvable
group is solvable.
Theorem 5.17. Let H 4 G, let K = G/H, and let f : G -+ K be the natural
map. Then K(") = 1 implies f (G'")) = 1, and hence G(")5 H. If H(") = 1, then
(G("))(")5 H(") = 1. Finally, one proves by induction on n 2 1 that G('"+")5
(G("))("')for all myand so G is solvable.
Definition. A normal subgroup H of a group G is a minimal rormal subgroup
if H # 1 and there is no normal subgroup I< of G with 1 iK i H.
Minimal normal subgroups always exist in nontrivial finite groups.
Theorem 5.24. If G is a finite solvable group, then every minimal normal sub-
group is elementary abelian.
 5. Normal Series
Proof. Let V be a minimal normal subgroup of G. Now Lemma 5.20(ii) says
that if H char V , then H a G; since V is minimal, either H = 1 or H = V. In
particular, V' char so that either V' = 1 or V' = V. Since G is solvable, so
is its subgroup I/. It follows from Theorem 5.23 that V' = V cannot occur
here, so that V' = 1 and so V is abelian. Since V is abelian, a Sylow p-
subgroup P of V, for any prime p, is characteristic in V; therefore, V is an
abelian p-group. But {x E V: x P = 1) char V, and so V is elementary abelian.
Corollary 5.25. If V is a minimal normal szrbgroup of a finite solvable group G,
then G acts on V as a group of linear transformations.
Proof. By the theorem, V is an elementary p-group; by Exercise 2.78, V is a
vector space over Z,and every homomorphism on V is a linear transforma-
tion. Define a homomorphism G -' GL(V) by a I-+ q,, where qa(v) = ava-'
for all v E V (normality of V shows that rp,(v) E V). Moreover, each q, is an
injection, being the restriction of an automorphism (namely, conjugation),
and every injection on a finite-dimensional vector space is a surjection; hence
each (o, is nonsingular. b3
What are the groups G whose only characteristic subgroups are G and I?
Such groups are sometimes called characteristically simple.
Theorem 5.26. A finite group G with no characteristic subgrozrps other than G
and 1 is either simple or a direct product of isomorphic simple groups.
PuooJ Choose a minimal normal subgroup H of G whose order is minimal
among all nontrivial normal subgroups. Write H = HI, and consider all sub-
groups of G of the form H, x H, x ... x H,,, where n > 1, Hi a G, and Hi 2
H. Let M be such a subgroup of largest possible order. We show that M = G
by showing that M char G; to see this, it suffices to show that rp(Hi) IM for
every i and every automorphism rp of G. Of course, q(Hi) E H = H,. We
show that rp(Hi) 4 G. If a E G, then a = q(b) for some b E G, and aq(Hi)a-' =
q(b)q(Hi)q(b)-l = q(aH,a-') 5 (!(Hi), because Hi a G. If q(Hi) M, then
q(Hi) n M $ q(Hi) and lq(Hi) n MI < Iq(Hi)l = IHl. But q(Hi) n M 4G,
and so the minimality of (HI shows that ?(Hi) n M = 1. The subgroup
(M, q(Hi)) = M x q(Hi) is a subgroup of the same type as M but of larger
order, a contradiction. We conclude that A4 char G, and so M = G. Finally,
H = H, must be simple: if N is a nontrivial normal subgroup of H, then N is
a normal subgroup of M = H, x H, x ..- x H,, = G, and this contradicts
the minimal choice of H. EZ!
Corollary 5.27. A minimal normal subgroup H of a finite group G is either
simple or n direct product of isomorphic simple subgroups.
Solvable Groups
Proof. If N char H, then N a G, by Lemma 5.20(ii), so that either N = 1 or
N = H (because H is a minimal normal subgroup). Therefore, H has no
proper characteristic subgroups, and Theorem 5.26 gives the result.
Ei
This last corollary gives another proof of Theorem 5.24, for a finite simple
group G is solvable if and only if it is cyclic of prime order.
5.15. Every refinement of a solvable series is a solvable series.
5.16. A solvable group having a composition series must be finite.
5.17. If G has a composition series and if H a G, then G has a composition series one
of whose terms is H.
5.18. (i) If S and Tare solvable subgroups of G with S a G, then ST is solvable.
(ii) Every finite group G has a unique maximal normal solvable subgroup Y(G);
moreover, G/Y(G) has no nontrivial normal solvable subgroups.
5.19. (i) If p and q are primes, then every group of order pZqis solvable.
(ii) If p and q are primes with p < q, then every group of order pq" is solvable.
(Hint. Use Sylow's theorems.)
5.20. (i) Show that S, has no series
S4= G o > GI > . . a > G,= 1
such that all factor groups are cyclic and each Gi is a normal subgroup of G.
(A group G with such a series is called supersoluable, and we now see that
not every solvable group is supersolvable.)
(ii) Show that every finite p-group is supersolvable.
5.21. If G is a group with IG/ < 60, then G is solvable. (Hint. Use Exercise 4.36.)
5.22. Burnside proved (using Representation Theory) that the number of elements in
a conjugacy class of a finite simple group can never be a prime power larger
than 1. Use this fact to prove Bumside's fhcorem: If p and q are primes, then
every group of order p"qn is solvable.
5.23. Prove that the following two statements are equivalent:
(i) every group of odd order is solvable;
jii) every finite simple group has even order.
(In 1963, Feit and Thompson proved (i); the original proof is 274 pages long.)
5.24. Let G be a finite group of order > 1. If G is solvable, then G contains a nontrivial
normal abelian subgroup; if G is not solvable, then G contains a nontrivial
normal subgroup H with H = H'.
5.25. For every group G, its center Z(G) is characteristic in G.
5.26. If H a G and (]HI, [G : HI)= 1, then H char G.
5.27. If H char G and H K 5 G, then K/H char G/H implies K char G.
 5. Normal Series
5.28. Give an example of a group G containing a normal subgroup that is not a
characteristic subgroup. (Hiizt. Let G be abelian.)
Definition. A subgroup H of G isfully invariant if cp(H) 5 H for every homo-
morphism cp: G -+ G.
Of course, every fully invariant subgroup is characteristic and hence
normal.
/
'
5.29. Prove that the higher commutator subgroups are fully invariant.
5.30. Show that Z(G) may not be fully invariant. (Hint: Let G = Z,x S,.) (See Exer-
cise 5.25.)
Two Theorems of P.Hall
The main results of this section are generalizations of the Sylow theorems
that hold for (and, in fact, characterize) finite solvable groups.
Theorem 5.28 (P.Hall, 1928). If G is a solvable group of order ab, where
(a, b) = 1, then G contains a subgroup of order a. Moreover, any two subgroups
of order a are conjugate.
Proof. The proof is by induction on IGI; as usual, the base step is trivially
true.
Case 1. G contains a normal subgroup H of order a'b', where a'la, b'lb, and
b' < b.
Existence. In this case, G/H is a solvable group of order (a/al)(b/b'), which is
strictly less than ab; by induction, G/H has a subgroup A/H of order a/al.
Now A has order (a/al)1 HI = ab' < ab; since A is solvable, it has a subgroup
of order a, as desired.
Conjugacy. Let A and A' be subgroups of G of order a. Let us compute
k = IAH1. Since AH 5 G, Lagrange's theorem gives 1 AH1 = ap, where ala
and /3l b. Since (a, b) = 1 and A 5 AH, we have slap, so that, a = a; since
H 5 AH, we have a'b'lap, so that b11/3.But the second isomorphism theorem
(actually, the product formula) gives k(aa1b',so that p(bl.We conclude that
(AH1= k = ab'. A similar calculation shows that IA'HI = ab' as well. Thus,
AHIH and AIH/H are subgroups of G/fI of order a/al. As / G/HI = (a/al)(b/b'),
these subgroups are conjugate, by induction, say by x H E G/H. It is quickly
checked that xAHx-' = A'H. Therefore, xAx-' and A' are subgroups of A'H
of order a, and so they are conjugate, by induction. This completes Case 1.
If there is some proper normal subgroup of G whose order is not divisible
Two Theorems of P. Hall 109
by b, then the theorem has been proved. We may, therefore, assume that
bl/Hl for every proper normal subgroup H. If H is a minimal normal sub-
group, however, then Theorem 5.24 says that H is an (elementary) abelian
p-group for some prime p. It may thus be assumed that b = pm,so that H is
a Sylow p-subgroup of G. Normality of H forces H to be unique (for all Sylow
p-subgroups are conjugate). The problem has now been reduced to the fol-
lowing case.
Case 2. I GI = ap", where playG has a normal abelian Sylow p-subgroup H,
and H is the unique minimal normal subgroup in G.
Remark. We shall complete the proof here, but we note that this case follows
immediately from the Schur-Zassenhaus lemma to be proved in Chapter 7.
Existence. The group G/H is a solvable group of order a. If K/H is a minimal
normal subgroup of G/H, then JI<lHJ = q" for some prime q # p, and so
= pmqn;if Q is a Sylow q-subgroup of K, then K = HQ. Let N* = NG(Q)
/I</
and let N = N* n K = NK(Q).We claim that IN*/ = a.
The Frattini argument (Theorem 4.18) gives G = KN*. Since
we have IN* 1 = I GI 1 N 111K 1. But I<= HQ and Q INIK gives K = HN;
='[HW = IHI[NI/IH n Nl, so that
hence /I</
Hence 1 N* 1 = a if H n N = 1. We show that H n N = 1 in two stages:
(i) H n N < Z(K); and (ii) Z(I<)= 1.
(i) Let x E H n N. Every k E I( = HQ has the form 12 = hs for h E H and
s E Q. Now x commutes with h, for H is abelian, and so it suffices to show
that x commutes with s. But (xsx-')s-' E Q, because x normalizes Q, and
x(sx-Is-') E H, because H is normal; therefore, xsx-Is-' E Q n H = 1.
(ii) By Lemma 5.20(ii), Z(IC) a G. If Z(K) f 1, then it contains a minimal
subgroup which must be a minimal normal subgroup of G. Hence H 5 Z(K),
for H is the unique minimal normal subgroup of G. But since K = HQ, it
follows that Q char K. Thus Q a G, by Lemma 5.20(ii), and so H s Q, a
cont~adiction.Therefore, Z(K) = 1, H n N = 1, and I N*l = a.
Conjugacy. We keep the notation of the existence proof. Recall that N* =
NG(Q)has order a; let A be another subgroup of G of order a. Since IAI<Jis
divisible by a and by IK/ = p"qn, it follows that IAKI = ab = IGI, AK = G,
I G/K = AK/IC r A/(A n K),
and I A n KI = qn.By the Sylow theorem, A n K is conjugate to Q. As conju-
gate subgroups have conjugate normalizers, by Exercise 3.9(i), N* = NG(Q)
is conjugate to NG(An K), and so a = 1 N* 1 = I NG(An K) 1. Since A n K a A,

5. Normal Series
110
we have A 5 N,(A n K) and so A = N G ( An K) (both have order a). There-
fore, A is conjugate to N*. Dl
The following definition is made in tribute to this theorem.
Definition. If G is a finite group, then a Hall subgroup H of G is a subgroup
whose order and index are relatively prime; that is, ((HI,[G : HI) = 1.
It is sometimes convenient to display the prime divisors of the order of a
group, If n is a set of primes, then a n-number is an integer n all of whose
prime factors lie in n; the complement of n is denoted by n', and so a
a'-number is an integer n none of whose prime factors lie in n.
Definition. If n is a set of primes, then a group G is a n-group if the order of
each of its elements is a n-number; a group is a nt-groupif the order of each of
its elements is a n'-number.
Of course if a is a n-number and b is a nl-number, then a and b are
relatively prime. If n consists of a single prime p, then n-groups are just
p-groups, while pt-groups have no elements of order a power of y. It follows
from (Sylow's) Theorem 4.14 that every Sylow p-subgroup in a finite group is
a Hall p-subgroup (Hall's theorem says that Hall n-subgroups always exist in
finite solvable groups). In contrast to Sylow p-subgroups, however, Hall n-
subgroups (with lnl 2 2) of a group G need not exist. For example, let G = A,
and n = { 3 , 5 ) ; since lA,l = 60, a Hall n-subgroup would have index 4 and
order 15, and Corollary 3.16 shows that no such subgroup exists.
Definition. If p is a prime, and G is a finite group of order up", where a is a
of-number,then ap-complement of G is a subgroup of order a.
Hall's theorem implies that a finite solvable group has a om^^^^^^^ lor
every prime p. ~f G is a group of order pmq",then G has a ~ - c o m ~ l e ~ ~ ~ ~ ,
namely, a sylow q-subgroup, and a q-complement, namely, a S ~ l o wPsub-
group. the coming proof of the converse of Theorem 5.28, we shall use
BuvnsiJeYs theorern (see Exercise 5.22): Every group of order pmq"is
Theorem 5-29(p. Wall, 1937). ~fG is a finite group having a ~ - c o mfor~ l ~ ~ ~ ~ ~ isomorphic and hence may not be conjugate.
every prime p, then G is ~olvable-
proof. We proceed by induction on /GI; assume, on the contrary, that there
are nonsolvable groups satisfying the hypotheses, and choose one such, say
G, of smallest order. If G has a nontrivial normal subgroup N, and if is any
Hall pt-subgroup of G, then checking orders shows that H n N is a Hal1
p'-subgroup of N and HN/N is a Hall p'-subgroup of GIN. Since both N and
have order smaller than 1 G1, it follows that both N and GIN are solvable*
~~t Theorem 5.17 now shows that G is solvable, a contradiction.
I
i Two Theorems of P. Hall
i
We may assume, therefore, that G is simple. Let I GI = pi1. ..p,'.., where
the pi are distinct primes and ei z 0 for all i. For each i, let Hi be a Hall
H3 n . .. n HI,, then [G : Dl = n;=,
pi-subgroup of G, so that [G : Hi] = pfi, and thus /Hi]= p,'i. If D =
p1fi,by Exercise 3.3 1(ii), and so IDl =
pilp;'. Now D is a solvable group, by Burnside's theorem. If N is a minimal
normal subgroup of D,then Theorem 5.24 says that N is elementary abelian;
for notation, assume that N is a pl-group. Exercise 3.31(ii) shows that
[G: D n H,] = n;=, pfi, so that 1 D n H21 = p;' and D n H2 is a Sylow pl-
subgroup of D. By Exercise 4.16, N 5 D n H2 and so N 5 H,. But, as above,
ID n H, I = pZ2, and comparison of orders gives G = H2(D n HI). If g E G,
then g = hd, where h E H, and d E D n H I ; if x E N, then gxg-I =
hdxd-'h-' = hyh" (where y = dxd-' E N, because N u D) and hyh-I E H,
(because N 5 H,). Therefore, NG 5 H,, where NGis the normal subgroup of
G generated by N. Since H, < G, N # 1 is a proper normal subgroup of G,
and this contradicts the assumption that G is simple. BI
This proof exhibits two recurring themes in finite group theory. Many
theorems having the form "If a group G has property P, then it also has
property Q" are proved by induction on JGJin the following style: assume
that G is a group of smallest possible order which has property P but not
property Q, and then obtain a contradiction. In R. Baer's suggestive phrase,
we assume that G is a "least criminal."
The other theme is the reduction of a problem to the special case when a
simple group is involved. Many theorems are proved by such a reduction,
a n d this is one reason why the classificaton of the finite simple groups is so
important.
I
EXERCISES
5'31. If is a finite(not necessarily solvable) group and H is a normal ~ ~subgroup,
1 1
then H char G.
5.32. (i) If G is a bite solvable group, then for every set of primes n, a maximal
z-subgroup is a Hall n-subgroup.
(ii) Let = i3, 51. Show that both z3and Z,are maximal n-subgroups of s,.
when G is not solvable, that maximal n-subgroups may not be
If = i2, 51, then a maximal ?i-subgroup of S, is not a ~ ~ n-subgroup,
1 1
(I-lint. Exercise 3.25.)
DefiniGon*If is a set of primes, define O,(G) to be the subgroup of
generated by all the normal n-subgroups of G.
5.33- that O,(G) is a characteristic subgroup of G.
5'34. Show that O,(G) is the intersection of all the maximal n-subgroups of G.
112 5. Normal Series
Central Series and Nilpotent Groups
The Sylow theorems show that knowledge of p-groups gives information
about arbitrary finite groups. Moreover, p-groups have a rich supply of nor-
mal subgroups, and this suggests that normal series might be a powerful tool
in their study. It turns out that the same methods giving theorems about
p-groups also apply to a larger class, the nilpotent groups, which may be
regarded as generalized p-groups.
Definition. If H, I<5 G, then
[H, K] = ([h, k]: h E H and k E K),
where [h, k] is the commutator hkh-'k-'.
An example was given, in Exercise 2.43, showing that the set of all commu-
tators need not be a subgroup; in order that [H, K] be a subgroup, therefore,
we must take the subgroup generated by the indicated commutators. It is
obvious that [H, K] = [K, HI, for [h, k]-' = [k, h]. The commutator sub-
group G' is equal to [G, GI and, more generally, the higher commutator
subgroup G('+') is equal to [G"), G")].
We say that a subgroup K rrormalizes H if K 5 NG(H);it is easy to see that
K normalizes H if and only if [H, K] < H.
Definition. If H 5 G, the centralizer of H in G is
CG(H)= {x E G: x commutes with every h E H);
that is, CG(H)= {x E G: [x, h] = 1 for all h E H).
We say that a subgroup K eentralizcs H if K 5 C,(H); it is easy to see that
K centralizes H if and only if [H, K] = 1.
If x, y E G and [x, y] E K, where K a G, then x and y "commute mod X";
that is, xKyK = yKxK in GIK.
Lemma 5.30.
(i) If K a G and K 5 H 5 G, then 111,GI IK $ and only if H/K 5 Z(G/K).
(ii) If H, K 5 G and f : G -+ L is a homomorphism, then f([H, K]) =
Cf(H)Y f (K)1.
Pro05 (i) If h E H and g E G, then hKgK = gKhK if and only if [h, g]11 = K
if and only if [h, g] E K.
(ii) Both sides are generated by all f([h, k]) = [f(h), f(k)]. g3
Definition. Define characteristic subgroups yi(G) of G by induction:
Central Series and Nilpotent Groups 113
Notice that y2(G) = [yl(G), GI = [G, GI = G' = G"). It is easy to check
that yi+,(G) 5 yi(G). Moreover, Lemma 5.30(i) shows that [yi(G), G] = Y,+~(G)
gives yi(G)/yi+,(G)5 Z(G/yi+,(G)).
Definition. The lower central series (or descending central series) of G is the
series
G = yl(G) 2 y2(G) 2 ...
(this need not be a normal series because it may not reach 1).
There is another series of interest.
Definition. The higher centers Ci(G) are the characteristic subgroups of G
defined by induction:
that is, if vi: G -+ G/Ci(G) is the natural map, then iit'(G) is the inverse image
of the center.
Of course, ['(G) = Z(G).
Definition. The rrpper central series (or ascending central series) of G is
When no confusion can occur, we may abbreviate C1(G) by Ci and yi(G)
by Yi.
Theorem 5.31. If G is a group, then there is an integer c with Cc(G)= G if and
only if y,+,(G) = 1. Moreover, in this case,
yi+,(G) 5 re-'(6) for all i.
Proof: Assume that Cc = G, and let us prove that the inclusion holds by
re.
induction on i. If i = 0, then y, = G = If yi+l < CC-i,then
the last inclusion following from Lemma 5.30. We have shown that the inclu-
sion always holds; in particular, if i = c, then ye+, I C0 = 1.
Assume that y,+, = 1, and let us prove that Y,+,-~ I [ j by induction on j
(this is the same inclusion as in the statement: set j = c - i). If j = 0, then
yCil = 1 = lo. If yc+,-j I_< jj, then the third isomorphism theorem gives a
surjective homomorphism G/y,,-,-j -+ ~ / j j Now. [Y,-~,GI = so that
Lemma 5.30 gives +l-j 5 Z(G/y,+,-j). By Exercise 3.10 [if A 5 Z(G)
 5. Normal Series
114
and f : G --+ H is surjective, then f ( A ) 5 Z(H)], we have
Therefore, 5 5 [jt', as desired. We have shown that the inclusion
always holds; in particular, if j = c, then G = y, 5 cc.
Bl
The following result reflects another relationship between these two series.
Theorem 5.32 (Schur). If G is a group with G/Z(G) finite, then G' is also finite.
Pvoof(0mstein). Let g,, . . ., g , be representatives of the cosets of Z(G) in G;
that is, each x E G has the form x = g,z for some i and some z E Z(G). For all
x, y E G, [x, y] = [giz, gjzl] = [g,, gj]. Hence, every commutator has the form
[gi, gj] for some i, j, so that G' has a finite number (< n2) of generators.
Each element g' E G' can be written as a word c, c,, where each ci is a
commutator (no exponents are needed, for [x, y]-' = [y, x]). It suffices to
Drove that if a factorization of g' is chosen so that t = t(gl) is minimal, then
t(gf) < n3 for all g' E G'.
We Drove first, by induction on r 2 1, that if a, b E G, then [a, b]' =
(aba-'b'-')" = (abj"(a"b-l)'u, where u is a product of I- - 1 commutators.
This is obvious when r = 1. Note, for the inductive step, that if x, y E G, then
xy = yxx-ly-lxy = yx[x-', y-'1; that is, xy = yxc for some commutator c.
Thus, if r > 1, then
for some commutator c, as desired.
Since yx = x-'(xy)x, we have (yx)" = x-'(xy)"x = (xy)", because [G : Z(G)]
= n implies (ab)" E Z(G). Therefore, (a7'b-')D = ((ba)-')" = ((bay)-' =((ab)")-'.
It follows that
'
[a, b]" is a product of n - 1 commutators.
(4
Now xyx = (xyx-')x2, so that two x's can be brought together at the ex-
pense of replacing y by a conjugate of y. Take an expressionb'ofan element
gf E Gf as a product of commutators c, ...c,, where t is minimal. If t 2 n3,
then there is some commutator c occurring m times, where rn > n (for there
are fewer than n2 distinct commutators). By our remark above, all such fac-
tors can be brought together to c'" at the harmless expense of replacing
commutators by conjugates (which are still commutators); that is, the num-
ber of commutator factors in the expression is unchanged. By (*), the length
of the minimal expression for g' is shortened, and this is a contradiction.
Therefore, t < n3, and so G' is finite.
Central Series and Nilpotent Groups
115
/
Definition. A group G is nilpotent4 if there is an integer c such that y,+,(G) =
1; the least such c is called the class of the nilpotent group G.
Theorem 5.31 shows, for nilpotent groups, that the lower and upper cen-
tral series are normal series of the same length.
A group is nilpotent of class 1 if and only if it is abelian. By Theorem 5.31,
a nilpotent group G of class 2 is described by y,(G) = G' 5 Z(G) = ['(G).
Every nonabelian group of order p3 is nilpotent of class 2, by Exercise 4.7.
Theweare 5.33. Every finite p-group is nilpotent.
Proo/: Recall Theorem 4.4. Every finite p-group has a nontrivial center. If, for
some i, we have ci(G) < G, then Z(G/ci(G)) # 1 and so ci(G) < citl(G). Since
G is finite, there must be an integer i with ci(G) = G; that is, G is nilpotent.
This theorem is false without the finiteness hypothesis, for there exist infi-
nite p-groups that are not nilpotent (see Exercise 5.45 below); indeed, there is
an example of McLain (1954) of an infinite p-group G with Z(G) = 1, with
G' = G (so that G is not even solvable), and with no characteristic subgroups
other than G and 1.
(i) Every nilpotent group G is solvable.
(ii) If G # 1 is nilpotent, then Z(G) # 1.
(iii) S3 is a solvable group that is not nilpotent.
Proof. (i) An easy induction shows that G(') 5 yi(G) for all i. It follows that if
yCtl(G) = 1, then G('+') = 1; that is, if G is nilpotent (of class 5 c), then G is
solvable (with derived length < c 1). +
(ii) Assume that G # 1 is nilpotent of class c, so that y,,,(G) = 1 and
y,(G) # 1. By Theorem 5.31, 1 # y,(G) I cl(G) = Z(G).
(iii) The group G = S3 is solvable and Z(S3) = 1. Bl
Theurem 5.35. Every subgroup H of a nilpotent group G is nilpotent. Moreover,
if G is nilpotent of class c, then N is nilpotent of class 1_< c.
Proof. It is easily proved by induction that H 5 G implies yi(H) 5 yi(G)for all
i. Therefore, y,+,(G) = 1 forces y,,, (N) = 1.
There is an analogue of the descending central series for Lie algebras, and Engel's theorem says
that if the descending central series of a Lie algebra L reaches 0, then L is isomorphic to a Lie
algebra whose elements are nilpotent matrices. This is the reason such Lie algebras are called
nilpotent, and the term for groups is taken from Lie algebras.
 5. Normal Series
Theorem 5.36. If G is nilpotent of class c and H a G, then G/H is nilpotent of
class 5 c. \
Proof: If f : G 4 L is a surjective homomorphism, then Lemma 5.30 gives
yi(L) 5 f(yi(G)) for all i. Therefore, y,+,(G) = 1 forces y,+,(L) = 1. The theo-
rem follows by taking f to be the natural map. E13
We have proved the analogues for nilpotent groups of Theorems 5.15 and
5.16; is the analogue of Theorem 5.17 true? If H a G and both H and G/H are
nilpotent, then is G nilpotent? The answer is "no": we have already seen that
S, is not nilpotent, but both A, z B , and &/A3 g Z2 are abelian, hence
nilpotent. A positive result of this type is due to P. Hall. If H a G, then we
know that H' a G; Hall proved that if both H and G/Hf are nilpotent, then
G is nilpotent (a much simpler positive result is in Exercise 5.38 below). The
analogue of Corollary 5.18 is true, however.
Theorem 5.37. If H and K are nilpotent, then their direct product H x !I is
nilpotent.
ProoJ: An easy induction shows that yi(H x K) 2 yi(H) x y,(K) for all i. If
A4 = maxjc, d), where y,+,(H) = 1 = yd+,(I<), then y,+,(H x K) = 1 and
H x K is nilpotent. El
Theorem 5.38. If G is nilpotent, then it satisfies the normalizer condition: if
H < G, then H < NG(H).
PYOO$There exists an integer i with yi+,(G) 5 H and yi(G) $ H (this is true
for any descending series of subgroups starting at G and ending at 1). Now
[y,, HI 2 [y,, GI = y,,, 5 H , so that y, normalizes H; that is, yi 5 NG(H).
Therefore, H is a proper subgroup of NG(H). B!
The converse is also true; it is &ercise 5.37 below.
Tileorem 5.39. A finite group G is nilpotent if and only if it is the direct product
of its Sylow subgroups.
Proof. If G is the direct product of its Sylow subgroups, then it'is nilpotent,
by Theorems 5.32 and 5.36.
For the converse, let P be a Sylow p-subgroup of G for some prime p. By
Exercise 4.11, NG(P) is equal to its own normalizer. On the other hand, if
NG(P)< G, then Theorem 5.38 shows that NG(P)is a proper subgroup of its
own normalizer. Therefore, NG(P)= G and P a 6. The result now follows
from Exercise 4.12.
Of course, in any group, every subgroup of prime index is a maximal
Central Series and Nilpotent Groups
117
subgroup. The converse is false in general (S, has a maximal subgroup of
index 4, as the reader should check), but it is true for nilpotent groups.
Theorem 5.40. If G is a rzilpotent group, then every maximal subgroup H is
normal and has prime index.
Proof. B y Theorem 5.38, H < NG(H);since H is maximal, NG(H)= G, and so
H a G. Exercise 2.58 now shows that G/H has prime order. iIi8
Theorem 5.41. Let G be a nilpotent group.
(i) If H is a nontrivial normal subgroup, then H n Z(G) # 1.
(ii) If A is a maximal abelian normal subgroup of G, then A = CG(A).
ProoJ (i) Since ['(G) = 1 and G = ['(G) for some c, there is an integer i for
which H n ['(G) # 1; let m be the minimal such i. Now [H n ["(G), G] 5
H n [['(G), G] 5 Hn("-'(G) = 1, because H o G, and this says that 1 #
H n ["'(G) 5 H n Z(G).
(ii) Since A is abelian, A 5 CG(A).For the reverse inclusion, assume that
g E CG(A)and g $ A. It is easy to see, for any subgroup H (of any group G)
and for all g E G, that gCG(H)g-' = CG(g-'Hg). Since A o G, it follows that
LJC,(A)~-' = CG(A)for all g E G, and so CG(A)a G. Therefore, CG(A)/Ais
a nontrivial normal subgroup of the nilpotent group CIA; by (i), there is
Ax E (Co(A)/A)n Z(G/A). The correspondence theorem gives ( A , x) a nor-
mal abelian subgroup of G strictly containing A, and this contradicts the
maximality of A. El
5.35. If G is nilpotent of class 2 and if a E G, then the function G -r G, defined by
x I-+ [a, x], is a homomorphism. Conclude, in this case, that C,(a) a G.
5.36. If G is nilpotent of class c, then G/Z(G) is nilpotent of class c - 1.
5.37. Show that the following conditions on a finite group G are equivalent:
(i) G is nilpotent;
(ii) G satisfies the normalizer condition;
(iii) Every maximal subgroup of G is normal.
5.38. If H 5 Z(G) and if G/H is nilpotent, then G is nilpotent.
Definition. A normal series
with each G, a G and G,/G,+, IZ(G/G,+,) is called a central series.
5.39. (i) If G is nilpotent, then both the upper and lower central series of G are
central series.
118 5. Normal Series

(ii) Prove that a group G is nilpotent if and only if it has a central series
G = GI 2 G, 2 2 G, = 1. Moreover, if G is nilpotent of class c, then
yi+,(G)< Gi+, < lc-'(G) for all i. / There are many commutator iderztities that are quite useful even though they
5.40. If G is a nilpotent group and H is a minimal normal subgroup of G, then are quite elementary.
H 5 Z(G).
Lemma 5.42. Let x, y E G and assume that both x and y commute with [x, y].
5.41. The dihedral group D,, is nilpotent if and only if n is a power of 2. Then:
5.42. Let G be a finite nilpotent group of order n. If mln, then G has a subgroup of (i) [x, Y]" = [xn, y] = [x, yn] for all n E Z;and
order m. for all n 2 0.
(ii) (xy)" = [y, X]"("-')/~X"~"
5.43. (i) If H and K are normal nilpotent subgroups of a finite group G, then HK is
a normal nilpotent subgroup. ProoJ (i) We first prove (i)for nonnegative n by induction on n 2 0; of course,
(ii) Every finite group G has a unique maximal normal nilpotent subgroup it is true when n = 0. For the inductive step, note that
F(G) (which is called the fitti~rgsubgrozsp of G).
(iii) Show that P ( G ) char G when G is finite.
= x [xn, y] yx-' y-l, by induction
5.44. (i) Show yi(UT(n, Z,))consists of all upper triangular matrices with 1's on the
main diagonal and 0's on the i - 1 superdiagonals just above the main =x ( x " y ~ - ~ ~ -y-l
~)~x-~
diagonal (Hint. If A is unitriangular, consider powers of A - E, where E is
the identity matrix.) = [xn+', y].
(ii) The group UT(n, Z,)of all n x n unitriangular matrices over Z, is a p-group Now x[x, y] = [x, y]x, by hypothesis, so that ~ y x - ' ~ -= ' yx-'y-lx;
that is nilpotent of class n - 1. that is, [x, y]-' = [y, x-']-' = [x-', y]. Therefore, if n 2 0, then [x, y]-n =
5.45. For each n 2 1, let GI,be a finite p-group of class 72. Define H to be the group of [x-', y]" = [x-", y], as desired.
all sequences (g,, g,, . ..), with g, E GI,for all n and with g,, = 1 for all large n; (ii) The second identity is also proved by induction on n 2 0.
that is, g, # 1 for only a finite number of g,. Show that H is an infinite p-group
(xy)"(xy)= [y, ~ ] ~ ' " - ~ " ~ x " y " x ~
which is not nilpotent.

5.46. If x, y E G, denote yxy-l by xY.If x, y, z E G, prove

(Recall that [x, y] = xyx-'y-I.)

5.47 (Jacobi identity). If x, y, z E G, denote [x, [y, 231 by [x, y, z]. Prove that
[x, y-1, z]"[y, z-I, xIZ[z, x-1, ylX= 1. Theorem 5.43. If G is a p-group having a unique subgroup of order p and more
5.48. (i) Let H, K, L be subgroups of G, and let [H, K, L] = ([h, k, I]: h E H, lc E I<, than one cyclic subgroup of index p, then G r Q, the quaternions.
1E L). Show that if [H, I<, L] = 1 = [K, L, HI, then [L, H, I<]= 1.
(ii) (Three sllabgroulps lemma). If N a G and [H, K, L] [I<, L, HI 5 N, then ProoJ If A is a subgroup of G of index p, then A a G, by Theorem 5.40. Thus,
[L, H, K] < N. if x E G, then Ax E G/A, a group of order p, and so xP E A. Let A = (a)
(iii) If H, K, and L are all normal subgroups of G, then [L, H, K] < and B = ( b ) be distinct subgroups of index p, and let D = A n B; note that
[H, K, L] [I<, L, HI. (Hint. Set N = [H, K, L] [K, L, HI.) D a G, for it is the intersection of normal subgroups. Our initial remarks
show that the subset
5.49. If G is a group with G = G', then G/Z(G) is centerless. (Hint. Use the three
subgroups lemma with H = C2(G)and K = L = G.) GP = {xP:x E G )
is contained in D. Since A and B are distinct maximal subgroups, it follows
5.50. Prove that [yi(G), yj(G)] < Y,+~(G)
for all i, j. (Hint. Use the three subgroups
that AB = G, and so the product formula gives [G : Dl = p2. Hence, G/D is
lemma.)
abelian and G' I D, by Theorem 2.23. As G = AB, each x E G is a product of
5.51. If H ti G and H n G' = 1, then H < Z(G) (and so H is abelian). a power of a and a power of b; but every element of D is simultaneously a
 5. Normal Series p-Groups 121
120

5' r - 1 mod 2'. Since m 2 3, this congruence implies 5' EZ - 1 mod 4; but
power of a and a power of by and so it commutes with each x E G; that is,
5 = 1 mod 4 implies 5' = 1 mod 4, a contradiction. It follows that these two
DI Z(G). We have seen that
G' I D I Z(G), I cyclic subgroups generate their direct product, which is a subgroup of order
at least 2 x 2" 2 2 x 2"-' = 2'I-l = ~ ( 2 " ) . This subgroup is thus all of
U(Z2m).
so that the hypothesis of Lemma 5.42(i) holds. Hence, for every x, y E G,
[y, x]P = [yp, x]. But yP E D 5 Z(G), and so [y, xlP= 1. Now Lemma 5.42(ii)
Corollary 5.45. Let G be a group containing elements x and y such that x has
gives (xy)' = [y, ~ ] ~ ( ~ - ~If ) p~is~odd,
x ~ then
y ~ plp(p
. - 1)/2, and ( ~ y =) ~ order 2"' (where m 2 3), y2 = xZr,and yxyU1= x'. Then
xpyp. By Exercise 2.55, if G[p] = {x E G: x p = 1) and GP = {xp: x E G) (as
defined above), then both these subsets are subgroups and [G : G[p]] = 1 GPI.
Thus,
~ G [ ~ ] ~ = [ G : G P ] = [ G : D ] [ D : G ~ ~ ~ P ~ ~ I In the latter two cases, G contains at least two inuolutions.

and G[p] contains a subgroup E of order p2; but E must be elementary ProoJ Since y2 = xZrcommutes with x, we have
abelian, so that G[p], hence G, contains more than one subgroup of order p.
We conclude that p = 2.
When p = 2, we have D = (a2) = G2 5 Z(G), [G : Dl = 4, and since so that t2 = 1 mod 2", and the congruence class [t] is an element of order 2
[y, XI' = 1 for all x, y E G, in U(Z,,,,). If m 2 3, the lemma exhibits the only four such elements, and this
gives the first statement.
(xy)4 = [y, xI6x4y4= x4y4. +
One involution in G is x2"-I. Suppose t = 1 2"'-'. For any integer lc,
Hence ( G[2] / = [G : G4] = [G : Dl [D : G4] = 8, because D = (a2 ) and G4 = (xky)2 = xk(yxky-~)y2= X k + k ' + 2 r = x 2 ~
(a4). If G[2] had only one cyclic subgroup of order 4, then it would contain
more than one involution (for every element of G[2] has order either 1,2,
+ +
where s = k(1 2rn-2) 2'-'. Since m 2 3, 1 + 2'-2 is odd, and we can
solve the congruence
or 4); there are thus two cyclic subgroups (u) and (u) of order 4 in G[2].
If a4 # 1, we may take (u) 5 (a2) 5 Z(G), and so (u) (v) is an abelian I +
s = k(l 2n'-2) + 2'-l r 0 mod 2"-'.
subgroup of G. But (u) (v) contains at least two involutions: either u2 # v2 For this choice of k, we have ( x " ~ =) ~xZS= x2'" = 1, SO that xky is a second
or u2 # uv-'; this contradiction shows that a4 = 1. It follows that ID1 = 2 involution (lest y E (x)).
and IGl = 8. By Exercise 4.34, G z Q or G z Z,; but only Q has more than I
one subgroup of index 2. @I Suppose that t = -1
X k + k t +2' -
+2"-'. As above, for any integer iz, ( X ~ Y=) ~
- x ~ ~ ~ Rewrite
- ' + ~ the
~ .exponent
We do an exercise in congruences before giving the next theorem.

Theorem 5.44. Let U(ZZm)be the multiplicative group

I and choose k so that k2"-'-' -
1 ~ 2 " -+
~ 2' = 2'(lc2"'-'-1 - 1 )7

1 mod 2'"-'; that is, there is an integer 1with

lc2rtt-r-1 - 1 = E2"'-'. For this choice of k, we have
U(Z2,) = ([a] E Z,,: a is odd). (Xky)2= X2r(k2m-r-1 - 1) = X12m
= 1,
If m 2 3, then
U(Z2.") = ([-I], [5]) z Z2 x Z2m-1. I and so G contains a second involution. Dl

Theorem 5.46. A finite p-group G having a unique subgrozrp of order p is either

Remark. U(B2,) is the group of units in the ring Z,... cyclic or generalized quaternion.
Proof. By Exercise 2.23, IU(Z2,.)\ = (~(2"')
= 2"'-'. Induction and the bino-
ProoJ The proof is by induction on n, where I GI = pH;of course, the theorem
mial theorem show that
is true when n = 0.
52"-) = (1 + 4)2.'-' 5 1 + 2t)l-l mod 2". Assume first that p is odd. If n > 0, then G has a subgroup H of index p, by
Exercise 4.2, and H is cyclic, by induction. There can be no other subgroup
Since U(Z2") is a 2-group, [5] has order 2', for some s 2 m - 2 (because
of index p, lest G be the quaternions (Theorem 5-43), which is a 2-group.
1 + 2"-I f. 1 mod 2"). Of course, [-l] has order 2. We claim that
Therefore, H is the unique maximal subgroup of G, and so it contains every
<[51) n([-11) = 1. If not, then [5'] = [-l] for some t; that is,
 5. Normal Series
122
proper subgroup of G. But if G is not cyclic, then (x) is a proper subgroup of
G for every x E G, and so G 5 H, which is absurd.
Assume now that G is a 2-group. If G is abelian, then Theorem 2.19 shows
that G is cyclic; therefore, we may assume that G is not abelian. Let A be a
maximal normal abelian subgroup of G. Since A has a unique involution, A
is cyclic, by Theorem 2.19, say, A = (a). We claim that A has index 2. As-
sume, on the contrary, that 1 G/A/ 2 4. If G/A does not have exponent 2, then
there is Ab E G/A with b2 4 A. Consider H = (a, b2) < (a, b) 5 G. If H is
abelian, then b2 centralizes A, contradicting Theorem 5.41(ii). As H is not
abelian, it must be generalized quaternion, by inducvon. We may thus as-
sume that b2ab-2 = ad'. Now (a) a G gives babe' = a' for some i, so that
-
and i2 - 1 mod 2", where 2' is the order of a. Note that e 2 2, for A prop-
erly contains Z(G). But there is no such congruence: if e > 3, then Theorem
5.44 shows that this congruence never holds; if e = 2, then - 1 is not a square
mod 4. It follows that G/A must have exponent 2. Since IG/AI > 4, G/A
contains a copy of V Therefore, there are elements c and d with c, d, c-'d 6A
and with (a, c), (a, d), and (a, cV'd) proper subgroups of G. Now none of
these can be abelian, lest c, d, or c-'d centralize A, so that all three are
generalized quaternion. But there are equations cac-' = a-' = dad-', giving
c-'d E Cc(A), a contradiction. We conclude that A = (a) must have index 2
in G.
Choose b E G with b2 E (a). Replacing a by another generator of A if
necessary, we may assume, by Exercise 2.20, that there is some r 5 n - 2 with
b2 = a2r.
Now bab-' = a' for some t, because (a) a G. Since G has only one involu-
tion, Corollary 5.45 gives t = & 1. But t = 1 says that a and b commute, so
that G is abelian, hence cyclic. Therefore, we may assume that t = - 1 and
G = (a, b), where
a2"-' = 1, bab-l = a-', b2 = a2'.
To complete the proof, we need only show that r = n - 2. This follows from
Theorem 5.44: since t = - 1, we have 2' 1 -2'mod 2"-', so that 2'" =
Omod2"-l,andr=n-2. Ed
It is not unusual that the prime 2 behaves differently than odd primes.
Definition. If G is a group, the its Frattini subgroup @(G)is defined as the
intersection of all the maximal subgroups of G.
If G is finite, then G always has maximal subgroups; if G is infinite, it may
have no maximal subgroups. For example, let G = Q, the additive group of
rationals. Since G is abelian, a maximal subgroup H of G would be normal,
and so G/H would be a simple abelian group; hence G/H would be finite and
of prime order. But it is easy to see that Q has no subgroups of finite index (it
has no finite homomorphic images).
If an (infinite) group G has no maximal subgroups, one defines @(G)= G.
It is clear that @(G)char G, and so @(G)a G.
Definition. An element x E G is called a rrongenevator if it can be omitted from
any generating set: if G = (x, Y), then G = (Y).
Theorem 5.47. For every group G, the Frattini subgroup @(G)is the set of all
nongerzerators.
Proof. Let x be a nongenerator of G, and let M be a maximal subgroup of G.
If x iji M , then G = (x, M) = Mya contradiction. Therefore x E M , for all M y
and so x E @(G).Conversely, if z E @(G),assume that G = (z, Y). If (Y) #
G, then there exists a maximal subgroup M with (Y) 5 M. But z E Myand
so G = (z, Y) 5 M, a contradiction. Therefore, z is a nongenerator.
Theorem 548.Let G be a finite group.
(i) (FrattiPni, 1885). @(G)is nilpotent.
(ii) If G is a finite p-group, then @(G)= G'Gp, where Gp is the subgroup of G
generated by all pth powers.
(iii) If G is a finite p-group, then G/@(G)is a vector space over Z,.
ProoJ: (i) Let P be a Sylow p-subgroup of @(G)for some p. Since @(G)a G,
the Frattini argument (!) gives G = @(G)NG(P).But @(G) consists of non-
generators, and so G = N,(P); that is, P a G and hence P a @(G).Therefore,
@(G)is the direct product of its Sylow subgroups; by Theorem 5.39, Q(G) is
nilpotent.
(ii) If A4 is a maximal subgroup of G, where G is now a p-group, then
Theorem 5.40 gives M 4 G and [G : A4-J = p. Thus, G/M is abelian, so that
G' I M; moreover, G' has exponent p, so that x p E M for all x E G. Therefore,
G'GP 5 B?(G).
For the reverse inclusion, observe that G/G'Gp is an abelian group of expo-
nent p, hence is elementary abelian, and hence is a vector space over Z,.
Clearly Q(G/G'Gp) = 1. If H a G and H 5 @(G),then it is easy to check that
@(G)is the inverse image (under the natural map) of @(G/H) (for maximal
subgroups correspond). It follows that @(G)= G'Gp.
(iii) Since G'GP = @(G),the quotient group G/@(G) is an abelian group of
exponent p; that is, it is a vector space over Z,.
Theorem 5.49 (Gaschiitz, 1953). For every (possibly infinite) group G, one has
G' n Z(G) I @(G).
Proof. Denote G' n Z(G) by D. If D $ @(G),there is a maximal subgroup M
of G with D $ M. Therefore, G = MD, so that each g E G has a factorization
124 5. Normal Series

g = md with rn E M and d E D. Since d E Z(G), gMgdl = rndMd-'rn-l = CHAPTER 6

mMm-' = M, and so A4 a G. By Exercise 2.58, G/M has prime order, hence
is abelian. Therefore, G' 5 M. But D 5 G' < M, contradicting D $ M. Ef Finite Direct Products
Dehition. A minimelgenerating set of a group G is a generating set X such
that no proper subset of X is a generating set of G.

There is a competing definition in a finite group: a generating set of smallest

cardinality. Notice that these two notions can be distinct. For example, let
G = (a) x (b), where a has order 2 and b has order 3. Now {a, b ) is a
minimal generating set, for it generates G and no proper subset of it gener-
ates. On the other hand, G is cyclic (of order 6) with generator ab, and so lab)
is a minimal generating set of smaller cardinality. In a finite p-group, how-
ever, there is no such problem.

Theorem 5.50 (Burnside Basis Theorem, 1912). If G is a finite p-group, then

any two minimal generating sets have the same cardinality, namely, dim G/@(G).
Moreover, every x $ Q(G) belongs to some minimal generating set of G.
The main result of this chapter is a complete description of all finite abelian
Proof. If (x,, ..., x,,} is a minimal generating set, then the family of cosets groups as direct products of cyclic p-groups. By passing from abelian groups
(Z,,. .., XI,}spans G/@(G)(where X denotes the coset x@(G)).If this family is to modules over a principal ideal domain, we show that this result gives
dependent, then one of them, say Z,,lies in (X,, . .., Z,,). There is thus y E canonical forms for matrices. The essential uniqueness of the factorization of
(x,, . ..,x,) s G with x, y-' E @(G). Clearly, {x,y-', x,, . . ., x,,} generates a finite abelian group as a direct product of cyclic p-groups is then generalized
G, so that G = (x,, . . ., x,), by Theorem 5.47, and this contradicts mini- to nonabelian groups that are direct products of "indecomposable" groups.
mality. Therefore, n = dim G/@(G),and all minimal generating sets have the
same cardinality.
If x $ @(G),then Z # 0 in the vector space G/@(G),\and so it is part of a
basis {X, Z,, ...,Z,,,).If xi represents the coset Zi, for i > 2, then G = (@(G),x,
x,, ... , x,) = (x, x,, ..., x,). Moreover, {x, x,, . . . ,x,,} is a minimal gener- The Basis Theorem
ating set, for the cosets of a proper subset do not generate G/Q(G). H
For the next few sections, we shall deal exclusively with abelian groups; as
is customary, we shift from multiplicative to additive notation. Here is a
dictionary of some common terms.
5.52. Every subgroup of Q,, is either cyclic or generalized quaternion.
5.53 (Wielamdt). A finite group G is nilpotent if and only if G' < @(G). ab a+b
1 0
5.54. If G is a finite p-group, then G is cyclic if and only if G/@(G)is cyclic. a-I -a
an na
Definition. A finite p-group G is extra-specia! if Z ( 6 ) is cyclic and @ ( G ) = ab-' a-b
Z(G) = GI. HK H+K
aH a+H
5.55. If G is extra-special, then G/Z(G) is an elementary abelian group. direct product direct sum
H x K WOK
5.56. Every nonabelian group of order p3 is extra-special. n H~ C Hi
5.57. (i) If rn is a power of 2, what is the class of nilpotency of D,,,?
(ii) What is the class of nilpotency of Q,,?(Hint.Exercise 4.42.) If a nonabelian group G = H x K is a direct product, then H is called a direct
126 6. Finite Direct Products
factor of G; in additive notation, one writes G = H @ K , and one calls H a
(direct) summand of G.
There are two remarks greatly facilitating the study of abelian groups.
+ +
First, if a, b E G and n E Z, then n(a b) = na nb (in multiplicative nota-
tion, (ab)" = anb",for a and b commute). Second, if X is a nonempty subset of
G, then (X) is the set of all linear combinations of elements in X having
coefficients in Z.(see Theorem 2.7: in additive notation, words on X become
linear combinations).
Definition. If G is an abelian p-group for some prime p, then G is also called
a p-primcavy group.
When working wholly in the context of abelian groups, one uses the term
p-primary; otherwise, the usage of p-group is preferred.
We have already proved, in Theorem 5.39, that every finite nilpotent group
is the direct product of its Sylow subgroups; since every abelian group is
nilpotent, the next theorem is an immediate consequence. However, we give
another proof here to put the reader in the abelian mode. The following
theorem was attributed to Gauss by G.A. Miller (1901).
Theorem 6.1 (Primary Decomposition). Every finite abelian group G is a direct
sum of p-primary groups.
PuooJ Since G is finite, it has exponent n for some n: we have nx = 0 for all
x E G. For each prime divisor p of n, define
G, = (x E G: pex = Ofor some e).
Now G, is a subgroup of G, for if pnx = 0 and pl"y = 0, where m 5 n, then
pn(x - y) = 0 (because G is abelian). We claim that G = G,, and we use the
criterion in Exercise 2.75(i).
Let n = p;l.. .,p: where the pi are distinct primes and ei > 0 for all i. Set
ni = n/pfi, and observe that the gcd (n,, ..., n,) = 1 (no pj divides every n,). By
Theorem VI.2 in Appendix VI, there are integers si with sini = 1, and so
x = (sinix). But sinixE Gpi, because ~ , ' ~ s , n= , xsinx = 0. Therefore, G is
generated by the family of G,'s.
Assume that x E G, n (U,+, G,). On the one hand, pex = 0 for some
e 2 0; on the other hand, x = x,, where qegx, = 0 for exponents e,. If
n
m = qeq, then m and pe are relatively prime, and there are integers r and s
+
with 1 = rm + spe.Therefore, x = rmx spex = 0, and so G, n (u,+, G,) =
0. \
Definition. The subgroups G, are called the p-pvirnouy comporrrrits of G.
Of course, G, is the Sylow p-subgroup of G, but the usage of p-primary
component is preferred when the works wholly in the context of abelian groups.
The Basis Theorem 127
We are going to show that every finite abelian group is a direct sum of
cyclic groups; it now suffices to assume that G is p-primary.
Definition. A set {x,, . . ., x,} of nonzero elements in an abelian group is
independent if, whenever there are integers nz,, ... , m, with XI,, mixi = 0,
then each mixi = 0.
When an abelian group G has exponent p, for some prime p, then it is a
vector space over Z,,and the notion of independence just defined coincides
with the usual notion of linear independence: mixi = 0 implies plm,, so that
the congruence class [in,] = 0 in Z,.Of course, if G has no elements of finite
order (as is the case, for example, when G is a vector space over Q, R, or C),
then ?nixi= 0 implies mi = 0, and so the definition of independence coincides
with that of linear independence in this case as well.
Lemma 6.2. If G is an abelian group, then a subset {x,, ..., x,) of nonzero
elements of G is independent if and only if (x,, . . .,x,) = (x, ) @ .. . @ (x,).
Proof. Assume independence; if y E (xi) n ({x,: j f i)), then there are
integers m,, ..., m, with y = - mixi = jii mjxj, and so EL=,mkxk= 0. By
independence, mkxk= 0 for all k; in particular, mixi = 0 and so y = - mixi =
0. Exercise 2.75(i) now shows that (x,, . . . ,x,) = (x,) @ ... @ (x,).
For the converse, assume that X mixi = 0. For each j, we have -mjx, =
Ex+ mkxkE (xj) n ((xk: k # j ) ) = 0. Therefore, each mjxj = 0 and {x,, .. .,
x,) is independent. R
Here is a solution to a part of Exercise 2.78.
Corollary 6.3. Every finite abelian group G of prime exponent p is an elemen-
tary abelian p-group.
Proof. As a vector space over Z,, G has a basis (x,,..., x,.}. Therefore, G =
(x,, .. .,x,), because a basis spans, and G = (x, ) @ -.- @ (x,), because a
basis is independent. Bill
Lemma 6.4. Let {x,, .. ., x,} be an independent subset of a p-primary abelian
group G.
(i) If {z,, .. .,q.}c G, where pz, = xi for all i, then {z,, ...,z,} is independent.
(ii) If k,, ..., k, are integers with kixi # 0 for all i, then {k,x,, ...,krxr) is also
independent.
Proof. An exercise for the reader.
DeP?nitiion.If G is an abelian group and in > 0 is an integer, then
mG = {mx: x E G}.
128 6. Finite Direct Products
It is easy to see that ntG is a subgroup of G; indeed, since G is abelian, the
function p,: G -+G, defined by x I---+mx, is a homomorphism (called nrultipli-
cation by RZ),and mG = im p,. We denote ker p , by G[mn]; that is,
Theorem 6.5 (Basis Theorem).' Every finite abelian group G is a direct sum of
primary cyclic groups.
h.oof. By Theorem 6.1, we may assume that G is p-primary for some prime
p. We prove the theorem by induction on n, where pnG = 0. If n = 1, then the
theorem is Corollary 6.3.
Suppose that f t l G = 0. If H = pG, then pnH = 0, so that induction gives
H = xi=,(yi). Since yi E H = pG, there are zi E G with pzi = y,. By Lemma
,
6.2, { y , . ..,y,) is independent; by Lemma 6.4 (i), (z, , . . ., z, ) is independent,
and so L = (z,, . .. ,z,) is a direct sum: L = xi=,(2,).
Here is the motivation for the next step. Were the theorem true, then
x
G = C,, where each C, is cyclic, and H = pG = PC,. In considering pG,
therefore, we have neglected all C, of order p, if any, for multiplication by p
destroys them. The construction of L has recaptured the C, of order greater
than p, and we must now revive the C, of order p.
For each i, let ki be the order of y,, so that kizi has order p. The linearly
independent subset {k,z,, . . .,k,z,) of the vector space G[p] can be extended
to a basis: there are elements {x,, ...,x,) so that {k,z,, . ..,k,z,, x,, . .. ,x,)
is a basis of G[p]. If M = (x,, . . .,x,), then independence gives M = (xj).
We now show that M consists of the resurrected summands of order p; that
is, G = L @ M, and this will complete the proof.
(i) L n M = 0. If g E L n M, then g = 1
because g E My and so x x
bizi = ajxj. Now pg = 0,
pbizi = 0. By independence, pbizi = biyi = 0
for all i. It follows from Exercise 2.13 that bi = bjk, for some bj. Therefore,
0 = bikizi - ajxj, and so independence of (k,z,, ..., krzr, x,, . . . , x,)
x
gives each term 0; hence g = ajxj = 0.
x
(ii) L + M = G. If g E G, then pg E pG = H , and so pg = ciyi = pcizi.
x x x
Hence, p(g - cizi) = 0 and g - cizi E G[p]. Therefore, g - cizi =
bikizi + ajxj, so that g = (c, + biki)zi + ajxj E L + M . E l
Corollary 6.6. Every finite abelian group G is a direct sum of cyclic groups:
G xi=,(xi), where xi has order mi, and
= +
m,Im,I.-.lm,-
Proof. Let the primary decomposition of G be G = xi=,
Gpi. By the basis
theorem, we may assume that each Gpi is a direct sum of cyclic groups; let
' The basis theorem was proved by E. Schering (1868) and, independently, by L. Kronecker
(1 870).
The Basis Theorem 129
Ci be a cyclic summand of Gpi of largest order, say, pfi. It follows that G =
K @ (C, @ .. @ C,), where K is the direct sum of the remaining cyclic sum-
mands. But C, @ - -.@ C, is cyclic of order m = p?, by Exercise 2.62(ii).
Now repeat this construction: let K = H @ D, where D is cyclic of order n,
say. If there is a cyclic summand S, in D arising from Gpi, that is, if Gpi # Ci,
then S, has order pp pfi, so that p:lpfi, for all i, and n/m. This process ends
in a finite number of steps.
Definition. If G has a decomposition as a direct sum G = C,, where Ci is
cyclic of order mi and m, Im2I.. .I m,, then one says that G has irrva~ia~r~tfactors
(ln, , . . ., int).
Theorem 4.7. I f p is an odd prime, the multiplicative group
U(Z,,,) = ( [a] E Zpn:(a, p) = I),
is cyclic of order (p - l)pJ1-'.
Remark. Theorem 5.44 computes this group for the prime 2.
ProoJ If n = 1, the result is Theorem 2.18, and so we may assume that n 2 2.
Let us denote U(Zp,) by G. By Exercise 2.23,/ GI = q(pn)= (p - l ) ~ " - ' .
It is easy to see that B = {[b] E G: b z 1 mod p) is a subgroup of 6. Every
integer b has a unique expression in base p: if 1 < b < pn, then
b = ao + a l p + + a,l-,p"-l, where 0 a, < p.
Since [b] E B if and only if a, = 1, it follows that IBJ = pn-l, and so B is
p-primary. By the primary decomposition, there is a subgroup A of G with
IAI=p-1 andwithG=A@B.Ifwecanshow thateachofAand Bis
cyclic, then Exercise 2.62(ii) will show that G is cyclic.
Consider f : G -+ U(Zp) defined by f([a]) = cls a (where [a] denotes the
congruence class of n mod p", and cls a denotes the congruence class of
a mod p). Clearly, f is a surjection and kerf = B, so that G/B z U(Zp) E
Zp-, . On the other hand, G/B = (A @ B)/B r A, and so A z Zp-, .
+
We shall show that B is cyclic by showing that [I p] is a generator. Let
us prove, by induction on m > 0, that
(1 + P ) ~ "= 1 mod pm+l and (1 + P)~'"$ 1 mod ptn+2.
+ +
If m = 0, then 1 p r 1 mod p and 1 p 1 mod p2. For the inductive
step, the assumed congruence gives (1 + p)*m*l= ((1 + p)pm)p= (1 + kpnril)p 7
for some integer lc; the assumed incongruence gives pKk. The binomial theo-
rem gives (1 + kp"+')P = 1 + 12ptnC2 for some I. Hence, (1 + p)p"'" z
1 mod p"t2 and (1 + p)""*' +
f lp1n+3
1 mod p ~ " ~ . It follows that (1 + p)"-' f
1 mod p", and so [I + p] has order pn-'. FB4
Here is another proof of the basis theorem.
 6. Finite Direct Products
130
Lemma 6.8. If G = (x,, . . . , x,,) and if a,, . . . , a,, are relatively prime integers,
then there is a generating set of G comprised of n elements one of which is
C;=, aixi.
Proof. By Lemma VI.4 (in Appendix VI), there is a unimodular n x n matrix
A with integer entries (i.e., det A = 1) whose first row is a,, . . ., a,. Define
Y = AX, where X is the column vector with entries x,, . . ., x,,. The entries
y,, . . . ,yn of the column vector Y are linear combinations of the xi, hence are
elements of G; moreover, y, = x:=,
aixi.Now X = A-'AX = A-' Y. Since A
is unimodular, all the entries of A-I are also integers. It follows that each xi
is a Z-linear combination of the y's, and so G = (y,, . . ., y,,).
Theorem 6.9 (Basis Theorem). Every finite abelian group G is a direct surn of
cyclic groups.
Proof(E. Schenkman). Choose n smallest such that G can be generated by a
set with n elements. Among all generating sets (x, ,. . . ,x,,] of size n, choose
one containing an element x, of smallest order k;that is, no generating set of
size n contains an element of order less than k. If H = (x,, ..., x,), then H is
a proper subgroup of G (by the minimal choice of n), so that an induction on
161 gives H a direct sum of cyclic groups. We claim that G = (x,) @ H. It
sufficestoshowthat(x,)nH=O,for(x,)+H=(xl ,..., x , , ) = G . I f z ~
( x l ) n H a n d z f 0, t h e n z = a , x , = C f = , a i x i , for a, ,... , ~ , , E and
Z O i
a, < k. If d is the gcd of a,, . .., a,,, define g = -(a,/d)x, +
Now the order of g is smaller than k , for dg = 0 and d 5 a, < k. But a,/d, .. .,
x7=, (ai/d)xi.
an/d are relatively prime, so that the lemma gives a generating set of G of size
n one of whose elements is g; this contradicts the minimality of k.
6.1. Use the basis theorem to show that if G is a finite abelian group of order n, and
if k[n, then G contains a subgroup of order Ic.
6.2. Use the basis theorem to give a new proof of Theorem 2.19.
6.3. A finite abelian group G is generated by its elements of largest order. Show, by
considering D,, that this may not be true of nonabelian groups.
6.4. If G is a finite p-primary abelian group, and if x E G has largest order, then (x)
is a direct summand of G.
6.5. If G is an abelian group with invariant factors (m,, .. .,m,), then the order of G is
n miand the minimal exponent of G is m,.
6.6. If G is a finite p-primary group, then @(G)= pG. Conclude, from the Burnside
basis theorem, that d(G),the minimal number of generators of G, is dim G/pG.
6.7. If G and H are elementary abelian p-groups, then d(G @ Hj = d(G) + d(H).
6.8. Let G be a direct sum of b cyclic groups of order pn'. If n < m, then ~ " G / ~ " +is' G
elementary and d(pnG/p"+l G) = b.
The Fundamental Theorem of Finite Abelian Groups 131
The Fundamental Theorem of Finite Abelian Groups
We have not yet answered a basic question about finite abelian groups: When
are two such groups G and H isomorphic? Since both G and H are direct
sums of cyclic groups, your first guess is that G z H if they have the same
number of summands of each kind; since H, z Z,@ H,, however, one had
better try to count primary cyclic summands of each kind. But this leads to a
serious problem. Now can we count summands at all? To do so would re-
quire that the number of primary cyclic summands of any given kind is the
same for every decomposition of G. That is, we seek an analogue of the
fundamental theorem of arithmetic in which the analogue of a prime number
is a primary cyclic group.
Lemma 6.10. If a p-primary abelian group G has a decomposition G =
Ci
into a direct sum of cyclic groups, then the number of Ci having order 2 p"" is
d(p"G/pnilG), the minimal number of generators of p"G/pnil~.
Proof. Let Bk be the direct sum of all Ci, if any, of order pk; say, there are
bk 2 0 such summands in 13,. Thus,
Now pnG= pnBU+,0...0 pnB,, because pnB, = . - -= pnB, = 0, and p " + ' ~=
p'i1B,,+2@-..~p11t1~,. Therefore, p n G / p n + 1 G ~ p " B n + , @ ( p " B , + 2 / p n t 1 ~ , , + 2 ) @
. . @ (p"B,/pntl~,),and so Exercise 6.7 gives d(p"G/pnilG) = bn+, + b,,,, +
... + b,. DL!
Definition. If G is a finite p-primary abelian group and n 2 0, then
The important thing to notice now is that Up(n, G) is a number depending
on G but not upon any particular decomposition of G into a direct sum of
cyclic groups.
Theorem 6.11. If G is a finite p-primary abelian group, then any two decompo-
sitions of G into direct sums of cyclic groups have the same number of summands
of each kind. More precisely, for every n 2 0, the number of cyclic summands
of order pnil is Up(n, G).
Proof. For any decomposition of G into a direct sum of cyclic groups, the
lemma shows that there are exactly Up(n, G) cyclic summands of order pnil.
The result follows, for U,(n, G) does not depend on the choice of decomposi-
tion.
CoroPlary 6.12. If 6 and H are finite p-primary abelian groups, then G r H if
and only if Up(n, G) = Up(n, H) for all rz 2 0.
132 6. Finite Direct Products
ProoJ If q : G -, H is an isomorphism, then q(pt1G)= pnH for all n 2 0, and
so q induces isomorphisms pnG/p"+'G z pnH/p"+'H for all n. Therefore,
Up(n, G) = U,(n, H) for all n.
Conversely, if G and H each have direct sum decompositions into cyclic
groups with the same number of summands of each kind, then it is easy to
construct an isomorphism G -+ H. El
We have only to delete the adjective p-primary to complete this discussion.
Definition. The orders of the primary cyclic summands of G, that is, the
numbers pnil with multiplicity Up(n, G) > 0 for all primes p and all n 2 0, are
called the elementary &visors of G.
For example, the elementary divisors of an elementary abelian group of
order p3 are (p, p, p), and the elementary divisors of Z,are (2, 3).
Theorem 6.13 (Fundamental Theorem of Finite Abelian group^).^ If G and H
are finite abelian groups, then G E' W if and only if, for all primes p, they have
the same elementary divisors.
Proof. The proof follows from two facts, whose easy proofs are left to the
reader:
(1) If q : G -+ H is a homomorphism, then q(Gp) 5 H, for all primes p;
(2) G z H if and only if G, E H, for all primes p.
Corollary 6.14. Let G be a finite abelian group.
(i) If G has invariant factors (m,, .. ., m,) and invariant factors ( k , , .. ., k,),
then s = t and 12, = mi for all i.
(ii) Two finite abelian groups G and H are isomorphic if and only if they have
the same invariant factors.
Proof. (i) The hypothesis gives two direct sum decompositions: G = zj=,Ci
and G = z:=, Dj, where C, is cyclic of order mi, Dj is cyclic of order kcj,
m, lm21.. .I m,, and k, 1 k, 1.. . I k,. By Exercise 6.5, m, = I<,, for each is the mini-
mal exponent of G. By Exercise 6.10(i) below, the complementary summands
x::
Ci and G = - . D,, are isomorphic, and the proof is completed by
induction on max(s, t ) .
(ii) This follows at once from (i).
If one arranges the elementary divisors of a p-primary group in ascending
order, then they coincide with the invariant factors of G. However, elemen-
tary divisors and invariant factors can differ for groups G which are not
This theorem was proved in 1879 by G. Frobenius and L. Stickelberger.
Canonical Forms; Existence
p-primary. For example, let
The elementary divisors of G are (2,2, 2,4; 3, 9), while the invariant factors of
G are (2,2, 6, 36).
6.9. If G and H are finite abelian groups, then
up(n, G O H) = Up(n, G) + U,(n, H)
for all primes p and all n 2 0.
6.10. (i) If A, Byand C are finite abelian groups with A @ C r B @ C, then A r B.
(Hint. Exercise 6.9.)
(ii) If A and B are finite abelian groups for which A @ A r B O B, then A r B.
6.1 1. (i) If p is a prime and e 2 1, then the number of nonisomorphic abelian groups
of order p" is 9 ( e ) , the number of partitions of e.
n ni
(ii) The number of nonisomorphic abelian groups of order n = pPi is
where the pi are distinct primes and the ei are positive integers.
9(ei),
(iii) How many abelian groups are there of order 864 = 2533?
6.12. (i) Let G = (a) x (b), where both (a) and (b) are cyclic of order p2. If H =
(pa) x (pb), compare Up(n,G) with Up(n,H) and Up(n, GIH).
(ii) Let G and H be finite abelian groups. If, for each k, both G and H have the
same number of elements of order k, then G r H. (Compare Exercise 4.33.)
6.13. If G is a finite abelian group and H < G, then G contains a subgroup isomorphic
to GIH. (Compare Exercise 4.29.)
Remark. The best solution to this exercise uses character groups; see Theo-
rem 10.55.
6.14. What are the elementary divisors of U(Z,,), the multiplicative group of all con-
gruence classes [a] mod n with (a, n) = l?
6.15. Use the Fundamental Theorem of Finite Abelian Groups to prove the Funda-
mental Theorem of Arithmetic. (Hint. If n = p,"'.. .p,"', then Exercise 2.62(ii)
gives ZlnZ r nf=,
Zlp7Z.)
Canonical Forms; Existence
We digress from the study of groups to apply the results of the preceding two
sections to Linear Algebra; we shall prove the existence and uniqueness of the
rational and Jordan canonical forms of a matrix. This material will not be
used until Chapter 8, but the reader will be pleased to see that the difficult
134 6. Finite Direct Products
portion of a first course in Linear Algebra can be done more easily from a
more advanced viewpoint (it is assumed that the reader has already learned
much of this, and so our pace is not leisurely). This project is one of transla-
tion, and so we first introduce a new vocabulary. The reader unfamiliar with
the rudiments of principal ideal domains can consult Appendix VI.
Definition. Let R be a ring. An abelian group V is an R-rnodzde if there is a
function s: R x V -+ V (called scalar multiplication and denoted by (a, v) t-,
av) such that, for every a, ,!I,1 E R and u, v E V :
(i) (ap)v = a(Pv);
+ +
(ii) (a p)v = av pv;
(iii) a(u + v) = au + av; and
(iv) l u = v.
When R is a field, an R-module is just a vector space. Thus, one may think
of an R-module as a vector space over a ring. Here we are concerned with
R-modules for R a principal ideal domain (we shall abbreviate "principal
ideal domain" to PID). Our favorite PID's are Z and k[x], the ring of all
polynomials in x with coefficients in a field k.
EXAMPLE 6.1. The terms abelian group and Z-module are synomyms. Every
abelian group is a Z-module, for axioms (i) through (iv) always hold for
scalars in Z.
EXAMPLE 6.2. Let k be a field and let R = k[x]. If V is a vector space over k
and T: V + V is a linear transformation, then V can be made into a k[x]-
module, denoted by VT, by defining f(x)v = f(T)v for all f(x) E Ic[x] and
v E I/. In more detail, iff (x) = a, + alx+ a2x2+ . .. + at,xnE k[x], define
where Ti is the composite of T with itself i times. The reader should check
that axioms (i) through (iv) do hold.
Just as a principal ideal domain is a generalization of Z,so is an R-module
a generalization of an abelian group. Almost any theorem holding for abelian
groups has a true analogue for R-modules when R is a PID; moreover, the
proofs of the generalizations are usually translations of the proofs for abelian
groups.
Definition. If V is an R-module, then a subgroup W of V is a sakbmod~dkif it
closed under scalar multiplication: if w E W and r E R, then rw E W. If W is a
submodule of V, then the quotient module V/W is the abelian group VJW
+
equipped with the scalar multiplication r(v W) = rv + W (the reader should
check that this is well defined).
Canonical Forms; Existence 135
6.16. A commutative ring R itself is an R-module (if r, s E R, define scalar multiplica-
tion rs to be the given product of two elements in R). Show that the submodules
of R are its ideals.
6.17. (i) The intersection of any family of submodules of an R-module V is itself a
submodule.
(ii) If X is a subset of V, let (X) denote the submodule generated by X; that is,
(X) is the intersection of all the submodules of V containing X. If X J; a,
show that (X) is the set of all R-linear combirtatioms of elements of X; that
is,
(X) = {finite sums 2 rixi:ri E R and xi E X) .
In particular, the cyclic subrnodule generated by v, denoted by (v), is
{rv: r E R).
6.18. An R-module V is called finitely generated if there is a finite subset X wth
v = (X).
(i) If R is a field, prove that an R-module V is finitely generated if and only if it
is finite-dimensional.
(ii) Prove that an abelian group (Z-module) G is finite if and only if G is finitely
generated and every element in G has finite order.
6.19. Let VT be the 1c[x]-module of Example 6.2. Prove that W is a submodule if
and only if W is a subspace of V for which T(W) I W (W is often called a
T-invariant subspace of V ) .
Definition. If V and W are R-modules, then their direct srm is the direct sum
V @ W of abelian groups equipped with the scalar multiplication r(v, w) =
(rv, rw).
6.20. If Wl , ..., W, are submodules of an R-module V, then V r W1 @ . @ W;, if and
only if V = W, + + W;, (i.e., every v E V is a sum v = wi, where wi E q )
and n (Uj,i q)= 0 for all i.
We are almost finished with the vocabulary lesson.
Definition. Let V be an R-module, where R is a PID, and let v G I/. The order
of v; denoted by ord(v), is ( r E R: ru = 0) (it is easily checked that ordfv) is an
ideal in R). One says that v has$nite order if ord(v) # 0, and one says that V
is p-primc9ry (where p E R is irreducible) if, for all v E V, ord(v) = (pm)for some
rn (where (p) is the principal ideal generated by p). An R-module V isfiite if
it is finitely generated and every element of V has finite order.
If V is an abelian group and v E V, then a generator of ord(v) is the smallest
positive integer m for which rnv = 0; that is, ord(v) = (m),where m is the order
of v in the usual sense. Exercise 6.18(ii) tells us that we have translated "finite
136 6. Finite Direct PIoducts
abelian group" correctly into the language of R-modules: a "finite Z-module"
is an abelian group of finite order.
We remark that Exercise 6.18(ii) is false if one drops the hypothesis that
the group G is abelian. The question (posed in 1902) whether a finitely gener-
ated group G of exponent e is finite became known as Burnside's problen~.~
Burnside proved that if G 5 GL(n, @) is finitely generated and has exponent
e, then G is finite. There is an "obvious" candidate for a counterexample, and
it was actually shown to be one, when e is a large odd number, by Adian and
Novikov in 1968 (in 1975, Adian showed that the group is infinite for all odd
e 2 665). The proof is very long and intricate; a much simpler "geometric"
proof was found by A. Ol'shanskii in 1982. In 1994, S. Ivanov showed that
there are infinite finitely generated groups of exponent e = 2kw, where k 2 48
and m 2 1 is any odd number.
Theorem 6.15. If V is a finite R-rnodzrle, where R is a PID, then there are
vl, ..., Vwith
v = (u, ) 0' ' ' 0(us).
Moreover, the cyclic summands may be chosen to satisfy either of the following
conditions. If ord(vi) = (r,), then either:
(i) each ri is a power of some irreducible element in R; or
(ii) r11r2/...I rs.
Proof. The proofs of the corresponding group-theoretic theorems translate
routinely to proofs for modules. The decomposition of the first type arises
from Corollary 6.12 (using the primary decomposition and elementary divi-
sors), and the decomposition of the second type arises from Corollary 6.6
(using invariant factors).
Corollary 6.16. Let T: V -+ V be a linear transformation on a finite-dimen-
sional vector space over a field k. Then V = W, 0... 0 Ws, where each K. =
(v,) is a cyclic T-invariant szlbspace. Moreover, the vectors v,, . .. , us can be
chosen with ord(vi) = (fi(x)), so that either:
(i) each f;(x) is a power of an irreducible polynomial in k[x]; or
(ii) f l ( ~ ) l f 2 ( ~ --lL(x)-
)1.
Proof. Regard V as a k[x]-module VT, as in Example 6.2. Since V is finite-
dimensional, it has a basis (w,, . . .,w,,); each vector v E V is a k-linear combi-
The restricter1 B~~rtisideProblena asks whether there is a function f(n, d) with IGi 5 f(n, d) for
every finite group G having minimal exponent n and d generators. P. Hall and G. Higman (1956)
proved that it suffices to find such a function for all prime powes n; A.L. Icostrikin (1959) found
such a function for all primes p; E.I. Zelmanov (1989) completed the proof by finding such a
function for all prime powers; see Vaughan-Lee (1993).
Canonical Forms; Existence 137
nation v = ajwj; a fortiori, each u is a k[x]-linear combination of the w's,
and so v T is a finitely generated k[x]-module. Also, every v E V is annihi-
lated by some nonzero polynomial (this follows from the Cayley-Hamilton
theorem or, more simply, from the observation that since dim(V) = n, the
n + 1 vectors v, Tv, T2v, . . ., Tt'v must be linearly dependent, and so there is
some nonzero polynomial in T that annihilates every vector v).
Since VT is a finite k[x]-module, Theorem 6.15 shows that it is a direct
sum of cyclic submodules. By Exercise 6.19, these summands are cyclic
T-invariant subspaces.
Lemma 6.17. Let T: V -+ V be a linear transformation on a finite-dimensional
vector space T/.
(i) A subspace W is a cyclic T-invariant subspace of V if and only i j there is
a vector v E W and an integer s 2 1 so that (v, Tv, T ~ U..,., T"-'v) is a
basis of W.
(ii) Moreover, if Tsv = Cf;; aiTiv, then ord(v) is generated by g(x) =
xs - 1;:;aixi.
Pro05 (i) Consider the sequence v, Tv, T ~ V. ., . in W; since V is finite-dimen-
sional, there is an integer s 2 1 and a linearly independent subset (v, Tv, T2v,
..., T"-'u) which becomes linearly dependent when Tsv is adjoined. There-
fore, there are ai E k with TSv= Cf; ai Tiv. If w E W, then w = f (T)v for
some f(x) E k[x], and an easy induction on degree f shows that w lies in the
subspace spanned by (v, Tv, T2v, ..., Ts-'v); it follows that the subset is a
basis of W.
(ii) It is clear that g(x) E ord(v). If h(x) E ord(v), that is, if h(T)v = 0, then the
division algorithm gives q(x), r(x) E k[x] with h(x) = q(x)g(x) + r(x) and
either r(x) = 0 or degree r(x) idegree g(x) = s. Now r(x) E ord(v); hence
r(x) = z:=, /3,xj, t 5 s - 1, and I:=,
independence of (v, Tv, T2v, . . ., Ts-'v).
P,T~V= 0, contradicting the linear
We remind the reader of the correspondence between linear transforma-
tions and matrices. Let V be a vector space over a field 12 with an ordered
basis (u,, . . ., u,), 'and let T: V -+ V be a linear transformation. For each j,
Tuj is a linear combination of the ui: there area, E k with
The rnnfrix of T relative to tire ordered basis (u,, .. ., u,) is defined to be
A = [a,]. Therefore, for each j, the coordinates of Tuj form the jth column
of A.
lD@finili@n-If f (x) = x r + a,-,xr-' + ... + a, E k[x], where r 2 2, then the
138
companiorr matrix off (x) is the r x r matrix C(f):
6. Finite Direct Products
i Canonical Forms; Existence
if f(x) = x - a, then C(f ) is the 1 x 1 matrix [a].
ILernrna 6.18. Let T: W 4 W be a linear transformation on a finite-dimensional
vector space W over k, and let W = W T = (v) be a cyclic k[x]-module. The
matrix of T relative to the basis {v, Tv, T2v, .... Ts-lv) of W is the s x s
companion matrix C(g), where g(x) is the monic generator of ord(v). More-
over, the characteristic polynomial of C(g) is &).
ProoJ If 0 5 i < s- 1, then T(Tiv)= Tit'v, and T(Ts-'v) = T"V =zf~AaiTiv,
so that the matrix of T relative to the given basis is a companion matrix. The
Cayley-Hamilton theorem gives x(T) = 0, where ~ ( x is) the characteristic
polynomial of C(g). Therefore ~ ( xE) ord(v) = (g(x)), so that g(x)l~(x).Hence
s = degree ~ ( x2) degree g(x) = s, and so ~ ( x= ) g(x).
Definition. Let A be an r x r matrix and let B be an s x s matrix; their direct
szm is the (r + s) x (r + s) matrix
Note that the direct sum of A and B is similar to the direct sum of B and A.
Theorem 6.19. Every n x n matrix A over a field k is similar to a direct sum of
companion matrices C(f,), .... C(fq). Moreover, the f;(x) may be chosen so that
either:
(i) the J(x) are powers of irreducible polynomials in k[x]; or
(ii) f,(x)lf2(x, I.. .If,(.,
I
Proof. Let V = kn, the vector space of all n-tuples of elements in k (viewed as
column vectors). The srnrrdard basis of V is {el, .... e,,), where ei is the n-tuple
with ith coordinate 1 and all other coordinates 0. The matrix A defines a
linear transformation T: V -+ V by T(v) = Av, where v is a column vector.
Note that A is the matrix of T relative to the standard basis, for Ae, is the ith
column of A. View V as the k[x]-module vT.By Corollary 6.13, V is a direct
sum of cyclic T-invariant subspaces: V = (v, ) @'. .. @ (v,); by Lemma 6.17,
there is a new basis of V:
{v,, To,, T2vl,...; v,, Tv,, T'v,, . ..; .. .; vq, Tv,, T ~ V ,..
, .).
The matrix B of T relative to this new basis is a direct sum of companion
matrices, by Lemma 6.18, and A is similar to B, for they represent the same
linear transformation relative to different ordered bases of V. Finally, Corol-
lary 6.16 shows that the vi can be chosen so that the polynomials j;(x) satisfy
either (i) or (ii). @
Definition. A raiiorral canonical form is a matrix B that is the direct sum of
companion matrices C(f,),.... C(&) with f,(x)]f,(x)l.. .I&(x). The polyno-
mials f,(x), f2(x), .... fq(x) are called the invariantfactors of B.
Theorem 6.19(ii) thus says that every matrix over a field k is similar to a
rational canonical form.
Recall that the minirnumpolynomial of a matrix A is the monic polynomial
m(x) of smallest degree with m(A) = 0. The reader should look again at Exer-
cise 6.5 to realize that the characteristic polynomial of a rational canonical
form B is analogous to the order of a finite abelian group and the minimum
polynomial is analogous to the minimal exponent.
In Chapter 8, we will study groups whose elements are nonsingular ma-
trices. Since the order of a group element is the same as the order of any of its
conjugates, the order of such a matrix is the order of its rational canonical
form. It is difficult to compute powers of companion matrices, and so we
introduce another canonical form (when the field of entries is large enough)
whose powers are easily calculated.
Definition. An s x s Jordan block is an s x s matrix of the form
A 1 x 1 Jordan block has the form [a].
Let K denote the n x n matrix consisting of all 0's except for 1's on the first
subdiagonal below the main diagonal; thus, a Jordan block has the form
+
aE K, where E is the n x n identity matrix. Note that K2 is all 0's except
for 1's on the second subdiagonal below the main diagonal, K3 is all 0's
except for 1's on the third subdiagonal, etc., and K" = 0.
 6. Finite Direct Products
140
Lemma 6.20. If J = aE + K is an n x n Jordan block, then
(we agree that
('r) = 0 if i > m.)
Proof. The binomial theorem applies because aE and K commute. The sum
is from 1 to n - 1 because Kn = 0.
Lemma 6.20 is very useful because the matrices Ki are "disjoint." For
example,
and
Theorem 6.21. If A is an n x n matrix over a field 12 which contains all the
eigenvalues of A, then A is similar to a direct sum of Jordan blocks.
Proof. Theorem 6.19(i) shows that it suffices to prove that a companion
matrix C(f), with f(x) a power of an irreducible polynomial, is similar to a
Jordan block. The hypothesis on k gives f(x) = (x - a)" for some a E k. Let
W be the subspace with basis {v, Tv, T2v, ..., Ts-'v), where T is the linear
transformation arising from the companion matrix C(f). Consider the sub-
set ~28= {u,, u,, ..., us-,) of W , where u, = v, u, = ( T - aE)v, ..., us-, =
( T - aE)"-'v. It is plain that 9i? spans W, for Tiv E (u,, .. . , ui) for all i; since
I9lI = s, it follows that 9l is an ordered basis of W.
+
Let us compute the matrix J of T relative to 9l.If j 1 5 s,
If j + 1 < s, then Tuj = auj + uj+,; if j + 1 = s, then ( T - ar~)j+'= ( T - ME)"=
0, by the Cayley-Hamilton theorem (Lemma 6.18 identifies f (x) = (x - a)I
as the characteristic polynomial of C(f), hence of T). Therefore Tu,-, =
aus-,. The matrix J is thus a Jordan block; it is similar to C(f) because both
represent the same linear transformation relative to different ordered bases
of W.
Canonical Forms; Uniqueness 141
Definition. A Jordan cnrtorticnlfim is a matrix 3 that is a direct sum of
Jordan blocks J,, . . ., J,. Each Ji determines a polynomial gi(x) which is a
power of an irreducible polynomial, and the polynomials g, (x), . . ., g,(x) are
called the elemerztary divisors of B.
Theorem 6.21 thus says that a matrix A is similar to a Jordan canonical
form if the ground field k contains all the eigenvalues of A. In particular, if k
is algebraically closed, then this is always the case.
Canonical Forms; Uniqueness
Our discussion is still incomplete, for we have not yet considered uniqueness;
can a matrix A be similar to several rational canonical forms? Can A be
similar to several Jordan canonical forms? We have used the module ana-
logue of the basis theorem; we are now going to use the module analogue of
the fundamental theorem.
Definition. If V and W are R-modules, then a function q : V -+ W is an
R-luomomorphispaz if
q(v + vl) = q(u) + (~(v')
and
~ ( a v=
) adv)
for all v, v' E V and a E R; if 9 is a bijection, then it is called an R-isomorphism.
Two modules V and W are called R-isornorplric, denoted by V E W, if there
exists an R-isomorphism q : V -+ W.
If R is a field, then an R-homomorphism is an R-linear transformation; if
V and W are abelian groups, then every homomorphism is a E-homomor-
phism. If R = k[x] and V and W are k[x]-modules, then q : V -+ W is a
k-linear transformation such that
q(f(x)v) = f(x)q(v)
for all v E V and f(x) E k[x].
6.21. Prove the first isomorphism theorem for modules. (Hint. Since modules are
abelian groups, the reader need check only that the isomorphism in Theorem
2.24 is an R-homomorphism.)
6.22. Every cyclic R-module V = (v) is R-isomorphic to R/ord(v). Conclude that two
cyclic modules are R-isomorphic if and only if they have generators with the
same order ideal.
6.23. If R is a PID and a, b E R are relatively prime, then R/(ab) 2 R/(a) @ R/(b).
 6. Finite Direct Products
142
Theorem 6.22 (Fundamental Theorem). If R is a PID and V and W are finite
R-modules (i.e., they are finitely generated and every element has finite order),
then V z W if and only if either they have the same invariant factors or the
same elementary divisors.
Proof. Translate Corollary 6.14 into the language of modules.
We continue the analysis of k[x]-modules in order to apply this theorem
to matrices.
Lemma 6.23. Let V and W be vector spaces over a field k, let T: V 4 V and
S: W 4 W be linear transformations, and let V T and W S be the corresponding
k[x]-modules. A function q : V 4 W is a k[x]-homomorphism if and only if q
is a linear transformation such that q(Tv) = Sq(v) for all v E V.
Proof. If q : V T + W S is a k[x]-homomorphism, then q(f(x)v) = f(x)qv for
all v E V and all f(x) E k[x]. In particular, this equation holds for every con-
stant "f(x),
. so that q is a k-linear transformation,
, - .and. for f(x) = x, so that
q(xv) = xqv. But xv = Tv and x(qv) = Sqv, as deared.
The converse is easy: if equality q (f (x)v) = f (x)qv holds for every v E V
whenever f (x) constant or f (x) = x, then it holds for every polynomial.
Theorem 6.24. If A and B are n x n matrices over a field k, then A and B are
similar if and only if the corresponding k[x]-modules they,determine are k[x]-
isomorphic.
Proof. Construct the modules determined by the matrices: let V be the vector
space of all column vectors of n-tuples of elements in k; define T, S: V -+ V by
Tv = Av and Sv = Bv. As usual, write V T to denote V made into a k[x]-
module by xv = T(v) = Av, and write vSto denote V made into a module by
xu = S(v) = Bv.
If A and B are similar, then there is a nonsingular matrix P with PAP-' =
B. Now P defines an invertible linear transformation q: V 4 V by qv = Pv.
We claim that q is a k[x]-isomorphism. By Lemma 6.23, it suffices to show
that q T = Sq, and this follows from the given matrix equation P T = SP.
Conversely, assume that there is a k[x]-isomorphism q : VT.+ V S . By
Lemma 6.23, q T = Sq; since rp is a bijection, qTq-' = S. If P is the matrix
of q relative to the standard basis of i/, this gives PAP-' = B, and so A and
B are similar. El
Theorem 6.25. Two n x n matrices A and B over a field k are similar if and
only if they have the same invariant factors. Moreover, a matrix is similar to
exactly one rational canonical form.
Canonical Forms; Uniqueness 143
Prooj: Only necessity needs proof. Since similarity is an equivalence relation,
we may assume that both A and B are rational canonical forms. By Theorem
6.24, A and B are similar if and only if V Tand vSare k[x]-isomorphic (where
T(v) = Av and S(v) = Bv). Recall that the invariant factors are monic polyno-
mials A(x) with f, (x)lf,(x) 1. . .I&(x);if V T = Ci, where Ci is a cyclic module
with order ideal (.&(x)),then Theorem 6.22 says that the invariant factors are,
indeed, invariant; that is, they do not change after an isomorphism is applied
to VT. Therefore, V T and V Shave the same invariant factors. But the invari-
ant factors of A and of B are just the polynomials determined by the last
columns of their companion matrices, and so A and B are equal.
lB4
Theorem 6.26. Two n x n matrices A and B over a field k containing their
eigenvalues are similar if and only if they have the same elementary divisors.
Moreover, if a matrix is similar to Jordan canonical forms J and J', then J and
J' have the same Jordan blocks.
Proof. The proof is essentially the same as that of Theorem 6.25, with com-
panion matrices replaced by Jordan blocks.
Note that the rational canonical form of a matrix A is absolutely unique,
whereas the Jordan canonical form is unique only up to a permutation of the
Jordan blocks occurring in it.
This discussion of canonical forms has the disadvantage of not showing
how to compute the invariant factors of any particular matrix, and a Linear
Algebra course should include a discussion of the Smith casronicalformwhich
provides an algorithm for displaying them; see Cohn (1982). Let A be an
n x n matrix over a field k. Using elementary row and column operations
over the ring k[x], one can put the matrix xE - A into diagonal form (d,(x),
..., d,,(x)), where each di(x) is a monic polynomial or 0, and d,(x)ld,(x)(. . .
Id,(x). The invariant factors of A turn out to be those di(x) which are neither
constant nor 0.
There are shorter proofs of Theorems 6.25 and 6.26, involving matrix com-
pu tations; for example, see Albert (194 1).
6.24. If A is a companion matrix, then its characteristic and minimum polynomials
are equal.
6.25. (i) Give an example of two nonisomorphic finite abelian groups having the
same order and the same minimal exponent.
(ii) Give an example of two complex n x n matrices which have the same char-
acteristic polynomials and the same minimum polynomials, yet which are
not similar.
144 6. Finite Direct Products
6.26. If b and b' are nonzero elements of a field k, then
are similar.
6.27. Let A and B be n x 11 matrices with entries in a field Ic. If fc is a subfield of a field
K, then A and B are similar over k if and only if they are similar over K. (Hint.
A rational canonical form for A over k is also a rational canonical form for A
over K.)
6.28 (Jordan Decompositions). Let k be an algebraically closed field.
+
(i) Every matrix A over k can be written A = D N, where D is diagonalizable
(i.e., similar to a diagonal matrix), N is nilpotent, and DN = ND. (It may be
shown that D and N are unique.)
(ii) Every nonsingular matrix A over k can be written A = DU, where D is
diagonalizable, U is ztnipotent (i.e., U - E is nilpotent), and DU = UD. (It
may be shown that D and U are unique.) (Hint: Define U = E + ND-I.)
The Mrull-Schmidt Theorem
If a (not necessarily abelian) group is a direct product of subgroups, each of
which cannot be decomposed further, are the factors unique to isomorphism?
The Krull-Schmidt theorem, the affirmative answer for a large class of groups
(which contains all the finite groups), is the main result of this section. This is
another instance in which the name of a theorem does not coincide with its
discoverers. The theorem was first stated by J.M.H. Wedderburn in 1909, but
his proof had an error. The first correct proof for finite groups was given by
R. Remak in 1911, with a simplification by O.J. Schmidt in 1912. The theo-
rem was extended to modules by W. Krull in 1925 and to operator groups by
Schmidt in 1928 (we shall discuss operator groups in the next section).
Let us return to the multiplicative notation for groups.
Definition. An e~ldomorpkismof a group G is a homomorphism rp: G -+ G.
There are certain endomorphisms of a group G that arise quite naturally
when G is a direct product (see Exercise 2.73).
Definition. If G = HI x - . x H,,,, then the maps ni: G 4 Hi, defined by
ni(hl . . .h,) = hi, are called p~ojecdio~ts.
If the inclusion Hi c,G is denoted by Ai, then the maps hixi are endomor-
phisms of G. Indeed, they are idempotent: Aini o A i q = Aini.
Definition. An endomorphism (p of a group G is normalif rp(axa-') = am(x)a-'
for all a, x E G.
The Krull-Schmidt Theorem 145
It is easy to see that if G is a direct product, then the maps Aixi are normal
endomorphisms of G.
Lemma 6.27.
(i) If q~ and are normal endomorphisms of a group G, then so is their
composite rp 0 $.
(ii) If rp is a normal endomorphism of G and if H 4 G, then q(H) o G.
(iii) If rp is a normal autornorphism of a group G, then ip-' is also normal.
Proof. The proofs are routine calculations.
Here is a new way to combine endomorphisms of a group; unfortunately,
the new function is not always an endomorphism.
Definition. If rp and $ are endomorphisms of a group G, then rp + $: G 4 G
is the function defined by x I+ rp(x)$(x).
+
If G is abelian, then rp $ is always an endomorphism. If G = S,, y, is
+
conjugation by (1 2 3), and I) is conjugation by (1 3 2), then (o $ is not an
endoinorphism of S,.
It is easy to see that if rp and $ are normal endomorphisms and if q + $ is
an endomorphism, then it is normal. The equation (i o q)(x)(j o p)(x) = x for
all x E G in Exercise 2.73 may now be written 1, = iq jp. +
Lemma 6.28. Let G = H, x x H,,, have projections ni: G -t Hi and inclu-
sions Ai: Hi c,G. Then the sum of any k distinct Aini is a normal endomor-
phism of G. Moreover, the sum of all the Aini is the identity function on G.
Proof: Note that Aini(h,.. .h,) = hi. If rp = zf=, Aiai (we consider the first k
maps for notational convenience), then rp(h, ...h,) = h, . .. h,; that is, rp = An,
where rr is the projection of G onto the direct factor HI x x H, and A is
the inclusion of HI x . - - x H, into G. It follows that rp is a normal endomor-
phism of G and, if k = mythat rp = 1,.
Definition. A group G is indecomposable if G # 1 and if G = H x K, then
either H = 1 or K = 1.
We now consider a condition on a group that will ensure that it is a direct
product of indecomposable groups (for there do exist groups without this
property).
Definition. A group G has A6C (ascendi~tgcltain condition) if every increasing
chain of normal subgroups stops; that is, if
 6. Finite Direct Productslhe Krull-Schmidt Theorem

, There is thus a strictly decreasing chain of rmmal

is a chain of normal subgroups of G, then there is an integer t for whichlirect of
Kt = Kt+, = K,p, = ;ubgroups of G
A group G has DCC (descending chain condition) if every decreasing chain G=Ho>H,>H2>...;
of normal subgroups stops; that is, if if G has DCC, we have reached a contradiction.
Hl>H22H3>.--. Suppose that G has ACC. Since each Hi is a direct factor of Hi-, , there are
normalsubgroups K i with Hi-, = Hi x K,. There is thus an ascending chain
is a chain of normal subgroups of G, then there is an integer s for whichofnormal subgroups
H, = H,+, = Hs+2= . - ..
A group G has Both chain corrditio~zsif it has both chain conditions! K , < K , x K 2 < K , x K, x K 3 < . . . ,
and we reach a contradiction in this case, too.
Every finite group has both chain conditions. The group Z has ACC but
not DCC; in Chapter 10, we shall meet a group Z(p") with DCC but not Lemma 6-31.Let G have both chain conditions. If is a normal endomorphism
ACC; the additive group of rationals Q has neither ACC nor DCC. of G, then q is an injection if and only if it is a surjection. (Thus, either Property
ensures that q is an automorphism.)
Lemma 6.29.
(i) If H Q G and both H and G/H have both chain conditions, then G jlns botlt ProO/. Suppose that q is an injection and that g # q(G). We prove7 by indue-
chain conditions. In particular, if H and K haue both chain conditio,ls, then tion, that qn(g)$ q " + ' ( ~ )Otherwise,
. there is an element h € G with ~ " ( g=)
so does H x K. pll+l(h), so that q ( ~ ~ ~ ~ = - 'q(qtZ(h)).
( ~ ) ) Since q is an injection, ql"'(g)) =
(ii) If G = H x K and G has both chain conditions,then each H and K has pl1(h): contradicting the inductive hypothesis. There is thus a strictly de-
both chain conditions. creasing chain of subgroups
G > q(G) > q2(G) > -...
$ If GI 2 G2 2 .-.is a chain of normal subgroups of G, then H n G , 2
P ~ Q (i)
H n G2 2 is a chain of normal subgroups of H and HG,/jy 2 HG,/H ... Now 9 normal implies 9 " is normal; by Lemma 6.27, qn(G) Q G for all nyand
is a chain of normal subgroups of G/H. By hypothesis, there is an integer SO the DCC is violated. Therefore, 9 is a surjection.
with H n Gt = H n Gt+, = - -., and there is an integer with HG,/H = Assume that 9 is a surjection. Define I<,= ker (PI';each K , is a normal
HG,+,/H = ; that is, HG, = HG,+, = ... . Let 1 = maxjs, t). B~ the subgroup of G because q" is a homomorphism (the normality of q n is here
Dedekind law (Exercise 2.49), for all i 2 1, irrelevant). There is an ascending chain of normal subgroups

= Gi+l(Hn Gi) = Gi+,(Hn ~i.1) k Gi+,, This chain stops because G has ACC; let t be the smallest integer for which
I<, = I<,+,= K, +,
= .- . We claim that t = 0, which will prove the result. If
and so GZ= Gz+, = ..-. A similar argument holds for ascending chains. 2 1, then there is x E K, x $ K t - , ; that is, <pi(x)= 1 and qi-'(x) f 1.
(ii) If G = H x Kythen every normal subgroup of H is also a normal Since is a surjection, there is g E G with x = q(g). Hence, 1 = qt(x) =
subgroup of G. heref fore, every (ascending or descending) chain of normal qr+l(g)yso that E K,+, = Kt. Therefore, 1 = qt(g) = qt-l(q(g)) = rpt-l(x)y
subgroups of H is a chain of normal subgroups of G, hence must stop. a contradiction. *bus, is an injection.
Lemma 6*30*If has either chain condition, then G is a direct product of a ~ ~ G ~An ie n~~ o im o~r p ~~ i s.m if there is a positive integer
of G is
finite number of indecomposable groups.
k such that q k = 0, where 0 denotes the endomorphism which sends every
Proo$ Call a group good if it satisfies the conclusion of this lemma; call it bad element of G into 1.
otherwise. An indecomposable group is good and, if both A and B are good
groups, then so is A x B. Therefore, a bad group G is a direct product, say, Theorem 6-32 (Fitting's Lemma, 1934). Let G have both chain conditions and
G = U x V, with both U and V proper subgroups, and with U or V bad. let 9 be a normal endomorphism of G. Then G = K x H, where K and H are
Suppose there is a bad group G. Define Ho = G. By induction, for every n, each invariant under q (i.e., rp(K) I K and q(H) 5 qIK is nilpotent, and
there are bad subgroups Ho, H I , . .., H, such that each Hi is a proper bad 9 I H is a surjection.

148 6. Finite Direct Products
ProoJ Let K n = ker q" and let H , = im qn.As in the proof of Lemma 6.31,
there are two chains of normal subgroups of G:
G>H,tH,>... and 1I K , IK21;..*.
Since G has both chain conditions, each of these chains stops: the H,, after t
steps, the K n after s steps. Let 1 be the larger of t and s, so that H, = H,,, =
Hli2 = ... and K , = K,,, = I<,+, = . Define H = H, and K = K,; it is
easy to check that both H and K are invariant under q.
Let x E H n K . Since x E H, there is g E G with x = Y ' ( ~ )since ; x E K,
rol(x) = q l ( x ) = 1, so that g E K,, = K,. Hence, x =
. . . = 1. Therefore, q21(g)
q ~ ' (=~ 1,
) and so H n K = 1.
If g E G, then q ' ( g E ) H, = H,,, so there is y E G with q'(g)= q2'(y). Ap-
olvine
A . - . to g m l ( y - l ) gives 1, so that g q l ( y - ' ) E I<, = K. Therefore, g =
) so G = K x H.
[ ~ ~ q ~ ( y - ~ ) ]EqK~H(, yand
Now q ( H ) = q ( H , ) = q ( q l ( G ) )= qIil(G) = HI+, = H, = H , so that q is a
surjection. Finally, if x E K,then q l ( x )E K n H = 1, and so q l K is nilpotent.
Corollary 6.33. If G is an indecomposable group having both chain conditions,
then every normal endomorphisrn q of G is either nilpotent or an automorphism.
ProoJ By Fitting's lemma, G = K x H with q l K nilpotent and q l H sur-
jective. Since G is indecomposable, either G = I< or G = H. In the first case,
U, is nilpotent. In the second case, U, is surjective and, by Lemma 6.31, (p is an
0
automorphism. E l
Lemma 6.34. Let G be an indecomposable group with both chain conditions,
and let q and $ be normal nilpotent endomorphisms of G. If q + $ is an
endomorphism of G, then it is nilpotent.
+
Proof. We have already observed that if U, $ is an endomorphism, then it
is normal, and so Corollary 6.33 says that it is either nilpotent or an auto-
morphism. If U, + II, is an automorphism, then Lemma 6.27 (iii) says that its
inverse y is also normal. For each x E G, x = ( q + $)yx = qy(x)$y(x), SO
that if we define A = q y and p = $y, then 1, = A + p. In particular, x-' =
A(x-')p(xml) and, taking inverses, x = p(x)A(x);that is, A + p = p + A. The
equation A(A + p) = ( A + p)A (which holds because A + p = lG)implies that
Ap = pa. If follows that the set of all endomorphisms of G obtained from A
and p forms an algebraic system4 in which the binomial theorem holds: for
every integer m > 0,
( A + p Y = CI
(9 qiirn-i.
This algebraic system (called a setnirirlg) is not a commutative ring because additive inverses
need not exist.
The Krull-Schmidt Theorem
149
Nilpotence of q and ij, implies nilpotence of L = q y and p = $y (they cannot
be automorphisms because they have nontrivial kernels); there are thus posi-
+
tive integers r and s with A' = 0 and ps = 0. If m = r s - 1, then either i > r
or m - i t s. It follows that each term Lipm-' in the binomial expansion of
+
(A p)" is 0: if i 2 r, then A' = 0; if rn - i 2 s, then ,urn-' = 0. Hence, (1,)" =
+
(A p))" = 0, and 1, = 0, forcing G = 1. This is a contradiction, for every
indecomposable group is nontrivial. &3
Corollary 6.35. Let G be an indecomposable group having both chain conditions.
i f q,, .. . ,q,, is a set of normal nilpotent endomorphisms of G such that every
+
sum of distinct q ' s is an endomorphism, then q , + -. rp, is nilpotent.
h o o J Induction on n.
Theorem 6.36 (Krull-Schmidt). Let G be a group having both chain conditions.
if
are two decompositions of G into indecomposable factors, then s = t and there
is a reindexing so that Hi E K i for all i. Moreover, given any r between 1 and
s, the reindexing may be chosen so that
G = H1 x x H, x K,+, x x K,.
Remark. The last conclusion is stronger than saying that the factors are
determined up to isomorphism; one can replace factors of one decomposition
by suitable factors from the other.
ProoJ We shall give the proof when r = 1; the reader may complete the proof
for general r by inducton. Given the first decomposition, we must find a
reindexing of the K's so that Hi E K i for all i and G = H , x K , x .-.x K t .
Let 71,: G -,Hi and A,: Hi c,G be the projections and inclusions from the first
decomposition, a n d let oj: G + K j and pj: K j c,G be the projections and
inclusions from the second decomposition. The maps Aini and pjoj are normal
endomorphisms of G.
By Lemma 6.28, every partial sum pjoj is a normal endomorphism of G.
Hence, every partial sum of
is a normal endomorphism of H I . Since lH1= rr, A, is not nilpotent, Lemma
6.29 and Corollary 6.35 give an index j with an automorphism. We
reindex so that n , p , o , A, is an automorphism of H I ; let y be its inverse.
We claim that 0-,A,: H , -, K , is an isomorphism. The definition of y gives
(yn,pl)(O-lLl)= I,,. To compute the composite in the reverse order, let
150 6. Finite Direct Products
8 = o,A, yn,p,: K, -+ K,, and note that O2 = 13:
(the term in brackets is lH1).Now lH1= I,, 0 lH1= yn,p, o, A,yn,p, o, A, =
yn,p, OG,~.,;it follows that I3 # 0 (lest l H 1= 0). Were 8 nilpotent, then 6' = 9
would force 8 = 0. Therefore, 8 is an automorphism of K,, and so O2 = 9
gives I3 = 1 (multiply each side by 8-I). It follows that oil.,: H, -+ K, is an
isomorphism (with inverse yn, pi).
Now o, sends K, x . - -x Kt into 1 while o,A, restricts to an isomorphism
on H,. Therefore
HI n(K, x ... x K,) = 1.
If we define G* = (H,, K, x x Kt) IGythen
If x E G, then x = k,k,.. .kt, where kj E Kj. Since n,p, is an isomorphism, the
map p: G -+ G, defined by x H n,p, (k,)k, .. .k,, is an injection with image
G*. By Lemma 6.31, fl is a surjection; that is, G = G* = H, x K, x . .. x Kt.
Finally,
K , x ... x K, z G/H, r H, x x H,,
so that the remaining uniqueness assertions follow by induction on maxjs, t ) .
6.29. Show that the following groups are indecomposable: Z; Z,,; Q; S,,; I),,,;Q,;
simple groups; nonabelian groups of order p3; A*; the group T of order 12 (see
Theorem 4.24).
6.30. Assuming the Basis Theorem, use the ICrull-Schmidt theorem to prove the
Fundamental Theorem of Finite Abelian Groups.
6.31. If G has both chain conditions, then there is no proper subgroup H of G with
G r H, and there is no proper normal subgroup K of G with GIK r G'.
6.32. Assume that G has both chain conditions. If there is a group H with G x G r
H x H, then G r H. (Hint.Use Lemma 6.29(ii).)
6.33.5 Let G have both chain conditions. If G E A x B and G r A x C, then B z C.
6.34. Let G be the additive group of Z[x]. Prove that G x Z r G x Z x Z, but that
Z*ZxZ.
Definition. A subgroup H I G is subnormal if there is a normal series
R. Hirshon (Atner. Math. Monthly 76 (1969), pp. 1037-1039) proves that if A is finite and B and
C are arbitrary groups, then A x B r A x C implies B r C.
Operator Groups
6.35. (i) Give an example of a subgroup that is not subnormal.
(ii) Give an example of a subnormal subgroup that is not a normal subgroup.
(iii) If G has a composition series and if H is subnormal in G, then G has a
composition series one of whose terms is H.
(iv) A group G has a composition series if and only if it has both chain condi-
tions on subnormal subgroups.
Operator Groups
There is a generalization of modules that allows us to extend some of our
earlier theorems in a straightforward way. We present new results having old
proofs.
Definition. Let Q be a set and let G be a group. Then Q is a set of operators
on G and G is an $2-gronq if there is a function Q x G -+ G, denoted by
(o, g) I-+ og, such that
= (og)(oh)
for all o E Q and g, h E G.
Definition. If G and H are Q-groups, then a function 9 : G -+ H is an $2-map if
9 is a homomorphism such that cp(wg) = o q ( g ) for all o E Q and g E G.
If G is an a-group, then a subgroup H 5 G is an admissale subg~oupif
o h E H for all o E Q and h E H.
EXAMPLE 6.3. If Q = @, then an Q-group is just a group, every homomor-
phism is an Q-map, and every subgroup is admissible.
EXAMPLE 6.4. If G is an abelian group and Q is a ring, then every Q-module
is an a-group, every Q-homomorphism is an Q-map, and every submodule is
admissible.
EXAMPLE 6.5. If Q is the set of all conjugations of a group G, then G is an
SZ-group whose admissible subgroups are the normal subgroups. An R-map
of G to itself is a normal endomorphism. An Q-isomorphism between Q-
groups is called a central isomorpbtism.
EXAMPLE 6.6. If Q is the set of all automorphisms of a group G, then G is an
SZ-group whose admissible subgroups are the characteristic subgroups.
EXAMPLE 6.7. If Q is the set of all endomorphisms of a group G, then G is an
a-group whose admissible subgroups are the fully invariant subgroups.
All the elementary results of Chapter 2 (and their proofs!) carry over for
Q-groups. For example, the intersection of admissible subgroups is admissi-
ble and, if H and I( are admissible subgroups at least one of which is normal,
152 6. Finite Direct Products
then HK is admissible. If H is a normal admissible subgroup of an R-group
G, then the quotient group G/H is an R-group (where one defines w(gH) =
(wg)H). The kernel of an R-map is an admissible normal subgroup, and the
first, hence the second and third, isomorphism theorems hold, as does the
correspondence theorem. The direct product of R-groups H and I< becomes
an R-group if one defines o(h, k) = (oh, wk).
Defioition. An R-group G is Q-simple if has no admissible subgroups other
than 1 and G.
An admissible normal subgroup H of an 0-group G is maximal such if and
only if G/H is R-simple.
If $2is a ring, then an R-module V with no submodules other than 0 and V
is an R-simple group. In particular, when 52 is a field, a one-dimensional
Q-module is R-simple. If R is the set of all conjugations of G, then R-simple
groups are just simple groups in the usual sense. If 0 is the set of all automor-
phisms of a finite group G, then G is a-simple (or characteristically simple) if
and only if it is a direct product of isomorphic simple groups (Theorem 5.26).
We call attention to three generalizations of results from groups to R-
groups (the proofs are routine adaptations of the proofs we have given for
groups).
Definition. Let G be an R-groups. An Q-series for G is a normal series
with each Gi admissible; an Q-composition series is an R-series whose factor
groups are Q-simple.
If V is abelian and R is a ring, then every normal series V = 2 V, 2
. . 2 1/, = 1 in which the V, are submodules is an Q-series; it is an Q-composi-
tion series if each V,/V,+, is Q-simple. In particular, if Q is a field, then V is a
vector space and the factors are one-dimensional spaces.
If R is the set of all conjugations of G, then an Q-series is a normal series
G = Go 2 G, 2 . - 2 G,, = 1 in which each Gi is a normal subgroup of G
(such a normal series is called a chiefseries or a principal series). If R is the set
of all automorphisms of G, then an R-series is a normal series in which each
term is a characteristic subgroup of G. Both the Zassenhaus lemka and the
Schreier refinement theorem carry over for a-groups; it follows that a gener-
alized Jordan-Holder the8rem is true.
Theorem 6.37 (Jordan-HGllder). Every two 92-composition series of an R-group
are equivalent.
One can now prove that if V is a finite-dimensional vector space over a
field k (that is, V is spanned by finitely many vectors), then the size n of a basis
Operator Groups 153
(v,, . . . , v,,) depends only on V. For i 2 0, define subspaces V, of V by =
( v ~ +... ~ ,, v,,), the subspace spanned by ( v , + ~..,., v,,). Then V = Vo 2 Vl 2
. 2 T/, = 0 is an Q-composition series for V (where R = I ) because the fac-
tor spaces are all isomorphic to 12, hence are R-simple. Therefore, the dimen-
sion of V is well defined, for it is the length of an R-composition series of V.
Here are two more applications. Any two chief composition series of a
group have centrally isomorphic factor groups (Q = conjugations), and any
two characteristic series in which the factor groups are products of isomor-
phic simple groups have isomorphic factor groups (Q = automorphisms).
Theorem 6.38 (Fitting's Lemma). Let G be an Q-group having both chain
conditions on admissible subgroups, and let q be an R-endomorphism of G.
Then G = H x K, where H and K are admissible subgroups, cpl H is nilpotent,
and cp 1 K is a surjec: lm.
Here is an application. Let V be a finite-dimensional vector space over a
field k and let T: V -+ V be a linear transformation. Then V = U @ W, where
U and W are T-invariant, TI U is nilpotent, and TI W is nonsingular. We have
talten Q = lc and observed that TI W surjective implies TI W nonsingular. The
matrix interpretation of Fitting's lemma thus says that every n x n matrix
over a field k is similar to the direct sum of a nilpotent matrix and a non-
singular matrix.
An Q-group G is Q-indecornposobIe if it is not the direct product of non-
trivial admissible subgroups.
. Theorem 6.39 (IIgruBB-ScQnmid&). Let G be an Q-group having both chain condi-
tions on admissible subgroups. If
are two decompositions of G into R-indecomposable factors, then s = t and
there is a reindexing so that Hi r Ki for all i. Moreover, given any r between 1
and s, the reindexing may be chosen so that G = H, x - - - x Hr x K,+, x
x K,.

CHAPTER 7
Extensions and Cohomology
A group G having a normal subgroup K can be "factored" into K and G/K.
The study of extensions involves the inverse question: Given K a G and G/K,
to what extent can one recapture G?
The Extension Problem
Definition. If K and Q are groups, then an exferrsion of K by Q is a group G
having a normal subgroup K , z K with GIK, z Q.
As a mnemonic device, K denotes kernel and Q denotes quotient. We think
of G as a "product" of K and Q.
EXAMPLE 7.1. Both Z6 and S, are extensions of Z, by H,. However, Z,is an
extension of b, by Z,, but S, is not such an extension (for S, has no normal
subgroup of order 2).
EXAMPLE 7.2. For any groups K and Q, the direct product K x Q is an
extension of K by Q as well as an extension of Q by K.
The extension problem (formulated by 0. Holder) is to find all exten-
sions of a given group K by a given group Q. We can better understand the
Jordan-Holder theorem in light of this problem. Let a group G have a com-
"position series
G=K,>K1~..->K,-,~Kn=1
The Extension Problem
and corresponding factor groups
Since K,, = I , we have Kt,-, = Q,, but something more interesting happens
at the next stage; K,-Z/K,,i = Q .-,,
so that Kt,-, is an extension of Kt,-, by
Q,-,. If we could solve the extension problem, then we could recapture K,,-,
-, -,
from K,, and Q,, ; that is, from Q,, and Q,,-, . Once we have Kt,-,, we can
attack K,,-, i n a similar manner, for Kt,-,/I<,,-, = Q,,-,. Thus, a solution of
the extension problem would allow us to recapture K t , - , from Q,, Q,,-,, and
Qn-2. Climbing up the composition series to KO= G, we could recapture
G from Q,,, ..., Q,. The group G is thus a "product" of the Qi, and the
Jordan-Holder theorem says that the simple groups Qi in this "factorization"
of G are uniquely determined by G. We could thus survey all finite groups if
we knew all finite simple groups and if we could solve the extension problem.
In particular, we could survey all finite solvable groups if we could solve the
extension problem.
A solution of theextension problem consists of determining from K and Q
all the groups G Is? which G/K zs Q. But what does "determining" a group
mean? We gave two answers to this question at the end of Chapter 1 when
we considered "knowing" a group. One answer is that a multiplication table
for a group G can be constructed; a second answer is that the isomorphism
class of G can be characterized. In 1926,O. Schreier determined all extensions
in the first sense (see Theorem 7.34). On the other hand, no solution is known
in the second sense. For example, given K and Q, Schreier's solution does not
allow us to compute the number of nonisomorphic extensions of K by Q
(though it does give an upper bound).
EXERCISES
7.1. If K and Q are finite, then every extension G of K by Q has order II<IIQI. If G has
n
a normal series with factor groups Q,,, ..., Q,, then IGI = IQil.
7.2. (i) Show that A, is an extension of V by Z 3 .
(ii) Find all the extensions of Z , by V.
7.3. If p is prime, every nonabelian group of order p3 is an extension of Z , by Z , x Z,.
(Hint.Exercise 4.7.)
7.4. Give an example of an extension of K by Q that does not contain a subgroup
isomorphic to Q.
7.5. If (a, b) = 1 and K and Q are abelian groups of orders a and b, respectively, then
there is only one (to isomorphism) abelian extension of K by Q.
7.6. Which of the following properties, when enjoyed by both IC and Q, is also enjoyed
by every extension of K by Q? (i) finite; (ii) p-group; (iii) abelian; (iv) cyclic; (v)
solvable; (vi) nilpotent; (vii) ACC; (viii) DCC; (ix)periodic (every element has finite
order); (x) torsioir-free (every element other than 1 has infinite order).

156 7. Extensions and Cohomology
Automorphism Groups
The coming construction is essential for the discussion of the extension prob-
lem; it is also of intrinsic interest.
Definition. The automorphism group of a group G, denoted by Aut(G), is the
set of all the automorphisms of G under the operation of composition.
It is easy to check that Aut(G) is a group; indeed, it is a subgroup of the
symmetric group S,.
Definition. An automorphism q of G is inner if it is conjugation by some
element of G; otherwise, it is outer. Denote the set of all inner automorphisms
of G by Inn(G).
Theorem 7.1.
(i) (NIC Lemma). If H 5 G, then CG(H)a NG(H)and NG(H)/CG(H)can be
imbedded in Aut(N).
(ii) Inn(G) a Aut(G) and G/Z(G) r Inn(G).
ProoJ (i) If a E G, let ya denote conjugation by a. Define q : NG(H)-+ Aut(H)
by a H yal H (note that yal H E Aut(H) because a E NG(H));q is easily seen to
be a homomorphism. The following statements are equivalent: a E ker q ;
y,lH is the identity on H; aha-' = h for all h E H; a E CG(H). By the first
isomorphism theorem, CG(H)a NG(H)and NG(H)/CG(H)s im q 5 Aut(H).
(ii) If H = G, then NG(G)= G, CG(G)= Z(G), and im q = Inn(G). There-
=
fore, G/Z(G) Inn(G) is a special case of the isomorphism just established.
To see that Inn(G) a Aut(G), take ya E Inn(G) and q E Aut(G). Then
qyoq-' = yqa E Inn(G), as the reader can check. E l
Definitiom. The group Aut(G)/Inn(G) is called the outer automovgrhism group
of G.
7.3. Aut(V) r S3 z Aut(S3).
EXAMPLE
The 4-group V consists of 3 involutions and 1, and so every q E Aut(V)
permutes the 3 involutions: if X = V - {1), then the map q H (PIX is a
homomorphism Aut(V) 4 S, z S,. The reader can painlessly check that this
map is an isomorphism.
The symmetric group S, consists of 3 involutions, 2 elements of order 3,
and the identity, and every q E Aut(S,) must permute the involutions: if Y =
{(l 2), (1 3), (2 3)), then the map q r ql Y is a homomorphism Aut(S,) -+
S, z S,; this map is easily seen to be an isomorphism.
Automorphism Groups
157
-
We conclude that nonisomorphic groups can have isomorphic automor-
phism groups.
EXAMPLE 7.4. If G is an elementary abelian group of order pH,then Aut(G)
Gun7 PI.
This follows from Exercise 2.78: G is a vector space over Z, and every
automorphism is a nonsingular linear transformation.
Let G = (x) be infinite cyclic. If q E Aut(G), then q(x) must be a generator
of G. Since the only generators of G are x and x-', there are only two auto-
morphisms of G, and so Aut(Z) z Aut(G) z Z,. Thus, an infinite group can
have a finite automorphism group.
7.6. Aut(G) = 1 if and only if I GI
EXAMPLE < 2.
It is clear that I GI < 2 implies Aut(G) = 1. Conversely, assume that Aut(G)
= 1. If a E G, then ya = 1 if and only if a E Z(G);it follows that G is abelian.
.-, The function a I-+ a-I is now an automorphism of G, so that G has exponent
2; that is, G is a vector space over Z,. If JGJz 2, then dim G 2 2 and there
exists a nonsingular linear transformation q : G -+G other than 1.
Recall that if R is a ring, then U ( R )denotes its group of units:
U(R) = (r E R: there is s E R with sr = 1 = rs].
Lemma 7.2. If G is a cyclic group of order n, then Aut(G) z U(Z,).
Proof. Let G = (a). If q E Aut(G), then q(a) = ak for some k; moreover, a'
must be a generator of G, so that (kc, n) = 1, by Exercise 2.20, and [k] E U(Z,).
It is routine to show that 69: Aut(G) + U(Zn), defined by O(q) = [k], is an
isomorphism.
Theorem 7.3.
(i) Aut(Z2)= 1; Aut(Z,) r Z,; if m 2 3, then Aut(Z,,) z Z, x Z,,-,.
(ii) If p is an odd prime, then Aut(Z,,) r Z,, where I = (p - l)pm-l.
(iii) If n = pi' ...p,e', where the pi are distinct primes and the ei > 0, then Aut(Z,)
Hi
r Aut(Z,), where q, = pri.
Proof. (i) U(Z2) = 1 and U(Z,) = {[I], [- 11) E Z,. If m 2 3, then the result
is Theorem 5.44.
(ii) This is Theorem 6.7.
(iii) If a ring R = R, x x R, is a direct product of rings (addition and
multiplication are coordinatewise), then it is easy to see that U(R) is the direct
158 7. Extensions and Cohomology
product of groups U(R,) x ... x U(R,); moreover, the primary decomposi-
tion of the cyclic group Z,= Z,, x -.. x Z, is also a decomposition of Z, as
a direct product of rings. !El
Theorem 7.1 suggests the following class of groups:
Definitio~.A group G is complete if it is centerless and every automorphism
of G is inner.
It follows from Theorem 7.1 (ii) that Aut(G) r G for every complete group.
We are now going to see that almost every symmetric group is complete.
Lemma '7.4. An automorphism cp of S, preserves transpositions (cp(z) is a trans-
position whenever z is) if and only if cp is inner.
P r o d If cp is inner, then it preserves the cycle structure of every permutation,
by Theorem 3.5.
We prove, by induction on t 2 2, that there exist conjugations y,, .. .,y,
such that y;l.. .y;'cp fixes (1 2), . . .,(1 t). If n E S,, we will denote q(n) by nv
in this proof. By hypothesis, (1 2)' = (i j) for some i, j; define y, to be conju-
gation by (1 i)(2 j) (if i = 1 or j = 2, then interpret (1 i) or (2 j) as the iden-
tity). By Lemma 3.4, the quick way of computing conjugates in S,,, we see that
(1 2)9 = (1 2)72, and so y,lq fixes (1 2).
Let y,, .. ., y, be given by the inductive hypothesis, so that $ = y;' .. . y;'q
fixes (1 2), . ..,(1 t). Since $ preserves transpositions, (1 t + I)$ = (1 k). Now
+
(1 2) and (1 k) cannot be disjoint, lest [(I 2)(1 t I)]$ = (1 2)$(1 t 1)" +
+
(1 2)(1 k) have order 2, while (1 2)(1 t 1) has order 3. Thus, (1 t l)$ = +
+
(1 k) or (1 t + I)$ = (2 k). If lc 5 t, then (1 t 1)" ((1 2), ..., (1 t)), and
+
hence it is fixed by $; this contradicts J, being injective, for either (1 t I)*
= (I b) = (1 k)* or (1 t + 1)' = (2 k) = (2 k)$. Hence, k > t + 1. Define y,,,
to be conjugation by (k t + 1). Now y,,, fixes (1 2), . .. , (1 t) and (1 t + l)Yt+l
= (1 t + I)$, so that y3, +
.- y i l q fixes (1 2), ...,(1 t I) and the induction
is complete. It follows that y,' . . y;lrp fixes (1 2), . .. , (1 n). But these trans-
positions generate S,, by Exercise 2.9(i), and so y i l ...yilcp is the identity.
Therefore, q = y2 . .y, E Inn(S,).
Theorem 7.5. If n # 2 or n # 6, then S, is complete.
Remark. S, z Z2 is not complete because it has a center; we shall see in
Theorem 7.9 that S, is not complete.
Proof. Let T, denote the conjugacy class in S,, consisting of all products of k
disjoint transpositions. By Exercise 1.16, a permutation in S,, is an involution
if and only if it lies in some T,. It follows that if 8 E Aut(S,), then 8(T1) = T,
for some Ic. We shall show that if n # 6, then [&I # I TI I for k # 1. Assuming
this, then @(TI)= TI, and Lemma 7.4 completes the proof.
Now I T, I = n(n - 1)/2. To count T,, observe first that there are
k-tuples of disjoint transpositions. Since disjoint transpositions commute and
there are k! orderings obtained from any k-tuple, we have
The question whether I T1I = I T,I leads to the question whether there is some
k > 1 such that
(4:) (n - 2)(n - 3)-*.(n- 2k + 1) = k!2k-1.
Since the right side of (*) is positive, we must have n 2 2k. Therefore, for
fixed n,
+
left side 2 (2k - 2)(2k - 3)-..(2k - 2k 1) = (2k - 2)!.
An easy induction shows that if k 2 4, then (2k - 2)! > /~!2~-',and so (*) can
hold only if k = 2 or Ic = 3. When k = 2, the right side is 4, and it easy to see
that equality never holds; we may assume, therefore, that k = 3. Since n 2 2k,
we must have n 2 6. If n > 6, then the left side of (*) 2 5 x 4 x 3 x 2 = 120,
while the right side is 24. We have shown that if n # 6, then ITI I # I T,I for all
k > 1, as desired. @I
Corollary 7.6. If 8 is an outer automorphism of S,, and if z E S, is a transposi-
tion, then 8(z) is a product of three disjoint transpositions.
Proof. If n = 6, then we saw in the proof of the theorem that (*) does not hold
if k # 3. (When k = 3, both sides of (*) equal 24.) Ll%f
Corollary 7.9. If n # 2 or n # 6, then Aut(S,) z S,,.
Proof. If G is complete, then Aut(G) r G.
We now show that S6 is a genuine exception. Recall that a subgroup P;( _<
S, is transitive if, for every pair x, y E X, there exists a E I< with a(x) = y. In
Theorem 3.14, we saw that if H 5 G, then the family X of all left cosets of H
is a G-set (where pa: gH I-+ agH for each a E G); indeed, X is a transitive
G-set: given gH and g'H, then p,(gH) = g'H, where a = g'g-I.
Lemma 7.8. There exists a transitive subgroup K 5 S6 of order 120 which
contains no transpositions.
Proof. If cr is a 5-cycle, then P = <a) is a Sylow 5-subgroup of S,. The Sylow
theorem says that if r is the number of conjugates of P, then r = 1 mod 5 and
r is a divisor of 120; it follows easily that r = 6. The representation of S, on
160 7. Extensions and Cohomology
X, the set of all left cosets of N = h 5 ( P ) ,is a homomorphism p: S, -+ Sx r
S,. Now X is a transitive &-set, by Exercise 4.1 1, and so lker pl I IS,//r =
(S,1/6, by Exercise 3.44(iii). Since the only normal subgroups of S, are S,, A , ,
and 1, it follows that ker p = 1 and p is an injection. Therefore, im p r S, is
a transitive subgroup of S, of order 120.
For notational convenience, let us write K < S, instead of im p 5 Sx.
Now K contains an element a of order 5 which must be a 5-cycle; say, a =
(1 2 3 4 5). If K contains a transposition (i j), then transitivity of K provides
p E K with p(j) = 6, and so p(i j)P-' = (Pi pj) = (1 6) for some 1 # 6 (of
course, E = Pi). Conjugating (1 6) E K by the powers of a shows that K con-
tains (16), (2 6), (3 6), (4 6), and (5 6). But these transpositions generate s,,
by Exercise 2.9(i), and this contradicts K (zS,) being a proper subgroup of
s6.
The "obvious" copy of S, in S, consists of all the permutations fixing 6;
plainly, it is not transitive, and it does contain transpositions.
Theorem 7.9 (KHGQder,8895). There exists an outer automorphisrn of S6.
HProo$. Let K be a transitive subgroup of S, of order 120, and let Y be the
family of its left cosets: Y = {a, K, ...,a,K). If 0: S, -+ S, is the representa-
tion of S6 on the left cosets of K , then lcer 0 < K is a normal subgroup of S6.
But A, is the only proper normal subgroup of S6, so that ker 0 = 1 and 0 is
an injection. Since S6 is finite, 0 must be a bijection, and so 0 E Aut(S,), for
S, z S,. Were 0 inner, then it would preserve the cycle structure of every
permutation in S6. In particular, 0(,,,, defined by 0(,,,: aiK ++ (1 2)aiK for
all i, is a transposition, and hence 0 fixes aiK for four different i. But if 0(,,,
fixes even one left coset, say aiK = (1 2)aiK, then a,l(l 2)ai is a transposi-
tion in K. This contradiction shows that 8 is an outer automorphism, fBl
Theorem 7.16). Aut(S,)/Inn(S,) r Z,,and so IAut(S6)I = 1440.
ProoJ Let T, be the class of all transpositions in S,, and let T, be the class of
all products of 3 disjoint transpositions. If 0 and $ are outer automorphisms
of S,, then both interchange T, and T,, by Corollary 7.6, and so O-l$(T,) =
TI. Therefore, 0-l$ E Inn(S,), by Lemma 7.4, and Aut(S,)/Inn(S,) has
order 2. fBl
This last theorem shows that there is essentially only one outer auto-
morphism 0 of s,; given an outer automorphism 0, then every other such
has the form y0 for some inner automorphism y. It follows that S, has exactly
720 outer automorphisms, for they comprise the other coset of Inn(S6) in
Aut(S,).
Automorphism Groups 161
Definition. A synthernel is asproduct of 3 disjoint transpositions. A pentad is
a family of 5 synthemes, no two of which have a common transposition.
If two synthemes have a common transposition, say, (a b)(c d)(e f ) and
(a b)(c e)(d f), then they commute. It is easy to see that the converse holds:
two commuting synthemes share a transposition.
Lemma 7.111. S, contains exactly 6 pentads. They are:
ProoJ There aqe exactly 15 synthemes, and each lies in at most two pentads.
There are thus at most 6 pentads, for 2 x 15 = 30 = 6 x 5; there are exactly
6 pentads, for they are displayed above. f.3
Theorem 4.12. If (a,, .. .,a,) is a pentad in some ordering, then there is a
unique outer automorphism 8 of S6 with 0: (1 i) ai for i = 2, 3,4, 5,6. More-
over, every outer automorphism of S6 has this form.
ProoJ Let X = ((1 2), (1 3), (1 4), (1 5), (1 6)). If 0 is an outer automor-
phism of S6, then Corollary 7.6 shows that each 0((1 i)) is a syntheme. Since
(1 i) and (1 j) do not commute for i # j, it follows that 0((1 i)) and 0((1 j))
do not commute; hence, Q(X)is a pentad. Let us count the number of possible
functions from X to pentads arising from outer automorphisms. Given an
outer automorphism 0, there are 6 choices of pentad for 0(X); given such a
pentad P, there are 5! = 120 bijections X -,P. Hence, there are at most 720
bijections from X to pentads which can possibly arise as restrictions of outer
automorphisms. But there are exactly 720 outer automorphisms, by Theo-
rem 7.10, and no two of them can restrict to the same bijection because X
gznerates S6. The statements of the theorem follow. Bl
Since every element in S, is a product of transpositions, the information in
the theorem allows one to evaluate Q(P)for every /?E S6.
Co~oIIary7.83. There is an outer automorphism of S6 which has order 2.
A syntheme is a partition of a set X into subsets Pi with lPil = iPjl for all i, j (this term is due to
J.J. Sylvester).
162 -1. Extensiotls and Cohomology
ProoJ Define" E h t ( S 6 ) by
A routine but long calcuation shows that t,b2 = 1. W
Here is another source of (possibly infinite) complete groups.
Theorem '7.14. If G is a simple nonabelian group, then Aut(G) is complete.
ProoJ Let I = Inn(G) a Aut(G) = A. Now Z(G) = 1, because G is simple and
nonabelian, and so Theorem 7.1 gives I r G. Now Z(A) I CA(I)= ( q E A:
qy, = y,cp for all g E G). We claim that C,(I) = 1; it will then follow that A
is centerless. If cpy, = y,rp for all g E G, then y, = (py,rp-' = y,(,,. Therefore,
q(g)g-' E Z(G) = 1 for all g E G, and so cp = 1.
It remains to show that every a E Aut(A) is inner. Now a(I) a A, because
I a A, and so I n a(I) a a(I). But o(I) r I E G is simple, so that either
I n a(I) = 1 or I n a(I) = a(I). Since both I and o(I) are normal, [I, a(I)] 5
I n a(I). In the first case, we have [I, a(I)] = 1; that is, a(I) 5 CA(I)= 1, and
this contradicts a(I) z I. Hence, I n a(I) = a(l), and so a(I) II. This inclu-
sion holds for every a E Aut(A); in particular, @-'(I) < I, and so @(I)= I for
every a E Aut(A). If g E G, then y, E I; there is thus a(g) E G with
The reader may check easily that the function a: G -+ G is a bijection. We
now show that a is an automorphism of G; that is, a E A. If g, lz E G, then
a ( ~ g ~= h )~ ( ~ g=h Ya(gh)-
) On the hand, a ( ~ g ~=
h )a(~g)o(~h)
= Ya(g)Ya(h) =
Ya(s)a(l,);hence a(gh) = a(g)a(h).
We claim that a = Fa, conjugation by a. To this end, define z = a 0 J?,-l.
Observe, for all h E G, that
In Lam, T.Y., and Leep, D.B., Combinatorial structure on the autornorphism group of S6,
Expo. Math. (1993),it is shown that the order of any outer automorphism cp of S6 is either 2,4, 8,
or 10, and they show how to determine the order of cp when it is given, as in Theorem 7.12, in
terms of its values on (1 i), for 2 < i < 6.
Automorphism Groups
Thus, z fixes everything in I. If P E A, then for every g E G,
Py,P-' = t(Py,/l-') (because /3ygP-' EI and z fixes I)
= z(P)y,.(P)-' (because z fixes I).
Hence z(P)P-' E CA(I)= 1, and z(P) = P. Therefore, z = 1, a = T,, and A is
complete.
It follows, for every nonabelian simple group G, that Aut(G) z Aut(Aut(G)).
There is a beautiful theorem of Wielandt (1939) with a similar conclusion. We
know, by Theorem 7.1, that every centerless group G can be imbedded in
Aut(G). Moreover, Aut(G) is also centerless, and so it can be imbedded in its
automorphism group Aut(Aut(G)).This process may thus be iterated to give
the automorphism tower of G:
Wielandt proved, for every finite centerless group G, that this tower is con-
stant from some point on. Since the last term of an autornorphisrn tower is a
complete group, it follows that every finite centerless group can be imbedded
is a complete group. Of course, there is an easier proof of a much stronger
fact: Cayley's theorem imbeds a finite group in some S,, with n > 6, and
Theorem 7.5 applies to show that S,, is complete.
The automorphism tower of an infinite centerless group need not stop after
a finite number of steps, but a transfinite automorphism tower (indexed by
ordinals) can be defined (taking unions at limit ordinals). S. Thomas (1985)
proved, for every centerless group, that this automorphism tower eventually
stops. As in the finite case, the last term of an autornorphism tower is corn-
plete, and so every centerless group can be imbedded in a complete group. It
is shown, in Exercise 11.56, that every group can be imbedded in a centerless
group, and so it follows that every group can be imbedded in a complete
group.
Tbeo~rem7.15. If K a G and K is complete, then K is a direct factor of G; that
is, there is a normal subgroup Q of G with G = K x Q.
ProoJ Define Q = CG(K)= (g E G: gk = kg for all lc E K). Now K n Q <
Z ( K ) = 1, and so K n Q = 1. To see that G = KQ, take g E G. Now y,(K) =
K, because K a G, and so y,lK E Aut(K) = Inn(K). There is thus lc E K with
gxg-' = kxk-' for all x E K. Hence k-'g E n,
CG(x)= CG(K)= Q, and so
g = k(lc-'g) E KQ. Finally, Q a G, for gQg-' = k(k-'g)Q(lc-'g)-'k-I = lcQk-'
(because k-" E Q), and kQk-' = Q (because every element of Q commutes
with k). Therefore, G = K x Q.
The converse of Theorem 7.15 is true, and we introduce a construction in
order to prove it. Recall that if a E K, then La:K -+ K denotes left translation
164 7. Extensions and Cohomology
by a; that is, L,(x) = ax for all a E K. As in the Cayley theorem (Theorem
3.12), K is isomorphic to K' = {La:a E K), which is a subgroup of SK. Simi-
larly, if R,: K -+K denotes right translation by a, that is, R,: x I-+ xu-', then
Kr = {R,: a E K} is also a subgroup of SKisomorphic to K.
The holomorph of a group K, denoted by Hol(K), is the subgroup
Defi'a~nitio~n.
/ 7.6, \Z(K)I I 2. If Z(K) r Z,, then it is isomorphic to a normal subgroup N
of SK generated by K' and Aut(K).
Notice, for all a E K, that R, = La-lyay so that Kr 5 Hol(K); indeed, it is
easy to see that Hol(K) = (Kr, Aut(K)).
Lemma 7.16. Let K be a group.
(i) K' a Hol(K), K' Aut(I<) = Hol(K), and K' n Aut(K) = 1.
(ii) Ho~(K)/K'z Aut(K).
(iii) CH,,(K,(K')= Kr.
Proof. (i) It is easy to see that cpL, (p-' = L,(,,, and that it lies in K' for every
a E K and cp E Aut(K); since Hol(K) = (K', Aut(K)), it follows that I<'a
Hol(K) and that Hol(K) = K' Aut(l<). If a E K, then La(l) = a; therefore,
La E Aut(K) if and only if a = 1; that is, K' n Aut(K) = 1.
(ii) Hol(K)/K1= K' A u t ( ~ ) l Kr ' Au~(K)/(K'n Aut(K)) r Aut(K).
(iii) If a, byx E I<, then LaRb(x)= a(xb-') and R,L,(x) = (ax)bW',so that
associativity gives Kr I c ~ , ~ ( ~ , ( KFor
' ) . the reverse inclusion, assume that
y E Hol(K) satisfies qLa = Lay for all a E K. Now y = L,y, for some b E K
and cp E Aut (K). If x E K, then yL,(x) = L, cp La(x) = bcp(a)cp(x) and Lay(x) =
L,L,cp(x) = aby,(x). Hence, bcp(a) = ab for all a E K; that is, y, = y,-I. It fol-
lows that y(x) = L,cp(x) = b(b-'xb) = xb, and so q = Rb-l E Kr, as desired.
BI
Here is the converse of Theorem 7.15.
Theorem 7.17. If a group K is a direct factor whenever it is (isomorphic to) a
normal subgroup of a group, then K is complete.
Proof. We identify I< with the subgroup K' < Hol(K). Since K' is normal, the
hypothesis gives a subgroup B with Hol(K) = K' x B. Now B 5 CH,,(KZ)=
I<', because every element of B commutes with each element of K'. It follows
that if cp E Aut(K) < Hol(K), then cp = LORbfor some a, b E K. Hence, cp(x)
= axb-' for all x E K. But now axyb-' = y, (x)y,(y) = axb-'ayb-l, so that
1 = b-'a; therefore, y, = y, E Inn(K).
Since Hol(K) = K' x B and B < Kr 5 Hol(K), Exercise 7.16 below shows
that Kr = B x (Kr n K'). If y, E Kzn Kr, then cp = La = R,, for a, b E K. For
all c E K, L,(c) = R,(C) gives ac = cb-l; if c = 1, then a = b-l, from which it
follows that a E Z(K). Therefore, Kr n K' = Z(Kr)and K z B x Z(K). If 1 #
Automorphism Groups 165
q E Aut(Z(K)), then it is easy to see that @: B x Z(K) -+B x Z(K), defined
(bycpz), is an automorphism of K; (?, must be outer, for conjugation
by (b, z) I-+
by (/I,c) E B x Z(K) z K sends (b, z) into (j,c)(b, z)(/I-', cV1)= (/Ib/I-l, 2).
But K has no outer automorphisms, so that Aut(Z(K)) = 1 and, by Example
of Z4 which is not a direct factor. But K is isomorphic to the normal sub-
group B x N of B x H4 which is not a direct factor, contradicting the
hypothesis. Therefore, Z(K) = 1 and K is complete.
The holomorph allows one to extend commutator notation. Recall that
the commutator [a, x] = axa-'x-' = xax-l. Now let G be a group and let
A = Aut(G). We may regard G and A as subgroups of Hol(G) (by identifying
G with G'). For x E G and a E A, define
[a, XI= a(x)x-l,
and define
[A, GI = ([a, x]: a E A, x E G).
The next lemma will be used to give examples of nilpotent groups arising
naturally.
Lemma 7.18. Let G and A be subgroups of a group H, and let G = Go 2 G, 2
. . be a series of normal subgroups of G such that [A, G,] I G,,, for all i.
Define A, = A and
A, = {a E A: [a, Gi] Gi+,for all i}.
Then [A,, A,] 5 A,,, for all j and 2, and [yj(A), G,] < Gi+, for all i and j.
Pro05 The definition of A, gives [A,, G,] < G,,, for all i. It follows that
[Ajy A,, Gi] = [A,, [ A l , G i l l r [A,, G,,,] < G,,,,,. Similarly, [A,, Ajy Gi] <
Gj+l+i.Now G,+,+, a (G, A), because both G and A normalize each Gi.
Since [A,, A,, G,] [A1, A,, G,] < G,,,,,, the three subgroups lemma (Exercise
5.48 (ii))gives [G,, [A,, At]] = [[A,, A'], Gi] I G,+'+,. Therefore, [A,, All <
A,,,,, for all i, by definition of A,,,. In particular, for i = 0, we have [A,, A,]
< A,,,. It follows that A = A, 2 A, 2 ... is a central series for A, and so
Exercise 5.38(ii) gives yj(A) 5 A, for all j. Therefore, [yj(A), GI 5 [A,, GI <
G, +, as desired. El!
Defiwiticra~a.Let G = Go 2 G, 2 ... T: Gr = 1 be a series of normal subgroups
of a group G. An automorphism a E Aut(G) stabilizes this series if a(G,x) =
G,x for all i and all x E Gi-, . The stabilizer A of this series is the subgroup
A = (a E Aut(G): a stabilizes the series} < Aut(G).
Thus, a stabilizes a normal series G = Go z GI 2 ... 2 Gr = 1 if and only
if a(G,) 5 Gi and the induced map Gi/Gi+, -+ G,/G,+, , defined by Gi+,x t-,
Gi+,a(x),is the identity map for each i.
166 7. Extensions and Cohomology
Theorem 7.19. The stabilizer A of a series of normal subgroups G = Go 2 G ,
2 . . . 2 G, = 1 is a nilpotent group of class I r - 1.
Proof. Regard both G and A as subgroups of Hol(G). For all i, if x E Gi and
a E A, then a(x) = gi+,x for some gi+, E Gi+,, and so a(x)x-% Gi+,. In com-
mutator notation, [A, GI < Gi+,. By Lemma 7.18, [yj(A), Gi] < Gi+jfor all i
and j. In particular, for i = 0 and j = r, we have [y,(A), GI < G, = 1; that is,
for all x E G and a E ?,(A), we have a(x)x-I = 1. Therefore, y,(A) = 1 and A is
nilpotent of class I r - 1.
For example, let (v,, . . ., u,,) be a basis of a vector space V over a field Ic,
and define KC/;:_,
= (vi, ...,on). Hence,
is a series of normal subgroups of the (additive abelian) group K If A 5
GL(V) is the group of automorphisms stabilizing this series, then A is a
nilpotent group of class I n - 1. If each a E A n GL(V) is regarded as a
matrix (relative to the given. basis), then it is easy to see that A n GL(V) =
UT(n, k), the group of all unitriangular matrices. Therefore, UT(n, k) is also
nilpotent of class I n - 1. Compare this with Exercise 5.44.
If G = Go 2 G, > ,-.> G, = 1 is any (not necessarily normal) series of a
group G (i.e., Gi need not be a normal subgroup of Gi-,), then P. Hall (1958)
proved that the stabilizer of this series is always nilpotent of class < $r(r - 1).
7.7. If G is a finite nonabelian p-group, then p2 / IAut(~)l.
7.8. If G is a finite abelian group, then Aut(G) is abelian if and only if G is cyclic.
7.9. (i) If G is a finite abelian group with I GI > 2, then Aut(G) has even order.
(ii) If G is not abelian, then Aut(G) is not cyclic. (Hint. Show that Inn(G) is not
cyclic.)
(iii) There is no finite'group G with Aut(G) cyclic of odd order > 1.
7.10. Show that IGL(2, p)l = (p2 - l)(p2- p). (Hint. How many ordered bases are in
a two-dimensional vector space over Z,?)
7.11. If G is a finite group and Aut(G) acts transitively on G# = G - {I), then G is an
elementary abelian group.
7.12. If H and K are finite groups whose orders are relatively prime, then Aut(H x K)
r Aut(H) x Aut(K). Show that this may fail if (/HI, IKI) > 1. (Hint. Take H =
Z, = K.)
7.13. Prove that Aut(Q) r S,. (Hint. Inn(Q) r V and it equals its own centralizer in
Aut(Q); use Theorem 7.1 with G = Aut(Q) and H = Inn(Q).)
7.14. (i) Show that Hol(Z,) r Z2, Hol(Z,) r S,, Hol(Z,) 2 D,, and Hol(Z,) r D12.
(ii) If P is a Sylow 5-subgroup of S,, then Hol(Z,) r Ns5(P).(Hint. See Exercise
4.20.)
Semidirect Products
7.15. Prove that Aut(D,) z D,, but that AutfD,,) $ Dl,.
7.16. Is Aut(A,) z S,? Is Aut(A,) r S,?
7.17. If G = B x K and B < L < G, then L = B x (LnK).
7.18. If H a G, prove that
( y E Aut(G): y fixes H pointwise and y(g)H = gH for all g E G)
is an abelian subgroup of Aut(G).
7.19. (i) Prove that the alternating groups A,, are never complete.
(ii) Show that if G is a complete group with G # G', then G is not the commuta-
tor subgroup of any group containing it. Conclude that S,,, for n # 2,6, is
never a commutator subgroup.
7.20. If G is a complete group, then Hol(G) = G' x Gr. Conclude, for n # 2 and n # 6,
that Hol(S,,)z S, x S,.
7.21. Prove that every automorphism of a group G is the restriction of an inner
automorphism of Hol(G).
7.22. Let G be a group and let f E S,. Prove that f E Hol(G) if and only if f(xy-'z) =
f (x)f (y)-'f (2) for all x, y, z E G.
Semidirect Products
Definition. Let K be a (not necessarily normal) subgroup of a group G. Then
a subgroup Q I G is a coinplement of I< in G if K n Q = 1 and KQ = G.
A subgroup K of a group G need not have a complement and, even if it
does, a complement need not be unique. In S,, for example, every subgroup
of order 2 serves as a complement to A,. On the other hand,.if they exist,
complements are unique to isomorphism, for
A group G is the direct product of two normal subgroups K and Q if
KnQ=landKQ=G.
Definition. A group G is a semidirect product of K by Q, denoted by G =
K x Q, if K a G and K has a complement Q, z Q. One also says that G
splits over K.
We do not assume that a complement Q, is a normal subgroup; indeed, if
Q, is a normal subgroup, then G is the direct product K x Q,.
In what follows, we denote elements of K by letters a, b, c in the first half
of the alphabet, and we denote elements of Q by letters x, y, z at the end of
the alphabet.
168 7. Extensions and Cohomology Semidirect Products
169

Before we give examples of semidirect products, let us give several different
descriptions of them.
Lemma 7.20. If K is a normal subgroup of a group G, then the following
statements are equivalent:
EXAMPLE 7.10. Let G be a solvable group of order mn, where (myn) = 1. If G
contains a normal subgroup of order In, then G is a semidirect product of K
by a subgroup Q of order n.
This follows from P. Hall's theorem (Theorem 5.28).

(i) G is a semidirect product of K by G/K (i.e., K has a complement in G); EXAMPLE

7.1 1. Aut(S6)is a semidirect product of S, by Z2.
(ii) there is a subgroup Q 5 G so that every element g E G has a unique expres-
sion g = ax, where a E K and x E Q; This follows from P. Hall's theorem (Theorem 5.28).
(iii) there exists n homomorphism s: G/K 4 G with vs = l,,,, where v: G -+
G/K is the natural map; and EXAMPLE 7.12. If G = (a) is cyclic of order 4 and K = (a2), then G is not a
(iv) there exists a homomorphism n: G G with ker n = K and a(x) = x for
-+ semidirect product of K by GIK.
all x E im n (such a map n is called a retraction of G and im n is called a
retract of G). Since normality is automatic in an abelian group, an abelian group G is a

-
P m J (i) (ii) Let Q be a complement of K in G . Let g E G. Since G = KQ,
there exist a E K and x E Q with 4 = ax. If g = by is a second such factoriza-
semidirect product if and only if it is a direct product. But G is not a direct
product. Indeed, it is easy to see that no primary cyclic group is a semidirect
product.
tion, then xy-I = a-'b E K n Q = 1. Hence b = a and y = x.
(ii) =E- (iii) Each g E G has a unique expression g = ax, where a E K and 7.13. Both S3 and Z,are semidirect products of Z3 by Z,.
EXAMPLE
x E Q. If Kg E GIK, then Kg = Kax = Kx; define s: G/K -+ G by s(Kg) = x.
Example 7.13 is a bit jarring at first, for it says, in contrast to direct prod-
The routine verification that s is a well defined homomorphism with vs =
uct, that a semidirect product of K by Q is not determined to isomorphism
lG1,is left as an exercise for the reader.
by the two subgroups. When we reflect on this, however, we see that a semi-
(111)=E- (iv) Define n: G -+ G by n = sv. If x = n(g), then n(x) = n(n(g)) =
direct product should depend on "how" K is normal in G.
svsv(g) = sv(g) = n(g) = x (because vs = I,/,). If a E K, then n(a) = sv(a) =
1, for K = ker v. For the reverse inclusion, assume that 1 = n(g) = sv(g) =
Lemma 7.21. If G is a semidirect product of K by Q, then there is a homomor-
s(Kg). Now s is an injection, by set theory, so that Kg = 1 and so g E K. phisin 6: Q -+ Aut(K), defined by 6, = y,l K; that is, for all x E Q and a E K,
(iv) --- (i) Define Q = im n. If g E Q, then n(g) = g; if g E K, then n(g) = 1;
a fortiori, if g E K n Q, then g = 1. If g E G, then gn(gd') E K = ker n, for %,(a) = xax-'.
n(gn(gd')) = 1. Since n(g) E Q, we have g = [gn(g-')]n(g) E KQ. Therefore, Moreover, for all x, y, 1 E Q and a E K ,
Q is a complement of K in G and G is a semidirect product of K by Q. RI

EXAMPLE
7.7. S, is a semidirect product of A, by Z,.
Proof. Normality of K gives y,(K) = K for all x E Q. The rest is routine.
Take Q = ((1 2)) to be a complement of A,.
Remark. It follows that K is a group with operators Q.
7.8. D2, is a semidirect product of Z, by Z2.
EXAMPLE )
he object of our study is to recapture G from K and Q. It is now clear that
G also involves a homomorphism 6: Q -+ Aut(K).
If D,,,= (a, x), where (a) s Zrf and (x) z Z,, then (a) is normal and
(x) is a complement of (a). Definition. Let Q and K be groups, and let 6: Q Aut(K) be a homomor-
phism. A semidirect product G of K by Q realizes 6 if, for all x E Q and a E K,
EXAMPLE 7.9. For any group K, Hol(K) is a semidirect product of K zby
Aut(K).
In this language, Lemma 7.21 says that every semidirect product G of K by
This is contained in Lemma 7.16. Q determines some 6 which it realizes. Intuitively, "realizing 6" is a way of
 Semidirect Products 171
170 7. Extensions and Cohomology

Pro05 Define OX(a)= xax-I (as in Lemma 7.21). By Lemma 7.20 (ii), each
describing how K is normal in G. For example, if 9 is the trivial map, that is, K and x E Q. Since multiplica-
g E G has a unique expression g = ax with a E
8, = 1, for every x E G, then a = BX(a)= xax-' for every a E K, and so K 5 tion in G satisfies
CG(Q). (ax)(by) = a(xbx-l)xy = abxxy,
Definition. Given groups Q and K and a homomorphism 9: Q -+ Aut(K), it is easy to see that the map K x, Q -+ G, defined by (a, x) I+ ax, is an
.
define G = K x, Q to be the set of all ordered pairs (a, x) E K x Q equipped
with the operation
isomorphism.

(a, x)(b, Y)= (flOX(b),xy). We now illustrate how this construction can be used.

EXAMPLE 7.14. The group T of order 12 (see Theorem 4.24) is a semidirect

Theorem 7.22. Given groups Q and K and a homomorphism 9: Q -+ Aut(K),
product of Z3 by Z,.
then G = K x, Q is a semidirect product of K by Q that realizes 9.
Let Z3 = (a), let Z4 = (x), and define 8: ;2,-+ Aut(Z,) E Z2 by sending a
Proof. We first prove that G is a group. Multiplication is associative: into the generator. In more detail,
ax = a2 and (a2)" = a,
while x2 acts on (a) as the identity automorphism: ax' = a.
The group G = Z3 x,Z, has order 12. If s = (a2,x2) and t = (1, x), then
the reader may check that
The formulas in Lemma 7.21 (K is a group with operators Q) show that the
final entries in each column are equal. s6 = 1 and t2 = s3 =
The identity element of G is (1, I), for which are the relations in T.
(1, x) = (19,(a), lx) = (a, x); EXAMPLE 7.15. Let p be a prime, let K = (a, b) be an elementary abelian
the inverse of (a, x) is ((9x-l(a))-', x-I), for group of order p2, and let Q = (x) be a cyclic group of order p. Define
e: Q 4 A U ~ ( Kr) GL(2, p) by
We have shown that G is a group.
Define a function n: G -+ Q by (a, x) I+ x. Since the only "twist" occurs in
the first coordinate, it is routine to check that n is a surjective homomorphism Thus, ax = ab and bx = b. The commutator axa-' is seen to be b. Therefore,
and that ker a = {(a, 1): a E K J ; of course, ker n is a normal subgroup of G. G = K x, Q is a group of order p3 with G = (a, b, x), and these generators
We identify K with ker n via the isomorphism a t+ (a, 1). It is also easy to satisfy relations
check that {(l, x): x E QJ is a subgroup of G isomorphic to Q (via x I--+ (1, x)), a P = b~ = X B = 1, b = [x, a], and [by a] = 1 = [b, x].
and we identify Q with this subgroup. Another easy calculation shows that
KQ = G and K n Q = 1, so that G is a semidirect product of K by Q. If p is odd, then we have the nonabelian group of order p3 and exponent p;
Finally, G does realize 9: if p = 2, then G r D, (as the reader may check). In Example 7.8, we saw that
) 1)(1, ~ 1 - l= (gX(a),x) (1, x-' ) = ($,(a), 1).
( 1 , ~(a, D, r Z, >a$,; we have just seen here that D, r V xe Z2. A group may thus
have distinct factorizations into a semidirect product.
Since K x, Q realizes 9, that is, BX(b)= xbx-', there can be no confusion if
we write bx = xbx-' instead of OX(b).The operation in K x, Q will henceforth EXAMPLE 7.16. Let p be an odd prime, let I< = (a) be cyclic of order p2, and
be written let Q = (x) be cyclic of order p. By Theorem 7.3, Aut(K) 2 Z,,,-,, s Zp-, x
Zp; indeed, by Theorem 6.9, the cyclic summand Z, = (a), where a(a) = altP.
(a, x)(b, Y)= (abX,xy).
If one defines 9: Q -+ Aut(K) by 8, = a, then the group G = K x, Q has order
p3, generators x, a, and relations xP = 1, ap2= 1, and xax-' = a x = a'+". We
Theorem 7.23. If G is a semidirect product of K by Q, then there exists 9: Q -t have constructed the second nonabelian group of order p3 (see Exercise 4.32).
Aut(i<) with G E I< x,Q.
172 7. Extensions and Cohomology Wreath Products

For each q E Q, define a permutation q* of A x R by

7.23. Show that the group Q, of generalized quaternions is not a semidirect product. 4*(AY 0') = (5 qw'),
7.24. If ] GI = mn, where (m, n) = 1, and if K 5 G has order m, then a subgroup Q I G and define
is a complement of K if and only if IQI.= n. Q* = (4:':: q E Q).
7.25. If k is a field, then GL(n, k) is a semidirect product of SL(n, k) by kX,where
kX= k - (0).
It is easy to see that Q* is a subgroup of SA
by q r--, q*, is an isomorphism.
., and that the map Q -+Q*, given

7.26. If M is the group of all motions of R", then M is a semidirect product of Tr(n, R)
Theorem 7.24. Given groups D and Q, a firzite Q-set 92, atzd a D-set A, then tfze
by 0 ( t z , R).
wreath product D ? Q is isomorphic to the subgroup
7.27. If K and Q are solvable, then K x, Q is also solvable.
7.28. Show that K x,Q is the direct product I< x Q if and only if 8: Q -, Aut(K) is
trivial (that is, 8, = 1 for all x E Q). and hence A x R is a (D 2 Q)-set.
7.29. If p and q are distinct primes, construct all semidirect products of Z, by Z,, and
compare your results to Theorem 4.20. (The condition q l p - 1 in that theorem
Pro05 We show first that K* = (U,., D,*) is the direct product n,,,
It is easy to see that D,*centralizes D,*, for all o' # w, and so D,* a K* for
D.:
should now be more understandable.) every o . Each dz E D: fixes all (A, a ' ) E A x R with o' # w, while each ele-
ment of ( U, D:,) fixes all (A, w) for all A E A. It follows that if d: E
D,* n (U,.#, D:.), then d: = 1.
Wreath Products If q E Q and co E Sn, then a routine computation gives
q*d* "-1 = d*
04 rjw
Let D and Q be groups, let Q be a finite Q-set, and let {D,: o E R) be a family
for each o E Q. Hence q*K*q*-I < I<* for each q E Q, so that K* a W
of isomorphic copies of D indexed by R.
(because W = (I<*, Q*)); it follows that W = K*Q*. T o see that W is a
semidirect product of K* by Q*, it suffices to show that K* n Q* = 1. Now
Definition. Let D and Q be groups, let R be a finite Q-set, and let K =
dz(A, a')= (dA, w') or (A, a'); in either case, d: fixes the second coordinate. If
e n Do, where D, z D for all o E R. Then the wreath product of D by Q,
q* E- Q*, then q*(A, w') = (A, qo') and q* fixes the first coordinate. Therefore,
denoted by D 2 Q (or by D wr Q), is the semidirect product of K by Q, where
any g E K* n Q* fixes every (A, w') and hence is the identity.
Q acts on K by q . (d,) = (d,,) for q E Q and (d,) E Ha.,
Do. The normal
It is now a simple matter to check that the map D 2 Q -+ W, given by
subgroup K of D Z Q is called the base of the wreath product.
(d,)q r--, (d:)q*, is an isomorphism. El
The notation D 2 Q is deficient, for it does not display the Q-set R; perhaps Call the subgroup W of SAX, the pernzrrtstion versiort of D Q; when we
one should write D Z,Q. wish to view D 2 Q acting on A x R, then we will think of it as W.
If D is finite, then 1K1 = I ~ l l ~ifl Q
; is also finite, then ID 2 QI = 1.K x QI =
IKIIQI = IDI'R'IQI.
If A is a D-set, then A x R can be made into a (D Z Q)-set. Given d E D and
o E Q, define a permutation d,* of A x R as follows: for each (A, a ' ) E A x R,
.,
Theorem '7.25. Let D and Q be groups, let R be a firzite Q-set, let A be a D-set,
and let W < SA be the perlnzitatiolz version of D 2 Q.

set (i) If R is a transitive Q-set and A is a tralzsitive D-set, thelz A x R is a

transitive (D Z Q)-set.
(ii) If o E R, then its stabilizer Q, acts orz R - {o). If (A, o ) E A x R arzd
D(A) < D is the stabilizer of A, then the stabilizer PI(,, of (A, o ) is isonzor-
It is easy to see that dzdh* = (dd'):, and so D,: defined by phic to D(A) x (D 2 Q,), and [W: Pt/;, ] = [D : D(A)] [Q : Q,].
D,. = id,*: d E D), Pro05 (i) Let (A, a ) , (A', a')E A x 52. Since D acts transitively, there is d E D
is a subgroup of S,;, indeed, for each o , the map D -+ Dz, given by d H dz, with dA = X;since Q acts transitively, there is q E Q with q o = w'. The reader
is an isomorphism. may now check that q8:d:(A, o ) = (A',a').
174 7. Extensions and Cohomology
(ii) Each element of W has the form (dz)q*, and (dz)q*(A, w) =
(nWrEn
a:.)(& qw) = d,*,(l, qw) = (d,,A, qw). It follows that (d,*)q* fixes
(A,a)if and only if q fixes o and d, fixes A. Let DZ(A) = {d: d E D(A)). Now
D,*(A) is disjoint from (n,.,,
q*d;q*-l - d,,* = d:); hence
D:,. Q:) and centralizes it: if q* E QZ, then
L
r D(A) x (D Z Q,).
It follows that I y,,,,I = JD(A)II lQ,l and
Cw : y,,,)1= l ~ l ' ~ Ql/l~(.l,l
ll lnll"i-llQ,l = [D : CQ : Q,l.
'kltaeo~em'7.26. Wreath product is associative: if both Q and A are finite, if
is a group, and if A is a T-set, then T 1 (D Z Q) G (T Z D) 1 Q.
.;
Proof. The permutation versions of both T Z (D Z Q) and ( T Z D) Z Q are sub-
groups of S A X A X we claim that they coincide. The group T Z (D Z Q) is gener-
ated by all tf*,,,,(for t E T and (A, a)E A x Q) and all f * (for f E D 2 Q). Note
that ts,,,: (8,2 , w') I-+ (td', A', a ' ) if (A', o f ) = (A,w), and fixes it otherwise;
also, f *: (6',;1', w') H (8,f(X, of)). Specializing f * to d,* and to q*, we see
that T Z (D 2 Q) is generated by all tT,,,,, dz, and q**, where dz: (8,A', w') t-,
(8,dX, w') if a'= w, and fixes it otherwise, and q**: (a', A', w') t-, (8,A', qw').
A similar analysis of ( T 2 D) 2 Q shows that it is generated by all q**, d,:
and (t,):, where (t,):: (6', A', o f )I+ (tat, A', w') if w' = w and X = A, and fixes
it otherwise. Since (t,): = tf*,,,,, the two wreath products coincide.
The best way to understand wreath products is by considering graphs.
Definition. A graph r is a nonempty set V, called vertices, together with an
-
adjace~tcyrelation on V, denoted by v u, that is symmetric (v u implies - (d,) EnxEQ
- +
u v for all u, v E V) and irreflexive (v v for all v E V).
One can draw pictures of finite graphs; regard the vertices as points and
join each adjacent pair of vertices with a line segment or edge. Notice that
our graphs are nondirected; that is, one can traverse an edge in either direc-
tion; moreover, there are no "loops"; every edge has two distinct endpoints.
An aastomorplzism of a graph with vertices V is a bijection cp: V -+ V such
that u, v E V are adjacent if and only if ~ ( uand
) cp(v) are adjacent. It is plain
that the set of all automorphisms of a graph T,denoted by Aut(T), is a group
under composition.
For example, consider the following graph I?:
Wreath Products
Figure 7.1
T If cp E Aut(T), then cp fixes vertex 0 (it is the only vertex adjacent to 5 vertices),
cp permutes the "inner ring" R = (1, 2, 3, 4, 51, and, for each i, either cp(n,) =
aPi and cp(b,) = bqi or cp(a,) = bpi and cp(bi) = a,,. It is now easy to see that
IAut(T)I = 2, >< 5!. Regard S , as acting on R and regard S, as acting on
A = (a, b). Identify the outer ring of all vertices (a,, b,: i E R) with A x R by
writing ai as (a, i) and bi as (b, i). If q E S5, then q permutes the inner ring:
q*(ai) = a,, and q*(b,) = b,,; that is, @(a, i) = (a, qi) and q*(b, i) = (b, qi). If
d E S2 and i E Q, then d:(a, i) = (da, i), d:(b, i) = (db, i), while d t fixes (a, j)
and (b, j) for j # i. For example, if d interchanges a and b, then dT(ai) = bi and
dT(bi) = a,, while dT fixes aj and bj for all j # i. Thus, both q* and d: corre-
spond to automorphisms of T.In Exercise 7.30 below, you will show that
Aut(T) S, ? S,.
A special case of the wreath product construction has R = Q regarded as a
Q-set acting on itself by left multiplication. In this case, we write W = D 1, Q,
and we call W the regisbar wre~thproditct.Thus, the base is the direct product
of 1 Ql copies of D, indexed by the elements of Q, and q E Q sends a I Q I-tuple
Dx into (d,). Note that ID ?, Q] = J D ] I ~ ~It~ Qis ]easy
. to see that
the formation of regular wreath products is not associative when all groups
arefinite,for ITl,(D Z,Q)l # I(Tl,D)?,QI.
If Q is an infinite set and (D,:co E R) is a family of groups, then there are
two direct product constructions. The first, sometimes called the coinplete
direct product, consists of all "vectors" (d,) in the cartesian product nwEQ
with "coordinatewise" multiplication: (d,)(dk) = (d,dL). The second, called
D,
the restricted direct product, is the subgroup of the first consisting of all those
(d,) with only finitely many coordinates d, # 1. Both versions coincide when
the index set is finite. The wreath product using the complete direct prod-
uct is called the co~npletewreath grodi[~ct;the wreath product using the re-
stricted direct product is called the restricted wreath prodrrct. We shall see
a use for the complete wreath product at the end of the next section. The
 Wreath Products
176 7. Extensions and Cohomology

the highest power of y dividing m!,for

first example of a (necessarily infipite) centerless p-group was given by D.H.
McLain (1954); it is a restricted wreath product of a group of prime order p m = a, + a l p + a2p2 + . . a + a,pt,
by Z(pm)(the latter group is discussed in Chapter 10; it is the multiplicative and so
group of all pth power roots of unity). McLain's example is thus a p-group
that is not nilpotent. CmIpI + Cm/p21 + Cm/p3] + +.- = (a,+ a,p + a3p2 + . - .+ a,pt-')
What is the order of a Sylow p-subgroup of the symmetric group S,? If + (a, + a3p + a4p2 + . . . + u , ~ ' - ~ )
k 5 m are positive integers, define t = [m/k], the greatest integer in m/k. Thus, +---
k, 2k, ..., tk I mywhile (t + l)k > myso that t is the number of integers i 5 m + (a, + a4p + .. . +
which are divisible by k. If p is a prime, what is the largest power u , of p
dividing m!? Think of m! as factored: m! = 2 x 3 x 4 x -.- x m. By our ini-
tial remark, [mnlp] factors of m! are divisible by p, [m/p2] factors are divisible
by p2, etc. Hence, if m! = ppm', where (m', p) = 1, then Thus, the direct product has the right order, and so it must be a Sylow
p-subgroup of Sx r S,.
For example, let us compute a Sylow 2-subgroup of S, (this has been done
For example, if p = 2, then [m/2] is the number of even integers I my[m/4] by hand in Exercise 4.15 (ii)). In base 2, we have 6 = 0 x 1 + 1 x 2 + 1 x 4.
is the number of multiples of 4 2 m, and so forth. (Notice, for example, that A Sylow 2-subgroup of S2 is Z,; a Sylow 2-subgroup of S, is b, 2 Z,.We
8 = 23 is counted three times by the formula for p.) In particular, if m = p", conclude that a Sylow 2-subgroup P of S, is Z,x (Z, Z Z,). By Exercise 7.31
then the largest power of p dividing p"! is below, Z2 Z Z, r D,, so that P E Z, x D,.

and so the order of a Sylow p-subgroup of the symmetric group S,. is p"").
Theorem 19.27 (Kaloujnine, 1948). If p is a prime, then a Sylow p-subgroup of
S,,, is an iterated regular wreath product Wn = Z, 2, Z, 2;. - 2, Z, of n copies of
Z,,where +, = Wn 2, Z,.
Proof: The proof is by induction on n, the case n = 1 holding because a Sylow
p-subgroup of S, has order p. Assume that n > 1. Let A be a set with pn
elements and let D be a Sylow p-subgroup of S,; thus, A is a D-set. Let
0 = (0, 1, . ..,p - I), and let Q = (q) be a cyclic group of order p acting on
SZ by qi = i + 1 mod p. The permutation version of the wreath product P =
D 2, b, is a subgroup of SAX.; of course, lA x a/ = p"tl. By induction, D is a
wreath product of n copies of Z, and so P is a wreath product of n + 1 copies
of Z,.To see that P is a Sylow p-subgroup, it suffices to see that its order is
pP("+'), where p(n + 1) = pn + pn-I + .-.+ p + 1. Now ]Dl = pp("),so that
/PI = ID 2, Z,( = ( p ~ ( n ) =
) ~pp~ p ( n ) f l = pp(n+l)s
Theorem 7.27 may be used to compute the Sylow p-subgroup of Srnfor any
m (not necessarily a power of p). First write m in base p:
m=ao+a,p+a2p2+...a,p', where O < a i < p - l .
Partition X = {I, 2, . ..,m) into a, singletons, a, p-subsets, a, p2-subsets, . . .,
and a, $-subsets. On each of these pi-subsets Y, construct a Sylow p-sub-
group of S,. Since disjoint permutations commute, the direct product of all
these Sylow subgroups is a subgroup of Sx of order pN, where N = a, +
a2jr(2) + - . + a, p(t) (recall that ,u(i) = pi-' + pi-2 + - . . + p + 1). But pN is
7.30. Prove that Aut(T) z S, 1 S, where r is the graph in Figure 7.1. (Hint. Every
cp E Aut(T) is completely determined by its behavior on the outer ring consisting
of all vertices of the form ai or b,.)
7.31. Prove that 72, 2 Z, r D,. (Hint. Z,2 Z, has several involutions.)
7.32. If both D and Q are solvable, then D 2 Q is solvable.
7.33. Definitiolra. Let D be a (multiplicative) group. A monomial matrix p over D is a
permutation matrix P whose nonzero entries have been replaced by elements of
D; we say that P is the support of p. If Q is a group of n x n permutation
matrices, then
M(D, Q) = (all monomial matrices p over D with support in Q).
(i) Prove that M(D, Q) is a group under matrix multiplication.
(ii) Prove that the subgroup Q r M(1, Q) IM(D, Q).
(iii) Prove that the diagonal M(D, 1) is isomorphic to the direct product
D x - .- x D (n times).
(iv) Prove that M(D, 1) a M(D, Q) and that M(D, Q) is a semidirect product of
M(D71) by MUyQ).
(v) Prove that M(D, Q) r D 2 Q.
7.34. (i) Fix a group Q and a finite Q-set Q. For all groups D and A and all homomor-
phisms f : D -+ A, there is a homomorphism M(f ): M (D, Q) 4 M (A, Q) such
that M(1,) = and, whenever g: A -,B, then M(gf) = M(g)M(f).
(Hint. Just replace every nonzero entry x of a monomial matrix over D by
f(x).) (In categorical language, this exercise shows that wreath product is a
functor.)
 7. Extensions and Cohomology ,
(ii) If D is abelian, show that determinant d: M(D, Q) -+D is a (well defined)
homomorphism.
7.35. If (a, x) E D 2 Q (so that a E K = n D,), then
(a, x)" = (aaxaxZ . ..ax"-', xn).
7.36. Let X = B , u ..- u B, be a partition of a set X in which each Bi has k elements.
If
G = {g E Sx:for each i, there is j with g ( B i ) = B j } ,
then G z Sk t Sm.
Factor Sets
Since there are nonsimple groups that are not semidirect products, our sur-
vey of extensions is still incomplete. Notice the kind of survey we already
have: if we know Q, K , and 0, then we know the semidirect product K x, Q
in the sense that we can write a multiplication table for it (its elements are
ordered pairs and we know how to multiply any two of them).
In discussing general extensions G of K by Q, it is convenient to use the
additive notation for G and its subgroup K (this is one of the rare instances
in which one uses additive notation for a nonabelian group). For example, if
k E K and g E G, we shall write the conjugate of k by g as g + k - g.
Definition.If K 5 G, then a (right) transversal of I< in G (or a complete set of
right coset representatives) is a subset T of G consisting of one element from
each right coset of K in G.
If T is a right transversal, then G is the disjoint union G = U,. K + t. , x(a
Thus, every element g E G has a unique factorization g = k + t for k E K and
t E T. There is a similar definition of left transversal; of course, these two
notions coincide when K is normal.
If G is a semidirect product and Q is a complement of K, then Q is a
transversal of K in G.
Definitiom.If n;: G 4 Q is surjective, then a lgtingr of x E Q is an element
l(x) E G with n;(I(x))= X.
If one chooses a lifting l(x) for each x E Q, then the set of all such is a
transversal of ker n.In this case, the function l: Q -+ G is also called a right
transversal (thus, both 1and its image I(Q) are called right transversals).
Theorem 7.28. Let G be an extension of K by Q, and let 1: Q -+ G be a transver-
sal.
If K is abelian, then there is a homomorphism 8: Q -r Aut(K) with
Factor Sets 179
for every a E K. Moreover, if 1,: Q -, G is another transversal, then
+ +
l(x) a - l(x) = 1,(x) a - l,(x)for all a E K and x E Q.
Proof. Since K a G, the restriction y,lK is an automorphism of K for all
g E G, where y, is conjugation by g. The function p: G -,Aut(K), given by
g I-+ y,jK, is easily seen to be a homomorphism; moreover, K I ker p, for K
being abelian implies that each conjugation by a E K is the identity. There-
fore, p induces a homomorphism p,: G/K -+ Aut(K), namely, Kg r p(g).
The first isomorphism theorem says more than Q z G/K; it gives an ex-
plicit isomorphism A: Q -+G/K: if 1: Q -+ G is a transversal, then A(x) = K +
+
l(x). If 1, : Q -, G is another transversal, then l(x) - 1, (x) E K, so that K l(x)
=K + I, (x)for every x E Q. It follows that A does not depend on the choice
of transversal. Let 0: Q Aut(K) be the composite: 6 = p , A. If x E Q, then
-+
Ox = p# A(x) = p, (K + l(x)) = p(l(x)) E Aut(K); therefore, if a E K,
does not depend on the choice of lifting E(x). E
J
There is a version of Theorem 7.28 that provides a homomorphism 6 when
K is not assumed to be abelian (in this case, 6: Q -+ Aut(K)/Inn(K)), but this
more general situation is rather complicated. Thus, we shall assume that K is
abelian for the rest of this chapter.
A homomorphism 6: Q -,Aut(K) makes K into a Q-set, where the action
is given by xu = @,(a).(For semidirect products, we denoted 6,(a) by ax; since
we are now writing K additively, however, the notation xu is more appropri-
ate.) The following formulas are valid for all x, y, 1 E Q and a, b E K :
+ b) = xa + xb,
Definition. Call an ordered triple (Q, I<, 0) data if K is an abelian group, Q is
a group, and 8: Q -,Aut(K) is a homomorphism. We say that a group G
realizes this data if G is an extension of K by Q and, for every transversal
1: Q -,G,
+
xa = 6,(a) = l(x) a - l(x)
for all x E Q and a E K.
Using these terms, Theorem 7.28 says that when K is abelian, every exten-
sion G of K by Q determines a homomorphism 6: Q --+ Aut(K), and G realizes
the data. The intuitive meaning of 0 is that it describes how K is a normal
subgroup of G. For example, if the abelian group K is a subgroup of the
center Z(G), then 6 is the trivial homomorphism with Ox = 1 for all x E Q (for
180 7. Extensions and Cohomology_ Factor Sets -GI-'-

+
then a = xa = l(x) a - l(x), and a commutes with all l(x), hence with all + +
Prooj: The definition of f gives l(x) l(y) = f(x, y) l(xy). In particular,
g = b + l(x) for b E K). The extension problem is now posed more precisely + +
l(1) l(y) = f(1, y) l(y); since we are assuming that l(1) = 0, we have
as follows: find all the extensions G which realize given data (Q, K, 8). Our f(1, y) = 0. A similar calculation shows that f(x, 1) = 0. The cocycle identity
aim is to write a multiplication table (rather, an addition table!) for all such G. follows from associativity:
Let a: G --+ Q be a surjective homomorphism with kernel K, and choose a
transversal I: Q --+ G with l(1) = 0. Once this transversal has been chosen,
every element g E G has a unique expression of the form on the other hand,

(after all, l(x) is a representative of a coset of K in G, and G is the disjoint = x f (y, 2) + f (xy, z) + 4xyz).
union of these cosets). There is a formula: for all x, y E Q, The cocycle identity follows. El

A more interesting result is that the converse of Theorem 7.29 is true when
because both l(x) + l(y) and l(xy) represent the same coset of K. K is abelian.

Definition. If n: G -+ Q is a surjective homomorphism with kernel K, and if
1: Q -+ G is a transversal with l(1) = 0, then the function f : Q x Q --+ K, de-
fined by (1) above, is called a factor set (or cocycle). (Of course, the factor set
f depends on the transversal 1.)
Consider the special case of a semidirect product G. Theorem 7.23 shows
that G z K x, Q, so that we may assume that G consists of all ordered pairs
(a, x) E K x Q and that multiplication is given by
In additive notation, this becomes
!(a, x) + (b, y) = (a + xb, XY).
If I: Q -+ G is the transversal defined by l(x) = (0, x), then 1 is a homomor-
phism-l(xy) = l(x) + l(y)-and so the factor set determined by this 1 is identi-
cally zero. Thus, one may think of a factor set as a "measure" of G's deviation
from being a semidirect product, for it describes the obstruction to the trans-
versal 1being a homomorphism.
Theorem 7.29. Let n: G 4 Q be a surjective homomorphism with kernel K, let
I: Q --+ G be a transversal with l(1) = 0, and let f : Q x Q --+ K be the corre-
sponding factor set. Then:
(i) for all x, y E Q,
f (1, Y) = 0 = f (x, 1);
(ii) the cocyele identity holds for every x, y, z E Q:
Tileorem 7.30. Given data (Q, I<, 0), a firnction f : Q x Q -+ IC is a factor set if
and only if it satisfies the cocycle identity
as well as f(1, y) = 0 = f(x, 1) for all x, y, z E Q. More precisely, there is an
extension G realizing the data and a transversal I: Q -+ G such that f is the
corresponding factor set.
Pro@$ Necessity is Theorem 7.29:To prove sufficiency, let G be the set of all
ordered pairs (a, x) E K x Q equipped with the operation
(note that if j'is identically 0, then this is the semidirect product K x, Q).
The proof that G is a group is similar to the proof of Theorem 7.22. The
cocycle identity is needed to prove associativity; the identity is (0, 1); inverses
are given by
-(a, X)= (-x-'a - x-'f(x, x-'), x-').
Define n: G --+ Q by (a, x) I--+ x. It is easy to see that n is a surjective homo-
morphism with kernel {(a, 1): k E I<>.If we identify K with ker n via a I--+
(a, l), then K a G and G is an extension of K by Q.
Does G realize the data? We must show, for every transversal 1: Q -+G,
+
that xu = l(x) a - l(x) for all x E Q and a E I<. Now we must have l(x) =
(b, x) for some b E K. Therefore,
l(x) + a - l(x) = (b, x) + (a, 1) - (b, x)
= (b + xa, x) + (-x-'b - x-lf(x, x-l), x-')

= (b + xa + x[-x-'b - x-'f (x, x-')I + f(x, x-l), 1).

182 7. Extensions and Cohomology Factor Sets

Since K is abelian, the last term simplifies to (xa, 1). As any element of K, we follows.
identify xa with (xa, l), and so G does realize the data.
Finally, define a transversal I: Q -+ K by l(x) = (0, x) for all x E Q. The l'(x) + UY)= CMx) + Wl + C ~ ( Y+) I(y)l
factor set F corresponding to this transversal satisfies F(x, y) = l(x) + l(y) - = h(x) + xh(y) + l(x) + l(y) (G realizes the data)
l(xy). But a straightforward calculation shows that F(x, y) = (f(x, y), I), and
so f is a factor set, as desired. = h(x) + W Y ) + f(x9 Y)+ Q ~Y)
= h(x) + xh(y) + f(x, y) - h(xy) + ll(xy).
Notation. Denote the extension G constructed in the proof of Theorem 7.30
by G,; it realizes (Q, K, 6) and it has f as a factor set (arising from the
+ +
Therefore, f '(x, y) = h(x) xh(y) f (x, y) - h(xy). The desired formula fol-
lows because each term lies in the abelian group I<. Bi
trailsversal l(x) = (0, x)).
DefiHlition. Given data (Q, X , 0), a cobounhry is a function g: Q x Q -+ K for
, 6) is the set of all factor sets f:Q x Q -+ K.
Definition. Z 2 ( ~K, which there exists h: Q -,Ii: with h(1) = 0 such that

Theorem 7.30 shows that Z2(Q, K, 0) is an abelian group under pointwise

addition: f + g: (x, y) I--+ f(x, y) + g(x, y). Iff and g are factor sets, then so is
+
f + g (for f g also satisfies the cocycle identity and vanishes on (1, y) and
(x, 1)). If G, and G, are the extensions constructed from them, then G,, is
also an extension; it follows that there is an abelian group structure on the
family of all extensions realizing the data (Q,K , 8) whose identity element is
the semidirect product (which is Go)! This group of all extensions is extrava-
gantly large, however, because the same extension occurs many times. After
all, take a fixed extension G realizing the data, and choose two different
transversals, say, 1 and 1'. Each transversal gives a factor set:
The set of all coboundaries is denoted by B ~ ( QK,
, 6).
It is easy to check that B2(Q, I<, 8 ) is a subgroup of Z2(Q, K, 8); that is,
every coboundary g satisfies the cocycle identity and g(x, 1) = 0 = g(1, x) for
all x E Q. Moreover, Lemma 7.31 says that factor sets f and f ' arising from
different transversals of the same extension satisfy f ' - f E B ~ ( QK,
, 6); that
is, they lie in the same coset of B2(Q,I(,8) in Z2(Q, K, 0). We have been led
to the following group and equivalence relation.
Definition. Given data (Q, I<, 8), then

l'(x) + 1' (y) = f '(x, y) + lf(xy).

Now the factor sets f and f ' are distinct, but both of them have arisen from
the same extension.
Lemma 9.31. Let G be an extension realizing (Q, K, 8), and let 1 and 1' be
transversals with l(1) = 0 = l'(1) giving rise to factor sets f and f ', respectively.
Then there is a function h: Q -+ K with h(1) = 0 such that
for all x, y E Q.
it is called the second cohomology group of the data.
Definition. Two extensions G and G' realizing data (Q, K, 8 ) are equivalent if
there are factor sets f of G and f ' of G' with f ' - f E B2(Q, K, 8); that is, the
factor sets determine the same element of H 2 ( ~I<, , 6).
Were is a characterization of equivalence in terms of the extensions.
A diagram of groups and homomorphisms commutes if, for each ordered
pair of groups G and W in the diagram, all composites of arrows from G to H
are equal."For example,

Proof. For each x E Q, both l(x) and ll(x)are representatives of the same coset
of K in G; there is thus an element h(x) E K with
fl\
B-C
B
.

Since l'(1) = 0 = 1(1), we have h(1) = 0. The main formula is derived as commutes if and only if h = gf.
184 7. Extensions and Cohomology
Theorem 7.32. Two extensions G and G' realizing data (Q, K, 0) are equivalent
if and only if there exists an isomorphism y making the following diagram
commute:
where i and i' are injective, n and n' are surjective, im i = ker n, and im i'
ker n'.
Remark. A homomorphism y making the diagram commute is necessarily an
isomorphism.
ProoJ Assume that G and G' are equivalent. There are thus factor sets f,
f': Q x Q -+I<, arising from liftings I, 1', respectively, and a function
h: Q 4 K with h(1) = 0 such that
(2) f '(x, Y)- f(x, Y)= xh(y) - W Y ) + h(x)
for all x, y E Q. Each element of G has a unique expression of the form
a + l(x), where a E IC and x E Q, and addition is given by
[a + l(x)] + [b + l(y)] = a + xb + f(x, y) + l(xy);
there is a similar description of addition in G'. Define y: G -+ G' by
y(a + l(x)) = a + h(x) + ll(x).
Since l(1) = 0, we have y(a) = y(a + l(1)) = a + h(1) + l'(1) = a, for all a E K,
because h(1) = 0; that is, y fixes K pointwise. Also, x = n(a + l(x)), while
nly(a + l(x)) = nl(a + h(x) + ll(x)) = zl(l'(x)) = X. We have shown that the
diagram commutes. It remains to show that y is a homomorphism. Now
Y ([a i- 4x)l + rb + K Y ) ~=) Y (a + xb + f(x, Y ) + K ~ Y ) )
=a + xb + f(x, y) + h(xy) + ll(xy),
while
+ 4 ~ )+) ~ ( +b 4 ~ )=) [a + h(x) + l'(x)l + Cb + h(y) + l'(y)l
=a + h(x) + xb + xh(y) + fl(x, y) + ll(xy).
The element ll(xy) is common to both expressions, and (2) shows that the
remaining elements of the abelian group K are equal; thus, y is a homomor-
phism.
Conversely, assume that there exists an homomorphism y as in the state-
ment. Commutativity of the diagram gives y(a) = a for all a E K. Moreover,
Factor Sets 185
if x E Q, then x = n(E(x)) = nly(l(x));that is, yl: Q -, G' is a lifting. Applying y
to the equation l(x) i- l ( y ) = f(x, y) + l(xy) shows that yf is the factor set
determined by the lifting yl. But yf(x, y) = f(x, y) for all x, y E Q, because
f(x, y) E K. Therefore yf = f ; that is, f is a factor set of G'. But f ' is also
a factor set of G' (arising from another lifting), and so Lemma 7.31 gives
f ' - f E B ~that
; is, G and G' are equivalent. Eil
Definition. If G is an extension of Pc' by Q, then y E Aut(G) stabilizes the
extension if the diagram in Theorem 7.32 (with G' replaced by G) commutes.
The group A of all such y is called the staBiHizZv of the extension.
=
Theorem 7.33. if I< and Q are (not necessarily abelian) groups and G is an
extension of K by Q, then the stabilizer A of tlze extension is abelian.
ProoJ We give two proofs.
(i) If y E A, then the hypothesis that y makes the diagram commute is
precisely the statement that y stabilizes the series G > K > 1. It follows from
Theorem 7.19 that the group A of all such y is nilpotent of class 5 1; that is,
A is abelian.
(ii)ere is a proof without Theorem 7.19. We first show that if cp E A and
g E G, then gcp(~~)-' E K. Let T be a transversal of K in G, and write g = at
for a E Hi; and t E T. By hypothesis, cp(g) = a't for some a' E K, and g ~ ( ~ )=- '
att-la'-' = aa'-I E K. Next, we show that gcp(g)-I E Z(K). If b E K, then
CgdS)-lY bl = gcp(g)-lbcp(g)g-'b-l
(cp(b) = b since b E K)
=: gcp(g)-'(~(b)c~(g)g-'b-l
The map 0: A -+ n,, 2,(where 2,z Z ( K )for all g E G) defined by O: cp H
(LJ-'~I(s)), has kernel (cp E A: gmlq(g)= 1 for all g E G). Thus, cp E ker O if
and only if ~ ( g=) g for all g E G; that is, q = 1,. Therefore, A is imbedded ,
n
in the abelian group Z,, and hence A is abelian. B
The next theorem summarizes the results of this section.
Theorem '7.341 (Schreier, 1926). There is a bijection from H2(Q, I<, 0) to the set
2 of all equivalence classes of extensions realizing data (Q, K, 0) taking tlze
identity 0 to the class of the semidirect product.
ProoJ Denote the equivalence class of an extension G realizing the data
(Q, K, 0) by [GI. Define cp: H2(QyK , 0) 4 E by (p(f + B2(Q,K, 0)) = [Gf],
where Gf is the extension constructed in Theorem 7.30 from a factor set f.
186 7. Extensions and Cohomology Factor Sets 187

First, cp is well defined, for iff and g are factor sets with f + B' = g + B2, then coordinate of z. Now the action of Q on K is given by z4(o)= ~(qco) for q E Q
f ' - g E BZ,Gf and G, are equivalent, and [GI] = [G,]. Conversely, q is and o E a. It follows that the coth coordinate of z q s the q-loth coordinate
+ +
an injection: if cp(f B2) = cp(g B2), then [G,] = [G,], f - g E B ~ and , of z. Thus,
+ +
f B2 = g B2. Now cp is a surjection: if [GI E E and f is a factor set (074) (0) = a(o)z(q-I a ) ,
(arising from some choice of transversal), then [GI = [Gf] and [GI =
cp( f + B2). The last part of the theorem follows from Exercise 7.39(ii) below: and multiplication in the wreath product is
an extension is a semidirect product if and only if it has a factor set in B ~ . H

By Exercise 1.44, there is a unique group structure on E making cp an Theorem 4.37 (lCaloujniwe and I<rrssner, 1951). If D and Q are groups with Q
isomorphism, namely, [Gf] + [G,] = [Gf +J. finite, then the regular wreath product D ?, Q contains an isomorphic copy of
every extension of D by Q.
Corollary 7.35. H2(e, K, 8) = 0 if and only if every extension G realizing data
(Q, K, 8) is a semidirect product. Remark. The group D may not be abelian.
Pro05 If G is an extension of D by Q, then there is a surjective homomor-
Deffnitiow. If both Q and K are abelian and 8: Q -+ Aut(K) is trivial, then
phism G -,Q with kernel D, which we denote by a F-+ n. Choose a transversal
Ext(Q, K) is the set of all equivalence classes of abelian extensions G of K
1: Q -P 4;.
by Q. For a E G, define a,: Q -,D by
Corolllary 7.36. If both Q and K are abelian and 8: Q -+ Aut(K) is trivial, then aa(x) = ~ ( x ) - ~ a l ( F x ) .
EX~(Q, K) 5 H ~ ( QK,, 8). If a, b E G, then
-
ProoJ Iff: Q x Q -+ K is a factor set, then the corresponding extension Gf is CJ-,(x)cJ-f(x)
= a,(x)ab(a-'x)
--
abelian if and only iff (x, y) = f (y, x) for all x, y E Q: since 0 is trivial, = l(~)~~al(a'x)l(a'x)-~bl(b~~a~~x)

= l(x)-l bl((nb)-'x) = %,(x).

Define q :G -,D 2, Q by
cp(4 = 5) (%7

for every a E G. We see that cp is a homomorphism, for

It is easy to see that the set S2(Q,K, 6 ) of all such "symmetric" factor sets = (a,,5)(%7 6)
(P(~)cp(b)
forms a subgroup of Z2(Q, K, 0), and = (%4,
nb)
= (aab, 2).
Finally, cp is injective. If a E ker cp, then a = 1 and aa(x) = 1 for all x E Q. The
first equation gives a E D s h e second equation gives aa(x)= ~(x)-~al(?x) =
The groups H2(Q, K, 8) and Ext(Q, K) (when Q is abelian and 0 is trivial) 1 for every x E Q. Since a-I = 1, we have l(x)-laE(x) = 1, and so a = 1.
are studied and computed in Homological Algebra and, in particular, in
Cohomology of Grcllps. Remark. Here is a way to view the proof just given. The lifting 1 determines
Here is an interesting use of the cocycle identity. Given a Q-set il, the a factor set f : Q x Q D (which determines the extension G to isomor-
and q E Q acts on (d,) E n, ., n,,,
wreath product D 2 Q is the semidirect product K x Q, where K = n,,,
D, by q(d,) = (dqo).NOWa formal description
D,
-+

phism). For each a E G, fixing one variable gives a function f, = f ( , 5): Q -+

D, namely, f,(x) = l(x)l(~)l(x~)-'.
Since l(E) is almost a, we see that a, is just
of the elements of the base K = D, is as functions a: il -+ D; that is, a a variation off,.
is the /ill-tuple whose wth coordinate is a(w); the multiplication in I< is
(az)(o) = a(o)z(o): the product of the o t h coordinate of a with the o t h Coroltlary 7.38. If is a class of finite grotips closed under stibgrozcps and
188 7. Extensions and Cohomology Theorems of Schur-Zassenhaus and Gaschiitz 189

semidirect products (i.e., if A E % and S < A, then S E %; if A, B E %', then Theorem 7.39. If PC is an abelian normal Hall subgroup of a finite group G (i.e.,
A >a, B E %' for all 8), then %? is closed under extensions. ((K1,[G : I<]) = I), then K has a complement.

Pro05 Assume that G is an extension of D by Q, where both D, Q E %. Since ProoJ Let I I< 1 = m, let Q = GIK, and let 1 Q 1 = n, so that (myn) = 1. It suffices
%?is closed under semidirect products, it is closed under finite direct products; to prove, by Corollary 7.35, that every factor set f : Q x Q -+ K is a cobound-
hence, n,,, (nu,,
D, E %? Since %' is closed under semidirect products, the wreath
product D 2, Q = D,) x Q E %?; since %' is closed under subgroups, the
ary. Define a : Q 4 K by
a(x) = C
f(x, u);
Y ~ Q
theorem gives G E Gk: !#I
a is well defined since Q is finite and K is abelian. Sum the cocycle identity
One may remove the finiteness hypothesis from Theorem 7.37 by using the
complete wreath product.
over all z E Q to obtain

7.37. If H is a subgroup of a group G and if U is a right transversal of H in G, then (as z ranges over all of Q, so does yz). Since (myn) = 1, there are integers s and
U-' = (u-': u E U ) is a left transversal of H in G. Conclude, for H a G, that t with sm + tn = 1. Define h: Q 4 IC by h(x) = ta(x). Then h(1) = 0 and
both U and U-' are transversals of H in G.
7.38. Prove that an extension G of K by Q is a semidirect product if and only if there
is a transversal I: Q -, G that is a homomorphism. But sf(x, y) E K, so that msf(x, y) = 0. Therefore, f is a coboundary. Bl
7.39. (i) Prove that any two semidirect pro2ucts realizing data (Q, K, 8 ) are equiva-
If I< 5 G has complements Q and Q', then Q z Q' (each is isomorphic to
lent.
(ii) Prove that an extension G realizing data (Q, K, 8) is a semidirect product if
GIK). One can say more if the orders of K and Q are relatively prime.
and only if it has a factor set f E B2.
Theorem 13.40. If K is an abelian normal Hall subgroup of a finite group G,
7.40. Let p be an odd prime. Give an example of two extensions G and G' realizing then any two complements of I< are conjugate.
(K, Q, 0) = (Z, Z,,8) with G E G' but with G and G' not equivalent. (Hint.Let
K = (a) be cyclic of order p, and let G = (g) and G' = (h) be cyclic of order
ProoJ Denote IPCl by m and I GIKI by n, so that (myn) = 1. Let Q, and Q, be
p2.Define i: K -+ G by i(a) = pg, and define it: K -, G' by it(a)= 2ph. Show that
there is no isomorphism y: G -t G' making the diagram commute.)
subgroups of G of order n. As we observed above, each of these subgroups is
a complement of K. By Exercise 7.38, there are transversals li: GIK -+ G, for
i = 1, 2, with l,(G/IC) = Qiand with each li a homomorphism. It follows that
Remark. It is plain that JH2(K,Q, 8)Jis an upper bound for the number of the factor sets A determined by li are identically zero. If h(x) is defined by
nonisomorphic extensions G of K by Q realizing 8. This last exercise shows l , ( x ) = h(x) + E2(x),then
that this bound need not be attained. After all, every extension of Z, by Z,
has order P 2 , hence is abelian, so that either G r ZP2or G 2 Z, x Z,; but it
can be shown that (H2(Z,, Z,,8)( = (Ext(Z,, Z,)I = p.
Summing over all y E G/IC gives the equation in K :
0 = xa, - a, + nh(x),
Theorems of Schur-Zassenhaus and Gaschiitz where a, = ~,,,,, h(y). Let sm + tn = 1 and define b, = ta,. Since K has
;exponent m,
We now apply Corollary 7.35 of Schreier's theorem. h(x) = h(x) - smh(x) = -tnh(x) = xta, - ta, = xb, - b,
Recall that a Hall subgroup of a finite group (when it exists) is a subgroup
whose order and index are relatively prime. Also recall Exercise 7.24: If for all x E GIK. We claim that -b, + Q, + b, = Q,. If l,(x) E Q,, then
(GI = mn, where (myn) = 1, and if K 5 G has order mythen a subgroup Q 5 G -bo + l,(x) + b, = -bo + xb, + l,(x) = -h(x) + l,(x) = l,(x) - l,(x) -I-
is a complement of K if and only if ( QI = n. ( ) = (x).
 190 7. Extensions and Cohornology
We now remove the hypothesis that K be abelian.
Theorem 7.41 (Schor-Zassenhaus Lemma, 1937).3A normal Hall subgroup K
of a finite group G has a complement (and so G is a semidirect product of K by
GIIC).
ProoJ Let 1K1 = m and let I GI = mn, where (m, n) = 1. We prove, by induc-
tion on m > 1, that G contains a subgroup of order n. The base step is
trivially true. If K contains a proper subgroup T which is also normal in G,
then KIT a G/T and (G/T)/(K/T) r G/K has order n; that is, KIT is a nor-
mal Hall subgroup of G/T If I KIT1 = m', then m' < m and [GIT :KIT] I = n.
The inductive hypothesis gives a subgroup NIT 5 G/T of order n. Now I N J
= n 1 Tl and (n, I TI) = 1 (for I TI divides m), so that T is a normal Hall sub-
group of N (with / T / < m and with index [N: T ] = n). By induction, N and
hence G contains a subgroup of order n.
We may now assume that K is a minimal normal subgroup of G. If p is a
prime dividing m and if P is a Sylow p-subgroup of K, then the Frattini
argument (Theorem 4.18) gives G = KNG(P). By the second isomorphism
theorem,
G/I< = I<NG(P)/KZ NG(P)/(Kn NG(P))= NG(P)/NK(P),
so that / N,(P)Jn = I N,(P)I I G/KI = ING(P)I.If NG(P)is a proper subgroup of
G, then I NG(P)I< m, and induction shows that NG(P)contains a subgroup of
order n. We may assume, therefore, that N,(P) = G; that is, P a G.
Since I<> P and K is a minimal normal subgroup of G, we have K = P.
Lemma 5.20(ii) now applies: Z(P) char P and P a G imply Z(P) a G. Mini-
mality applies again, and Z(P) = P (Z(P) $. 1 because P is a finite p-group).
But now P = K is abelian, and the proof follows from Theorem 7.39. BI
It follows that if a finite group G has a normal Sylow p-subgroup P, for
some prime p, then P has a complement and G is a semidirect product of P
by G/P.
We can prove part of the generalization of Theorem 7.41 for K nonabelian.
Theorem 7.42. Let K be a normal Hall subgroup of a finite group G. If either
I<or G/K is solvable, then any two complements of I<in G are conjugate.
Remark. The Feit-Thompson theorem says that every group of odd order
is solvable. Since 1 KI and IG/I<I are relatively prime, at least one of them
has odd order, and so complements of normal Hall subgroups are always
conjugate.
Proof. Let I K I = m, let IG/KI = n, and let Q, and Q, be complements of K in
G; of course, Q1 z GIK z Q2.
Assume first that K is solvable. By Lemma 5.20(ii), K' a G; moreover,
Schur (1904) proved this in the special case Q cyclic.
Theorems of Schur-Zassenhaus and Gaschiitz 191
Q,I<'/K1 z Q,/(Q, n K 1 )z Q, (because Q, n Kt 5 Q, n K = 1), so that
lQ,K1/l<'j = n. Now K' < K, because K is solvable. If K' = 1, then K is
abelian and the result is Theorem 7.40; otherwise, IG/KII < IGI, and induc-
tion on I GI shows that the subgroups Q,K1/K' and Q, K'/K1 are conjugate in
G/Kt. Thus, there is i
j E G/K1with g(Q, K'/Kt)g-' = Q2K'/Kt; that is, gQ,g-I
5 Q2K1(where K'g = $7). But K' < K gives jQ, K'I < /GI, and so the sub-
groups gQ,g-' and Q, of order n are conjugate in Q2K1,hence are conjugate
in G.
Assume now that G/IC is solvable. We do an induction on 1 GI that any two
complements of K are conjugate. Let M/K be a minimal normal subgroup of
G/K. Since K 5 M, the Dedekind law (~xercise2.49) gives
(4 M = MnG=MnQiI<=(MnQi)K for i = 1,2;
note also that M n Qi a Qi. Now solvability of G/K gives M/K a p-group for
some prime p, by Theorem 5.24. If M = G, then G/K is a p-group (indeed,
since MIK is a minimal normal subgroup of G/K, we must have / M/K I = p).
Therefore, Q, and Q, (zG/I<) are Sylow p-subgroups of G, and hence are
conjugate, by the Sylow theorem.
We may assume, therefore, that M < G. Now M n Qi is a complement of
I< in M, for i = 1 , 2 (because M = (M n Qi)K, by (*), and (M n Q,) n K 5
Qi n K = 1). By induction, there is x E M 5 G with M n Q, = x(M n Q,)x"
= M n xQ,x-'. Replacing Q, by its conjugate xQ2x-' if necessary, we may
assume that
MnQ, =MnQ,.
If J = n/r n Q, =M n Q,, then J a Qi for i = 1,2, and so
Two applications of the Dedekind law give
and
J[NG(J) n I<]n Qi = J([NG(J)n K] n Qi) = J
(because (NG(J)n K) n Qi I K n Qi = 1). Therefore, Q,/J and Q,/J are com- '
plements of J(NG(J)n K)/J in NG(J)/J. By induction, there is y E N,(J)/J
with Q,/J = y(Q,/J)y-'; it follows that Q, = yQ2y-', where Jy = 7, as
desired. El
The proof of the following theorem is a variation on the proof of Theorem
- 7.39. Recall Exercise 4.16; a normal p-subgroup K of a finite group G is
contained in every Sylow p-subgroup P of G.
Theorem 7.43 (Gipsehiitz, 1952). Let K be a normal abelian p-subgroup of a
finite group G, and let P be a Sylow p-subgroup of G. Then K has a cornplement
in G if and o~zlyif K lzas a cornplement in P.
192 7. Extensions and Cohomology Transfer and Burnside's Theorem 193

Proof. Exercise 7.17 shows that if Q is a complement of K in G, then Q n P is + + +

But y + z = K g K + ti = PC + g + ti = IC tni + q,, so that f(x, y + z)
a complement of K in P. +
= f(x, I< tni + qi) = f(x, K + tni), by (**); since n is a permutation,
For the converse, assume that Q is a complement of K in P, so that Q is a x,, T f (x, y + z) = ~ ( x )Therefore,
.
transversal of K in P. All groups in this proof will be written additively.
If U is a transversal of P in G (which need not be a subgroup!), then
xa(y) - a(x + y) + a(x) = [G: P I f(x, y).
P = U n E a K + q and G = U , G u P + u = U , , u K + q + u ; thus, Q + U = This is an equation in I< I P. Since ([G : PI, If</)= 1, there are integers a
(q + t i : q E Q, u E U} is a transversal of K in G. By Exercise 7.37, +
and b with a lKI b [G : P I = 1. Define lz: G/K -+ K by h(x) = ba(x). Then
- U - Q = - U + Q = ( - u + q: u E U, q E Q} is also a transversal of K in Iz(1) = 0 and
G (-Q = Q because Q is a subgroup). Let us denote - U by T, so that
xh(y) - h(x + Y)+ h(x) = f(x, y);
I TI = [G : PI and T + Q is a transversal of K in G.
Define 1: G/K + G by E(K + t + q) = t + q. The corresponding factor set that is, f is a coboundary and so G is a semidirect product.
f : G/I< x G/K + K is defined by
E(K + t'+ q') + [(I< + t + q) = f(K + t ' + q l , K + t + q)
+ E(K + t' + q' + t + q). 7.41. Use the Schur-Zassenhaus lemma and Exercise 7.29 to reclassify all groups of
order pq, where p and q are distinct primes.
In particular, if t = 0, then
(*) f(x,K+q)=O forall x e G / K , a l l q ~ Q 7.42. Prove that every group of order p2q, where p > q are primes, has a normal
Sylow p-subgroup, and classify all such groups.
(l(K + t' + q' + q) = t' + q' + q = 1(K + t' + q') + I(K + q) because q' + q 7.43. Using Lemma 4.23, reclassify all groups of order 12.
E Q). Consider the cocycle identity
7.44. Use factor sets to prove the existence of the generalized quaternions Q,. (Hint.
xf(y,z) -f(x +Y, 2) + f ( x , y + z ) -f(x,y) = 0
Exercise 4.42.)
for x, y E G/K and z = K + q with q E Q. Equation (*) shows that the first
two terms are zero, and so
( ) f(x, y + z) = f(x, y) for x, y E G/K and z = K + q. Transfer and Burnside's Theorem
Let T = (t,, . .., t,), where n = [G : PI. For fixed y = I( + g E GIK and
for any ti, K + g + ti lies in G/I<; since T + Q is a transversal of K in G, there We have seen several conditions guaranteeing that a group be a semidirect
is tniE T and qi E Q with K + g + ti = K + tni + qi. We claim that n is a product: P. Hall's theorem (Theorem 5.28) for solvable groups; the Schur-
permutation. If K + g + tj = K + tni + qj, then Zassenhaus lemma; Gaschiitz's theorem. In each of these theorems, one starts
+
Q + ti - (tni qi) E K and +
g tj - (tni+ qj) E K with a normal subgroup K of G and constructs a complement Q (rGIK). We
give now aim for a companion theorem, due to Burnside, that begins with a Sylow
subgroup Q of G and, in certain cases, constructs a normal complement K.4
g + ti - qi - tZi+ tni + qj - ti - g E K. It is natural to seek a homomorphism whose kernel is such a normal sub-
Since I< a G, we have ti - qi + qj - tj E K; since Q is a subgroup, -qi + qj group; it is called the transfer. The next lemma is in the spirit of a portion of ,
= q E Q, SO that ti + q - tj E K and K + tj = K + ti + q. It follows that tj = the proof of Gaschiitz's theorem given above.
ti + q, so that j = i (and q = 0). Therefore, 71 is an injection and hence a
permutation. Lemma '7.44. Let Q be a subgroup of finite index n in G, arzd let ( l , , .. ., I,,} and
Let T = {K + t: t E T } For x E G/K, define {h,, . . . , h,) be left transversals of Q in G. For fixed g E G nizd eaclz i, there is
a unique ~ ( i(1) 5 a(i) 5 71) and a unique xi E Q with
i

ghi = lgixi.
a is well defined because G/IC is finite and K is abelian. Summing the cocycle
identity gives Moreover, a is a permutatiorz of (1, . . . , n}.

Some other such theorems are quoted at the end of this section.
194 7. Extensions and Cohomology Transfer and Burnside's Theorem

ProoJ Since the left cosets of Q partition G, there is a unique left coset 4Q The uniqueness assertion of Lemma 7.44 and the definition of j give
containing ghi; the first statement follows by defining ai = j. Assume that
ci = ak = j. Then ghi = I,xi and gh, = I,x,; thus ghix;' = ghkxil, hi1hk =
xr1xkE Q, hiQ = hkQ, and i = k. Therefore, a is an injection of a finite set to Factors may be rearranged in the abelian group Q/Q1:thus
itself, hence is a permutation.

This lemma will be used in two cases. The first has li = hi for all i; that is,
n yiQ' n z S , , i x , i ~ i ~ ' n xaiQ',
= =

because a-'aa S,, and so the inverse of each zi occurs and cancels zi.
E
Finally, n xaiQ1 n xiQ1 since a
the transversals coincide. In this case,
= E We have shown that is indepen-
S,. V
dent of the choice of transversal.
Let g, g' E G and let {El, . . .,1,} be a left transversal of Q in G. Thus, gli =
where a E S,, and xi E Q. The second case has two transversals, but we set luixiand glEi = Eriyi,where xi, yi E Q. Then
g = 1. Now
hj = lajyj,
where a E St,and yj E Q.

Defimm'te'ona.If Q is a subgroup of finite index rz in a group G, then the transfer

is the function V: G -+ Q/Q1 defined by

where {l,, . . ., I,,) is a left transversal of Q in G and gEi = ljxi.

If a subgroup Q of finite index in a group G has a (not necessarily normal)
Retnark. The transfer V: G --+ Q/Qf is often denoted by VG,*, the letter V complement K, then K = {a,, .. . , a,} is a left transversal of Q in G. If b E K ,
abbreviating the original German term Verlagerung. then ba, E K for all i: ba, = a,,. But the general formula (Lemma 7.44) is
bai = aCixi,so that each xi = 1. We conclude that if b E K, then V(b) = 1; that
Theorem 7.45. If Q is a subgroup of finite index in a group G, then the transfer is, K 5 ker V. If Q is abelian, then Q' = 1 and we may identify Q/Q1 with Q;
V: G -+ Q/Qf is a homomorphism whose definition is independent of the choice thus, im V < Q in this case. These remarks indicate that K = ker V is a rea-
of left transversal of Q in G. sonable candidate for a normal complement of Q.
The following formula for the transfer says that V(g) is a product of conju-
Remark. See Exercise 7.45 below which shows that a transfer defined via gates of certain powers of g.
right transversals coincides with V.
Lemma 4.46. Let Q be a subgroup of finite index n in G, and let {I,, . .., I,,) be
ProuJ Let (l,, . . ., I,) and {h,, .. ., h,) be left transversals of Q in G. By a left transversal of Q in G. For each g E G, there exist elements It,, ..., h, of
Lemma 7.44, there are equations for each g E G: G and positive integers n,, . . . , n, (all depending on g) such that:
gli = ICixi a E S,,, xi E Q, (i) each h i € {Il, ..., I,);
(ii) hiignihi E Q;
ghi = hriyi, E Sn, Yi E Qy
(iii) ni = n = [G : Q]; and
hi = laizi, a E S,,, zi E Q. (iv) V(g) = (h;'gnihi)Q'.
,.
Now
ghi = glaizi= laaixaizi. ProoJ We know that gli = 1,,xi, where a E S, and xi E Q. Write the complete
factorization of a as a product of disjoint cycles (so there is one 1-cycle for
Defining j by a j = aai, we have hj = l,aizj, whence each fixed point): a = a, .. .a,. If ai = ( j , , . . . ,j,), then
 196 7. Extensions and Cohomology
and
1 ~ ' g r l j=~ xjv.. .xjl E Q.
Define 12, = ljl and ni = r; all the conclusions now follow. Btl
Theorem 7.47. If Q is an abelian subgroup of finite index n in a group G and
if Q I Z(G), then V(g) = gnfor all g E G.
ProoJ Since Q is abelian, we may regard the transfer as a homomorphism
V: G -+ Q. The condition Q : !Z(G) implies that Q is a normal subgroup of G.
If g E G and h-'grh E Q, then normality of Q gives gr = h(h-'grlz)h-l E Q. But
QI Z(G) now gives h-'grh = gr. The result now follows from formulas (iii)
and (iv) of Theorem 7.45. !!&I
CorolIary 7.48. If a group G has a subgroup Q of finite index n with Q IZ(G),
the11 g C , g" is a homomorphism.
Proof. We have just seen that this function is the transfer. iZ4I
The reader should try to prove this last corollary without using the trans-
fer; I do not know a simpler proof.
Lemma 7.49. Let Q be a Sylow p-subgroup of a finite group G (for some prime
p). If g, h E CG(Q)are conjugate in G, then they are conjugate in NG(Q).
Proof. If h = y-'gy for some y E G, then h E Y-'C,(Q)~ = C,(y-'Qy). Since Q
and yd'Qy are contained in CG(h),both are Sylow subgroups of it. The Sylow
theorem gives c E CG(b)with Q = c-'y-'Qyc. Clearly yc E NG(Q)and c-'y-'gyc
= c-'Jzc = Jz. BY
'Fheorem'7.50 (Burnside Normal Cornpllement Theorem, 1900). Let G be a
finite group and let Q be an abelian Sylow subgroup contained in the center of
its nornzalizer: Q 5 Z(NG(Q)).Then Q has a normal complement K (indeed, K
is even a characteristic subgroup of G).
Proof. Since Q is abelian, we may regard the transfer V as a homomorphism
from G to Q. Let us compute V(g) for each g E Q. By Lemma 7.46, V ( g ) =
n h;'gnihi. For each i, if gni and h;'gnihi lie in Q, then they are conjugate
elements which lie in CG(Q)(for Q abelian implies Q < CG(Q)). By Lemma
7.49, there is ci E NG(Q)with lz;'gnihi = c;'gnici. But Q I Z(NG(Q))implies
c;'gnici = gni. Hence, if n = [G : Q], then V(g) = gn for all g E Q. If lQl = q,
then (n, q) = 1 (because Q is a Sylow subgroup), and there are integers a and
p with an + pq = 1. Therefore, when g E Q, we have g = gangP4= (ga)", so
that V: G -+Q is surjective: if g E Q, then V(ga) = gan= g. The first iso-
morphism theorem gives G/K r Q, where K = ker K It follows that G = KQ
Transfer and Burnside's Theorem 197
and K n Q = 1 (because I I< / = n and so (1KI, 1 Ql) = 1). Therefore, K is a
normal complement of Q. Indeed, K char G because it is a normal Hall
subgroup. El
Dennition. If a Sylow p-subgroup of a finite group G has a normal p-comple-
ment, then G is called p-nilpotent.
Thus, Burnside's theorem says that if Q I Z(NG(Q)),then G is p-nilpotent.
Here are some consequences of Burnside's theorem.
:L .
Theorem 7.51. Let G be a finite group and let p be the smallest prime divisor of
\GI. If a Sylow p-subgroup Q of G is cyclic, then G is p-nilpotent.
Pro05 By Theorem 7.1, there is an imbedding NG(Q)/CG(Q)ct Aut(Q). Obvi-
ously, IN/CI divides I GI. Now Q is cyclic of order pm,say, and so Theorem 7.3
gives IAut(Q)/ = pn'-'(p - 1) (this is true even for p = 2). Since Q is a Sylow
subgroup of G and Q 5 C = CG(Q),p does not divide IN/CI. Hence, IN/CI
divides p - 1. But (p - 1, /GI) = 1, because p is the smallest prime divisor of
I GI, and so IN/CI = 1. Therefore NG(Q)= CG(Q). Since Q I G is abelian,
Q :!Z(CG(Q)),and since NG(Q)= CG(Q),we have Q 5 Z(CG(Q))= Z(NG(Q)).
Thus, the hypothesis of Burnside's theorem is satisfied, and so Q has a normal
complement. !El
Corollary 7.52. A nonabelian simple group cannot have a cyclic Sylow 2-
subgroup.
Remark. We have already seen this in Exercise 3.30.
Theorem 7-53 (HSlder, 1895). If every Sylow subgroup of a finite group G is
cyclic, then G is solvable.
Proo4; If p is the smallest prime divisor of 1 GI, then Theorem 7.5 1 provides a
normal complement K to a Sylow p-subgroup of G. and Q z G/K. By induc-
tion on lG1, G/K is solvable. Since K is solvable (it is cyclic, hence abelian),
Theorem 5.17 shows that G is solvable. El
Corollilgry 7.54. Every group G of squarefiee order is solvable.
A.oo$ Every Sylow subgroup of G must be cyclic.
" Coroillary 7.55. Let G be a nonabelian simple grozip, and let p be the smallest
prime divisor of I GI. Then either p3 divides I G 1 or 12 divides I GI.
P r o d Let Q be a Sylow p-subgroup of G. By Corollary 7.52, Q is not cyclic.
Hence, / Q /2 p2, SO that if p3 does not divide 1Q1, then Q is elementary abelian
of order p2. Since Q is a two-dimensional vector space over Z,,Exercise 7.10
198 7. Extensions and Cohomology
gives IAut(Q)I = (p2 - U(p2 - P) = P(P + I)(P - 02-Now NG(Q)/CG(Q)G
Aut(Q), so that 1 N/CI is a divisor of IAut(Q)j, and I N/CJ # 1 lest Burnside's
theorem apply, Since Q 5 C, p does not divide I N/CI; since p is the smallest
+
prime divisor of /GI, IN/CI must divide p 1. But this is impossible if p is
+
odd, for the smallest prime divisor of I N/CI is > p 2. (We have shown that
if p is odd, then p3 must divide 1 GI.) Moreover, if p = 2, then IAut(Q)I = 6 and
IN/CI = 3, so that /GI is divisible by 22 x 3 = 12. H
The simple group A , has order 60, and 60 is divisible by 12 but not by 8.
There is an infinite class of simple groups, the Suzuki groups, whose orders
are not divisible by 3, hence not by 12 (their orders are divisible by 8). The
magnificent result of Feit and Thompson says that every simple nonabelian
group G has even order, and so I GI is divisible by either 8 or 12.
In Chapter 5, we gave an elementary but ingenious proof of a theorem of
Schur (Theorem 5.32). We now give a straightforward proof of this theorem
using the transfer; indeed, it was Schur who invented the transfer in order to
give the forthcoming proof of Theorem 7.57.
It is not generally true (see Theorem 11.48) that a subgroup of a finitely
generated group G is itself finitely generated.
Lemma 7.56. If G is a finitely generated group and H is a subgroup of G of
finite index, then H is finitely generated.
Proo$ Let G = (g,, . . . , g,) and let t,, ... , tn be right coset representatives of
H with I , = I; thus, G = U;=, Ht,. Enlarging the generating set if necessary,
we may assume that if g, is a generator, then g,' = g, for some k.
For all i, j, there is h(i, j) E H with tigj = h(i, j)t,,i,j,. We claim that N is
generated by all the h(i, j). If a E H, then a = gilgi2...giS;no exponents are
needed since the inverse of a generator is also a generator. Now
= h(1, i,)tltgi2...gis some 1'
. .- gis
= h(1, i1)h(l1,i2)t2,gi3 some 2'
= h(1, i,)h(ll, . gis some 3'
i2)h(2', i3)t3,gi4.. 2 of a group 6;. If C,(E(P)) and NG(J(P))are p-nilpotent, then G is p-izilpoterzt.
= h(1, il)h(ll, i2).. .h((s - I)', is)ti.
Since a and all the h's lie in H, we have t: E H; therefore, ti = t, =
desired. El
Theorem 7.57 (Schur). If Z(G) has finite index in a group G, then G' is finite.
Pro05 As in the proof of Theorem 5.32, G' has a finite number of generators.
Now G'/(G1 n Z(G)) GIZ(G)/Z(G)5 G/Z(G), so that G'/(G1 n Z(G)) is
Transfer and Burnside's Theorem 199
finite. By the lemma, G' n Z(G) is finitely generated. As G' n Z(G) I Z(G) is
abelian, it is finite if it has finite exponent (this is not true for nonabelian
groups). Let V: G -+Z(G) be the transfer. If a E GI, then V(a) = 1; if a E Z(G),
then V(a) = an, where n = [G : Z(G)] (Theorem 7.47). Therefore, G' n Z(G)
has exponent n, and hence is finite. As G' is an extension of one finite group
by another finite group, it, too, is finite.
The reader should be aware of two theorems of Griin whose proofs involve
the transfer; if G is a finite group and P is a Sylow subgroup, then one may
often compute P n G' (see Robinson (1982, pp. 283-256)).
We state some other theorems guaranteeing p-nilpotence.
Theorem (Tate, 11964). Let G be a finite group, and let P I G be a Sylow
p-subgroup of G. If N a G and N n P 2 (D(P), then N is p-nilpoterzt.
ProoJ The original proof is short, using the 5-term exact sequence in co-
homology of groups; a longer proof using the transfer is in Huppert (1967),
p.431. Ell
Call a subgroup H of a group G p-local, for some prime p, if there is some
nontrivial p-subgroup Q of G with H = NG(Q).
Theorem (Frobenikas). A group G is p-nilpotent if and only if euery p-local
subgroup H of G is p-nilpotent.
ProoJ See [Aschbacher, p. 203). Pi!
If p is a prime and P is a Sylow p-subgroup of a group G, define E(P) =
{x E Z(P): x P = 1). If pd(P) is the largest order of an elementary abelian sub-
group of P, then the Tho~npnro~i sasbgrorsp J(P) is defined as the subgroup of G
generated by all the elementary abelian p-subgroups of G of order pd(P).
/
Theorem (Thompson). Let p be an o h priine, and let P be a Sylow p-subgroup ,
Pro05 See [Aschbacher, p. 2031. B
1, as
7.45. Let Q be a subgroup of finite index 11 in a group G, and let { y l , . . . , y,) be a right
transversal of Q in G. For a E G, yia = piy,, for pi E Q and z E S,. Prove that
R: G - - Q/Q1,defined by X(a) = piQ1,is the transfer; that is, R(a) = V(a) for
all a E G. (Hint. Exercise 7.37.)
 Projective Representations and the Schur Multiplier 20 1
200 7. Extensions and Cohomology

Projective Representations and the Schur Multiplier

7.46. (i) Let p, < p2 < .-.< p, be primes and let n = pl . ..p,. Prove that every group
G of order iz has a normal Sylow p,-subgroup. (Hint. Exercise 4.38.) We have already seen the usefulness of H2(Q, K, 0), where K is abelian and
(ii) If, in addition, (pi, pi - 1) = 1 for all i < j, then G must be cyclic. (One
0: Q --+ Aut(K) is a homomorphism. When 0 is triuial, that is, 0(x) = 1, for all
can characterize such integers n by (n, q(n)) = 1, where q is the Euler q -
function.) x E Q, then we drop it from the notation and write H2(Q, K).
(iii) Find all integers n such that every group G of order rz is abelian.
Defiwition. A central extension of K by Q is an extension G of I< by Q with
Remark. We sketch a proof that there exist arbitrarily large sets of primes
I< 5 Z(G).
( p l y ...,p,) which satisfy condition (ii) of Exercise 7.46. Let p, = 3, and sup-
pose that p, < p2 < ,.. < p, are primes with (pi, pj - 1) = 1 for all i < j. A It is easy to see that if G = K x , Q is a central extension, then G is the
theorem of Dirichlet states: If (a, b) = 1, then the arithmetic progression direct product K x Q.
a, a + +
by a 2b, ... contains infinitely many primes. Since (2, p, ...p,) = I,
+
there is a positive integer m such that p,,, = 2 mp, . . .p, is prime. The set Lemma 7.58. Given data (Q, K , %),then 0 is trivial if and only if every exten-
+,
{ p l y.. . ,pt ) satisfies all the desired conditions. sion realizing the data is a central extension.

7.47. (i) Let q and p be primes such that q E 1 mod pe. Show that the multiplicative HPJ.oo$ Recall that B arises from the equation: for all x E Q and a E K,
group Z," contains a cyclic subgroup isomorphic to Z,,. (Hint: Theorem
2.18.)
(ii) Let G = C, x ... x C,, where Ci is cyclic of order pPi for (not necessarily where I(x) is a lifting of x. Assume that 0 is trivial. Every g E G has the form
distinct) primes pi. Use Dirichlet's theorem (see the remark following Exer- g = b + l(x) for some b E K and x E Q. If a E K , then a commutes with l(x) for
cise 7.46) to show that there are n distinct primes qi with qi = 1 mod pfi for all x E Q; since I< is abelian, a commutes with g, and so n E Z(G). Conversely,
i = 1, ..., n.
if G is a central extension, each a E K commutes with every l(x) and so
(iii) If G is a finite abelian group, then G can be imbedded in Z i for some
n
m. (Hint. Let m = qi, and use Theorem 7.3 (iii).) (This proof is due to
G. McCormick.)
$,(a) = a for all x E Q and a E K.

7.48. (i) If V: G -, QIQ' is the transfer, then G' 5 ker V and V induces a homomor-
phism v: GIG' -+ QIQ', namely, G'a H V(a).
(ii) Prove that the transfer is transitive: if P < Q < G are subgroups of finite
index, and if T: G -+ QIQ', U:G -,PIP', and V: Q + PIP' are transfers, then
8 = VT.
Theorem 9.59. There is a bijection from the set of all equivalence classes of
central extensions realizing data (Q, K , 0), where 6 is trivial, to H2(Q, K).
A.oo$ Thzorem 7,34 specializes to this result once we take account of the
lemma. El ,/'
/'

7.49. If Q has index iz in G, and if K 5 G satisfies G = KQ and Q I CG(K), then
V(a) = anQ' for all a E G. (Hint. There is a transversal of Q contained in I<; if
a E K and g E K, then g-larg E Q I CG(K),and this implies a' E C,(I<)~-' =
CG(gfGl-') = CG(K1.1
7.50. If G is a finite group of order mn, where (nz, n) = 1, and if Q I Z(G) is a Hall
subgroup of order my then K = ker V is a normal complement of Q (where
V: G -, Q/Q' is the transfer), and G = K x Q.
7.51. Let G be a torsion-free group having a cyclic subgroup of finite index. Prove
that G is cyclic.
7.52. If p and q are primes, prove that every group of order p2qZis solvable.
7.53. If G is a nonabelian group with /GI < 100 and IGI # 60, then G is solvable. (See
Exercise 4.36.) (The next order of a nonabelian simple group is 168.)
Definition. If Q is a group, then its Schur m t s b ~ l i e r(or
. nzultiplicator) is the
abelian group .
M(Q) = H2(Q, C x 1,
where C x denotes the multiplicative group of nonzero complex numbers.
Since C x is written multiplicatively, it is more convenient to write
M(Q) = H2(Q, C x ) multiplicatively as well. Thus, with 8 trivial, a function
f : Q x Q -,C x is a factor set if and only if, for all x, y E Q:
f ( l , Y) = 1 = f(x, 1);
f (x, y)f (xy,z)-lf(x, yz)f(x, Y)-l = 1;
a function g: Q x Q -+C is a coboundary if and only if there is a function

202 7. Extensions and Cohomology
h: Q -+C with h(1) = 1 such that
g(xY Y)= h(y)h(xy)-'h(x).
Two factor sets f and g are equivalent if and only if fg-I is a coboundary.
Definition. If G is a finite group, its minimal exponent, denoted by exp(G), is
the least positive integer e for which xe = 1 for all x E G.
Theorem 7.60. If Q is a finite group, then M(Q) is a finite abelian group and
exp(M(Q)) divides 1 QI.
Remark. The first paragraph of the proof is just a repetition, in multiplicative
notation, of a portion of the proof given in Theorem 7.39.
Pro05 Iff: Q x Q --+ C is a factor set, define a: Q --+ @ " by
Note that a(1) = 1. Now multiply the cocycle identity
f (Y7 z)f (xy, z)-'f (x, yzlf (x, Y)-' = 1
over all z E Q to obtain
where 12 = IQI (as z rangesrbver all of Q, so does yz). But this equation says
that (f B2(Q, C " ))" = 1; that is, M(Q)" = 1. Thus, the minimal exponent of
M(Q) divides n when Q is finite.
For each x E Q, define h: Q --+ C x by choosing h(1) = 1 and h(x) to be
some nth root of G(x)-': h(x)" = a(x)-'. Define g: Q x Q --+ C x by g(x, y) =
f(x, y)h(y)h(xy)-'h(x). Clearly, f and g are equivalent, for they differ by a
coboundary. On the other hand,
Therefore, each element [f ] E M(Q) determines a function g: Q x Q -+ Z,,
where 2, denotes the subgroup of C x consisting of all the nth roots of unity.
The result follows, for there are only finitely many such functions g. Bl!
Corollary 7-61. I f Q is a finite p-group, then M(Q) is a finite abelian p-group.
The Schur multiplier arises in Representation Theory; more precisely, in
the study of projective representations.
T: -
Definition. A projective representation of a group Q is a homomorphism
Q PGL(n, C) = GL(n, @)/Z(n,@), the group of all nonsingular n x n
Projective Representations and the Schur Multiplier 203
complex matrices modulo its center, the normal subgroup Z(n, C) of all non-
zero scalar matrices.
Just as GL(n, C) consists of automorphisms of a vector space, we shall see
in Chapter 9 that PGL(n, C) consists of automorphisms of a projective space.
An important example of a central extension is provided by GL(n, C), which
is a central extension of C x by PGL(n, C). Note that Z(n, C) r C x .
"Ordinary" representations of Q are homomorphisms Q -+ GL(n, C); they
have been well studied and contain valuable information about Q. The ques-
tion whether a given irreducible representation cp of a normal subgroup H a
Q can be extended to a representation of Q leads, in a natural way, to a
projective representation of Q. If x E Q, one can define a new representation
cpx: H -, GL(n, @) by cpX(h)= ~(xhx-');this makes sense because x11x-' E H
for all 12 E H. To make it easier to extend cp, let us assume (as would be the
case were cp extendable to G) that the representations cp and cpx are all similar;
that is, there is a nonsingular matrix Px with cp(x1zx-') = Pxcp(lz)P;' for all
h E H (of course, we may choose PI to be the identity matrix E). Now choose
a transversal T of H in Q, so that each x E Q has a unique expression of the
form x = th for t E T and h E H. The obvious candidate for an extension
0 : Q 3 GL(n, C) of cp is @(th)= P,cp(h); it is clear that cD is a well defined
function that extends cp. If h' E H and x = th E Q (where t E T and 12 E H),
then
cp(xlifx-' ) = q(thhlh-It-')
It follows that if x, y E Q, then
On the other hand,
Hence, @(xy)-l@(x)@(y) centralizes cp(hl)for all 12' E H. Now Schur's lemma (!)
states that if q is irreducible, then any matrix A centralizing every cp(hl)
must be a scalar matrix. Therefore, @(xy)-'@(x)@(y)is a scalar; that is, there
is f(x, y) E C with
where E is the 12 x n identity matrix. This equation says that @ defines a
homomorphism Q -, PGL(n, C), namely, x t-, @(x)Z(n,C).
It will be more convenient for the coming computations to think of projec-
204 7. Extensions and Cohomology
tive representations z: Q -+ PGL(n, C) not as homomorphisms whose values
are cosets mod Z(n, 6 ) but, as they have just occurred above, as matrix-
valgedfianctiorrs T satisfying the equation
for all x, y E Q, where f,: Q x Q -+ C x is a function. Of course, one can al-
ways obtain this latter form by arbitrarily choosing (matrix) representatives
of the cosets z(x) in PGL(n, C); that is, if n: GL(n, C) 4 PGL(n, C) is the
natural map, choose T(x) with nT(x) = z(x). We always choose the identity
matrix E as the representative of ~ ( 1 )that
; is, T(1) = E.
Lemma 7.62. Let z: Q -+ PGL(n, C) be a projective representation. If T: Q -,
GL(n, C) is a matrix-valued function of z, that is, T(x)T(y) = f,(x, y)T(xy),
then f, is a factor set. If T' is another matrix-valued function of z, so that
T'(x)Tf(y)= g,(x, y)Tt(xy), then f,g,' is a coboundary. Therefore, z deter-
mines a unique element [f,] E M(Q).
Remark. This result is the reason factor sets are so called.
ProoJ Since T(l) = E, we have f,(l, y) = 1 = f,(x, I), while the cocycle iden-
tity follows from associativity: T(x(yz)) = T((xy)z).
Since nT(x) = Z(X)= nT'(x) for all x E Q, T(x)T'(x)-' = h(x)E, where h(x) E
C ". It is easy to see, as in Lemma 7.31, that f,(x, y)g,(x, y)-' = h(y)h(xy)-'h(x);
that is, hgr1is a coboundary. B I
Schur proved, for every finite group Q, that there is a "cover" U with
Q = UIM such that every projective representation of Q determines an ordi-
nary representation of U; moreover, it turns out that M r M(Q) and U is a
central extension of M(Q) by Q. The reader can now see why M(Q) is of
interest; for example, if M(Q) = 1, then U = Q, and every projective represen-
tation of Q gives a representation of Q itself. We now begin the proof of these
assertions.
Defi~itioaa.Let v : U + Q be a surjective homomorphism with kernel K, and
assume that U is a central extension of K by Q. If z: Q 4 PGL(n, C) is a
projective representation, then z can be lrlfted to U if there exists a homomor-
phism z" making the following diagram commute;
Projective Representations and the Schur Multiplier 205
We say that U has the projective lbt$tirzg property if every projective repre-
sentation of Q can be lifted to U .
By Theorem 7.30, the elements of the central extension U may be viewed
as ordered pairs (a, x) E K x Q, where K = ker v. With this understanding, z
) all (a, x) E U . As
can be lifted to U if and only if nz"(a,x) = -cv(a,x) = ~ ( xfor
usual, we replace z by a matrix-valued function T ; the equation nZ((a, x)) =
z(x) in PGL(n, C) is now replaced by a matrix equaton in GL(n, C):
for some function p: U -, C ", where U = K x Q.
Definition. If K and A are groups with A abelian, then
Hom(K, A) = (all homomorphism I< -, A).
If q, $ E Hom(I<, A), define q$: I< 4 A by q$: x c, q(x) + $(x) for all x E
K; note that q $ is a homomorphism because A is abelian. It is easy to see
that Hom(K, A) is an abelian group under this operation.
In particular, if K is also an abelian group, define its character group
K* = Hom(K, C ").
Since C x is a multiplicative group, it is more convenient to write K* as a
multiplicative abelian group: if a E K and q , $ E K*, then q$: a t-, q(a)$(a).
We shall prove, in Theorem 10.56, that I< z K* for every finite abelian
group I<.
If U is any central extension of Q with kernel K, define a function 6: K* -,
H2(Q, C x ) = M(Q) as follows. By Theorem 7.34, U corresponds to an ele-
ment [el E kf2(Q, I<), where e: Q x Q -, K is a factor set, and U consists of
all (a, X)E K x Q, where (a, x)(b, y) = (abe(x, y), xy). If q E K*, it is routine to
check that the composite q o e: Q x Q -+ C x is a factor set, and so [q o el E
H ~ ( QC, "). Moreover, if e and e' are equivalent factor sets, then cp o e and
q o e' are also equivalent: if e'(x, y) = e(x, y)h(y)lz(xy)-'lz(x),then
Definition. Let U be a central extension of K by Q. If e: Q x Q -, K is a factor
set of U, then the trwnsgressio~t6 = 6 is the homomorphism 6: I<* -, M(Q)
defined by 6(q) = [q o el.
The preceding discussion shows that the transgression does not depend on
the choice of factor set e arising from the central extension U .
Lemma 9.43. Let U be a central extension of K by Q. Then the transgression
6: K* !
: M(Q) is surjective if and only if U has the projective lijtirzg property.
ProoJ By Theorem 7.34, the central extension U of K by Q determines [el E
H2(Q, K), where 2: Q x Q -+K is a factor set. The strategy is to show that if
206 7. Extensions and Cohomology Projective Representations and the Schur Multiplier

7:Q --+ PGL corresponds to a matrix-valued function T with T(x)T(y) = If a = b = 1 in (4), then
h(x, y)T(xy), then r can be lifted to U if and only if [A] G im 6; that is, there
is y E K* with 6(y) = [ y o el = [A].
If [A] E im 6, then, for all x, y E Q, there is rp E I<* and h: Q 4 C x with Finally, setting b = 1 and x = 1 gives ~ ( ay), = y(a)p(y). Hence
(3) Y O 4%Y)= A(x, y)h(y)h(xy)-lh(x).
As in Theorem 7.30, regard the elements of U as ordered pairs (a, x) E IC x Q. and so rp 0 e(x, y)A(x, y)-' is a factor set. Therefore, [f,] = [y o el = 6(y),
Define z": U -+ GL(n, C) by as desired.
?(a, x) = y(a)h(x)T(x).
Several properties of character groups, proved in Chapter 10, will be used
We compute: in the next proof. A (multiplicative) abelian group D is called divisible if every
element d E D has an nth root in D; that is, for every n > 0, there exsts x E D
with x n = d. For example, C x is a divisible group.

Lemma 7.64. Let U be a central extension of K by Q. Then the transgression

On the other hand, since C x is abelian,
?(fa7 x))z"((b,Y))= rp(a)h(x)T(x)rp(b)h(y)T(y)
By (3), 7 is a homomorphism. But rv(a, x) = r(x) and ni(a, x) = nrp(a)h(x)T(x)
= -nT(x)= r(x), and so r can be lifted to U.
Conversely, assume that r can be lifted to U: there is a homomorphism
7 : U -+ GL(n, C) and a function p: K x Q -+ C x with
8: I<* 1: M(Q) is injective if and only if K _< U'.
fiooJ We first show that if y and q(K n U') = 1, then S(y) = [y o el
E K*
= 1 in M(Q). The second isomorphism theorem gives KIK n U' z KU'IU'.
Define @: K U'IU' 1: C " by @(aU1)= y (a) for all a E K; @ is a well defined
homomorphism because K n U' 5 ker y. But KU'IU' is a subgroup of the
abelian group U/U', so that the injective property (Theorem 10.23) of the
divisible group C x gives a homomorphism Y: U/U' --+ C " extending @. Now,
for all x, y E Q,

."(a, = p(a, x) T(x)

for a l l a E K and x E Q. Since ? is a homomorphism, we may evaluate

?(a, x)i(b, y) in two ways: on the one hand, it is
P(aY x) T ( ~ ) P ( ~Y)
YT(y); Define h: Q --+ C x by h(x) = Y((1, x) U'). A routine calculation shows that
on the other hand, it is z"(abe(x,y), xy), which is

so that 6(y) = [y-o el = 1 in M(Q). Since 6 is injective, y = 1. But Theorem

But T(x)T(y) = L(x, y) T(xy), so that, for all a, b E K and x, y E Q, 10.58 shows that if K n U' < K, then there exists y E K* with y # 1 and
K n U' I ker y. We conclude that K n U' = K; that is, K 5 U'.
Conversely, if 9 E ker 6, then 1 = 6((p) = [y o el, where e is a factor set of
If x = y = 1, then (4) gives the extension U of K by Q. Thus, 9 o e is a coboundary: there is a function
Aab9 1) = P(~Yl)p(b, 1); h: Q --+ C with h(1) = 1 such that, for all x, y E Q,

that is, the function y: K --+ C ", defined by . 9 0 e(x, y) = Iz (y)h(xy)-'h(x).

If we define @: U -, C x by @(a,x) = y(a)Iz(x),then a routine calculation shows
that @ is a homomorphism with @ JIC = y. As C x is abelian, U' 5 ker cD, and
lies in the character group K*. Define p: Q 4 C Xby so U' n K 5 ker y. The hypothesis K < U' gives U' n K = K , and so y is
trivial; that is, 6 is injective. B
208 7. Extensions and Cohomology
Definition. If Q is a group, then a cover (or representation group) of Q is a
central extension U of K by Q (for some abelian group K) with the projective
lifting property and with K I U'.
The following lemma will be used in proving the existence of covers. It
follows from the preceding two lemmas that if a cover U of a finite group Q
exists, where U is a central extension of K by Q, then K r M(Q).
Lemma 7.65. If Q is a finite group, then the subgroup B2(Q, C ') has a finite
complement M 2 M(Q) in Z 2(Q, C " ).
Proof. We first show that B ~ ( QC, " ) is a divisible group. If f E B2(Q, C x),
then there is h: Q -+ C x with h(1) = 1 and, for all x, y E Q,
-
For n > 0 and each x E Q, let k(x) be an nth root of h(x) with 141) = 1. Then
g: Q x Q C x , defined by g(x, y) = k(y)k(xy)-'k(x), is a coboundary with
g" = f. By Corollary 10.24, B2(Q, a) " ) has a complement M in Z2(Q, C " ).
But M z Z2(Q, C " )/B2(Q, @ ") = M(Q), which is finite, by Theorem 7.60.
Theorem 7.66 (Schur, 1904). Every finite group Q has a cover U which is a
central extension of M(Q) by Q.
Proof. Let M be a complement of B ~ ( QC , X )in Z2(Q, C X ) ,as in the lemma;
note that the elements of M are factor sets Q x Q -+ C x . Define I< = M*. We
first define a certain factor set s: Q x Q -+ M*, and then construct the desired
central extension from it.
For each (x, y) E Q x Q, define s(x, y): M -+ C x by f H f (x, y); thus, s(x, y)
E M*, and there is a function s: Q x Q 4 M* given by (x, y) H s(x, y). Now s
satisfies the cocycle identity: for all x, y, z E Q.
and the last term is 1 because every f E M is a factor set. Similarly, s(1, y) =
1 = s(x, 1)for all x, y E Q because, for example, s(1, y): f r f(1, y) = 1. There-
fore, s E Z2(Q, M*). Let U be the corresponding central extension of M* by Q.
To see that U is a cover of Q, it suffices,by Lemmas 7.61 and 7.62, to prove that
the transgression 6: M** 4 M(Q) is an isomorphism. If [f ] E M(Q), then
f E Z2(Q, a) ); since Z 2 = B2 x Myf has a unique expression f = bf ', where
b~ B2(Q, ex),f r ~ M , a n d [ f ] = [bf'] = [ f t ] . N o w p ( f ' ) ~ C Xmakessense
for all p E M* (since f ' E M), and it is easy to check that q : M** 4 6 " , given
by q:p r p(f '), is a homomorphism. By definition, 6(q) = [q 0 s]. The
composite q o s: Q x Q -+M* 4 C x sends (x, y) r S(X,y) F+ q(s(x, y)) =
S(X,y) (f ') = f '(x, y); that is, q o s = f ' and 6(q) = [q o s] = [f '1 = [f 1.
Therefore, 6 is surjective. But IM**I = IM(Q)I, since M r M(Q) is finite, and
so 6 must be an isomorphism. El
Projective Representations and the Schur Multiplier 209
It should not be surprising that calculations of the multiplier are best done
using techniques of Homological Algebra, Cohomology of Groups, and Rep-
resentation Theory; for example, see Huppert (1967,§V.25) and Karpilovsky
(1987). Indeed, the proofs given above fit into a general scheme (e.g., the
transgression homomorphism 6 arises in the "five-term exact sequence" as
the map H1(K, C " ) -+H ~ ( Qa), " )). When Q is finite, we mention that M(Q)
has been calculated in many cases; sometimes it is 1; sometimes it is not.
EXAMPLE
7.17. Covers U of a finite group Q need not be unique.
If Y is the Cgroup, consider the central extensions U of K E 2, by V,
where U z D, or U z Q, the quaternions. In each case, K = Z(U) = U', so
that U is a central extension with K I U'. It follows from Lemma 7.64 that
M(V) # 1. We shall see in Chapter 11 that M(V) = if2; it will then follow
from Lemma 7.63 that both D, and Q are (nonisomorphic) covers of V.
The following discussion comparing two central extensions of a group Q
will be completed in Chapter 11 when we show that covers of perfect groups,
in particular, covers of simple groups, are unique. Consider the commutative
diagram
where both rows are central extensions. We claim that a(K) 5 L. If a E K,
then 1 = v(a) = pa(a), so that a(a) E ker p = L. Denote the restriction of a by
p: K L. !f p: K -+ L is any homomorphism, then there is a homomorphism
-+
p*: L* -+ K* given by $ F+ 11o p, v;here 11:L -+ C x is in L*.
The diagram gives two transgressions: denote them by 6 ': K* -+ M ( Q ) and
by aV:L* -+ M(Q).
Lemma 7.67. Consider the cornnzutative diagram
whose rows are central extensions. Then 6 'j?* = 6 '.
Proof: Let e: Q x Q -+ K be a factor set of the top extension; we may
assume that U consists of all (a, x) E K x Q with multiplication (a, x)(b, y) =
(abe(x, y), xy) and with v(a, x) = x. Similarly, let f : Q x Q -+ L be a factor set
210 7. Extensions and Cohomology Derivations 21 1

of the bottom extension, so that V consists of all (c, x) E L x Q with (c, x)(d, y) Theorem 7.69. If every Sylow subgroup of a finite group Q is cyclic, then
= (cdf (x, y), xy) and with p(c, x) = x. M(Q) = 1.
Now a((1, x)) = (h(x), 1(x)) E V, and x = v(1, x) = pa((1, x)) = p((h(x), 1(x))
= E(x); hence, x = I(x), and a(1, x) = (h(x),x). Note also that (a, x) = (a, 1)(1, x) Proof. For such a group, exp(Q) = 1 Ql. By Theorem 7.68, exp(M(Q)) = 1, and
for all a E I<and x E Q. soM(Q) = 1. E 3
For all x, y E Q,
Corollary 7.768.
(i) For every n 2 1, M(Z,,) = 1.
(ii) If Q has squarefree order, then M(Q) = 1.

There is a more general result which implies Theorem 7.69: if p is a prime

divisor of 1 Q 1 and P is a Sylow p-subgroup of Q, then the p-primary compo-
because a E K is identified with (a, 1) and P((a, 1)) E L. On the other hand, nent of M(Q) can be imbedded in M(P).
We shall return to the Schur multiplier in Chapter 11, for it is also related
to presentations of groups.

Therefore,
P O 4x9 Y)= f(x7 y)h(y)h(xy)-lh(x);
that is,
CP 0 el = CfI E w e ) -
It follows easily, for any $ E L*, that [$ o /3 o el = [$ of]. Hence, SUP*($)
= SU($o ,!?) = [$ o P o el = [$ o f ] = SV($),as desired. B l
Theorem '7.68 (Allperin-Kuo, 1967). If Q is a finite group, e = exp(M(Q)),and
e' = exp(Q), then ee' divides I QI.
Proof (Brandis). We first show that if U is a finite group with subgroup
I<I U' n Z(U), then exp(K) exp(U/K) divides I UlKl. Let A be an abelian
subgroup with K I A < U, and let V: U -+ A be the transfer. Since A is
abelian, I< < U' < ker V , so that u E K implies V(u) = 1. By Theorem 7.47, if
u E Z(U), then V(u) = un.Therefore, if u E K 5 U' nZ(U), then x n = 1, and so
e = exp(I<)divides n = [U : A] = [UIK : AIK]. Hence, elA/KI divides ] UIKI.
In particular, this holds for all cyclic subgroups AIK < UIK (for I U/KI does
not depend on n). But e' = exp(U/K) divides IAIKI for all cyclic AIK, and so
ee' divides I UIK I.
If U is a cover of Q, then M(Q) K < Z(U), for U is a central extension.
Now K < U', by definition of cover, so that K < U' n Z(U). By the first
paragraph, exp(K) exp(U/K) divides I UIKI. But K z M(Q) and UIK z Q,
and this gives the result. Bl
We remark that Schur proved that if e = exp(M(Q)),then e2 divides IQI.
Derivations
We again consider general data (Q, K, 8) consisting of a group Q, an (addi-
tive) abelian group K, and a not necessarily trivial homomorphism 0: Q -+
Aut(I<) giving an action of Q on K.
~efinition.Given data (Q, K, 0), a deriuntion (or crossed homomorphism) is a
function d: Q -+ K such that
/
+
d(xy) = xd(y) d(x).
The set Der(Q, K, 8) of all derivations is an abelian group under the follow-
ing operation: if d, d' E Der(Q, K, 0), then d + dl: x H d(x) + df(x).In partic-
ular, if 8 is trivial, then Der(Q, K) = Hom(Q, K).
EXAMPLE 7.18. 'Let data (Q, K, 8) be given. For fixed a E K , the function
d,: Q -+ I<, defined by x H a - xa is easily checked to be a derivation: it is
called the puinckak de~iaationdetermined by a.
EXAMPLE 7-19. Let G be an extension realizing data (Q, K , 0), so that G con-
+ + +
sists of all (a, x) E K x Q with (a, x) (b, y) = (a xb f ( x , y), xy) for a
factor set f. If y: 6 -+ G is a stabilizing automorphism, then y: (a, x) H
+
(a d(x), x) for some d(x) E K, and the reader may checlc that d is a derivation.
Theorem 7.71. Let G be an extension realizing data (Q, K, 8). If A is the
stabilizer of this extension, then
212 7. Extensions and Cohomology Derivations

Proo$. Regard G as all (a, x) E K x Q. As in Example 7.19, if y E A, then PDer(Q, ICY0). Hence
y: (a, x) I--+ (a + d(x), x), where d is a derivation. Define q: A -+ Der(Q, K, 8)
by q(y) = d. Now q is a homomorphism: if y' E A, then y': (a, x) H
(a + dr(x),x), and y'y: (a, x) H (a + d(x) + dl(x), x). We see that y? is an iso-
morphism by constructing its inverse: if d is a derivation, define y: G -+ G by
+
y : (a, x) I-+ (a d (x), x). &!
Since we will be dealing with two transversals in the next theorem, let us
Lemma 4.72. Let G be a semidirect product K x,Q, and let y: (a, x) I-+ write the elements of a semidirect product G = K >a, Q not as ordered pairs
+
(a d(x), x) be a stabilizing automorphism of G. Then y is an inner auto- +
(a, x), but rather as sums a l(x), where 1: Q -, G is a transversal.
morphism if and only if d is a principal derivation.
Theorem 7.74. Let G = K >a, Q, where K is abelian, and let C and C' be corn-
ProoJ If d is a principal derivation, then there is b E K with d(x) = b - xb. plemerzts of K in G. If H'(Q, K , 0) = 0, then C and C' are conjugate.
Hence y(a, x) = (a + b - xb, x). But
(by1) + (a, x) - (by1) = (b + a, x) Remark. By Theorem 7.34, the hypothesis is satisfied if H2(Q, K, 0) = 0.
= (b + a - xb, x), Proo$. Since C and C' are complements of I<, they are isomorphic: C g
GIK z Q e C'; choose isomorphisms I: Q 4 C and 1': Q 4 C'. Both C and C'
so that y is conjugation by (b, 1). are also transversals of IC in G, and so they determine factor sets f and f '
Conversely, if y is conjugation by (byy), then for all (a, x) E G, (where, for example, E(x) + l(y) = f(x, y) + E(xy)). By Lemma 7.31, there is a
?(a, 4 = (b, Y)+ (a, X)- (by Y) function h: Q 4 K, namely, h(x) = ll(x) - l(x), with
= (b + ya, yx) + (- y-I b, y-l)
= (b + ya - yxy-'by yxy-'). Since 1 and 1' are homomorphisms, it follows that both f and J" are identi-
Since y is stabilizing, we must have yxy-I = x, so that
+
cally zero. Thus, f '(x, y) - f(x, y) is identically zero, h(xy) = xh(y) h(x),
and h is a derivation. The hypothesis H1(Q, K, 0) = 0 says that h is a princi-
y(a, x) = (ya + b - xb, x). pal derivation; that is, there is b E I<with
Finally, if we choose x = 1, then (a, 1) = y(a, 1) (because y is stabilizing), and lr(x) - l(x) = Iz(x) = b - xb.
so (a, 1) = (yay1). Therefore, ya = a for all a, and so y (a, x) = (a + b - xb, x) Therefore,
= (a + d(x), x), where d(x) = b - xb is a principal derivation. El
ll(x) = b + (-xb + l(x)) = b + l(x) - b;
It is easy to see that PDer(Q, K, 8), the set of all principal derivations, is a
subgroup of Der(Q, K, 0).
that is, C' =b + C - b is a conjugate of C. H

Definit~iadw.If d: Q + K is a derivation, then its kernel is

Definition. Given data (Q, K, 0), the first cokomology group H1(Q, K, 0) =
Der(Q, K, Q)lPDer(Q, K, 8). lter d = (x E Q: d(x) = 0).

If 0 is trivial, then H1(Q, K, 0) = Horn(Q, K). In particular, if Q is abelian For example, if do is a principal derivation, then ker d, is the set of all x E Q
and K = C x , then H1(Q, K) is the character group Hom(Q, C x ) = Q*. which fix a.

Theorem 7.73. If (Q, K, 0) are data and G = H x, Q, then Lemma 7.75. Let (Q, K , 0) be data, and let d: Q 4 K be a derivation.
H1(Q, K, 0) 4 Aut(G)/Inn(G). (i) d(1) = 0.
(ii) d(x-l) = -x-ld(x).
ProoJ Let A be the stabilizer of the extension and let q : A 4 Der(Q, K, 0) (iii) lter d is a szibgroup of Q.
be the isomorphism of Theorem 7.71. By Lemma 7.72, q ( A n Inn(G)) = (iv) d(x) = d(y) if and only if x-ly E lter d.
214 7. Extensions and Cohomology Derivations 215

ProoPf. (i) d(1) = d(l '1) = Id(1) d(1).+
(ii) 0 = d(1) = d(x-lx) = x-ld(x) + d(x-I).
(iii) 1 E ker d, by (i). If x, y E ker d, then d(x) = 0 = d(y). Hence, d(xy-') =
xd(y-') + d(x) = - xy-'d(y) + d(x) = 0. Therefore, xy-' E ker d and so ker d
is a subgroup of Q.
(iv) Now d(xA1y)= x-ld(y) + d(xml)= x-'d(y) - x-'d(x). Hence, if d(x)
= d(y), then d(xM1y)= 0. Conversely, if d(x-'y) = 0, then x-'d(y) = x-'d(x)
and d (y) = d(x). H
The subgroup ker d need not be a normal subgroup.
We are going to use derivations to give another proof of the Schur-
Zassenhaus lemma. Given data (Q, I<, 0), it is now more convenient to write
the abelian group K multiplicatively. Assume that an abelian normal sub-
group K has finite index n in a group E, and let I,, . . ., 1, be a left transversal
of K in G; that is, E = U:=, 1,K. As in Lemma 7.44, if e E E, then there is a
unique rc,(e) E K with el, = E,,rc,(e), where o is a permutation (depending on e).
Define ai(e) = laiici(e)l;l, so that
eli = ai(e)lai,
and define a function d: E -,K by
(ii) If Ic E K, then K a E implies lilkli E I<.Thus, kli = li(l[lkli) so that
rc,(lc) = l~llcliand the permutation a determined by Ic is the identity. But
a,(k) = lirci(k)l;' = k, for all i, and so d(k) = n;=, ai(k) = Ic".
(iii) Recall that el, = l,,rc,(e); if e = lj, then ljli = l,1 for some 1, E L, because
L is now assumed to be a subgroup. Hence, rci(lj) = 1 = ai(lj)for all i (because
a, is a conjugate of rc,), and so d(lj) = 1 for every j.
(iv) First, note that d(k-l) =d(k)-l, by Lemma 7.75 (ii), because both k-I and
d(Ic)-' lie in the abelian group K. Wow, d(k-lek) = d(ek)k-'d(k-') = d(eIc)d(k-I),
because both d(ek) and k-I lie in the abelian group I<, and d(ek)d(kw1)=
d(k)ed(e)d(k-l)=d(k)ed(e)d(lc)-l. El
The definition of the derivation d is reminiscent of the transfer; indeed, it is
now easy to see that dlK is the transfer.
The following proof is due to I<. Gruenberg and B.A.F. Wehrfritz.
Theorem 7.77 (=Theorems 7.40 and 19.49). Given data (Q, K, 8) with Q a finite
group, K a finite abelian group, and (IKI, IQI) = 1, then every extension G of
K by Q realizing the data is a senzidirect product; moreover, any two comple-
rnents of K in G are cotzjugate.

Pros$. Let E be an extension of I< by Q realizing the data and let I Q I = tz.
Define L = lter d, where d: Q -,K is Cruenberg's derivation. By Lemma 7.75
(iii), L is a subgroup of Q. We claim that L n K = 1. If a E I<, then d(a) = an,
Lemma 7.76 (Granenberg). Let K be an abelian normal subgroup of finite index
by Lemma 7.76 (ii); if a E L, then d(a) = 1, by definition of kernel; hence,
in a group E, and let L = {l,, ...,I,) be a left transversal of K in E.
a" = 1 for all a E L n K. But (lKI, n) = 1, by hypothesis, so that a = 1. Let us
(i) If 0: E -+ Aut(K) is defined by %(k) = eke-', then the function d defined now set that E = KL. If e E E, then d(e) E I<. Since (I K 1, n) = 1, there is k E K
above is a derivation. with d(e) = k-", by Exercise 1.31. Hence d(e) = k-" =f(k-I), by Lemma 7.76
(ii) If k E I<, then d(k) = kn. (ii), and so Ice E L = ker d, by Lemma 7.75 (iv). Therefore, e = kR1(ke)E KL,
(iii) If L is a complement of K in E, then L Iker d. and E is a semidirect product.
(iv) If k E I( and e E E, then d(kblek) = d(k)"d(e)d(k)-'. We now show that if Y is another complement of K in E, then Y and L are
conjugate. Note that each of Y and L is a transversal of I< in E. Let Y =
ProoJ (i) Iff E E, then fl, = a,(f )lri,where z is a permutation. Now {y,, . . . ,y,) and L = {I,, . . . , I,), and write y, = c,l, for ci E K. The derivation
determined by L is d(e) =
Hi
ni a,(e); the derivation determined by Y is 6(e) =
ai(e), where ai(e) E K and ey, = ai(e)yri.But
ey, = ecili = (ecie-')eli = cfai(e)lai= cFai(e)cz/y,,;
since (ef)li = ai(ef)lWi,for some permutation o,so that ai(ef) = a,(f)"ari(e).
Since each a,(e) lies in the abelian group K, for e E E and 1 5 i 5 n, and since it follows that z = a, ct,(e) = cfai(e)c,l, and
z is a permutation, it follows that

If we define c = n ci, then c E I<and

Since (IKI, tz) = 1, there is k E K with c = lcn,by Exercise 1.31, and so c = d(k),
Therefore, d is a derivation. by Lemma 7.76 (ii). Hence, d(k-lek) = d(k)"df(e)d(Ic)- = ced(e)c-', by Lemma
216 7. Extensions and Cohomology

7.76 (iv). Therefore,

6(e) = d(kW1ek), CHAPTER 8
, Lemma 7.75 (iii). But ILl = I YI,
so that L = ker d = lc(ker 6)k-I 2 k ~ k - l by
for both are complements of K , and so L = kYk-'. kd Some Simple Linear Groups

7.54. If G = li; x, Q, then every transversal I: Q -+ G has the form 1(x) = (d(x), x) for
some d(x) E K. Show that I is a homomorphism if and only if d is a derivation.
7.55. Give an example of a derivation d: Q -+ K whose kernel is not a normal sub-
group of Q.
7.56. If 0: Q -+ Aut(K) is trivial, then H1(Q, K) z Hom(Q, K). Conclude that if B is an
abelian group, then H1(B,C X) r B*, the character group of B.
7.57. Regard the elements of an extension G realizing data (Q, K, 8) as ordered pairs
(a, x) E K x Q, let 1: Q -+ G be the transversal 1(x) = (0, x), and let f : Q x Q -+ K
be the corresponding factor set. If d: G -+ K is Gruenberg's derivation, show
that d(a, x) = Iza + CyEa f(x, y), where n = IQI.
7.58. Let h 3 = (a), let Z, = (x), and define an action 0 of B, on h 3 by ax = a-l. The Jordan-Hijlder theorem tells us that once we know extensions and
Show that H1(Z,, Z,, 8) = 0. (Hint.S3 r Z3 x Z,, and Aut(S3)/Inn(S3)= 1.)
simple groups, then we know all finite groups. There are several infinite
7.59. Let 9 : G -+ G be a homomorphism, and let G act on itself by conjugation: if families of finite simple groups (in addition to the cyclic groups of prime
g, x E G, then gx = xgx-'. Show that d: G -+ G, defined by d(g) = gq(g)-', satis- order and the large alternating groups), and our main concern in this chapter
fies the equation is the most "obvious" of these, the projective unimodular groups, which arise
d(gg') = d(g')gd(g). naturally from the group of all matrices of determinant 1 over a field K. Since
(Thus, d would be a derivation if G were abelian. This function d has already these groups are finite only when the field IC is finite, let us begin by examining
arisen in Exercise 1.50 and in the proof of Theorem 7.33.) the fidite fields.
/
Given data (Q, K, 8), there are cohomology groups Hi(Q, K, 8) for all i 2 0
and hoinology groups Hi(Q, K, 0) for all i 2 0. Cohomology of Groups is
the interpretation and computation of these groups. We have already Finite Fields
discussed the cohomology groups Hi(Q, K, 8) for i = 1, 2. The group
HO(Q,K, 0) consists of fixed points: Definition. If I< is a field, a subfield of K is a subring k of K which contains
the inverse of every nonzero element of k. A prime field is a field lc having no
HO(Q,K, 0) = (a E K: xa = a for all x E Q); proper subfields.
the group H3(Q, I(,0) is involved with obstructions: what sorts of data can
extensions with nonabelian kernels realize. Now H,(Q, K) is the maximal Theorem 8.1. Every field I< contains a unique prirne subfi'elcl li, nrzd eitlzer
Q-trivial quotient of K. If K = Z and 0 is trivial, then the first homology k r or k E Z, for some prirne p.
group H1(Q, Z) r Q/Q' and, when Q is finite, the second homology group
H,(Q, Z) r M(Q). There are also some standard homomorphisms between ProoJ If lc is the intersection of all the subfields of K, then it is easy to check
these groups, one of which, corestriction, generalizes the transfer. A funda- that k is the unique prime subfield of K. Define X: Z -+K by ~ ( n= ) rzl, where
mental idea is to construct the group ring ZG of a group G and to observe 1 denotes the "one" in K. It is easily checked that x is a ring homomorphism
that G acting on an abelian group K corresponds to K being a ZG-module. with im x c lc. Since K is a field, im x is a domain and ker x must be a prime
Indeed, our notation Hi(Q, K, 0) is usually abbreviated to Hi(Q, K) by ideal in Z. Therefore, either ker x = 0 or lter 31 = (p) for some prime p. In the
assuming at the outset that K is a ZG-module. first case, im x r Z and k contains an isomorphic copy F of the fraction field
of Z, namely, Q; as 1c is a prime field, k = F r Q. In the second case, k
218 8. Some Simple Linear Groups
contains an isomorphic copy E of Z/ker x = Z,, which is a field; as k is .a
prime field, k = E r Z,. BI
Defiaition. If K is a field with prime field k, then K has characteristic 0 if
lz r Q and K has characteristicp if lz E Z,.
Observe that if K has characteristic p 2 0, then pa =0 for all a E K.
Corollary 8.2. If K is a finite field, then 1 Kl = pn for some prime p and some
n 2 1.
ProoJ If Iz is the prime field of K, then k $ Q because I< is finite; therefore,
k r Z,for some prime p. We may view K as a vector space over Z, (the
"vectors" are the elements of K, the "scalars" are the elements of k, and the
"scalar multiplication" aa, for a E 12 and a E K, is just their product in K); if K
has dimension n, then 111' = pn.
There exist infinite fields of prime characteristic; for example, the field of all
rational functions over Zp (i.e., the fraction field of Z,[x]) is such a field.
The existence and uniqueness of finite fields are proven in Appendix VI: for
every prime p and every integer n 2 1, there exists a field with pn elements
(Theorem VI.19); two finite fields are isomorphic if and only if they have the
same number of elements (Theorem VI.20). Finite fields are called Gabis
fields after their discoverer; we thus denote the field with q = pn elements by
GF(q) (another common notation for this field is IF,) though we usually
denote GF(p) by Z,.
Recall that if E is a field, k is a subfield, and n E E, then k(n), the subfield of
E obtained by adjoining n to k, is the smallest subfield of E containing k and
n; that is, k(n) is the intersection of all the subfields of E containing k and n.
Definition. A primitive element of a finite field K is an element n E K with
K = lz(n), where k is the prime field.
Lemma 8.3. There exists a primitive element n of GF(pn);moreover, n may be
chosen to be a root of an irreducible polynomial g(x) E Z,[x] of degree n.
ProoJ Let q = p" and let K = GF(q). By Theorem 2.18(ii), the multiplicative
group K x is cyclic; clearly, any generator n of K x is a primitive element of K
(there can be primitive elements of I ' that are not generators of Kx). By
Lagrange's theorem, nq-I = 1 (for I K xI = q - I), and so n is a root off (x) =
x4-I - 1. Factoring f(x) into a product of irreducible polynomials in k[x]
(where k r Zp is the prime field) provides an irreducible g(x) E k[x] having n
as a root. If g(x) has degree d, then k(n) is a subfield of K with [k(n) : k] = d
(Theorem VI.21 in Appendix VI); therefore, llz(n)( = pd. But k(n) = K (be-
cause n is a primitive element) and so d = n. BB
The General Linear Group 219
Theorem 8.4. If p is a prinze, then the group Aut(GF(pn))of all field atitomor-
phisrns of GF(pn)is cyclic of order n.
ProoJ Let k be the prime field of K = GF(pn).If n is a primitive element of
K, as in the lemma, then ihere is an irreducible polynomial g(x) E k[x] of
degree n having 7c as a root. Since every q E Aut(K) must fix Iz pointwise
(because cp(1) = I), Lemma 5.1 shows that cp(n) is also a root of g(x). As
K = k(n), Lemma 5.2 shows that q is completely determined by cp(n). It
follows that IAut(l<)l 5 n, because g(x), having degree n, has at most n roots.
The map a: K -,K, given by o(a) = aP,is an automorphism of K. If 1 I i<n
and a i = 1, then a = api for every a E K. In particular, npi-l = 1, contra-
dicting n having order pn - 1 in I<". Therefore, (a) < Aut(I<) is cyclic of
order n, and so Aut(K) = (a).
We remark that Aut(GF(pn))is the Galois group Gal(GF(pn)/Zp),for
every cp E Aut(GF(pn))fixes the prime field Zp pointwise.
The General Linear Group
Groups of nonsingular matrices are as natural an object of study as groups
of permutations: the latter consists of "automorphisms" of a set; the former
consists of automorphisms of a vector space.
Deffnwition.If V is an m-dimensional vector space over a field K , then the
genirlak h e a r group GL(V) is the group of all nonsingular linear transforma-
tions on V (with composite as operation).
If one chooses an ordered basis (el, .. ., em) of V, then each T E GL(V)
xi
determines a matrix A = [aij], where Te, = aijei(the jth colurnn of A con-
sists of the coordinates of Tej). The function T w A is an isomorphism
GL(V) -+ GL(mn, K), where GL(m, I<) is the multiplicative group of all nz x In
nonsingular matrices over K. When I< = GF(q), we may write GL(m, q) ,
instead of GL(m, K).
ProoJ Let V be an m-dimensional vector space over a field I<, and let
(el, . . . , em)be an ordered basis of V. If B denotes the family of all ordered
bases of V, then there is a bijection GL(V) B: if T is nonsingular, then
-+
(Te,, . . ., Ten,) is an ordered basis of V; if (v,, . . ., v,) is an ordered basis,
then there exists a unique nonsingular T with Tei = vi for all i.
Let (v,, . . . , v,) be an ordered basis of V. Since there are qn' vectors in F/,
there are qm- 1 candidates for v, (the zero vector is not a candidate). Having
220 8. Some Simple Linear Groups
chosen v,, the candidates for v, are those vectors in V not in (v,), the
subspace spanned by v,; there are thus qm- q candidates for v,. More gener-
ally, having chosen an independent set {v,, ... ,v,), we may choose v,+, to be
any vector not in (v,, ...,v,), and so there are qm- qi candidates for vi+,.
The result follows.
Notation. If V is an m-dimensional vector space over K = GF(q), if t is a
nonnegative integer, and if ?I is a primitive element of K, then
M(t) = {A E GL(V): det A is a power of ?it).
Lemma 8.6. If Q = IGL(m, q)] and fi t is a divisor of q - 1, then M(t) is a
normal subgroup of GL(m, q) of order Q/t. Moreover, if q - 1 = p1 . ..p,, where
the pi are (not necessarily distinct) primes, then the following normal series is
the beginning of a composition series:
Proof. Let K = GF(q). Use the correspondence theorem in the setting
det: GL(m, q) -+ K x
If t divides q - 1 = IK " I, then the cyclic subgroup (nr) of K " is normal (K "
is abelian), has order (q - l)/t, and has index t. Since M(t) is the subgroup of
GL(m, q) corresponding to (?it), it is a normal subgroup of index t hence
orderQ/t. Now IM(p1 i'i)/M(i'l Pi+l)I = (Q/Pl - p , . . . ~ i ) / (...Pi+l)
Q / ~ ~ = Pi+,;
since the factor groups have prime order, they are simple.
Definition. A matrix (or linear transformation) having determinant 1 is called
u~airpror%lllar.
The subgroup M(q - 1) consists of all the unimodular matrices, for
n"l = 1.
Definition. If V is an m-dimensional vector space over a field K, then the
special linear group SL(V) is the subgroup of GL(V) consisting of all the
unimodular transformations.
Choosing an ordered basis of V gives an isomorphism SL(V) -+ SL(m, K),
the group of all unimodular matrices. If K = GF(q), we may denote SL(m, K)
by SL(nz, q).
The following elementary matrices are introduced to analyze the structure
of SL(r72, K).
Definition. Let A be a nonzero element of a field K, and let i # j be integers
between 1 and m. An elementary transvection BJL) is the m x m matrix differ-
ing from the identity matrix E in that it has L as its ij entry. A transtrection is
The General Linear Group 221
a matrix B that is similar to some B&); that is, B is a conjugate of some B,(A)
in GL(m, K).
Every transvection is unimodular. Note that the inverse of an elementary
transvection is another such: Bij(A)-I = Bij(- A); it follows that the inverse of
any transvection is also a transvection.
If A E GL(m, K), then Bij(R)A is the matrix obtained from A by adding A
times its jth row to its ith row.
Lemma 8.7. Let K be a field. If A E GL(mri, K ) and det A = p, then A =
UD(,u), where U is a product of elementary transvections and D =
diag{l, 1, ..., 1, p).
Proof. We prove, by induction on t 5 m - 1, that A can be transformed, by
a sequence of elementary operations which add a multiple of one row to
another, into a matrix of the form
where E, is the t x t identity matrix.
For the base step, note that the first column of A is not zero (A is non-
singular). Adding some row to the second row if necessary, we may assume
that a,, # 0. NOWadd a,i(l - a,,) times row 2 to row 1 get entry 1 in the
upper left corner. We may now make the other entries in column 1 equal
corner. We may now make the other entries in column 1 equal to zero by
adding suitable multiples of row 1 to the other rows, and so A has been
transformed into A , .
For the inductive step, we may assume that A has been transformed into a
matrix At as displayed above. Note that C is nonsingular (for det A, = det C).
Assuming that C has at least two rows, we may further assume, as in the base
step, that its upper left corner y,+l,t+l = 1 (this involves only the rows of C,
hence does not'disturb the top t rows of A,). Adding on a suitable multiple of
row t + 1 to the other rows of A, yields a matrix A,+,.
We may now assume that A has been transformed into
where p E K and ,u # 0. Adding suitable multiples of the last row to the other
rows cleans out the last column, leaving D(p).
In terms of matrix multiplication, we have shown that there is a matrix P,
which is a product of elementary transvections, with PA = D(p). Therefore,
A = P-'D(,u); this completes the proof because the inverse of an elementary
transvection is another such. Pl
222 8. Some Simple Linear Groups
Theorem 8.8.
(i) CL(m, K) is a semidirect product of SL(m, K) by I<".
(ii) SL(m, K) is generated by elementary transuections.
Proof. (i) We know that SL a C L (because SL = lter det), and it is easy to see
that A = (D(p): u , EK ( z K ') is a complement of SL.
(ii) By (i), each A E C L has a unique factorization A = UD(p), where
U E SE, D ( p ) E A, and det A = p. Therefore, A is unimodular if and only if
A = U . The result now follows from Lemma 8.7.
Notation. If V is an m-dimensional vector space over a field K, let Z(V) denote
the subgroup of GL(V) consisting of all scalar transformations, and let SZ(V)
be the subgroup of Z(V) consisting of all unimodular scalar transformations.
Let Z(m, K) z Z(V) denote the subgroup of all m x m scalar matrices aE,
and let SZ(m, K) r SZ(V) denote the subgroup of all aE with a m= 1. If
K = GF(q), we may also denote these subgroups by Z(m, q) and SZ(m, q),
respectively.
Theorem 8.9.
(i) The center of GL(V) is Z(V).
(ii) The center of SL(m, K) is SZ(m, K).
PI'QQJ(i) If T E CL(V) is not a scalar transformation, then there is u E V with
(v, Tu) independent; extend this to a basis (v, Tv, u3, ...,urn)of K It is easy to
see that {u, v + Tu, u3, ..., u,) is also a basis of V, so that there is a (non-
+
singular) linear transformation S: V -+ V with Sv = v, S(Tu) = u Tv, and
Su, = ui for all i 2 3. Now T and S do not commute, for TS(v) = Tu while
ST(v) = v + TV. Therefore, T 4 Z(GL(V)), and it follows that Z(GL(V)) =
Z(
(ii) Assume now that T E SL(V), that T is not scalar, and that S is the
linear transformation constructed in (i). The matrix of S relative to the basis
(v, Tu, u3, ..., urn)is the elementary transvection B,,(l), so that det(S) = 1
and S E SL(V). As in (i), T 4 Z(SL(V)); that is, if T E Z(SL(V)), then T = aE
for some a E K. Finally, det(aE) = am, and so a m= 1, so that SZ(V) =
Z(SL(V)).
Theorem 8.10. jSZ(m, q)l = d, where d = (m, q - 1).
ProoJ Let K = GF(q). We first show, for all a E K x , that a m= 1 if and only
if ad = 1. Since d divides m, a d = 1 implies a m= 1. Conversely, there are
The General Linear Group
integers a and b with d = am + b(q - 1). Thus
because aY-' = 1. Hence a m= 1 gives 1 = am"= ad.
It follows that SZ(m, q) z {a E I<" : an' = 11 z {a E K x :ad = 1). Therefore,
if n is a generator of K ", then SZ(m, q) z (n"Id)and hence I SZ(n1, q)/ =
d.
Our preceding discussion allows us to lengthen the normal series in
Lemma 5.6 as follows:
The center SZ(m, q) is abelian and so its composition factors are no secret
(they are cyclic groups of prime order, occurring with multiplicity, for all
primes dividing q - 1). We now consider the last factor group in this series.
Definition. If V is an m-dimensional vector spaces over a field I<, the pvojec-
iigle rsnimodrslar group PSL(V) is the group SL(V)/SZ(V).
A choice of ordered basis of V induces an isomorphism q : SL(V) 2
SL(m, K ) with q(SZ(V)) = SZ(nz, K), so that PSL(V) 2 SL(n2, K)/SZ(m,K).
The latter group is denoted by PSE(m, I().When I( = GF(q), we may denote
PSL(m, K) by PSL(m, q).
we shall see, in Chapter 9, that these groups are intimately related to
projective geometry, whence their name.
Theorem 8.11. If d = (m, q - I), then
IPSL(n2, q)l = (qm- l)(qm- q) ...(qm- qm-')Id.
ProoJ Immediate from Theorems 8.5 and 8.10. F3
8.1. Let H d SL(2, I<),and let A E H. Using the factorization A = U D ( p ) (in the
proof of Theorem 8.81, show that if A is similar to [Y !], then there is p E K
such that W contains
,:[ p s p ] n
8.2. Let B = Bij(l) E GL(m, K)= G. Prove that CG(B)consists of all those nonsingu-
lar matrices A = [aij]whose ith column, aside from a,,, and whose jth row, aside
from ajj, consist of all 0's.
226 8. Some Simple Linear Groups
that a # + 1, and C is the desired matrix. In the second case, Exercise 8.1
If T = diag(a-', a), where a is to be chosen, then H contains the commutator
We are done if a-2 # rfr. 1; that is, if a4 # 1. If q > 5, then such an a exists, for
a field contains at most four roots of x4 - 1. If q = 4, then every ,u E K
satisfies the equation x4 - x = 0, SO that a # 1 implies a4 # 1.
Only the case GF(5) EZ Z5 remains. Consider the factor P occurring in the
lower left corner R = ,u/3(a2- 1) of U . If p # 0, choose a = [2] E Z5; then
a 2 - 1 # 0 and U = B,,(A). Hence H contains the elementary transvection
U2 = B2,(- 2A) and we are done. If P = 0, then
Therefore, the normal subgroup H contains
for all v E Z5. If v = 2,u-', then the top row of this last matrix is [2 01, and
the theorem is proved. B l
Corollary 8.14. If K is an infinite field which either has characteristic # 2 or
is per$ect of characteristic 2, then PSL(2, K) is a sirnple group.
Proof. The finiteness of I< in the proof of the theorem was used only to satisfy
the hypotheses of Lemma 8.12.
Rernark. In Theorem 9.48, we will prove that PSL(2, K) is simple for every
infinite field.
Corolllsary 8.85. SL(2, 5) is not solvable.
Proof. Every quotient of a solvable group is solvable. if4
We have exhibited an infinite family of simple groups. Are any of its mern-
bers distinct from simple groups we already know? Using Theorem 8.11, we
see that both PSL(2,4) and PSL(2, 5) have order 60. By Exercise 4.37, all
simple groups of order 60 are isomorphic:
If q = 7, however, then we do get a new simple group, for IPSL(2, 7)1 = 168,
which is neither prime nor i n ! . If we take q = 8, we see that there is a simple
group of order 504; if q = 11, we see a simple group of order 660. (It is known
that the only other isomorphisms involving An's and PSLs, aside from those
displayed above, are Exercise 8.12: PSL(2, 9) r A , (these groups have order
360); Exercise 9.26: PSL(2, 7) E PSL(3, 2) (these groups have order 168);The-
=
orem 9.73: PSL(4,2) A , (these groups have order 20, 160).)
EXERCISES
8.4. Show that the Sylow p-subgroups of SL(2, 5) are either cyclic (when p is odd) or
quaternion (when p = 2). Conclude that SL(2, 5) $ S,.
8.5. What is the Sylow 2-subgroup of SL(2, 3)?
8.6. (i) Show that PSL(2, 2) r S,.
(ii) Show that SL(2,3) $k S, but that PSL(2, 3) r A,.
8.7. What are the composition factors of GL(2, 7)?
8.8. Show that if H a GL(2, K), where K has more than three elements, then either
H < Z(GL(2, K)) or SL(2, I<) IH.
8.9. (i) What is the commutator subgroup of GL(2,2)?
(ii) What is the commutator subgroup of GL(2,3)?
(iii) If q > 3, prove that the commutator subgroup of GL(2, q) is SL(2, q).
8.10. Prove, for every field K, that all transvections are conjugate in GL(2, K).
8.11. Let A be a unimodular matrix. Show that A determines an involution in
PSL(2, K) if and only if A has trace 0, and that A determines an element of order
+
3 in PSL(2, I<) if and only if A has trace 1. (Hirzt. Use canonical forms.)
8.12. Prove that any two simple groups of order 360 are isomorphic, and conclude
that PSL(2,9) s A A ((Hint. Show that a Sylow 5-subgroup has six conjugates.)
The simplicity of PSL(m, K) for all m 2 3 and all fields I< will be proved in
this section. In 1870, C. Jordan proved this theorem for I< = Z,,and L.E.
Diclcson extended the result to all finite fields K in 1897, four years after
Moore had proved the result for m = 2. The proof we present, due to E.
Artin, is much more elegant than matrix manipulations (though we prefer
matrices when m = 2).
An m x m elementary transvection Bij(/Z)represents a linear transforma-
tion T on an m-dimensional vector space V over K. There is an ordered basis
(v ,,..., v,) of Vwith Tv, = v,for all 1 # i and with TV,= vi + Loj. Note that
T fixes every vector in the ( i ~ z- 1)-dimensional subspace I1 spanned by all
v, # vi*
228 8. Some Simple Linear Groups

Definition. If V is an m-dimensional vector space over a field K , then a are there any others? Choose w E V with w 4 H; since T fixes Pi pointwise,
hypwpbaae I l in V is a subspace of dimension m - 1.
Tw = pw + h, where h E H.
The linear transformation T arising from an elementary transvection fixes If v E V and v 6 H, the lemma gives
the hyperplane H pointwise. If w E V and w & H, then ( w ) = { p w : p E K ) is
a transversal of H in V : the vector space T/, considered as an additive group, Tv = pv + h',
is the disjoint union of the cosets H + pw. Hence, every vector v E V has a where h' = (1 - p)h" + Ah, E H. If v is an eigenvector of T, then 'Ti, = pv for
unique expression of the form some p E M.But Tv = pv if and only if P = p and Ah = ( p - 1)h": sufficiency
is obvious; conversely, if flu = pu + hr,then ( P - p)v = h' E ( v ) n H = 0.
(i) If T is a dilatation, then p - 1 # 0 and, h" = A(p - 1)-lh. It follows
Lemma 8.14. Let H be a hyperplane in V and let T E G L ( V )fix H pointwise. If that v = w + ( p - 1)-'h is an eigenvector of T for the eigenvalue p. If
w E V and w 4 H, then (v,, . .., v,-,} is a basis of H, then adjoining v gives a basis of V , and the
T ( w ) = pw + Iz, matrix of T relative to this basis is D ( p ) = diag ( 1 , . . . , 1, p}.
(ii) If T is a transvection, then p = 1. Choose w 4 H so that Tw = w + lz,
for soine p E K and h, E H. Moreover, given any v E T/, where ii E H and h # 0. If v 6 H is an eigenvector of T, then av = Tv = u + h
T ( v ) = pw + h', for some a E K ; hence, (a - 1)v E ( 0 ) n H = 0, so that a = 1 and Tv = v. It
follows that T = I,, contradicting the proviso in the definition of transvec-
for sonle 12' E H. tion excluding the identity. Therefore, T has no eigenvectors outside of N.If
{h, h,, .. .,h,} is a basis of H , then adjoining w as the first vector gives an
Proof. We observed above that every vector in V has an expression of the ordered basis of V , and the matrix of T relative to this basis is Bl,(l). @!4
+
form 2.w h. In particular, T ( w ) has such an expression. If v E T/, then v =
Aw + 11" for some A E K and 12" E H. Since T fixes H, Cora~llary8.18. All transvections in G L ( m , I<) are conjugate.
+ h"
T ( v ) = /ZT(w)+ h" = /Z(pw + h,)
ProoJ Since transvections are, by definition, conjugates of elementary trans-
= p(Aw + h") + [(I - p)h" + Ah,] vections, it suffices to prove that any two elementary transvections are conju-
= pv + 12'.
gate to B2,(1). Let V be an m-dimensional vector space over K with basis
{v,, ..'i,v,}, and let T be the linear transformation with Tv, = v , + v, and
The scalar p = p ( T ) in Lemma 8.16 is thus determined uniquely by any T Tv, = v, for all 1 2 2. If i # j and A # 0, define a new ordered basis { u , , . . ., u,}
fixing a hyperplane pointwise. of V as follows: put v , in position i, put A-lv, in position j, and fill the
remaining m - 2 positions with v,, . . ., v, in this order (e.g., if m = 5, i = 2,
Defimitiom. Let T E G L ( V )fix a hyperplane H pointwise, and let p = p(T). If and j = 4, then {u,, ..., u,) = {v,, u,, v,, A-'v,, v,)). The matrix of T rela-
p # I , then T is called a dilatation; if p = 1 and if T # I,, then T is called a tive to this new ordered basis is easily seen to be Bij(l).Therefore B,, ( 1 ) and
transvection. B,j(A)are similar, for they represent the same linear transformation relative to
different choices of ordered basis. !El
The next theorem and its corollary show that the transvections just defined
are precisely those linear transformations arising from matrix transvections. If T E G L ( V ) is a transvection fixing a hyperplane H and if tv $ H , then
Tw = w'+ h for some nonzero h E H. If v E V, then v = ;Iw+ h" for some
Theorem 8.17. Let T E G L ( V )fix a hyperplane H pointwise, and let p = p(T). A E K and 12" E H, and (*) in the proof of Lemma 8.16 gives Tv = v + AIz
(because 1 - p = 0). The function y : V 4 K , defined by y ( v ) = y(Aw + h) =
(i) If T is a dilatation, then T has a matrix D ( p ) = diag(1, .. ., 1, p} (relative R is a K-linear transformation (i.e., it is a binetau$~szctiond)with kernel H. For
to a suitable basis of V ) . each transvection T, there is thus a linear functional cp and a vector Iz E ker cp
(ii) If T is a traizsvection, then T has matrix B12(1)(relative to a suitable basis with
of V ) . Moreover, T has no eigenvectors outside of H in this case. Tv=v+y(v)h forall U E K

PB'OOJEvery nonzero vector in H is an eigenvector of T (with eigenvalue 1); Notation. Given a nonzero linear functional cp on V and a nonzero vector
230 8. Some Simple Linear Groups

tl E ker q , define (q, h): V 4 V by exception to the theorem. In this case,

{y, h): vwv + y(v)h.
and (S,)' = A,, a proper subgroup.
It is clear that (q, h) is a transvection; moreover, for every transvection T, We have seen that any two transvections are conjugate in GL. It is easy to
there exist q # 0 and h # 0 with T = {q, h). see that

Lemma 8.19. Let V be a vector space over I<.

(i) Jf q and $ are Iinear functionals on V , and if h, 1 E V satisfy q(h) = $(h) = are not conjugate in SL(2, 3); indeed, these transvections are not conjugate in
q(E), then SE(2, K) for any field IC in which - 1 is not a square. The assumption m 2 3
(q, h ) {q, 1) = (q, + I) '2nd (4%h) ($7 h) = {cp + $, h). in the next result is thus essential.

(ii) For all a E K ',
(@Y,h) = (9, ah).
(iii) (q, h) = ($, I) if and only fi there is a scalar a E K with
$=ay and h=aE.
(iv) If S E GL(V), then
S{q, h}S" = (qS-', Sh).
Proof. All are routine. For example, let us prove half of (iii). If {q, h) = {$, I),
then q(v)Iz = $(v)l for all v E V. Since y f 0, there is v E V with q(v) 0, SO
that h = y(v)-'tb(v)I; if a = q(v)-'$(v), then h = al. To see that $(u) = aq(u)
for all u E V, note that q(u) = 0 if and only if $(u) = 0 (because both h, 1 # 0).
If $(u) and y(u) are nonzero, then h = q(u)-'$(u)l implies q(u)-'$(u) =
q(v)-'$(v) = a, and so $ = ay. BI
Theorem 8.20. The commutator subgroup of GL(V) is SL(V) unless V is a
two-clii?zensionalvector space over Z,.
Proof. Now det: GL 4 K " has kernel SL and GL/SL z I< " ; since K x is
abelian, (GL)' I SL.
For the reverse inclusion, let v: GL -+ GL/(GL)' be the natural map. By
Corollary 8.18, all transvections are conjugate in GL, and so v(T) = v(TJ)for
all transvections T and T'; let d denote their common value. Let T = {q, h)
be a transvection. If we avoid the exceptional case in the statement, then H
contains a nonzero vector 1 (not necessarily distinct from h) with h + 1 # 0.
+
By the lemma, (9, h) 0 (9, I) = {q, h l) (these are transvections because
1 # 0 and h + 1 # 0). Applying v to this equation gives d2 = d in GL/(GL)',
whence d = 1. Thus, every transvection T E ker v = (GL)'. But SL is gener-
ated by the transvections, by Theorem 8.8(ii), and so SL < (GL)'. IBI
If V is a two-dimensional vector space over Z,,then GL(V) is a genuine
Theorem 8.21. If m 2 3, then all transvections are conjugate in SL(V).
I$Yoo+Let (9, h) and ($, 1) be transvections, and let H = ker cp and L =
ker t,b be the hyperplanes fixed by each. Choose v, u E V with q(u) = 1 = $(u)
(hence v 6H and u L). There are bases (h, h,, . . ., h,,z-,) and (I, I,, ... , Em-,)
of H and L, respectively, and adjoining v and u gives bases (v, h, h,, . . . , hm-, )
and (u, 1, l,, . . ., Em-') of 1V. If S E GL(V) takes the first of these ordered bases
to the second, then
(4 S(v) = u, S(H) = L, and S(h) = 1.
+ Let ,det S = d; we now show that we can force S to have determinant 1.
jn
Since 2 3, the first basis of V constructed above contains at least one other
vector\ (say, hm-,) besides v and h. Redefine S so that S(lt,,-,) = d-llm-, .
Relative to the basis {v, Iz, h,, . . . , hm-,), the matrix of the new transforma-
tion differs from the matrix of the original one in that its last column is
multiplied by d-'. The new S thus has determinant 1 as well as the other
properties (*) of S.
Now S(cp, h)S-' = {qS-', Sh) = (qS-', I), by Lemma 8.19(iv). Since qS-'
and $ agree on the basis (u, 1, I,, . . ., 1,-,) of V, they are equal. Therefore
{cp, h) and {$, 1). are conjugate in SLYas desired.
Notation. If H is a hyperplane in a vector space V, then
F ( H ) = (all transvections fixing H ) v (1,).
Lemma 8.22. Let H be a hyperplane in an m-dimensional vector space V over
K.
(i) There is a linear functional 40 with H = lcer q so that
(ii) F ( H ) is alz (abeliaiz) subgroup of SL(V), and F ( H ) r H.
(iii) The centralizer CsL(Y(H))= S Z ( V ) T ( H ) .
232 8. Some Simple Linear Groups
Proof. (i) Observe that linear functionals q and $ have the same kernel if and
only if there is a nonzero a E K with $ = aq. Clearly $ = a q implies ker $ =
ker q. Conversely, if H is their common kernel, choose w E V with w 4 H.
Now $(w) = aq(w) for some a E K x . If v E I/, then v = Lw + h, for some
1 E K and h E H, and $(v) = R$(w) = Laq(w) = aq(Lw + h) = crq(v).
If {q, h), {$, I) E F(H), then Lemma 8.19(ii) gives ($, 1) = {aq, 1) =
{q,al}. Since (q, h)-' = {q, - h), Lemma 8.19(i) gives {q, h) 0 {$, I)-' =
{cp, h - al) E F(H). Therefore, F ( H ) 5 SL(V).
(ii) Let q be a linear functional with H = ker p. By (i), each T E F ( H ) has
the form T = {q, h) for some h E H, and this form is unique, by Lemma
8.19(iii). It is now easy to see that the function F ( H ) -+ H, given by {q, h) r
12, is an isomorphism.
(iii) Since F ( H ) is abelian, SZ(V)F(H) 5: CsL(F(H)). For the reverse in-
clusion, assume that S E SL(V) commutes with every {q, h): for all h E H,
S {(p, h ) S-' = (9, h). By Lemma 8.19(iv), S(q, h) S-' = {qS-', Sh), and so
Lemma 8.19(iii)gives a E K x with
(**) qS-' =aq and Sh = a-'h,
Hence crS fixes H pointwise, so that crS is either a transvection or a dilatation.
If aS is a transvection, then aS E F(H), and so S = a-'(as) E SZ(V)F(N). If
aS is a dilatation, then it has an eigenvector w outside of H, and aSw = pw,
where 1 # ,u = det aS = a m(for det S = 1); hence, Sw = am-'w. But qS-'w =
rp(a-"+'w) = a-"+'q(w), so that (**) give q(w) = crmq(w).Since q(w) # 0 (be-
cause w & H), we reach the contradiction a m= 1. B!
Theorern 8.23 (Jordan-Dickson). If m 2 3 and V is an m-dimensional vector
space over a field K, then the groups PSL(V) are simple.
Proof. We show that if N is a normal subgroup of SL(V) containing some A
not in SZ(V), then N = SL(V); by Theorem 8.17, it suffices to show that N
contains a transvection.
Since SL(V) is generated by transvections, there exists a transvection T
which does not commute with A: the commutator B = T-'A-'TA # 1. Note
that N a SL gives B E N. Thus
where each T, is a transvection. Now T, = {qi, hi), where hi E Hi = ker qi for
i = 1,2; that is,
T,(v) = v + qi(v)hi for a11 u E V:
Let W be the subspace (h,, h2) 2 I/, SO that dim W < 2. Since dim V 2 3,
there is a hyperplane L of V containing W. We claim that B(L) I
then
B(1) = Tl T2(1) = T2(1) + cpl(TZ(l))hl
L. If 1 E L,
We now claim that HI n H , # 0. This is surely true if H, = H,. If HI # H,,
then H, $ H, = V (hyperplanes are maximal subspaces) and dim(H, + H,) =
rn. Since
dim HI + dim N, = dim(H, + N,) + dim(H, n H,),
we have dim(H, n Hz) = m - 2 2 1.
If z E N1 n H , with z # 0, then
B (z) = T1 T, (z) = z.
We may assume that B is not a transvection (or we are done); therefore,
B & F(L), which is wholly comprised of transvections. If B = as, where
S E F(L), then z is an eigenvector of S (z = Bz = aSz, and so Sz = a-'2). As
eigenvectors of transvections lie in the fixed hyperplane, z E L and so a = 1,
giving the contradiction S = B. Therefore, B $ SZ(V)Y(L) = C,,(F(L)), so
there exists U E F ( L ) not commuting with B:
of course, C = (UBUF1)B-I E N. If 1 E L, then
because B-'(1) E L and U-I E F ( L ) fixes L. Therefore, the transformation C
fixes the hyperplane L, and so C is either a transvection or a dilatation. But
C is not a dilatation because det C = 1. Therefore C is a transvection in N,
and the proof is complete. Ell
We shall give different proofs of Theorems 8.13 and 5.22 in Chapter 9.
Obseqve that IPSL(3,4)1 = 20,160 = $8!, so that PSL(3,4) and A , are
simple d o u p s of the same order.
\
Theorem 8.24 (ScRottenfeBs, 11900). PSL(3, 4) and A8 are noizisornorphic simple
groups of tht! same order.
P r o d The permutations (1 2) (3 4) and (1 2) (3 4) (5 6) (7 8) are even (hence
lie in A,), are involutions, and are not conjugate in A, (indeed, they are not
even conjugate in S8 for they have different cycle structures). We prove the
theorem by showing that all involutions in PSL(3,4) are conjugate.
A nonscalar matrix A E SL(3,4) corresponds to an involution in PSL(3,4)
if and only if A2 is scalar, and A2 is scalar if and only if (PAP-')' is scalar for
every ncnsingular matrix P. Thus A can be replaced by anything similar to
it, and so we may assume that A is a rational canonical form. If A is a direct
sum of 1 x 1 companion matrices, then A = diagja, p, y}. But A2 scalar im-
plies a2 = p' = y2;as GF(4) has characteristic 2, this gives a = /3 = y and A
is scalar, a contradiction. If A is a 3 x 3 companion matrix,
A=
[, r S]
1 a O ,
234 8. Some Simple Linear Groups
then A' has 1 as the entry in position (3, I), and so A2 is not scalar. We
conclude that A is a direct sum of a 1 x 1 companion matrix and a 2 x 2
companion matrix:
Now det A = 1 = a p (remember that - 1 = 1 here), so that p = a-', and A2
scalar forces y = 0. Thus,
There are only three such matrices; if n is a primitive element of GF(4), they
are
0 1 0 0 1 0
Note that A2 = E, B2 = n2E, and C2 = nE. It follows that if 1M E SL(2, 3)
and M2 = E (a stronger condition, of course, than M2 being scalar), then M
is similar to A; that is, M = PAP-' for some P E GL(3,4). In particular, n2B
and nC are involutions, so there are P, Q E GL(3,4) with
PAP-' = n2B and QAQ-I = nC.
Since [GL(3,4): SL(3,4)] = 3 (for GL/SL r GF(4)") and since the matrix
diagjn, 1, 1) of determinant n # 1 commutes with A, Exercise 3.7 allows us
to assume that P and Q lie in SL(3,4). It follows that A, B, and C become
conjugate in PSL(3,4), as desired. B l
Theorem 8.24 can also be proved by showing that PSL(3,4) contains no
element of order 15, while A, does contain such an element, namely,
(1 2 3)(4 5 6 7 8).
One can display infinitely many pairs of nonisomorphic simple groups
having the same finite order, but the classification of the finite simple groups
shows that there do not exist three nonisomorphic simple groups of the same
order.
Classical Groups
At the end of the nineteenth century, the investigation of solutions of systems
of differential equations led to complex Lie groups which are intimately re-
lated to simple Lie algebras of matrices over C. There are analogues of these
Classical Groups 235
Lie groups and Lie algebras which are defined over more general fields, and
we now discuss them (not proving all results).
In what follows, all vector spaces are assumed to be finite-dimensional.
Defenitiow. If V is a vector space over a field K, a function f : V x V -r K is
called a bilinear form if, for each v E V, the functions f (v, ) and f ( ,v) are
linear functionals on V.
A bilinear form f is called symmetric iff (v, u) = f (24 v) for all u, v E V , and
it is called alternating iff (v, v) = 0 for all v E V.
+
Iff is alternating and u, v E V, then 0 = f(u v, zr + v) = f(u, a) + f(u, v) +
+ +
f(v, U) f (u, v) = f(u, v) f (v, u), SO that f (v, u) = -f (u, v). Conversely, iff
is a bilinear form for which f(v, u) = -f(u, v), then 2f(v, v) = 0 for all v E I/. If
K has characteristic # 2, then f is alternating; if K has characteristic 2, then
f is symmetric.
There is another interesting type of form, not quite bilinear (Bourbaki calls
it ccsesquilinear").
Definition. If K is a field having an automorphism a of order 2 (denoted by
a: a- a"), then a hea~rmitianfarmon a vector space V over K is a function
!t: V x V 4 I< such that, for all u, v E V:
(i) h(u, v) = h(v, u)";
(ii) h(au, v) = ah(u, v) for all a E I<; and
(iii) h(u i- v, w) = h(u, W)+ h(v, w).
Note that if h is hermitian, then h(u, pv) = h(pv, rr)" = (ph(v, u))" =
puh(v,u)" = puh(u, v). Moreover, h is additive in the second variable, for
+
h(u, v w) = h(v + w, u)" = (h(v, u) + h(w, u))" = h(v, u)" + h(w, u)" = lr(tr, v) +
h(u, IV). \
Complex conjugation z l-;.Z is an automorphism of C of order 2. If V is a
complex vector space with basis (x,, . . . , x,), if x = z a j x j , and if p = xPjxj
(where aj, bj E C), then
h(x, y) = ajj8,
is a hermitian form.
If K is a finite field, then it has an automorphism u of order 2 if and only if
K 2 GF(q2) for some prime power q, in which case a" = a q (this can be
shown using Theorem 8.4).
Definition. If f : V x V -r IC is either symmetric, alternating, or hermitian,
then we call the ordered pair (V,f ) an inrzev prodirct space.
Definition. Let (V, f ) be an inner product space. If (v,, . . . , v,,) is an ordered
basis of k: then the intnnerprodidct matrix off relative to this basis is
236 8. Some Simple Linear Groups Classical Groups 237

It is clear that f is completely determined by an inner product matrix, for Conversely, if u satisfies f (u, v) = 0 for all v E I/, then f (u, ui) = 0 for all vi
if ti = aiui and w = Pivi, then (where {v,, . . . , v.1 is a basis of V). If u = pjvj, then x j p jf(vj, vi) = 0; if
Y = ( p l ,. . . ,prl)is the column vector of u, then Y # 0 and A Y = 0. Hence A
(u, W ) = C aiPjf (vi, vj)-
i,j
is singular. !El

What happens to the inner product matrix after changing basis? Definition. An isometry of a nondegenerate space (I/,f ) is a linear trans-
formation T: V 4 V such that
Lemma 8.25. Let (I/, f ) be an inner product space, let {v,, ..., vnJ and
{u,, . . . , u,J be ordered bases of V, and let the corresponding inner product
f (Tu, Tu) = f (24, v)
matrices be A and B. for all u, v E Y;

(i) Iff is bilinear, then A and B are congruent; that is, there is a nonsingular Lemma 8.27. If (v f ) is a nondegenerate space, then every iso~netryis non-
matrix P with singular, nnd so all the isometries form a subgroup Isom(V, f ) I GL(V).
B = P'AP.
(ii) If f is hermitian, then A and B are a-congruent; that is, there is a ProoJ If T is an isometry and Tu = 0, then f (u, v) = f (Tu, Tv) = f (0, Tv) = 0
nonsingular inatrix P = [pijJ with for all v E K Since f is nondegenerate, it follows that u = 0 and T is an
injection; since V is finite-dimensional, T is nonsingular, B
B = PtAPa,
where Po = [(pij)"]. Lemma 8.28. Let (V, f ) be a nondegenerate space, let A be the inner product
In either case, B is nonsingular if and only fi A is nonsingular. matrix off relative to an ordered basis {v,, . . .,u,,} of arzd let T be a linear
transfortnatiorz on K
ProoJ. (i) Write uj = zpijvi; the matrix P = [pljJ, being a transition matrix (i) Iff is bilinear, then T is an isometry fi and only if its matrix M = [mu]
between bases, is nonsingular. Now relative to the ordered basis satisfies
MtAhl = A;
in this case, det M = )1.
in matrix terms, this is the desired equation (the transpose is needed because (ii) If J; is hermitian, then T is an isometry fi arzd only fi
the indices 12, i in the first factor must be switched to make the equation '\ MtAMa = A,
correspond to matrix multiplication). The last statement follows from det B =
det(PtAP) = det(P)2 det(A). where M" = [(mij)o]; in this case, (det M)(det Ma) = 1.

(ii) f (ui7 uj) = C ~ (vk, v~)PP,.

k i f
k, 1 ProoJ (i)

Definition. An inner product space (I!f ) is nondegenerate (or nonsingular) if = 1mlif ( ~ rvk)mkj
, = f (vi, vj)-
one (and hence any) of the inner product matrices off is nonsingular. 1, k

Lemma 8.26. An inner product space (V,f ) over a field K is nondegenerate if
and only if f(u, v) = 0 for all v E V implies u = 0.
Proof. Let {v,, . . ., v,) be a basis of K If an inner product matrix A off is
singular, then there is a nonzero column vector Y with AY = 0; that is, if
Y = (p,, . . . , p,), where pi E K, then u = pivi is a nonzero vector with
f(v,u) = XtAY = 0
for all v = x Aivi (where X = (4,. .., A,)).
After translating into matrix terms, this is the desired equation. It follows
t det A. Moreover, nondegeneracy of ( V , f ) g'~vesnon-
that (det ~ ) ~ ( dA)e =
singularity of A, and so det M = f 1.
(ii) The obvious modification of the equations above leads to the desired
matrix equation and its consequence for determinants. tFl
The group GL(V) acts on F ( V ) ,the set of all functions V x V 4 K : i f f is
a function and P E CL(V), define
238 8. Some Simple Linear Groups
it is easily checked that if Q E GL(V), then f = (f ')Q (this is the reason for
the inverse). Notice that iff is either symmetric, alternating, or hermitian,
then so is f'.
Theorem 8.29. Let V be a vector space over a field K , and let F ( V ) be the
GL(V)-set of all functions V x V -+ K. When f is either symmetric, alternat-
ing, or hernzitian, then the stabilizer off is Isom(V, f ); moreover, if g is in the
orbit off, then Isom(V, g) is isomorphic to Isom(V, f ) (indeed, they are conjzr-
gate subgroups of GL(V)).
Proof. The stabilizer GL(V), of f is { P E GL(V): f P = f 1, so that P E
GL(V)f if and only if f(u, v) = f (P-'u, P-'v) for all u, v E V; that is, P-' E
Isom(V, f ) and hence P E Isom(V, f). By Exercise 3.37, Isom(V, f P ) =
P Isom(V, f)P-'. !El
Definition. Two functions f, g E F ( V ) are called eqirivalent if g =f for some
P E GL(V).
If follows from the theorem that equivalent symmetric, alternating, or
hermitian forms determine isomorphic groups of isometries.
Lemma 8.30. Two bilinear forrns f, g E F ( V ) are equivalent if and only if they
have inner product matrices A and B which are congruent. Two hermitian forms
(relative to the same automorphism o)are equivalent if and only if their inner
product matrices are a-congruent.
Proof. By Lemma 8.25(i), we may assume that all inner product matrices are
determined by the same basis of I/.
Let X and Y be column vectors. Iff and g are bilinear and g = f P, then
g(X, Y) = XtBY (see the proof of Lemma 8.26). By definition, f '(x, Y) =
f (P-I X, P-I Y) = (P-'X)'A(P-' Y) = X' [(P-')'AP"] Y. Since this equation
holds for all X and Y, it follows that B = (P-')'AP-'; hence, A and B are
congruent.
Conversely, if B = QtAQ for some nonsingular Q, then XtBY =
X'QtAQY = (QX)'A (Q Y), SO that g (X, Y) = f (Q-'X, Q-' Y).
This argument, rnzrtatis mutandis, works in the hermitian case as well.
Let (V, f ) be a nondegenerate space. We are going to see that all alternat-
ing forms f are equivalent, and so there is, to isomorphism, just one isometry
group Isom(l/, f); it is called the symplectic group. If V is an n-dimensional
vector space over K, then Isom(v f ) is denoted by Sp(V) or by Sp(n, K); if
K = CF(q), one writes Sp(n, q).
It is true that all hermitian forms are equivalent, and so there is, to isomor-
phism, just one isometry group Isom(v f ) in this case as well; it is called the
rrrritary grorrp. The group Isom(V, f )is now denoted by U(V) or by U(n, K);
Classical Groups 239
when K = GF(q2), one writes U(n, q2) (recall that the only finite fields for
which hermitian forms are defined are of the form GF(q2)).
It is not true that all symmetric forms over a finite field are equivalent, and
it turns out that inequivalent forms give nonisomorphic groups. The groups
Isom(V;f ) are called orthogoaalgroarps; in odd dimensions over a finite field
of odd characteristic, there is only one orthogonal group, but in even dimen-
sions over finite fields of any characteristic, there are two orthogonal groups,
denoted by 0 + ( V )and by 0 - ( V ) .
Definition. A group is a classicalgrorrp if it is either general linear, symplectic,
unitary, or orthogonal.
We remark that the term classical group is usually not so precisely defined;
for most authors, it also encompasses important groups closely related to
these as, say, SL(V) or PSL(V).
We now discuss symplectic groups; afterwards we will describe corres-
ponding results for the unitary and orthogonal groups.
Lemma 8.31. If (I/, f ) is a nondegenerate space, then for every linear functional
gE V* (the dual space of V), there exists a unique x E V with g = f(x, ).
Prooj: We first prove that if {v,, . . . , v , ] is a basis of K then (f(v,, ), . . . ,
f(vn, )) is a basis of V*. Since dim V* = n (by standard linear algebra), it
suffices to prove that these n linear functionals are independent. Otherwise,
1
there are scalars /Ii, not all 0, with /Iif(vi, ) = 0; that is, ,Iif(ui, x) = 0 for
1
all x E V. If z = Aivi, then f(z, x) = 0 for all x E K Thus, z = 0, because f is
nondedenerate, and this contradicts the independence of the vi.
E Vr, there are scalars pi with g = pif (v,, ), and g(v) = f (x, v) for
all v E V. TGprove uniqueness of x, suppose that f (x, v) = f (y, v) for all v E K
Then f (x - y, v) = 0 for all v E I/, and so nondegeneracy gives x - y = 0. B l
Definition. Let (I/, f ) be an inner product space. If x,y E V, then x and y are
orthogonal iff (x, y) = 0. If Y is a nonempty subset of V, then the ortliogon~l
complement of Y, denoted by YL, is defined by
y L = {v E V: f(v, y) = 0 for all y E Y).
It is easy to see that YL is always a subspace of V. Using this notation, an
inner product space (I/,f ) is nondegenerate if and only if VL = 0.
Let ( V , f ) be a nondegenerate space. If W 5 V is a subspace, then it is
possible that the restriction f l(W x W) is degenerate. For example, let V =
(x, y) be a two-dimensional space and let f have inner product matrix A
relative to this basis:
240 8. Some Simple Linear Groups
thus, f is symmetric and nondegenerate. However, if W = (x), the restric-
tion f l(W x W )is identically zero.
Lemma 8.32. Let (V,f ) be an inner product space, and let W be a subspace of
V.
(i) Iff l(W x W) is nondegenerate, then
v = w@wl
(ii) If ( V ,f ) is a nondegenerate space and V = W @ W', then f l(W' x w') is
rzorzdegenerate.
Proof. (i) If x E W n W', then f(x, W) = 0, and so nondegeneracy gives x =
0. If v E V, then the restriction g = f(v, )I W is a linear functional on W, and
so there is wo E W with g(w) = f(v, W)= f(wo, W)for all w E W. But v = wo
( u - w,), where wo E W, and v - w, E WL.
(ii) If {v,, . . . , v,) is a basis of W and {v,,,, . . ., v,,) is a basis of W', then
the inner product matrix A off relative to the basis (v,, ...,v,) has the form
so that det A = (det B)(det C). But A nonsingular implies C nonsingular; that
is, the restriction off is nondegenerate. El
Assume that (I/, f ) is an inner product space with f alternating. Iff is not
identically zero, there are vectors x and y with f (x, y) = a #
by a-'x if necessary, we may assume that f(x, y) = 1. If dim
inner product matrix is thus
Definition. A hyperbolic plane is a two-dimensional nondegenerate space
(V, f ) with f alternating.
We have just seen that every two-dimensional inner product space (V,f )
with f alternating and not identically zero is a hyperbolic plane.
Theorem 8.33. If (V,f ) is a nondegenerate space with f alternating, then V is
even-dimensional. Indeed,
v = w, @ . . . @ w;,
where each M: is a hyperbolic plane and the summands are p~irwiseorthogonal;
that is, if i # j, then f ( q . , y ) = 0.
Proof. We proceed by induction on dim V 2 0, the base step being trivial. If
Classical Groups 24 1
dim V > 0, then our discussion above shows that there exist x,, y, E V with
f(xl, y,) = 1. If W = (x,, y,), then Wis a hyperbolic plane and f / ( W x W)
is nondegenerate. By Lemma 8.32(i), V = W @ WL; by Lemma 8.32(ii),
f j(wL x w') is nondegenerate, and so the inductive hypothesis shows that
W' is an orthogonal direct sum of hyperbolic planes.
Definition. If (Kf ) is a nondegenerate space with f alternating, then a sym-
plectic basis of V is an ordered basis
with f (xi, xj) = 0 = f ( ~ iyj)
, for all i, j, and f (xi,yj) = 6,, = -f (y,, xi), where
= 1 if i = j and aij = 0 if i f j.
+ Thus, all inner products are 0 except f (xi, y i ) = 1 = -f (y,, xi) for all i.
Theorem 8.34. Let (V,f ) be a nondegenerate space with f alternating.
(i) V has a symplectic basis {x, , y, , x,, y,, .. . , x,, y,).
(ii) The inner product matrix A off relative to this ordered basis is the matrix
J which is the direct sum of 2 x 2 bloclcs
[ 11.
(iii) If u = (aixi + Piyi) and v = 1(yixi+ 6,yi), then
1
f (u, v) = (aiai - Piyi).
(iv) All nondegenerate alternating forms on V are equivalent, and so the sym-
plectic groups Isom(V, f), to isomorphism, do not depend on f.
Prooj: (i) If V is the orthogonal direct sum of hyperbolic planes w,
then the
union of the bases of the is a symplectic basis of K
(ii) and (iii) are now routine calculations.
(iv) If g is a nondegenerate alternating form, then there is a symplectic basis
of V relative to g, so that any inner product matrix of g is also congruent to '
J, and hence to any inner product matrix off. Therefore, the isometry group
Isom(% f ) does not depend on f, by Theorem 5.29.
Notice that alternating bilinear forms are thus sums of 2 x 2 determinants.
Note also that if one reorders a symplectic basis so that all the xi precede all
the y,, then the matrix J is congruent to
where E is the 1 x I identity matrix.
242 8. Some Simple Linear Groups Classical Groups

Definition. If (V, f ) is a nondegenerate space, then the adjoint of T, denoted
by T*, is a linear transformation on V for which
f(Tx,y)=f(x,T*y) forall x , y ~ V .
Lemma 8.35. Let (I/, f ) be a nondegenerate space, and let T be a linear trans-
formation on V having an adjoint T*. Then T is an isometry if aizd only if
T"T = 1,.
It follows that if T* has matrix
then M = -BE. Similar calculations give the other three blocks. For existence
of T*, it is routine to checlc that the matrix construction in the statement
defines a linear transformation that behaves as an adjoint must. The last
statement follo-tvs from Lemma 8.35.

ProoJ: If T has an adjoint with T*T = I,, then, for all x, y E V, f(Tx, Ty) = By Lemma 8.28, symplectic matrices have determinant $- 1, but it can
f (x, T*Ty) = f (x, y), so that T is an isometry. be shown, in fact, that every symplectic matrix has determinant 1; that is,
Conversely, for all x, y E V, Sp(V) 5 SL(V).

f (x, T*Ty - Y)= f (x, T* TY)- f (x, Y)

=f(Tx, TY)- f ( x , Y)= 0, 8.13. Show that Sp(2, M) = SL(2, K) for all fields K.
because T is an isometry. Since f is nondegenerate, T*Ty =y for all y E V, 8.14. Let (I.:f ) be a nondegenerate space with f alternating.
and so T*T = 1,. blil (i) Show that T E GL(V) is symplectic if and only if T carries symplectic bases
to symplectic bases.
One can prove uniqueness of adjoints in general, but we may observe here (ii) If V is a vector space over a finite field K , show that ISp(V)I is the number
that uniqueness holds for adjoints of isometries T because T* = T-I. It fol- of ordered symplectic bases.
lows that TT* = 1,.
How can one recognize a syknplectic matrix? The isometry group Isom(V,f ) is called a unitary grorsp when f is non-
degenerate hermitian, and it is denoted by U(V), U(n, I<), or U ( n , q2) (when
\\ K is finite, a hermitian form requires 1K1 to be a square).
Theorem 8.36. Let T E GL(V) a h let {x,, y,, . . . , x,, y,) be a symplectic basis We now state without proofs the analogues of Theorem 8.34 and 8.35 for
of V. If the matrix Q of T relative to this basis is decomposed into 1 x 1 unitary groups.
blocks
Definition. If (V, f ) is an inner product space, then an ovtho~sormalbasis is an
ordered basis (v,, . . ., v,) with S(vi, vj) = 6,.

then T* exists and has matrix Theorem. Let (V,f ) be a nondegenerate space with f hermitian.
(i) V has an orthorzormal basis (x,, . . . ,x,).
(ii) The inner product matrix off relative to this basis is the identity matrix. ,

(iii) If u = Aixi and v = Cpixi, then

moreover, Q E Sp(21, K) if and only if Q*Q = E.
ffu, V ) = zAip,P-
Pro06 Assume that T* exists. If Txi = zv(orvixv+ PViy,),then (iv) All nondegenerate alternating forins on V are equivalent, and so unitary
groups Isom(V, f ) do not depend on f.
f (xi, T*xj) = f (Txi, xj) (v) If M E GL(V), then M* = (Mt)", aizd so M E U(V) if and only if
= 1avif(xv9 xj) + C Bvif(yv7 xj) = -Pji.
v v
M(Mt)" = E.
(vi) A linear transformation T E GL(V) is unitary if and only if it takes ortho-
On the other hand, if T*xj = zr(Arjxr+ prjyr),then normal bases into orthonormal bases.

The isometry groups Isom(V, f ) are called orthogonal groups when f is

nondegenerate symmetric; we restrict the discussion to finite fields of scalars
244 8. Some Simple Linear Groups
K. If I<= GF(q) has odd characteristic and dim V = 21 + 1 is odd, there is a
basis of I/ relative to which f has inner product matrix
where E is the 1 x 1identity matrix. All forms are thus equivalent (for all inner
product matrices are congruent to J), and so there is, to isomorphism, just
one group Isom(V; f )in this case. It is denoted by O(V), 0(21+ 1, K), or by
+
O(21 1, q).
If dim V = 21 is even and K = GF(q) has odd characteristic, then there are
exactly two inequivalent forms: every inner product matrix is congruent to
E O O O
JE=/0 0 1 0 ' 1 irreducible in K [ t ] . One proves that there are only two nonisomorphic
LO o o EJ
+
where E is the ( I - 1) x (1 - 1) identity matrix and 8 = 1. The correspond-
ing isometry groups are not isomorphic; they are denoted by O f (21, K) and
by 0-(21, K) (or by 0+(21, q) and by 0-(21, q) when K = GF(q)).
Before describing the orthogonal groups in characteristic 2, let us recall the
definition of the real orthogonal group O(n, R) given in Chapter 3 in terms of
distance: if T E O(n, R), then/'1/Tvll = Ilvil for all v E Rn. In Theorem 3.29,
however, it was shown that T is orthogonal in this sense if and only if
(zr, v) = (Tu, Tu) for all u, v E IWn (where (u, v) is the usual dot product of vec-
tors in Rn). Given any symmetric bilinear form f, define the analogue of
llv112 = (u, v) to be Q(v) = f(u, v). If U, v E V; then
f (u + VY u + 0) = f (u, U)+ 2f (u, v) + f (v, v);
that is, Q(u + v) - Q(u) - Q(v) = 2f(u, 0).
Definition. A quadraticform on a vector space V over a field K is a function
Q: V -+ K such that:
(i) Q(au) = a2Q(u)for all u E V and a E I<; and
(ii) Q(u + v) - Q(u) - Q(v) = g(u, v), where g is a bilinear form on V (one
calls g a bilinear form associated to Q).
A quadratic form Q is nondegenerate if (V, f ) is nondegenerate, where f is
an associated bilinear form.
If K has characteristic # 2, one can recapture f from Q: f(u, u) =
i(Q(u + v) - Q(u) - Q(u)),and there is just one bilinear form associated to Q.
The proof of Theorem 3.28 shows that a linear transformation T on V lies in
0( Y )(i.e., f ( Tu, Tv) = f (u, v) for all u, v E V) if and only if Q (Tu) = Q(u) for all
u E V. Thus, orthogonal groups in odd characteristic could have been defined
in terms of quadratic forms.
Classical Groups 245
Assume now that K = GF(q) does have characteristic 2. Bilinear forms g
associated to a given quadratic form Q are no longer uniquely determined
by Q; however, g satisfies two extra conditions: first, g(u, u) = 0 (because
u + u = 0); second, g is symmetric. Since - a = a! for all a E K , it follows that
g is alternating. Theorem 8.33 shows that if (V; g) is to be nondegenerate, then
V must be even-dimensional, say, dim V = 21. If K = GF(q) has characteris-
tic 2 and Q is a nondegenerate quadratic form on a 21-dimensional vector
space V over I<, then the orthogonal group O(V,Q) is defined as
O(V;Q) = { T E GL(V): Q(Tz4) = Q(u) for all u E V}.
Given a nondegenerate quadratic form Q, there is always a symplectic basis
(x,, yl, . . . , x,, y,} such that Q(xi) = 0 = Q(yi)for all i < 1; moreover, either
(+): Q(x,) = 0 = Q(yr) or (-): Q(yl) = 1 and Q(x,) = y, where t 2 + t + y is
groups for a given V arising from different quadratic forms Q (one from each
possibility just described); they are denoted by O'(21, K) and by 0-(21, I<); as
usual, one may replace I< by q in the notation when K = GF(q).
Each type of classical group gives rise to a family of simple groups; we
describe the finite such.
The center of Sp(21, q) consists of $. E. Define
+
PSp(21, q) = Sp(21, q)/( E ) .
These groups are simple unless (2E, q) = (2, 2), (2, 3), or (4, 2). Moreover,
where d = (2, q - 1).
The center of U(n, q2) consists of all scalar transformations LEYwhere
M" = 1. Let SU(n, q2) 1 U(n, q2) consist of all unitary transformations of
determinant 1, and define
These groups are simple unless (n, q2) = (2,4), (2, 9), or (3,4). Moreover,
where E = (n, q + 1).
Assume that q is a power of an odd prime. Let the commutator subgroup
of O(n, q) be denoted by R(n, q) (it is usually a proper subgroup of SO(n, q),
the subgroup consisting of all orthogonal transformations of determinant 1).
The center of R(n, q) consists of diagonal matrices having diagonal entries
$- 1, and one defines
PR(n, q) = R(n, q)/center.
When rz 2 5 is odd, then these groups are simple, and
246 8. Some Simple Linear Croups

i
where d = (2, q - 1). Notice that JPR(21+ 1, q)l = JPSp(21,q2)/; these two ( CHAPTER 9
groups are not isomorphic if I > 2 (Theorem 8.24, proved in 1900, exhibited 1

the first pair of nonisomorphic simple groups of the same order; here is an
infinite family of such examples).
When dim V = 21 is even, there are two orthogonal groups. If we denote
+
the commutator subgroup by RE,where E = 1, and the quotient Elcenter
by PaE,then the orders are:
I Permutations and the Mahhieu Groups

where g = (4, q' - E).

When q is a power of 2, then the groups PRc(21,q) are simple for 1 > 3, and
their orders are given by the same formulas as when q is odd.
For proofs of these results, the reader is referred to Artin (1957), Carter
(1972), and Dieudonnk (1958).
Every complex Lie group G determines a finite-dimensional Lie algebra
L(G) over C, and G simple implies L(G) simple. These simple Lie algebras
were classified by E. Cartan and W. Killing about 100 years ago; there are
four infinite classes of them and five "sporadic" such. In 1955, Chevalley
showed, for every finite field GF(q), how to construct analogues of these The Mathieu groups are five remarkable simple groups discovered by E.
simple Lie algebras over C; he also showed how to construct simple finite I Mathieu in 1861 and 1873; they belong to no infinite family of simple groups,
groups from them. These families of simple groups (four are doubly indexed 1 as do all the simple groups we have so far exhibited, and they are the first
by the dimension and the field GF(q); five of them, arising from the sporadic 1i examples of what are nowadays called sporadic simple groups. This chapter
Lie algebras, are singly indexed by q) are now called Chevalley groaps. There is devoted to proving their existence and displaying some of their interesting
are thus nine such families, which include all those arising from the classical II properties.
groups (special linear, symplectic orthogonal) except those arising from ti
unitary groups (however, the finite simple groups arising from the unitary
groups were known to Dickson in 1900). Simple Lie algebras over C are
classified by certain graphs, called Dynkin diagrams. In 1959, Steinberg Multiple Transitivity
showed that automorphisms of these graphs can be used to construct new
finite simple groups. There are four infinite classes of these simple groups, Even though permutation groups, orbits, and stabilizers were discussed in
called Steinberg groz~ps,two of which are the families PRc((m,q) for E = f1 Chapter 3, we repeat the basic definitions here for the reader's convenience.
arising from the unitary groups. In 1960, Suzuki discovered a new class Sz(q)
Definition. If X is a set and G is a group, then X is a G-set if there is a function
of simple groups, where q is a power of 2 (these are the only simple groups
a: G x X -+ X (called an action), denoted by or: (g, x) t+ gx, such that:
whose orders are not divisible by 3); in 1961, Ree discovered two more
infinite classes whose construction is related to that of the Suzuki groups; (i) l x = x for all x E X; and
these are the S~tzuIcigroacps and the Ree gro~ps.(The interested reader s h o ~ ~ l d (ii) g(hx) = (gh)x for all g, 32 E G and x E X.
consult the books of Carter and of Gorenstein for more details.) Collectively, One also says that G sets on X. If 1x1 = n, then n is called the degree of the
these 16 classes of simple groups are called the groups of Lie type. G-set X.
The classificatiorl theorem of finite simple g r o ~ p ssays that there are ex-
actly 18 infinite classes of them: the cyclic groups of prime order, the alternat- We assume throughout this chapter that all groups G and all G-sets X
ing groups, and the groups of Lie type; moreover, there are exactly 26 "spo- are finite; moreover, all vector spaces considered are assumed to be finite-
radic" simple groups, 5 of which are the Mathieu groups to be discussed in dimensional.
the next chapter. This theorem is one of the highest achievements of mathe- It is clear that if X is a G-set and H < G, then X is also an H-set (just
matics; it is the culmination of the work of about 100 mathematicians be- restrict the action G x X -, X to H x X). Let us also recall that G-sets are
tween 1955 and 1985, and it consists of thousands of journal pages. just another way of viewing perm~ltationrepresentations.
248 9. Permutations and the Mathieu Groups
Theorem 3.89. If X is a G-set with action a, then there is a Izomomorphism
8: G -+ Sx given by ai(g): x H gx = a(g, x). Conversely, every homomorpl~ism
q: G -. Sx defines an action, namely, gx = ga(g)(x).
Definition. A G-set X with action a is faithful if a": G -+ Sx is injective.
Thus, a G-set X is faithful if and only if gx = x for all x E X implies g = 1.
If X is a G-set with action a, the subgroup im a" Sx is a permutation
group; hence, if X is a faithful G-set, then G can be identified with im 8, and
we may view G itself as a permutation group. Cayley's theorem says that
every group G of order n has a faithful representation cp: G 4 S,; in this case,
G itself is a faithful G-set of degree n. Since IS,( = n! is so much larger than n,
however, faithful G-sets of smaller degree are more valuable.
EXAMPLE 9.1. This is a generic example. Assume that a set X has some "struc-
ture" and that Aut(X) is the group of all permutations of X which preserve
the structure. Then X is a faithful Aut(X)-set; indeed, X is a faithful G-set for
every G 5 Aut(X).
If X is a G-set and gx = x, where g E G and x E X, then one says that g
fires x. A G-set X in which each g E G fixes every x E X is called a triuial
G-set.
EXAMPLE 9.2. If X is a G-set with action a, let N = (g E G: gx = x for all
x E X). Then N = keri?, so that N a G, and it is easy to see that X is a
faithful (GIN)-set (where (Ng)x = gx).
Recall that a G-set X is transitive if, for every x, y E X, there exists g E G
with y = gx. If x E X, we denote its G-orbit {gx: g E G) either by Gx or by
O(x); thus, X is transitive if X = Gx for some x E X. The first observation
when analyzing G-sets is that one may focus on transitive G-sets.
Theorem 9.1. Every G-set X has a unique decomposition into transitive G-sets.
More precisely, X is partitioned into its G-orbits, each of which is a transitive
G-set. Conversely, if a G-set X is partitioned into transitive G-sets (Xi: i E I),
then tlze Xi are the G-orbits of X.
Proof. It is clear that the orbits partition X into transitive G-sets. Conversely,
it sufficesto prove that each Xi is an orbit. If xi E Xi, then the orbit Gxi c Xi
because Xiis a G-set. For the reverse inclusion, if y E Xi, then transitivity of
Xi gives y = gx, for some g E G. Hence, y E Gxi, and Xi c Gxi. Eli
Recall that the stabilizer ofx, denoted by G,, is defined by
Gx = {g E G: gx = x).
Multiple Transitivity
Theorem 9.2. If X is a transitive 6-set of degree n, and if x E X, then
IGI = rzlGxI.
If X is a faithful G-set, then I G,I is a divisor of (12 - I)!.
ProoJ By Theorem 3.20, 1 Gxl = [G : G,]. Since X is transitive, Gx = X, and
so n = IGl/lG,/. If X is faithful, then G, 5 Sx-i,), and the latter group has
order (n - I)!. E l
Here is a solution to Exercise 3.4. If a finite group G has only two con-
jugacy classes, then every two nonidentity elements are conjugate; that is,
G# = G - (1) is a transitive G-set (where G acts on G" by conjugation).
Theorem 9.2 says that I GI - 1 is a divisor of I GI; this can happen only when
IGl = 2.
Theorem 9.3. Let X be a transitive G-set, and let x, y E X.
(i) If tx = y for some t E G, then G,, = G,, = tG,t-I.
(ii) X has the same number of G,-orbits as of Gy-orbits.
ProoJ (i) If g fixes x, then tgt-'y = tgx = tx = y and tgt-I fixes y. Therefore,
tG,t-I 5 G,,. Similarly, t-'G,,t 5 G,, and this gives the reverse inclusion.
(ii) Since X is transitive, there is t E G with tx = y. Denote the G,-orbits by
(Gxzi:i E I}, where ziE X. If wi = tz, E X, then we shall show that the subsets
Gywiare Gy-orbits of X. Clearly, these subsets are transitive Gy-sets.Further,
Since t is a permutation of X, it carries partitions into partitions. The result
now follows from Theorem 9.1.
Definition. If X is a transitive G-set, then the rank of X is the number of
G,-orbits of X.
Theorem 9.3 shows that the rank of X does not depend on the choice of ,
x E X. Of course, {x) is a G,-orbit of X, and so rank X is really describing the
behavior of G, on X - (x).
Theorem 9.4. If X is a transitive G-set and x E X, then rank X is tlze iizci~zbei*
of (Gx-G,)-double cosets in G.
Proof: Define f : (G,-orbits) 4 ((G,-G,)-double cosets) by f(G,y) = GxgG,,
where gx = y. Now f is well defined, for if hx = y, then gx = hx, g-'h E G,,
h = lg(g-'lz) E G,gG,, and GxgG, = G,hG,. Let us show that f' is injective.
If f(G,y) = G,gGx = G,hG, = f(G,z), where gx = y and hx = z, then there
are a, b E G, with g = ahb. hTow y = gx = ahbx = alzx = nz E G,z; it follows
250 9. Permutations and the Mathieu Groups Multiple Transitivity 25 1

that G,y = G,z. Finally, f is surjective: if g E G, then gx = y and f(Gxy) = It is easy to see that GXl, . . . , , t = ni
Gxi.In particular, if x, y E X, then X is
GxgGx. E J also a G,-set and a Gy-set,and

See Exercise 9.12 below for another characterization of rank X.

Observe that if X is transitive and 1x12 2, then rank X 2 2 (otherwise
G = GxGx, and every g E G fixes x, contradicting transitivity). Let us now
consider the minimal case when rank X = 2; the G,-orbits are {x) and
X - {x), and so Gx acts transitively on X - (x).
De6"arsition.Let X be a G-set of degree n and let k < n be a positive integer.
Then X is k-transitive if, for every pair of k-tuples having distinct entries
in X, say, (x,, .. ., x,) and (y,, . .., y,), there is g E G with gxi = yi for
i = 1, ..., k.
The proof of Theorem 9.3(i) generalizes: if X is a k-transitive G-set, then
stabilizers of lc distinct points are conjugate; that is, if (x,, . . . , x,) and
(y,, . .. , y,) are k-tuples of distinct points and tx, = yifor all i, then Gyl,...,,k -
tGxl,...,xkt-l'
There is another stabilizer not to be confused with Gx,,.,.,,t: if Y =
{x,, . .. , x,), define 6, = {g E G: g(Y) = Y). Thus, g E Gy if and only if g per-
mutes Y, whereas g E GX1,. , . [, if and only if g fixes Y pointwise. Hence,
G, , , . . . , , t IG,, with strict inclusion almost always.
The next result generalizes Theorem 9.2.

Of course, 1-transitivity is ordinary transitivity. If k > 1, then every
k-transitive G-set is (k - 1)-transitive. A k-transitive G-set X is called doubly
transitive (or multiply transitive) if k 2 2, triply transitive if k 2 3, . . . , and so
forth.
It is quite common to say that a group G is k-transitive if there exists a
k-transitive G-set. This usage applies to other adjectives as well.
The easiest example of a multiply transitive G-set is provided by X =
(1, ..., k) and G = S,; it is plain that X is a k-transitive Sk-set.
Theorem 9.7. If X is a k-transitive G-set of degree n, then
for every choice of k distinct elements x,, . . . , x,. If X is faithful, then
1 G,, ,...,Xk 1 is a divisor of (n - k) !.
P r o d If x, E X, then JGI = tz]GXlj,by Theorem 9.2. Since Gxl acts (k - 1)-
transitively on X - {x,), by Lemma 9.5, induction gives

Lemma 9.5. Let X be a G-set. If k 2 2, then X is k-transitive if and only if, for lGxl/ = (n - ')'*'(' - + l)lGxl,...,~kl,
each x E X, the G,-set X - (x) is (k - 1)-transitive. and this gives the result. When G acts faithfully, then the result follows from
G being imbedded in SX-IXl,.. .,,kJ. Dl
Proof: Assume that X is k-transitive, and let (x,, ..., x,-,) and (y,, . . . , y,-,)
be (k - 1)-tuples of distinct elements of X - {x). There is thus g E G with DeGnition. A k-transitive G-set X is sharply k-transitive if only the identity
gx = x (so that g E Gx) and gx, = yi for all i 2 1. fixes k distinct elements of X.
Conversely, let (x,, . . ., x,) and (y,, ..., y,) be k-tuples of distinct ele-
ments of X. By hypothesis, there is g E Gxk with g(x,, .. .,x,-,, xk) = Theorem 9.8. The following conditions are equivalent for a faithful Ic-transitive
(y,, . .., y,-,, x,), and there is h E G,, with h(y,, y,, ..., y,-,, x,) = G-set X of degree n.
(y,, y,, . . . ,Y,-~,y,). Therefore, Jzg E G carries (x,, ..., x,) to (y,, .. ., y,). BI
(i) X is sharply k-transitive.
Theorem 9.6. Every doubly transitive G-set has rank 2, and if x E X and g 4 G,, (ii) If (x,, . . ., x,) and (y,, . . . , y,) are k-tuples of distinct elements in X , then
then G = G, u GxgGx. there is a unique g E G with gxi = yi for all i.
(iii) ]GI = n(n - 1)...( TZ - k 1). +
B)I.ooJ Since G acts k-transitively on X for k 2 2, we have Gx acting (k - 1)- (iv) The stabilizer of any k elements in X is trivial.
transitively on X - {xi; hence G, acts transitively on X - {x). Therefore, as If k 2 2, then these conditions are equivalent to:
a Gx-set, X has two orbits: {x) and X - {x); thus rank X = 2. By Theorem
9.4, there are exactly two (Gx-Gx)-doublecosets in G. E l (v) For every x E X, the Gx-set X - {x) is sharply ( k - 1)-transitive.

Definition. If X is a G-set and x,, .. ., x,E X, then the stabilizer is the Pro@$ All verifications are routine and are left are exercises for the reader.
subgroup
G, = {g E G: gx, = xi for i = 1, ..., t). Theorem 9.9. For every n, the symnzetric group S,, acts sharply tz-transitively on
252 9. Permutations and the Mathieu Groups
X = {1, ..., n); for every n 2 3, the alternating group A, acts sharply (n - 2)-
transitively on X.
Proof. The first statement is obvious, for Sncontains every permutation of X.
We prove the second statement by induction on n 2 3. If n = 3, then A, =
((1 2 3)) acts sharply transitively on X = {l, 2, 3). If n z 3, then (A,),, the
stabilizer of i, where 1 Ii 2 n, is isomorphic to A,-,; by induction, it acts
sharply (n - 3)-transitively on X - (i). Theorem 9.8(v) now completes the
proof.
EXAMPLE 9.3. Let f (x) E Q [x] be a polynomial of degree n with distinct roots
X = (a,, . . . , a,) c C; let E = Q(a,, ..., a,) be the splitting field of f(x), and
let G = Gal(E/Q) be the Galois group off (x). By Lemma 5.2, X is a G-set; by
Theorem 5.7, X is transitive if and only iff (x) is irreducible over Q. Now f (x)
factors in Q(a,)[x]: f(x) = (x - a,)f,(x). Moreover, G, = Gal(E/Q(a,)) 5
Gal(E/Q) = G is the stabilizer of a,, and so G, acts on X - {a,); indeed, G,
is also the Galois group of f,(x). By Theorem 5.7, G, acts transitively if and
only if f,(x) is irreducible (over Q(a,)). Hence, G = Gal(E/Q) acts doubly
transitively on X if and only if both f (x) and f,(x) are irreducible (over Q and
Q(a,), respectively). Of course, this procedure can be iterated. The set X =
{a,, . .. , an) of all roots of f(x) is a k-transitive G-set if and only if the poly-
nomials f (x), f, (x), ...,f,-, (x) are all irreducible (over Q, Q (a,), ..., and
Q(a,, ..., a,-,), respectively).
It can be proved that there are no faithful k-transitive groups for k > 5
other than the symmetric and alternating groups. It follows that if the Galois
group G of a polynomial is 6-transitive, then G is either a symmetric group or
an alternating group. Indeed, if G is 4-transitive, then the only additional
possibilities for G are four of the Mathieu groups.
Sharp k-transitivity is interesting for small k.
Definition. A sharply 1-transitive G-set X is called regular.
A faithful G-set X is regular if and only if it is transitive and only the
identity has a fixed point. It is now easy to see that the left regular representa-
tion of a group G in the Cayley theorem makes G itself into a regular G-set.
Our discussion of sharply Ztransitive G-sets begins with a technical
definition.
DeG'awition. If X is a G-set, then the Pi'obenirss kernel N of G is the subset
N = { l) u {g E G: g has no fixed points).
The Frobenius kernel may not be a subgroup of G, for it may not be closed
under multiplication.
Multiple Transitivity 253
Lemma 9.110. If X is a faithful sharply 2-transitive G-set of degree n, then the
Frobenius kernel N of G has exactly n elenzents.
Roo$ By Theorem 9.8(iii), IGl = n(n - 1). For each x E X, the stabilizer G,
has order n - 1 (because I GI = n 1 G,l), so that I G,XI = rz - 2, where G
: =
G, - {l). If x # y, then G, n G, = G,,, = 1, by Theorem 9.8(iv). Hence,
{G: x E X) is a disjoint family, and IU,., G/: = n(n - 2). Since N is the
complement of this union, IN( = n(n - 1) - n(n - 2) = n.
The basic philosophy is that permutation representations of a group G can
yield important information about 6 . We now illustrate this by classifying all
those groups G having a G-set X as in Lemma 9.10 in the special case when
n = 1x1is odd (the classification when n is even is known, but it is more
difficult).
Theorem 9.811. Let X be a faithful sharply 2-transitiue G-set of odd degree n.
(i) Each G, contains a unique involution.
(ii) G, has a center of even order, and a Sylow 2-subgroup of G, is either
cyclic or generalized quaternion.
(iii) The Frobenius kernel N of G is a normal subgroup of G.
(iv) The degree n is a powei. of an odd prime p.
(v) N is an elementary abelian p-group.
(vi) G is a semidirect product of N by 6,.
ProoJ (i) Since ]GI = n(n - 1) is even, G contains an involution g. Now g,
being a permutation of the n points of X, has a cycle decomposition; indeed,
Exercise 1.16 shows that g is a product of disjoint transpositions z,. . . z,.
Since 1x1 is odd, g must fix some x E X ; that is, g E G,; because G acts
sharply, g can fix nothing else, so that m = $-(n- 1). If h is another involution
in G, then h E G, for some y E X (perhaps y = x) and h = a, . . . a,, a product
of disjoint transpositions. Note that g and Iz can have no factors in common:
otherwise, we may assume that z, = a, = (a b) (because disjoint transposi-
tions commute); then gh-l fixes a and b, hence is the identity. Therefore,
1 = gh-I and g = h.
Assume that G, has t involutions. Sharp 2-transitivity implies that {G:
x E Xj is a disjoint family, and so G contains nt such elements, each of
which involves :n = $(n - 1) transpositions. Collectively, there are thus tm =
$nt(n - 1) distinct transpositions occurring as factors of these involutions.
But there are only $n(n - 1) transpositions in S, s S,,, and so t = 1, as
desired.
(ii) Let g, h E Gx, and assume that g is an involution. Then hgh-' is an
involution, so that the uniqueness in (i) gives hglz-' = g; that is, g E Z(G,).
The second statement follows from Theorem 5.46.
(iii) If T is the set of all involutions in G, then we claim that T T c N (the
254 9. Permutations and the Mathieu Groups
Frobenius kernel). Otherwise, there exist g, h E T with gh # 1 and gh $ N ;
that is, gh fixes some y E X. If z = hy, then both g and h have the transposi-
tion (y z) as a factor: hy = z and hz = hhy = y (h2 = 1); gy = hy = z (for
ghy = y gives hy = g-'y = gy) and gz = ghy = y. By (i), g = h, giving the
contradiction gh = 1.
For fixed g E T, the functions T -,N, defined by h I-+ gh and by h I-,hg, are
both injective. By (i), I TI = n, for there is exactly one involution in each G$
as x varies over X; by Lemma 9.10, IN1 = n = I TI. It follows that both injec-
tions above must be surjections; that is, g T = N = Tg for all g E T.
The Frobenius kernel always contains 1 and it is closed under inverses and
conjugations by elements in G. Here N is closed under multiplication: choose
g E T and observe that
(iv) Choose an element h .s N of prime order p; since N is a subgroup of
odd order n, p is odd. We claim that CG(h)5 N. Iff E GE commutes with h,
then hfh-' = f. Iff E G, for some x E X, then
because hx f. x. Therefore, f 6 u,., G,, so that f E N. We conclude that
CG(h)5 N, and SO [G : CG(h)]= [G : N] [N : CG(h)]2 [G : N] = n - I. But
[G : CG(h)] is the number of conjugates of h, all of which lie in the normal
subgroup N of order n. It follows that N # consists precisely of all the conju-
gates of h; that is, N is a p-group (of exponent p), and so n = I NI is a power
of p.
(v) If g E G is an involution, then conjugation by g is an automorphism of
N satisfying the conditions in Exercise 1.50(ii). Therefore, N is an abelian
group of exponent p, hence it is a vector space over Z, and hence it is an
elementary abelian p-group.
(vi) We know that N a G, N n G, = 1, and NG, = G (by the product rule).
(One could also use the Schur-Zassenhaus lemma, for the orders of N and
G, are consecutive integers, hence are relatively prime.)
Here is the proper context for this theorem. Let X be a faithful transitive
G-set with each g E GE having at most one fixed point. If no such g has a fixed
point, then X is a regular G-set; if some g does have a fixed point, then G is
called a Frobenius group. (An equivalent description of a Frobenius group is
given in Exercise 9.9 below: a group G is a Frobenius group if and only if it
contains a proper subgroup H # 1, called a Frobenius complement, such that
H n gHg-' = 1 for all g $ H.) It is easy to see that if there exists a faithful
sharply 2-transitive G-set X of any (not necessarily odd) degree, then G is
a Frobenius group: set H = G,, and note that G, n gGxg-' = G, n G,, =
G,,,, = 1 for all g $ G,. In 1901, Frobenius proved that Frobenius kernels of
Frobenius groups are (normal) subgroups (in 1959, Thompson proved that
Frobenius kernels are always nilpotent), and that every Frobenius group G is
Multiple Transitivity 255
a semidirect product of its Frobenius kernel by G, (which is a Frobenius
complement) (see Isaacs (1976),p. 100). Every Sylow subgroup of a Frobenius
complement is either cyclic or generalized quaternion; such groups are usu-
ally solvable (see Theorem 7.53), but there do exist nonsolvable ones. The
group SL(2, 5), which is not solvable, by Corollary 8.15, is a Frobenius com-
plement. Let Y be a two-dimensional vector space over GF(29), and let
GL(2, 29) act on V by matrix multiplication on the elements of V regarded as
column vectors. There is a copy H of SL(2, 5) imbedded in GL(2,29), and the
group G = V x H is a Frobenius group with Frobenius complement H r
SE(2, 5) (see Passman (1968), p. 202).
9.1. If V is a vector space, then GL(V) acts faithfully on V and on V# = V - (0).
9.2. Let H be a proper subgroup of G.
(i) Show that the representation of G on the cosets of H (Theorem 3.14) makes
the set of cosets GJH into a transitive G-set of degree [G: HI. Show that
GIH need not be faithful.
(ii) Show that the representation of G on the conjugates of H (Theorem 3.17)
makes the family of all conjugates of H into a transitive G-set of degree
[G :NG(H)]which need not be faithful.
9.3. Let n 2 5 and 2 < t < n.
(i) Show that Sncannot act transitively on a set with t elements. Conclude that
every orbit of an &-set with more than two elements has at least rt elements.
(ii) Show that S, has no subgroups of index t.
9.4. If X is a G-set and H IG, then every G-orbit is a disjoint union of H-orbits.
9.5. (i) If G is a finite group with G* = G - (1) a transitive Aut(G)-set, then G is
elementary abelian.
(ii) Show that Q # is a transitive Aut(Q)-set.
9.6. If X is a transitive G-set and N a G, then X is an N-set and I NxI = 1 N,I for all
x,y E X.
9.7. Let G be a group with /GI < n. Prove that G is isomorphic to a transitive
subgroup of Sn if and only if G contains a subgroup H of index n such that ,
neither H nor any proper subgroup of H is normal in G. (Hint. For necessity, use
Theorem 3.12; for sufficiency, take H to be the stabilizer of any symbol.)
9.8. Let X be a G-set and let H _< G. If the H-orbits of X are (O,, .. . , 0,), then the
orbits of gHg-I are {go,, . . . ,gO,). Use this result to give a new proof of Theo-
rem 9.3(ii).
9.9. A finite group G acting on a set X is a Frobenius group if and only if there is a
subgroup H with 1 < H < G and H n gHg-' = 1 for all g $ H. (Hint. Take H to
be the stabilizer of a point.)
9.10. If G is abelian, then a faithful transitive G-set is regular. (Hint. If x, y E X, then
G, = G,.)
256 9. Permutations and the Mathieu Groups Primitive G-Sets 257

9.11. If X is a sharply k-transitive G-set, then X is not (k + 1)-transitive.(In particu-

lar, no regular G-set is doubly transitive.)
9.12. If X is a transitive G-set, prove that IGI rank X =
number of x E X fixed by g. (Hint.
gy = y ) in two different ways.)
x,,, ~ ( g )where
~ , F(g) is the
Count the set ((9, x, y) E G x X x X: gx = x,
Every partition (B,, . .. ,B,,) of a set X determines an equivalence rela-
tion r on X whose equivalence classes are the B,. In particular, if X is a G-set

Primitive G-Sets
If G is a group of order n, then Cayley's theorem gives a faithful permutation
representation of G of degree n; that is, G may be regarded as a subgroup of
S,. The coming discussion can often give representations of smaller degree.
Definition. If X is a G-set, then a block is a subset B of X such that, for each
g E G, either gB = B or gB n B = @ (of course, gB = (gx: x E B)).
Uninteresting examples of blocks are 0, X, and one-point subsets; any
other block is called nontrivial.
Consider the following example of a G-set of degree 6, where G E S,.
G = ( 1, (1 2 3) (a b c), (1 3 2) (a c b), (1 b) (2 a) (3 c),
(1 4 2 c)(3 b)Y (1 c)(2 b)(3 a))
-
and B is a block, then there is an equivalence relation on X given by x y
if there is some giB containing both x and y. This equivalence relation is
G-invariant: that is, if x = y, then gx gy for all g E G. Conversely, if F is a
-
G-invariant equivalence relation on a G-set X, then any one of its equivalence
classes is a bloclt. Thus, a G-set is primitive if and only if it admits no non-
trivial 6-invariant equivalence relations.
Theorem 9.12. Every doubly transitive G-set X is primitive.
Proof. If X has a nontrivial block Bythen there are elements x, y, z E X with
x, y E B and z Ff: B. Since X is doubly transitive, there is g E G with gx = x
and gy = z. Hence, x E B n gB and B # gB, a contradiction. BI
Theorem 9.13. Let X be a transitive G-set of degree n and let B be a nontrivial
bloclc of X.
(i) If g E G, then gB is a block.
(ii) There are elements g,, . . . , g,, of G such that
is a partition of X.
(iii) IBI divides n, and Y is a transitive G-set of degree m = n/lBI.

P r o d (i) If gB n hgB f @ for some Iz E G, then B n g-lhgB # @; since B is

a bloclt, B = c~-lhgBand gB = hgB.
(ii) Choose b E B and x, q! B. Since G acts transitively, there is g, E G with
g,b = x,. Now B # g l B implies B n g l B = @, because B is a block. If X =
B u jlB, we are done. Otherwise, choose x, Ff: B u glB, and choose g, E G
with g2b = x,. It is easy to sce that g, B is distinct from B and from g ,B, so
that g2B n (B u g,B) = @ because B and g, B are blocks. The proof is com-
pleted by iterating this procedure.
(iii) Since J B J= lgiBl for all i, n = 1x1= rnlBl and IYI = m = rz/lBI. To
see that Y is a transitive G-set, let giB, gjB E Y and choose x E B. Since
X is transitive, there is g E G with ggix = gjx; that is, @ # ggiB n gjB =
( g ~ ~ ~ g ; l ) ~ ~ B Since
n g ~ BgjB
. is a block, it follows that ggiB = gjB, as
&sired. E l
Figure 9.1
Defi'nnition. If B is a block of a transitive G-set X and if Y = (g, B, .. . , gmB) is
a partition of X (where g,, . . . , g, E G), then Y is called the i~pmpri~nitive
system2
It is easy to see that either "triangle" (1, 2, 3) or (a, b, c) is a block; moreover, generated by B.
(1, a), (2, b), and (3, c) are also blocks.
Corollary 9.14. A transitive G-set of prime degree is prilnitive.
Definition. A transitive G-set X is primitiue if it contains no nontrivial block;
otherwise, it is irpeprimitive. ProoJ This follows from Theorem 9.13(iii).
258 9. Permutations and the Mathieu Groups
Here is a characterization of primitive G-sets.
Theorem 9.15. Let X be a transitive G-set. Then X is primitive if and only if,
for each x E X , the stabilizer Gx is a nzaximal subgroup.
Proof. If G, is not maximal, there is a subgroup H with G, < H < G, and we
will show that Hx = (gx: g E H) is a nontrivial block; that is, X is imprimi-
tive. If g E G and Hx ngHx # @, then hx = gh'x for h, h' E H. Since h-lgh'
fixes x, we have h-lgh' E G, < H and so g E H; hence, gHx = Hx, and Hx is
a block. It remains to show that H x is nontrivial. Clearly Hx is nonempty.
Choose g E G with g 4 H. If Hx = X, then for every y E X, there is h E: H with
y = hx; in particular, gx = hx for some h E H. Therefore g-'h E G, < H and
g E H, a contradiction. Finally, if Hx is a singleton, then H 5 G,, contra-
dicting G, < H. Therefore, X is imprimitive.
Assume that every G, is a maximal subgroup, yet there exists a nontrivial
block B in X. Define a subgroup H of G:
H = (g E G: gB = B).
Choose x E B. If gx = x, then x E B n gB and so gB = B (because B is a
block); therefore, G, < H. Since B is nontrivial, there is y E B with y $. x.
Transitivity provides g E G with gx = y; hence y E B n gB and so gB = B.
Thus, g E H while g 4 G,; that is, G, < H. If H = G, then gB = B for all g E G,
and this contradicts X # B being a transitive G-set. Therefore G, < H < G,
contradicting the maximality of G,.
EXAMPLE 9.4. If X is a transitive G-set and 1 # H a G, then it need not be
true that X is a transitive H-set. For example, if V is a vector space, then
V# = V - (0) is a transitive GL(V)-set. However, if H is the center of GL(V)
(i.e., if H is the subgroup of all the scalar transformations), then V# is not a
transitive H-set.
Lemma 9.16. Let X be a G-set, and let x, y E X
(i) If H _< G, then Hx n Hy Z @ implies H x = Hy.
(ii) If N a G, then the subsets Hx are blocks of X.
Pro05 (i) It is easy to see that Hy = H x if and only if y E Hx. If Hy n H x #
a, then there are Iz, h' E H with hy = h'x. Hence y = h-lh'x E Hx and
Wy = Hx.
+
(ii) Assume gHx n H x @. Since H a G, gHx n Hx = Mgx n Hx. There
are h, h' E H with hgx = Iz'x, and so gx = Iz-lh'x E Hx. Therefore, gNx =
Hx. B
Theorem 8.17.
(i) If X is a faitltful primitive G-set of degree n 2 2, if N a G and if H # 1,
then X is a transitive H-set.
(ii) iz divides I H 1.
Simplicity Criteria 259
Proof. (i) The lemma shows that Hx is a block for every x E X. Since X
is primitive, either Hx = @ (plainly impossible), Hx = (x), or Hx = X. If
Mx = {x) for some x E X, then N _< G,. But if g E G, then normality of H
gives H = g ~ g - l5 gG,g-I = G,,. Since X is transitive, H 5 Oy,,
Gy = 1,
for X is faithful, and this is a contradiction. Therefore Hx = X and X is a
transitive H-set.
(ii) This follows from Theorem 9.2.
Using this theorem, we see that the GL(V)-set V# in Example 9.4 is transi-
tive but not primitive.
Corollary 9.18. Let X be a faithful prir~zitiveG-set of degree n. If G is solvnble,
then n = pmfor some prime divisor p of I GI; if G is nilpotent, then n is a prirne
divisor of GI.
PuooJ If 6; is solvable, a minimal normal subgroup H of G is elementary
abelian of order pk, by Theorem 5.24. The theorem now gives lz a divisor of
pk, and so n, too, is a power of p. If G is nilpotent, then G has a normal
subgroup H of prime order p (e.g., take H = (g), where g is an element of
order p in Z(G)). The theorem gives n a divisor of p; that is, n = p.
9.13. Let X be an imprimitive G-set and let B be a maximal nontrivial bloclc of X;
that is, B is not a proper subset of a nontrivial block. Show that the imprimitive
system Y generated by B is a primitive G-set. Give an example with X faithful
and Y not faithful.
9.14. (i) Let X be a transitive G-set, let x E X, and let A be a nonempty subset of X.
Show that the intersection of all gA containing x, where g E G, is a block.
(ii) Let X be a primitive G-set and let A be a nonempty proper subset of X. If x
and y are distinct elements of X, then there exists g E G with x E gA and
y 4 gA. (Hilzt. The block in part (i) must be {x).)
9.15. (i) Prove that if a group G has a faithful primitive G-set, then its Frattini
subgroup @(G)= 1.
(ii) Prove that a p-group G that is not elementary abelian has no faithful primi-
tive G-set. (Hint. Theorem 5.47.)
Simplicity Criteria
We now prepare the way for new proofs showing that the alternating groups
and the projective unimodular groups are almost always simple.
Definitio~a.If X is a G-set and H a G, then H is a regular norr~zalsrrbgvoisp if
X is a regular H-set.
260 9. Permutations and the Mathieu Groups
Xf H is a regular normal subgroup, then I HI = 1x1. Thus, all regular normal
subgroups have the same order.
Theorem 9.19. Let X be a faithful primitive G-set with G, a simple group. Then
either G is simple or every nontrivial normal subgroup H of G is a regular
rzormal subgroup.
ProoJ If H a G and H # 1, then Theorem 9.17(i) says that X is a transitive
H-set. We have H n G, a G, for every x E X, SO that simplicity of G, gives
either H n G, = 1 and X is regular or H n G, = G,; that is, G, 5 H for some
x E X. In the latter event, Theorem 9.15 gives G, a maximal subgroup of G,
so that either G, = H or H = G. The first case cannot occur because H acts
transitively, so that H = G and G is simple. ill
It is proved in Burnside (191I), p. 202, Theorem XIII, that if a group G has
a faithful doubly transitive G-set X whose degree is not a prime power, then
either G is simple or G has a simple normal subgroup. (This result may be
false when the degree is a prime power; S, is a counterexample.)
Here is the appropriate notion of homomorphism of G-sets.
Definition. If X and Y are G-sets, then a function f : X -+ Y is a G-map if
f (gx) = gf (x) for all x E X and g E G; iff is also a bijection, then f is called a
G-isomorphism. Two G-sets X and Y are isomorphic, denoted by X r Y, if
there is a G-isomorphism f : X -, Y.
By Theorem 9.1, every G-set X determines a homomorphism q : G -+ S,.
Usually there is no confusion in saying that X is a G-set and not displaying
q, but because we now wish to compare two G-sets, let us denote X more
precisely by (X, cp). The action of g E G on x E X is now denoted by cp,x
instead of by gx. The definition of G-map f : (X, q ) -+ (Y, $) now reads, for all
g E G and x E X, as
f (cp,(x)) = $&f(XI).
EXAMPLE 9.5. Let G be a group, and let A, p: G -+ SG be the left and right
regular representations of G (recall that A,: x H gx and p,: x H xg-' for all
x, g E G). We claim that (G, A) and (G, p) are isomorphic G-sets. Define
f : G 4 G by f(x) = x-l; clearly f is a bijection. Let us see that f is a G-map.
9.6. CIzinese Rernairzder Theorenz.
EXAMPLE
If S < G is any (not necessarily normal) subgroup, we denote the family of
all left cosets of S in G by G/S; it is a G-set with action g(xS) = (gx)S (as in
Theorem 3.12). If X and Y are G-sets, then their cartesian product X x Y
may be regarded as a G-set with diagonal action: g(x, y) = (gx, gy).
Simplicity Criteria 26 1
If G is a (finite) group and H, K 5 G are such that H K = G, then there is a
G-isomorphism f : G/(H n K) 2; (GIH) x (GIK), where the latter has diagonal
action. Define f by x(H n K) ++ (xH, XI<). It is straightforward to show
that f is a well defined injective G-map. Since HK = G, the product formula
lHKl IH n KI = IHI /I<\ gives IGIIIHI IKF(I= 1/IH n KI; multiplying both sides
by /GI gives [G : H I [G : K] = [G : H n K], and so f must be surjective as
well. Therefore, f is a G-isomorphism.
Theorem 9.20. Every transitive G-set X is isomorphic to the G-set GIG, of all
left cosets of G, on which G acts by left mmztltiplication.
Proof. Let X = {x,, . . . , x,,), let H = G,,, and, for each i, choose g, E G with
gixl = xi (which is possible because X is transitive). The routine argument
that f : X -+ GIH, given by f(xi) = giH, is a well defined bijection is left to the
reader (recall that n = 10(xl)I = [G : HI). To check that f is a G-map, note
that if g E G, then for all i there is j with gx, = xj, and so
On the other hand,
gf (xi) = ggiH.
But ggixl = gxi = xj = gjxl; hence g,rlggi E G,, = H, and so gjH = ggiH, as
desired. E l
(i) if H, K _< G, then the G-sets GIH and GIK (with G acting by lefl tlzultipli-
cation) are isomorphic if and only if H and K are conjugate in G.
(ii) Two transitive G-sets (X, q ) and (Y, $) are isomorphic if and only if stabi-
lizers of points in each are conjugate in G.
ProoJ (i)Assume that there is a G-isomorphism f : G/H 4 GIK. In particular,
there is g E G with f(H) = gK. If h E H, then
Therefore, g-lhg E K and g-'Ng 5 K. Now f (g-'H) = g-'f (N) = g-'gl< =
K gives f -l(K) = g-IN. The above argument, using f -' instead off, gives
the reverse inclusion gKg-' _< H.
Conversely, if g - l ~ g= K, define f : G/H 4 G/K by f(aH) = agK. It is rou-
tine to check that f is a well defined G-isomorphism.
(ii) Let H and K be stabilizers of points in (X, cp) and ( Y , I
)
), respectively.
By Theorem 9.20, (X, q ) r GIN and (x tk) r G/I<. The result now follows
from part (i). !I3
Corol1lary 9.22. If G is solvable, then every nzaxirnal subgroup has index a prime
power; if G is nilpotent, then every maximal subgroztp has prirlze index.
262 9. Permutations and the Mathieu Croups
Remark. The second statement was proved in Theorem 5.40.
Pros$ If H < G, then the stabilizer of the point (H) in the transitive
G-set G/H is the subgroup H. If H is a maximal subgroup of G, then G/H
is a primitive G-set, by Theorem 9.15, and so IGIHI = [G: H I is a prime
power, by Corollary 9.18. A similar argument gives the result when G is
nilpotent.
Lemma 9.23. Let X be a transitive G-set and let H be a regular normal sub-
group of G. Choose x E X and let G, act on H" by conjugation. Then the
G,-sets H # and X - (x) are isomorphic.
ProoJ Define f : H # -+ X - {x) by f(h) = hx (notice that hx f x because H
is regular). Iff (A) = f (k), then h-lk E H, = 1 (by regularity), and so f is injec-
tive. Now 1x1= /HI (regularity again), IH*I = IX - {x) I, and so f is surjec-
tive. It remains to show that f is a Gx-map. If g E G, and h E H", denote the
action of g on h by g * h = ghg-l. Therefore,
f(g * h) = f(ghg-') = ghg-'x = ghx,
because g-l E G,; on the other hand, g. f(h) = g(hx), and so f(g * h) =
g.f(h).
Lemma 9.24. Let k 2 2 and let X be a k-transitive G-set of degree n. If G has
a regular normal subgroup H, then 1c < 4. Moreover:
(i) if k >_ 2, then H is an elementary abelian p-group for some prime p and n
is a power of p;
(ii) if k 2 3, then either H r Z3 and n = 3 or H is an elementary abelian
2-group and n is a power of 2; and
(iii) fi k > 4, then H r V and n = 4.
Pros$ By Lemma 9.5, the G,-set X - (x) is (k - 1)-transitive for each fixed
x E X; by Lemma 9.23, H # is a (k - 1)-transitive Gx-set, where G, acts by
conjugation.
(i) Since k > 2, H # is a transitive G,-set. The stabilizer Gx acts by conjuga-
tion, which is an automorphism, so that all the elements of H # have the same
(necessarily prime) order p, and H is a group of exponent p. Now Z(H) a G,
because Z(H) is a nontrivial characteristic subgroup, so that I XI = IZ(H)I =
IHI, for Z(H) and H are regular normal subgroups of G'. Therefore, Z(H) =
H, H is elementary abelian, and ]XI is a power of p.
(ii) If h E H", then it is easy to see that {h, h-') is a block. If k 2 3, then
H * is a doubly transitive, hence primitive, G,-set, so that either {h, h-I ) =
H# or (h, h-') = (h). In the first case, lH[ = 3, H r Z,, and n = 3. In the
second case, h has order 2, and so the prime p in part (i) must be 2.
(iii) If Ic 2: 4, k - 1 2: 3 and IH#l 2 3; it follows that both H z Z, and
H r Z,are excluded. Therefore, H contains a copy of V; say, (1, h, k, hk).
Simplicity Criteria 263
Now (G,), acts doubly transitively, hence primitively, on H # - { h ) . It is easy
to see, however, that (k, hlc) is now a block, and so H * - {h) = {k, hk). We
conclude that AT = (1, h, lc, hk) r V and n = 4.
Finally, we cannot have k 2: 5 because lz I 4. BI
Of course, the case k = 4 does occur (G = S4 and H = V). Compare the
case k = 2 with Theorem 9.1 1.
Theorem 9.25. Let X be a faithful k-transitive G-set, where k r: 2, and asstime
that G, is simple for some x E X.
(i) If k 2: 4, then G is simple.
(ii) If k 2: 3 and 1x1is not a power of 2, tlzerz either G r S3 or G is simple.
(iii) If Ic 2 2 and 1x1is not a prime power, then G is simple.
ProoJ By Theorem 9.19, either G is simple or G has a regular normal sub-
group H. In the latter case, Lemma 9.24 gives k < 4; moreover, if k = 4, then
N E $7 and 1x1= 4. Now the only 4-transitive subgroup of S4 is S4 itself, but
the stabilizer of a point is the nonsimple group S,. Therefore, no such H
exists, and so G must be simple. The other two cases are also easy conse-
quences of the lemma (note that the stabilizer of a point of an S3-set is the
simple group S2 r Z2 so that S, is a genuine exception in part (ii)).
Here is another proof of the simplicity of the large alternating groups.
Theorem 9.24. A, is simple for all iz 2 5.
ProoJ The proof is by induction on n 2 5. If rz = 5, then the result is Lemma
3.8. By Theorem 9.9, A, acts (n - 2)-transitively on X = (1, 2, . . ., n); hence,
if n 2 6, then A, acts k-transitively, where k 2 4. The stabilizer (A,),, of 11 is
just (for it consists of all the even permutations of (1, . . . , n - I)), and
so it is simple, by induction. Therefore, A, is simple, by Theorem 9.25(i).
Here is another simplicity criterion. It shall be used later to give another
proof of the shplicity of the PSLs.
Theorem 9.27 (Iwasawisra, 19411). Let G = G' (such a group is called perfect) and
let X be a faithful primitive G-set. If there is x E X and an abelian normal
subgroup K a G, whose conjugates {gKg-l: g E G) generate G, then G is
simple.
Proox Let H # 1 be a normal subgroup of G. By Theorem 9.17, H acts
n
transitively on X. By hypothesis, each g E G has the form g = g i k i g ~ ' ,
where g, E G and k, E PC. Now G = HG,, by Exercise 4.9(i), so that gi = hisi
for each i, where hi E H and si E G,. Normality of K in G, now gives
g= nhisilzis;lh;l E HKH IHI<

264 9. Permutations and the Mathieu Groups
(because H lies in the subgroup HK), and so G = HK. Since K is abelian,
G/H = HK/H G K/(H n K) is abelian, and H 2 G' = G. Therefore, G is
simple.
9.16. If X is a G-set, let Aut(X) be the group of all G-isomorphisms of X with itself.
cp -
Prove that if X is a transitive G-set and x E X, then Aut(X) z N,(G,)/G,. (Hint.
If y, E Aut(X), there is g E G with gx = q(x); the desired isomorphism is
g-lGx.1
9.17. Let X be a transitive G-set, and let x, y E X. Prove that G, = G,,if and only if
there is y, E Aut(X) with q ( x ) = y.
Affine Geometry
All vector spaces in this section are assumed to be finite-dimensional.
Theorem 9.28. If V is an n-dimensional vector space over a field K, then
V# = V - {O) is a transitive GL(V)-set that is regular when n = 1. If n 2 2,
then V" is doubly transitive if and only if K = Z,.
PYOOJ GL(V) acts transitively on V", for every nonzero vector is part of a
basis and GL(V) acts transitively on the set of all ordered bases of V. If n = 1,
only the identity can fix a nonzero vector, and so V" = K x is regular.
Assume that n 2 2, and that {y, z) is a linearly independent subset. If
K # Z,, there exists A E K with d # 0, 1; if x E V', then (x, Ax) is a linearly
dependent set, and there is no g E G with gx = y and gAx = z. Therefore,
GL(V) does not act doubly transitively in this case. If K = Z,, then every
pair of distinct nonzero vectors is linearly independent, hence is part of a
basis, and double transitivity follows from GL(V) acting transitively on the
set of all ordered bases of V.
Definition. If V is a vector space and y E then translation by y is the func-
tion t,: V --+ V defined by
t,(x) = x +y
for all x E V. Let Tr(V) denote the group of all translations under composi-
tion (we may also write Tr(n, K) or Tr(n, q)).
Definition. If V is a vector space over K, then the a . e groalp, denoted by
Aff(V), is the group (under composition) of all functions a: V -+ V (called
affinities) for which there is y E V and g E GL(V) such that
Affine Geometry 265
for all x E V (Of course, a is the composite t,g.) We may also denote Aff(V)
by Aff(n, I<) or by Aff(n, q).
Since gtYg-' = t,,, we have Tr(V) a Aff(V); moreover, the map t, H y is an
isomorphism of the group Tr(V) with the additive abelian group V.
Theorem 9.29. A vector space V of dimension n over a field K is a dozrbly
transitive Aff(V)-space; it is sharply 2-transitive when tz = 1.
%"roo$ Since Tr(Y) acts transitively on V (as is easily seen), it suffices to show
that if x # 0 and if y # z are vectors, then there exists a E Aff(V) with a(x) = y
and a(0) = z. There exists g E GL(V) with g(x) = y - z, by Theorem 9.28, and
one checks that a E Aff(V), defined by a: U H g(v) + z, sends (x, 0) to (y, z). If
n = 1, then g is unique, hence a is unique, and so the action is sharp.
When n = 1 and K = GF(q), then / Aff(1, q)l = q(q - I), for Aff(1, q) acts
sharply 2-transitively on a set with q elements. Indeed, Aff(1, q) is the semi-
direct product K xo K ",where 0: K " -+ K is defined by 0: x~multiplication
by x. If q is a power of an odd prime, then K = GF(q) is a sharply 2-transitive
Aff(1, q)-set of odd degree, and so Aff(1, q) is described by Theorem 9.1 1.
Let us now consider linear subsets of a vector space V: lines, planes, etc.,
not necessarily passing through the origin.
DeFnnition. If S is an m-dimensional subspace of a vector space V , then a coset
+
W = S v, where v E V, is called an affine m-saabspace of I/. The dintension of
S -I- v is defined to be the dimension of the subspace S.
There are special names for certain affine m-subspaces: if rn = 0, 1, or 2,
they are calledpohts, affi~telines, and affi~eplanes, respectively; if dim V = n,
then affine (n - 1)-subspaces of V are called affine hyperplanes.
We are going to focus attention on the affine subspaces of a vector space
V-its "geometry (that is, we focus on the lattice of all affine subspaces)."
Here is a description of this aspect of V.
Definition. Let V be an n-dimensional vector space over a field K , let A be a
set, and, for 0 5 m I n, let Ym(A)be a family of subsets of A (called affine
m-subspaces). If a: V -+ A is a bijection such that a subset W of V is an afine
m-subspace ( W = S + v) if and only if a(W) E Yn,(A),then (A, Y*(A), a) is
called an afjne n-space over I< with associated uectov space V.
EXAMPLE9.7. If V is an n-dimensional vector space over a field K , define
gm(V)to be the family of all affine m-subspaces S + v in V (where S is a
sub-vector space) and define a: V -+ V to be the identity. Then (V, ,Y*(V), I,)
is an affine space, called the standard affie space of V over K.
EXAMPLE9.8. Let (V, Y*(V), 1,) be a standard affine space over K and let
266 9. Permutations and the Mathieu Groups
v E V. Then (V, Y*(V), t,) is also an affine space over K, where t, is translation
by u.
We have not given the most important examples of affine spaces: certain
subsets of projective spaces (to be defined in the next section). A projective
space X is obtained from a vector space V by adjoining a "hyperplane at
infinity"; the original V inside of X is called the "finiteyyportion of X; that is,
it is called aSfine.
Definition. If (A, Y,(A), a) and (B, 9*(B), /I) are affine spaces over K, then a
bijection f:A + B is an affirrte isomorphism if, for all m, a subset W of A lies
in Ym(A)if and only if f(W) lies in Yn,(B).One says that (A, &(A), a) and
(B, LY*(B), p) are isomorphic if there is an affine isomorphism between them.
If (A, g*(A), a) is an affine space over K, then the set of all affine automor-
phisms of A is a group under composition, denoted by
If (A, Y*(A), a) is an affine space over K, then a: V -+ A is an affine isomor-
phism from (V, 9*(V), 1,) to (A, 9*(A), a), and so every affine space is iso-
morphic to a standard affine space.
Theorem 9.30. If (A, .5?*(A), a) is an affine space associated to a vector space T/,
then
Aut(V, 9*(V), 1,) z Aut(A, Z..(A), a).
Moreover, two affine spaces (A, 9,(A), a) and (8, Y*(B), /I) over K are isornor-
phic if and only $ they have associated vector spaces of the same dimension.
Proof. It is easy to see that f Hafa-l is an isomorphism Aut(V, &(I/), 1,) -+
Aut(A, =%*(A),a).
If (A, Y*(A), a) is an affine space with associated vector space V, then
dim V is the largest m for which gm(A)is defined. Therefore, isomorphic
affine spaces have associated vector spaces of the same dimension.
Conversely, if dim V = dim U, then there is a nonsingular linear transfor-
mation g : V + U, and jga-': A -+ B is an affine isomorphism, for it is a
composite of such. H
As a consequence of Theorem 9.30, we abbreviate notation and write
If V is an n-dimensional vector space over a field K, we may write Aut(n, K),
and if K = GF(q), we may write Aut(n, q).
It is clear that Aff(V) 5 Aut(V), for affinities preserve alfine subspaces. Are
there any other affine automorphisms?
Affine Geometry 267
EXAMPLE 9.9. If K = C, the map f : Cn Cn (the vector space of all n-tuples
-+
of complex numbers), given by (z,, . .. , z,,) I-+ (Z,, .. . , 2,) (where denotes the
complex conjugate of z), is easily seen to be an affine automorphism. More
generally, if a is any automorphism of a field K, then a,: Kn Kn,defined by
-+
is an affine automorphism, where A" denotes the image of A under a.
Here is a coordinate-free description of a,.
Definition. Let V and V' be vector spaces over k.A function f:V -+ V' is a
semilinear transformation if there exists a E Aut (K) such that, for all x, y E V
and all A E I<,
f (x 4-Y)= f (x) + f (Y)
and
f(ax) = A ~ ( x ) .
A semilinear transformation f is nonsingular if it is a bijection.
All linear transformations are semilinear (take a = I,), as are all the func-
tions a, of Example 9.9, and it is easy to see that every nonsingular semilinear
transformation V -, V is an affine automorphism of the standard affine space.
We claim that each nonzero semilinear transformation f : V -+ U deter-
mines a unique automorphism a of K. Assume that there is an automorphism
z of I<for which f(Au) = A"f(v) = A'f(v). Since f is nonzero, there is a vector
vo E V with f(vo) # 0. For each A E K, /Zbf(uo)= ASf(vO);hence, A" = A' and
a = z.Iff and g are semilinear transformations V + V with field automor-
phisms a and z, respectively, then their composite fg is semilinear with field
automorphism az; iff is a nonsingular semilinear transformation with field
-'
automorphism a, then f is semilinear with field automorphism a-l.
Defilraitisrra. All nonsingular semilinear transformations on a vector space V
form a group under composition, denoted by TL(V). If V is an n-dimensional
vector space over K, then we may write TL(n, I<), and if I<= GF(q), we may
write TL(n, q).
Of course, TL(V) 5 Aut(V).
The remarkable fact is that we have essentially displayed all possible affine
automorphisms. Here are two lemmas needed to prove this.
Lemma 9.34. Let V be a vector space of dimension 2 2 over a field K.
+ +
(i) Two distinct lines C, = Kx, y, and t, = Icx, y, are either disjoint or
intersect in a unique point.
(ii) Two distinct points x,y E V lie in a unique line, namely, K(x - y) + y.
268 9. Permutations and the Mathieu Groups
(iii) ~f{x, y} is linearly independent, then
+ y) = (Kx + y) n (KY+ x).
{x
(iv) Two distinct hyperplanes H + x and J + y are disjoint if and only if H =
J. In particular, fi dim V = 2, distinct lines Kx + y and Kz + w are dis-
joint if and only fi Kx = Kz.
Proof. (i) If z E el n t , , then C, nC,=(Kx, + +
y1)n(Kx2 y,)=(Kx, nKx,)+
z, by Exercise 2.26. Now Kx, = Kx, cannot occur lest the lines C, and C,
be distinct cosets of the same subgroup Kx, and hence disjoint. Therefore
Kx, n Kx, = {Oj and 8, n C, = {z).
+
(ii) Clearly K(x - y) y is a line containing x and y; a second such line
intersects this one in at least two points, contradicting (i).
+
(iii) Independence of {x, y) implies that the lines Kx y and Ky x are + f(0) = 0; it suffices to prove that such an affine isomorphism is semilinear.
+ +
distinct. But x + y E (Kx y) n (Ky x), and (i) now gives the result.
(iv) If H = J , then distinct hyperplanes are distinct cosets of H , and hence
are disjoint. If H # J , then H + J = V (because H and J are maximal sub-
spaces). Hence y - x = h + j for some h E H and j E J. Therefore, h x = +
+
- j + y E (H + x) n (J y), and the hyperplanes intersect.
The last statement follows, for a hyperplane in a two-dimensional vector
space is a line. !&!I
Lemma 9.32. Let V and U be uector spaces over K, let f : V 4 U be an affne
isomorphism of standard afline spaces with f(0) = 0, and let W be a two-
dimensional subspace of V with basis {x, y). Then:
(i) f (Kx) = Kf (x);
(ii) { f (x), f (y)} is linearly independent;
(iii) f (W) = (f (x), f (y)), the subspace spanned by f(x) and f (y); and
(iv) f 1 W: W -+ f (W) is an affine isomorphism.
Proo/. (i) Kx is the unique line containing x and 0, and f(Kx) is the unique
line containing f (x) and f (0) = 0.
(ii) If (f (x), f(y)} is dependent, then f(x), f(y), and 0 are collinear; ap-
-'
plying the affine isomorphism f gives x, y, and 0 collinear, contradicting
the independence of {x, y).
(iii) For each ordered pair A, p E K, not both zero, denote the line contain-
ing Ax and py by /(A, p). Since W = U,,,e(A, p), it follows that f ( W ) =
U,, f(L(A, p)). Denote (f(x), f(y)) by W'. To see that f(W) = W', choose a
line /(A, p). By (i), there are a, /3 E K with f (Ax) = af (x) and f (py) = Pf (y).
Now f (Ax),f (py) E W' for all A and ,u, hence f ({(A, p ) ) c W', and so f (W) c
W'. For the reverse inclusion, consider the affine isomorphism f -'and the
subset { f(x), f(y)}, which is independent, by (ii). As above, we see that
f -'(wl) c CY, so that W' c f(W).
(iv) By (iii),f(W) is a two-dimensional vector space, and so its affine struc-
ture is the same as that of W, by Theorem 9.30.
Affine Geometry 269
Theorem 3.33. if V and U are isornorplzic vector spaces of dirllerisiorz 2 2 over
a field I<, then every afine isomorphistn f : V -+U has the forrn f = t,,g for
sorne zr E U and some rzonsingular semilinear transfor~nationg; that is, for all
XE &/,
f(x) = g(x) + u.
Remarks. 1. One must assume that dim V 2 2, for every bijection between
one-dimensional vector spaces is an affine isomorphism.
2. We will not use the full force of the hypothesis, namely, f(,S?,,(V))=
Zm(U)for all m I n; we shall assume only that f(Zl(V)) = Zl(U); that is, f
carries lines to lines.
Proo4; Composing f with the translation x +-+ x - f (0), we may assume that
/,
Since f is a bijection, it preserves intersections; in particular, if and L, are
lines, then f (8, n L,) = f (C, ) n f (L,).
If (x, y} is independent, then we claim that
+
By Lemma 9.32(iii), both f (Kx y) and f(Kx) are lines contained in
(f(x), f(y)); indeed, they are disjoint because Kx + y and I<x are disjoint.
Since f (Kx) = Kf (x), by Lemma 9.31(i), we may apply Lemma 9.31(iv) to
(f (x),f (Y)) to obtain
+
f(I<x y) = ITf (.x) z +
for some z. In particular, there are scalars a, P E K with f(y) = af(x) + z and
+
f(Ax + y) = /?f(x) z, where A G K and /? depends on A. Thus
Therefore f(Kx + y) c Kf(x) + f(y), and equality holds because both are
lines.
+
We now prove that f(x y) = f(x) + f(y) when (x, y} is independent. By
Lemma 9.31(iii), (Kx + y) n (Icy + x) = (x + y}. Since f preserves intersections,
the last equality because (f (x),f (y)} is independent (Lemma 9.32(ii)).There-
fore, f(x + y) = f(x) + f(y) if (x, y} is independent. It remains to show that
f(Ax + px) = f (Ax) + f (px), and we do this in two steps (as f (0) = 0, we may
assume that A # 0 and p # 0). Since dim V 2 2, we may choose w so that
 i
270 9. Permutations and the Mathieu Groups ; Affine Geometry 27 1

{w, X) is independent; it follows that ( x + w, -x) is also independent. Now Proox This is essentially contained in the next to last paragraph of the proof
of the theorem.
f (w) = f ((x + W)- X)
= f (x + w ) + f (- x) (independence of (x + w, - x)) Corollary 9.35. Let (A, Y : + ( A )a)
, and (8,S?..:(B),P) be afline spaces of dirnen-
sion 2 2 with associated vector spaces V and U, respectively. Iff: A -+ B is an
= f (x) + f(w) + f (-x) (independence of (x, w)) affine isomorphism, then
It follows that f ( -x) = -f (x). Consequently, if A + p = 0, then f (Ax + px) = f = ptzga-l,
+
0 = f (Ax) + f(px). Finally, if A + p # 0, then {Ax w, px - w), {Ax, w), and where g: V --+ U is a nonsingular semilinear transformation and z = P-'fa(0).
{px, - w] are independent sets (for A # 0 and p # O), and so

-
Proof. The function g': V -+ U, defined by g' = ,!?-'fa, is an affine isomor-
phism; if g is defined by g = t-,gl, then g: V U is an affine isomorphism
with g(0) = 0. By Theorem 9.33, g is a nonsingular semilinear transformation.
Therefore g = t-=g' = t-,P-lfa, and the result follows. El
We have proved that f is additive.
If x # 0 and A E I<, then f (Kx) = Kf(x) implies that there is ox@)E K with If V is a vector space over a field I<, we have written Aut(V) for the group
f(Ax) = ox(A)f (x). The function ox: K --+ K is a bijection (because f(Kx) = of all affine automorphisms of I/; If dim V 2 2, then Theorem 9.33 shows that
Kf (x)) with ox(l) = 1. Additivity off implies f E Aut(V) if and only iff (x) = g(x) + y, where g is a nonsingular semilinear
transformation and y E V. If dim V = 1, then we have already remarked that
every permutation of V is an affine automorphism: the group of all afine
automorphisms is the symmetric group on J!
Since f(x) # 0, we see that ox is additive.
Tl~eorem9.36. If V is an n-dir~zensionalvector space over a field K, therl TL(V)
Next, we show that cx does not depend on x. Choose w so that {x, w) is
is a sernidirect product of GL(V) by Aut(K). If K = GL(q) = GF(pr),then
independent. Now
f (Ax + Aw) = f (Ax) + f (Awl = ox(4f ( 4 + ~w(Alf(w).
On the other hand, Proof. We have already seen that each g E TL(V) determines a unique o E
Aut(K); this function TL(V) --+ Aut(l<) is a surjective homomorphism with
kernel GL(V). As in Example 9.9, choose a basis of V and, for each o E
Equating coefficients, Aut(K), consider the semilinear trarlsformation o,. It is easy to see that Q =
%(A) = %+,(A>= ow(4. {G,: 0 E Aut(K)) TL(V), Q z Aut(K), and Q is a complement of GL(V).
Lastly, if p E K ', then {px, w) is independent and we see, with px in place of Therefore, FL(V) E GL(V) x Aut(K).
- .

x, that opx(A)= %(A). When K = GF(q), then lrL(V)l = IAut(I<)(IGL(V)I.The result now fol-
It remains to show that o: I< --+ K is multiplicative (the subscript may now lows from he or em 8.4. El
be omitted). But
f (Apx) = ~ x ( k ~ ) f (x) When V is finite, IGL(V)Jis given by Theorem 8.5

and also Theorem 9.37. Let V be a vector space of dimension n over a field I<.
f ( t x ) = o,,(Alf (4.
(rx) = opx(A)cx(~)f
(i) Aut(V), the group of afline automorphisms of V , is a srnlidirect product of
Therefore, cx(Ap)= opx(A)ox(p);as o does not depend on the subscript, Tr(V) by TL(V).
= o(A)o(p),and so o E Aut(K). It follows that f is semilinear.
El
~(41) (ii) If I< = GF(q), then
IAut(V)l = qnlTL(V)I.
Corollary 9.34. Let V and W be vector spaces over a field K with dim(V) =
dim(W) 2 2. If g: V -+ W is an additive function for which g(v) = A,v for all (iii) V is a doubly transitive Aut(V)-set.
v E V , where A, E I<, then all the A, are equal and g is a scalar transforination. (iv) Aff(V) = Aut(V) if and only if Aut(K) = 1.
272 9. Permutations and the Mathieu Groups
Proof. (i) By Theorem 9.33, if f E Aut(V), then f (x) = g(x) + y for some g E
TL(V) and some y E V. Define ?r: Aut(V) -+ TL(V) by n(f ) = g. Now ker n =
Tr(V) and n fixes TL(V) pointwise, so that n is a retraction. Lemma 7.20 now
gives Aut(V) z Tr(V) >a TL(V).
(ii) This follows from (i) after recalling that Tr(V) r K
(iii) This follows from Theorem 9.29, for Aff(V) 5 Aut(V).
(iv) Aut(K) = 1 if and only if every semilinear transformation is linear. E9
There exist fields K for which Aut(K) = 1. For example, the prime fields Q
and Z,,for all primes p, and the real numbers R.
9.18. Choose a basis of an n-dimensional vector space V over a field K. To the affin-
ity a E Aff(V) with a(x) = g(x) + y (where g E GL(V) and y E V), assign the
where A is the matrix of g (relative to the chosen basis) and b is the column
vector whose entries are the coordinates of y. Show that this assignment gives
an injective homomorphism Aff(V) = Aff(n, K) -+GL(n + 1, K) whose image
consists of all matrices whose bottom row is [0 .- - 0 11.
9.19. Show that Aff(V) is a semidirect product of Tr(V) by GL(V) and that
IAff(n,dl = q nIGL(n, dl.
9.20. Let V and U be rz-dimensional vector spaces over a field K, where n 2 2, and let
(V, Y*(V), ly)and (U, 9*(U), 1,) be standard affine rz-spaces over K. Show that
every affine isomorphism f : V -+ U has the form f = t,,gtv,where u E U, v E V,
and g: V -+ U is a nonsingular semilinear transformation.
9.21. Show that the center of TL(V) consists of all the nonzero scalar transforma-
tions.
9.22. Choose a basis of a vector space V over I< and, for a E Aut(K),define a,: V -, V
as in Example 9.9. If g E GL(V), let [Aij] be the matrix of g relative to the
chosen basis. Show that cr*gc~,' is linear and has matrix [a(lij)]. Conclude that
det(a*g%') = a(det g), and hence that SL(V) a rL(V).
Projective Geometry
We now turn from affine geometry to projective geometry. The need for
projective geometry was felt by artists obliged to understand perspective in
order to paint replicas of three-dimensional scenes on two-dimensional can-
vas. If a viewer's eyes are regarded as a vertex, then the problem of drawing
in perspective amounts to analyzing conical projections from this vertex onto
Projective Geometry 273
planes (see Coxeter (1987), 81.2); this is the etymology of the adjective "projec-
tive." To a viewer, two parallel lines appear to meet at the horizon, and
projective space was invented to actually make this happen. "Points at infin-
ity" (which constitute a "horizon") are adjoined to ordinary space: for each
line in ordinary space, there is a new point serving as the common meeting
point of all lines parallel to it. It is more efficient for us, however, to reverse
this procedure; we begin with the larger space and then observe that ordinary
(affine) space exists inside of it.
Let V be a vector space over a field I<. Define an equivalence relation on
V# = V - (0) by x -- y if there exists A E K," with y = Ax; if x E V#, denote
its equivalence class by [XI. If x E V " , then [x] is the family of all the nonzero
points on the line through x and the origin.
Definitiow. If V is an (n + 1)-dimensional vector space over a field K, then
P(V) = {[x]: x E V") is called projective n-space; one says that P(V) has
projecthe dimension n.
If V = Kn+l,then we denote P(V) by Pn(K); if K = GF(q),we may denote
Pn(K) by Pn(q). If x = (x,, x,, . . ., x,) G Knfl, then we may denote [x] by
its homogeneous cooudi~zates[x, , x, , . ..,x,,]. Of course, homogeneous co-
ordinates are not coordinates at all; if R # 0, then [Axo, Axl, . . . ,Ax,] =
ExO,x l , . ..,xn]. However, it does make sense to say whether the ith homoge-
neous coordinate is 0, for xi = 0 implies Axi = 0 for all A E K ".
Affine spaces have no algebraic operations; one can neither add nor scalar
multiply. However, using its associated vector space, one can restore these
operation to (A, L?*(A), a) with the bijection a. Projective spaces P(V) have
also lost the algebraic operations of I/, but they cannot be restored; the
only vestiges are homogeneous coordinates and projective subspaces, defined
below.
Definition. If W is a subset of V, define
If W is an (m + 1)-dimensional sub-vector space of V, then [ W ] is called a
projective m-subspace; one says that [W] has projective dimension in.
Here, too, there are names for special subspaces: if m = 0, 1, or 2, projective
m-subspaces are called pvojectiue paints, p~o>ctive lines, or projective planes;
respectively; projective (n - 1)-subspaces of a projective n-space are called
projective hyperplanes.
The reason for lowering dimension in passing from V to P(V) should now
be apparent: a line in V (through the origin) becomes a projective point, for
all the nonzero points on the line are equivalent. A pIane in V (through the
origin) becomes a projective line, and so forth.
274 9. Permutations and the Mathieu Groups
Theorem 9.38. Let V be a vector space over a field K of dimension n 2 2.
(i) If x, y E V# = V- {0), then [x] # [y] if and only if {x, y) is linearly
independent.
(ii) Every pair of distinct points [x], [y] E P(V) lie on a unique projective line.
(iii) If H is a projective hyperplane and L is a projective line not contained in
H, then H n L is a projective point.
Proof. (i) The following are equivalent: [x] = [y]; y = Ax for some A E K x ;
(x, y) is linearly dependent.
(ii) Since [x] # [y], the subspace W = (x, y) of V is two-dimensional, and
so [W] is a projective line containing [x] and [y]. This line is unique, for
if [W'] is another such line, then W' is a two-dimensional subspace of V
containing the independent set {x, y), and so W' = W.
(iii) Write H = [U] and L = [W], where U and W are subspaces of V of
dimension n and 2, respectively. Now W $ U because L $ H. Therefore,
u + w=v,
and [U n W] has projective dimension 0; that is, [U n W] = [U] n [W] = points determine a line, 8' coincides with the line joining [x] and [y] origi-
H n L is a projective point.
Definitiom. Let V and V' be vector spaces. A colheation (or projective isomor-
phism) is a bijection 8: P(V) -+ P(V1)such that a subset S of P(V) is a projec-
tive m-subspace if and only if 8(S) is a projective m-subspace of P(V1). Two
projective spaces are isomorphic if there is a collineation between them.
Notation. If g: V -, V' is a nonsingular semilinear transformation, then the
function
P(g): P(V) + P(V1),
defined by g([x]) = [g(x)], is easily seen to be a collineation. If g is a linear
transformation, then P(g) is called a projectiuity.
We are going to see that if dim(V) 2 3, every collineation on P(V) has the
form P(g) for some g E TL(V).
TPneo~t-em9.39. Two projective spaces P(V) and P(V1) are isomorphic if and
only if dim V = dim V'.
P r o 6 Necessity is obvious: for all m, P(V) has a projective rn-subspace if and
only if P ( V 1 ) does. Conversely, if dim V = dim V', there is a (linear) iso-
morphism g: V -+ V', and hence there is a collineation (even a projectivity)
P(g): P(V) -+ P(V1).
Projective Geometry 275
In view of this theorem, every projective 1.2-spaceover K is isomorphic to
P(Kn+l)= P"(K), where Kn+' is the vector space of all ( n + 1)-tuples of ele-
ments of I<.
(i) For every n 2 0 and every prime power q,
+
In particular, every projective line in Pn(q)has q 1 points.
(ii) The number of projective lines in P2(q)is the same as the number of projec-
tive points, namely, q2 + q + 1.
Proof. (i) If V = Kn+', where K = CF(q), then I V#l = qnfl - 1. Since V# is
partitioned into equivalence classes [x] each of which has q - 1 elements, we
have
/Pfl(q)l= (qn+l- l)/(q - 1) = qn + qn-l + ... + q + 1.
(ii) We claim that there are exactly q + 1 lines passing through any point
[x]. Choose a line d not containing 1x1; for each of the q + 1 points [y] on
/, there is a line joining it to [x]. If d' is any line containing [x], however,
Theorem 9.38(iii) shows that there exists some point [y] E L" n L'. Since two
nally counted.
+
Choose a line t o ; for each of the q 1 points [x] on do, there are exactly
q lines (other than do) passing through [x]. We have displayed q(q 1) + 1+
distinct lines (the extra 1 counts the line do). Since every line t' meets t o , we
have counted all the lines in P2(q).
Here is the important example of an affine space.
Theorem 9.41. If [W] is a projective hyperplane in a projective n-space P(V)
and fi x E V - W, then A = P(V) - [W] can be given the structure of an afJi'tze
n-space (A, 9*(A), a) with associated vector space W, where a: W - , A is
defined by a(w) = [w + XI for all w E W.
ProoJ If 0 5 m 5 n, define
Ym(A)= {[U] n A: U is a subspace of V with dim(U) = n.2 + 1).
We show that a is a bijection by exhibiting its inverse. Since x 4 W, every
+
v E V has a unique expression of the form v = Ax w for w E W and A E K. If
v q! W, then A # 0 and there is an element v' equivalent to v with v' - x E W
(namely, u' = A-'v). This element v' is the unique such: if v" = pv, p # 1-',
and v" - x E W,then (v' - x) - (v" - x) = (A-' - p)v forces v E W, a contra-
diction. The function P: A -, Ty, defined by P([v]) = v' - x (where v' is the
unique scalar multiple of v for which v' - x is in W), is thus well defined, and
276 9. Permutations and the Mathieu Croups
/3 is easily seen to be aM1.It is now routine to show that (A, T m ( A )a)
, is an
aEfine n-space with associated vector space W,
We illustrate Theorem 9.41 with K = GF(2) and V = K2. In this case, the
complement of a projective line (which we now call a "line at infinity") is an
affine plane consisting of 4 points and 6 lines.
Figure 9.2
There are three pairs of parallel lines in A (e.g., 12 and 34), so that P2(2)
requires 3 points at infinity to force every pair of lines to meet. Since IP2(2)1=
2' + 2 + 1 = 7, the picture of P2(2)is
Figure 9.3
There are now 7 lines instead of 6 (the circle {a, b, c) is a line; "line" in this
pictorial representation of P2(2) has nothing to do with euclidean lines drawn
on a sheet of paper). We have adjoined one "infinite point" to each of the lines
in A so that any pair of extended lines now meet; {a, b, c) is the horizon.
Projective Geometry 277
Lemma 9.42. Let V and V' be vector spaces of dirnensiorz 2 2, let H be a
projective hyperplane in P(V), and let A be the affine space P(V) - H. If
f, h: P(V) -+ P(V') are collineations with f 1 A = hl A, then f = 11.
ProoJ Let & be an affine Iine in A, and choose two distinct points x and y on
&. Since P(V) is a projective space, there is a unique projective line &* con-
taining x and y (and /). Since f(x) = h(x) and f(y) = h(y), it follows that
f(/*) = h(/*). By Theorem 9.38(iii), H n /* is a projective point z. Now
f(z) = f(Hn L*) = f(H) n f(/*) = h ( H )n h(L*) = Iz(z), so that f and h also
agree on all the points of H as well. .
Theorem 9.43 (Fuwdamentae Theorem of Projective Geometry). If V and V'
are isomorphic vector spaces over a field I< of dinzension n + 1 2 3, then every
collineationf : P(V) -+ P(V') has the form f = P(g) for some norzsirzgular semi-
linear transforrnation g: V -+ V'.
Remarks. 1. One needs dim(V) 2 3, for then P(V) has projective dimension
2 2. Every bijection between projective lines is a collineation.
2. There is a stronger version of this theorem in which K is assumed only
to be a division ring. This version does not affect finite groups, however, for
a theorem of Wedderburn (1905) asserts that every finite division ring is a
field.
ProoJ Let W be an n-dimensional subspace of ;V, so that [W] is a projective
hyperplane in P(V). As in Theorem 9.41, A = P(V) - [W] is an affine
n-space with affine isomorphism a: W -> A given by a(w) = [w + x] for
any x E V - W. Now f([IW]) is a hyperplane in P(V') (since f is a collin-
eation), so there is an n-dimensional subspace W of V' such that A' =
P(V1)- f([W]) = [W'] is an affine 11-space with bijection P: W' 4 A' given
by P(w') = [w" + y] for y E V' - W'; as any such y will serve, choose y so
that [y] = f([x]). Since f is a collineation, its restriction f I A: A -+ A' is an
affine isomorphism. By Corollary 9.35, f 1 A = Pt,ga-l, where g: W -+ W' is a
nonsingular semilinear transformation and z = P-lfa(0). Now
If U' E Y', then P-'([v']) = v" - y, where v" is the unique scalar multiple of v'
for which v" - y E W'. If folloufsthat P-l([y]) = y - y = 0, and so t, = to =
I,, . Therefore,
f 1 A = Pga-'.
Now V = (x) O W. If o is the field automorphism determined by the
semilinear transformation g, define 0: V -+V' by
where A E K and w E It is routine to check that y" is a semilinear
278 9. Permutations and the Mathieu Groups
transformation with g"(x) = y and g"l W = g; also, g" is nonsingular because g
is nonsingular and y 4 W'. We claim that f = P(8). For all [v] E A,
f (Cvl) = Ps"~-~(Cvl)
= P(ij(Au - x)) (where Av - x E W )
= P(g")([ul).
Therefore, f 1 A = P(g")IA, and so f = P(8), by Lemma 9.42.
There are some interesting groups acting on projective space.
Naptation. If V is a vector space over K , denote the group of all nonzero
scalar transformations on V by Z(V). If dim(V) = n, we may denote Z(V) by
Z(n, K); if I< = GF(q), we may denote Z(V) by Z(n, q).
Theorem 9.44. If dim(V) 2 3, the group TL(V)/Z(V) is isomorphic to
the group C of all collineations of P(V) with itsel$, if dim(V) = 2, then
TL(V)/Z(V) is isomorphic to a subgroup of the symmetric group on P(V) (the
symmetric group is the full collineation group of P(V) in this case).
Pro05 Define a homomorphism n: TL(V) -+ C by n(g) = P(g). If g E ker n,
then P(g) = 1; that is, [g(v)] = [v] for all v E V; so there are scalars A, with
g(v) = A,v for all v. By Corollary 9.34, g is scalar, and so n induces an injec-
tion PTL(V) -+ C. If dim(V) 2 3, then the Fundamental Theorem 9.43 gives
n surjective, so that TL(V)/Z(V) E C. If dim(V) = 2, then C is the symmetric
group on P(V). B!l
Definition. If V is a vector space over a field K, then the quotient group
TL(V)/Z(V) is denoted by PTL(V) and its subgroup GE(V)/Z(V) is denoted
by PCL(V). As usual, one may replace V by (n, K) or by (n, q) when
appropriate.
Theorem 9.44 shows that PTL(V) is (isomorphic to) the collineation group
of the projective space P(V) (when dim(V) > 2), and it is easy to see that
PGL(V) is the subgroup of all projectivities. The projective unimodular
group PSL(V) = SL(V)/SZ(V) can be imbedded in PGL(V): after all,
Projective Geometry
9.10. PTL(2,4) z S , and PGL(2,4) z A , .
EXAMPLE
For all 12 r. 1, the group PFL(n + 1, q) is the group of all collineations of
Pn(q), so that PTL(n + 1, q) (and each of its subgroups) acts faithfully on
Pn(q). In particular, PTL(2, 4) acts faithfully on P1(4). Since IP1(4)1 = 5,
PGL(2, 4) < PTL(2, 4) < S,. By Theorems 9.36 and 8.5, IPTL(2,4)) = 120
and lPGL(2,4)1 = 60. The isomorphisms follow.
Theorem 9.45. For every vector space V , PSL(V) (and hence the larger groups
PGE and PTL) acts doubly transitively on P(1V.).
PrkaoJ If ([x], [y]) and ([x'], [y']) are ordered pairs of distinct elements
of P(V), then {x, y) and {x', y') are linearly independent, by Theorem 9.38.
Each of these independent sets can be extended to bases of V, say, {x, y,
z 3 , . .. , zn) and (x', y', z;, . .., 2 ; ) . There exists g E GL(V) with g(x) = x',
g(y) = y', and g(zi) = zf for all i 2 3. Hence P(g) [x] = [x'] and P(g) [y] =
[y']. If detfg) = A, define h E GL(V) by h(x) = A-lx', h(y) = y', and h(zi) = zf
for all i. Then det(h) = 1, so that h E SL(V), P(h)[x] = [A-'x'] = [x'], and
P(h)[y] = [y']. Therefore, PSL(V) acts doubly transitively on P(V).
We now give a second proof of the simplicity of the PSL's. Recall the
discussion preceding Lemma 8.19: for every transvection T E SL(V), where V
is a vector space over K , there is a linear functional q : V -+ K and a nonzero
vector h E kerf such that T = {cp, h), where
{q, h ) : v-v + q(v)h.
Theorem 9.46. PSL(n, K) is simple if (n, K) # (2, Z,) and (n, K) # (2, Z,).
ProoJ We use Iwasawayscriterion, Theorem 9.27. Let G = PSL(V), where V
is a vector space over I<. By Theorem 9.45, P(V) is a faithful doubly transi-
tive, hence primitive, G-set.
Choose h E ViS"and define a subgroup of the stabilizer GLhlby
H = {P({q, h ) ) :q(h) = 0) u {I).
Applying P to the formula {q, h ) {$, h) = {q + $, h) of Lemma 8.19(i),
we see that H is abelian. Recall Lemma 8.19(iv): If S E GL(V), then
S h yh)S-I = {qs-l, Sh); this is true, in particular, for S E SL(V). Now
P(S) E G[,, if and only if Slz = Ah for some d E K. But {$, Ah) = {A$, h ) , by
Lemma 8.19(ii), and this shows that H a GI,].
We now show that the conjugates of H generate G, and it suffices, by
Lemma 8.8(ii) (SL(V) is generated by the transvections), to show that every
P(($, k)) is a conjugate of some P({q, h)). Choose S E SL(V) with Slz = k.
Then, for any {q, h),
280 9. Permutations and the Mathieu Groups
As q varies over all linear functionals annihilating h, the linear functional
IS-' varies over all those annihilating k = Sh. In particular, given $, there is
y with q ( h ) = 0 and $ = qS-'.
It remains to prove that G is perfect: G = G'. Suppose some transvection T
is a commutator: T = [My N] for some M, N E SL(V). If T' is another trans-
vection, then T' and T are conjugate in GL(V), by Corollary 8.18. There-
fore, there is U E GL(V) with T' = T U= [My NIU = [MU,NU]. But both
M', Nu E SL(V), because SL(V) a GL(V), and so T', too, is a commutator.
It follows that SL(V) is perfect, and so its quotient PSL(V) is perfect as well.
Let n 2 3 and let (el, .. . , en) be a basis of K Define T E GL(V) by T(ei) =
e, for all i # 3 and T(e,) = e, - e, - e l .
Note that T = {q,h), where h = -e, - el and q~ is the linear functional
which selects the third coordinate: qJ(zRiei)= 1,. Define M = B,,(- 1) and
define N by Ne, = -e2, Ne, = el, and Ne, = ei for all i 2 3. Both M and N
lie in SL(V), and a routine calculation shows that [M, N] = MNM-'N-' =
T:
0 1 0 1 0 1 0 - 1
0 0 1 0 0 1 0
If n =2 and IK I > 3, then there exists R E K with A2 # 1. But
and so the (elementary) transvection B12(R2- 1) is a commutator. All the
conditions of Iwasawa's criterion have been verified, and we conclude that
PSL(tz, K) is simple with two exceptions.
Note that the proof of Corollary 8.14 does not work for nonperfect fields
of characteristic 2; the proof above has no such limitation.
Here is a chart of the groups that have arisen in this chapter.
Here is a summary of the various relations that have arisen among these
Sharply 3-Transitive Groups
groups, where V is a vector space over I<.
9.23. Show that if 11 2 2, then PSL(iz + 1, K) acts faithfully and transitively on the set
of all projective lines in Pn(I<).(Hint. Two points determine a unique line.)
9.24. Let f : P(V) -+ P(V) be a collineation, where dim(V) 2 3. If there exists a projec-
tive line L in P(V) for which f Id is a projectivity, then f is a projectivity.
9.25. Prove that PGL(2, 5) r S, and PSL(2, 5) s A , . (Hiizt. IP1(5)1= 6.)
9.26. Prove that any two simple groups of order 168 are isomorphic, and conclude
that PSL(2, 7) r PSL(3,2). (Hint. If G is a simple group of order 168, then a
Sylow 2-subgroup P of G has 7 conjugates and NG(P)/Pz Z,. Construct a
projective plane whose points are the conjugates of P and whose lines are those
subsets ( a ~ a - ' ,PPP-', yPy-l) for which {a, P, y) is a transversal of P in N,(P).)
Sharply 3-Transitive Groups
We have seen that the groups PTL(n, K) are interesting for all n 2 3: they
are collineation groups of projective (n - 1)-space. Let us now see that
PTL(2, K) and its subgroup PGL(2, K) are also interesting.
Definition. If K is a field, let k = K u (ca), where m is a new symbol. If
a E Aut(K) and ad - bc # 0, then a semilinear fiactiot~altratrsfortnatio~tis a
function f: I? -+ I? given by
f(A) = (anu + b)/(c;lu + d) for A E I?.
The "extreme" values are defined as follows: f(A) = co if cRu + d = 0; f(co) =
co if c = 0; f ( m ) = ac-I if c # 0. If o is the identity,'then f is called a litte~v
fr~ctionaltrcansforrnaPion.
These functions arise in complex variables; there, I< = C and o is con~plex
conjugation.
It is easy to see that all the semilinear fractional transformations on I?
form a group under composition.
282 9. Permutations and the Mathieu Groups
Notation. TLF(I<) denotes the group of all semilinear fractional transforma-
tions on k,and LF(K) denotes the subgroup of all the linear fractional
transformations.
Theorem 9.47. For every field K, PTL(2, K) g TLF(K) and PGL(2, K) r
LF(K).
Proof. Choose a basis of a two-dimensional vector space V over K. Using
Theorem 9.36, one sees that each f E TL(2, K) has a unique factorization
f = ga,, where g E GL(2, I<) and a E Aut(I<). If the matrix of g relative to the
chosen'basis is
define 0:TL(2, K) -+ TLF(K) by go, H (aRu + b)/(cRa + d). It is easy to see
that $ is a surjective homomorphism whose kernel consists of all the non-
zero scalar matrices. The second isomorphism is just the restriction of this
one.
We have seen that P1(K) is a PTL(2, K)-set and that I? is a TLF(K)-set.
There is an isomorphism $: PTL(2, K) -+ TLF(K) and there is an obvious
bijection 6: P1(K) -+ I?, namely, [A, 11r l if 1 E K and [I, 0] H m. If y E
PTL(2, K) and A E K, when is it reasonable to identify the action of y on
[A, 11 with the action of $(y) on 6([A, I])? More generally, assume that there
is a G-set X (with action a: G -+ S,), an H-set Y (with action p: H -+S,), and
a bijection 6: X -+ Y. As in Exercise 1.39,8 gives an isomorphism 0,: Sx -+ Sy
by n: H On:O-'. Finally, assume that there is an isomorphism $: G -+ H. There
are now two possible ways to view Y as a G-set: via 6,a or via p$.
Definition. With the notation above, the G-set X and the H-set Y are isornor-
phic if the diagram commutes; that is, if 6,a = P$.
EXAMPLE 9.11. The TLF(K)-set I? and the PTL(2, I<)-set P1(I<) are
isomorphic.
Let $: PTL(2, I<) 4 TLF(I<) be the isomorphism of Theorem 9.47, and let
9: P1(I<)-+ I? be the bijection given by [A, 11H A and [0, 11H m. Now each
y E PTL(2, I<) is the coset of some semilinear transformation g; relative to the
standard basis of K2, g = ho,, where a is the corresponding field automor-
Sharply 3-Transitive Groups
phism and
The action of y on [A, p] is essentially matrix multiplication: g[A, p] =
ha,[R, p] = lz[lu, pu] (we are regarding [A", pu] as a column vector).
To see that @,a, = P$,, it must be shown that 9,a, and P$, agree
+ +
on I<u (co). If A E K and c l u d 0, then B,a,(A) = 9a,6-1(A) =
6a,([A, 11) = Q(P(g)[A, 11) = Q(P(ha,)[A, 11) = B([aAu + +
bycRu d l ) =
+
9[(aRU + b)/(cAa d), 11 = (aAu + b)/(cT + d) = /?$,(A) ($, is this semilinear
fractional transformation, and P$:,(A) is its evaluation at A).
The reader may check that O,a,,(;l) = ,8$,(A) when c;lu + d = 0 and also
that 6,a,(co) = P$,(co).
Theorem 9.48. For every field I<, the projective line P1(I<) is a faithful sharply
3-transitive PGL(2, I()-set.
ProoJ AS in Example 9.11 (with the restriction PGL(2, K) -+ LF(I<) re-
placing the isomorphism PTL(2, K) -+ TLF(K)), it suffices to consider linear
fractional transformations acting on I?. The stabilizer of co is Aff(1, K) =
+
( l H a l b), the subgroup of LF(K) consisting of all numerators. By Theo-
rem 9.29, I<is a sharply Ztransitive Aff(1, I<)-set. It suffices, by Theorem 9.8,
to show that LF(K) acts transitively on W. But if p E K, then f ( l ) = l + p
sends 0 to p and f(A) = 1/R sends 0 to co. E4l
If K = GF(q), then Theorem 9.S(iii) gives another proof that 1 PGL(2, q)l =
(4 + l)q(q - 1).
Let us display a second family of sharply 3-transitive G-sets. The groups
occur as subgroups of the semilinear fractional transformations TLF(q). If
+ +
h = (aRO b)/(cAu d) E TLF(K), then ad - bc # 0. Multiplying numerator
and denominator by p E I<' does not change lz, but it does change the
"determinant" to p2(ad - bc). If q is a power of an odd prime p, then the
nonzero squares form a subgroup of index 2 in GF(q)", namely, (n2), where
n is a primitive element of GF(q) (if q is a power of 2, then every element of
GF(q) is a square). In this odd prime case, it thus makes sense to say that
det(lz) is or is not a square.
A second ingredient in the coming construction is an automorphism a of
GF(q) of order 2; a exists (uniquely) if and only if q = p2", in which case
Lo = AP".
Definition. Let p be an odd prime, let q = p2", and let a E Aut(GF(q)) have
order 2. Define M(q) 5 I"EF(q) = S u 17 where
S = {AH (a2 + b)/(cA + d)l ad - bc is a square)
and
T = {AH (aA" + b)/(cAu + d ) /ad - bc is not a square).
284 9. Permutations and the Mathieu Groups
It is easy to check that M(q) is a subgroup of TLF(K); indeed, S is a
subgroup of index 2 in M(q), and T is its other coset. Since it is a subgroup
of semilinear fractional transformations, M(q) acts faithfully on 2, where
K = GF(q).
Theorem 9.49. Let p be an odd prime, let q = p2", and let K = GF(q). Then l?
is a faitlfzd sharply 3-transitive M(q)-set.
ProoJ If G = M(q), then G = S u T implies G, = S, u T,, where
S, = (A -+i a 1 + b ( a is a square)
and
+
T, = (AHa/Zu bl a is not a square),
Let cc and P be distinct elements of K. If a - P is a square, define h E S,
by h(1) = (a - P)/Z+ P; if a - P is not a square, define h E T, by h(A) =
(a - P)AU+ P. In either case, h(1) = cc and h(0) = P, so that G, acts doubly
transitively on K. But, in each of S, and T, there are q choices for b and
*(q - 1) choices for a, so that I G, 1 = q(q - 1).By Theorem 9.8(iii), this action
is sharp. Finally, G acts transitively on f?, for A w - 1/A lies in G (the negative
sign gives determinant I, and 1 is always a square), and - 1/A interchanges 0
and co. The result now follows from Theorem 9.8(v). E l l
Zassenhaus (1936) proved that the actions of PGL(2, q) and of M(q) on the
projective line (if we identify GF(q) u (a)with the projective line) are the
only faithful sharply 3-transitive G-sets; the first family of groups is defined
for all prime powers q; the second is defined for all even powers of odd
primes. As each of PGL(2, q) and M(q) acts sharply 3-transitively on a set
+
with q + 1 elements, their common order is (q I)q(q - 1) when q = p2" and
p is odd. It is true that PGL(2, q) $ M(q) in this case; here is the smallest
instance of this fact.
Notation. The group M(9) is denoted by MI, (and it is often called the
Mathieu group of degree 10).
In the next section, we shall construct five more Mathieu groups: MI,,
MI,, M,,, M23, M2, (the subscripts indicate the degree of each group's usual
representation as a permutation group). However, the phrase "the Mathieu
groups" generally refers to these five groups and not to MI,.
Theorem 9.50. PGL(2,9) and MI, are nonisornorphic groups of order 720
actirzg sharply 3-transitively on I?, where K = GF(9).
ProoPf. We already know that each of these groups acts sharply 3-transitively
on k, a set with 10 elements, and so each group has order 10 x 9 x 8 = 720.
Let G = PGL(2,9). The double stabilizer Go,, = (Iz: Ai-taA/a # 0) r
Sharply 3-Transitive Groups 285
GF(9)" r Z,;indeed, a generator of Go,, is g: At-+ n/Z,where n is a primitive
element of GF(9). There is an involution t E G, namely, t: A H 1/A, such that
tgt = g-'. It follows that (t, Go,,) z D16 is a Sylow 2-subgroup of G.
Let H = MI,. The automorphism a of GF(9) in the definition of Mlo is
A HA3. The double stabilizer
Kl,o3 = S0,m LJ TO,@
= (h: A i-t a2/Z/a # 0) u (h: /Z H aA31 a nonsquare).
It is now easy to see that Ho,, is a nonabelian group of order 8 having a
--
unique involution; hence, Hot, Q, the quaternions. Since Dl, has no qua-
ternion subgroups, it follows that G and H have nonisomorphic Sylow 2-
subgroups, and so G $ H.
T.Y. Lam and D.B. Leep found that every subgroup of index 2 in Aut(S6),
a group of order 1440, is isomorphic to either S6, MLO,or PGL(2, 9), and
each of these does occur.
The polynomial p(x) = x2 + x - 1 is irreducible in Z3[x] (for it is a quadratic having
no roots in Z,). Now GF(9)contains a root n of p(x);indeed, GF(9) = Z3(n),so that
n is a primitive element of GF(9). In the following exercises, n denotes a primitive
element of GF(9)for which n2 + n = 1.
9.27. Prove that a Sylow 3-subgroup of M,,is elementary abelian of order 9.
9.28. Prove that (M,,), is a group of order 72 having a normal Sylow 3-subgroup.
Conclude that (M,,), is a semidirect product (Z,x 12,) xQ. (Hiizt. For every
b E GF(9)", A H A + b has order 3; the inverse of A H x2'A + b is A H
z6'A - n6'b, and the inverse of A H n2'+lA3+ b is A H n2i+5A- x2i+5
b .)
9.29. There are exactly 8 elements of (MI,), of order 3, and they are conjugate to one
another in (M,,),.
9.30. Prove that the subgroup S of M(q) is isomorphic to PSL(2, q). Conclude that
Mlo is neither simple nor solvable. (Hint.If h(A) = (aL + b)/(cL + d) and
ad - bc = ,u2,multiply numerator and denominator by p-'.)
9.31. Show that MI, is not a semidirect product of S by Z2. (Hint.T contains no
involutions.)
9.32. Regard GF(9) as a two-dimensional vector space over Z, with basis { l , n).
Verify the following coordinates for the elements of GF(9):
1 = (1,0)7 n4 = (-1, O),
286 9. Permutations and the Mathieu Groups Mathieu Croups 287

9.33. Show that MI, = (a,, a,, a,, Q, a,), where Recall Lemma 9.5: If X is a k-transitive G-set, then 3 is a (k + 1)-transitive
(?-set (should 2 exist).

Theorem 9.51. Let G be a doubly transitive pernzutatiotz group on a set X.

Suppose there is x E X, m $ X,g E G, and a permutation h of 2 = X u ( m )
9.34. Prove that MI, consists of even permutations of P1(9).(Hint. Write each of the such that:
generators oi, 1 Ii I 5, as a product of disjoint cycles.)
(9 9 E Gx;
9.35. Let a6 and a, be the permutations of GF(9) defined by a,(A) = n2A + nA3 and (ii) h(m) E X;
o,(;l) = A3.Regarding GF(9)as a vector space over Z, prove that a, and a, are (iii) h2 E G and (gh)3 E G; and
Iinear transformations. (iv) hGxh = Gx.
9.36. Prove that GL(2,3) r (a4, a,, a,, a,) (where a4 and a, are as in Exercise 9.33,
and a6 and a, are as in Exercise 9.35). (Hint. Using the coordinates in Exercise
Then c = (G, h) 5 S? is a transitive extension of G.
9.32, one has
Proof. Condition (ii) shows that acts transitively on 2.It suffices to prove,
as Theorem 9.4 predicts, that G = G u GhG, for then gw= G (because noth-
ing in GhG fixes co).
By Corollary 2.4, G u GhG is a group if it is closed under multiplication.
Now
(G u GhG)(G u GhG) c GG u GGhG u GhGG u GhGGhG
c G u GhG u GhGhG,

Mathieu Groups because GG = G. It must be shown that GhGhG c G u GhG, and this will
follow if we show that hGh c G u GhG.
We have already seen some doubly and triply transitive groups. In this sec- Since G acts doubly transitively on X, Theorem 9.4 gives G = Gx u GxgGx
tion, we construct the five simple Mathieu groups; one is 3-transitive, two ( f ~ gr $ G,). The hypothesis gives y, 6 E G with h2 = y and (~$2)~= 6. It fol-
are 4-transitive, and two are 5-transitive. In 1873, Jordan proved there are lows that hy-I = h-' = y-'h and hgh = g-'h-'g-'6. Let us now compute.
no sharply 6-transitive groups (other than the symmetric and alternating
groups). One consequence of the classification of all finite simple groups is
that no 6-transitive groups exist other than the symmetric and alternating
groups; indeed, all multiply transitive groups are now known (see the survey
= hGxh u (hGxh)h-'gh-l(hG,h)
article [P.J. Cameron, Finite permutation groups and finite simple groups,
Bull. London Math. Soc. 13 (1981), pp. 1-22]). = G, u ~ , h - ' ~ h - ' ~ , (condition (iv))
All G-sets in this section are faithful and, from now on, we shall call such
= Gx v Gx(y-'h)g(hy-l)G,
groups G permutation groups; that is, G _< S, for some set X. Indeed, we
finally succumb to the irresistible urge of applying to groups G those adjec- = G, u Gxy-' (g-'h-'g-16)y-1 G,
tives heretofore reserved for G-sets. For example, we will say "G is a doubly
c G u GIz-'6
transitive group of degree n" meaning that there is a (faithful) doubly transi-
tive G-set X having n elements. = G uG ~ - % G
We know that if X is a k-transitive G-set and if x E X , then X - {x) is a
= G u GhG.
(k - 1)-transitive Gx-set. Is the converse true? Is it possible to begin with a
+
k-transitive Gy-set X and construct a (k 1)-transitive G-set X v {y}?
One can say a bit about the cycle structure of lz. If lz(co) = a E X, then
Definition. Let G be a permutation group on X and let 2 = X u {m), where h2 E G = cwimplies /$(a)= h2(m) = m; hence, h = (m a)hl, where h' E Ga,,
oo 4 X . A transitive permutation group G on 8 is a transitive extension of G is disjoint from (m a). Similarly, one can see that gh has a 3-cycle in its
if^ Gand GW= G. factorization into disjoint cycles.
288 9. Permutations and the Mathieu Groups
The reader will better understand the choices in the coming constructions
once the relation between the Mathieu groups and Steiner systems is seen.
Theorem 9.52. There exists a sharply 4-transitive group M,, of degree 11 and
order 7920 = 11 -10- 9 . 8 = Z4. 3' - 5 el1 szrch that the stabilizer of a poirzt is
M10.
ProoJ By Theorem 9.49, M,, acts sharply 3-transitively on X = GF(9) u
{a). We construct a transitive extension of M,, acting on 2 = {X, w),
where w is a new symbol. If n is a primitive element of GF(9) with
+
n2 n = 1, define
and
h = ( a LU)(X
n2)(n3 n7)(n5 n6) = ( 0 CO)(T~,
where a,(A) = n2A + nA3 (use Exercise 9.32 to verify this).
The element g lies in M,,, for det(g) = - 1 = .n4, which is a square in
GF(9). It is clear that g $ (MI,), (for g(co) = O), h(w) = co E X,and h2 =
1 E G. Moreover, (gh)3 = 1 because gh = (w 0 m)(n n6 n3)(n2 n7 n5).
To satisfy the last condition of Theorem 9.51, observe that iff E (M,,),,
then
hfh(a) = hf(o) = h(o) = co,
so that Iz(M,,),lz = (M,,), if we can show that hflz E Mlo. Now (M,,), =
S, u T,, so that either f = n2'A + a or f = n2i+1A3+ a, where i 2 0 and
a E GF(9). In the first case (computing with the second form of h = (o a)(T6),
+ ~ ~ ~) " +~ E(~ 1" ~ respec-
The coefficients of A and l3are 7 - ~ ~ ' + ~7( 1 ~ and ),
tively. When i = 2j is even, the second coefficient is 0 and the first coefficient
is n4j'4, which is a square; hence, hfh E S, < MI, in this case. When
i = 2j + 1 is odd, the first coefficient is 0 and the second coefficient is
n4jn5, which is a nonsquare, so that hfh E T, c M,,. The second case
(f = n2'+'A3 + a) is similar; the reader may now calculate that
an expression which can be treated as the similar expression in the first case.
It follows from Theorem 9.8(v) that M,,, defined as (M,,, h), acts sharply
4-transitively on 2,and so I M l l 1 = 7920.
Note, for later use, that both g and h are even permutations, so that Exer-
,
cise 9.34 gives M, 5 A, ,.
This procedure can be repeated; again, the difficulty is discovering a good
permutation to adjoin.
Mathieu Groups 289
Theorem 9.53. There exists a sharply 5-transitive group M,, of degree 12
and order 95,040 = 12.11 . 10.9 - 8 = 2 6 . 33. 5 . 11 strclz that the stabilizer of a
point is MI,.
Pro05 By Theorem 9.52, M,, acts sharply 4-transitively on Y =
{GF(9), m, a). We construct a transitive extension of M,, acting on =
{Y, a), where R is a new symbol. If n is a primitive element of GF(9) with
n2 + 71: = 1, define
and
k = ( o Q)(n n3)(n2 n6)(7-c5 n7) = (w 52)d3 = (W Q)07
(note that this is the same 12 occurring in the construction of M,,).
Clearly k(52) = o E Y and h 4 (MI,), = M,,. Also, k2 = 1 and hk =
(w 92 m)(n n7 n6)(n2 n5 n3) has order 3. To satisfy the last condition of
Theorem 9.51, observe first that iff E (M,,), = M,, = S u T, then kfk also
fixes w. Finally, kfk E M,, : if f (A) = (aA + b)/(cA + d) E S, then kJc(A) =
(a3A + b3)/(c3A + d3) has determinant a3d3- b3c3 = (ad - b ~ )which ~ , is a
square because ad - bc is; a similar argument holds when f E 7: Thus,
kMlOk = MI,.
It follows from Theorem 9.8(v) that MI,, defined as (M,,, Ic), acts sharply
5-transitively on l?, and so IM,, 1 = 95,040. 89
Note that k is an even permutation, so that M,, < A,,.
The theorem of Jordan mentioned at the beginning of this section can now
be stated precisely: The only sharply 4-transitive groups are S,, S,, A,, and
MI,; the only sharply 5-transitive groups are S,, S,, A,, and M,,; if Ic 2 6,
then the only sharply k-transitive groups are S,', S,+, , and A,,,. We remind
the reader that Zassenhaus (1936) classified all sharply 3-transitive groups
(there are only PGL(2, q) and M(p2") for odd primes p). If p is a prime
and q = p", then Aut(1, q) is a solvable doubly transitive group of degree q.
Zassenhaus (1936) proved that every sharply 2-transitive group, with only
finitely many exceptions, can be imbedded in Aut(1, q) for some q; Huppert .
(1957) generalized this by proving that any faithful doubly transitive solvable
group can, with only finitely many more exceptions, be imbedded in Aut(1, q)
for some q. Thompson completed the classification of sharply 2-transitive
groups as certain Frobenius groups. The classification of all finite simple
groups can be used to give an explicit enumeration of all faithfill doubly
transitive groups. The classification of all sharply 1-transitive groups, that is,
of all regular groups, is, by Cayley's theorem, the classification of all finite
groups.
The "large" Mathieu groups are also constructed as a sequence of transi-
tive extensions, but now beginning with PSL(3,4) (which acts doubly transi-
290 9. Permutations and the Mathieu Groups

tively on p2(4)) instead of with MI,. Since Ip2(4)1= 42 + 4 + 1 = 21, one

II
Mathieu Groups 29 1

II
[A, p, v]. If v = 0, then the right side is [p2, A2, 01; since Aji # 0, by our initial
begins with a permutation group of degree 21. We describe elements of P2(4) choice of [A, p, v], we have [ji2, ,I2, 0) = [(;lP)y2, (A,u)/I2,01 = [A, p, 01. The
by their homogeneous coordinates. reader may show that (LJ~,)~also fixes co, [I, 0,0], and [0, 1,0], so that
( ~ h ,= ) ~1.
Lemma 9.54. Let P be a primitive element of GF(4,).The futzctions fi: P2(4)-, Finally, assume that k E G, I PSL(3,4), so that k is the coset (mod scalar
P2(4),for i = 1,2, 3, defined by matrices) of
fl CAY P, v l = [A2 + PV, P2, v21,
f2CAY PYv l = [A2, p2, Pv21,
P v1 = [A2, p2, v219
f3C4 Y (because Ic fixes [I, 0,0]). Now det(k) = 1 = ad - bc. The reader may now
are involutions which fix [I, 0, 01. Moreover, calculate that hllchl, mod scalars, is

f
ProoJ The proof is left as an exercise for the reader (with the reminder that
all 3-tuples are regarded as column vectors). A hint for the second statement 1
r
is that PSL(3,4) a PTL(3,4), PTL(3,4)/PSL(3,4) r S,, and, if the unique which fixes [I, 0,0] and whose determinant is a2d2 - b2cZ= (ad - b ~ =) 1.~
nontrivial automorphism of GF(4) is o: AH ,I2, then f3 = o, and Thus hlGxh, = G,, and Theorem 9.51 shows that M2, = (PSL(3,4), h,)
acts 3-transitively on 3 = PZ(4)u { a ) with (M,,), = PSL(3,4).
By Theorem 9.7, I Mz2(= 22 - 21 .20.1H 1, where H is the stabilizer in M2,
of three points. Since (M,,), = PSL(3,4), we may consider H as the
stabilizer in PSL(3,4) of two points, say, [I, 0,0] and [0, 1,0]. If A E SL(3,4)
I sends (1,0,O) to (a, 0,O) and (0, 1, 0) to (0, p, O), then A has the form
Theorem 9.55. There exists a 3-transitive group M2, of degree 22 and order
443,520 = 22.21 -20- 48 = Z7 32.5 . 7 - 11 such that the stabilizer of a point is
I
PSL(3,4).
II
1
Proof. We show that G = PSL(3,4) acting on X = P2(4) has a transitive where q = (ap)-l. There are 3 choices for each of a and P, and 4 choices for
extension. Let each of y and 6, so that there are 144 such matrices A. Dividing by SZ(3,4)
i
x = [I, 0,01, z (which has order 3), we see that IH( = 48. El
sCA, p, v1 = CP, 2, v1, i
1 Theorem 9.56. There exists a 4-trarzsitiue group M23 of degree 23 and order
i'
h, =( a Cl, 0,Ol)fl. i 10,200,960 = 23 22.21 .20 48 = 27 3" 5 7 11.23 such that the stabilizer

[:: :]
1x1 matrix form, of a point is MZ2.

g= 1 0 0 , ProoyC The proof is similar to that for M,,, and so we only provide the
necessary ingredients. Adjoin a new symbol o to Y2(4)v (a),
and let
so that det(g) = - 1 = 1 E GF(4) and g E PSL(3,4). It is plain that g does not
fix x = [I, 0,0] and, by the lemma, that h: = 1. The following computation g = (co [I, 0, 01)f, = the former 1z,,
shows that (gh,)3 = 1. If [A, p, v] # coy[I, 0,0], or [0, 1,0], then
h2 = ( o co>S2.
( s ~ , ) ~ [ Ap,, V] = [Av + p2(v3 + I), pv + A2(v3+ I), v2].
The reader may apply Theorem 9.51 to show that M23 = (M,,, Iz,) is a
If v # 0, then v3 = 1 and v3 + 1 = 0, SO that the right side is [Av, pv, v2] =
transitive extension of M2,. PI
 292 9. Permutations and the Mathieu Groups
Theorem 9.57. There exists a 5-transitive group M2, of degree 24 and order
244,823,040 = 24.23-22.21 -20.48 = 210-33.5.7.11~ 2 such 3 that thestabi-
lizer of a point is M2,.
ProoJ Adjoin a new symbol LI to P2(4)u ( a , o),and define
g = ( a a ) f 2 = the former h,,
The reader may check that Theorem 9.51 gives M2, = (M23, h,) a transitive
extension of MZ3. B!
Theorem 9.58 (Miller, 1904)). The Mathieu groups M,,, &,, and M2, are
simple groups.
PBl.004: Since M2, is 3-transitive of degree 22 (which is not a power of 2) and
since the stabilizer of a point is the simple group PSL(3,4), Theorem 9.25(ii)
gives simplicity of M,,. The group M2, is 4-transitive and the stabilizer of a
point is the simple group M,,, so that Theorem 9.25(i) gives simplicity of
M2,. Finally, M2, is 5-transitive and the stabilizer of a point is the simple
group M2,, so that Theorem 9.25(i) applies again to give simplicity of
Mz4. El
Theorem 9.59 (Cole, 1896; MilBer, 1899). The Mathieu groups Mll and M12
are sirnple.
ProoJ Theorem 9.25(i) will give simplicity of M12 once we prove that MI, is
simple. The simplicity of MI, cannot be proved in this way because the
stabilizer of a point is MI,, which is not a simple group.
Let H be a nontrivial normal subgroup of M,,. By Theorem 9.17, H is
transitive of degree 11, so that ]HI is divisible by 11. Let P be a Sylow
11-subgroup of H. Since (1 does not divide IM,, 1, P is also a Sylow 11-
,,
subgroup of M , and P is cyclic of order 11.
We claim that P # NH(P). Otherwise, P abelian implies P I C,(P) I
NH(P)and NH(P)/CH(P)= 1. Burnside's normal complement theorem (Theo-
rem 7.50) applies: P has a normal complement Q in H. Now / QI is not divisi-
ble by 11, so that Q char H; as H a MI,, Lemma 5.20(ii) gives Q a MI,. If
Q # 1, then Theorem 9.1 7 shows that 1 Q j is divisible by 11, a contradiction.
If Q = 1, then P = H. In this case, H is abelian, and Exercise 9.10 gives H a
regular normal subgroup, contradicting Lemma 9.24.
Let us compute b l I ( P ) . In Sll, there are 11!/11 = lo! 11-cycles, and
hence 9! cyclic subgroups of order 11 (each of which consists of 10 11-
cycles and the identity). Therefore [Sf,: Nsl(P)] = 9! and I Nsf(P)I = 110.
,
Now NAfl(P) = NsIl(P)n M, . We may assume that P = (a), where a =
Steiner Systems 293
(1 2 . . . 10 11); if z = (1 11)(2 10)(3 9)(4 8 ) ( 5 7), then r is an involution
with soz = o-I and z E Nsll(P).But i is an odd permutation, whereas M,,2
A , , , so that INMll(P)I= 11 or 55. Now P 2 &(P) 5 NMll(P),so that either
P = N'(P) or NH(P) = NMll(P).The first paragraph eliminated the first pos-
sibility, and so NH(P) = NMll(P) (and their common order is 55). The Frattini
argument now gives M,, = HNMll(P)= HNH(P)= H (for NH(P)5 H), and
so M,, is simple.
EXERCISES
9.37. Show that the 4-group V has no transitive extension. (Hirtt. If h E S , has order
5, then (V, h ) 2 A,.)
9.38. Let W = {g E M I 2 :g permutes {cc,o R)). Show that there is a homomor-
phism of W onto S3with kernel (M,,),,,,,. Conclude that 1 = 6 x 72.
9.39. Prove that Aut(2, 3), the group of all affine automorphisms of a two-dimen-
sional vector space over Z3, is isomorphic to the subgroup W of M I , in the
previous exercise. (Hint.Regard GF(9) as a vector space over Z3.)
9.40. Show that (PSL(3, 4), h,, h,) IMz4 is isomorphic to PTL(3,4). (Nirzt. Lemma
9.54.)
Steiner Systems
A Steiner system, defined below, is a set together with a family of subsets
which can be thought of as generalized lines; it can thus be viewed as a kind
of geometry, generalizing the notion of affine space, for example. If X is a set
with 1x1 = u, and if k Iv, then a it-srrbset of X is a subset B c X with I Bl = k.
Definition. Let 1 < t < k < v be integers. A Steiner syste~ttof type S(t, Ic, u ) is
an ordered pair (X, B), where X is a set with u elements, 97 is a family of
k-subsets of X, called bdociis, such that every t elements of X lie in a unique
block.
EXAMPLE 9.12. Let X be an affine plane over the field GF(q), and let iB be the
family of all affine lines in X. Then every line has q points and every two
points determine a unique line, so that (X, B) is a Steiner system of type
q, q2).
EXAMPLE 9.13. Let X = P2tq)and let 93' be the family of all projective lines in
X. Then every line has q + 1 points and every two points determine a unique
line, so that (X, B) is a Steiner system of type S(2, q + 1, q2 + q + 1)).
EXAMPLE 9.14. Let X be an m-dimensional vector space over Z2, where 171 2
3, and let 93 be the family of all planes (affine 2-subsets of X). Since three
294 9. Permutations and the Mathieu Groups Steiner Systems

distinct points cannot be collinear, it is easy to see that (X, B ) is a Steiner and
system of type S(3,4, 2").

One assumes strict inequalities 1 < t < k < v to eliminate uninteresting

cases. If t = 1, every point lies in a unique block, and so X is just a set
partitioned into k-subsets; if t = k, then every t-subset is a block; if k = v, then
there is only one block. In the first case, all "lines" (blocks) are parallel; in the
second case, there are too many blocks; in the third case, there are too few
blocks.
Given parameters 1 < t < k < v, it is an open problem whether there exists
a Steiner system of type S(t, k, v). For example, one defines a projective plane
of o v h n to be a Steiner system of type S(2, n + 1, n2 + n + 1). It is conjec-
tured that iz must be a prime power, but it is still unknown whether there
exists a projective plane of order 12. (There is a theorem of Bruck and Ryser
(1949) saying that if n r 1 or 2 mod 4 and n is not a sum of two squares, then
there is no projective plane of order n; note that n = 10 is the first integer
which neither satisfies this hypothesis nor is a prime power. In 1988, C. Lam
proved, using massive amounts of computer time, that there is no projective
plane of order 10.)
Definition. If (X, B ) is a Steiner system and x E X, then
Pros$ If Y is the family of all t-subsets of X, then I YI = "v choose t" =
+
v(v - 1)...(v - t l)/t!. Define W c Y x 99 to consist of all ((x,, . . ., x,), B)
with (x,, ... , x,) c B. Since every t-subset lies in a unique block,
# ((x,, .. . ,x,), ) = 1; since each block B is a lc-subset, # ( , B) = "k choose
t" = k(1c - I) - - (k - t -I- l)/t!. The counting principle now gives the desired
formula for 1B1.
The formula for r follows from that for 1 91 because r is the number of
blocks in the contraction (X',9') (where X' = X - (x)), which is a Steiner
system of type S(t - 1, k - 1, v - 1). It follows that r does not depend on the
choice of x. EY
Remarks. 1. The proof just given holds for all t 2 2 (of course, (XI, 97')
is not
a Steiner system when t = 2 since t - 1 = 1).
2. The same proof gives a formula for the number of blocks in a Steiner
system of type S(t, k, v) containing two points x and y. If (X', 2 ' ) is the
contraction (with X' = X - {x)), then the number r' of blocks in (X', B')
containing y is the same as the number of blocks in (X, B ) containing x and
y. Therefore,

Theorem 9.60. Let (X, B) be a Steiner system of type S(t, k , v), where t 2 3. If
x E X, define X' = X - (x) and B' = { B - (x}: B E star(x)}. Then (X',3') is
a Steiner system of type S(t - 1, k - 1, v - 1) (called the contraction of (X, 99) Similarly, the number r(P)of blocks in (X, $3') containing p points, where
at x). l_<pit,is
,(P) = (v - p)(v - p - l).,.(v - t + 1)
Pros$ The routine proof is left to the reader. (12 - p)(k - p - l ) . - . ( k- t + 1)'

A contraction of (X, B ) may depend on the point x.

3. That the numbers 1231 = r, r', . . . r(P),. . ., I.(') are integers is, of course, a
constraint on t, k, v.
Let Y and Z be finite sets, and let W c Y x Z. For each y E Y, define
#(y, )=I(z E 2: (y, Z)E W}/ and define # ( , z)=j(y E Y (y, z) E W)(. Clearly,
Definitiolm. If (X, 9J)and (Y,V ) are Steiner systems, then an isomorpltism is a
'
bijection f : X -+ Y such that B E B if and only iff (B) E 44;: If (X, 93)= (Y,%),
then f is called an automovphtsm.
We deduce a corsrsting principle: If # (y, ) = rn for all y E Y and if # ( , z) = n
for all z E 2, then For certain parameters t, k, and v, there is a unique, to isomorphism,
i7z1YI = nIZ/. Steiner system of type S(t, 12, v), but there may exist nonisomorphic Steiner
systems of the same type. For example, it is known that there are exactly four
Theorem 9.61. Let (X, B ) be a Steiner systenz of type S(t, k, u). Then the num- projective planes of order 9; that is, there are exactly four Steiner systems of
ber of blocks is type S(2, 10, 91).

'Ifheorem 9.62. All the automorphist~zsof n Steiner system (X, B)form a group
if r is the number of blocks containing a point x E X, then r is independent of x Aut(X, B ) r S,.
296 9. Permutations and the Mathieu Groups
Proof. The only point needing discussion is whether the inverse of an auto-
morphism h is itself an automorphism. But S, is finite, and so h-' = hm for
some m 2 1. The result follows, for it is obvious that the composite of auto-
rnorphisms is an automorphism. fl
Theorem 9.63. If (X, B ) is a Steiner system, then Aut(X, B) acts faithfully on
iJ3.
Proof. If cp E Aut(X, 93)and q(B) = B for all blocks 13, then it must be shown
that cp = 1,.
For x E X,let r = Istar(x)l, the number of blocks containing x. Since cp is an
automorphism, cp(star(x))= star(&)); since cp fixes every block, cp(star(x))=
star(x), so that star(x) = star(cp(x)).Thus, ~ ( xand
) x lie in exactly the same
blocks, and so the number r' of blocks containing (cp(x), x) is the same as the
+
number r of blocks containing x. If cp(x) x, however, r' = r gives k = v
(using the formulas in Theorem 9.61 and the remark thereafter), contradict-
ing k < v. Therefore, ~ ( x=) x for all x E X.
Corollary 9.64. If (X, 99)is a Steiner system and x E X, then 0, Es,,,, B = {x).
Proof. Let x, y E X. If star(x) = star(y), then the argument above gives the
contradiction r' = r. Therefore, if y # x, there is a block B with x E B and
4 B, so that y 4 nBEsta,,,B.
We are going to see that multiply transitive groups may determine Steiner
systems.
Notation. If X is a G-set and U 4 G is a subgroup, then
F ( U ) = (x E X: gx = x for all g E U).
Recall that if U 5 G and g E G, then the conjugate gUg-' may be denoted
by Ug.
Lemma 9.65. If X is a G-set and U < G is a subgroup, then
F ( U g )= g F ( U ) for all g E G.
PPOO$, The following statements are equivalent for x E X: x E F(Ug);
gug-'(x) = x for all u E U; ug-'(x) = gP1(x) for all u E U; g-'(x) E F(U);
x EgF(U).
Theorem 9.66. Let X be a faithful t-transitive G-set, where t 2 2, let H be the
stabilizer of t points x,, . . ., x, in X, and let U be a Sylow p-subgroup of H for
some prime p.
Steiner Systems 297
(i) NG(U)acts t-transitively on F ( U ) .
(ii) (hnniehael? 1931; Witt, 1938). If k = IF(U)I > t and U is n nontrivial
nornzal subgroup of H, then (X, B) is a Steiner system of type S(t, k, v),
where 1x1 = v and
98 = {gF(U): g E G} = {F(Ug):g E G).
Pro05 (i) Note that 9 ( U ) is a NG(U)-set: if g E NG(U), then U = Ug and
F ( U ) = 9 ( U g )= gP(U). NOW(xl, . . . x,) c F ( U ) because U < H, the sta-
bilizer of x,, . .., x,; hence 12 = IF(U)j 2 t. If y,, . . . , y, are distinct elements
of F ( U ) , then t-transitivity of G gives g E G with gy, = xi for all i. If u E U,
then gug-'xi = guy, = gyi = xi (because y, E F(U)); that is, Ug < H . By the
Sylow theorem, there exists h E H with Ug = u". Therefore 1 2 - l ~E~NG(U)and
(h-'g)yi = h-'xi = xi for all i.
(ii) The hypothesis gives 1 < t < k _< v. If k = v, then F ( U ) = X; but U #
1, contradicting G acting faithfully on X. It is also clear that k = IF(U)j =
Ig9(U)I for all g E G.
If y,, .. . , y, are distinct elements of X, then there is g E G with gxi = yi for
all i, and so {y,, . . ., y,} c gS"(U). It remains to show that g F ( U ) is the
unique block containing the yi. If {y,, . . . ,y,) c hF(U), then there are z,, . . . ,
z, E F ( U ) with yi = hzi for all i. By (i), there is a E NG(U)with zi = oxi for all
i, and so gxi = y, = haxi for all i. Hence g-'ha fixes all xi and g-'ha E H. Now
H s NG(U),because U u H, so that g-'ha E NG(U)and g-'h E NG(U).There-
fore, Ug = Uh and g F ( U ) = 9 ( U g )= F ( u " ) = hF(U), as desired.
Lemma 9.67. Let H < Ad2, be the stabilizer of the five points
(i) H is a group of order 48 having a normal elementary abelinn Sylow 2-
subgroup U of order 16.
(ii) F ( U ) = L' u ( a , o , 921, where LP is the projective line v = 0, and so
I9(U)l = 8.
(iii) Only the identity of M2, fixes more than 8 points.
Proof. (i) Consider the group fi of all matrices over GF(4) of the form
[a : 1
A=A 0 y /? ,
where 2, y # 0. There are 3 choices for each of A and y, and 4 choices for each
of a and p, so that = 3 x 48. Clearly R / Z ( ~4), has order 48, lies in
PSL(3,4) 1 M2,, and fixes the five listed points, so that H = @/2(3,4) (we
know that IHJ = 48 from Theorem 9.57). Define G 4 l? to be all those matri-
ces A above for which y = 1. Then U = 0/2(3,4) has order 16 and consists
of involutions; that is, U is elementary abelian. But 6 u l?,being the kernel
298 9. Permutations and the Mathieu Groups
of the map a + SL(3,4) given by
so that U a H.
(ii) Assume that [A, p, v] E F ( U ) . If h E U, then y = 1 and
<
for some E GF(4)'. If v = 0, then all projective points of the form [A, p, 01
(which form a projective line L having 4 + 1 = 5 points) are fixed by h. If
v # 0, then these equations have no solution, and so h fixes no other projec-
tive points. Therefore, every h E U fixes L, co,w, R, and nothing else, so that
B ( U ) = t' u (as, w, 0) and IF(U)I = 8.
(iii) By 5-transitivity of M2,, it suffices to show that h E H # can fix at most
3 projective points in addition to [I, 0,0] and [O, l,O]. Consider the equa-
tions for 5 E GF(4) :
If v = 0, then we may assume that A # 0 (for LO, 1,0] is already on the list of
five). Now 1, = li. + av = and p = yp + pv = 5p give y = 1; hence h E U
and 12 fixes exactly 8 elements, as we saw in (ii). If v # 0, then v = y-'v = t v
implies 5 = y-l; we may assume that y # 1 lest h E U . The equations can now
be solved uniquely for A and p (A = (y-' - 1)-'av and p = (y-' - y)-lflv), so
that h 4 U can fix only one projective point other than [I, 0,0] and [0, 1,0];
that is, such an h can fix at most 4 points.
Theorem 9.68. Neither MI, nor M2, has a transitive extension.
Proof. In order to show that MI, has no transitive extension, it suffices to
show that there is no sharply 6-transitive group G of degree 13. Now such a
group G would have order 13.12.11 . l o . 9.8. If g E G has order 5, then g is
a product of two 5-cycles and hence fixes 3 points (g cannot be a 5-cycle lest
it fix 8 > 6 points). Denote these fixed points by (a, by c), and let H = G,,,,,.
Now (g) is a Sylow 5-subgroup of H ((g) is even a Sylow 5-subgroup of G),
so that Theorem 9.66(i) gives N = NG((g)) acting 3-transitively on F ( ( g ) ) =
{a, byc); that is, there is a surjective homomorphism cp: N -+ S,. We claim
that C = C,((g)) $ ker cp. Otherwise, q~induces a surjective map 9,:N/C -+
S,. By Theorem 7.1, N/C I Aut((g)), which is abelian, so that N/C and
hence S, are abelian, a contradiction. Now C a N forces y(C) a y(N) = S3,
Steiner Systems 299
so that q(C) = A3 (we have just seen that q(C) # 1) and so 3 divides ICI.
There is thus an element h E C of order 3. Since g and h commute, the element
gh has order 15. Now gh cannot be a 15-cycle (G has degree 13), and so its
cycle structure is either (5, 5, 3), (5, 3, 3), or (5, 3). Hence (gh)5, being either a
3-cycle or a product of 2 disjoint 3-cycles, fixes more than 6 points. This
contradiction shows that no such G can exist.
A transitive extension G of M,, would have degree 25 and order 25.24.
23.22.21 .20.48. If g E G has order 11, then g is a product of 2 disjoint
11-cycles (it cannot be an 11-cycle lest it fix 14 > 8 points, contradicting
Lemma 9.67(iii)). Arguing as above, there is an element lz E G of order 3
commuting with g, and so gh has order 33. Since G has degree 25, glz is not a
33-cycle, and so its cycle structure is either of the form (11, 11, 3) or one
11-cycle and several 3-cycles. In either case, (gh)" has order 3 and fixes more
than 8 points, contradicting Lemma 9.67.
Theorem 9.69.
(i) Let X = P2(4)u (as, w, $2) be regarded as an M2,-set, let U be a Sylow
2-subgroup of H (the stabilizer of 5 points), and let 33' = (gB(U): g E
M2,). Then (X, g)is a Steiner.system of type S(5, 8, 24).
(ii) If g F ( U ) contains (coyw, R), then its renzaining 5 points form a projective
line. Conversely, for every projective line f', there is g E PSL(3,4) _< M2,
with g g ( U ) = L' u ( a , o , 0).
ProoJ (i) Lemma 9.67 verifies that the conditions stated in Theorem 9.66 do
hold.
(ii) The remark after Theorem 9.61 gives a formula for the number r."
of blocks containing 3 points; in particular, there are 21 blocks containing
(coy w, R). If t' c F ( U ) is the projective line v = 0, and if g E PSL(3,4) =
(M24),,,,Q, then g F ( U ) = g(8) u {as, 0 , a).But PSL(3,4) acts transitively
on the lines of P2(4)(Exercise 9.23) and P2(4)has exactly 21 lines (Theorem
9.40(ii)). It follows that the 21 blocks containing the 3 infinite points coyo , $2
are as described. El
The coming results relating Mathieu groups to Steiner systems are due to ' .
R.D. Carmichael and E. Witt.
Theorem 9.70. M2, z Aut(X, B), where (X, 9 ) is a Steiner system of type
S(5, 8, 24).
Remark. There is only one Steiner system with these parameters.
Pro@+Let (X, g)be the Steiner system of Theorem 9.69: X = P2(4)u
( m yw, 0) and B = (gB(U): g E M2,), where F ( U ) = C u { a , w, R) (here C
is the projective line v = 0).
300 9. Permutations and the Mathieu Groups
It is clear that every g E M2" is a permutation of X that carries blocks
to blocks, so that M2, < Aut(X, 93). For the reverse inclusion, let q E
Aut(X, B). Multiplying cp by an element of M2, if necessary, we may assume
that cp fixes {m, o, a ) and, hence, that q l P2(4): P2(4) 4 P2(4). By Theorem
9.69(ii), q carries projective lines to projective lines, and so q is a collineation
of P2(4). But M2, contains a copy of PTL(3,4), the collineation group of
P2(4), by Exercise 9.40. There is thus g E M2, with gl P2(4) = q 1 P2(4),and
cpg-' E Aut(X, B ) (because M2, 5 Aut(X, 9)). Now qg-I can permute only
m, o , R. Since every block has 8 elements qg-' must fix at least 5 elements;
as each block is determined by any 5 of its elements, qg-' must fix every
block, and so Theorem 9.63 shows that qg-' = 1; that is, q = g E M2,, as
desired.
We interrupt this discussion to prove a result mentioned in Chapter 8.
Proof. The Sylow 2-subgroup U in H, the stabilizer of 5 points in M2,, is
elementary abelian of order 16; thus, U is a 4-dimensional vector space
over Z2. Therefore, Aut(U) z GL(4,2) and, by Theorem 8.5, / Aut(U)/ =
(2" - 1)(2" - 2)(24 - 4)(2" - 8) = 8!/2.
Let N = NMZ4(U). By Theorem 9.66(ii), N acts 5-transitively (and faithfully)
on g ( U ) , a set with 8 elements. Therefore, I Nl = 8 - 7 - 6 . 5 - 4 s, where s 2
6 = IS,/. If we identify the symmetric group on F ( U ) with S8, then [S, :N] =
t 5 6 (where t = 61s). By Exercise 9.3(ii), S, has no subgroups of index t with
2 < t < 8. Therefore, t = 1 or t = 2; that is, N = S, or N = A,. Now there is
a homomorphism q : N -+ Aut(U) given by g Hy, = conjugation by g. Since
A, is simple, the only possibilities for im q are S,, A,, Z2, or 1. We cannot
have im q r S, (since JAut(U)]= 8!/2); we cannot have lim q l 1 2 (for H < N,
because U a H, and it is easy to find h E H of odd order and u E U with
Izzih-" # u). We conclude that N = A, and that y : N -+ Aut(U) r GL(4,2) is
an isomorphism.
Theorem 9.72. M2, z Aut(Xt, g'), where (X', B') is a Steiner system of type
S(4, 7,23).
Remark. There is only one Steiner system with these parameters.
Pvooj: Let X' = P2(4)u (m, o ) , let B' = B1(C)= C v (m, w ) , where L' is the
projective line v = 0, and let 9Y'= {g(B1):g E kfZ3). It is easy to see that
(X', -9')is the contraction at Q of the Steiner system (X, 9?)in Theorem 9.69,
so that it is a Steiner system of type S(4, 7,23).
It is clear that M2, 5 Aut(X', 93'). For the reverse inclusion, let q E
Aut(Xt, 2'), and regard q as a permutation of X with q(Q) = Zl. Multiplying
by an element of M2, if necessary, we may assume that q fixes m and a.
Steiner Systems 301
Since (X', B') is a contraction of (X, B), a block in 93" containing m and w
has the form C' u {a, w), where C' is a projective line. As in the proof of
Theorem 9.70, q/P2(4)preserves lines and hence is a collineation of P2(4).
Since M24 contains a copy of PTL(3,4), there is g G M2, with glP2(4)=
qlP2(4). Therefore, g and q can only disagree on the infinite points co,w,
and Q.
If B E star@) (i.e., if B is a block in 9
7 containing Zl), then q(B) and g(B) are
blocks; moreover, /q(B)n g(B)I 2 5, for blocks have 8 points, while (o and g
can disagree on at most 3 points. Since 5 points determine a block, however,
q(B) = g(B) for all B E star(0). By Corollary 9.64,
Hence g(f2) = !2 and g E (M24)f2= &,. The argument now ends as that in
Theorem 9.70: qg" E Aut(X1,2 ' ) since M2, 5 Aut(Xt,&?'), qg-I fixes B',
a n d q = g ~ M ~El~ .
Theorem 9.73. M2, is a subgroup of index 2 in Aut(XU,B"),
where ( X u ,B") is
a Steiner system of type S(3, 6, 22).
Xetlzark. There is only one Steiner system with these parameters.
ProoJ Let X" = X - (R, o ) , let b" = F ( U ) - (R, a ) , and let B" = (gb":
g E M12;. It is easy to see that ( X u ,93"') is doubly contracted from (X, B),so
that it is a Steiner system of type S(3, 6, 22).
Clearly M,, 5 Aut(XV,a").For the reverse inclusion, let q E Aut(XU,B")
be regarded as a perinutation of X which fixes Q and w. As in the proof of
Theorem 9.72, we may assume that q(m) = co and that (oIP2(4) is a
collineation. There is thus g E M2, with g /P2(4)= qlP2(4). Moreover, con- ,
sideration of star(o), as in the proof of Theorem 9.72, gives g(w) = w.
Therefore, qg-' is a permutation of X fixing P2(4)u {w). If yg-' fixes R,
then qg-' = I, and q = g E (M24)n,, = M2,. The other possibility is that
qlg-l = (m Q).
We claim that [Aut(XU,2"): Mi2] 5 2. If q,, y2 E Aut(XU,B") and q,,
q2 q! M,,, then we have just seen that qi = (cn R)gi for i = 1, 2, where
gi E M24. But g;'g2 = q;'q2 E (M24)f2,,= M2, (since both q i fix R and o);
there are thus at most two cosets of M,, in Aut(XU, . .
L$Y"'.
,
Recall the definitions of the elementsh, and h3 in M,,: h, = (w cn)f2 and
h, = (R w)f,, where f2, f3 act on P2(4) and fix co,o,and fi. Note that h,
fixes Q and h, fixes m. Define g = h3k2h , = (R a) f3 f, f3, and define
302 9. Permutations and the Mathieu Groups
cp: X" -+ X" to be the function with q(m) = m and qlP2(4) = f3 fZ By f3.
Lemma 9.54, qlp2(4) is a collineation; since 9 fixes m, it follows that
q E Aut(XU,93"). On the other hand, rp $ M,,, lest qg-' = (Q m ) E M24,
contradicting Lemma 9.67(iii). We have shown that M,, has index 2 in
Aut(X", .9?").
Corollary 9.74. M,, has an outer automorphism of order 2 and Aut(XU,B") z
M22 x z2.
h o o f . The automorphism 9 E Aut(XU,B")with q $ MZ2constructed at the
end of the proof of Theorem 9.73 has order 2, for both f2 and f3 are
involutions (Lemma 9.54), hence the conjugate f3f2 f3is also an involution. It
follows that Aut(XU,93") is a semidirect product M I , x Z2. NOW(P is an
automorphism of M,,: if a E M,,, then am= qaq-' E M,,. Were an inner
automorphism, there would be b E M,, with qaV-' = babe' for all a E M,,;
that is, 90-I would centralize M,,. But a routine calculation shows that rp
does not commute with h, = (m [I, 0, O])fl E M,,, and SO q is an outer
automorphism of M,,. Bl
The "small" Mathieu groups MI, and MI, are also intimately related to
Steiner systems, but we cannot use Theorem 9.66 because the action is now
sharp.
Lemma 9.75. Regard X = GF(9) u {m, w, Q) as an MI,-set. There is a
srrbgrozip Z 5 MI,, isomorphic to S,, having two orbits of size 6, say, Z and Z',
and which acts sharply 6-transitively on 2. Moreover,
Proof. Denote the 5-set {m, w, R, 1, - 1) by Y. For each permutation r of Y,
sharp 5-transitivity of MI, provides a unique r* E M,, with r*j Y = r. It is
easy to see that the function S , -+ M,,, given by r r r * , is an injective
homomorphism; we denote its image (isomorphic to S,) by Q.
Let us now compute the Q-orbits of X. One of them, of course, is Y. If r is
the 3-cycle (aw R), then r* E Q has order 3 and fixes 1 and - 1. Now r* is
a product of three disjoint 3-cycles (fewer than three would fix too many
points of X), so that the (r*)-orbits of the 7-set X - Y have sizes (3,3, 1).
Since the Q-orbits of X (and of X - Y) are disjoint unions of (r*)-orbits
(Exercise 9.4), the Q-orbits of X - Y have possible sizes (3, 3, I), (6, I), (3,4),
or 7. If Q has one orbit of size 7, then Q acts transitively on X - Y; this is
impossible, for 7 does not divide /QI = 120. Furthermore, Exercise 9.3(i) says
that Q has no orbits of size t, where 2 < t < 5. We conclude that X - Y has
two Q-orbits of sizes 6 and 1, respectively. There is thus a unique point in
X - Y, namely, the orbit of size 1, that is fixed by every element of Q. If G E S,
is the transposition (1 - I), then its correspondent a* E Q fixes m, w, Q and
Steiner Systems 303
interchanges 1 and - 1. But [: GF(9) -+ GF(9), defined by [: At---+-A, lies in
MI, (for - 1 is a square in GF(9)) and [I Y = o, so that [ = O-*.Since the only
other point fixed by [ is 0, the one-point Q-orbit of X - Y must be {O).
Define Z = Y u (0) = (myo,Q, 1, - 1,O). We saw, in Exercise 9.33,
that MI, 5 M,, contains o : ( 9 + ( 9 , where a : 1 w - 1 is
(0 m)(l - l)(n3 n)(n5 n7). Let us see that the subgroup Z = (Q, o,) EZ S,.
The set Z is both a Q-set and a (o,)-set, hence it is also a C-set. As C acts
transitively on Z and the stabilizer of 0 is Q (which acts sharply 5-transitively
on Z - (0) = Y), we have 2 acting sharply 6-transitively on the 6-point set
Z, and so Z z S,. Finally, the 6 points X - Z comprise the other C-orbit of
X (for we have already seen that X - Z is a Q-orbit).
If P E Q, then P(Y) = Y and P(0) = 0, so that P(Z) = 2.Since o,(Z) = 2, it
follows that o(Z) = Z for all u- E C. Conversely, suppose p E M,, and p(Z) =
Z. Since Z acts 6-transitively on 2, there is o E C with o Z = p ]2. But pa-'
fixes 6 points, hence is the identity, and ,u = a E C. B
Theorem 9.76. If X = GF(9) u {a, w, a ) is regarded as an MI,-set arzd L%? =
{gZ: g E MI,}, where Z = {m, w, Q, 1, - 1,O), then (X, B)is a Steiner system
of type S(5, 6, 12).
Proof. It is clear that every block gZ has 6 points. If x,, . . ., x, are any
five distinct points in X, then 5-transitivity of MI, provides g E MI, with
{x, , . . . , x, ) c gZ. It remains to prove uniqueness of a block containing five
given points, and it sufices to show that if Z and g Z have five points in
common, then Z = gZ. Now if Z = {z,, .. . , Z6), then gZ = {gzl,.. . , yz,),
where gz,, .. . , gz, E Z. By Lemma 9.75, there is O- E Z s MI, with O-zz,=
gz,, .. . , oz, = gz,. Note that 0.2= 2,for Z is a Corbit. On the other hand,
CJ and g agree on five points of X, so that sharp 5-transitivity of M,, gives
a = g. Therefore Z = O-Z= gZ.
If GF(9) is regarded as an affine plane over Z,,then the blocks of the Steiner
system constructed above can be examined from a geometric viewpoint.
Lemma 9.77. Let (X, 9) be the Steiner system corzstrircted fiorn MI, in Theo-
rem 9.76. A szrbset B of X containing T = {m, w, Q) is a block if arzd only if
B = T u C, where C is a line in GF(9) regarded as an affine plane over Z3.
Proof: Note that Z = TuC,, where C, = {l, - 1,0), and C, is the line con-
sisting of the scalar multiples of 1. By Exercises 9.38 and 9.39, MI, contains
a subgroup W z Aut(2,3) each of whose elements permutes 7: Hence, for
every g E W, g Z = T u gCo, and gd, is an affine line. But one may count
exactly 12 affine lines in the affine plane, so that there are 12 blocks of the
form T u C. On the other hand, the remark after Theorem 9.61 shows that
there exactly 12 blocks containing the 3-point set 7: I Y
304 9. Permutations and the Mathieu Groups
Theorem 3.78. M,, r Aut(X, B), where (X, 9 ) is a Steiner systern of type
S(5, 6, 12).
Renzark. There is only one Steiner system with these parameters.
DooJ Let (X, B ) be the Steiner system constructed in Theorem 9.76. Now
MI, 5 Aut(X, 98) because every g E M,, carries blocks to blocks. For the
reverse inclusion, let cp E Aut(X, B).Composing with an element of M,, if
necessary, we may assume that rp permutes T = ( m yw, Q) and cp permutes
GF(9). Regarding GF(9) as an affine plane over Z,,we see from Lemma 9.77
that qlCF(9) is an affine automorphism. By Exercise 9.39, there is g E M,,
which permutes T and with g JGF(9) = cp 1 GF(9). Now cpg-I E Aut(X, B), for
M,, 5 Aut(X, g),rpg-' permutes T, and cpg-I fixes the other 9 points of X.
We claim that cpg-' fixes every block B in B. This is clear if IB n TI = 0, 1, or
3. In the remaining case, say, B = { m yo , x,, ...,x,), then cpgW1(B)must
contain either m or o as well as the xi, so that ( Bn rpg-'(B)[ 2 5. Since
5 points determine a block, B = cpgV1(B),as claimed. Theorem 9.63 forces
cpg-' = 1, and so cp = g E M,,, as desired. E l
Theorem 9.79. M,, r Aut(X', 9'), where (X', 93')is a Steiner system of type
S(4, 5, 11).
Remark. There is only one Steiner system with these parameters.
Proof. Let ( X ' , B') be the contraction at R of the Steiner system (X, a)of
Theorem 9.76. It is clear that MI, 2 Aut(X1,98'). For the reverse inclusion,
regard q E Aut(X1,98') as a permutation of X with cp(Q) = a. Multiplying by
an element of M,, if necessary, we may assume that cp permutes (a, a).
By Lemma 9.77, a block B' E 98' containing co and o has the form B' =
{m, o ) u C, where C is a line in the affine plane over Z,.As in the proof of
Theorem 9.78, pJGF(9)is an affine isomorphism, so there is g E M,, with
gIGF(9) = (plGF(9). As in the proof of Theorem 9.72, an examination of
g(star(Q)) shows that g(f2) = R, so that g E (M,,), = &,. The argument
now finishes as that for Theorem 9.78: cpg-I E Aut(X1,98'); cpg-' fixes W';
cp = g e M l l . !%I
The subgroup structures of the Mathieu groups are interesting. There are
other simple groups imbedded in them: for example, M,, contains copies of
A,, PSL(2,9), and PSL(2, ll), while M2, contains copies of M,,, A,, and
PSL(2,23). The copy C of S, in M,, leads to another proof of the existence
of an outer automorphism of S,.
Theorem 9.80. S, lzas an outer autornorplzism of order 2.
Remark. See Corollary 7.13 for another proof.
Steiner Systems 305
ProoJ Recall from Lemma 9.75 that if X = { a , w,Q) u GF(9) and C ( 2 S,)
is the subgroup of M,, in Lemma 9.75, then X has two C-orbits, say, Z =
Y u (0) and Z' = Y' u (0'1, each of which has 6 points. If a E C has order 5,
then a is a product of two disjoint 5-cycles (only one 5-cycle fixes too many
points), hence it fixes, say, 0 and 0'. It follows that if U = (a), then each of Z
and Z' consists of two U-orbits, one of size 5 and one of size 1. Now H =
(M12)o,ol E Ml0, and U is a Sylow 5-subgroup of H. By Theorem 9.66, N =
NMIZ(U) acts 2-transitively on B ( U ) = (0, 0'), so there is a E N of order 2
which interchanges 0 and 0'.
Since a has order 2, a = z, .. . z,, where the zi are disjoint transpositions
and m I 6. But M,, is sharply 5-transitive, so that 4 _< m; also, M,, I A I 2 ,
so that m = 4 or rn = 6..
We cIaim that a interchanges the sets Z = Y u (0) and 2'= Y'u {Of).
Otherwise, there is y E Y with a(y) = z E X Now aaa = a ' for some i (because
a normalizes (a)). If ai(y) = u and a(z) = v, then u, v E Y because Y u (0) is a
X-orbit. But u = ai(y) = aaa(y) = a&) = a(v), and it is easy to see that y, z,
u, and v are all distinct. Therefore, the cycle decomposition of a involves
(0 Or), (y z), and (v 21). There is only one point remaining in Y, say a, and
there are two cases: either a(a) = a or %(a)E Y'. If a fixes a, then there is
y' E Y' moved by a, say, a(yl) = z' E Y'. Repeat the argument above: there
are points u', v' E Y' with transpositions (y' z') and (0' u') involved in the
cycle decomposition of a. If a' is the remaining point in Y', then the transpo-
sition (a a') must also occur in the factorization of a because a is not a
product of 5 disjoint transpositions. In either case, we have a E Y and a' E Y'
with a = (0 O')(y z)(v u)(a a')& where /? permutes Y' - (a'). But aaa(a) =
ai(a) E 2 ; on the other hand, if a(af)= b' E Y', say, then aaa(a) = aa(al) =
a(bl), so that a(bl) E Y. Since a' is the only element of Y' that a moves to Y,
b' = a' and a(al) = b' = a'; that is, a fixes a'. This is a contradiction, for a
fixes only 0 and 0'.
It is easy to see that a normalizes C. Recall that a E C if and only if a(Z) =
Z (and hence 42') = 2'). Now aaa(Z) = aa(Z1)= a(Z') = Z, so that aaa E
C. Therefore, y = y, (conjugation by a) is an automorphism of C.
Suppose there is fl E C with aa*a = Ba"fl-' for all a* E C; that is, P-la E
C = CMI2(C).If C = 1, then a = fl E ;S, and this contradiction would show
that y is an outer automorphism. If a* E C, then a* = aa', where CT permutes
Z and fixes Z' and a' permutes 2' and fixes Z. Schematically,
if p E M1,, then (as any element of S, ,),
In particular, if p E C (so that pa*,u-' = a*), then either p(Z) = Z and
p(Z') = 2'or p switches Z and 2'.In the first case, p E C, by Lemma 9.75,
and p E C n C = Z(C) = 1. In the second case, pap-' = a' (and pa'p-' = a),
so that a and a' have the same cycle structure for all a* = aa' E C. But there
306 9. Permutations and the Mathieu Groups

is o-* E C with rs a transposition. If such p exists, then IT* would be a product 1 CHAPTER 10
of two disjoint transpositions and hence would fix 8 points, contradicting
MI, being sharply 5-transitive. Abelian Groups
There is a similar argument, using an imbedding of MI, into M2,, which
exhibits an outer automorphism of MI,. There are several other proofs of the
existence of the outer automorphism of S,; for example, see Conway and
Sloane (1993).
The Steiner systems of types S(5,6, 12) and S(5, 8,24) arise in algebraic I
coding theory, being the key ingredients of (ternary and binary) Golay odes.
The Steiner system of type S(5,8,24) is also used to define the Leech lattice,
11
a configuration
- in iW24 arising in certain sphere-packing problems as well as i
in the construction of other simple sporadic groups.

1
J

! Commutativity is a strong hypothesis, so strong that all finite abelian groups

! are completely classified. In this chapter, we focus on finitely generated and,
!1 more generally, countable abelian groups.
sr"

Basics
A valuable viewpoint in studying an abelian group G is to consider it as an
extension of simpler groups. Of course, this reduces the study of G to a study
of the simpler groups and an extension problem.
In this chapter, we assume that allgroups are abelian and we again adopt
additive notation.

Defiwitioaa. A sequence of groups and homomorphisms

...-+A~B<C-%D-+..-
is an sequence if the image of each map is the kernel of the next map. A
short exact sequence is an exact sequence of the form

There is no need to label the arrow 0 - + A , for there is only one such
homomorphism, namely, O r 0; similarly, there is no need to label the only
possible homomorphism C -+ 0: it must be the constant map x r 0. In the
short exact sequence above, 0 = im(0 -+ A) = kerf says that f is an injec-
tion and A E im f ; also, im g = ker(C 0) = C says that g is a surjection.
-+

Finally, the first isomorphism theorem gives B/im f = B/ker g im g = C.

308 10. Abelian G r o u p s Basics 309

If A 5 B and f is the inclusion, then im f = A and B/A E C. Thus, B is infinite and infinitely many A, # 0, then the direct sum is a proper subgroup
an extension of A by C if and only if there is a short exact sequence of the direct product.
O-,A-+B-,C-,O.
Defi~ition.If x E G and n is a nonzero integer, then x is divisible by $1 in G if
Definiticb~.If G is a group, its torsion sakbgrolsp is there is g E G with ng = x.
tG = {x E G: nx = 0 for some nonzero integer n).
Were the operation in G written multiplicatively, then one would say that
Note that tG is a fully invariant subgroup of G. x has an nth root in G. Exercise 1.31 shows that an element of order In is
When G is not abelian, then tG may not be a subgroup. For example, divisible by every n with (n, m) = 1.
Exercise 2.17 shows that tG is not a subgroup when G = GL(2, Q).
Theorem 80.23. There exists a group G whose torsion subgroup is not a direct
Definition. A group G is torsion if tG = G; it is torsion-free if tG = 0. summand of G.

The term torsion is taken from Algebraic Topology; the homology groups
of a "twisted" manifold have elements of finite order.
Proof. Let P be the set of all primes, and let G = n,,,Z,.If q is a prime and
x = (x,) E G is divisible by q, then there is y = (y,) with qy, = x, for all p; it
follows that x, = 0. Therefore, if x is divisible by every prime, then x = 0.
Theorem 10.1. The quotient group G/tG is torsion-fiee, and so every group G is We claim that G/tG contains a nonzero element which is divisible by every
an extension of a torsion group by a torsion-free group. prime. If this were true, then G # tG @ H for some subgroup H, because
H z G/tG. If a, E Z, is a generator, then a = (a,) has infinite order: if nu = 0,
Proof. If n(g + tG) = 0 in G/tG for some n # 0, then ng E tG, and so there is then nap = 0 for all p, so that p divides n for all p and hence n = 0. Therefore
m # 0 with m(ng) = 0. Since mn # 0, g E tG, g + tG = 0 in G/tG, and G/tG is +
a $ tG, and its coset a tG is a nonzero element of GItG. If q is a prime, then
torsion-free. a, is divisible by q in Z, for all p # q, by Exercise 1.31; there is thus y, E Z,
with qy, = a, for all p # q. Define y, = 0 and define y = (y,). Now a - qy E
If an abelian group is a semidirect product, then it is a direct product or, in tG (for its coordinates are all 0 except for a, in position q). Hence
additive terminology, it is a direct sum. The first question is whether the
extension problem above is only virtual or if there exists a group G whose q(y + tG) = qy + tG = a - (a - qy) + tG =n + tG,
torsion subgroup is not a direct summand of G (i.e., there is no subgroup
AI G with G = tG @ A). Let us first generalize one of the constructions we
and so a + tG is divisible by every prime q. El
have already studied.
We restate Lemma 7.20 for abelian groups.
Definition. Let K be a possibly infinite set and let {A,: k E K ) be a family of
groups1 indexed by K. The direct produd (or complete direct sum or strong Lemma 10.3. If G is an abelian group and A I G, then the followitzg statements
direct sum), denoted by n,,,
A,, is the group whose elements are all "vec-
tors" (a,) in the cartesian product of the A, and whose operation is
are equivalent.
(i) A is a direct summand of G (there is a subgroup B I G with A n B = 0 and
A + B = G).
The direct sum (or weak direct sum), denoted by
n,, zk,,
A,, is the subgroup of
A, consisting of all those elements (a,) for which there are only finitely
(ii) There is a subgroup B 5 G so that each g E G has a unique expression
+
g = a b with a E A and b E B.
(iii) There exists a homomorphism s: GIA -t G with vs = I,,,, where v: G -,
many k with a, # 0. G/A is the natural map.
If the index set I< is finite, then n,,, xkE,
A, = A,; if the index set I< is
(iv) There exists a retraction n: G -, A; that is, n is n homomorphism with
n(a) = a for all a E A.
These constructions make sense for nonabelian groups as well. They have already arisen, for
example, in our remark in Chapter 7 that different wreath products K 2 Q (complete and re- The following criterion is a generalization of Exercise 2.75 (which charac-
stricted) arise when Q acts on an infinite set !2. terizes finite direct sums).
3 10 10. Abelian Groups Basics 311

Lemma 10.4. Let (A,: k E K] be a family of subgroups of a group G. The Theorem 10.7 (Primary Decomposition). Every torsion group G is a direct sum
following statements are equivalent. of p-primary groups.
(i) GzCkEXAk. 1
(ii) Every g E G has a unique expression of the form I Pro05 For a prime p, define

I G, = {x E G: pnx = 0 for some tz >. 0)

where a,
(iii) G = (U,,,
E A,, the k are distinct, and a, # 0 for only finitely many k.
A,) and, for each j E K, A j n <UkzjAk)= 0.
1
1
L
(Gpis called thep-primary component of G.) The proof of Theorem 6.1, nzutatis
murandis, show that G z Z,
b,. O

Theorem 10.8. If G and H are torsion groups, then G 2 H fi and only if

Pro05 Routine. El G, z H, for all primes p.

Theorem 10.5. If V is a vector space over a field I<, then, as an additive group,
V is a direct sum of copies of K.
HPE.oo$ Let X be a basis of V. For each x E X, the one-dimensional subspace
Kx spanned by x, is, as a group, isomorphic to K. The reader may check,
using Lemma 10.4, that V = C,,,KX.
There is a notion of independence for abelian groups,
! Because of these last two results, most questions about torsion groups can
Defil~lliticen.A finite subset X = (x,, . . . , x,) of nonzero elements of a group G
is independent if, for all mi E Z, x m i x i = 0 implies mixi = 0 for each i. An
infinite set X of nonzero elements in G is indepenbnt if every finite subset is
independent.
ProoJ If q : G 4 H is a homomorphism, then y(G,) 5 H, for all primes p. In
particular, if ip is an isomorphism, then ip(G,) 5 H, and q-'(H,) G, for all
p. It follows easily that q 1 G, is an isomorphism G, 4 H,.
Conversely, assume that there are isomorphisms q,: Gp -+H, for all primes
p. By Theorem 10.4(ii), each g E G has a unique expression of the form g =
C,a,, where only a finite number of the a, # 0. Then q : G 4 H , defined by
y ( x a,) = C qp(ap),is easily seen to be an isomorphism. B
be reduced to questions about p-primary groups.
Here are two technical results about direct sums and products that will be
useful.

If X is an independent subset of a group G and m,x = 0, then m,x = 0

If Theorem 10.9. Let G be an abelian group, let {A,: k E K ) be a family of abelian
groups, and let {i,: A, 4 G: k E K ) be a family of homomorphisms. Then G z
for each x. If G is torsion-free, then mx = 0 for all x; however, if G has torsion, I1 CkrRAXfi and only if, given any abelian group H and any family of homo-
then one may conclude only that mx is a multiple of the order of x.
i;
morphisms { f,: A, 4 H: k E K), then there exists a unique homonzorphism
q : G 4 H making the following diagrams commute (qi, = f,):
i
Lemma 10.6. A set X of nonzero elements of a group G is independent if and
only if
<x>= C (x). xEX

83uooJ Assume that X is independent. If x, E X and y E (x,) n (X - (x,)),

then y = mx, and y = z m i x i , where the xi are distinct elements of X not Prooji We show first that G = A, has the stated property. Define
j,: A, + G to be the inclusion. By Lemma 10.4, every g E G has a unique
equal to x,. Hence
-mx, f Z mixi = 0, expression of the form g = x,,,
a,, with a, # 0 for only finitely many k. It
follows that $(g) = Cf,(a,) is a well defined function; it is easily checked that
so that independence gives each term 0; in particular, 0 = mx, = y. t,b is a homomorphism making the kth diagram commute for all k; that is,
The proof of the converse, also routine, is left to the reader. RI $jk = for all k.
Assume now that G is any group satisfying the stated property, and
Recall, in the context of abelian groups, that p-groups are called p-primary choose the diagram with H = G and f , = j,. By hypothesis, there is a map
groups. 1
cp: G 4 A, making the diagrams commute.
312 10. Abelian Groups Free Abelian Groups 313

Finally, we show that $y and y$ are identities. Both $9 and 1, complete

the diagram
We allow the possibility X = a,in which case F = 0.
It follows at once from Lemma 10.4 that if X is a basis of a free abelian
group F, then each u E F has a unique expression of the form u = Em,x,
where nz, E Z and m, = 0 for "almost all" x E X; that is, rn, # 0 for only a
finite number of x.
Notice that a basis X of a free abelian group is independent, by Lemma
10.6.
and so the uniqueness hypothesis gives $y = 1,. A similar diagram shows
that y$ is the identity on A,. Theorem 10.11. Let F be a free abelian group with basis X, let G be any abelian
group, and let f : X -+ G be any function. Then.there is a unique homomorphisrn
Theorem 10.10. Let G be an abelian group, let {A,: k E K ) be a family of y: F -+ G extending f;that is,
abeliatz groups, and let (p,: G -+ A,: k E K ) be a family of homomorphisms.
Then G r LK A, if and only if, given any abelian group H and any family
of honzomorplzisrns {f,:H -+ A,: k E K ) , there exists a unique homomorphism Indeed, if u = x m x x E F , then y(u) = 1m, f (u).
y: H -, G making the following diagrams commute for all k:

Proopf. The argument is similar to the one just given if one defines
p,:nlEK A, -+ H as the projection of a "vector" onto its kth coordinate. El
10.1. Let {A,: k E K ) be a family of torsion groups.
(i) The direct sum Eke, A, is torsion.
(ii) If n is a positive integer and if each A, has exponent n (i.e., nA, = 0 for all k),
thenn,,, A, is torsion.
10.2. If x E G, then any two solutions to the equation ny = x differ by an element z
with nz = 0. Conclude that y is unique if G is torsion-free.
10.3. If G is a torsion-free group and X is a maximal independent subset, then G / ( X )
is torsion.
PuooJ If u E F, then uniqueness of the expression u = x m x x shows that
y: uHZ mxf (u) is a well defined function. That y is a homomorphism ex-
tending f is obvious; y is unique because homomorphisms agreeing on a set
of generators must be equal.
Here is a fancy proof. For each x E X, there is a unique homomorphism
yx: ( x ) 4 G defined by m x w i n f ( x ) . The result now follows from Lemma
10.6 and Theorem 10.10. E l
Corollary 10.12. Every (abelian)group G is a quotient of a free abelian group.
ProoJ Let F be the direct sum of j G I copies of Z,and let x, denote a generator
of the 9th copy of Z, where g E G. Of course, F is a free abelian group with
basis X = {x,: g E GI. Define a function f : X -+ G by f(x,) = g for all g E G. ,

By Theorem 10.11, there is a homomorphism y: F -+ G extending f. Now y

10.4. (i) If G =
(ii) If G =
tions.
nA,, prove
A,,
that the maps A, G in Theorem 10.9 are injectioils.
i,: -+
prove that the maps p,: G A, in Theorem 10.10 are surjec-
--+ is surjective, because f is surjective, and so G r Filter y, as desired.

The construction of a free abelian group in the proof of Corollary 10.12

can be modified: one may identify x, with g. If X is any set, one may thus
construct a free abelian group F having X itself as a basis. This is quite
Free Abelian Groups convenient. For example, in Algebraic Topology, one wishes to consider
formal Z-linear combinations of continuous maps between topological
Defiwitioaa. An abelian group F is*ee abelian if it is a direct sum of infinite spaces; this can be done by forming the free abelian group with basis the set
cyclic groups. More precisely, there is a subset X c F of elements of infinite
order, called a basis of F, with F = Ex,,
( x ) ; i.e., F E Z. x of all such functions.
The corollary provides a way of describing abelian groups.
314 10. Abelian Groups
Definition. An abelian group G has generotors X and relations A if G z FIR,
where F is the free abelian group with basis X , A is a set of Z-linear combina-
tions of elements of X, and R is the subgroup of F generated by A. If X can
be chosen finite, then G is calledfinitely generated.
EXAMPLE
10.1. G = Z6 has generator x and relation 6x.
EXAMPLE
10.2. G = Z6 has generators {x, y), and relations {2x, 3 ~ ) .
EXAMPLE10.3. G = Q has generators {x,, ..., x,, ...) and relations
{x, - 2x2, x, - 3x3, ..., x,-, - nx,, ...).
EXAMPLE 10.4. If G is free abelian with basis X, then G has generators X and
no relations (recall that 0 is the subgroup generated by the empty set). The
etymology of the term free should now be apparent.
We have just seen that one can describe a known group by generators and
relations. One can also use generators and relations to construct a group
with prescribed properties.
Theorem 10.13. There is an infinite p-primary group G = Z(pW)each of whose
proper subgroups is finite (and cyclic).
ProoJ Define a group G having
generators: X = {x,, x,, .. . , x,, ...I
and
relations: {px,, xo - px,, x, - px,, . . .,x,-, - px,, ...).
Let F be the free abelian group on X, let R < F be generated by the relations,
+
and let a, = x, R E FIR = G. Then pa, = 0 and a,-, = pa, for all n 2 1, so
that pn+'a, = pa, = 0. It follows that G is p-primary, for pt+l z;=,m,a, = 0,
where nz, E 72. A typical relation (i-e., a typical element of R) has the form:
If a, = 0, then x, E R, and independence of X gives the equations 1 = mop +
m, and m,,, = pm, for all n > 1. Since R 5 F and F is a direct sum, m,, = 0
for large n. But m,,, = pl'm, for all n, and so m, = 0. Therefore, 1 = mop, and
+
this contradicts p 2 2. A similar argument shows that a, 0 for all n. We
now show that all a , are distinct, which will show that G is infinite. If a, = a,
for k > n, then a,-, = pa, implies a, = pk-na,, and this gives (1 - pk--")a,=
0; since G is p-primary, this contradicts a, # 0.
Let H 5 G. If H contains infinitely many a,, then it contains all of them,
and H = G. If H involves only a,, ... , a,, then H < (a,, .. ., a,) 5 (a,).
Thus, H is a subgroup of a finite cyclic group, and hence H is also a finite
cyclic group. Ell
Free Abelian Groups 315
The group Z ( p W )has other interesting properties (see Exercise 10.5 below),
and we shall return to it in a later section.
Theorem 10.14. Two free abelian groups F = Exex
(x) and G = EYE(y)
XI = 1 YI.
are isomorphic fi and only if I
ProoJ Since 1x1= I YI, there is a bijection f : X + Y c G, and f determines a
homomorphism cp: F + G with cp(x) = f(x) for all x E X. Similarly, there is a
homomorphism $: G -,F with $(y) = f -l(y) for all y E Y. But cp$ and $9
are identities because each fixes every element in a basis, and so cp: F -+ G is
an isomorphism.
Conversely, if p is a prime, then V = F/pF is a vector space over Z,. We
claim that = {x + pF: x E X) is a basis of V. It is clear that spans V.
Assume that C [m,] (x + pF) = 0, where [m,] E Z,and not all [m,] = [O]. If
m, is a representative of [m,], then Em,(x + pF) = 0. In F, this equation
becomes zm,x E pF; that is, there are integers n, with C m x x = x p n x x .
Independence of a basis gives m, = pn, for all x, and so [m,] = [0] for all x.
This contradiction shows that X is independent, and hence it is a basis of K
We have shown that dim FIpF = 1x1 = 1x1. In a similar way, one shows that
dimF/pF=IYI,sothatIXI=IYI.
Definition. The rank of a free abelian group is the cardinal of a basis.
Theorem 10.14 says that two free abelian groups are isomorphic if and
only if they have the same rank. The reader will not be misled by the analogy:
vector space-free abelian group; dimension-rank.
It is clear that if F and G are free abelian, then
for a basis of F @ G can be chosen as the union of a basis of F and a basis
of G.
Theearern 10.15 (Projective Property). Let P: B 4 C be a surjective hornomor-
phism of groups. If F is j7ee abelian and fi a: F -, C is a homomorphism, then ..
there exists a homomorphism y: F 4 B making the diagram below cornmute (i.e., . '
Remark. The converse is also true.
ProoJ Let X be a basis of F. For each x E X, surjectivity of a provides b, EB
3 16 10. Abelian Groups
with P(b,) = a(x). Define a function f : X -+ B by f(x) = b,. By Theorem
10.11 there is a homomorphism y: F -+B with y(x) = b, for all x. It follows
that fly = cc, for they agree on a generating set of F: if x E X, then py (x) =
P(b,) = a(x). El
Corollary 10.16. If H 5 G and G/H is free abelian, then H is a direct summand
of G; that is, G = H @ K, where K < G and K z G/H.
ProoJ Let F = G/H and let P: G -+ F be the natural map. Consider the
diagram
where 1, is the identity map. Since F has the projective property, there is a
homomorphism y: F -+ B with by = 1,. Define K = im y. The equivalence of
(i) and (iii) in Lemma 10.3 gives B = ker P @ im y = H @ I<.
We give two proofs of the next result. The first is a special case of the
second, but it contains the essential idea; the second involves infinite methods
which, though routine, may obscure the simple idea.
Theorem 10.1'7. Every subgroup H of a free abelian group F of finite rank n is
itseEffree abelian; moreover, rank(H) < rank(F).
ProoJ The proof is by induction on n. If n = 1, then F E Z. Since every
subgroup H of a cyclic group is cyclic, either N = 0 or H r Z, and so H is
free abelian of rank I1. For the inductive step, let (x,, .. . , x,,,) be a basis
of F. Define F' = (x,, . . . , x,) an9 H' = H n F'. By induction, H' is free
abelian of rank < n. Now
H/H' = H/(H n F') r (H + F1)/F' < FIF' r Z.
By the base step, either H/H' = 0 or H/H' z Z.In the first case, H = H'
and we are done; in the second case, Corollary 10.16 gives H = H' @ ( h )
for some 12 E H, where (h) z Z, and so H is free abelian and rank(H) =
rank(H1@Z)=rank(H')+ 1 i n + 1.
We now remove the finiteness hypothesis.
Tlneosem 10.18 Every subgroup H of a free abelian group F is free abelian, and
rank(N) < rank(F).
ProoJ That every nonempty set can somehow be well-ordered is equivalent
Free Abelian Groups 3 17
to the Axiom of Choice (see Appendix IV). Let (x,: k E I<) be a basis of F,
and assume that K is well-ordered.
For each k E K, define Fi = (xj: j < k) and Fk = (xj: j < Ic) = F; @ (x,);
define Hi = H n FL and Hk = H n F,. Note that F = 0 F, and H = U H,.
Now Hi = H n FL = & n FLYand so
By Corollary 10.14, either Hk = Hi or H, = Hi @ (h,), where (h,) r Z. We
claim that H is free abelian with basis the set of all h,; it will then follow that
rank(H) i rank(F), for the set of h, has cardinal < I K I = rank(F).
U
Since F = F,, each h E H (as any element of F) lies in some F,; define
p(h) to be the smallest index k for which h E F, (we are using the fact that K
is well-ordered). Let H* be the subgroup of H generated by all the 12,. Sup-
pose that H* is a proper subgroup of H. Let j be the smallest index in
(p(h): h E H and h $ H*),
and choose h' E H, h' 4 PI* with p(hl) = j. Now ,u(hl)= j gives h' E H n 4, so
that
h' =a + r?zhj, a E Hj and nz E Z.
Thus, a = h' - mhj E H, a 6 H* (lest h' E H"), and p(a) < j, a contradiction.
Therefore, H = H*.
By Lemma 10.4(ii), it remains to show that linear combinations of the h,
are unique. It suffices to show that if
where k, < . . . < k,, then each m i= 0. Of course, we may assume that m,, # 0.
But then m,hkn E (hkn) n HLn = 0, a contradiction. It follows that H is free
abelian.
'
10.5. (i) Prove, for each n 2 I, that Z(pm)has a unique subgroup of order pn.
(ii) Prove that the set of all subgroups of Z(pm)is well-ordered by inclusion.
(iii) Prove that Z(pm)has the DCC but not the ACC.
(iv) Let R, = {e2"ik1pn:k E Z, n 2 0) IC be the multiplicative group of all pth
power roots of unity. Prove that Z(pm)r R,.
10.6. (i) Prove that the group G having generators {x,, x,, x,, ...) and relations
{px,, x, - pnxn,all n '2 1) is an infinite p-primary group with pnG #
0.
(ii) Prove that the group G in (i) is not isomorphic to Z(pm).
10.7. (i) Prove that an abelian group G is finitely generated if and only if it is a
quotient of a free abelian group of finite rank.
 10. Abelian Croups
(ii) Every subgroup H of a finitely generated abelian group G is itself finitely
generated. Moreover, if G can be generated by r elements, then H can be
generated by r or fewer elements.
10.8. Prove that the multiplicative group of positive rationals is free abelian (of
countably infinite rank). (Hint. Exercise 1.52(ii).)
10.9. If F is a free abelian group of rank n, then Aut(F) is isomorphic to the multi-
plicative group of all n x n matrices over Z with determinant = 1. +
10.10. An abelian group is free abelian if and only if it has the projective property.
10.11. If F is a free abelian group of rank n and H is a subgroup of rank k < n, then
F/H has an element of infinite order.
f
10.12. (i) If A + B 3 C 5 D is an exact sequence of free abelian groups, prove that
B r i m f Okerh.
(ii) If n 2 1 and 0 -,F, -,... -,F, -,F, -+ 0 is an exact sequence of free abelian
groups of finite rank, thenx;='=,
rank(Fi) = 0.
10.13. Prove the converse of Corollary 10.16: If a group G is (isomorphic) to a direct
summand whenever it is a homomorphic image, then G is free abelian.
10.14. A torsion-free abelian group G having a free abelian subgroup of finite index is
itself free abelian.
Finitely Generated Abelian Groups
We now classify all finitely generated abelian groups.
Tirheorrewn 10.19. Every finitely generated torsion-free abelian group G is fiee
abelian.
Pro05 We prove the theorem by induction on n, where G = (x,, ...,x,). If
n = 1 and 6 # 0, then G is cyclic; G z Z because it is torsion-free.
Define H = (g E G: mg E (x,) for some positive integer m). Now H is a
subgroup of G and GIH is torsion-free: if x E G and k(x + H) = 0, then kx E
H, m(kx) E (x,), and so x E H. Since GIH is a torsion-free group that can be
generated by fewer than n elements, it is free abelian, by induction. By Corol-
lary 10.16, G = F @ H, where F r G/H, and so it suffices to prove that H is
cyclic. Note that H is finitely generated, being a summand (and hence a
quotient) of the finitely generated group G.
If g E H and g # 0, then mg = kx, for some nonzero integers m and k. It is
routine to check that the function cp: H -+ Q, given by g- k/m, is a well
defined injective homomorphism; that is, H is (isomorphic to) a finitely gener-
ated subgroup of 0, say, H = (a,/b,, .. ., a,/b,). If b = n;=,
bi, then the map
(1: fl -+ Z, given by h I+ bh, is an injection (because H is torsion-free). There-
Finitely Generated Abelia~iGroups 319
fore, H is isomorphic to a nonzero subgroup of Z,and hence it is infinite
cyclic.
Theorem 146.20 (Fundamental Theorem). Every finitely generated abeliaiz
group 6 is a direct sum of primary and infinite cyclic groups, and the nuinber
of summands of each kind depends only on G.
ProoJ Theorem 10.19 shows that G/tG is free abelian, so that Corollary 10.16
gives G = tG @ F, where F r GltG. Now tG is finitely generated, being a
summand and hence a quotient of G, and Exercise 6.18(ii) shows that tG is
finite, The basis theorem for finite abelian groups says that tG is a direct sum
of primary cyclic groups.
The uniqueness of the number of primary cyclic summands is precisely
Theorem 6.11; the number of infinite cyclic summands is just rank(G/tG), and
so it, too, depends only on G.
The next result will give a second proof of the basis theorem.
Theorem 10.21 (SimuBtaneous Bases). Let H be a subgroup of finite index in a
free abelian group F of finite rank n. Then there exist bases (y,, . . . ,y,)
of F
and (h,, . . ., h,) of H such that hi E (yi) for all i.
Pvooj: If (x,, .. ., x,) is an ordered basis of F, then each element h E H has
coordinates. Choose an ordered basis and an element h so that, amo'ng all
such choices, the first coordinate of h is positive and minimal such. If h =
+
k,x, -). .. . k,x,, then we claim that k, divides ki for all i 2 2. The division
algorithm gives lci = qik, + ri, where 0 _< ri < k,. Therefore,
Define y, = x, + q2x2 + . - . + q,,xn,and note that (y,, x,, . . . , x,) is an or-
dered basis of F. Now h = k, y, + r2x, + - . . + r,x,. If ri # 0 for some i, then
the first coordinate of h relative to the ordered basis (xi, y l , . .., x,) violates
the minimality-of our initial choice. Therefore, ri = 0 for all i >_ 2 and k,
divides ki for all i 2 2.
If h' = m,y, + m2x2 + + in,x, is any element of H, we claim that k,
divides m,. For if m, = qk, + r, where 0 r < k,, then k' - qh E H has first
coordinate r < k,, a contradiction. It follows that the map n:H -+ H, given
by lz'~-+m,y,, is a retraction with image ( 1 2 ) . By Lemma 10.3, H = (h) @
lter n = ( h ) @ (H n (x,, . . ., x,)). Since (x,, . . . , x,) is free abelian of rank
n - 1 and H n (x,, . . . ,x,) is a subgroup of finite index, the proof can be
completed by induction on n. EY
Corollary 163.22 (Basis Theorem). Every finite abelian group G is a direct szttn
of cyclic groups.
 320 10. Abelian Groups
Proof. Write G as FIR, where F is free abelian of finite rank n, say. By
Theorem 10.21, there are bases {y,, . ..,y,) and {h,, ...,h,) of F and R,
respectively, with hi = kiyi for all i. By Theorem 2.30, G z Gi. &I x7=,
10.15. If F is a free abelian group of finite rank n, then a subgroup H of F has finite
index if and only if H is free abelian of rank n.
10.16. Let (x,, ..., x,) be a basis of a free abelian group F. If k,, ..., k, are inte-
gers with gcd(k,, ..., k,) = 1, then there are elements y,, ..., y, such that
+ +
( k ,x, - - - k,x,, y,, ...,y,) is a basis of F.
10.17. Let F be free abelian of rank n and let H be a subgroup of the same rank. Let
(x, , .. .,x,) be a basis of F, let (y,, ...,y,) be a basis of H, and let yj = miixi-
Prove that
[F : HI = ldet[mii]l.
(Hint. Show that Idet[rnij] 1 is independent of the choice of bases of F and of H.)
Divisible and Reduced Groups
A reader of Chapter 1, asked to give examples of infinite abelian groups,
probably would have responded with Z, Q, R, and C.We now study a com-
mon generalization of the latter three groups.
Definition. k group G is divisible if each x E G is divisible by every integer
rz > 2; that is, there exists g, E G with ng, = x for all n 2 2.
EXAMPLE 10.5. The following groups are divisible: Q; R; C;the circle group T;
Z(pW);the multiplicative group F x of all nonzero elements of an algebraically
closed field F (in particular, @ ").
10.6. Every quotient of a divisible group is divisible.
EXAMPLE
EXAMPLE 10.7. If {A,: k E K ) is a family of groups, Then each ofx,. A, (and
nkEKA,) is divisible if and only if every A, is divisible.
10.8. A torsion-free divisible group G is a vector space over Q.
EXAMPLE
If x E G and n > 0, then there is a unique y E G with ny = x, by Exercise
10.2. There is thus a function Q x G -+ G, given by (m/n, x)t+my (where
ny = x), which is a scalar multiplication satisfying the axioms in the defini-
tion of vector space.
Theorem 10.23 (Injective Property, Baea, 1940). Let D be a divisible group and
let A be a subgroup of a grotrp B. If f : A D is a homomorphism, then f
Divisible and Reduced Groups 321
can be extended to a homornorphisin cp: B 4 D; that is, the following diagral?~
commutes:
Proof. We use Zorn's lemma. Consider the set Y of all pairs (S, h), where
A 5 S < B and h: S -+ D is a homomorphism with hlA = f. Note that 9 #
@ because (A,f ) E Y. Partially order 9 by decreeing that (S, h) 5 (S', h') if
S 5 S'and h' extends h; that is, Iz'lS = h. If % = {(S,, ha)) is a simply ordered
Sa and = U, ha (this makes sense if one
subset of 9, define (g, h) by S =Ua
if
realizes that a function is a graph; in concrete terms, s _ $,~ then s E Sa for
some a, and i(s) = ha(s)).The reader may check that (S, h) E 9 and that it is
an upper bound of % By Zorn's lemma, there exists a maximal pair (M, g) E
9.We now show that M = Byand this will complete the proof.
Suppose that there is b E B with b 4 M. If M' = (My b), then M < MI, and
so it suffices to define h': M' 4 D extending g to reach a contradiction.
Case 1. M n (b) = 0.
In this case, M' = M @ (b), and one can define lz' as the map rn + lzb H
s(m).
Case 2. M n (b) # 0.
If k is the smallest positive integer for which kb E M , then each y E M' has
a unique expression of the form y = m + tb, where 0 < t < k. Since D is
divisible, there is an element d E D with kd = h(1cb) (kb E M implies h(kb) is
defined). Define h': M' 4 D by m + tb c g(m) + td. It is a routine calculation,
left for the reader, that h' is a homomorphism extending g.
Corollary 161.24. If a divisible group D is a szrbgrotrp of a group G, then D is a
direct summand of G.
ProoJ Consider the diagram:
D
?'\.
\
\
O - D r i , B
where 1, is the identity map. By the injective property, there is a homomor-
phism cp: G 4 D with (pi = 1, (where i is the inclusion D % G); that is,
q(d) = d for all d E D. By Lemma 10.3, D is a direct summand of G.
Definition. If G is a group, then dG is the subgroup generated by all the
divisible subgroups of G.
322 10. Abelian Groups
Note that dG is a fully invariant subgroup, for every image of a divisible
group is divisible.
Lemma 10.25. For any group G, dG is the unique maximal divisible subgroup
sf 6 .
ProoJ It suffices to prove that dG is divisible. Let x E dG and let n > 0. Now
x = dl + ... + d,, where each di E Di, a divisible subgroup of G. Since Di is
+
divisible, there is y, E D, with ny, = di for all i. Hence y, + . . - y, E dG and
n(yl + + y,) = x, as desired. dI!
Defi~aition.A group G is reduced if dG = 0.
Of course, G is divisible if and only if dG = G.
Theorem 10.26. For every group G, there is a decomposition
G=dG@R.
where R is reduced.
PYQQ$Since dG is divisible, Corollary 10.24 gives the existence of R. If D < R
is divisible, then D < R n dG = 0, by Lemma 10.25.
The reader should compare the roles of the subgroups tG and dG. Every
abelian group G is an extension of the torsion group tG by a torsion-free
group (but tG need not be a direct summand); G is an extension of the
divisible group dG by a reduced group, and dG is always a direct summand.
Recall that if G is a group, then G[n] = (x E G: nx = 0).
Lemma 10.27. I f G and H are divisible p-primary groups, then G r H if and
only if G[p] r H[p].
ProoJ Necessity follows easily from the fact that q(G[p]) < H[p] for every
homomorphism q : G + H.
For sufficiency, assume that q : G [p] -,H [p] is an isomorphism; com-
posing with the inclusion H [p] c+ H, we may assume that cp: G [p] --+ H. The
injective property gives the existence of a homomorphism @: G -, H extend-
ing q ; we claim that @ is an isomorphism.
(i) @ is injective.
We show by induction on n 2 1 that if x E G has order pn, then @(x)= 0.
If n = 1, then x E G [PI, SO that @(x)= cp(x) = 0 implies x = 0 (because q is
injective). Assume that x has order pnfl and @(x)= 0. Now @(px)= 0 and
px has order pn, so that px = 0, by induction, and this contradicts x having
order pnf '.
(ii) @ is surjective.
Divisible and Reduced Groups 323
We show, by induction on n 2 1, that if y E H has order p", then y E im @,
If n = 1, then y E N[p] = im q I im 0.Suppose now that y has order pnfl.
Since p"y E N[p], there is x E G with @(x)= pity; since G is divisible, there is
g E G with png = x. Thus, pn(y - @(x))= 0, so that induction provides z 6 G
with @(z) = y - @(g).Therefore, y = @(y + g), as desired.
Theorem 10.28. Every divisible group D is a direct sum of copies of Q and of
copies of Z(pm)for various primes p.
ProoJ It is easy to see that tD is divisible, so that D = tD @ V, where V r
DltD. Now V is torsion-free and divisible, so it is a vector space over Q, by
Example 10.8; by Theorem 10.5, V is a direct sum of copies of Q.
For every prime p, the p-primary component G of tD is divisible (it is a
summand of a divisible group). Let r = dim G[p] (as a vector space over Z,),
and let H be the direct sum of r copies of Z(pm). Now H is a p-primary
divisible group with G [p] r H [p], and so G r H, by the lemma. EJ
Notation. If D is a divisible group, let 8,(D) = dimQ DltD and let J,(D) =
DCpl.
The proof of the next theorem is left as an exercise.
Theorem 110.29.If D and D' are divisible groups, then D r D' if and only if
6,(D) =. 6,(D1) and, for all primes p, 6,(D) = 6,(D1).
There is an analogy between theorems about free abelian groups and theo-
rems about divisible groups that may be formalized as follows. Given a com-
mutative diagram containing exact sequences, then its dual d ~ g r a mis the
diagram obtained from it by reversing all arrows. For example, the dual
diagram of 0 --+ A -+ B is B --+ A --+ 0, and this leads one to say that "sub-
group" and "quotient group" arc dual notions. The notion of short exact
sequence is self-dual, Theorems 10.9 and 10.10 show that "direct sum" and
"direct product" are dual, and the projective property is dual to the injective
property (suggesting that free abelian groups are dual to divisible groups).
The next result should be compared to Corollary 10.12.
Theorem 10.30. Every group G can be imbedded in a divisible group.
ProoJ Write G = FIR, where F is free abelian. Now F = Z Z , SO that F I
Q (just imbed each copy of Z into Q). Hence G = FIR = Z)/R 5
(zQ)/R, and the last group is divisible, being a quotient of a divisible
(z
group. El
I The next result should be compared to Exercise 10.12.
I
i
! CoroEllary BO.31. A grozrp G is divisible if and only if it is a direct sunzi?laizdof
1
any group co~ztainingif.
i
324 10. Abelian Groups Torsion Groups 325

PrmJ Necessity is Corollary 10.24. For sufficiency, Theorem 10.30 says that Torsion Groups
there is a divisible group D containing G as a subgroup. By hypothesis, G is
a direct summand of D, and so G is divisible. Torsion groups can be quite complicated, but there are two special classes of
torsion groups that are quite manageable: divisible groups and direct sums of
cyclic groups. We shall prove that every torsion group is an extension of a
direct sum of cyclics by a divisible. The proof involves a special Icind of
10.18. Show that the group G/tG in Theorem 10.2 is divisible. subgroup that we now investigate.
10.19. If 0 -t A -+ B -+ C -+ 0 is an exact sequence and if A and C are reduced, then B
is reduced. Definitions. A subgroup S 2 G is a pawe subgroup if, for every integer 1.1,
10.20. (i) If G is the group in Exercise 10.6, then G/(a,) is a direct sum of cyclic S n nG = rzS.
groups.
(ii) Show that G is reduced. It is always true that S n nG T: nS, and so it is only the reverse inclusion
10.21. (i) Prove that Q/Z r z,
Z(pm).(Hint. Use Exercise 10.5.)
(ii) Prove that (Q/Z)[n] r Z,.
that is significant: if s E S n nG, then s E nS; that is, if s E S and s = rzg for
some g E G, then there exists s' E S with s = rzs'.
10.22. Prove that a group D is divisible if and only if it has the injective property.
(Hilzt. (n) < Z for all n 2 1; extend homomorphisms (n) -+ D to homomor- EXAMPLE
10.9. Every direct summand is pure.
phisms Z 4 D.)
10.23. (i) A group G is divisible if and only if G = pG for all primes p.
+
Let G = A @ B. If a E A and a = ng, then g = a' b', for a' E A and b' E B.
Wow nb' = 0, for nb' = a - rza' E A n 3 = 0. Hence a = rza' and A is pure.
(ii) A p-primary group is divisible if and only if G = pG. (Hint. Use Exercise
1.31.)
EXAMPLE
10.10. If S 2 G and G/S is torsion-free, then S is pure.
10.24. If G and H are groups, prove that G r H if and only if dG r dN and G/dG r
H/dH. If s = ng, then g + S E G/S has finite order; since G/S is torsion-free,
10.25. The following conditions on a group G are equivalent: g + S = S, and g E S.
(i) G is divisible;
(ii) Every nonzero quotient of G is infinite; and EXAMPLE
10.11. tG is a pure subgroup of G that may not be a direct summand
(iii) G has no maximal subgroups. of G.
10.26. If G and H are divisible groups each of which is isomorphic to a subgroup
of the other, then G r H. Is this true if we drop the adjective "divisible"? By Theorem 10.1, G/tG is torsicn-free, and so Example 10.10 shows that tG
10.27. If G and H are divisible groups for which G @ G r H @ H, then G r H. is pure. It follows from Theorem 10.2 that tG need not be a direct summand.
10.28. (i) Prove that the following groups are all isomorphic: R/Z; the circle group
n,
'If; Z(prn);R 0(Q/Z); @ x .
(ii) Prove that t(CX)E Q/Z.
Lemma 10.32. Let S be a pure subgroup of G, and let v: G -+ G/S be the rzatural
map. If y E GIs, then there is x E G with v ( x ) = y sttclz that x arzcl y have the
10.29. Prove that every countable abelian group G can be imbedded in
Dir QS @ (Q/Z) for all i.
x~"=,~,
where
sa.me order.

ProoJ Surjectivity of the natural map provides x E G with v(x) = y. If y has

10.30. (i) Every torsion-free group G can be imbedded in a vector space over 0 .
(Hint. Imbed G in a divisible group D, consider the natural map D -+ DltD,
and use Example 10.8.)
(ii) If a maximal independent subset of a torsion-free group G has n elements,
then G can be imbedded in a vector space over Q of dimension n.
10.31. If A is a group and m E Z,define in,: A + A by a-ma.
(i) Show that A is torsion-free if and only if m, is an injection for all m # 0.
(ii) Show that A is divisible if and only if in, is a surjection for every in # 0.
(iii) Show that A is a vector space over Q if and only if m,, is an automorphism
for every m # 0.
infinite order, then so does x. If y has finite order rz, then v(izx) = tzv(x) =
ny = 0, so that n x E ker v = S. Since S is pure, there is s' E S with n x = rzs'. If
z = x - s', then nz = 0 and v(z) = v ( x - s') = y. But n divides the order of z,
by Exercise 2.14, and so z has order n. El
Lemma 10.33. Let T 5 G be pure. If T 5 S 5 G, then SIT is pzii-e in G / T fi
and only if S is pure in G.
Pros$ Suppose that SIT is pure in G/T. Assume that s E S and s = n g for
326 10. Abelian Croups
some g E G. In G/T, S = rzg (where bar denotes coset mod T), and so there is
+
s' E S with S = nS'; that is, there is t E T with s = ns' t. Thus t = n(g - s'),
and the purity of T gives t' E T with ng - ns' = nt'. Hence s = ng =
+
n(sl t'). But s' + t' E S, because T I S, and so S is pure in G.
Conversely, suppose that S is pure in G. If S E SIT and S = rzg in G/T, then
+ +
ng = s t for some t E T. Since T 5 S, we have s t E S, and purity gives
s' E S with s + t = ns'. Therefore, S = nS' and SIT is pure.
Lemma 10.34. A p-primary group G that is not divisible contains a pure non-
zero cyclic subgroup.
Proof. Assume first that there is x E G[p] that is divisible by pk but not by
pk+l,and let x = pky.We let the reader prove that (y) is pure in G (Exercise
1.31 shows that one need check only powers of p).
We may, therefore, assume that every x E G[p] is divisible by every power
of p. In this case, we prove by induction on k > 1 that if x E G and pkx = 0,
then x is divisible by p. If k = 1, then x E G[p], and the result holds. If
p"+lx = 0, then pkx E G [PI, and so there is z E G with pk+lz= pkx. Hence
pk(pz - x) = 0. By induction, there is w E G with pw = pz - x, and x =
p(z - w), as desired.
We have shown that G = pG, and so Exercise 10.25(ii)gives G divisible, a
contradiction. Ed
DeGr~ition.A subset X of a group G is pure-independent if it is independent
and (X) is a pure subgroup of G.
Lemma 10.35. Let G be a p-primary group. If X is a maximal pure-independent
subset of G (i.e., X is contained in no larger such), then G/(X) is divisible.
Proof. If G/(X) is not divisible, then Lemma 10.34 shows that it contains a
pure nonzero cyclic subgroup (7); by Lemma 10.32, we may assume that
y E G and 7 E G/(X) have the same order (where y t-+ 7 under the natural
map). We claim that (X, y) is pure-independent.
Now (X) 5 (X, y) 5 G, and (X, y)/(X) = (7) is pure in G/(X); by
Lemma 10.32, (X, y) is pure in G.
Suppose that my + z r n i x i = 0, where xi E X and m, mi E Z. In G/(X), this
equation becomes my = 0. But y and 7 have the same order, so that my = 0.
Hence 1 mixi = 0, and independence of X gives mixi = 0 for all i. Therefore
(X, Y) is independent, and by the preceding paragraph, it is pure-indepen-
dent, contradicting the maximality of X.
Definition. A subgroup B of a torsion group G is a basic subgroaap if:
(i) B is a direct sum of cyclic groups;
(ii) B is a pure subgroup of G; and
(iii) G/B is divisible.
Torsion Groups 327
Theorem 80.36 (Kilealikoat, 1945). Every torsion group G has a basic subgroup,
and so G is an extension of a direct sum of cyclic groups by a divisible group.
PuooJ Let G = Gp be the primary decomposition of G. If G, has a basic
subgroup of Bp, then it is easy to see that B, is a basic subgroup of G. Thus,
we may assume that G is p-primary.
If G is divisible, then B = 0 is a basic subgroup. If G is not divisible, then it
contains a pure nonzero cyclic subgroup, by Lemma 10.34; that is, G does
have pure-independent subsets. Since both purity and independence are pre-
served by ascending unions (see Exercise 10.28(i)),Zorn's lemma applies to
show that there is a maximal pure-independent subset X of G. But Lemmas
10.6 and 10.33 show that B = (X) is a basic subgroup.
The following theorem was proved by H. Priifer in 1923 for G countable;
the general case was proved by W. Baer in 1934.
Corollary 10.37 (Priifer-Baer). If G is a group of bounded order (i.e., nG = 0
for some n > 0), then G is a direct surn of cyclic groups.
Remark. Were G nonabelian, we would say "G has finite exponent" instead
of "G is of bounded order."
Proof. By Theorem 10.28, a bounded divisible group must be 0. Therefore, if
B is a basic subgroup of G, then GIB = 0 and B = G. Eil
Assume that G is a direct sum of p-primary cyclic groups. Let B, be the
direct sum of all those summands of order p", if any, so that G = B, @
B, @ .. . When G is finite, we proved (in Chapter 6) that d, = dim p"G/pnf 'G
is the number of cyclic summands of order 2 pn+l, so that the number of
cyclic summands of order precisely pn+l is just d,, - dn+,. This formula
does not generalize to infinite groups because one cannot subtract infinite
cardinals.
If G is an infinite direct sum of p-primary cyclic groups, it is still true that
,
d, is the number of cyclic summands of order >_ pn+l.How can we distin-
guish those elements in pnG coming from cyclic summands of order pn+l from
those cyclic summands of larger order? The elementary observation that a
cyclic summand (a) has order pn+l if and only if pna has order p suggests that
we focus on elements of order p; that is, replace pnG by pnG n G[p].
Definition. If G is a p-primary group and rz 2 0, then
Lemma 10.38. If G is a direct sum of p-primary cyclic groups, then U (n, G) is
the number of cyclic summands of order pnf l.
 Torsion Groups 329
328 10. Abelian Groups

Proof. Let Bn be the direct sum of all those cyclic summands of order pn, if
any (in the given decomposition of G), so that G = B, @ B, @ . . @ Bk @ . . .;
let bk be the number of summands in B, (of course, bk may be 0). It is easy to
see that
G[p] = B, @ pB2 @ p 2 ~ @ , . - -@ p k - ' ~ , @ - - .
and
pnG = pnB,,+,@ . . . @ pnBk@ .. . .
Hence, for all n 2 0,
+ +
v: G -+ G/(S nG) be the natural map. The group G/(S nG) is of bounded
order (it has exponent n) so that it is a direct sum of cyclic groups, by Corol-
+
lary 10.33. Write G/(S nG) = xi{Zi), where Zi has order ,; for each i,
choose xi E G with v(xi) = xi. Now
riXi = si + ng,,
where si E S and gi E G. But ri divides (the exponent) n, so that
si = ri(xi - (n/ri)gi).
Since S is pure, there is ti E S with si = riti. Define

and so
n GCpl) r P " ~ + I .
(pnGn GCPI)/(P"+~G Now v(y,) = Zi and riyi = ng,. Let K be the subgroup generated by nG and
Therefore, U{n, G) = dim(pnBn+,)= b,,,, as desired. all the y,; we shall show that G = S @ PC.
(i) S n K = 0.
Theorem 10.39. If G and H are direct sums of p-primary cyclic groups, then
G z H if and only if U(n, G) = U{n, H) for all n 2 0.
x
Let s E S n K ; since s E K, s = miyi + nh; since s E S, V(S)= 0. Thus, 0 =
znziZi, and independence gives mixi = 0; hence, I;. divides mi for all i. We
have chosen yi so that riyi E nG, hence miyi E nG. Therefore, s = iniyi + 1
Proof. The numbers U (n, G) depend only on G and not upon the decomposi- nh E nG. Since S is pure, there is s' E S with s = izs' E nS = 0.
tion. B!
(ii) S + K = G.
1
Theorem 10.660. Any two basic subgroups of a p-primary group G are
isomorphic.
If g E G, then v(g) = liZi. Since v ( x liyi) = liZi, we have g - liyi E
+ 1 +
ker v = S nC; say, g - liyi = s nh. Thus, g = s + (nh + liyi)E
S-t-K. El
x1
Proof. Let B be a basic subgroup of G. The number bn of cyclic summands of
B of order pn is equal to the number of such summands of Blpn+'B, and so it C o r ~ l l ~ 10.42.
ry If tG is of bounded order, then tG is a direct szrnzmaizd of G.
suffices to show that this latter quotient depends only on G. In particular, tG is a direct sz.lmmand when it is finite.
We claim, for every n 2 1, that G = B + pnG. If g E G, then divisibility of
+ +
G/B gives g B = pnx B for some x E G, so there is some b E B with g = Cabroliary 10.43. A torsion group G that is not divisible has a p-primary cyclic
direct summand (for some prime p).
b + p"x E B + pnG. It follows that
G/pn+'G = (B + pnCIG)/pn+'G Pvoo/:Since G is not divisible, at least one of its primary components, say, G,,
r B/(B n pn+'G) is not divisible. By Lemma 10.34, G, has a pure nonzero cyclic summand C,
and C must be a summand of G, by the theorem. El
= B/p"+'By by purity.

Therefore, B/pnflB is independent of the choice of B. Corolllary 10.44. An indecomposable group G is either torsion or torsion-3ee.

An example is given in Exercise 10.41 below showing that the divisible Proo$ Assume that 0 < tG < G. Wow tG is not divisible, lest it be a summand
quotient GIB, where B is a basic subgroup, is not determined by G. of G, so that Corollary 10.43 shows that G has a (cyclic) summand, contra-
Were is a condition forcing a pure subgroup to be a summand. dicting indecomposability. E3

Corollary 10.41 (PrEfer, 1923). A pure subgroup S of bounded order is a direct Mere are three lovely results (see Fuchs, Griffith, or Kaplansky for proofs).
summand. All countable torsion groups are classified by UZm's Tlzeorern (1933): if G is a
p-primary group, there is a transfinite version of the numbers U{ir, G) (the
Proof. Assume that S < G is pure and that nS = 0 for some n > 0. Let Ulin invariants) with n varying over ordinal nurnbers, and two countable
330 10. Abelian Groups
torsion groups are isomorphic if and only if their respective primary com-
ponents have the same Ulm invariants. (There are uncountable p-primary
groups having the same Ulm invariants which are not isomorphic; see Exer-
cise 10.39(ii) below.) A theorem of Priifer (1923) says that a countable p-
primary group G is a direct sum of cyclic groups if and only if pnG = 0
(this is false for uncountable groups, as is shown in Exercise 10.39(iii) below).
Mulikov (1941) has characterized direct sums of cyclic groups, and one conse-
quence of his criterion is that every subgroup of a direct sum of cyclic groups
is another such.
10.32. If G is torsion-free, then a subgroup S is pure if and only if G/S is torsion-free.
10.33. (i) Given an example of an intersection of two pure subgroups of a group G
not being pure (Hint: Take G = Z, @ Z,.)
(ii) Given an example in which the subgroup generated by two pure subgroups
is not pure. (Hint. Look within a free abelian group of rank 2.)
10.34. (i) If G is torsion-free, then any intersection of pure subgroups is pure, and one
can define the prire subgrorip generafed by a subset X (as the intersection of
all the pure subgroups containing X).
(ii) Let G be torsion-free, and let x E G. Show that the pure subgroup gener-
ated by x is:
{g E G: nzg E (x)}.
(We have rediscovered the subgroup H in the proof of Theorem 10.19.)
10.35. A pure subgroup of a divisible group is a direct summand.
10.36. (i) Show that an ascending union of pure subgroups is always pure.
(ii) Show that an ascending union of direct summands need not be a direct
summand. (Hint. Consider n,,,Z,.)
10.37. (i) If G = t(n;& Z,,), then G is an uncountable group with U{n, G} = 1 for
all n 2 0.
(ii) Show that Ulm's theorem does not classify uncountable torsion groups.
(Hint.t(n,"P=, Z,,) $ xz, Zen.)
(iii) Prove that G = t(n,"', Z,,) is not a direct sum of cyclic groups.
10.38. Prove that a torsion group is indecomposable if and only if it is isomorphic to
a subgroup of Z(pm)for some prime p.
10.39. Show that n , Z , is not a direct sum of (possibly infinitely many) indecom-
posable groups.
10.40. (Kapllarasky). In this exercise, G is an infinite group.
(i) If every proper subgroup of G is finite, then G r Z(pm)for some prime p.
(ii) If G is isomorphic to every proper subgroup, then G r Z.
(iii) If G is isomorphic to every nonzero quotient, then G z Z(pm).
(iv) If every proper quotient is finite, then G z Z.
Subgroups of Q 33 1
10.41. Let G =
n 2 1) I
x,"=,
(a,), where n, has order pn+', and let B = (pa,, a,, - pa,,,,,
G. Show that B is a basic subgroup of G. As G is a basic subgroup of
itself, conclude that the quotient G/B, where B is basic in G, is not an invariant
of G.
Subgroups of Q
The notion of rank can be generalized from free abelian groups to arbitrary
torsion-free groups.
Ifheapreim 10.45. If G is a torsion-free group, then any two ilzaxi~nalindependent
sets in G have the same number of elements. If a maxitnal independent subset
of G has r elements, then G is an additive subgroup of an r-dirnensional vector
space over Q.
ProoJ Let X be a maximal independent subset of G. By Exercise 10.30, G can
be imbedded in a vector space W over Q; let V be the subspace of W spanned
by G. If X spans V , then it is a basis of V; this will prove both statements in
the theorem, for then all maximal independent subsets of G will have dim V
elements.
1
If v E V , then there is a finite sum v = qigi, where qi E Q and gi E G; if b
is the product of the denominators of the q,, then bu E G. By Exercise 10.3,
maximality of X gives G/(X) torsion There is thus a nonzero integer in with
mbv E (X); that is, rnbv is a Z-linear combination of elements in X, and so v
is a Q-linear combination of elements in X.
Definitiarum. The rank of a torsion-free group G is the number of elements p(G)
in a maximal independent subszt. Define the raank p(G) of an arbitrary
abelian group G to be p(G/tG).
Theorem 10.45 shows that the rank is independent of the choice of maxi-
mal independent subset; when G is torsion-free, it characterizes p(G) as the
(minimal) dimension of a vector space V over Q containing G. Thus, a tor-
,
sion-free group of finite rank is just a subgroup of a finite-dimensional vector
space'over Q; in particular, a torsion-free group of rank 1 is just a nonzero
subgroup of the additive group of rational numbers Q.
Mere are three subgroups of Q.
A: all rationals having squarefree denominators;
B: all dyadic rationals; that is, all rationals of the form ~ 1 / 2 ~ .
C: all rationals whose decimal expansion is finite.
No two of these groups are isomorphic. For example, A $ B because 3 con-
tains a nonzero solution x to the system of equations 2kyk= x, while A does
332 10. Abelian Groups
not. One can also describe C as all rationals whose denominators are re-
stricted to be powers of 10, and the same reasoning shows that A $ C and
B $ C.
Definition. Let G be a torsion-free group and let x E G. If p is a prime, then
the p-height of x, denoted by hp(x),is the highest power of p dividing x in G:
more precisely, if png,,= x is solvable in G for all n, then h,(x) = a ; if k is the
largest integer n for which png = x is solvable in G, then h,(x) = k.
Each nonzero x in a torsion-free group G determines its height seqeaence
h(x) = (h,(x), h3(x), . . . , h,(x), ...), which is a sequence of nonnegative inte-
gers and the symbol a . For example, each of the groups Z,Q, A, Byand C
contains x = 1; its height sequence in each group is:
Different elements in the same group G of rank 1 may have different height
sequences. For example, the height sequence of x = 168 = 2 3 .3 . 7 in each of
the groups above is:
z : (3, l,O, l,O,O ,...);
Q: ( a , coycoya , . . .);
A: (4, 2, 1, 2, 1, 1, 1, . ..);
B: ( a , 1,0, 1,0,0,0, ...); and
C: ( a , 1, coy1,0,0,0, ...).
We have been led to the following definitions.
Definition. A characteristic is a sequence of nonnegative integers and the
symbol co. Two characteristics are equiuabernt if:
(i) they have co in the same coordinates; and
(ii) they differ in at most a finite number of (other) coordinates.
An equivalence class of characteristics is called a type.
Lemma 10.46. Let G be a torsion-free group of rank 1. If x, y E G are nonzero,
then their height sequences are equivalent.
Jf y = nx, where ~z = p ; ~ . .. p?, then Iz,(x) = h,(y) for all primes p #
PYOQ.~;
Subgroups of Q 333
+
ply . . . ,pt, and hpi(y)= ei lzpi(x)for i = 1, . . . , t (we agree that co + k = co).
Thus, the result is true in this case.
For the general case, note that if G is isomorphic to a subgroup of Q, then
there are nonzero integers m and rz with my = nx. Thus, the height sequences
of y, my = nx, and x are all equivalent. D
As a result of the lemma, one may define the type z(G) of a torsion-free
group G of rank 1 as the equivalence class of the height sequence of any of its
nonzero elements.
Theorem 10.47. If G and G' are torsion-free of rank 1, then G z G' fi
and only fi z(G) = z(G1).
ProoJ If x E G and y : G -+ G' is a homomorphism, then hp(x) < Iz,(y(x)) for
all primes p; if y" exists, then h,(y(x)) I h,(y7'(q(x))) = hp(x). Hence, if y
is an isomorphism and x E G, then x and y(x) have the same height sequence,
so that z(G) = z(Gf).
For the converse, the hypothesis says that nonzero elements x E G and
x' E G' have equivalent height sequences. Let P be the finite set of primes p
for which hp(x) < h,(xl), and let Q be the finite set of primes for which
h,(x) > hP(x1)(of course, P or Q may be empty). For p E P, define e, =
hP(x1)- hp(x) (the definition of equivalence says that both hp(x) and IzP(x1)
are finite); for q E Q, define f, = h,(x) - hP(xf).If m = n,,,
pep and n =
n,,,qfq, then it is easy to see that nzx and nx' have the same height
sequence.
Let us now assume that both G and G' are subgroups of Q containing
elements y and y', respectively, having the same height sequence; let y = n/b
and y' = a'/bl. The subgroup (b/a)G of Q is isomorphic to G, and the sub-
group (b'la') G' is isomorphic to G'. Replacing G by (b/a) G and G' by (b'/a1)G',
we may assume that 1 lies in both G and G' and that it has the same height
sequence in each group. But it is now an easy exercise that G = G'. Etl
There is an existence theorem complementing the uniqueness theorem just
proved.
Theorem 18.48. For every type z, there exists n torsion-free group G of rank 1
with z(G)'= z.
HPuooJ If (lcZ,k3, . .. , kp, . . .) is a characteristic in z, define G as the subgroup
of Q generated by all rationals of the form l/pe, where e I k, if lc, is finite,
and e is any positive integer if Ic, = co. It is easy to see that the height
sequence of 1 E G is the given characteristic. El
Torsion-free groups of rank 2 2 are not classified (though there do exist
several kinds of description of them). After L. Fuchs (1971) showed that there
334 10. Abelian Groups Character Groups 335

are indecomposable groups of all ranks I., where r is smaller than the first
strongly inaccessible cardinal (should such exist), S. Shelah (1974) showed
that indecomposables of every rank r exist.
There are groups of infinite rank that are not direct sums of indecomposable
groups. Every group of finite rank is either indecomposable or a direct sum
of indecomposables; if all the summands have rank 1, then R. Baer (1937)
showed that the summands are unique to isomorphism; otherwise, the sum-
mands need not be unique; indeed, not even the ranks of the summands are
determined. B. Jbnsson (1957) introduced the notion of quasi-isomorphism2:
two torsion-free groups of finite rank are quasi-isomorphic if each is iso-
morphic to a subgroup of the other having finite index. There is a corre-
sponding notion of indecomposable: a group is strongly indecomposable if it
is not quasi-isomorphic to a direct sum of two nonzero groups. He proved
that every torsion-free group of finite rank is quasi-isomorphic to a direct
sum of strongly indecomposable summands and that these summands are
unique in the sense of the Krull-Schmidt theorem.
Definlitio~m.If G is an abelian group, then its endomorphism ring End(G) is the
set of all endomorphisms of G with composition as multiplication and point-
+
wise addition (if y, $ E End(G), then y $: g t-,y(g) + +(g)).
There is a remarkable theorem of A.L.S. Corner (1963). Let R 'be a
countable ring whose additive group is torsion-free and reduced; then there
exists a countable group G, which is also torsion-free and reduced, with
End(G) E R. In the proof of the Krull-Schmidt theorem, we saw a close
connection between decompositions of a group and endomorphisms, and
one can thus use Corner's theorem to produce strange examples of torsion-
free groups from pathological rings. For example, there are nonisomorphic
countable torsion-free groups, each isomorphic to a direct summand of the
other; there is a countable torsion-free group which has no indecomposable
direct summands.
10.45. (i) Show that if q :Q * Q is a (group) homomorphism, then there exists q E Q
with q(x) = qx for all x E Q.
(ii) If G and G' are subgroups of Q and if q: G -+ G' is a homomorphism, then
there is q E Q with q ( x ) = qx for all x E G. (Hitzt. Use the injective property
of 69.)
10.46. If G is torsion-free of rank 1 and type z, prove that End(G) is a subring of Q
and find its type.
10.47. If R and S are subrings of Q, then R r S as rings if and only if R s S as abelian
groups (by definition, both R and S contain 1). Conclude that there are un-
countably many nonisomorphic subrings of Q.
10.48. Give an example of nonisomorphic torsion-free groups of rank 1 having iso-
morphic endomorphism rings.
10.49. Let A denote the dyadic rationals and let B denote the triadic rationals:
B = (q E Q: q = 0/3~,a E Zand k 0).
Let G be the subgroup of Q @ Q generated by
((a, 0): a E A) u ((0, b): b E B ) v (($, $)).
Prove that G is an indecomposable group of rank 2.
10.50. (i) Use Theorem 10.45 to show that p(A) = dim(V), where V is a vector space
over Q of smallest dimension containing AItA.
(ii) If 0 -,A -,B -, C -,0 is an exact sequence of abelian groups, prove that
P(B) = P(A) +
Character Groups
In Chapter 1, we raised the twin questions of describing groups and of de-
scribing homomorphisms, and we now focus on the latter.

DeG'aanition. Let d denote the class of all abelian groups. A function T: d -, d

is a covarianf (additive)fiunctor if, for every homomorphism q : A B, there -+

10.42. If G is torsion-free of rank 1 and x E G is nonzero, then G/(x) is torsion. is a homomorphism T(q): T(A) -+ T(B)such that:
Describe G/(x) in terms of the height sequence of x. (Hint.G/(x) IQ/Z.)
ti) T t L ) = Iq-4);
10.43. If A and B are subgroups of Q, show that there is an exact sequence 0 -, (ii) if 0: B -, C, then T(9q) = T(e)T(cp);and
AnB+A@B--+A+B-+O. (iii) if $: A -, B, then T(y + $1 = T(q) + T($).
10.44. If G is a subring of Q, then G is also a torsion-free group of rank 1. Show that
the height sequence of 1 consists of 0's and a's. EXAMPLE
10.12. If G and A are abelian groups, then
Non1(G, A) = (homomorphisms q : G -+ A);
In his disseriation submitted in 1914 (which was not well lcnown because of World War I),
F.W. Levi defined the characteristic of an element in a torsion-free group, classified the sub- Hom(G, A) is an abelian group under pointwise addition: if cp, E Hom(G, A), +
groups of Q, introduced quasi-isomorphism, and gave the first examples of torsion-free groups + +
then cp $: g I-+ q(g) $(g). (If A = G, then Hom(G, G) is just the additive
having different direct sum decompositions into indecomposables. group of the endomorphism ring End(G).)
 10. Abelian Croups Character Groups

If G is a fixed group, define T: d + d by implies exactness of

T(A) = Hom(G, A); 0 -+ S(C) 2S(B) %S(A).

if cp: A 4 By define T(q): Hom(G, A) -+ Hom(G, B) by a I-+qa, where a E There are also right exact functors in Homological Algebra.
Hom(G, A). The reader should verify that T is a covariant functor.
EXAMPLE 10.13. The iskentidy fidmctor J: d --+ d , defined by J(A) = A and Theorem 110.49. If G is a group, then S = Mom( , G) and T = Hom(G, ) are
J ( q ) = cp, is a covariant functor. left exact functors.

EXAMPLE 10.14. The torsion subgroup defines a covariant functor t: d --+ d. HPk'6soJ We shall prove that S is left exact; the reader may prove that T is left
Define t(A) to be the torsion subgroup tA, and if cp: A -+ B, then define exact.
t(q) = cpltA.
In a similar way, the maximal divisible subgroup defines a covariant
functor d: d' -+ d. 0 -
We must show exactness of
Hom(C, G) 3Hom(B, G) 3Hom(A, G);
that is, we must show that S(P) is injective, im S(P) 5 ker S(a), and
Definition. A function S: d 4 d is a contr'davcgriant (additive) Janctov if, for ker S(a) < irn S(P).
every homomorphism q : A 4 Bythere is a homomorphism S(q): S(B) 4 S(A)
such that: (i) ker S(Q) = 0.
Iff: C -+ G and S(P)f = fp = 0, then f annihilates im P; as Q is surjective,
(i) s ( l ~ )= ~s(A); i m p = Cand f =0.
(ii) if 8: B + C, then S(0q) = S(q)S(B); and (ii) im S(Q) I ker S(a).
(iii) if t,b: A + Bythen S(q + $) = S(q) + S($). Iff: C 4 G , then S(a)S(/?)f = fpa = 0 because pa = 0.
(iii) ker S(a) < im S ( P ) .
Note that contravariant functors reverse the direction of arrows. Suppose that g: B 4 G and S(a) = ga = 0, so that g annihilates im a. De-
fine g,: C -+ G by g,(c) = g(b), where P(b) = c (/I is surjective). Now g, is
EXAMPLE
10.15. If G is a fixed group, define S: at' -+ d by well defined: if j?(bf) = c, then b - b' E ker P = im a; that is, b - b' = aa, and
S(A) = Hom(A, G); so g(b - b') = ga(a) = 0. But S(P)g, = g,Q = g, for if b~ B and Q(b) = c,
theng+$(b)=g,(c)=g(b).
if q: A + B y define S(q): Hom(B, G) 4 Hom(A, G) by P H P ~ ,where j3 E
Hom(B, G). The reader can verify that S is a contravariant functor. Here is the answer to the question when "+ O" occurs at the right end of
the functored sequence.
To see how functors behave, one must recast definitions in a form recog-
nizable by them. For example, instead of saying that an isomorphism is a Theorem 10.50. A group G is free abelian if and only if9for every exact se-
homomorphism q : A --+ B that is an injection and a surjection, one should quence 0 --+ A 5 B +
B
C + 0, there is an exact sequence
say instead that it is a homomorphism for which there exists 8: B 4 A such
that
840 = 1, and q 8 = 1,.
ProoJ Assume that G is free abelian. To prove that T(P) is surjective, we
We can now see that if cp is an isomorphism and T is a functor (contra or co), must show that if g E Hom(G, C), there is f E Hom(G, B) with T(P)f = g;
then T(q) is also an isomorphism: just apply T to the pair of equations above. that is, Qf = g. Let us draw a diagram.

Definition. A covariant functor T is left exact if exactness of (t): 04A 5

-
P
B -+ C implies exactness of
0 T(A) -% T(B) 2T(C);
P
a contravariant functor S is lefr exact if exactness of (x;*): A B + C -+ 0
338 10. Abelian Groups Character Groups
339

The projective property of G (Theorem 10.15) gives the existence off (indeed, Theorem 10.54. If m E 8 and 01,: A A is multiplicatiorl by my then
--+

this is how the projective property was born). T(n1,): T(A) -+ T(A) is also multiplication by m.
Conversely, a similar argument shows that a group G for which T(P)
is always surjective must have the projective property. But Exercise 10.10 ProoJ By Exercise 10.51 below, the result is true when m = 0. If m > 0, then
shows that such a group must be free abelian.
T(mA)= T(lA + ..- + I,)
Theorem 10.511. A group G is divisible if and only if, for every exact sequence
= T(1,) + ... + T(1,)
0 -+ A -% B -% C -+ 0, there is an exact sequence
= TO, + . + TO,
*

-
-~T(A).
ProoJ Use the injective property and Exercise 10.22.
The reader may easily see that T(- 1,) = - lTcA,,
so that the result is true for
Theorem 10.52. Given a group G and a family of groups {A,: j E J), then all m E Z. El

Hom
(jeJ
A,, G z
) n jeJ
Hom(Aj, G). EXAMPLE 10.16. For every group G, Hom(G, Q) is torsion-free divisible, and
hence it is a vector space over Q.
Proof. For each jOE J , let i,, ~t CjsJA, be the inclusion. Define

direction by (f;)
n
z G) -+ Hom(A,, G) by f r (fi,). Define a map $ in the reverse
8: ~ o m (A,,
rf, where f is the unique map 1
A, -+ G for which fi, =
By Exercise 10.31, H = Hom(G, Q) is torsion-free divisible if and only if
rn, is an automorphism for all m # 0. Thus, the result follows from Theorem
10.54.
4 for all j (Theorem 10.10). The reader may check that 8 and $ are The same argument shows that Hom(Q, G) is also a vector space over Q.
inverses. lB

If the index set J is finite, then the functor Hom( , G) sends the finite direct EXAMPLE
10.17. Hom(Z, G) 2 G for every group G.
sum x7=, A, into the direct sumx;=, A, Hom(Aj, G); that is,
It is easy to see that f H f (1) is an isomorphism.
Hom
(
x )z 1
j
A,, G
jll
Hom(A,, G).
EXAMPLE
10.18. For every group G,
Theorem 10.53. Given a group G and a family of groups {A,: j E J], then Nom(Zn, G) E G[n] = ( g G:~ ng = 0).

Hom G,
( n )z njeJ
A,
jcJ
Hom(G, A,). Apply the (contravariant) functor S = Hom( , G) to the exact sequence
0 -+ Z Z -+ Z, -+ 0, where the first map is multiplication by n, to obtain the
n
Proof. For each j, E J, let pjo: A, -+ A, be the projection onto the joth
coordinate. An argument similar to that of Theorem 10.52, using Theorem
exact sequence -

n
10.11, shows that the map 8: Hom(G, A,) 4 Hom(G, A,), defined by
f w (p, f ), is an isomorphism. 'El
fl Thus Hom(Z,, G) E ker S(n). But there is a commutative diagram

If the index set J is finite, then the functor Hom(G, ) sends the (finite)direct
sum x;=, A, into the direction sum z7=,
Hom(G, A,); that is,

Hom G,
( C A,
j:l ) g 1 Hom(G, A,).
j:,

If m is an integer and A is a group, let m,: A -+ A be multiplication by m; where the downward arrows are the isomorphisms of Example 10.17. It fol-
that is, m,: a Hma. Note that m, = 1, + - - .+ 1, (m addends) if m > 0. lows easily that ker S(n) 2 ker n = GCn].
340 10. Abelian Groups
I
Defini~on.If G is a group, its character group G* is
G* = Hom(G, Q/Z).
The next lemma shows, when G is finite, that this definition coincides with
our earlier definition of character group in Chapter 7.
Lemma 10.55. If G is finite, then G* z Hom(G, C ").
Proof, By Exercise 10.28(ii), the torsion subgroup t(Cx)r Q/Z. Therefore,
C x z (Q/Z) D, where D is torsion-free divisible. Hence, Hom(G, C ") r
Hom(G, Q/Z) @ Hom(G, D) r Hom(G, Q/Z), for Hom(G, D) = 0 because G
is finite. Is
Remark. There is another common definition of character group: = c Theorem 10.59. For every group G, the evaluation map E: G + G** is an injec-
c
Hom(G, T). Since T 2 C ", by Exercise 10.28(i), we see that r G*. The
character groups 6 arise in Pontrjagin duality, where G is assumed to be a
locally compact abelian topological group and is the group of all continu-
ous homomorphisms (when G is a discrete group, then every homomorphism
is continuous).
The following properties of character groups were used in our discussion
of the Schur multiplier in Chapter 7.
Theorem 10.56. If G is finite, then G E G*.
Proof, If G r Z, then Example 10.18 gives G* z (Q/Z)[n] r Z, (Exercise
1
10.21(ii)), as desired. By the basis theorem, G = Ci, where Ci is finite cyclic,
1
so that Theorem 10.52 gives G* z (Ci)* z Ci = G. H
Let us now solve Exercise 6.13.
Theorem 10.57. If G is a finite abelian group and S IG, then G contains a
subgroup isomorphic to G/S.
ProoJ There is an exact sequence 0 --+ S -+ G 4 G/S 4 0. Since Q/Z is divisi-
ble, Theorem 10.51 gives an exact sequence
Hence G/S r (G/S)* is isomorphic to a subgroup of G* r G. ill
Theorem 10.58. Let G be an abelian group, let a E G, and let S be a subgroup
of G with a gC: S. If D is a divisible group with Q/Z ID, then there exists
cp: G D with cp(S) = 0 and cp(a) f: 0.
-+
Remark. The important cases are D = Q/Z and D = T z C x .
Character Groups 34 1
+
Proof. If [a] = a S E G/S, then there is a homomorphism $: ([a]) 4 D
with $([a]) # 0: if [a] has finite order n, define $([a]) = l/n Z E Q/Z I D; +
+
if [a] has infinite order, define $([a]) = 1 Z. By the injective property of D,
$ extends to a homomorphism Y: G/S -+ D. If cp: G 4 D is defined to be Yv,
where v : G -,G/S is the natural map, then cp(S) = 0 and q(a) = Y([a]) #
0. Ia
If G is finite, then we know that G r G* 2 (G*)*; let us denote the latter
group, the "double dual," by G**. We now exhibit a specific isomorphism.
Definition. Let G be a group. For each x E G, define a homomorphism
Ex: G* Q/Z by Ex(cp)= q(x); thus, Ex E G**. The evaluation map
-+
E: G G*" is defined by x I--,Ex.
-+
tion; fi G is finite, then E is an isomorphism.
Proof. If x E ker E, then Ex(q)= cp(x) = 0 for all q E G*; by Theorem 10.58,
x = 0, and so E is an injection. If G is finite, then G** r G, by Theorem 10.56,
so that / G** 1 = I GI and E is an isomorphism.
EXERCISES
10.51. Show that if T is an (additive) functor, then T(0)= 0, where 0 denotes either
the trivial group 0 or the (constant) map which sends every element into 0.
10.52. Prove that Hom(A, 3)= 0 in the following cases:
(i) A is torsion and B is torsion-free;
(ii) A is divisible and 3 is reduced;
(iii) A is p-primary and Ed is q-primary, where p i: q.
10.53. (i) Show that A is reduced if and only if Hom(Q, A) = 0.
(ii) Use Theorem 10.50 to show that if 0 4 A -+ B 4 C -+ 0 is exact and A and
C are reduced, then B is reduced.
10.54. If G is finite and S IG, define
SL = {f E G*: f(s) =0 for all s E S).
(i) Show that SL is a subgroup of G* and that SL r (G/S)*.
(ii) If G is a finite group of order n and if k is a divisor of n, then G has the same
number of subgroups of order n as of index n. (Hint: The function St+ SL
is a bijection.)
10.55. Here are examples related to Theorems 10.52 and 10.53. Let P denote the set
of all primes.
(i) Prove that Hom(n,,,Z,, Q) f 0. Conclude that Hom(nZ,, Q) is iso-
morphic to neither Hom(Z,, Q) nor Hom(Z,, Q).
 10. Abelian Groups

(ii) If G = CpSp
isomorphic.
Z,, then Hom(G, z,,,
7,) and x
Hom(G, $) are not ' CHAPTER 11
(iii) Hom(Z, xPEp n
7,) and Hom(7, 7,) are not isomorphic.
Free Groups and Free Products
10.56. Let f:A 4 B be a homomorphism. Show that f *: B* -,A* is a surjection
(where A* = Hom(A, Q/Z) and f *: q H qf for every q E B*) if and only iff is
an injection; show that f *: B* 4 A* is an injection i f and only i f f is a sur-
jection. Conclude that f * is an isomorphism if and only if f is an isomorphism.
10.57. Consider the commutative diagram of not necessarily abelian groups in which
the rows are exact sequences (thus, the kernels are assumed to be normal
subgroups):

Show that if P is an isomorphism, then a is an isomorphism.

Generators and Relations

The notion of generators and relations can be extended from abelian groups
to arbitrary groups once we have a nonabelian analogue of free abelian
roups. We use the property appearing in Theorem 10.11 as our starting
oint.
efinition. If X is a subset of a group F, then F is a free group with basis X
for every group G and every function f : X -+ G, there exists a unique
omomorphisrn q:F -,C extending f

We shall see later that X must generate F.

Observe that a basis in a free group behaves precisely as does a basis
B = (v, , ...,v,) of a finite-dimensionalvector space V. The theorem of linear
algebra showing that matrices correspond to linear transformations rests on
the fact that if W is any vector space and w, ,...,w, E W, then there exists a
unique linear transformation T: V -,W with T(vi) = wi for all i.
The following construction will be used in proving that free groups exist.
Let X be a set and let X-' be a set, disjoint from X, for which there is a
bijection X + X-l, which we denote by XI-x-l. Let X' be a singleton set
disjoint from X u X-' whose only element is denoted by 1. If x E X, then x1
may denote x and x0 may denote 1.
344 11. Free Groups and Free Products
Definition. A word on X is a sequence w = (a,, a,, ...), where
a, E X v X-' v (1) for all i, such that all ai = 1 from some point on; that is,
there is an integer n 2 0 with ai = 1 for all i > n. In particular, the constant
sequence
is a word, called the empty word, and it is also denoted by 1.
Since words contain only a finite number of letters before they become
constant, we use the more suggestive notation for nonempty words:
+
where xi E X, E, = + 1, - 1, or 0, and E, = 1. Observe that this spelling of a
word is unique: two sequences (a,) and (bi)are equal if and only if ai = b, for
all i. The length of the empty word is defined to be 0; the length of w =
xi1x?.. .x> is defined to be n.
Definition. If w = x", . ..x? is a word, then its inverse is the word w-l =
xnEn...xTEf.
Definition.A word w on X is reduced if either w is empty or w = x",x?. ..x,".,
+
where all xi E X, all E, = 1, and x and x-I are never adjacent.
The empty word is reduced, and the inverse of a reduced word is reduced.
Definition. A sabword of w = x", x? ...x,". is either the empty word or a word
oftheformv=xFi ...x?,wherel < i < j < n .
Thus, v is a subword of w if there are (possibly empty) subwords w' and w"
with w = w'vwN.A nonempty word w is reduced if and only if it contains no
subwords of the form xEx-'or xO.
There is a multiplication of words: if w = x",x? . . . x,". and u = y f l yi2.. .y>,
then wu = x",x?. . .x$yflyi2.. .yim. This multiplication does not define a
product on the set of all reduced words on X because wu need not be reduced
(even when both w and u are). One can define a new multiplication of reduced
words w and u as the reduced word obtained from wu after cancellations.
More precisely, there is a (possibly empty) subword v of w with w = w'v
such that v-' is a subword of u with u = v-lu" and such that wfu" is reduced.
Define a product of reduced words, calledjuxtaposition, by
Theorem 11.1. Given a set X, there exists a JFee group F with basis X.
ProoJ Let F be the set of all the reduced words on X. One can show that F
is a group under juxtaposition, but verifying associativity involves tedious
case analyses. Instead, we use the van deu Waerden trick (1945).
Generators and Relations 345
For each x E X, consider the functions 1x1: F -+ F and Ix-ll: F -+ F, defined
as follows: for E = 1,
Since lxEl0 Ix-'/ and IX-'~ 0 oxE/are both equal to the identity 1,: F -,F, it
follows that IxElis a permutation of F with inverse Ix-']. Let SF be the sym-
metric group on F, and let 9 be the subgroup of SF generated by [XI =
(1x1: x E X). We claim that F is a free group with basis [XI. Note that there
is a bijection (: [XI -+ X, namely, I x 1 I-+ x.
An arbitrary element g E F (other than the identity) has a factorization
(*I g = IxE,lI 0 jxPJ 0 - - - 0 /xnE,I,
where ci = f1 and IxEland IX-'~ are never adjacent (or we can cancel). Such
a factorization of g is unique, for g(1) = x;'x?. ..xnE,, and we have already
noted that the spelling of a (reduced) word is unique.
To see that F is free with basis [XI, assume that G is a group and that
f : [XI -, G is a function. Since the factorization (*) is unique, the function
q: 9-,G, given by q(lx",\ 0 lx?l 0 -.-o 1x21) = f(lx?1l)f(lx?I)... f(lx?l), is
well defined and extends f. Since [XI generates P, it suffices to show that q
is a homomorphism, for uniqueness of cp would then follow from the fact that
two homomorphisms agreeing on a generating set must be equal.
Let w and u be reduced words on [XI. It is obvious that q(w 0 u) =
w)cp(u) whenever the word wu (obtained from w 0 u by deleting vertical
rs) is reduced. Write w = w' 0 v and u = v-' 0 u", as in the defini-
tion of juxtaposition. Now q(w) = q(wl)q(v) and q(u) = q(v-')q(uU) =
u)-I q(uN)(because w' 0 v and v-I 0 u" are reduced). Therefore, q(w)q(u) =
w')q(v)q(v)-'q(u") = q(wf)q(u").On the other hand, q(w 0 u) = q(wr 0 u")
q(w')q(uU)(because w' 0 u" is reduced), and so q is a homomorphism.
We have shown that 5" is a free group with basis [ X I . Since 5" -+ F,
defined by IxB1I 0 Ix?l 0 0 I x ~ n I ~ x ~...
- - a l xx~,isabijection
? with[([X]) =
(([XI) = X, Exercise 1-44shows that we may regard F as a group isomor-
phic to F ; thus, F is a free group with basis X (moreover, X generates F
because [XI generates 9 ) .
Corollary 11.2. Every group G is a quotient ofa free group.
Proof. Construct a set X = (x,: g E G) so that f: x, t-+ g is a bijection X G.
-+
If F is free with basis X, then there is a homomorghism q : F -+ G extending
f,and q is a surjection because f is. Therefore, G r F/ker q. ill
Definition. Let X be a set and let A be a family of words on X. A group G has
generators X and relations A if G E FIR, where F is the free group with basis
X and R is the normal subgroup of F generated by A. The ordered pair (XIA)
is called a presentation of G.
346 11. Free Groups and Free Products Generators and Relations 347

A relation1 r E A is often written as r = 1 to convey its significance in the holds, for each element in G has a factorization x ' y j ~with 0 5 i < n and
quotient group G being presented. 0 < j < 2. Thus, 1 GI = 212,and we are now entitled to write G z D,,.
There are two reasons forcing us to define R as the normal subgroup of F A description of a group by generators and relations is flawed in that the
generated by A: if r E A and w E F, then r = 1 in G implies wrw-' = 1 in G; order of the presented group is difficult to determine. This is not a minor
we wish to form a quotient group. difficulty, for we shall see in the next chapter that it is even an unsolvable
problem (in the logicians' precise sense) to determine, from an arbitrary
EXAMPLE
11.1. G = Z6 has generator x and relation x6 = 1. presentation, the order of the presented group. Indeed, it is an unsolvable
problem to determine whether a presentation defines a group of order 1. The
reader should also see the next section on coset enumeration.
A free group F = (x) on one generator is infinite cyclic, and (x)/(x6)
Z,. A presentation of G is (x 1x6).
Let us continue the list of examples.
EXAMPLE
11.2. Another presentation of G = Z6 is EXAMPLE
11.4. The group of quaternions has presentations
Z6 = ( X , ~ /=X1,y2
~ = 1, XYX-'Y-' = 1). Q = (a, bla4 = 1, b2 = a2, bab-l = a-l)
and
When we described a presentation of Z6 as an abelian group in Example
Q = (x, y lxyx = y, x2 = y2).
10.2 (i.e., when we viewed Z, as a quotient of a free abelian group), the only
relations were x3 and y2. Now we must also have the commutator as a
In each case, an argument is needed to show that the presented group has
relation to force the images of x and y to commute in FIR.
order 8.
EXAMPLE
11.3. The dihedral group D2, has a presentation 11.5. Given positive integers 1, m, and n, define
EXAMPLE
D2,,= (x, ylxn = 1, y2 = 1, yxy = x-l). P(1, rn, n) = (s, tls' = t m= (st)" = 1).

It is acceptable to write a relation as yxy = x-l instead of xyxy = 1. In Example 11.3 shows that P(n, 2,2) = D,, and, using Exercise 3.52, one can
particular, compare the presentation of D6 with that of Z6 in Example 11.2. show that P(2, 3,3) r A,, P(2,3,4) 2 S,, and P(2, 3, 5) r A,. These groups
We have passed over a point needing more discussion. By definition, D,, is are called polyhedral groups, and they are finite only in the cases just listed
a group of order 2n having generators S and T satisfying the given relations. (see Coxeter-Moser).
If G = FIR, where F is the free group with basis (x, y) and R is the normal EXAIMPEE
11.4. The braid grd~apBm has the presentation
subgroup generated by (x*, y", xyxy), does G have order 2n? We have seen
various concrete versions of D,,; for example, Theorem 3.3 1 displays it as the
symmetry group of a regular n-gon. The definition of free group gives a Braid groups were introduced by E. Artin (1925) and are related to knot
surjective homomorphism q: F -,D2, with q(x) = S and q ( y ) = T, More- theory.
over, R < ker q , because S and T satisfy the relations, so that the third
EXAMPLE11.7. A free abelian group G with basis X has presentation
isomorphism theorem gives a surjection FIR -+F/ker 9;that is, there is a
surjection2 G = FIR -+ D,,. Hence, [GI >_ 2n. The reverse inequality also G = (Xlxyx-ly-l = 1 for all x, y E X);
a free group F with basis X has presentation
Many authors use the words "relation" and "relator" interchangeably.
W. von Dyck (1882) invented free groups and used them to give the first precise definition of F = (XI@).
presentations. The version of the third isomorphism theorem used here is often called von Dyek's
Theorem: Let G have a presentation Having proved that free groups exist, let us now consider their uniqueness;
G = (xl, ..., Xnlrj(x1,..., x,),j E J ) that is, when are two free groups isomorphic.
so that G = FIR, where F is the free group with basis (x,, ...,x,) and R is the normal subgroup
generated by the 5. If H is a group with H = (y,, ..., y,) and if q(y,, ...,y,) = 1 for all j, then Lemma 11.3. If F is a free group with basis X, then F/Ffis a JFee abelian group
there is a surjective homomorphism G -+ H with X ~ Hyi for all i. with basisir, = (xFf: x E X).
348 11. Free Groups and Free Products Semigroup Interlude

Proof. Assume that A is an abelian group and that f : X, -+ A is a function. homomorphism y: F -+ B making the diagram below commute (i.e., fly = a):
Define f,: X -+ A by x I-+ f(xF'). Since F is free with basis X, there is a
F
homomorphism q: F -+ A extending f,. But F' < ker q, because A is abelian, // I
so that there is a homomorphism @: F/F'-+A, defined by wF't+q(w),
extending f.
We claim that the extension 4 is unique. Suppose that 0: F/F' -+ A and
0(xFf)= f(xF'). If v: F -+ F/F' is the natural map, then 8v: F -+ A is a
homomorphism with E)v(x)= 8(xFf)= f (xFf)= q(x) for all x E X. Since X is Proof. The proof is identical to that given for free abelian groups in Theorem
a basis of F, Ov = q = q v ; since v is surjective, 8 = 4. Therefore, F/F' is free 10.15. H
abelian with basis X,.
We shall see in Exercise 11.46 below that the converse of Theorem 11.6 is
Theorem 11.4. Let F and G be free groups with bases X and Y, respectively. also true: a group G is free if and only if it has the projective property.
Then F E G if and only if 1x1= I YI.

Proof. If q: F -+ G is an isomorphism, then F/F' g GIG'. By the lemma, F/F' Semigroup Interlude
is free abelian with basis X, = (xF': x E X). As IX , I = I X I, it follows that
I XI = rank(F/Ff).Similarly, I YI = rank(G/Gf),and so IX I = I YI, by Theorem We are now going to construct free semigroups; the formal definition is no
10.14. surprise.
If I XI = I YI, there is a bijection f : X Y which, upon composing with the
-+

inclusion Y c,G, may be regarded as a function X -+ G. Since F is free with Definition. If X is a subset of a semigroup Z, then C is afree semigroup with
basis X, there is a unique homomorphism q: F -+ G extending f. Similarly, basis X if, for every semigroup S and every function f : X -+ S, there exists a
there is a unique homomorphism $: G -+ F extending f -': Y -,X. The com- unique homomorphism cp: C -+ S extending f.
posite $cp: F 4 F is a homomorphism which fixes X pointwise; that is, $40
extends the inclusion function I : X + F. But the identity 1, also extends I ,
and so uniqueness of extension gives 11/50 = IF. Similarly, q11/ = I,, so that
q : F -+ G is an isomorphism. H

DeGfa~aitianm.The rank of a free group P is the number of elements in a basis

of F.
Theorem 11.4 says that rank(F) does not depend on the choice of basis
of F.
Corollary 11.5. If F is free with basis X, then F is generated by X.
Proof. Choose a set Y with I Y1 = I X I and a bijection f : Y -+ X. The free
group G with basis Y constructed in Theorem 11.1 (as the set of all reduced
words on Y) is generated by Y. As in the proof of Theorem 11.4, the
homomorphism t,b: G --+ F extending f is an isomorphism, so that G = (Y)
implies F = ($(Y)) = <f(Y)) = (X). El
Theorem 11.6 (Projective Property). Let P: B -+ C be a surjective homomor-
phism. If F is free and if a: F -+ C is a homomorphism, then there exists a
Defirmitia~rm.A word w on X is positive if either w = f or w = xitx?. ..xin,
where all exponents E~ > 0.
The set ): of all positive words on X is a free semigroup with basis X (the
product of positive words is positive and, with no cancellation possible, it is
easy to prove that multiplication is associative). It follows that every semi-
group is a homomorphic image of a free semigroup. Before defining presenta-
tions of semigroups, however, we first define quotients.
Definition. A congruence on a semigroup S is an equivalence relation = on S
such that
a r a' and b = b' imply ab r arb'.
If -. is a congruence on a semigroup S, then the quotient semigroup is the
set of all equivalence classes, denoted by S/ r ,with the operation
Cal Cbl = Cab17
350 11. Free Groups and Free Products
where [a] denotes the equivalence class of a E S (this operation is well defined
because r is a congruence).
There are two general constructions of congruences. The first arises from a
homomorphism cp: S -+ T of semigroups; define a = b if cp(a) = cp(b). This
congruence is called ker 9,and it is straightforward to prove the first isomor-
phism theorem:
S/ker cp r im cp
(if S and T are groups and K = (s E S: ~ ( s= ) 11, then ker cp is the equiva-
lence relation on S whose equivalence classes are the cosets of K). Here is a
second construction. As any relation on S, a congruence is a subset of S x S.
It is easy to see that any intersection of congruences is itself a congruence.
Since S x S is a congruence, one may thus define the congruence generated
by any subset E of S x S as-theintersection of all the congruences containing
E. If C is the free semigroup with basis X and if (w, = ui: i E I} is a family of
equations, where w,, u, E X, then define = to be the congruence generated by
((w,, u,): i E I} c X x C. The quotient semigroup X/= is said to have the
presentation
(XI wi = ui for all i E I).
11.1. Use presentations to prove the existence of the nonabelian groups of order p3,
where p is prime. (See Exercise 4.32.)
11.2. Prove that a free group of rank 2 2 is a centerless torsion-free group.
11.3. Prove that the group G = (x, ylxm,yn)is infinite when m,n 2 2.
11.4 (Baer). Prove that a group E has the injective property if and only if E = 1.
(Hint. D.L. Johnson). Let A be free with basis (x, y) and let B be the semidirect
product B = A x (z), where z is an involution acting by zxz = y and zyz = x.)
11.5. Let X be the disjoint union X = Y u Z. If F is free with basis X and N is the
normal subgroup generated by Y, then FIN is free with basis (zN: z E 2).
11.6. Show that a free group F of rank 2 2 has an automorphism cp with cp(cp(w))=
w for all w E F and with no fixed points (cp(w) = w implies w = 1). (Compare
Exercise 1.50.)
11.7. If H a G and G/H is free, then G is a semidirect product of H by GIN. (Hint.
Corollary 10.16 and Lemma 7.20.)
11.8. Let G be a group, let {ti: i E I) c G, and let S = (ti: i E I) 5 G. If there is a
homomorphism 9:G 4 F (where F is the free group with basis X = {xi: i E I))
with cp(ti)= xi for all i, then S is a free group with basis (ti: i E I).
11.9. The binary tetrahedral group B is the group having the presentation
B = (r, s, tlr2 = s3 = t3 = rst).
(i) Prove that rst E Z(B) and that B/(rst) r A, (the tetrahedral group).
(ii) Prove that B has order 24.
(iii) Prove that B has no subgroup of order 12.
Coset Enumeration
11.10. The dicyclic group DC, is the group having the presentation
DC, = (r, S,tlr2 = s2 = tn = rst).
(i) If n = 2m-2,then DC, r Q,, the generalized quaternion group (see Exercise
4.40).
(ii) Show that DC, has order 4n.
11.11. Show that (o,o, ...c~,)"+'E Z(B,,,), where Bmis the braid group (see Example
11.6). It is known that Z(B,) is the infinite cyclic group generated by this
element.
11.12. (i) Show that a free semigroup with a basis having at least two elements is not
commutative.
(ii) Show that a subsemigroupof a free semigroup need not be free. (Hint. Find
an appropriate subsemigroup of the multiplicative semigroup of positive
integers.)
Coset Enumeration
The method of coset enumeration, distilled by Todd and Coxeter (1936) from
earlier particular cases, is a mechanical way to find the order of a given group
from a presentation. It does not always work (nor can any such algorithm
always work, as we shall see in the next chapter), but it does work whenever
the presented group is finite. The method rests on the following elementary
lemma.
Lemma 111.7. Let G be a finite group, X a set of generators of G, H IG
a subgroup, and Hw,, . .., Hw,, some distinct cosets of H. If U:=, Hw, is
closed under right multiplication by every a E X u X-', then G = Hw,, U;=,
[G : PI] = n, and 161= nlHI.
PmoJ If Y is any nonenpty subset of G with Yu c Y for all cs E X u X-',
then Y = G (because X generates G and w E Y for every word w on X). In
particular, G = U:=, Hw,, so that every coset of H must appear as Hwi for
some i; that is, [G : HI = n. Ed
We illustrate the method in a specific case before describing it in general.
Let G be the group having the presentation
G = (s, tls3 = t2 = 1, tst = s2).
Write each of the relations as a word with all exponents IfI 1:
sss; tt; tsts-'s-l.
For each of these relation words, begin making a relation table by putting a
vertical line under each of its letters.
 I
354 11. Free Groups and Free Products
but the auxiliary table already contains
This is an instance of coset collapse; delete row 7 and replace all other occur-
rences of 7 by the smaller number 6, including the entries in the auxiliary
table. Continuing this procedure ultimately leads to the completed tables
The procedure now stops because all the relation tables are complete.
According to the next theorem, the conclusion is that the presented group G
has order 6 (of course, G r S,).
Theorem 11.8 (Coset Enumeration). Let G have a presentation with a finite
number of generators and relations. Set up one table for each relation as above,
add new integer entries and enlarge the auxiliary table as above whenever
possible, and delete any larger numbers involved in coset collapse. If the proce-
dure ends with all relation tables complete and having n rows, then the presented
group G has order n.
Proo$. Let 1 denote the identity element of G, and assume that the other
integers i in the tables, where 1 < i 5 n, denote other elements of G. The entry
in any relation table is interpreted as the equation ia =j in G. This explains
the twin entries in the auxiliary table: if r'a = j, then ja-I = i. The construction
of the relation tables is a naming of elements of G. If there is a blank square
to the right of i, with line labeled a between, then j is the element ia; if the
blank square is to the left, then j is the element ia-I. Coset collapse occurs
when ia =j and ia = k, in which case j = k.
Let Y be the set of elements in G that have been denoted by some i with
1 5 i 5 n. That all the tables are complete says that right multiplication by
any a E X u X-' produces only elements of Y. Therefore, Lemma 11.7 applies
to Y (with H taken to be the trivial subgroup), and so /GI = n. Rl
Coset Enumeration 355
Notice the hypothesis "If the procedure ends"; one does not know in ad-
vance whether the procedure will end.
There is a gereralization of the algorithm from which the name "coset
enumeration" arises. Consider the binary tetrahedral group (of order 24)
given in Exercise 11.9:
B = (r, s, tlr2 = s3 = t3 = rst).
First rewrite the presentation to display relations equal to 1:
B = (r, s, tlr-lst = r-2s3 = r-I s-It2 = 1).
One could use Theorem 11.8 to show that B has order 24, but tables with 24
rows are tedious to do. Instead, let us choose a subgroup H 5 G for which
generators are known. For example, we might choose H = ( s ) in this exam-
ple (cyclic subgroups are simplest). The idea is to use a slight variant of
Theorem 11.8 to enumerate the cosets of H in G (instead of the elements of G).
This is done as follows. In addition to relation tables, draw subgroup genera-
tor tables, one for each generator of H. For example, there are two such tables
if we choose H = (rst, s ) ; there is just one such table if we choose H = ( s ) .
New tables consist of one row, and they are called complete once all their
squares are filled without drawing any new rows under them. In our example,
there is just one subgroup generator table, and it is already complete.
In the general case, the rows of the subgroup generator tables are completed
first, giving pairs of entries to the auxiliary table (in our example, the entries
in the auxiliary table arising from the subgroup generator table are
S s-I
1I 1 and 1 I 1).
After completing these one-rowed tables, the relation tables are completed as
before. The numbers i now denote right cosets of H in 6, with 1 denoting H.
The entry
a
i lj
in a table means that if i = Hw, then j = Hwa. When all the tables are com-
pleted, Lemma 11.7 applies to calculate [G : HI,and hence I G 1 is known if (HI
is. This version actually does enumerate the cosets of H.
In Exercise 11.13 below, the reader is asked to use coset enumeration to
show that the order of the binary tetrahedral group B is 24. One must com-
pute /HI;that is, one must compute the order of s (it is 6) and then see that
the relation tables are complete with 4 rows.
There are two unexpected consequences of coset enumeration. When H =
356 11. Free Groups and Free Products Coset Enumeration 357

1, the completed relation tables can be used to construct the regular represen- Definition. Let G be a group and let X be a set of generators of G. The
tation of G. For example, we saw above that the presentation of G = S,, Cayley graph r = T(G, X) is the directed graph with vertices the elements
G = (s, tls3 = t2 = 1, tst = s2), of G and with a directed edge from g to h if h = gx for some x E X.

has relation tables: If coset enumeration of a presentation (X 1 A) of a group G yields complete

S S S t t t s t s-I s-I relation tables, then one can record the information in these tables as the
Cayley graph T(G, X). For example, here is the Cayley graph of S, obtained
from the presentation above.

The first column of the first table displays the values of right multiplication
by s (as a permutation of { 1, ...,6)), and the first column of the second table
does this for t. Right multiplication by s and t are:
s ++ (1 2 3)(4 5 6) and t ++ (1 4)(2 6)(3 5), 3 2
so that the right regular representation has R, = (1 3 2)(4 6 5) (because Figure 11.1
R,: i t-, is-') and R,= (1 4)(2 6)(3 5). More generally, when one enumer-
ates the cosets of a subgroup H of G, then one obtains the representation of
G on the cosets of H (the construction above differs from that of Theorem The Cayley graph of a group and a generating set is always defined, wheth-
3.14 only in giving the representation on the right cosets of H instead of on er or not coset enumeration can be completed. Notice that the Cayley graph
the left cosets as in that theorem). does depend on the choice of generating set. For example, a loop is an edge of
The information contained in completed relation tables can also be used to the form (u, u). If we take G itself as a generating set, then F(G, G) contains
draw a directed graph. the loop (1, I), while r(G,X) has no loops if 1 8 X. The Cayley graph is
the beginning of a rich and fruitful geometric way of viewing presentations
Defi'aanitioira.A da'recbedgraph T is a set V, called ~ertiws,together with a subset (see Burnside (1911), Dicks and Dunwoody (1989), Gersten (1987), k p d o n
E c V x V ; ordered pairs (u, v) E E are called directed edges. A directed and Schupp (1977), and Serre (1980)).
graph yields an associatedgvaph T':both I-' and IT' have the same vertices,
and u and v are called adjacent in T' if u fi v and either (u, u ) or (u, u) is a
directed edge in T.

One can picture a finite directed graph r by drawing V as points and 11.13. (i) In the presentation of the binary tetrahedral group B given above, show
that s has order 6 in B.
drawing an arrow from u to v if (u, v) E E. In contrast to graphs, which have (ii) Use coset enumeration relative to the subgroup H = (s) to compute the
at most one edge between any pair of vertices, a directed graph may have two order of B.
edges between a pair of vertices, one in each direction (given u, u E V, it may . - , cosets of H.
(iiil Find the remesentation of B on the (right)
\ I

happen that both (u, u) and (u, u) E E). However, even if both (u, v) and (v, u)
11.14. Describe the group G to isomorphism if G has the presentation
are directed edges in l?, there is only edge between them in the associated
graph IT'. (There is a notion of multigraph, directed or nondirected, which (q, r, S,tlrqrW'= q2, rtr-' = t2,s-lrs = r2, tst-' = s2,rt = tr).
allows many edges between a given pair of vertices, but we do not need 11.15. Let (XIA) be a presentation of a group G. Show that the Cayley graph T(G, X)
them here.) has no loops if and only if 1 $ X.
358 11. Free Groups and Free Products
Definition. The degree of a vertex v in a graph lr is the number of vertices
adjacent to it; the degree of a vertex v in a directed graph I'is its degree in the
associated graph T'.A graph or directed graph is regular of degree k if every
vertex has the same degree, namely, k.
11.16. If X is a finite generating set of a group G with 1 4 X, then the Cayley graph
T(G, X) is regular of degree 2 1x1.(Hint. If g E G and x E X, then (gx-', g) and
(g, gx) are directed edges.)
11.17. Draw the Cayley graph T(G, X) if G is a free abelian group of rank 2 and X is
a basis.
11.18. Draw the Cayley graph T(G,X) if G is a free group of rank 2 and X is a basis.
Presentations and the Schur Multiplier
The Schur multiplier M(Q) of a group Q is discussed in Chapter 7 (the reader
is advised to reread the appropriate section); it is related to presentations of
Q because of the following isomorphism.
Hopf's Formula. If Q r FIR is a finite3 group, where F is free, then
M(Q) r (R n F')/[F, R].
Remark. An "aspherical" topological space X has the property that its
homology groups are completely determined by its fundamental group
n,(X). Hopf (1942) proved that H,(X) r (R n F')/[F, R], where F is free and
FIR E n,(X). Schur (1907) proved that M(Q) r (R n F')/[F, R] when Q is
finite (i.e., Schur proved Hopf's formula in this case!). Comparison of Hopf's
formula to Schur's theorem led Eilelnberg and Mac Lane to their creation of
Cohomology of Groups; the homology group H,(X) of the aspherical space
X is the homology group H2(n,(%), Z) of the fundamental group nl(X).
When nl(X) is finite, H2(z1(X),Z)is isomorphic to the second cohomology
group H2(n,(X), C x )= M(n,(X)).
We will prove Hopf's formula for all finite groups Q, but we first consider
a special class of groups.
Definition. A group Q is perfect if Q = Q'.
Every simple group is perfect. The proofs of Theorems 8.13 and 8.23 show
that the groups SL(n, q) are perfect unless (n, q) = (2,2) or (2, 3).
Let us explain the finiteness hypothesis in Hopf's formula. In Chapter 7, we defined M(Q) as
the cohomology group H2(Q, C '). Nowadays, after defining homology groups of Q, one defines
M(Q) as the second homology group H2(Q,Z). There is always an isomorphism H2(Q,Z) r
(H2(Q, Cx))*, where * denotes character group. When Q is finite, the abelian group H ~ also S
finite, and hence it is isomorphic to its own character group, by Theorem 10.54.
Presentations and the Schur Multiplier
The definition of exact sequence
. . - + A $B-%c+...
I
(the image of each homomorphism is equal to the kernel of the next one)
makes sense if the groups are nonabelian. Of course, every image, being a
kernel of a homomorphism, must be a normal subgroup.
Lemma 11.9. Let v: U -, Q be a central extension of a group K = ker v by a
group Q. If Q is perfect, then U' is perfect and v / U':U' -+ Q is surjective.
Proof. Since v is surjective, v(U1)= Q'; as Q is perfect, v(U1)= Q, and v ]U' is
surjective. If u E U, there is thus u' E U' with v(uf)= v(u); hence, there is
z E K < Z(U) with u = u'z. To see that U' is perfect, it suffices to show that
U' < U". But if [u, v] is a generator of U', then there are u', u' E U' and
I
central elements z,, z, with [u, v] = [u'z,, v1z2] = [u', v'] E U". iB
Theorem 11.10. If Q is a perfect finite group and fi Q r FIR, where F is free,
then M(Q) E (R n F')/[F, R]. Moreover, F'/[F, R] is a cover of Q.
Proof. Since R a F, we have [F, R] IR; moreover, [F, R] a F. There is
thus an exact sequence
1 -+ R/[F, R] -+ FIEF, R] 4 FIR 4 1,
which is plainly a central extension. It is easily checked that (F/[F, R])' =
F'/[F, R]. Since Q = FIR is perfect, Lemma 11.9 gives an exact sequence 5
(with v' the restriction of v)
5: 1 -+( R n F')/[F, R] -+ F1/[F,RJ % F/R -+ 1
(for ( R nF1)/[F, R] = (R/[i;,R]) n (F'/[F, R]) = ker v'). Let us denote
(R n F')/[F, R] by K. As F'/[F, R] is perfect, by Lemma 11.9, we have
K 5 (F'/[F, R])'. Therefore, the transgression 6: K* -+ M ( Q ) is injective, by
Lemma 7.64. But Q is finite, by hypothesis, so that M(Q) is also finite, by
Theorem 7.60; hence K is finite and K r K*, by Theorem 10.56. To see that
6 is surjective, it suaces, by Lemma 7.63, to prove that the central extension
has the projective lifting property. Consider the diagram with exact rows
- - I' -
1,- Cx GL PGL 1.
Since F is free, it has the projective property (in the diagram in Theorem 11.6,
let a = zn); there exists a homomorphism a: F -+ GL making the diagram
commute; moreover, since the bottom extension is central, it is easy to see
 i
Presentations and the Schur Multiplier 361
360 11. Free Groups and Free Products

Theorem 11.11. If Q is a finite pe$ect group, then its cover U = F'/[F, R] is a

that [F, R] 5 ker o. There results a commutative diagram with exact rows universal central extension of Q.

Proof. Consider the portion of the proof of Theorem 11.10 showing the exis-
tence of a map F'/[F, R] -+ GL making the diagram commute; this portion
holds if one replaces the central extension 1 -t C x GL 4 PGL -+ 1 by any
-+

where of: u[F, R] H o(u) for all u E F. Since Q is perfect, we may replace the
top row by 6 and the downward map a' by its restriction i = dlF'/[F, R].
Thus, z can be lifted, 5 has the projective lifting property, and the injection
6: K* -+ M(Q) is also a surjection, as desired. Finally, F'/[F, R] is a cover of
Q, by Lemmas 7.63 and 7.64. W
A central extension U of K by Q has the projective lifting property if, for
every homomorphism T: Q -+ PGL, there is a homomorphism i : U -t GL
making the following diagram commute:
central extension and the map z by any map. In particular, one may replace
the central extension by 1 L -+ V Q -+ 1 and the map i by l p , where V is
-+ -+
a central extension of L by Q. Thus, it only remains to prove the uniqueness
of such a map.
Let v: U -, Q and p: V -+ Q be the given surjections. Suppose that a,
p: U -+ V are homomorphisms with pa = v = pP. If u E U, then pa(u) = pP(u),
so that a(u)/3(u)-' E ker p I Z ( V ) ; there is thus z E Z(V) with u(u) = p(u)z.
Similarly, if u' E U', there is z' E Z(V) with a(ul) = P(ur)z'. Therefore,
a(Cu9 u l l ) = Ca(u), a(ul)l = [P(u)z, P(uf)z'l = CP(4, P(uf>l = P(C% u'l).
Since U is perfect, by Lemma 11.9, it is generated by all commutators. There-
fore, a = p, as desired.

The converse of Theorem 11.11 is true: a finite group Q has a universal

1 -.C X - - 4
GL
1
PGL ------+ 1.
Of course, the bottom row is a central extension. Are there central extensions
central extension if and only if Q is perfect (see Milnor (1971), 55). More
generally, a possibly infinite group Q has a universal central extension if and
only if Q/Qf is free abelian (see Gruenberg (1970), p. 214).

of Q which have a projective lifting property with respect to other central
extensions?
Definition. A central extension U of Q is a gniversal central extension if, for
every central extension V of Q, there is a unique homomorphism ar making
the following diagram commute:
Coroalary 11.12. Every finite perfect group Q has a unique cover U which is
itself a finite perfect group.
HProoJ By Theorem 11.11, the cover U = Fr/[F, R] is a universal central
extension of Q; if a central extension V of it by Q is a cover, then there is a
commutative diagram

The uniqueness of the homomorphism a will be used to show that if Q has

a universal central extension, then U is unique to isomorphism.
Observe that commutativity of the diagram implies that a(K) IL; that
is, insertion of the map P = alK: K -+ L yields an augmented commutative
diagram
By Lemma 7.67, the transgressions are related by dUP*= dV.As both U and
V are covers, however, both transgressions are isomorphisms (Lemmas 7.63
and 7.64), and so P* is an isomorphism. By Exercise 10.55, P is an isomor-
phism, and by Exercise 10.56, a is an isomorphism. Finally, Lemma 11.9
shows that U is, perfect. H

We are now going to prove that Hopf's formula holds for every (not neces-
sarily perfect) finite group Q.
 11. Free Groups and Free Products
Lemma 11.313. If Q is a finite group and
l-+K-+E-+Q-+l
is a central extension, then K n E' is finite.
pmo$ since K I Z(E),we have [E : Z(E)] I CE : KI = lQ1 < a.Thus Z(E)
has finite index in E, and so Schur's Theorem (Theorem 7.57) gives E' finite.
Therefore K n E' <_ E' is also finite.
Lemma 11.14. ~fQ = FIR is a finite group, where F is free, then there is a
central extension
1 -,R/[F, R] 4 F/[F, R] Q -, 1. -+
Moreover, if F is finitely generated, then R/[F, R] is also finitely generated.
proof. We have already noted (in Theorem 11.10) that the sequence is a
central extension. NOW F finitely generated implies that FIEF, R1 is finitely
generated. As Q is finite, R/[F, R] 5 FIEF, R1 is a subgroup of finite index,
and so Lemma 7.56 shows that R/[F, R] is finitely generated. ill
Lemma 11.15. Let Q = FIR be a finite group, where F is a finitely generated
free group. The torsion subgroup of R/[F, R] is (R n F')/[F, R], and there is a
subgroup S with [F, R] 5 S 5 R, with S a F , and with
RICK R1 = (R n F')/CF, R] O S/[F, R].
Proof. Let T = (R n F')/[F, R]. Since T = (R/[F, R]) n (F'/[F, R]), Lemma
11.13 shows that T is finite, and so T 5 t(R/[F, R]). For the reverse inclu-
sion, note that (R/[F,R])/T E (R/[F,R])/((R n F')/[F, R ] )s R/(R n F') z
F'RIF' I FIF', which is free abelian. By Corollary 10.16, there is a subgroup
S with [F, R] si: S si: R such that R/[F, R] = T @ S/[F, R]. Therefore,
t(R/[F, R]) = (R n F')/[F, R]. Finally, R/[F, R] 5 Z(F/[F, R]), so that all
its subgroups are normal in F/[F, R]; in particular, S / [ F , R] a FIEF, R] and
S a F . FR4
Lemma 11.16. Let Q be a finite group and let Q = FIR, where F is a finitely
generated free group.
(i) There is a central extension
l-+K-,E-,Q+l
with K <_ E' and with K (R n F')/[FY R].
(ii) j(R n F')/[F, R1 I 2 I M(Q)I.
proof: (i) Choose S, as in Lemma 11.15, with S 4 F, [F, R1 5 S 5 R, and
R/[F, RI = (R n F)/[F, R] 0S/[F, R]. Consider the exact sequence
1 -+ R/S -+ F/S -+ Q -+ 1-
resentations and the Schur Multiplier
Since R/[F, R1 is central in F/[F, R], it follows that R/S g
(R/[FYRI)/(S/[F, R1) is central in F/S E (F/[F, R])/(S/[F, R]); moreover, the
definition of S gives (RICF, R])/(S/[F, R]) 2 (R n F')/[F, R]. Finally, R/S =
(R n F')S/S I FrS/S = (FIS)'.
-
(ii) By Lemma 11.13, R/S is finite. By Lemma 7.64, the transgression
6: (R/S)* M(Q) is an injection, so that I(R n F')/[F, R]J = 1 R/SI =
I(R/s>*l IM(Q)I.
Theorem 11.17 (Schur, 1907). If Q is a finite group with Q = FIR, where F is
a finitely generated free group, then
M(Q) z (R n F')/[F, R].
Proof. Let 1 -+L+ U 5 Q -+ 1 be a central extension with L 5 u',let
I , ...,Y,) be a basis of F, and let v: F -+Q be a surjective hornomor-
ism. For all i, choose ui E U with n(ui) = v(yi). Since F is free with basis
yl, ..., y,), there is a homomorphism cr: F -, U with .no. = v. Now L =
L n U' 2 Z(U) n U' I(D(U),by Theorem 5.49. Therefore,
u = ( ~ 1 .,..,u,, L) l (u1, .. .,u,, (D(U)) = (ul, ...,u,),
by Theorem 5.47, so that a is surjective.
If a E L, then a = a(w) for some w E F (for a is surjective), and so 1 =
= I[G(W)= v(a). Hence, w E ker v = R, and so a = a(w) E o(R); that is,
o(R). For the reverse inclusion, if r E R, then ncr(r) = v(r) = 1, so that
E ker n = L. Thus, L = a(R). Note that cr([F, R]) = [o(F), a(R)] =
,L] = 1, because L is central, so that a induces a homomorphism
a: F/[F, R] -+ U. But Z((R n F')/[F, R])) = o(R) n o(F') = L n U' = L, and
0 I@ n P')/CP, Rll 2 ILI.
By Theorem 7.66, there is a central. extension
l+L+U+Q-+l
with L 2 M(Q) and L s U'.Therefore, Lemma 11.16(ii) gives
.
IM(Q)I = ILI 5 I@ n F')ICF, RI l 2 lIlrf(Q)l-
Returning to Lemma 11.16 with K Z (R n F')/[F, R], the injection
6: K* -+ M(Q) must be surjective, and so M(Q) 2 (R n F)/[F, R].
Corollary 11.18. For every finite group Q, if Q = FIR, where F is a finitely
generated free group, then (R n F)/[F, R] is independent of the finite presen-
tation FIR of Q.
Proof: We have (R nF)/[Fy R] sz M(Q), and the Schur multiplier M (Q) is
defined independently of a presentation.
Definition. If 21 is a finitely generated abelian group, let d(A) denote the
number of elements in a smallest generating set of A; that is, A can be gener-
364 11. Free Groups and Free Products Presentations and the Schur Multiplier

ated by some set of d ( ~elements,

) but it cannot be generated by any set of sequence of Theorem 10.17). Now
size d(A) - 1. R/(R n F') z RFr/F' IF/F1.
~f p ( ~denotes
) the rank of a finitely generated abelian group A, that is, the BY Ixmma 11.3, F/F' is free abelian of rank n; by Theorem 10.17, its sub-
rank of the free abelian group A/tA, then it is easy to see that group R/(R F') is also free abelian, and so Corollary 10.16 shows that the
exact sequence (1) splits. Thus,
p(A) Id(4;
with equality if and only if A is free abelian; moreover, p(A) = 0 if and only if R/[F, R] r M(Q) O RF'IF';
A is fite. The reader may prove that if A and B are finitely generated abelian since RF'/F' is free abelian, d(M(Q)) + d(RF'/Ff) = d(R/[F, R]) 5 r, and so
groups, then
P(A O B) = p(A) + p(B);
indeed, Exercise 10.50(ii) shows that if there is an exact Sequence 0 -+ A + Since RF'/Ft is free abelian, d(RFf/F') = p(RF'/F1), and so
E -t B -+ 0 of abelian groups, then p(E) = p(A) + dB). On the other hand, (2) p(RFtIF') 5 r - d(M(Q)).
this is not generally true if p is replaced by d. For example, Z6 Z2 @ Z,,
and 1 = d(Z,) = d(Z, 6 Z,) # d(Z,) + d(Z3) = 2. However, if F is a finitely Now Qf = (FIR)' = RFf/R, SO that Q/Qt = (F/R)/(RFt/R) = F/RF' and
generated free abelian group, then (3) p(F/RF') = p(QIQr).
d(A @ F) = d(A) + d(F). There is another exact sequence
Lemma 11.19. Assume that Q has a finite presentation 0 + RF'/F' -+ F/F' -+ F/RF' -+ 0,
Q = (xl, ..., ~ , I Y I ,
- - - 7 Y,).
~f F is the free group with basis (x,, ...,x,) and R is the normal subgrou
by (y,, ...,y,), then R/[F, R] is a finitely generated abelian grou n - p(Q/Qr)= p(RF'/F') Ir - ~(M(Q)),
and d(R/[F, R]) I r. hich is the desired inequality.

Proof. We have R' = [R, R] 5 [F, R] 5 R, since R F, so that R/[F, R] is Corolany 11.21. If Q is a finite group having a presentation with n generators
abelian. The proof is completed by showing that it is generated by the cosets and r relations, then
of the y's. Now R is generated by (fyif -': f E F, i = 1, . . . ,I) (R is the normal d(l%f(Q))-< i. - a.
subgroup of F generated by { y,, . .. , y,)). But fyif -'y;' E IF, R], so that
fyif [F, R] = yi[F, R], as desired. 3 Proof: Since Q is finite, Q/Qf is finite and p(Q/Qt) = 0.

Theorem 11.20. If Q has a finite presentation Since d(M(Q)) 2 0, it follows that r 2 n for every finite presentation of a
finite group Q; that is, there are always more relations than generators. We
Q = (XI,. - . , x ~ I Y I ~~ r ) , give a name to the extreme case.
then M(Q) is finitely generated, d(M(Q)) d r, and
Definition. A group is balanced if it has a finite presentation with the same
n - r 5 p(Q/Qr)- d(M(Q)).
number of generators as relations.
Proof: Let F be free with basis (x,, . . ., x,), and let R be the normal subgroup
generated by { y, , ...,y,). By Lemma 11.19, there is an exact sequence of 11-22. If Q is a finite balanced group, then M(Q) = 1.
finitely generated abelian groups
(1) 0 -+(R n Fr)/[F, R] -t R/[F, R1-+ R/(R n F') 0. +

Therefore, d(M(Q)) = d((Rn F')/[F, R]) 5 r (by Exercise 10.7, an easy con-
The converse of this corollary is false.

Corollary 11-23.If V is the 4-group, then M(V) - Z2.

 I
11. Free Groups and Free Products
366
proof.~n Example 7.17, we saw that M(V) # 1, and l 3 m ~ e m7-68shows that
exp(M(V)) = 2. There is a presentation V = (a, bla2 = 1, b2 = 1, [a, b1 = 1).
~y ~~rollar- 1 1.21, d(M(V))S 3 - 2 = 1, and so M(V) is cyclic-
We have now completed Example 7.17: in contrast to perfect groups, the
4-group V does not have a unique cover.
It is a theorem of J.A. Green (1956) that if p is a prime and Q is a group of
order pn, then IM(Q)l 5 pn(n-1)12. One can also show that this bound is best
possible: equality holds if Q is an elementary abelian group of order pn; of
course, a special case of this is Q = V.
We have proved two theorems helping us to compute the Schur multiplier
of a finite group Q: the theorem of Alperin-Kuo (Theorem 7.68) giving a
bound on exp(M(Q));Corollary 11.21 giving a bound on d(M(Q)).
11.19. Prove that M(Q,) = 1, where Q, is the group of generalized quaternions.
11.20. Prove that M(D,,) = 1 if n is odd and has order 2 2 if n is even. (It is known
that M(D,,) r 12, if n is even.)
11.21. Prove that IM(A,)I 5 2. (It is known that M(A,) r Z,.)
11.22. (i) As in Example 11.5, show that A, has a presentation
A4 = (s, tls2 = 1, t3 = 1, = 1).
(ii) Show that the binary tetrahedral group B is a cover of A,.
(iii) Prove that M(A,) r 12,.
11.23. Show that M(S4)is cyclic of order 5 2. (Hint. Example 11.5.) (It is known that
fW(S,,)z Z2 for all n 2 4.)
11.24. Show that SL(2,4) is the cover of PSL(2,rf).
Fundamental Groups of Complexes
The theory of covering spaces in Algebraic Topology contains an analogue o_f
Galois Theory: there is a bijection from the family of all covering spaces X
mapping onto a topological space X and the family of all subgroups of the
fundamental group a,(X). This theory was used by Baer and Levi to Prove
the Nielsen-Schreier theorem: Every subgroup of a free group is itself free.
We mimic the topological theorems here in a purely algebraic setting.
Definition. A K (or abstract simplicial Complex) is a family of
empty finite subsets, called simplexes, of a set V = Vert(K), called vertices,
such that:
(i) if v E I/, then (v) is a simplex;
(ii) if s is a simplex, then so is every nonempty subset of 3.
Fundamental Groups of Complexes
+
A simplex s = {Do, vl, - . . , v,) with q 1 vertices is called a q-simplex; one
says that s has dkzension q, and one writes dim(s) = q. If n is the largest
dimension of a simplex in K, then K is called an n-complex and one writes
dim(K) = n (if there is no simplex of largest dimension, then dim(K) = a).
A 0-complex is a set of points, and a 1-complex is a graph: define u, u E V
to be adjacent if and only if (u, v) is a 1-simplex. It turns out that Zcomplexes
are sufficiently complicated to serve all of our needs (see Exercise 11.21
below).
Even though no topology enters into the forthcoming discussion, the read-
er should know the geometric background behind the definition of complex
in that setting. A 0-simplex is a point; regard a 1-simples {u, v) as an edge
with endpoints u and v; regard a 2-simplex (u, v, w) as a (two-dimensional)
triangle with vertices u, v, and w; regard a 3-simplex as a (solid) tetrahedron;
and so forth. A complex may now be regarded as a space built by gluing
simplexes together in a nice way.
Figure 11.2
A complex L is a subcomplex of a complex K if Vert(L) c VertiKI and if
every simplex in L is also a simplex in we , , \
recognize the empty set 0
being a subcomplex). A subcomplex L of K isfull if every simplex in K having
,
all its vertices in L already lies in L. Thus, a full subcomplex L is determined
by its vertices Vert(L).
For example, if s is a simplex, then the subcomplex Is],consisting of and
all its nonemPtY subsets, is full. For each q 2 0, the q-skeleton, defined by
Kq = {simplexes s E K: dim(s) < qj,
is a subcomplex. Thus, Vert(K) = KO c K1 = KO {all l-simplexes) c
K 2 K 3 C ... If dim(K) = n and q < n, then Kq is not a full subcomplex.
Definition. An edge in a complex K is an ordered pair E = (u, v) of (not neces-
sarily distinct) vertices lying in a simplex. If u and v are vertices in a complex
368 11. Free Groups and Free Products j Fundamental Groups of Complexes 369

K, then a path a of length n from u to v is a sequence of n edges Definition. The connected subcomplexes Ki occurring in the disjoint union
a = (u, vl)(vl, v2) - - - (vn-2, v,,-l)(vn-l, 4. K = U Ki are called the components of K.
Call u the origin of a and denote it by o(a); call v the end of a and denote it by We are now going to define a multiplication of paths reminiscent of juxta-
e(a). A closedpath at u is a path a for which o(a) = e(a). position of words in a free group.

Definition. A complex K is connected if there is a path between any pair of its Definition. If a = E ~ ... E, and p = q, ...q, are paths in a complex K, where
vertices. the E, and qj are edges, and if e(a) = o(P), then their prodrsct is the path

Definition. If (L,: i E I) is a family of subcomplexes of a complex K, then the

union (JLi is the subcomplex consisting of all those simplexes s lying in The path ap is a path from o(a) to e(p). This multiplication is associative
least one L,, and the intersection Li is the subcomplex consisting of when defined, but every other group axiom fails.
those simplexes s lying in every L,. Two subcomplexes L and L' are disjoznt
ifLnLf=@. Definition. There are two types of elementary moves on a path a in a complex
K. The first replaces a pair of adjacent edges (u, v)(v, w) in a by (u, w) if
Li) = U Vert(Li)and ~ e r t ( rLi)
It is easy to see that ~ e r t ( U / = r/ Vert (u, v, w) is a simplex in K; the second is the inverse operation replacing (u, w)
in particular, L and L' are disjoint if and only if Vert(L) n Vert(L1)= 0. by (u, v)(v, w) in this case. Paths a and P in K are homotopic, denoted by
a rz p, if one can be obtained from the other by a finite number of elementary
Theorem 11.24. Every complex K is the disjoint union of connected su
complexes K = U K,, and the Ki are uniquely determined by K. Moreove
each K, is a full maximal connected subcomplex.

Proof. The relation on V = Vert(K) defined by u r v if there is a path in

from u to v is easily seen to be an equivalence relation; let ( y :i E 1
family of equivalence classes, and let K ibe the full subcomplex of K
vertex set K. Clearly, K is the union U K,. If a simplex s in K has a v
in K,, then there is a path from u to each vertex of s, and so s c '&I;:
Y
s E K ibecause is full. Mow I< is the disjoint union U Xi,
X W
for if s E Ki n
where i + j, then s c n = @, a contradiction. To see that ISi is co Figure 11.3
nected, assume that there is an edge (u, v) in M,where u E 'Zr,. Then s = (u,
is a simplex, and so the argument above shows that s c B/i and v E
If u, w E K, then u = w, and so there is a path in K from u to w. An indu For example, let K be the 2-complex drawn above, let a =
tion on the length of the path shows that the path lies in Ki, and so Ki i ,U)(u, w)(w,y), and let p = (x, u)(u, u)(v, w)(w, y). If K contains the simplex
connected. u, v, w), then a - p; if K does not contain this simplex, then a $P.i
U
To prove uniqueness, let K = Lj be a disjoint union, where each Lj is a It is easy to check that homotopy defines an equivalence relation on the
connected subcomplex. It is easy to see that each Lj is a full subcomplex; it family of all paths in K.
follows, for each simplex in K, that there is a unique Lj containing all its
vertices. In particular, there is no simplex (u, v) with u E Vert(Lj) and v $ Definition. If a is a path in a complex K, then the equivalence class of a,
Vert(Lj); this shows that each Lj is a maximal connected subcomplex, for denoted by [a], is called apath class.
there are no paths leading outside of it.
Choose some Lj. If s E L,, then there is a unique Ki with s E Ki. If t E Lj is If a -- P, then o(a) = o(P) and e(a) = e(P) (for only "interior" vertices are
another simplex, then t E K, for some 1. However, the presence of a path changed by the elementary moves in a homotopy). Hence, one may define the
between a vertex of s and a vertex of t shows, as above, that 1 = i. Therefore, origin and end of a path class [a], denoted by o[a] and e[a], respectively.
t E K iand Lj is contained in Ki. Maximality of Lj gives Lj = K,. B! Homotopy is compatible with the multiplication of paths: if a 2. P, a' 1 p,
370 11. Free Groups and Free Products Fundamental Groups of Complexes

and e(a) = o(P), then the reader may check that aP I.a'P'; that is, if Theorem 11.27.
e(a) = o(P), then multiplication of path classes, given by (i) If (K, v) is a complex with basepoint, and if L is the component of K
containing v, then
n(K, u) = n(L, v).
is well defined. (ii) If K is a connected complex with basepoints v and v', then
If K is a complex and v E Vert(K), then the trivial path at v is (v, u). If
E = (u, u), define E-I = (v, u) and, if a = E, ...E,, is a path, define its inoersepath n(K, u) r a(K, u').
a-I = ...E;'.
Proof. (i) Since every (closed) path with origin v has all its vertices in Vert(L),
Lemma 61.25. The set of all path classes of a complex K has the following the underlying sets of the two groups are equal. As the multiplications on
properties: each coincide as well, the groups themselves are equal.
(ii) Since K is connected, there is a path y in K from v to u'. Define
(i) if o [a] =u and e [a] = v, then
f : a(K, V)-+a(K, v') by [a] H [y-'1 [a] [y] = [y-lay]. Note that the latter
multiplication takes place in the groupoid of all path classes in K; the prod-
uct, however, lies in n(K, 0'). It is a simple matter, using Lemma 11.25, to

.
check that f is an isomorphism with inverse [PI ct [y] [PI [y-'1.
and
Definition. If K and L are complexes, then a simplicial map q: K -+ L is a
Gal = [(u, 41. function q: Vert(K) -+ Vert(L) such that (quo, qv,, ...,qv,) is a simplex in L
(ii) if a, P, and y are paths and one of ([a] [PI) [y] or [a] ([PI [y] ) is defined, whenever (v,, v,, . .,v,) is a simplex in K. A sirnplicial map q is an isomor-
then so is the other and they are equal. phism if it is a bijection whose inverse is also a sirnplicial map.

Proof. Straightforward. 8% The identity on Vert(K) is a simplicial map. It is easy to see that the
composite of simplicial maps, when defined, is a simplicial map. If q: K -+ L
The set of all path classes in K with its (not always defined) multiplica- is a sirnplicial map and (v,, v,, ...,v,) is a simplex, then there may be re-
tion is called a groupoid. We extract groups from a groupoid in the obvious peated vertices in the simplex (quo, qv, ,..., qv,].
way. Let q : K -* L be a sirnplicial map. If E = (u, v) is an edge in K, then qe =
(qu, qv) is an edge in L (because (rpu, rpv) is a simplex in L). If a = . .. E,, is
a path, then define

DeGni~ow.A basaigoimt of a complex X is some chosen vertex v. The Jurlda-
mentalgroup of a complex with basepoint v is
a(K, v) = ([a]: a is a closed path at v).
Theorem 11.26. For every vertex v in a complex K,n(I<, v) is a group with
identity the path class of the trivial path at v.
- -
which is a path in L. If u P are paths in K, then q a qP in L,for if (u, v, w)
is a simplex in K, then (qu, qv, qw) is a simplex in L.
Theorem 11.28. If q : K -+ L is a simplicial map, then q,: n(K, v) n(L,qv),
-+

defined by [ a ] ~ [ q a ] , is a homomorphism. Moreover, a is a (covariant)

functor: (I,), is the identity, and if $: L -+ M is a simplicial map, then (t,bq), =
Proof. This follows at once from the lemma, for multiplication is now always
defined. $#q#: a(K, U) a(M, $qv).
-+

Proof. Routine. 88
Remark. There is a topological space IK I which is the "geometric realization"
of a complex K , and a(K, v) is isomorphic to the fundamental group of IKI I

defined in Algebraic Topology (see Rotman (1988), Theorem 7.36). Definition. A path a = E, ...E,, is reduced if either a is trivial or no ci = (u, v) is
The next result shows that the fundamental group of a connected complex adjacent to its inverse (v, u) and no is a trivial path. A circuit is a reduced
does not depend on the choice of basepoint. closed path.'
372 11. Free Groups and Free Products
Let us show that every path a in a complex is homotopic to either a
-
reduced path or a trivial path. If a contains a subpath (u, v)(v, u), then a d,
where a' is obtained from a by replacing (u, v)(v, u) by the trivial path (u, u). If
a' is not trivial and a' contains a trivial path (u, u), then a' a", where a" is
obtained from a' by deleting (u, u). These steps can be iterated. Since each
path obtained is shorter than its predecessor, the process eventually ends,
and the last path is either reduced or trivial. In particular, every closed path
is homotopic to either a circuit or a trivial path.
Definition. A tree is a connected complex of dimension < 1having no circuits
(the only zero-dimensional tree has a single vertex).
Let us show that if u and u are distinct vertices in a tree T, then there is a
unique reduced path from u to v. Connectivity provides a path a from u to u,
which we may assume is reduced. If P f a is another reduced path from u to
v, then a and p contain a (possibly empty) subpath y such that a = a'y,
p = Ply, and the last edge of a' is distinct from the last edge of P . It follows
that a1/3'-I is reduced, and hence it is a circuit in T.This contradiction shows
that a = p.
Definition. A complex K is simply connected4 if it is connected and n(K, v) =
1 for some v E Vert(K).
By Theorem 11.27(ii), this definition does not depend on the choice of
basepoint v in K.
Every tree T is simply connected: we have just noted that every closed path
is homotopic to either a circuit or a trivial path, and there are no circuits in
a tree.
Theorem 11.29. Let L be a simply connected subcomplex of a complex M.If a
is a closed path in K at v all of whose edges lie in L, then [a] = 1 in n ( K , v).
This is true, in particular, when L is a tree.
Pro05 The inclusion 'p: Vert(L) Vert(K) is a simplicia1 map L -t K, and
it induces a homomorphism 9,: n(L, v) --+ n(K, u). The hypothesis gives
'p,([a]) = ['pa] = [a], SO that [a] E im p',. But L simply connected gives
n(L, v) = 1, hence [a] = 1.
If L is a subcomplex of a complex K , then the homomorphism
p
' ,:n(L, v) -t n(K, v ) induced by the inclusion 'p: Vert(L) c,Vert(K) need
not be injective. For example, it is easy to see that a 2-simplex K is simply
connected, but we shall soon see that its perimeter is not.
Some authors do not insist that simply connected complexes be connected. For them, a com-
plex is simply connected if all its components are simply connected in our sense.
Fundamental Groups of Complexes 373
Definition. A subcomplex T of a complex K is a maximal tree if T is a tree
which is contained in no larger tree in K.
Lemma 11.30. If K is a connected complex, then a tree T in K is a maximal
tree if and only if Vert(T) = Vert(K).
Proof. Suppose that T is a maximal tree and there exists a vertex v 4 Vert(T).
Choose a vertex vo in T; since K is connected. There is a path E , ...8, in K
~ , Since v, is in T and v is not in IS there
from vo to u = v,; let gi = ( u ~ -vi).
must be an i with vi-, in T and vi not in T. Consider the subcomplex T'
obtained from T by adjoining the vertex vi and the simplex {vi-,, vi). Clearly
T' is connected, and any possible circuit in 7''must involve the new vertex ui.
There are only two nontrivial edges in T' involving ui, namely, E = ( u ~ -vi) ~,
and E-f, and so any closed path involving vi as an "interior" vertex is not
reduced, while any circuit at vi would yield a circuit in T at vi-, . Thus T'is a
tree properly containing T, contradicting the maximality of T.
The proof of the converse, similar to that just given, is left to the
reader.
Every complex K has a maximal tree (this is obvious when K is finite, and
a routine Zorn7slemma argument shows it to be true in general). Usually, a
complex has many maximal trees.
efinition. Let K be a complex and let 9 = {Xi: i E I) be a partition
of Vert(K). The quotient complex K / 9 has vertices the subsets 4, and
simplex if there are vertices uij E Xij such that {ui,. .., uiq)
Of course, one can construct a quotient complex of K modulo an equiva-
lence relation on Vert(K), for the equivalence classes partition Vert(K).
EXERCISES
11.25. Prove that a complex K is connected if and only if its 1-skeleton K1is con-
nected.
11.26. If s is a simplex, then the complex Is1 (consisting of s and all its nonempty
subsets) is simply connected.
11.27. Prove that the inclusion K% K induces an isomorphism n(K2,u) r n(K, v).
Conclude that every fundamental group arises as the fundamental group of a
2-complex.
11.28. Let In be' the 1-complex having vertices {to,..., tn} and simplexes {to,t, j,
{t,, t,), ..., (tn-,, t,}. Prove that a path in a complex K of length n is a
simpliclal map In4 K.
374 11. Free Groups and Free Products Tietze7sTheorem 375

11.29. Let T be a finite tree. If v(T) is the number of vertices in T and e(T) is the
number of edges in T, then
v(T) - e(T) = 1.
11.30. Prove that a 1-complex K is simply connected if and only if it is a tree.
11.31. (i) Let K be a complex and let T and S be trees in K. If T n S # 0, then T n s
is a tree if and only if T n S is connected.
(ii) If {q:i E I ) is a family of trees in a complex K with q n Zj a tree for all i
and j, then U is a tree.
Since T is a maximal tree in K, there is a unique reduced path A, in T from
w to each v E Vert(T) - (w} = Vert(K) - (w}; define 1, = (w, w). ~ e f i n ea
function f : X -,n(K, w) by
(u, C~,(U,v)A,ll
(which is the path class of a closed path at w), and let rp: F -+ n(K, w) be the
unique homomorphism it defines. We claim that R I ker 9.
Type (a): If (u, v) is in T, then the path A,(u, v)l;' lies in T, and hence

-
11.32. Let K be a 1-complex with basepoint w, let T be a tree in K, and let (u, v) be an
edge not in T. If a = al(u, v)a" and a = P(u, v)/?" are closed paths in K at w
with a', a", p', and p" paths in T, then a P.
11.33. Let G be a free group of rank 2 with basis X. Show that the graph associated
[AU(u,v)A;'] = 1 in n(K, w), by Theorem 11.29.
Type (b): If (u, v, x) is a simplex in K, then

to the Cayley graph T(G,X) is a tree.

11.34. If cp: K -t L is a simplicia1 map, then im cp is a subcomplex of L; moreover, if = [A,(u, x ) ~ , ~ ] .
K is connected, then im cp is connected.
11.35. If K is a complex and 9 = ( X i :i E I ) is a partition of its vertices, then the
Therefore, rp induces a homomorphism @ r ( K , T) -+ n(K, w) with
natural map v: K -,KIP, which sends each vertex into the unique Xicontain-
ing it, is a simplicial map.
@: (u, V)R ,+[A,(u, v)A;~].
We prove that Q is an isomorphism by constructing its inverse. If E l ...En is
11.36. Let K be a connected complex, and let L be a subcomplex that is a disjoint
union of trees. Show that there is a maximal tree of K containing L. (Hint. Use a closed path in K at w, define
Exercise 11.35.) O(cl. ..E,) = E ~ .&,R
. E y ( K , T).
Observe that if a and /3 are homotopic closed paths, then the relations in
Y(K, T) of Type (b) show that B(a) = B(/3): for example, if a = y(u, v)(v, x)S
Tietze's Theorem and /3 = y(u, x)6, where (u, v, x) is a simplex in K, then

Tietze9stheorem gives a presentation for the fundamental group of a con- p-la = 6-'(za, x)-' (u, u) (v, x)S E R.
nected complex. There is thus a homomorphism 8:z(X, w ) -+F ( K , T ) given by
Definition If T is a maximal tree in a connected complex I<, then F ( K , T )is O: .E,] -@(El. . .E,) = - .. E,,R.
the group having the presentation: . ..E,] E z(K, w), then
Let us compute composites. If
generators: all edges (u, v) in K;
.E,R) = R). ..(D(E,R)
relations: Type (a): (u, v) = 1 if (u, v) is an edge in T;
Type (b): (u, v)(v, x) = (u, x) if (u, v, x) is a simplex in K. = Crp(~1). -.d & , ) l
Theorem 11.31 (H. Tietze, 1908). If K is a connected complex and T is a
maximal tree in K, then
n(K, w) 2 F ( K , T).
because A, is a trivial path. Therefore, 00 is the identity. We now compute
ark. Since K is connected, different choices of basepoint w for K yield the other composite. If (u, v) is an edge in K, then
isomorphic groups. @@((u,v)R) = @(rp(u,4 )

PPOO$Let F be the free group with basis X = all edges (u, v) in K and let R
be the normal subgroup of relations, so that Y(K, T) = FIR.
376 11. Free Groups and Free Products
But A, and A,' lie in R, since their edges do, so that
A,(u, v)Ayl R = A,(U, v)R = (u, v)R,
the last equation arising from the normality of R (Exercise 2.30(ii)). The
composite Od,thus fixes a generating set of F ( K , T), hence it is the identity.
Therefore, n(K, w) z T ( K , T).
Corollary 11.32. If K is a connected 1-complex, then n(K, w) is a free group.
Moreover, if T is a maximal tree in K, then
rank n(K, w) = I { 1-simplexes in K not in T) I;
indeed, F has a basis consisting of all [d,(u, v)A;'], where one edge (u, u) is
chosen from each 1-simplex (u, v) 4 T.
Proof. By the theorem, it suffices to examine F ( K , T). The relations of Type
(a) show that Y ( K , T) is generated by those edges (u, v) in K which are not in
T. A smaller generating set is obtained by discarding, for each 1-simplex
{u, u), one of the two edges (u, u) or (v, u), for (v, u)R = (u, V)-'R in Y ( K , T).
If {u, v, w) is a simplex in K, then at least two of the vertices are equal, for
dim K = 1. Thus, the relations of Type (b) have the form:
All of these are trivial: since (v, u) = 1 and (a, u) = (u, u)-l, the subgroup 41 of
relations is 1. Therefore, T ( K , T) is free with basis as described. EiI
Here is a simple example of a complex whose fundamental group is free.
Figure 11.4
Covering Complexes 377
Definition. If I is a set, then a bouquet of 1 1 circles is the 1-complex BI
with distinct vertices {u,, v,: i E I) v (w) and 1-simplexes (w, ui), {w,vi), and
{v,, u,) for all i E I.
Corollary 11.33. If I is a set and BI is a bouquet of I circles, then n(BI, w) is a
free group ofrank I I I.
Proof. It is easy to see that P3, is a connected 1-complex and that all the
1-simplexescontaining w form a maximal tree. Therefore, n(BI, w) is free with
basis
{ C ( w , ~ i ) ( ~ i , v i ) ( v i , ~ ) I : i E I881) .
In Exercise 11.14, we gave a presentation of the trivial group G = 1. Using
Tietze's theorem, we see that fundamental groups of simply connected com-
plexes produce many such examples.
Covering Complexes
The last section associated a group to a complex; in this section, we associate
a complex to a group.
Definition. Let p: K -+ K' be a simplicial map. If L' is a subcomplex of K',
then its inverse image is
p-l(L') = (simplexes s E K : p(s) E L').
It is easy to check that p-I (L') is a subcomplex of K which is full if E' is full.
In particular, if s is a simplex in R', then the subcomplex Is/, consisting of s
and all its nonempty subsets, is full, and so p-'(Is/) is a full subcomplex of K.
In what follows, we will write s instead of Is/.
Definition. Let K be a complex. A connected complex I? is a covering complex
of K if there is a simplicia1 map p: I? -+ K such that, for every simplex s in K,
the inverse image p-'(s) is a disjoint union of simplexes,
p-'(s) = u S,,
is1
with p 1 .Ti:5 -,s an isomorphism for each i E I. The map p is called the projec-
tion and the simplexes 5, in are called the sheets over s.
If K has a covering space, then K must be connected, for a projection
p: I? -+ K is a surjection, and Exercise 11.28 shows that the image of a con-
nected complex is connected.
378 11. Free Groups and Free Products Covering Complexes 379

Notice that every simplex Sin R is isomorphic to a simplex in K; it follows 1-simplex, (G, iJ) is an edge, and p(G, 6) = (w, v). To prove uniqueness, sup-
that dim = dim K. pose that (G, ii) is also a lifting (so that {W, ii) is a 1-simplex).Then {W, u") and
The picture to keep in mind is {I?, ii) are sheets over {w, v) that are not disjoint, contradicting the definition
of covering complex. Therefore, ii = 5.
If n > 1, then a = (w, v)P, where p is a path of length n - 1 beginning at v.
By the base step, there is a unique (W,6)lifting (w, v); by induction, there is a
unique lifting fl of P beginning at L Thus, (G, iJ)fl is the unique lifting of a
beginning at 15. Bll

Lemma 11.35. Let p: $ -.K be a covering complex, and let %I be a basepoint in

I? with p(G) = w. If a and p are homotopic paths in K with origin w, then their
lijtings oi and having origin G are homotopic and e(&)= e()).

Proof. It suffices to prove that if (6, iJ)(5,3) is a lifting of (u, v)(v, x) and if
\
.- s = {u, v, x) is a simplex in K, then (6, $2) is a simplex in I?. Let S =

PI {u", ii', 3') be the sheet over s containing 5 and let t" = {ii, u"",1")be the sheet
over s containing u", where pu" = pu"" = v and pii = pu"' = u. Now (ii, 6 ) and

~a s
Figure 11.5
EWLE 11.8. Let K be the "circle" having vertices (v,, v,, vz) and 1-
simplexes {v,, v,), (v,, v,), and {v,, v,), and let d be the "line" having
vertices {ti: i E Z) and 1-simplexes {ti, ti+, ] for all i E Z.Define p: + K by
p(t,) = v,, where j E i mod 3. The reader may check that p: 17 -+ K is a
(ii, 6") are liftings of (u, v) beginning at 12, so that uniqueness of lifting gives
u" = u"". The sheets S and t" over s are not disjoint, and so s" = t"; that is, u" = u"",
ii' = ii, and 1' = 3". A similar argument comparing s" with the sheet over s
containing 3 shows that 1 = 3'. E I!
Theorem 11.36. Let p: I? -+ K be a covering complex, and let W be a basepoint
in d with p(G) = w. Then p,: n(x, 6 ) + n(K, w) is an injection.
Proof: Assume that [A], [B] E x(W, G) and that p,[A] p#[B]; that is,
[PA] = [pB]. If a = pA and P = pB, then A = a" and 43 = P. The hypothesis
--
gives ar /3, and so Lemma 11.35 gives A = oi -- = B; that is, [A] = [B].
6 FTa

covering complex.

Theorem 11.34. Let p: 3 -+ K be a covering complex, and let G be a basepoint
in I? with p(W) = w. Given a path a in K with origin w, there exists a unique
path oi in l? with origin W and with poi = a.
Remark. One calls a" the liffing of a because of the picture
l?
What happens to the subgroup p,n(W, G) as the basepoint is changed?
Theorem 11.37. Let p: I? -+ K be a covering complex, and let G be a basepoint
in d with p(G) = w. If p(u")= w, then p,n(I?, G) and pp,(l?, ti) are conjugate
subgroups of n(K, w). Conversely, if H a(K, w) is conjugate to p,n(E, W),
then H = p,n(I?, 6) for some P with p(ii) = w.

I;--tK
Ip Proof. Since is connected, there is a path B from G to 2. Then P = pB is a
closed path at w, [PI E n(K, w), and Theorem 11.27(ii)gives
[B-'1n(I?, G) [B] = n(I?, 6);
hence I

Proof. We prove the lemma by induction on n = length a. If n = 1, then

c P - ' I P # ~ ( K$)[PI = p#n(I?, 6).
a = (w, v), where s = (w, v) is a simplex; we may assume that v # w, so that s
is a 1-simplex. If S is the sheet over s containing 6 , then i = {G, 5) is a Conversely, assume that H = [alp,a(I?, @)[a]-'. If fl is the lifting of a-'
380 1 1. Free Groups and Free Products
with origin @, and if e(p) = ii, then p(ti) = w. By Theorem 11.27(ii),
[P-'1n(R, @) [PI = n(E, ti),
so that p,n(R, ti) = p#([fl-']n(K, +)[PI) = [alPen(k, @)[a]-' = H, as
desired. W
Definition. If p: i? -+ K is a simplicia1 map and if w is a vertex in K, then
p-l(w) is called the fiber over w.
In the next theorem, we observe that the fundamental group G = n(K, W)
acts on the fiber p-'(w); more precisely, p-'(w) is a right G-set. Recall that one
can always convert a right G-set X into a left action by defining gx to be xg-l.
Theorem 11.38. Let p: k 9 K be a covering complex and let w be a vertex in
K. Then the fiber p-'(w) is a transitive (right) n(K, w)-set and the stabilizer of
a point @ is p,n(l?, @).
Proof. If 2 E p-'(w) and [a] E n(K, w), define :[a] = e[a"], where a" is the
lifting of a having origin 2;since homotopic paths have the same end, Lemma
11.35 shows that this definition does not depend on the choice of path in [a].
We now verify the axioms of a G-set. The identity element of n(K, W)is
[(w, w)]; the lifting of (w, w) with origin 2 is obviously (2, 2) whose end is 2.
Let [a], [PI E n(K, w), let a" be the lifting of a having origin 2, and let j = e(a").
If p is the lifting of /3 with origin j, then a"p is a lifting of ap with origin 2. B
uniqueness of lifting, a"p is the lifting of a/l having origin 2, and so 2[ap]
e[@] = e[p] On the other hand, 2[a] [PI = (e[a"])[P] = j[p] = ev].
Now n(K, w) acts transitively: if 2, j E p-'(w), then connectivity of K shows
there is a path A in R from 2 to 7. If a = pA, then [a] E n(l<, w) and ?[a] =
e[A] = P.
Finally, the stabilizer of a point 2 E p-' (w) consists of all [a] E n(X, W)
for which e[E] = 2. But e[E] = 2 if and only if [ E l E x(l?, 6 ) if and only if
[a] ~ p ~ n ( E , 6 )81.
CoroUary 11.39. Let p: R -+ K be a covering complex.
(i) If w is a basepoint in K and @ E p-'(w), then
(ii) If w and u are basepoints in K, then Ip-'(w) 1 = ]p-'(u)l.
Proof. (i) The number of elements in a transitive G-set is the index of the
stabilizer of a point.
(ii) If ti E pv1(u),then there is a path B in from 6 to G; let P = pB. Defi
homomorphisms @: n(k, 6 ) n(R, ti) and 9:n(K, w) -+ n(K, u) by [A]
-+
[B-'AB] and [a] I-,[P-lap], respectively. It is easy to check that the follow-
Covering Complexes
ing diagram commutes:
Since <f, and cp are isomorphisms, it follows that the index of im p, on the left
isequal to theindexofimp, on theright. I
We are now going to construct some covering complexes.
Definition. Let K be a complex with basepoint w and let n be a subgroup of
n(K, w). Define a relation on the set of all paths in K having origin w by
a r, P if e(a) = e(p) and Cap-'] 6 n.
Notation. It is easy to see that =, is an equivalence relation. Denote the
equivalence class of a path a by %a, and denote the family of all such classes
by K,-
We now make K, into a complex. Let s be a simplex in K, and let a be a
path in K with o(a) = w and e(a) E s. A continuation of or in s is a path p = aa',
where a' is a path wholly in s. Define a simplex in K, to be
[s, %a] = {%p: P is a continuation of a in s),
where s is a simplex in K and a xis a path from w to a vertex in s. Thus,
%p E [s, %a] if and only if there is a path a' wholly in s with P r, aa'.
Lemma lB.4BD. Let K be a connected complex, let w be a basepoint in R,and let
nI n(K, w). Then K , is a complex and the function p: a"=, -,K, defPned by
%a++ e(a), is a simplicial map.
Proof. Straightforward. Bl
Define @ = %(w, w) E K,, and choose G as a basepoint.
Lemma 11.41. If K is a connected complex, then every path a in K with origin
w can be lifted to a path A in K, $-omG to %a.
Proof. Let a = (w, v,)(v,, v,) . ..(v,_,, v), and define "partial paths" ai =
~ ,for i ;L 1. If si is the simplex {vi,ui+, 1, then both %ai
(w, vl)(vl, v2)...( v ~ -vi)
and %ai+,lie in [si, %ai], so that (%ai, %ai+,)is an edge in K,. Therefore,
A = (@, %al)(%al,%a2)...(%a;-', %a)
is a lifting of a having origin @. W
382 11. Free Groups and Free Products
Corollary 11.42. If K is a connected complex, then K, is a connected complex.
Proof. There is a path in K, from 6 to every Wa. Ril
Theorem 11.43. If K is a connected complex with basepoint w, and if n 2
n(K, w), then p: K , -,K is a covering complex with p,(K,, 6 ) = n.
Proof. If s is a simplex in K, then p' = pl [s, Wa] is an isomorphism [s, Wa] -+ s
if and only if it is a bijection. To see that p' is an injection, suppose that
WP,Wy E [s, Wa] and e(B) = p(%P) = p(Wy) = e(y). Thus, P = ap' and y = ayl,
where p' and y' are paths wholly in s, and
([a] [p' yf-l] [a-'1 E [ol]n(s, e(a))[a-'1 = 1, by Theorem 11.29). Thus,
[fly-'] = 1 E n, P =, y, %$? = Wy, and pf is an injection. To see that p' is a
surjection, let u be a vertex in s and let a' be a path in s from e(a) to v. Define
p = aaf,and note that p(%P)'= e(P) = v.
If s is a simplex in K, then it is easy to see that
as sets; but since s is a full subcomplex of K, its inverse image is a full
subcomplex of K,, and so it is completely determined by its vertices. To
prove that p: K , -,K is a covering complex, it remains to prove that the
sheets are painvise disjoint. If Wy E [s, %a:] n [s, WP], then y 5, aa' and y r,
Pp', where a' and p' lie wholly in s. It follows that e(al) = e(p') and, as in the
preceding paragraph, [aollP'-'p-'] = [afl-l]. But [aalp'-lP-'] = [yy-'1 =
1 E x, SO that a =, p, %a = %P,' and [s, %a] = [s, %'PI.
Finally, we show that p,n(K,, 6 ) = n. If a is a closed path in K a t w, then
p: K , -,K being a covering complex implies, by Theorem 11.34, that there is
a unique lifting d of a having origin G. But we constructed such a lifting A in
Lemma 11.41, so that a" = A and e(d) = e(A) = %'a.The following stata~ments
are equivalent: [a] E p,n(K,, 6); [a] = [PA], where [A] E x(K,, 6); e(A) =
o(A) = 6 ; %a = 6 ; [a(w, w)-'1 E n; [a] E n. This completes the proof. !Bl
Remark. Here is a sketch of the analogy with Galois Theory. If p: I? -+ K is
a covering complex, then a simplicia1 map q:i? 2 i? is called a covering map
(or deck transformation) if pq = p. The set Cov(K/K) of all covering maps is
a group under composition.
Recall that if k is a subfield of a field K, then the Galois group Gal(K/k) is
defined as the set of all automorphisms a: K -+ K which fix k pointwise. To
say it another way, if i: k +K is the inclusion, then Gal(K/k) consists of all
those automorphisms a of K for which ai = i. In the analogy, therefore, all
arrows are reversed.
Every connected complex K has a covering complex p: U -+ K with
The Nielsen-Schreier Theorem 383
U simply connected (Exercise 11.34 below), and it can be shown that
Cov(U/K) E z(K, W) (note that @ov(U/I<)is defined without choosing a
basepoint). It is true that U is a aniuersal covering complex in the sense that
whenever q: I? -+ K is a covering complex, then there is a projection r: U -t
I? which is a covering complex of K. (One may thus regard U as the analogue
of the algebraic closure of a field k, for every algebraic extension of k can be
imbedded in E.) Moreover, the function I? c @ov(U/I?)is a bijection from
the family of all covering complexes of K to the family of all subgroups of the
fundamental group n(K, w). For proofs of these results, the reader is referred
to my expository paper, Rocky Mountain Journal of Mathematics, Covering
complexes with applications to algebra, 3 (1973), 641-674.
EXERCISES
11.37. Let p: I? + K be a covering complex. If L_is a connected subcomplex of K and
if is a component of p-'(L), then pl L: L -+L is a covering complex.
11.38. (i) if p: I? -+ K is a covering complex and T is a tree in K, then p-l(T) is a
disjoint union of trees. (Hint: Show that every component of p-'(T) is a
tree.)
(ii) Show that there is a maximal tree of I? containing p-'(T). (Hint: Use
Exercise 11.30.)
11.39. Let p: I? -+ K be a covering complex, and suppose that there are j points in
every fiber p-'(v), where v is a vertex in K. Show that there are exactly j sheets
over each simplex s in K.
11.40. (i) Every connected complex K has a universal covering space p: I? -+ K; that
is, K is simply connected. (Hint. Let n be the trivial subgroup of n(K, w).)
(ii) If K is a connected 1-complex,then its universal covering complex is a tree.
(This last result may well have been the impetus for the "Bass-Serre the-
ory" of groups acting on trees (see Serre, 1980).
The Nieken-Schreie-s Theorem
In 1921, J. Nielsen proved that every finitely generated subgroup H of a free
group F is itself free (he also gave an algorithm deciding whether or not a
word a in F lies in H). In 1927, 0. Schreier eliminated the hypothesis that H
be finitely generated. There are, today, several different proofs of this theo-
rem, some "algebraic" and some "geometric." The first geometric proof was
given by R. Baer and F.W. Levi in 1936, and this is the proof we present.
There is another elegant geometric proof, due to J.-P. Serre (1970), using the
notion of groups acting on trees.
Theorem 11.44 (?Xielsen-Schreier). Every subgroup H of a free group F is
itself@ee. -
384 11. Free Groups and Free Products
Proof. If F has rank 111, and if K is a bouquet of II1 circles, then Corollary
11.33 allows us to identify F with n(K, w). As in Theorem 11.43, there is
a covering complex p: KH -+K with p#z(KH,6 ) = H. NOWp# is an injec-
tion, by Theorem 11.36, so that H r z(KH,G). But dim(K) = 1 implies
dim(KH)= 1, and so n(KH, G) is free, by Corollary 11.32. W
Theorem 11.45. I f F is a free group of finite rank n and H is a subgroup of
finite index j, then rank(H) = jn - j + 1.
Proof. If K is a finite connected graph, denote the number of its vertices by
u(K) and the number of its 1-simplexes by e(T). If T is a maximal tree in
K, the v(T) = v(K), by Lemma 11.30. We saw in Exercise 11.23 that
u(T) - e(T) = 1. By Lemma 11.30, the number of 1-simplexes in K - T is
If, now, Bn is a bouquet of n circles, then v(Bn)= 2n + 1 and e(Bn)= 3n.
After identifying F with n(Bn,w), let p: KH + Bn be the covering complex
corresponding to H. By Corollary 11.39, j = [F : H] = Ip-'(w)]. Therefore,
v(K,) = jv(Bn) and, by Exercise 11.33, e(KH)= je(Bn).We compute the num-
ber of 1-simplexesin KH outside a maximal tree T:
Corollary 11.32 completes the proof. @
! Let 7 be a maximal tree in X, containing p-'(T) (whose existence is
Remark. If K is an n-complex and bi denotes the number of its i-simplexes,
then its Eulev-Poinear2 eharoeieristic x ( K ) iszy=,
j- l)'bi. Thus, rank(H) =
-x(~H)-
We have shown that a subgroup H of a free group F is free; can we find a
basis of H?
Recall that a right transversal of a subgroup H in a group F consists of one
element chosen from each right coset Ha; denote the chosen element by
l(Ha). If F is free with basis X and if x E X, then both l(Hax) and l(Ha)x lie
in flax, and so the element ha,, defined by
lies in H (these elements should be indexed by (Ha, x), but we have abbrevi-
ated this to (a, x).) We are going to see that if a transversal of H in F is chosen
nicely, then the set of all nontrivial ha., is a basis of H.
The Nielsen-Schreier Theorem 385
Here is a notion that arose in Schreier's proof of the subgroup theorem.
Definition. Let F be a free group with basis X, and let H be a subgroup of F.
A Schreier transversal of H in F is a right transversal S with the property that
+
whenever x", x y ...x,". lies in S (where xiE X and g i = I), then every initial
segment xE1xy . .. x? (for k 5 n) also lies in S.
We will soon prove the existence of Schreier transversals.
Lemma 11.46. Let p: 3 -t K be a covering complex and let T be a maximal tree
in K. If w is a basepoint in K, then each component of P-'(T) meets the fiber
P-"w)-
ProoJ If C were a component of p - ' ( ~ )disjoint from the fiber, then w # p(C).
There would then exist an edge (u, u) with u E p(C) and v # p(C); of course,
v E Vert(T) = Vert(K). By Theorem 11.34, this edge may be lifted to an edge
(u", 3, where u" E C and v" # C. Since v" E pW1(T),this contradicts C being a
maximal connected subcomplex of p-'(T). IBs
Theorem 11.47. Let F be a free group with basis X, and let H be a subgroup of
F. There exists a Schreier transversal S = (&Ha):a E F); moreover, a basis for
H consists of all those ha,, = l(Ha)xl(Hax)-' that are distinct from 1, where
x E X.
ProoJ5 Identify F with n(K, w), where K is a bouquet of circles, and let T be
a maximal tree in K. Let p: K H + K be the covering complex corresponding
to H and choose some % E p-'(w) as the basepoint in K,.
guaranteed by Exercise 11.32(ii)).For each vertex V in K H ,there is a unique
reduced path A, in from $ to ?l In particular, there are such paths for every
V E p-'(w), and p,[lv] E x(K, w). We claim that the family
S = (1) u (ail p,[Av]: V E p-'(7') and A, not in F)
is a right transversal of H in n(K, w). Given a coset HCar], let oi be the lifting
of a having origin G. If V = e(oi), then V E pW'(w) and [&A,'] E n(KH,G).
Here is an algebraic proof of the existence of Schreier transversals. Define the length A(Hg) of
a coset Hg to be the minimum of the lengths of its elements (relative to a basis X of F). We prove,
by induction on n 2 0, that every coset Hg with A(Hg) 2 n has a representative of the form
x",x?. ..x,". each of whose initial segments is also a representative (of a coset of shorter length).
Begin by choosing 1to be the representative of H . For the inductive step, let A(Hg) = n + 1, and
let uxe E Hg be an element of minimal length, where E = $. 1. The coset H u has length n (or we
could find a shorter element of Hg), so that it has a representative v of the desired type, by
induction. Now Wg = (Hu)x" (Hv)xE,and so vxEis a representative of Hg of the desired type.
An algebraic proof that the nontrivial ha,, form a basis of H,however, does not follow as
quickly from the existence of a Schreier transversal as it does in the geometric proof above.
 11. Free Groups and Free Products
Figure 11.6
Applying p, gives [alp# [I,'] E p,n(KH, 8 ) = H,and so H [a] = Hp, [I,],
as desired.
We now show that S is a Schreier transversal. Subdivide I, into subpaths:
I, = A, A,. ..A,, where each Ai contains exactly one edge not in p-'(T) (if
there is no such subdivision, then I, lies wholly in P-'(T) c 'i and p,[A,] =
1). If U;: = e(Ei), then Ui lies in some component Ci of p-'(7'). By Lemma
11.46, there is a vertex f( E Ci n p-'(w) and, since Ci is a tree, there is a unique
reduced path Pi in Ci from Ui to q. Consider the new path Dl D, . . .Dn,where
Dl = A,/3,, and Di = P;; AiPifor i 2 2; of course, I,-= Dl D, . . .Dn. For each
i, the initial segment Dl D, .. .Di is a path in the tree T from 8 to K. Since AVi
is also such a path, these paths are homotopic: [Ivi] = [Dl D, ...D,], and so
p#[D,D,.. .Di] lies in S for all i. But Exercise 11.26 shows that each pDi
determines a generator (or its inverse) of n(K, w). Therefore, S is a Schreier
transversal.
By Corollary 11.32, a basis for n(KH,8 ) is
since p, is an injection, a basis of H is the family of all
where (U, V) is an edge corresponding to a 1-simplex {U, V ) not in and
D,,, 1: P(U, V)/3' for paths P and /3' having all edges in ? IH
Theorem 11.48. I f F is a @ee group of rank 2, then its commutator subgroup 9;'
is free of infinite rank.
PmoJ If (x, y) is a basis of F, then Lemma 11.3 shows that F/F' is free
abelian with basis {xF',yF'). Therefore, every right coset of F'a has a unique
representative of the form xmyn,where m, n E Z. The transversal l(F'a) =
1(F'xmyn)= xmynis a Schreier transversal (e.g., if m and n are positive, write
xmynas x . . .xy.. .y). If n > 0, then l(F'yn) = yn while l(F'ynx) # ynx. There-
fore l(F'yn)xl(F'ynx)-' # 1 for all n > 0, and so there are infinitely many
ha,, # 1. BBI
The Nielsen-Schreier Theorem 387
We have seen, in a finitely generated free group F, that a subgroup of finite
index is also finitely generated but, in contrast to abelian groups, we now see
that there exist subgroups of finitely generated groups that are not finitely
generated.
11.41. Let G be a noncyclic finite group with G s FIR, where F is free of finite rank.
Prove that rank(R) > rank(F).
11.42. Let G have a presentation with n generators and r relations, where n > r. Prove
that G has an element of infinite order. Conclude that n < r when G is finite.
(Hint. Map a free group on n generators onto a free abelian group on n genera-
tors, and use Exercise 10.11.) Equality n = r can occur; for example, the group
Q of quaternions is a finite balanced group.
11.43. Prove that a free group of rank > 1 is not solvable.
11.44. Exhibit infinitely many bases of a free group of rank 2.
11.45. If F is free on {x, y), then {x, y-lxy, ...,y-"xyn, ...) is a basis of the subgroup
it generates.
11.46. Prove that a group is free if and only if it has the projective property.
11.47. Use Theorem 11.45 to give another proof of Lemma 7.56: if G is a finitely
generated group and H is a subgroup of fmite index, then H is finitely
generated.
11.48. Show that a finitely generated group G has only finitely many subgroups of
any given (finite) index m. (Mint. There are only finitely many homomorphisms
cp: G -+S, (for there are only finitely many places to send the generators of G).
If H < G has index rn, then the representation of 6 on the cosets of W is such
a homomorphism q, and ker q 5 N.Apply the correspondence theorem to the
finite group G/ker q.)
11.49 (M. Hall). If G is a finitely generated group and H < G has finite index, then
there is K I H with [G: K ] finite and with K char G. (Hint. If cp E: Aut(G),
then [G : rp(H)] = [G :HI.By Exercise 11.48, there are only finitely many sub-
groups of the form q(H); let K be their intersection.)
11.50. If F is free and R a F, then FIR' is torsion-free, where R' = [R, R]. (Hint
(Russet). First reduce to the case FIR cyclic of prime order p. Let x E F satisfy
x P E R'; if x E R, its coset has finite order in R/R'; if x 4 R, then x 4 F' (since
F' I R), and X P .$ F', hence x P 4 R'.)
11.51. If F is a free group of rank 2 2, then its commutator subgroup F' is free of
infinite rank.
11.52. Let F be free on {x, y), and define cp: F -+ S3 by q(x) = (1 2) and ~ ( y=
)
(1 2 3). Exhibit a basis for ker q.
11.53. If F is free on {a, b, c, d), prove that [a, b] [c, d l is not a commutator.
 388 11. Free Groups and Free Products
Free Products
We now generalize the notion of free group to that of free product. As with
free groups, free products will be defined with a diagram; that is, they will be
defined as solutions to a certain "universal mapping problem." Once exis-
tence and uniqueness are settled, then we shall give concrete descriptions of
free products in terms of their elements and in terms of presentations.
Definition. Let {A,: i E I ) be a family of groups. Afvee prodact of the Ai is a
group P and a family of homomorphisms j,: A, -,P such that, for every
group G and every family of homomorphisms J;:: A, G, there exists a
-+
unique homomorphism q : P -,G with qj, = fi for all i.
The reader should compare this definition with Theorem 10.9, the analo-
gous property of direct sums of abelian groups.
Lemma 11.49. If P is a free product of {A,: i E I), then the homomorphisms j,
are injections.
Proof. For fixed i E I, consider the diagram in which G = A,, J;: is the identity
and, for k # i, the maps f,:A, -+ A, are trivial.
P
/7A,
A,
Then qj, = lAi,and so j, is an injection. MI
In light of this lemma, the maps j,: A, -+ P are called the irnheddings.
EXAMPLE
11.9. A free group F is a free product of infinite cyclic groups.
If X is a basis of F, then (x) is infinite cyclic for each x E X; define
j,: (x) F to be the inclusion. If G is a group, then a function f : X -+ G
determines a family of homomorphisms f,: (x) -+ G, namely, xnI--+ f (x)".
Also, the unique homomorphism q : F G which extends the function f
-+
clearly extends each of the homomorphisms f,; that is, qj, = f, for all x E X.
Here is the uniqueness theorem.
Free Products 389
Tltneorm 11.58. Let {A,: i E I ) be a family of groups. If P and Q are each a free
product of the A,, then P z Q.
Proof. Let j,: A, -+ P and k,: A, 4 Q be the imbeddings. Since P is a free
product of the A,, there is a homomorphism q: P -+ Q with qj, = k, for all i.
Similarly, there is a map $: Q -+ P with $ki = j, for all i.
Consider the new diagram
Both $q and 1, are maps making this diagram commute. By hypothesis,
there can be only one such map, and so $cp = 1,. Similarly, cp$ = lQ, and
so q : P 4 Q is an isomorphism.
Because of Theorem 11.50, we may speak of the free product P of
{A,: i E I); it is denoted by
P= * A,;
is1
if there are only finitely many A,'s, one usually denotes the free product by
Theorem 11.51. Given a family { A , :i E I) of groups, a fiee product exists.
Proof. The proof is so similar to the proof of the existence of a free group
(Theorem 11.1) that we present only its highlights,
Assume that the sets "A A, - {1) are pairwise disjoint, call (U,Af) u
(1) the alphabet, call its elements letters, and form words w on these letters;
that is, w = a,. ..a,, where each a, lies in some A" {I). A word w is reduced
if either w = 1 or w = a,. .. a,, where each letter aj E A t and adjacent letters
lie in distinct A" Let the elements of the free product be all the reduced
words, and let the multiplication be "juxtaposition." In more detail, assume
that a, ...a,, and b, ...bm are reduced. If a, and b, lie in distinct Af, then
a, ...a,b, ...'bmis reduced, and it is the product. If a, and b, lie in the same
A"nd a, b, # 1 (i.e., a, b, E A#), then a, ...(a,b, ) ...bmis reduced and it is
the product. If a, and b, lie in the same A# and a,b, = 1, then cancel and
 390 11. Free Croups and Free Products The Kurosh Theorem 391

repeat this process for a, . ..an-, and b, ...b,-,; ultimately, one arrives at a
reduced word, and it is the product. It is easy to see that 1 is the identity and
that the inverse of a reduced word is reduced; an obvious analogue of the van
der Waerden trick can be used to avoid a case analysis in the verification of
associativity.
11.56. Prove that every group E can be imbedded in a centerless group, and use this
result to prove that E can be imbedded in Aut(H) for some H.
11.57 (= 11.4). Prove that a group E has the injective property if and only if E = 1.
(Hint (Humphreys).Show that E is not normal in the semidirect product H >a E,
where H is as in Exercise 11.56.)

If P is the free product of two groups A and B, and if f : A -t G and 11.58. The operation of free product is commutative and associative: for all groups
g: B --+ G are homomorphisms, then the homomorphism q: P + G in the def- A, B, and C.
inition of free product is given by A*B r B*A and A*(B*C) z(A*B)*C.
11.59. If N is the normal subgroup of A * B generated by A, then (A * B)/N r B
(compare Exercise 11.5).
Uniqueness of the spelling of reduced words shows that q is a well defined
function, and it is not difficult to show (as in the proof of Theorem 11.1) that 11.60. Show that there is a unique homomorphism of A, * - - - * A, onto A, x - - x A,
cp is a homomorphism. which acts as the identity on each Ai.
11.61. Let A,, ...,A,, B,, ..., Bm be indecomposable groups having both chain
Theorem 11.52 (Normal Form). If g E *,,,
A, and g # 1, then g has a unique conditions. If A, * - - * A, z B1 *. .- * Bm,then n = m and there is a permuta-
factorization tion a of (1,2, ...,n) such that B,(,, r Ai for all i.
g = a, . .. a,, 11.62. If G' is the commutator subgroup of G = *i,,
(compare Lemma 11.3).
Ai, then GIG' r z,,,(A,/A{)
where adjacent factors lie in distinct A#.
Definition. The infinite dihedral group D, is the group with presentation
Proof. The free product constructed in Theorem 11.51 has as its elements all
reduced words. W
11.63. (i) Prove that D, z Z,* Z,.
Theorem 11.53. Let (A,: i E I ) be a family of groups, and let a presentation of (ii) Prove that the Schur multiplier M(D,) = 1. (Hint. Theorem 11.20.)
A, be (XilAi),where the sets (Xi: i E I] are pairwise disjoint. Then a presenta- Definition. The modular group is PSL(2, Z) = SL(2, Z)/(rt I).
tion of *,, U A,).
I A, is ( U Xi]
11.64. (i) The group G with presentation (a, blaz = b3 = 1)is isomorphic to Z, * Z,.
PzooJ Exercise 11.54 below shows that if F, is the free group with basis Xi, (ii) Prove that PSL(2, Z) r Z, * Z,. (Hint. Exercise 2.17(ii).)
then F = *,, , ,,
Fiis the free group with basis U Xi.Let { j,: A,G %i,,Ai) (iii) Show that the Schur multiplier M'(PSE(2, Z)) = 1. (Bint. Theorem 11.20.)

.
be the imbeddings. If Riis the normal subgroup of Figenerated by the rela- 11.65 (Braer-kevi). Prove that no group G can be decomposed as a free product and
tions A,, and if v,: Fi-+Aiis a surjection with ker v, = R,, then the map as a direct product (i.e., there are not nontrivial groups with A* B = G =
q: F --+ *,,,
A, extending all F, -+A, ct *,,
I A, has kernel the normal sub- C x D). (Hint (P.M. Beurnam). If G = A* B and a E A and b E B are non-
group generated by U , A,. ,, trivial, then CG(ab)z ;2; if G = C x D, choose ab = cd, for c E C# and d E DK,
to show that CG(ab)is a direct product.)
One can give a second proof of the existence of free products by showing
that the group presented by ( U X i / U A,) satisfies the conditions in the
definition. The Kurosh Theorem
Kurosh proved that every subgroup of a free product is itself a free product;
we will now prove this theorem using covering complexes. To appreciate this
11.54. Assume that the sets {Xi: i E I) are pairwise disjoint. If F, is the free group with
geometric proof, the reader should first look at the combinatorial proof in
basis Xi, then *ieIF, is the free group with basis U,.,X,. Kurosh, vol. 2, pp. 17-26.
11.55. If a E A and b E B are nontrivial elements in A * B, then aba-l b-I has infinite
order. Conclude that A * B is an infinite centerless group. Theorem 1K54. Let K be a connected complex hauing connected subcomplexes
392 11. Free Groups and Free Products The ICurosh Theorem

K,, i E I, such that K = U K,. If there is a tree T in K with T = K in Kj f i r Exercise

* *
Ti in Kii.
11.36). Observe that f n Eij= zj,
lest we violate the maximality of
all i # j, then
n(K, W)E * i:(Ki, wi)
ieI
-consider, for ali i and j, the subcomplexes 2and kiju 17" of K,. Clearly KH
is the union of these, while the intersection of any pair of them is the tree f
for vertices w in K and wi in Ki. By Theorem 11.54,

Proof. For each i, choose a maximal tree 17. in K, containing T By Exercise 7T(KH,6 ) = n(E, fi) * * 7E(kij
(is,
U E Gij)
11.31(ii), T* = U, ,
, is a tree in K; it is a maximal tree because it contains
where Gij E pE1(w)n gij.Now n(z, G) is free, because dim(2) = 1. Since f is
every vertex of K.
By Tietze's theorem (Theorem 11.36), n(K,, wi) has a presentation (E,I A,),
where E, is the set of all edges in K iand A, consists of relations of Type (a):
(u, V)= 1 if (u, V)E T ; Type (b): (u, u)(v, x)(u, x)-' = 1 if (u, v, x) is a simplex
-
a maximal tree in k, u Tietze's theorem gives n(Riju fiij) z ~ ( k , ,G,)
for each i and j. By Exercise 11.37, pll?o: kij Ki is a covering complex, and
Theorem 11.36 shows that n(Rij, fiij) is isomorphic (via (P/k,),) to a sub-
group of n(Ki, w). Finally, Theorem 11.37 shows that this subgroup is equal
in K,. There is a similar presentation for n(K, w), namely, (EIA), where E is
to a conjugate of a subgroup of n(K,, w), as desired.
the set of all the edges in K, and A consists of all edges (u, u) in T* together
with all (u, v)(v, x)(u, x)-I 'with {u, u, x) a simplex in K. Theorem 11.53 says
that a presentation for ?r(K,,w,) is ( U Eil U A,). It follows that n(K, w) 2 Corollary 11.56. If G = A,, where each A, is torsion, then every torsion-
n(K,, w,), for E = U,,,
E,, T* = Vie,
7;, and (u, v, x) lies in a simplex fiee subgroup of G is a free group.
of K if and only if it lies in some K,.
ProoJ Every nontrivial subgroup of any A, contains elements of finite
In the next section we will prove that every group G is isomorphic to the order. RJl
fundamental group of some complex. Assuming this result, we now prove the
Kurosh theorem. Corollary 11.57. If G = A,, then every finite subgroup is conjugate to a
subgroup of some Ai. In particular, every element of finite order in G is conju-
Theorem 11.55 (Knrosh, 1934). If H 5 $,,,
A,, then H = F * (*A HA)for gate to an element of finite order in some A,.
some (possibly empty) index set A, where F is a @ee group and each HAis a
conjugate of a subgroup of some A,. ProoJ Every nontrivial free product contains elements of infinite order.
Pro@ Choose connected complexes Riwith n(Ri, w,) g Ai.Defme a new
The last corollary shows how dramatically the Sylow theorems can fail for
complex I( by adjoining a new vertex w to the disjoint union U Vert(K,) ,, infinite groups. If A and B are any two finite p-groups, then each is a Sylow
and new I-simplexes (w, w,) for ail i E I. If T is the tree consisting of these
p-subgroup of A * B; thus, Sylow subgroups need not be isomorphic, let
new simplexes, then Theorem 11.54 gives
alone conjugate.
We only state the following important result of Grushko (1940) (see
Massey for a proof using fundamental groups). If G is a finitely generated
But Tietze7stheorem (Theorem 11.36) gives n(Ki u T, w,) r n(Ki, wi) E A, for group, define p(G) to be the minimal number of generators of G. Grushko's
each i, so that theorem states that if A and B are finitely generated groups, then
n(K, w) r * A,.
ieI
P(A * B) = P(A) + P(B).

Identity n(K, w) with A,, let p: K , -,I( be the covering complex

corresponding to the subgroup H 5 *,,,
A,, and choose @ E p-l(w) with
p,n(KH, @) = H. For each gp-' (5,) is the disjoint union of its components 11.66. Show that if an element T E SL(2, Z) has finite order n, then n = 1,2,3,4, or 6
I?,; choose a maximal tree Tj in Kij. Define E to be the 1-complex: (see Exercise 11.64(ii)).Moreover, T is conjugate to either + I, + A , or + 3,
E=U zj LJ p-' (T).
where
A=[: -:] and B=[-y :I.
Finally, let 17" be a maximal tree in containing p-'(T) (which exists, by
394 11. Free Groups and Free Products
11.67. Prove that the modular group has a free subgroup of index 6. (Hint. The kernel
of Z, * Z3 + Z2 x Z3 is torsion-free.)
11.68. Show that the commutator subgroup of the modular group is free.
11.69. Prove that the modular group contains a free subgroup of infinite rank.
11.70. (i) If f : A 4 G and g: B 9 H are homomorphisms, prove that there is a
unique homomorphism q : A * B + G * H with q ( A = f and q l B = g ,
Denote cp b y f * g.
(ii) Given a group A, show that there is a functor T with T(G)= A * G and, if
g: G+H, then T(g) = l,*g: A * G - t A * H .
The van Kampen Theorem
The van Kampen theorem answers the following natural question: If a com-
plex K is the union of subcomplexes L, and L,,can one compute z(K, w) =
x(L1 u L2, W)in terms of z(L1, W)and x(L,, w)? The key idea is to realize
that n is a functor, and since functors recognize only complexes and
simplicia1 maps (but not vertices or simplexes), the notion of union should be
described in terms of diagrams.
If S is a set that is a union of two subsets, say, S = A u B, then there is a
commutative diagram (with all arrows inclusions)
AnB c A
N'oreover, A U B is the "best" southeast corner sf such a diagram in the
following sense: given any set X in a commutative diagram
AnB c----, A Proofi It is easy to see that (P,j', it) is a solution, where P = (B * C)/N,jr(b) =
where f and g are (not necessarily injective) functions agreeing on A n B, then
there is a (unique) function q: A u B -,X with qlA = f and qlB = g. We
have been led to the following definition.
Defi~lition.Let A, B, and C be groups, and let i: A 4 B and j: A 4 C be (not
necessarily injective) homomorphisms. A solution (G, f, g) of the data is a
The van Kampen Theorem
commutative diagram
B'A
where G is a group and f and g are homomorphisms. A pushout of this data
is a solution (P,j', if)
B.A
'
C-P
such that, for every solution (G, f, g), there exists a unique q : P -+ G with
f = qjf and g = qi'.
If a pushout (P, j', it) of the data exists, then P is unique to isomorphism,
for if (Q,j", i") is another pushout, then the homomorphisms q: P -, Q and
$: Q -+ P provided by the definition are easily seen to be inverses.
Even though a pushout is a triple (P,j', if),one usually calls the group P
a pushout of the data.
Theorem 8 1.58.
(i) A pushout (P,j', if)exists for the data i: A -+ B and j: A -+ C.
(ii) The pushout P is isomorphic to (B * C)/lV, where N is the normal subgroup
of B 9 C generated by (i(a)j(a-I): a E A ) . Iudeed, if A has a presentation
(XIA) and B has a presentation (YIT),then P has a presentation
P = ( X u Y IA u r u (i(a)j(adl):a E A ) ) .
bN, and if(c)= cN.
Suppose that (G, f, g) is a solution of the data. The definition of free prod-
uct gives a unique homomorphism $: B * C -+ G with $ IB = f and $1 C = g:
if b E B and c E C, then $(bc) = f (b)g(c). For all a E A,
because fi = gj, and so N Iker $. Therefore, $ induces a homomorphism
9: P = (B * C)/N -,G with qjl(b) = q(bN) = $(b) =f(b) and qif(c)= q(cN) =
$(c) = g(c). The map q is unique because P is generated by (the cosets of)
B u C. It is plain that P has the desired presentation. 88
396 11. Free Groups and Free Products
Corollary 11.59.
(i) ~fi: A + B, j: A -+ C, and C = I, then the pushout P of this data is P =
BIN, where N is the normal subgroup of B generated by im i.
(ii) If A = 1, then P r B * C.
If A is an infinite cyclic group with generator x, then A * A is a free group
of rank 2 with basis, {x,y). Obviously, it is necessary to write y for the second
generator to avoid confusing it with x. More generally, if groups Ai have
presentations (Xi\ A,,) for i = 1,2, then we assume that the sets Xiare disjoint
when we say that a presentation for A , * A, is (XI u X,IA, u 8,). If the sets
X , and X , are not disjoint, then new notation must be introduced to make
them disjoint.
Theorem 11.60 (van ~arnpen).~ Let K be a connected complex having con-,
nected subcomplexes L, and L, with K = L, u L,. If L, nL, is connected
and w E Vert(L1 nL,), then n(K, W)is the pushout of the data
where ji: L, n L, +Li is the inclusion for i = 1,2.
Moreover, if a presentation of n(Li, w) is (EilAiu A!) as in Tietze's theorem
(E, is the set of edges in Li, A; are the relations of Type (a), and A: are the
relations of Type (b)),then a presentation for n(K, w) is
where E, is the set ofedges in %, n L,.
Remark. The hypothesis that Pi: is connected is redundant.
Proof. Choose a maximal tree T in L, n L, and, for i = 1 and 2, choose a
maximal tree T, in Li containing T.By Exercise 11.31(ii), T, v T, is a tree; it
is a maximal tree in K because it contains every vertex. Tietze's theorem gives
the presentation n(K, W)= ( E l k u A'), where E is the set of all edges (u, v) in
K, A' = E n (T, u T,), and
A" = { (u, V)(v, X)(u, x)-I : {u, V,X) is a simplex in K).
There are similar presentations for both n(Li, w), as in the statement of the
theorem. Make the sets El and E, disjoint by affixing the symbols j, and j2.
This theorem was first proved by Siefert and found later, independently, by van Kampen. This
is another instance when the name of a theorem does not coincide with its discoverer.
The van Kampen Theorem
By Theorem 11.58(ii),a presentation for the pushout is
(jlE1 uj2E21j l A i ujlAY u j 2 4v j 2 A i u {(jle)(j2e)-I:e E E,))
The generators may be rewritten as
L E Ou j l ( E l - EO)uj2E0 uj2(E2- E,).
The relations include j, E, = j2Eo, and so one of these is superfluous. Next,
A; = Ei n q = (Ein T) u (E, n ( q - T)), and this gives a decomposition of
j,A; u j2A', into four subsets, one of which is superfluous. Further, A" =
h; v A;, for if (u, v)(v, x)(u, x)-' E A, then s = {u, v, x) E K = L, u L,, and
hence s E Li for i = 1 or i = 2. Now transform the presentation as follows:
(i) isolate those generators and relations involving L, n L,;
(ii) delete superfluous generators and relations involving L, n L, (e.g., delete
all such having symbol j,);
(iii) erase the (now unnecessary) symbols j, and j,.
It is now apparent that both n(K, w) and the pushout have'the same presen-
tation, hence they are isomorphic. iI4
Corollary 11.61. If K is a connected complex having connected subcomplexes
L, and L, such that L, u L, = K and L, nL, is simply connected, then
n(K, w) r n(L,, w) * n(L,, w).
Proof. Immediate from the van Kampen theorem and Corollary 11.59(ii).
Caslra~Blaugr11.62. Let K be a connected complex having connected subcomplexes
k , and L, such that E l v L, = K and L, n L, is connected. If w E
Vert(L, n E,) and if L, is simply connected, then
n(K, W)N= n(L1,w)/iY,
where M is the normal subgroup generated by the image of n(L, nL,, w).
Moreover, in the notation of the van Kampen theorem, there is a presentation
Proof. Since n(L2, w) = 1, the result is immediate from the van Kampen
theorem and Corollary 11.59(i). 81
We now exploit Corollary 11.62. Let K be a connected 2-complex with
basepoint w, and let
~=e,...e,,=(w,v,)(u,,u,) ...(V , - ~ , W )
be a closed path in K at w,
Define a tiiangulated polygon D(a) to be the 2-complex with vertices
 11. Free Groups and Free Products The van Kampen Theorem 399

follows that n(L, n L,, w) zz Z. Now L,, being isomorphic to the full
subcomplex of D(a) with vertices {r, q,, ..., qn-,}, is simply connected. The
inclusion j: M G L, induces an isomorphism n(K, w) -+n(L, ,w). Define a
function $ : Vert (L, ) -+ Vert(K) by $(v) = u for all v E Vert (K) and $(qi) =
qa(pi) for all i. It is easy to check that $ is a simplicial map and that
j$: L, --+ L, is homotopic to the identity. Since rc is a functor, the induced
map $, is the inverse ofj,. The proof is completed by Corollary 11.62, for the
image of the infinite cyclic group n(L, n L,, w) is generated by [a].

The next construction is needed to attach a family of 2-cells.

Definition. Let {K,: i E I} be a family of complexes, and let wi be a basepoint

in K,. The wedge ViEz K, is the disjoint union of the K, in which all the
basepoints wi are identified to a common point b.

Figure 11.7 For example, a bouquet of circles is a wedge of 1-complexes. Theorem

11.54 shows that n(VieZKipb) r n(Kiyw,).
The next theorem was used in the proof of the Kurosh theorem.
(PO,..- Pn-1, 40, - - ,qn-1 r} and 2-sim~lexes { r qi,4i+l),
Y ~ {qi7 qi+t pi+,),
Y
Theassem 11.44. Given a group G, there exists a connected 2-complex K with
and (q,, pi, pi+,), where 0 5 i < n - 1 and subscripts are read mod n. Let
aD(a) denote the boundary of D(a); that is, aD(a) is the full subcomplex of G r n(K, w).
D(a) having vertices { p,, ...,pn-, ). Define the attaching map qa: aD(a) -+ K
BProoJ Let (XIA) be a presentation of G and let B be a bouquet of 1x1circles
by %(pi) = vi for all 0 5 i I n - 1 (where we define v, = w). Clearly, 50,
having vertices {w, ux, vx: x E X). If each x E X is identified with the path
carries the boundary path (p,, p,)(p,, p , ) . ..(p,-,, p,) onto the path a.
(w,uX)(uX, vX)(vX,w), then each relation in A may be regarded as a closed
Let K be a complex with basepoint w, let a be a closed path in K at w, and
path in B at w. For example, xyx-I is viewed as the path
let cp,: D(a) + K be the attaching map. Let 9 be the partition of the disjoint
union Vert(K) u Vert(D(a)) arising from the equivalence relation which (w, uX)(uX,
vX)(vX,
w)(w, uY)(uY,
vY)(vY,
w)(w, vX)(vX,
uX)(uX,
w).
identifies each pi with %(pi). Then the quotient complex Xa = ( K u D(a))/Y
For each path a a A, Bet the triangulated polygon D(a) have vertices
is called the comp1e:r obtained from K by a t a a c h i ~a 2-cell g2lloag a. Notice
(ra9p;, py, ...,q;, q;",. . .), and let qa: aD(a) --+ B be the attaching map. Let
that all the pi in D(a) have been identified with vertices of X, but the "interior"
D = VaE,D(a) (in which all the vertices p: are identified to a common base-
0
vertices r, q,, q,, ... are untouched. Thus, if is the full subcomplex of Ka
point, denoted by p,). If a(VaE,D(a)) is defined to be the full subcomplex
with vertices jr, q,, q,, ...), then & is simply connected. On the other hand,
with vertices all the pf, then there is a simplicial map q: a(V,,,D(a)) -+ B
if we delete the vertex r, and if Q is the full subcomplex with vertices
with pldD(a) = 9, for all a E A. Define K as the quotient complex of the
{qo, 41, - - - )thenn(Q,
, qo) r z. disjoint union B u D in which p, is identified with w and, for all a E A and all
i > 0, each vertex pq is identified with a ( p f ) . We have "draped" 2-simplexes
Theorem 11.63. Let a be a closed path at w in a complex K, and let Ka be
on a bouquet of circles. Thus,
obtained from K by attaching a 2-cell along a. Then

\ae A /
where N is the normal subgroup generated by [a].
Let T be the tree in K with vertices {w, ux: x E X}, let ea
be the full sub-
ProoJ Define L, to be the full subcomplex of Ka with vertices Vert(K)v complex with vertices {q:, qy, ...), and let @ be the full subcomplex with
{q,, ...,qn-,}, and define L, to be the full subcomplex with vertices {r, w, vertices {rayq:, q;, . ..). Define L, to be the full subcomplex with vertices
q,, ..., qn-,). Note that L, v L, = Ka and that L, n L, is the 1-complex Vert (B) u (U,, A Vert(P)), and define L, to be full subcomplex with vertices
with vertices {w, q,, .. .,qn-,). Since there is just one reduced circuit, it Vert(T) u (u Vert(&)). Each @ is simply connected, so that (with suitable
 11. Free Groups and Free Products
basepoints # ) n(L,, w) z n(Va@, #), by Theorem 11.34, and ?r(V @, #) 2
+a n(@, # ) = 1, by Theorem 11.54; thus, n(L,, w) = 1. We now show that
the inclusionj: B C+ L, induces an isomorphism n(B, v,) -+ n(L1, v,). Define
a function $: Vert(L,) -+ Vert(B) by $(v) = v for all v E Vert(B) and $(qq) =
(o,(pq) for all a and i. It is easy to see that II/ is a simplicia1 map and that
j$: L, -,L, is homotopic to the identity. Since n is a functor, the induced
map $, is the inverse of j,. Now L, u L, = K and L, n L, = T u ( V a p ) ,
so that n(L, n L,, w) E n(VaQa, # ) . :*a n ( p , # ) is free of rank IAl.
By Corollary 11.62, n(K, w) 2 n(L,, w)/N, where N is the image of
n(L, n L,,w). Therefore, n(K, w) r G, for they have the same presentation.
I central extension which allows z to be lifted:
Definition. A group G is finitely presented (or finitely related) if it has a
presentation with a finite number of generators and a finite number of
relations.
There are uncountably many nonisomorphic finitely generated groups, as
we shall see in Theorem 11.73. Since there are only countably many finitely
presented groups, there exist finitely generated groups that are not finitel
presented. We shall give an explicit example of such a group at the end of th
chapter.
Corollary 11.65. A group G is finitely presented fi and only fi there is a finite
connected 2-complex K (i.e., Vert(K) is finite) with G E n(K, w).
Pro05 By Tietze's theorem, n(K, W) is finitely presented. Conversely, if G
is finitely presented, then the complex M constructed in the theorem is
finite.
There is a construction dual to that of pushout. A sohsim of the diagram
A
I
(*)
B-G P
la
is a triple (D, a', p) making the following diagram commute:
Definition. A pull6oeh of diagram (*) is a solution (D, a', P ) that is "best" in
Amalgams
the following sense: if (X, a", p ) is any other solution, then there is a unique
homomorphism 0: X + D with a" = a' 0 and /3" = PO.
It is easy to see that if (D, a', p') is a pullback, then D is unique to isomor-
phism. One often abuses notation and calls the group D the pullback of
diagram (*). Some properties of pullbacks are given in Exercises 11.75 and
11.76 below.
Pullbacks have already arisen in the discussion of the projective lifting
property in Chapter 7: given a projective representation z: Q -+ PGL(n, C),
then the pullback U of z and the natural map n: GL(n, C) + PGL(n, C) is a
GL 7PGL.
11.71. Prove that every finitely generated free group is finitely presented.
11.72. Prove that every finite group is finitely presented.
11.73. Prove that every finitely generated abelian group is finitely presented.
11.74. Prove that a group having a presentation with a finite number of relations is
the free product of a finitely presented group and a free group.
11.75. Let A and B be subgroups of a group G, and let a and P be the respective
inclusions. Show that A n B is the pullback.
11.76. Prove that the pullbaclc of diagram (*) always exists. (Hint. Define B =
((a, b)JP(b)= a(a)) I A x 3, define a': (a, b) I-+ b, and define j3': (a, b) I-+ a.)
Amalgams
Amalgams arise from the special case of the van Kampen theorem in
which the maps ji,: n(L, nL,, w) -+n(Li, W ) induced from the inclusions
ji: L , n L, C+ Li (for i = 1,2) are injections. That an inclusion L ct K need
not induce an injection between fundamental groups can be seen when K is
simply connected and L is not. The advantage of this extra hypothesis, as
we shall see, is that a normal form is then available for the elements of the
pushout.
Definition. Let A, and A, be groups having isomorophic subgroups B, and
B,, respectively; let 8: B, -+ B, be an isomorphism. The amalgam of A, and
402 11. Free Groups and Free Products
A, over I9 (often called the fiee product with amalgamated subgroup) is the
pushout of the diagram
- A,
where i and j are inclusions.
Recall that pushouts are the diagrammatic version of unions.
We proved in Theorem 11.58 that the amalgam exists: it is (A, * A,)/,?/,
where N is the normal subgroup of A, * A, generated by (bO(b-I): b E 43,);
moreover, any two pushouts of the same data are isomorphic. We denote the
amalgam by
A, *e A2
(a less precise notation replaces the subscript I9 by the subscript B,). An
amalgam, as any pushout, is an ordered triple (A, *, A,, A,, A,); the maps
A,: A, + A, *, A, are given, as in Theorem 11.58, by a, I+ aiN.
It is clear that each b E B1 is identified in A, *, A, with e(b); it is not clear
whether other identifications are consequences of the amalgamation. For
example, is it obvious whether the maps A, are injections? Is it obvious
whether A, *, A, # I? These questions can be answered once we give a more
concrete description of an amalgam in terms of its elements.
For each i = 1,2, choose a left transversal of Biin A, subject only to the
condition that the representative of the coset B, is 1. For a E A,, denote the
chosen representative of aBi by l(a), so that a = l(a)b for some uniquely deter-
mined b E 23, (depending on a).
DeG~itiann.A normaBf~~rn
is an element of A, .i:A, of the form
d(a,)E(a,) .. . i(a,) b,
where b E B,, n 2 0, the elements l(aj) lie in the chosen transversals of Bij in
Aij, and adjacent l(aj) lie in distinct A,.
In the special case that B, (and hence B,) is trivial, the amalgam is the free
product and every reduced word is a normal form.
Theorem 11.66 (Normal Form). Let A, and A, be groups, let Bibe a subgroup
of A, for i = 1,2, and let 0: B, -+ B, be an isomorphism. Then for each element
wW E A, *, A,, there is a unique normal form F(w) with wM = F(w)N.
Proof. By Theorem 11.58, A, *, A, = (A, * A,)/N, where N is the normal
subgroup of A, * A, generated by [bI9(b-I): b E 3,). Each coset of N has a
representative w = x, y, ...xnynin the free product, where xi E A,, yi E A,,
Amalgams 403
and only x, or y, is allowed to be 1. We now give an algorithm assigning a
normal form F(w) to each coset wN in the amalgam such that F(w)N = wN.
Let w = x,y,. If y, = 1, then x, = l(x,)b = F(w), where b E B,, and we are
done. If x, = 1, then w = y, = l(y,)b for b E B, and so l(y,)I9-'(b) = F(w) is
a normal form in wN. If x, # 1 and y, # 1, then
but z = O(b)y, E A,, so that z = l(z)b, for some b, E B,. Therefore,
xlyl = l(xl)l(z)b2 = l ( ~ , ) l ( z ) 8 - ~ ( b ~ )in A, *, A,,
and the last element is a normal form F(w) in wN. This procedure can be
iterated, ending with a normal form. (Observe that the penultimate step
produces a last factor b lying in either B, or B,; if b E B,, one has a normal
form; if b E B,, then it must be replaced by 0-'(b) E B,.)
The van der Waerden trick will prove uniqueness of this normal form by
constructing a homomorphism b, with domain A, *, A, which takes different
values on different normal forms: if F(w) # F(wf), then (D(F(w)N)#
(D(F(wf)N).Let M be the set of all normal forms; by Theorem 11.52, different
normal forms have different spellings. If a E A,, define a function lal: M 4 A4
by
lal(@1)4a2). 4an)b) = F(al(alY(a2) - 4an)b)
(if a and l(a,) lie in distinct A,, then l(a,)l(a,) ... l(a,)b and also
al(a,)l(a,). ..l(a,)b have the form x,y, ...xnyn,and the algorithm F can be
applied; if a and l(a,) lie in the same A,, then the algorithm applies to
[al(a,) J E(a,) ... l(a,)b). Clearly 11I is the identity function on M, and
consideration of several cases (depending on where initial factors live) shows
that if a, a' E A, u A,, then
Therefore, la-' 1 = 1 al-' and each la1 is a permutation of M. If S, is the group
of all permutations of M , then a~ la1 is a homomorphism Ai -+ SM for
i = 1,2. In particular, if b E B, 5 A , , Ibl: M -t A4 is defined and (bl = lB(b)l.
The defining property of free product allows us to assemble these two
homomorphisms into a single homomorphism
with q(l(a,)l(a,). ..l(an)b)= ]l(a,)/ 0 Il(a,)l 0 . - -0 ll(a,)( 0 (bl. For all b E B,,
Ibl = /I9(b)/ gives bI9(b-') E ker q, and so q induces a homomorphism
iD: Al *, A2 = (A1 * A2)/N SM by
Now I l(a,)l 0 I l(a,)/ 0 - . o 1 l(a,)l 0 1 bl(1) = l(a,)l(a,) ...l(an)b; that is,
dD(wN): 1 -F(w). Thus, if G(w) were another normal form in wN, then
404 11. Free Groups and Free Products Amalgams

@(wN):1 f-, G(w), and so G(w) = F(w). Therefore, for each element wN of the If L, = (G, uAuF1) IK,, then
amalgam, there is a unique normal form F(w) with wN = F(w)N. !M

Theorem 11.67. Let A, and A, be groups, let B, and B, be isomorphic for there can be no equation in K,, a fortiori in L,, of the form
subgroups of A, and A,, respectively, and let 0: B, -,B, be an isomorphism. .gnu-lanu = 1.
g1~-1alug2u-1a2u..
(i) The (pushout)homomorphisms A,: A, -,A, *, A, are injectionsfor i = 1,2. Similarly, if L, = (G, uBv-l) 2 K,, then
(ii) If A: = /Zi(Ai),then (A;, A',) = A, *, A, and A; n A', = A,(B,) = A2(B2).

Proof. (i) If a, E A, is not 1, then F(a,) # 1 and @(a,N)# 1; but
@(aiN)= q/Zi(ai)# 1 implies Li(ai)# 1, and so A, is an injection.
(ii) It follows from A, *, A, = (A, * A,)/N that (A;, A',) = A, *, A,. If
u E A; n A',, then there are a, E A, with a, N = u = a, N. Now F(a,) = I(a,)b
and F(a,) = l(a,)b', so that the uniqueness of the normal form gives
l(a,) = l(a,) and b = b'. But l(a,) = l(a,) can occur only when both are 1, lest
we have equality of the distinct normal forms l(a,)l(a,) and l(a,)l(a,). Hence
bN = a,N = u = a,N = b'N, and u E A,(B,) = A,(B,). For the reverse
inclusion, it is easy to see that if b E B,, then bN E A; n A',. H
In view of the last theorem, it is customary to regard the elements of
A, *, A, as normal forms and the maps A, as inclusions. The statement of
Theorem 11.67 now simplifies to (A,, A,) = A, *, A, and A, n A, = B, =
B,. This is the point of view taken in the next corollary.
Corollary 11.68. Let E r A, *, A, have amalgamated subgroup B. If y,, ...,
y, E E with yj E Aij, where i, # i,,, ,and if yj # B for all j, then y, ...y, # 1.
Proof. Immediate from the normal form theorem.
Thea~sem11.69 (Torsiioira TheadreM. An element in A, *, A, has finite order if
and only fi it is conjugate to an element of finite order in A, or in A,.
ProoJ This follows easily from the normal form theorem. H
We are now going to apply amalgams to prove some imbedding theorems.
The following two theorems are due to G. Higrnan, B.H. Neumann, and
Hanna Neumann.
Theorem 11.70 (Hgman, Neumann, and Neumann, 1949). Let G be a group
and let 9:A B be an isomorphism between subgroups A and B of G. Then
-)
there exists a group H containing G and an element t E H with
q(a) = t-I at for all a E A.
Proof. Let (u) and (v) be (disjoint) infinite cyclic groups, and define groups
K,=G*(u) and K,=G*(v).
By Exercise 11.70, there is an isomorphism 8: L, -,L, with 0IG the identity
and 0(u-'au) = v-I q(a)v.
Define H = K, *, K,. By Theorem 11.67, H contains a subgroup isomor-
phic to L, (and G 5 L,). For each a E A, u-lau = v-lq(a)v, so that if t E H is
defined by t = uv-I, then
t-'at = q(a) for all a E A. ssll
If G is a countable group, then G is a homomorphic image of a free group
F of countable rank: G r FIR. By Theorem 11.48, F can be imbedded in a
free group F* of rank 2. Were R a normal subgroup of F* (it is not!), then
G = FIR would be imbedded in a group F*/R having two generators. This
proof is fictitious, but the theorem is true.
Theorem 11.71 (Higman, Neumann, and Neumann, 1949). Every countable
group G can be imbedded in a group H having two generators.
Remark. This follows from Exercise 3.28 when G is finite.
Pro05 Let go = 1 and let go, g,, ..., g,, ... be a list of all the elements of G.
Let H = G + F, where F is free with basis (x, y). Consider the subgroups of
N:
A = (y, x-I yx, .. . , x-nyxn, ... )
and
B = (x, 9, y-'xy, ..., g,y-"xyn, ...).
Now A is free with basis the displayed generators, by Exercise 11.45, and the
map q : A -+ B given by
q:~-"yx"t-,g,y-~xy" forall n > 0
is easily seen to be an isomorphism. Theorem 11.70 gives a group Hn
containing H and an element t E H" such that
q(a) = t-I at for all a E A.
We claim that (y, t) I H" contains G, and this will-complete the proof. Now
x = q(y) = t-'yt E (y, t). Moreover, t-'x-"yxnt = q(x-"yxn) = gny-nxyn,
and this shows that g, E (x, y, t) = (y, t) for all n > 1. lBss
406 11. Free Groups and Free Products
Corollary 111.72. If G is a countable group, then there is a Zgenerator group E
containing it such that, for all n 2 1, E contains an element oforder n if and
only if G contains an element of order n.
Proof. Observe, in the proof of Theorem 11.71, that the Zgenerator group
containing G is obtained in two steps. First, we formed the amalgam H =
G *, F, where F is free. By the torsion theorem (Theorem 11.69), the only
integers n which are orders of elements of finite order in H are those arising
from G. The second step uses Theorem 11.70, where the ultimate group is a
subgroup of an amalgam K, *, K,, where each Ki H * Z,and the torsion
theorem applies again. H
Theorem 11.73 (B.H. Neumann, 1953). There are uncountably many non-
isomorphic finitely generated groups.
Proof (Schupp). If S is a set of primes, define G(S) = x,,,
B,;as in Corollary
11.72, let H(S) be a Zgenerator group containing G(S) which has an element
of prime order p if and only if p E S. It follows that if T is a subset of the
primes, then T # S implies H(T) $ H(S), for there is a prime p in T that is
not in S (or vice versa); thus H(T) has an element of order p and H(S) does
not. As there are uncountably many subsets of the primes, there are thus
uncountably many nonisomorphic 2-generator groups.
11.77 (Schupp). Prove that Corollary 11.68 implies the normal form theorem (Theo-
rem 11.66).
11.78. For every torsion-free group G, there exists a (necessarily simple) group H
containing G which has exactly two conjugacy classes. (Compare Exercise 3.4.)
11.79. Prove that there exists a 2-generator group G which contains an isomorphic
copy of every countable abelian group. (Hint.Exercise 10.31.)
11.80. Prove that there is a Zgenerator group containing an isomorphic copy of
every finite group.
11.81. Prove that a finitely presented group can be imbedded in a finitely presented
group having two generators.
11.82. Consider the diagram
If A , and A, are finitely presented and if B is finitely generated, then the
pushout is finitely presented.
HNN Extensions
HNN Extensions
There is another construction, closely related to amalgams, that we will use
in the next chapter to prove that there exists a finitely presented group having
unsolvable word problem.
We adopt the following notation. If a group G has a presentation (XI A),
then
GR = (G: YIA')
denotes the group with presentation (X u YI A u A'), where it is understood
that X and Y are disjoint. In particular, if Y = 0, then we are merely
adjoining additional relations to G, so that GR is a quotient of G. For
example, in the notation of Theorem 11.58, the pushout P has the presenta-
tion (B * CJi(a)j(a-'), a E A).
Definition. Let G be a group with isomorphic subgroups A and B, and let
q: A -+ B be an isomorphism. Then the group having the presentation
(G; pl p-' ap = q(a) for all a E A)
is called an HNN extension of G; it is denoted by G a, A or, less precisely, by
G n A. The group G is called the base and the generator p is called the stable
letter of G a, A.
The next theorem shows that HNN extensions appear in Theorem 11.70 of
Higman, Weumann, and Weumann.
Theorem 8 8.74. The subgroup ( 6 ,t ) 5 K , *, K , in Theorem 11.70 is an NMN
extension with base G and stable letter t.
BDuooJ Let us recall the notation of Theorem 11.70. Begin with a group G and
two subgroups A and B isomorphic via an isomorphism cp: A -+ B. Let K, =
G * (u), K , = G * (v), L, = (G, u-'Au) _< G * (u), L2 = (G, v-'Bv) _<
G * (v), and 8: L, -+ L2 the isomorphism which carries each g E G into itself
and which sends u-lau into v-'q(a)v. If a presentation of G is (XIA), then
Theorem 11.58(ii) says that a presentation of the amalgam K, *, K, is
(X, u, vl A, u-' au = v-' q(a)u for all a E A).
As t = uu-l, we see that a presentation for (G, t) is
(X, tlA, t-'at = q(a) for all a E A);
that is, (G,-t) is an HNN extension with base G and stable letter t. IH
 11. Free Groups and Free Products
HNN Extensions 409

Note the resemblance between amalgams and HNN extensions. Both be-
(a x 0, b x 0), (a x 1, b x I), and (a x 0, b x 1). It follows easily that A x I
gin with a pair of isomorphic subgroups; the amalgam is a group in which the
is connected if A is connected.
two subgroups are made equal; the HNN extension is a group in which the For example, the cartesian product 1 x I has Zsimplexes pictured below.
two subgroups are made conjugate. This observation is important in further
study (see Dicks and Dunwoody, Lyndon and Schupp, Serre (1980), and
Stallings).
Here is a geometric context in which HNN extensions arise. Consider a
connected topological space X with homeomorphic disjoint subspaces A and
B, and let cp: A B be a homeomorphism. We are going to "add a handlem
-+

to X. Define a new space Xi2 as the quotient space of the disjoint union
X u (A x I) (where I is the closed unit interval) by identifying each a E A with
(a, 0) and each q(a) with (a, 1). The picture is as shown in Figure 11.8. Remark. If a = (w, a,). ..(a,, w) is a closed path in A, then a x 0 and a x 1
This construction can be carried out for complexes: the role of the closed are closed paths in A x I, where a x i = (w x i,a, x i) ...( a, x i, w x i) for
unit interval I is played by the l-simplex (also denoted by I) having two i E (0, 1). Moreover, if p is the edge (w x 1, w x 0), then
vertices 0 and 1. If we denote the vertices in Vert(A x I) by a x 0 and a x 1,
where a E Vert(A), then A x I is made into a complex by "triangulating" it: if
+
dim(A) = n, then dim(A x 1) = n + 1 and its (n 1)-simplexes are defined to
be all subsets of the form It is now easy to show that the injection j: A -+ A x I, given by a- a x 0,
induces an isomorphism
(a, x O,..., a i x O , a i x 1,..., a , I),
~
j,: n(A, w) 7 n(A x I, w x 0).
where (a,, ...,a,) is an n-simplex in A and 0 < i < n. Notice that if (a, b)
is a l-simplex in A, then A x 1 has edges (a x 0, a x I), (b x 0, b x I), Definition. Let A and B be disjoint subcomplexes of a connected complex
K , and let 9 : A -+ B be an isomorphism. The complex obtained by adding
a handle to I< is
Ki2= (K u (A.x 1[))/9',

where 9'identifies a E Vert(A) with a x 0 and q(a) with a x 1.

Theorem 11.75. Let K be a connected complex with disjoint isomorphic sub-

complexes A and B; let w E Vert(A) be a basepoint, and let 9: A -+ B be an
isomorphism. If KR is obtained from K by adding a handle according to this
data, then n(KR,w) is an H N N extension with base n(K, w).

Remark. It is this result that suggests the notation G Q A for HNN exten-
sions.

Proof. Since K is connected, there is a path y in K from w to q(w); since A,

hence A x 1, is connected, there is a path ,8 in the handle from q(w) to w.
Define H as the union of y and the handle (regard a path as the l-complex
consisting of its edges and their vertices); note that K u H = K~ and
K n H = AvBuy.
Figure 11.8
The van. Kampen theorem (which applies because K n H is connected)
 11. Free Groups and Free Products
Figure 11.9
shows that n(Kn, w) is the pushout of the diagram
Since A n (B u y) = y, which is simply connected, Corollary 11.61 gives
n(A u B u y, w) z n(A, w) * n(B u y, w). Wow H = y@ u (W - y), so that
n(H, W)z n(H - y, w) * n(yP, w), by Corollary 11.41. As A and B are disjoint,
iH - y z A x I, where A is identified with A x 0 and B is identified with
A x 1. Thus, each closed path a at w in A is identified with the closed path
a x 0 at w x 0. By the remark above,
loop,
-
where y(a x 1)y-' is a closed path at w x 0 in B u y . But a x 1 = cp,a (be-
, that cr = a x 0 (yP)-' ycp,ay-'(y/3). Since y@ is a
cause a = cp(a) in K ~ )SO
n(H, W)2 n(H - y, W)* n(yp, W)2 n(A x I, W)* (t),
where t = [yP] is a generator of the infinite cyclic n(yp, w). Finally, the injec-
tion j: a H a x 0 induces a isomorphism
j,: n(A, W)z n(A x I, w x 0).
Thus, n(H, W)2 n(A, W)* (t), and [a] = t-' cp, [a] t, where cp,: n(A, w) -+
n(B u y, w) is induced by the isomorphism cp. Now use the presentation given
in van Kampen's theorem. ! I E * F is ~ ~ H extension
HNN Extensions 411
Notice that the definition of HNN extension involves two isomorphic sub-
groups A and B of a group G. In contrast to the geometric situation in
Theorem 11.75, there is no hypothesis that the subgroups be disjoint in any
sense; indeed, A = B is allowed.
We are going to generalize the definition of HNN extension so that it
involves a set of stable letters.
Definition. Let a group E have a presentation (XIA), and let {pi: i E I ) be a
nonempty set disjoint from X. Assume that there is an index set J and, for
each i E I, there is a family (aij, b,,: j E J) of pairs of words on X. Then the
group E* has a presentation with base E and stable letters {pi: i E I ) if
En = (E; pi, i E I[p;'aijpi = bij for all i, j).
We allow a,, and b,, to be 1 so that the number of "honest" relations
involving pi (i.e., both av and bij distinct from 1) may be distinct from the
number of honest relations involving p, for some k # i.
Lemma 11.76. If En has a presentation with base E and stable letters
{pi: i E I), then (pi: i E I) IEn is a free group with basis {pi: i E I).
Proof. Let (zi: i E I} be a set and let F be the free group with basis {zi:i E I).
If E has the presentation E = (XlA), then En has the presentation
En = (X, pi, i E I I A, p;' a,,pi = b,, for all i, j).
Define a homomorphism cp: En -+ F by q(x) = 1 for all x E X and cp(pi) = zi
for all i; note that cp is well defined because it sends all the relations into 1.
The lemma now follows from Exercise 11.8. BI
Nsta~sn.Let IZR have a presentation with base E and stable letters {pi: i E I ) .
For each i, define two subgroups of E:
Ai = (av: j E J) and Bi = (b,,: j E J);
now define
A=(UA~) and B=(UB~).
Definition. A group E* having a presentation with base E and stable letters
{pi: i E I ) is an HNN extension if, for each i, there is an isomorphism
cpi: Ai Bi with cpi(aij) = bij for all j.
-+
If there is only one stable letter p, then we can recapture the original
definition of HNN extension by setting {aj:j E J ) = A (there is now no need
to have the first index i) and bj = cp(aj) for all j.
EXAMPLE 11.10. If E is a group and F is a free group with basis {pi: i E I ) then
N N of E with stable letters (pi: i E I).
412 11. Free Groups and Free Products
This is the "trivial" example in which all aij and bij are 1; of course, ER =
E x (t) r E x Z in this case.
EXAMPLE 11.11. I f E = (X1A)and E" = (E;t/t-'wit = wi, ~ E I )where
, themi
are words on X, the E" is an HNN extension with base E and stable letter t.
In this case, A =B = ( a i , i E I), and the isomorphism p: A -,B
identity.
EXAMPLE 11.12. Let E be the free group with basis {w, x), and let E" have the
presentation
E" = (w, x, y, zly-lxy = w, y-lw-'xwy = xw-', z-'wxz = w).
Now E" has a presentation with base E and stable letters {y, z). The various
subgroups are: A, = (x, w-'xw); By = (w, xw-I); AZ = (wx); 4 = (w).
There is an isomorphism q,,:A, By with x F+ w and w-'xw F+ xw-l,
-+
because both groups are free with bases the displayed generating sets; there is
also an isomorphism cp,: A, B, with w x w,~because both A, = (wx) and
-+
B, = (w) are infinite cyclic. It follows that En is an HNN extension with base
E and stable letters {y, z).
There are two natural questions about an HNN extension E" of E. Is E
imbedded in E"? Is there a normal form for the elements of E"?
The next lemma shows that an HNN extension with several stable letters
can be viewed as the end result of a sequence of HNW extensions, each
involving a single stable letter.
Eernms 1137. If E" is an HNN extension of E with stable letters {p,, . . . ,p,),
then there is a group E7 which is an HNN extension of E with stable letters
(pl, ...,pn-,) such that E" is an HNN extension of Eb with stable letter p,,.
PmoJ Define
E~ = ( E ; p l ,..., pn-llp;laijpi = bij, ~ El
1 5 i s n - 1 , j J).
Theorem 11.78. If E" is an HNN extension with base E and stable letters
( p i , . ..,pn), then E can be imbedded in E". In particular, i f A and B are
isomorphic subgroups of a group E and if p: A -+ B is an isomorphism, then
E 5 E a, A.
Proof. We prove the theorem by induction on the number n of stable letters.
If n = 1, the result follows from Theorem 11.74. For the inductive step, the
lemma gives E" an HNN extension with base E b and stable letter p,. The
inductive hypothesis gives E < Eb, and the step n = 1 gives Eb < E". ld
HNN Extensions 413
Corollary 11.79. If K is a connected complex with basepoint w and if Kn is
obtained from K by adding a handle, then
n(K, w) 5 .n(K", w).
Here is a sharper version of Theorem 11.71.
is the Corollary 11.80. Every countable group G can be imbedded in a 2-generator
group H = (t, x) in which both t and x have infinite order.
Proof. Let F be a free group with basis (x, y). Enumerate the elements of
',
G: go = I, g ...,gn, . ..,and define
H = (G * F; tlt-'x-"yxnt = gny-"xyn,n 2 0).
The group H thus has a presentation with base G + F and stable letter t. In
the proof of Theorem 11.71, we saw that
A = (y, x-' yx, ...,x-"yxn, ...)
and
B = (x, gly-lxy, ..., g,y-"xyn, ...)
are each free with bases the displayed generators; there is thus an isomor-
phism cp: A -+B with cp(x) = y and cp(x-"yxn) = gny-"xyn for all n 2 1.
Therefore, H is an HNN extension of G * F, so that G G * F < H, and
G I(t, x) IH, as in the proof of Theorem 11.71. Finally, x E F < H has
infinite order, for F is free, while t has infinite order, by Lemma 11.76.
Notation. Let F be a free group with basis X , let N be a normal subgroup of
F, and let G = FIN. If w and w' are (not necessarily reduced) words on M,
then we write
w = of in G
if wN = o'M in F/N. We write
w r w'
if w and w' have exactly the same spelling.
For example, if X = (a, b, c), w = ac and w' = abb-'c, then w = ofin G
(even in F), but w f w'.
In this new notation, if w is a word on X = {x,, ...,x,), then a word p is
a subwovd of w if there are possibly empty words a and y with w = spy. A
word w involves xi if either xi or x i ' is a subword of w.
Definition. Let ER be an HNN extension with base E = (XIA) and stable
letters (pi: i E I). A pinch is a word of the form
414 11. Free Groups and Free Products HNN Extensions 415

where g is a word on X such that the element pfgp;' lies in Ai I

or the element pf gpie lies in Bi < B if e = + 1.
A if e =

Theorem 11.$1 (Britton's Lemma, 1963). Let En be an H N N extension with

base E = (XIA) and stable letters {pi: i E I). If o is a word with o = 1 in EQ
-1
-
element of M = (E, v-'Bv) only when gi lies in B (in the proof of Theorem
11.70, we saw that (E, v-'Bv) E * v-'Bv); of course, v-'giv arises in o from
the subword u~-'g,vu-~ pgip-l; that is, from the pinch pegip-" with e =
+ 1. A conjugate u-'giu arises from an element of H = (E, u-'Au) only
when gi lies in A; of course, umlgiuarises from vu-'g,uv-' = p-'gip; that is,
and which involves at least one stable letter, then o has a pinch as a subword. from the pinch pegip-" with e = - 1.
The general case follows by induction. As in the proof of Lemma 11.77,
Proof (Schugp). Assume first that there is only one stable letter p, = p, so that define
the presentation of En is E b = (E;pl,..., p,-,jp;laijpi = bij, 1 I i I n - 1 , j J),
~
E~ = (x, pIA, p-'ajp = bj, j E J). and note that En is an HNN extension with base E~ and stable letter p,. If
o = 1does not involve p,, then it contains a pinch pfgpie for some i < n and
Let A and B be isomorphic subgroups of E, and let cp: A -,B be an isomor- induction completes the proof. If i = n, then the base step of the induction
phism, so that the presentation may be rewritten gives a pinch p,'gpie as a subword of a.If g is a word on X, we are done. If
E~ =-(X,pl A, p-I ap = cp(a), a E A). g involves some p,, ...,p ,-,, then there is a word g' on X, namely, a word on
b,, j E J , with g = g1in ~ b Thus,
. gg'-l = 1 in Eb, and induction provides a
We may assume that pinch n which is a subword of ggr-'. Since gr-' contains no stable letters, n
o = g0pe1g1...pengn, must be a subword of g, as desired. fil
+
where n 2 1, each ei = 1, and the gi are (possibly empty) words on X.
As in Theorem 11.67, we view the amalgam K, *, K2 (where K, = Definition. Let En be an HNN extension with base E = (XIA) and stable
E * (u), K, = E * (v)) as having amalgamated subgroup H = K, nK2 = letters {p,, . . .,p,). A word o on X v (p,, ...,p,) is pi-reduced for some i if it
(E, u-~Au)= (E, v-'Bv) satisfying u-'au = ~ - ' ~ ( a ) for
v all a E A. By contains no pinch pf wpie as a subword.
Theorem 11.75, E" can be identified with (E, p) I K , *, K,, where p = uv".
We are going to use the normal form theorem for elements in amalgams, as Note, in particular, that pi-reduced words contains no subwords of the
given in Corollary 11.68. form pfpie.
Return to the word o.Since p = uv-l,
Cor~111ary14.82. Let En be an HMN extension with base E = (%/A) and stable
letters (p,, . . ., p,). Asswine that
where gi E E. Reassociate o according to the following instructions: assume
that el, ...,ei have the same sign, but that e,,, has opposite sign. If el > 0, a = yopfly, ...pfmym and /IB a0pf 16, ...p(.an
reassociate arc pi-reduced words for some i, where each ej, f;, = fl and none of the (possi-
g O ( ~ ~ - l ) e(uu-ly'.
'gl ..(uv-I p g i ( ~ ~ - l p + l bly empty) words yj or 6, involve pi.
If a = P in E ~ then, rn = PI, (el, ..., em)= (f,,..., f,), and the word
pfmym6,-1pifn is a pinch.

if el < 0,interchange the symbols u and v in the last expression; continue this Proof. Since ap-' = 1 in E,' Britton's lemma says that the word a/?-'
rewriting process if ei+,,...,ekhave the same sign and e,,, has opposite sign, contains a pinch as a subword. As each of a and P (hence P-') are pi-reduced,
until all of o has been reassociated according to this scheme. Note that there the pinch must occur at the interface; that is, the subword pPym',sn-'p;fn is a
are conjugates v-'giv or u-'g,u wherever exponents change sign, and that pinch. It follows that the exponents emand -fn have opposite sign, and so
adjacent factors (in the new association) lie in distinct K,. Since o = 1 in the
amalgam, by hypothesis, Corollary 11.68 says that at least one of the factors em = f,.
The proof is completed by induction on maxim, n). In the pinch
must lie in the amalgamated subgroup H. Now the factors uv",vv",giu, giv, pfmymd;lP;fn, Britton's lemma says that
+
where E = 1, do not lie in H = (E, uW1Au)= (E, BV),because the sum
of the exponents of u or of v in any such word is not 0. Therefore, one of the
conjugates v-'giv or u-'giu lies in H. A conjugate v-'giv arises from an
416 11. Free Groups and Free Products HNN Extensions 417

where aGE A,, bGE B,, and h, = 1. If em= - 1, pn suffice to prove all $(ai) = 1. For notational convenience, let us denote
these by p2, . . p,.
- 9

Since q(pn)= 1 in G, for each n,

We have eliminated one pi from a and one from 8, and so the remainder of
the proof follows from the inductive hypothesis. The proof in the other case
em= + 1 is similar.
The normal form theorem for HNN extensions has its most succinct state-
ment in the special case when there is only one stable letter; the statement
and proof of the generalization for arbitrary HNW extensions is left to the
reader.
Theorem 111.83 (Normal Form). Let ER be an H N N extension with base
E = (X IA) and stable letter p. Then each word w on (X, p) is equal in En to a
p-reduced word
o0pe1o1...penon;
moreover, the length n and the sequence (el, ..., en) of exponents are uniquely
determined by o.
Proof. If w contains a pinch 71 as a subword, then the relations in En allow
one to replace 71 by a subword involving two fewer occurrences of the stable
letter p. The uniqueness of the length and exponent sequence follow at once
from Corollary 11.82.
It follows from Theorem 11.73 that there exist finitely generated groups
that are not finitely presented, for there are only countably many finitely
presented groups. Here is an explicit example of such a group.
(~(pn)
=~n(al, . .- , q),
where onis a word on conjugates of a,, a;, ...,at. Therefore,
But this equation says that every relation pn lies in the normal subgroup
generated by p,, p,, . . .,p,, which is what is to be proved. E i
The following explicit example was found by W.W. Boone.
Theorem 11.85. Let F be the free group with basis (a, b), and let its comrnuta-
..., o n ,...). Then the group G having
tor subgroup F' be free with basis {a,,
the presentation
G = (a, b, plp-lonp = o n ,n 2 1)
is a finitely generated group that is not finitely presented.
Proof.Recall first that Theorem 11.48 shows that F' is a free group of infinite
rank. Were G finitely presented, then the lemma would say all but a finite
number of the relations could be deleted; that is, there is some N with
G = (a, b, p(p-'o,p = o,, ..., p-'co,-,p = oNdl).
The displayed presentation exhibits G as an MNN extension with base F =
(a, b) and stable letter p. Since p-lo,p = w, in G. Britton's lemma provides
a word on (o,, . . . ,co,-,) with o, = p in P; = (a, b), and this contradicts
the fact that (o,, . .., o,) is a free group with basis (co,, . . . , o,). g

Lemma 11.84. If a group G has a presentation 11.83. Let a E G have infinite order. Show that there is a group H containing G in
which ( a ) and (a2) are conjugate. Conclude that a conjugate of a subgroup S
(XI,. . - , x ~ I P I~, 2..-,
, Pn, .-.) may be a proper subgroup of S.
as well as a finite presentation 11.84 (Schupp). Use the normal form theorem for HNN extensions to prove the
normal form theorem for amalgams.

then all but a finite number of the relations pn = 1 in the first presentation are
superfluous.

Proof. Let G, be the group defined by the first presentation, let G, be the
group defined by the second presentation, and let cp: G, -+ G2 be an isomor-
phism with inverse $. Since +(ai) = 1 in G, for each i, it is a word on (finitely
many) conjugates of pn's; as there are only finitely many a,, only finitely many

CHAPTER 12
The Word Problem
Introduction
Novikov, Boone, and Britton proved, independently, that there is a finitely
presented group B for which no computer can ever exist that can decide
whether an arbitrary word on the generators of B is 1. We shall prove this
remarkable result in this chapter.
Informally, if 9 is a list of questions, then a decisionprocess (or algorithm)
for 9 is a uniform set of directions which, when applied to any of the ques-
tions in 9 , gives the correct answer "yes" or "no" after a finite number of
steps, never at any stage of the process leaving the user in doubt as to what
to do next.
Suppose now that G is a finitely generated group with the presentation
every (not necessarily reduced) word o on X = {x,, ...,x,) determines an
element of G (namely, oR, where F is the free group with basis X and R is
the normal subgroup of F generated by {q,j 2 1)). We say that G has a
solulrble wordproblem if there exists a decision process for the set 9 of all
questions of the form: If o is a word on X, is w = 1 in G? (It appears that
solvability of the word problem depends on the presentation. However, it can
be shown that if G is finitely generated and if its word problem is solvable for
one presentation, then it is solvable for every presentation with a finite num-
ber of generators.)
Arrange all the words on {x,, ...,x,] in a list as follows: Recall that the
+
length of a (not necessarily reduced) word o = xyl.. .xzm, where ei = 1, is
m. For example, the empty word 1 has length 0, but the word xx-I has length
2. Now list all the words on X as follows: first the empty word, then the
Introduction 419
words of length 1 in the order x, ,x;', ..., x,, xi1, then the words of length 2
in "lexicographic" order (as in a dictionary): x,x, < x,x;' < x,x, < ..-<
xy1xl < xf1xy1 < ... c x i l x i l , then the words of length 3 in lexicographic
order, and so forth. Use this ordering of words: o,, u,, o,, . .. to define the
list 9 whose kth question asks whether o,= 1 in G.
We illustrate by sketching a proof that a free group
has a solvable word problem. Here is a decision process.
1. If length(@,) = 0 or 1, proceed to Step 3. If length(cok)2 2, underline the
first adjacent pair of letters, if any, of the form xix;' or x;'xi; if there is no
such pair, underline the final two letters; proceed to Step 2.
2. If the underlined pair of letters has the form xixyl or xilxi, erase it and
proceed to Step 1; otherwise, proceed to Step 3.
3. If the word is empty, write co, = 1 and stop; if the word is not empty, write
ok# 1 and stop.
The reader should agree, even without a formal definition, that the set of
directions above is a decision process showing that the free group G has a
solvable word problem.
The proof of the Novikov-Boone-Britton theorem can be split in half.
The initial portion is really Mathematical Logic, and it is a theorem, proved
independently by Markov and Post, that there exists a finitely presented
semigroup S having an unsolvable word problem. The more difficult portion
of the proof consists of constructing a finitely presented group B and show-
ing that if had a solvable word problem, then S would have a solvable
word problem. Nowhere in the reduction of the group problem to the semi-
group problem is a technical definition of a solvable word problem used, so
that the reader knowing only our informal discussion above can follow this
part of the proof. Nevertheless, we do include a precise definition below.
There are several good reasons for doing so: the word problem can be prop-
erly stated; a proof of the Marltov-Post theorem can be given (and so the
generators and relations of the Markov-Post semigroup can be understood);
a beautiful theorem of C.Higman (characterizing the finitely generated sub-
groups of finitely presented groups) can be given. Here are two interesting
consequences: Theorem 12.30 (Boone-Higman): there is a purely algebraic
characterization of groups having a solvable word problem; Theorem 12.32
(Adian-Rabin): given almost any interesting property P, there is no decision
process which can decide, given an arbitrary finite presentation, whether or
not the presented group enjoys P.
12.1. Sketch a proof that every finite group has a solvable word problem.
12.2. Sketch a proof that every finitely generated abelian group has a solvable word
420 12. The Word Problem
problem. (Hint. Use the fundamental theorem of finitely generated abelian
groups.)
12.3. Sketch proofs that if each of G and H have a solvable word problem, then the
same is true of their free product G * H and their direct product G x H.
12.4. Sketch a proof that if G = (x,, ..., x,lrj = 1,j 2 1) has a solvable word problem
and if H is a finitely generated subgroup of G, then H has a solvable word
problem. (Hint. If H = (h,, ..., h,), write each hi as a word in the x.)
Turing Machines
Call a subset E of a (countable) set Q "enumerable" if there is a computer that
can recognize every element of E and no others. Of course, the nature of such
a well-behaved subset E should not depend on any accidental physical con-
straints affecting a real computer; for example, it should not depend on the
number of memory cells being less than the total number of atoms in the
universe. We thus define an idealized computer, called a Turing machine, after
its inventor A. Turing (1912-1954), which abstracts the essential features of a
real computer and which enumerates only those subsets E that, intuitively,
"ought" to be enumerable.
Informally, a Turing machine can be pictured as a box with a tape running
through it. The tape consists of a series of squares, which is as long to the left
and to the right as desired. The box is capable of printing a finite number of
symbols, say, so, s,, ...,s,, and of being in any one of a finite number of
states, say, q,, q , , ...,q,. At any fixed moment, the box is in some state qi as
it "scans" a particular square of the tape that bears a single symbol sj (we
agree that so means blank). The next move of the machine is determined by
q, and sj and its initial structure: it goes into some state q, after obeying one
of the following instructions:
1. Replace the symbol sj by the symbol sk and scan the same square.
2. Move one square to the right and scan this square.
3. Move one square to the left and scan this square.
The machine is now ready for its next move. The machine is started in the
first place by being given a tape, which may have some nonblank symbols
printed on it, one to a square, and by being set to scan some one square while
in "starting state" q,. The machine may eventually stop (we agree that qo
means "stop"; that is, the machine stops when it enters state q,) or it may
continue working indefinitely.
Here are the formal definitions; after each definition, we shall give an infor-
mal interpretation. Choose, once and for all, two infinite lists of letters:
s0,s1,s2,... and qo,q,,q2,....
Turing Machines
Definitionm. A quadruple is a Qtuple of one of the following three types:
qisjRq~,
qisjLq,.
A Turing machine T is a finite set of quadruples no two of which have the
same first two letters. The alphabet of T is the set (so, s,, . . .,s,) of all
s-letters occurring in its quadruples.
The three types of quadruples correspond to the three types of moves in
the informal description given above. For example, qisjRq, may be inter-
preted as being the instruction: "When in state qi and scanning symbol sj,
move right one square and enter state q,." The "initial structure" of the
Turing machine is the set of all such instructions.
Recall that a word is positive if it is empty or if it has only positive
exponents. If an alphabet A is a disjoint union S u T, where S = IS,: i E I),
then an s-word is a word on S.
Definition. An instantaneous description a is a positive word of the form a =
aqiz, where a and z are s-words and z is not empty.
For example, the instantaneous description a = s2soq,s,s,s2 is to be inter-
preted: the symbols on the tape are s2s,s5s2s,, with blanks everywhere else,
and the machine is in state q , scanning s, .
Defimifi0~1.Let T be a Turing machine. An ordered pair (a, P) of instanta-
neous descriptions is a basic moue of T, denoted by
CI p,
if there are (possibly empty) positive s-words a and a' such that one of the
following conditions hold:
(i) a = aqisjol and p = oq,skal,where qisjskq,E T;
(ii) a = aqisjskafand /? = asjqzskd,where qisjRqlE T;
(iii) a = aqisj and /? = asjqrs0,where qisjRq, E T;
(iv) a = askqisjaland /? = aq,s,sjd, where qisjLq, E T; and
(v) a = qisjafand /? = q,s0s,a1,where qisjLq, E T.
If a describes the tape at a given time, the state qi of T, and the symbol sj
being scanned, then P describes the tape, the next state of T, and the symbol
being scanned after the machine's next move. The proviso in the definition
of a Turing machine that no two quadruples have the same first two symbols
422 12. The Word Problem
means that there is never ambiguity about a machine's next move: if a -+ P
and a -+ y, then /?= y.
Some further explanation is needed to interpret basic moves of types (iii)
and (v). Tapes are finite, but when the machine comes to an end of the tape,
the tape is lengthened by adjoining a blank square. Since so means blank,
these two rules thus correspond to the case when T is scanning either the last
symbol on the tape or the first symbol.
Definition. An instantaneous description a is terminal if there is no instanta-
neous description P with a P. If o is a positive word on the alphabet of T,
-+
then T computes o if there is a finite sequence of instantaneous descriptions
a, = q,o, a,, . .., a,, where ai -+ a,,,, for all i I t - 1, and a, is terminal.
Informally, o is printed on the tape and T is in starting state q, while
scanning the first square. The running of T is a possibly infinite sequence of
instantaneous descriptions q, co -,a, + a, -+ - . . This sequence stops if T
computes o ; otherwise, T runs forever.
Definition. Let S2 be the set of all positive words on symbols S = (s,, ...,s,).
If T is a Turing machine whose alphabet contains S, define
e(T) = ( o E Ck T computes o ) ,
and say that T enumerates e(T). A subset E of S2 is r.e. (recursively
enumerable) if there is some Turing machine T that enumerates E.
The notion of an r.e. subset of Q can be specialized to subsets of the natural
numbers N = (n E Z:n 2 0) by identifying each n E N with the positive word
s;". Thus, a subset E of N is an roe.subset sf N if there is a Turing machine
T with s, in its alphabet such that E = {n E N: T computes s;+l).
Every Turing machine T defines an r.e. subset E = e ( T ) c R, the set of all
positive words on its alphabet. Wow can we tell whether o E R lies in E? Feed
q, o into T and wait; that is, perform the basic moves q, o -+ a, a, -+ .- .
-+
If o E E, then T computes o and so T will eventually stop. However, for a
given o , there is no way of knowing, a priori, whether T will stop. Certainly
this is unsatisfactory for an impatient person, but, more important, it
suggests a new idea.
Definition. Let Cl be the set of all positive words on (so, s,, .. ., s,). A subset
E of i2 is recursive if both E and its complement SZ - E are r.e. subsets.
If E is recursive, there is never an "infinite wait" to decide whether or not
a positive word o lies in E. If T is a Turing machine with e(T) = E and if T'
is a Turing machine with e(T1)= Q - E, then, for each o E Q, either T or T'
computes o. Thus, it can be decided in a finite length of time whether or not
Turing Machines 423
a given word o lies in E: just feed o into each machine and let T and T' run
simultaneously.
Recall the informal discussion in the introduction. If 2 is a list of ques-
tions, then a decision process for 2 is a uniform set of directions which, when
applied to any of the questions in 9, gives the correct answer "yes" or "no"
after a finite number of steps, never at any stage of the process leaving the
user in doubt as to what to do next. It is no loss of generality to assume
that the list 2 has been encoded as positive words on an alphabet, that E
consists of all words for which the answer is "yes," and that its complement
consists of all words for which the answer is "no." We propose that recursive
sets are precisely those subsets admitting a decision process. Of course, this
proposition (called Church's thesis) can never be proved, for it is a question
of translating an intuitive notion into precise terms. There have been other
attempts to formalize the notion of decision process (e.g., using a Turing-like
machine that can read a two-dimensional tape; or, avoiding Turing machines
altogether and beginning with a notion of computable function). So far, every
alternative definition of "decision process" which recognizes all recursive sets
has been proved to recognize only these sets.
Theorem 12.1. There exists an r.e. subset of the natural numbers N that is not
recursive.
Proof. There are only countably many Turing machines, for a Turing
machine is a finite set of quadruples based on the countable set of letters
(R,L, so, s, , . ..;q,, q, ,...). Assign natural numbers to these letters in the
following way:
If T is a Turing machine having rn quadruples, juxtapose them in some order
to form a word w(T) of length 4m; note that T # T' implies w(T) # w(Tf).
Define the Godel number
where pi is the ith prime and ei is the natural number assigned above to the
ith letter in w(T). The Fundamental Theorem of Arithmetic implies that
distinct Turing machines have distinct Godel numbers. All Turing machines
can now be enumerated: To, TI, . .., T,, . ..:let T precede T' if G(T) < G ( T f ) .
Define
E = {n E N: T, computes s:+']
. . --
(thus, n E E if and only if the nth Turing machine computes n).
We claim that E is an r.e. set. Consider the following figure reminiscent of

424 12. The Word Problem
the proof that the set of all rational numbers is countable:
The nth column consists of the sequence of basic moves of the nth Turing
machine T, beginning with q,s;f+l.It is intuitively clear that there is an enu-
meration of the natural numbers n lying in E: follow the arrows in the figure,
and put n in E as soon as one reachers a terminal instantaneous description
a,, in column M. A Turing machine T*,can be constructed to carry out these
instructions (by Exercise 12.11 below, such a T* exists having stopping state
q,; that is, terminal instantaneous descriptions, and only these, involve q,.)
Thus, E is an r.e. subset of N.
The argument showing that E is not recursive is a variation of Cantor's
diagonal argument proving that the set of reals is uncountable. It suffices to
prove that the complement
E' = {n E N: n $ E) = {n E N: does not compute sYi1)
is not an r.e. subset of N. Suppose there were a Turing machine T' enumerat-
ing E'; since all Turing machines have been listed, T' = T, for some m E N. If
rn E E' = e(Tt) = e(T,), then T, computes sy'l, and so rn E E, a contradic-
tion. Pf m q! E', then m E E and so T, computes sy+l (definition of E); hence
m E e(Tm)= e(T') = E', a contradiction. Therefore, E' is not an r.e. set and E
Is not recursive. BI
EXERCISES
12.5. Prove that there are subsets of N that are not r.e. (Hint. There are only
countably many Turing machines.)
12.6. Prove that the set of all even natural numbers is r.e.
12.7. Give an example of a Turing machine T, having s, in its alphabet, which does
not compute s, .
12.8. Let SZ be the set of all positive words on (so,sly...,.),s If El and E, are r.e.
subsets of Q, then both El u E2 and El n E2are also r.e. subsets.
12.9. Let Q be the set of all positive words on (so,s,, ..., s,). If El and E, are
The Markov-Post Theorem 425
recursive subsets of Q, then both El u E, and El n E, are also recursive sub-
sets. Conclude that all recursive subsets of Q form a Boolean algebra.
12.10. If El and E , are recursive subsets of N, then El x E, is a recursive subset of
N x N. (Hint. First imbed N x N into N by "encoding" the ordered pair (m, n)
as 2"3".)
12.11. If T is a Turing machine enumerating a set E, then there is a Turing machine
T* having the same alphabet and with stopping state qo that also enumerates
E.
The Markov-Post Theorem
We now link these ideas to algebra.
If r is a semigroup with generators X = (x,, ...,x,) and if Q is the set of
all positive words on X, then the semigroup r has a soluable wordproblem if
there is a decision process to determine, for an arbitrary pair of words o ,
co' E Q, whether o = o' in T.This (informal) definition gives a precise defini-
tion of unsolvability.
Definition. Let r be a semigroup with generators X = {x,, ...,x,), and let Q
be the set of all positive words on X. The semigroup I? has an unsoluable word
problem if there is a word wo E SZ such that ( o E fi o = o, in I?) is not
recursive.
If F is the free group with basis X = {x,, ...,x,), then we shall view the set
Q of all (not necessarily positive) words on X as the set of all positive words
on thi: alphabet
(x, , x;,, .. . Xn,.,;I).
y
DefipnitIow. Let G be a group with presentation (x,, .. .,x,jA), and let Q be
the set of all words on x,, ...,x, (viewed as the set of positive words on
(x,, x;', ..., x, xi1)). Then G has a solvable wordproblem if { o E fi w = 1
in G ) is recursive.
The distinction between r.e. sets and recursive sets persists-ingroup theory.
Theorem 12.2. Let G be a finitely presented group with presentation
G = (x,, ..., x,lrl ,..., r,).
If Q is the set of all words on x,, . ..,x,, then E = { o E Q: w = 1 in G) is r.e.
Proof. List the words wo, a , , .. . in Ln as we did in the Introduction: first the
empty word, then the words of length 1 in order x,, x;', .. .,x,, xi1, then the
426 12. The Word Problem
words of length 2 in lexicographic order, then the words of length 3 in lexico-
graphic order, and so forth. Similarly, list all the words on {r,, . .., r,): p,,
p,, .. . . As in the proof of Theorem 12.1, following the arrows in the figure
below enumerates E.
It follows that a finitely presented group G has solvable word problem if
and only if ( o E a: w # 1 in G ) is r.e.
Recall the following notation introduced in Chapter 11. If o and of are
(not necessarily reduced) words on an alphabet X, then we write
if o and o' have exactly the same spelling.
Suppose that a semigroup has a presentation
T = (Xlaj = pj, j E J).
If w and o' are positive words on X, then it is easy to see that o = w' in r if
and only if there is a finite sequence
where wi -+ mi+, is an elementary operation; that is, either oiEE cajr and
mi+, = aDjr for some j, where c and r are positive words on X or wi+, = .Bji
and mi = eaji.
Let us now associate a semigroup to a Turing machine T having stopping
state q,. For notational convenience, assume that the s-letters and q-letters
involved in the quadruples of T are so, s,, . .., s,, and q,, q,, ...,q,. Let q
and h be new letters.
Definition. If T is a Turing machine having stopping state q,, then its asso-
ciated semigroup T(T) has the presentation:
where the relations R(T) are
The Markov-Post Theorem
qisj = qlsk if qisjskq,E T,
forallj? =0, 1, ..., M :
The first five types of relations are just the obvious ones suggested by the
basic moves of T; the new letter h enables one to distinguish basic move (ii)
(in the definition of a Turing machine) from basic move (iii)and to distinguish
basic move (iv) from basic move (v). One may thus interpret h as marking the
ends of the tape, so that the following words are of interest.
Definition. A word is h-special if it has the form hah, where a is an instanta-
neous description.
Since T has stopping state q,, each hah (with a terminal) has the form
hcq,rh, where a and r are s-words and i is not empty. Therefore, the last
three relations allow us to write hah = q in T(T) whenever a is terminal.
Lemma 12.3. Let T be a Turing machine with stopping state qo and associated
semigroup
T ( T )= (4, h, so, sly -.. , SMY 40, 41, . - ., 4xlR(T)).
(i) Let o and o' be words on (so, s,, ..., s,, a,, q,, .. ., q N ) with o f q and
cot f q. If o -+ a' is an elementary operation, then o is h-special if and only
if ofis h-special.
-
(ii) If o = hah is h-special, o' f q, and o -+ ofis an elementary operation of
one of the first five types, then o' hph, where either a -,P or j? -,a is a
basic move of T.
Proof. (i) This is true because the only relation that creates or destroys h is
hq,h = q.
(ii) By the first part, we know that oris h-special, say, o' = hph. Now an
elementary move in a semigroup is a substitution using an equation in a
defining relation; such a relation in T(T) of one of the first five types corre-
sponds to a quadruple of T, and a quadruple corresponds to a basic move.
Thus, either a -+j? or p -,a. 8le
Lemma f 2.4.-Let T be a Turing machine with stopping state q,, let be the set
428 12. The Word Problem The Markov-Post Theorem

of all positive words on the alphabet of T, and let E = e(T). If o E Q, then define
oEE if and only if hq,oh =q in T(T).
Define q:Q Ti by w I+ hq,wh, and identify Q with its image $2, c a; the
-+

Proof. If o E E, then there are instantaneous descriptions a, = q,w, a,, .. ., subset E of Q is now identified with
a,, where a, -+ai+,, and a, involves q,. Using the elementary operations in
T(T) of the first five types, one sees that hq, o h = ha,h in F(T); using the last El = (hq,wh: o E E).
three relations, one sees that ha,h = q in T ( T ) . It is plain that El is a recursive subset of R, if and only if E is a recursive
The proof of sufficiency is of a different nature than the proof of necessity subset of Q. In this notation, Lemma 12.4 reads:
just given, for equality in T(T) is, of course, a symmetric relation, whereas
a -+ /? a basic move does not imply that /? -+ a is a basic move.
If hq,oh = q in T(T), then there are words w,, ..., co, on (h, so, s,, ..., s,, Now assume that T is the Turing machine T* (with stopping state q,) of
qO,ql, . ..,qN) and elementary operations Theorem 12.1, so that E, hence El, is r.e. but not recursive. Were E recursive,
then Exercise 12.9 would give El, hence E, recursive, and this is a contradic-
tion. Therefore, y = T(T*) has an unsolvable word problem.
By Lemma 12.3(i), each mi is h-special: o, = haih for some instantaneous (ii) Define
description a,. By Lemma 12.3(ii),either a, -+ a,,, or ai+, 4 a,. We prove, by
induction on t > 2, that all the arrows go to the right; that is, for all i It - 1,
S = {h-special words hah: hah = q in T(T*)).
a, -+ a,,, . It will then follow that q, o + a, -+ . .- -+ a, is a sequence of basic Were S a recursive subset of a,then 3 n Q, would be a recursive subset of Q,,
moves with a, terminal (for a, involves q,, the stopping state); hence T by Exercise 12.9. But n $2,= El. 81
computes o and w E E. It is always true that a,-, -+ a,, for a, is terminal and
hence a,-, t a, cannot occur. In particular, this shows that the induction For later use, we rewrite the generators and relations of the Markov-Post
begins when t = 2. Suppose that t > 2 and some arrow goes to the left. Since semigroup y (T*).
the last arrow a,-, -+ a, points right, moving backward until one reaches an
arrow pointing left gives an index i with Corollary 12.6.
(i) There is a finitely presented semigroup

-
But there is never ambiguity about the next move of a Turing machine, so
that a,-, = ai+, and oi-, = hai.-,h r ha,,, h wi+,. We may thus eliminate
mi and oi+,, thereby reducing rn, and the proof is completed by induction.
la
with an unsolvable word problem, where 6 , G,, Hi,
positive s-words and qil, qi, E (q, qO,. .., qNj.
Xi are (possibly empty)

(ii) There is no decision process which determines, for arbitrary qij and positive
s-words X and Y,whether XqijY = q in r.
Theorem 12.5 (Markov-Post, 1947).
(i) There is a finitely presented semigroup Proof. (i) Regard the generator h of the semigroup y = T(T*) as the last
s-letter and re-index these s-letters so that h = s,. The rewritten relations in
R(T*) now have the described form.
(ii) Let Q, be the set of all positive words on the rewritten generators of T,
with an unsolvable word problem.
let
(ii) There is no decision process which determines, for an arbitrary h-special
word hah, whether hah = q in y. = {XqijY: X, Y are positive words on rewritten
s-letters and Xq,, Y = q in r ) ,
Proof. (i) If T is a Turing machine with stopping state q, and with alphabet and let
A = {so,s,, . ..,s,), then let Q be all the positive words on A and let E =
e(T) c Q. Define Q to be all the positive words on A u {q, h, q,, q,, ...,qN}, S, = {s,as,: a = oqr, where o and r are positive
where q,, q,, ..., q, are the q-letters occurring in the quadruples of T, and words on so, ...,sM-, and sMas, = q in r}
 12. The Word Problem The Novikov-Boone-Britton Theorem: Sufficiency of Boone's Lemma

s2
(remember that h has been rewritten as s,). Of course, is just the subset generators: q, q,, . . .,q,, so, . ..,s,, ri, i E I, X,t, k;
of the theorem rewritten in the new notation. Now A n Q2 = S2;since 3, is relations: for all i E I and all p = 0, . . ., M ,
not recursive. Exercise 12.9 shows that is not recursive. H
xsp = spx2,
1 1
risp = spxrix,
r1-1 &# qilGiri= iYTqi2Ki, A2

The Novikov-hone-Bsitton Theorem: tri = rit,

Sufficiency of Boone's Lemma tx = xt,

The word problem for groups was first considered by M. Dehn (1910) and by kri = rik,
A. Thue (1914). The solution was given by P.S. Novikov (1955) and, indepen-- kx = xk,
dently, by W.W. Boone (1954-1957) and by J.L. Britton (1958). In 1959,
Boone exhibited a much-simpler finitely presented group than any of those k(q-l tq) = (q-ltq)k.
previously given, and he proved it has an unsolvable word problem. In con- The subsets 8, c A, c A, of the relations are labeled for future reference.
trast to the "combinatorial" proofs of Novikov and Boone, Britton's proof
relies on properties of KNN extensions (which led him to discover Britton's If X and Y are s-words, define
lemma). In 1963, Britton gave a much simpler and shorter proof
group; we present his proof here, incorporating later impro (XqjY)* = X#qj Y;
Boone, D.J. Collins, and C.F. Miller, 111. We assure the reader where qj E {qy40, .- - q ~ ) .
Mathematical Logic required in the proof has already appeared; we need
only Corollary 12.6, a paraphrase of the Markov-Post theorem, that exhibits Definition. A word C is special if I: = X#qjY, where X and Y are positive
a particular finitely presented semigroup r with an unsolvable word s-words and qj E (q, qO,.. ., q,).
problem.
Remember that the proof is going to reduce equality of words in a group
to equality of words in a semigroup. It is thus essential to keep track of
exponents, for while arbitrary words make sense in a group, only positive
- -
If C is special, then E r X # q j x where X and Y are positive s-words, and
so Z* (X#qjY)* XqjY is a positive word; therefore, C* determines an
element of the semigroup T.
The reduction to the klarkov-Post theorem is accompllished by the fol-
words make sense in a semigroup.
lowing lemma:

-
Notation. If X zz sj: ...si; is a (not necessarily positive) s-word, then X" r
silel. . .si'm. Note that if X and Y are s-words, then (X")" X and (XY)'
X#Y#.
- Lemma 12.7 (BOOB+.Let T be a Turing machine with stopping state qo and
associated semigroup lr = T(T) (rewritten as in Corollary 12.6). If E is a
special word, then
k(C-l tC) = (E-I t C)k in B = B(T)
Recall, for every Turing machine T, that there is a semigroup l? = T(T)
with the presentation if and only fi C* = q in T(T).

r = (4, 40, - - . ,q ~So,

, sMIFiqilGi = Hiqi2Ki,i E I), Theorem 12.8 (Novikov-Boone-Britton). There exists a finitely presented
group 9?with an unsolvable word problem.
where F,, Gi, Hi, Ki are (possibly empty) positive s-words and qil, qi2E
(4, 40, q~).
m e . ,
Proof. Choose T to be the Turing machine T* of the Markov-Post theorem.
For every Turing machine T, we now define a group B = B(T) that will be If there were a decision process to determine, for an arbitrary special word C,
shown to have an unsolvable word problem if T is chosen to be the Turing whether kC-I tCk-' C-' t-I C - 1 in B(T*), then this same decision process
machine T* in the Markov-Post theorem. The group B(T) has the presenta- determines whether C* = q in T(T*). But Corollary 12.6(ii) asserts that no
tion: such decision process for r(T*) exists. BII
432 12. The Word Problem Cancellation Diagrams 433

Corollary 12.9. Let T be a Turing machine with stopping state q, enumerating Since wv = ov+, in r implies o,*= o*,, in B, by the relations labeled A,, it
a subset E of 0 (the set of all positive words on the alphabet of T). If o E: Q, follows, for each v, that
then o E E if and only if k(hw1q, oh) = (h-I q, oh)k in B(T).
a,* = Lvo;+,Rv i n g
Proof. By Lemma 12.4, co E E if and only if hq, o h = q in T(T). But, in B(T), for words L, and R, on some rivand x. The words L = L, ...L,-, and R =
R,-, ...R, are thus words on {x, ri, i E I), and

-
(hq,oh)* = h-lq,wh (which is a special word), and Boone's lemma shows
that (h-lql oh)# = hq, o h = q in T(T) if and only if k(h-I q, wh) = (h-lq1 wh)k oT = Lw:R in B.
in B(T). But oT = (C*)*= Z: and o,T zz q* q, so that
The proof below is valid for any Turing machine T with stopping state q,, C = LqR in@.
We abbreviate B(T) to B and T(T) to T. Since the generators t and k commute with x and all the ri, they commute
The proof of Boone's lemma in one direction is straightforward. with L and R. Therefore,
kz-1 tCk-lz-l t -1 c = k ~ - l q - lL-, t ~ q ~ kR-,
- l q-l L-I t-' L ~ R
Lemma 12.10.
= k ~ - l ~ - l t ~ k - l ~ -q~
lt-f
(i) If V is a positive s-word, then
= R-l(kq-l tqk-lq-'t-lq)R
ri V = VR . in B and rL:' V = VR' in 99,
where R and R' are words on (r,, x] with R positive. because the last word is a conjugate of a relation. W
(ii) If U is a positive s-word, then
U#r,' = LU* in B and U#ri = LfU# in B, Observe that the last relation of the group 93 appears only in the last step
of the proof.
where L and L' are words on (ri, x).

-
Pro05 We prove that riV = VR in @ by induction on m > 0, where V =
spl...sp,. This is certainly true when m = 0.If rn > 0, write V = V'sam;by
induction, riV ri Vspm= VfR'ssm,where R' is a positive word on (ri, x).
Using the relations xss = ssx2 and riss = ssxrix, we see that there is a posi-
tive word R on (ri, x) with R = RfsDm in a.
The proofs of the other three equations are similar. B
Proof of Safficieacy in Boone9s Lemma. If 2 is a special word with C*
Xqj Y = q in r, then there is a sequence of elementary operations
where, for each v, one of the words w, and w,+, has the form UFiqilGiV with
U and V positive s-words, and the other has the form UHiqi,KiK By the
lemma, there are equations in B:
where L' and R' are words on {r,, x). In a similar manner, one sees that there
are words L" and R on {ri, x) with
Cancellation Diagrams
We interupt the proof of Boone's lemma (and the Novikov-Boone-Britton
theorem) to discuss a geometric method of studying presentations of groups,
essentially due to R. Lyndon, that uses diagrams in the plane. Sinc-,,we are
only going to use diagrams in a descriptive way (and not as steps in a proof),
= we may write informally. For a more serious account, we refer the reader to
Lyndon and Schupp (1977, Chap. V) with the caveat that our terminology
does not always coincide with theirs.
When we speak of a polygon in the plane, we mean the usual geometric
figure including its interior; of course, its boundary (or perimeter) consists of
finitely many edges and vertices. A directed polygon is a polygon each of
whose (boundary) edges is given a direction, indicated by an arrow. Finally,
given a presentation (XI A) of a group, a labeled divectedpolygonis a directed
polygon each of whose (directed)edges is labeled by a generator in X.
Given a presentation (XlA) of a group, we are going to construct a labeled
directed polygon for (almost) every word
o = xgl ...x?,
where x,, ..:, x, are (not necessarily distinct) generators and each ei = + 1.
For technical reasons mentioned below, o is restricted a bit.
434 12. The Word Problem
Definition. Let F be a free group with basis X. A word w E x;l.. .x,". on X
with each ei = +_ 1 is called fieely redtrsced if it contains no subwords of the-
form xx-' or x-'x with x E X.
A cyclic gerantpfation of w = X I I . . . x? is a word of the form
x? ...x,".x;t. .. xri-j.1 (by Exercise 3.8, a cyclic permutation of w is a conjugate
of it). A word w is cyclic~l&gt reduced if every cyclic permutation of it is freely
reduced.
If w E x;' . ..x,". is cycli~allyreduced, construct a labeled directed polygon
as follows: draw an n-gon in the plane; choose an edge and label it x,; label
successive edges x,, x,, ...,x, as one proceeds counterclockwise around the
boundary; direct the ith edge with an arrow according to the sign of ei (we
agree that the positive direction is counterclockwise). For example, if k and x
commute, then the labeled directed polygon is the square in Figure 12.1.
Figure 12.1
Figure 12.2
As a second example, consider the last relation in Boone's group B: o =
kq-' tqk-l q-' t-' q. The labeled directed polygon for w is the octagon whose
first edge is the top k-edge in Figure 12.2. If o is not cyclically reduced, this
construction gives a polygon having two adjacent edges with the same label
and which point in opposite directions, and such polygons complicate proofs.
However, there is no loss in generality in assuming that every relation in a
Cancellation Diagrams 435
presentation is cyclically reduced, for every word has some cyclically reduced
conjugate, and one may harmlessly replace a relation by any of its conju-
gates. Every cyclically reduced relation thus yields a labeled directed polygon
called its relator polygon.
We can now draw a picture of a presentation (XIA) of a group G (with
cyclically reduced relations A) by listing the generators X and by displaying
a relator polygon of each relation in A. These polygons are easier to grasp
(especially when viewing several of them simultaneously) if distinct genera-
tors are given distinct colors. The presentation of the group B in Boone's
lemma is pictured in Plate 1 (inside front cover). There are six types of genera-
tors: q; s; r; x ; t ; k, and each has been given a different color.
There is a presentation of a group called B, which is pictured in Plate 3.
This group will occur in our proof of the Higman imbedding theorem.
Another example is provided by an HNN extension: a relation involving a
stable letter p has the form apebp-"c, where e = f1. If the corresponding
relator polygon is drawn so that the p-edges are parallel, then they point in
the same direction.
Let D be a labeled directed polygon. Starting at some edge on the bound-
ary of D, we obtain a word w as we read the edge labels (and the edge
directions) while making a complete (counterclockwise) tour of D's boundary.
Such a word o is called a boundary word of D. (Another choice of starting
edge gives another boundary word of D, but it is just a cyclic permutation,
hence a conjugate, of w. A clockwise tour of D's boundary gives a conjugate
of o-'.)
Definition. A diagram is a labeled directed polygon whose interior may be
subdivided into finitely many labeled directed polygons, called regions; we
insist that any pair of edges which intersect do so in a vertex.
We quote the fundamental theorem in this context; a proof can be found
in Lyndon and Schupp.
Fundamental Theorem of Coanbinatoriall Group Theory. Let G have a finite
presentation (XI&),where A satisfies the following conditions:
(i) each 6 E A is cyclically reduced;
(ii) i f 6 E A, then 6-I E A;
(iii) i f 6 E A, then every cyclic permutation of 6 lies in A.
If w is a cyclically reduced word on X, then w = 1 in G if and only i f there is
a diagram having a boundary word o and whose regions are relator polygons of
relations in A.
An immediate consequence of this theorem is a conjugacy criterion. As-
sume that w and w' are cyclically reduced words-on X, and consider the
annulus with outer boundary word w' and inner boundary word o, as in
Figure 12.3.
 12. The Word Problem Cancellation Diagrams

The hypothesis provides a sequence of elementary operations

C*fo1+02-+----+o,=q inr.
The proof begins by showing that each w3 has the form U,#qivV;,where
v 5 n - 1 and Uv and V; are positive s-words; moreover, there are words L,
and R, on (x, ri, i E I) such that, for all v,
o: = Lvo:+l R, in 97.

Figure 12.3

Corollary. The elements o and ofare conjugate in G if and only if the interior
of the annulus can be subdivided into relator polygons.

The proof consists in finding a path ,ti' from o' to w and cutting along
/3 to form a diagram as in Figure 12.4. A boundary word of the new diagram
is w'/30-1,ti'-', and the fundamental theorem says that this word is 1 in G.
Conversely, if o'/3o-'/3-' = 1 in G, one may form an annulus by identifying
the edges labeled /3; that is, start with the diagram on the above right and
glue the p's together to obtain the annulus on the left.

Figure 12.5

Figure 12.5 pictures all of these equarioms; we have not drawn the subdivisions
of each interior polygon into relator polygons, and we have taken the liberty
of labeling segments comprised of many s-edges by a single label P: %, &/,, or
uv -
The reader should now look at Plate 2; it is a diagram having wiC) as a
boundary word. In the center is the octagon corresponding to the octagonal
relation w(q) = kq-' tqk-lq-' t-' q, and there are four (almost identical) quad-
Figure 12.4 rants as drawn above, involving either C or C-' on the outer boundary and
q or q-' on the octagon (actually, adjacent quadrants are mirror images). The
commutativity of k with x and each ri allows one to insert sequences of
An example will reveal how these diagrams can illustrate the various steps squares connecting k-edges on the outer boundary to k-edges on the octagon;
taken in rewriting a word using the relations of a given presentation. The similarly, the commutativity of t with x and each ri inserts sequences
proof of sufficiency of Boone's lemma requires one to prove, for a special connecting t-edges on the outer boundary with t-edges on th!: octagon. Since
word C, that the quadrants have already been subdivided into relator polygons, the four
quadrants together with the four border sequences, form a diagram. There-
w(C)r kZ-'tCk-'Z-'t-'Z= 1 inB. fore, w(Z) = 1 in B, as asserted by the fundamental theorem.
438 12. The Word Problem
Define BAto be the group having the same presentation as B except that
the octagonal relation is missing. Now regard Plate 2 as an annulus having
the octagonal relation as the inner boundary word. This annulus has just
been subdivided into relator polygons, and so the corollary of the fundamen-
tal theorem says that w(Z) is conjugate to w(q) in BA.This last result is a
reflection of the fact that the octagonal relation enters the given proof of the
sufficiencyof Boone's lemma at the last step.
The Novikov-Boone-Britton Theorem:
Necessity of Boone's Lemma
We now turn to the proof of the more difficult half of Boone's lemma. Geo-
metrically, the problem is to subdivide the labeled directed polygon with
boundary word w(Z) into a diagram whose regions are relator polygons of B,
The conjugacy of w(C) and the octagonal relation w(q) in the group BA
(mentioned above) suggests a strategy to prove the necessity of Boone's
lemma: subdivide the annulus with outer boundary w(Z) and inner boundary
w(q) using the relations of BA(thereby allowing us to avoid further use of the
octagonal relation w(q)), trying to make the annulus look like Plate 2. We
shall give formal algebraic proofs, but, after the proof of each lemma, we shall
give informal geometric descriptions. (It was the idea of E. Rips to describe
this proof geometrically, and he constructed the diagrams for the Novikov-
Boone-Britton theorem as well as for the coming proof of the Higman im-
bedding theorem. He has kindly allowed me to use his description here.)
Define groups Bo, B, ,B2, and B3as follows:
Bo = (XI@),the infinite cyclic group with generator x;
(recall that we labeled certain subsets of the relations of B as A, c A2 c A,
when we defined B; recall also that this notation means that we are adjoining
the displayed generators and relations to the given presentation of go);
where Q is free with basis (q, qo, ..., q,);
Lemma 12.11. In the chain
each group is an HNN extension of its predecessor; moreover, B, * Q is an
HNN extension of go.In more detail:
The Novikov-Boone-Britton Theorem: Necessity of Boone's Lemma 439
(i) B, is an HNN extension with base Bo and stable letters {so,. . .,s,);
(ii') B, * Q is an HNN extension with base Boand stable letters {so,...,s,) v
(4, 40,- 9 qN);
(ii) B, * Q is an HNN extension with base B, and stable letters {q,qo, ...,q,);
(iii) B2is an HNN extension with base B, * Q and stable letters (ri: i E I);
(iv) B3is an HNN extension with base B, and stable letter t; and
(v) B is an HNN extension with base B3 and stable letter k.
Proof. (i) The presentation
B, = (x, so, . ..,s, /s;'xsa = x2, all p)
shows that B, has base (x) = Boand stable letters (so, ...,s,). Since x has
infinite order, Ap = (x) z (x2) = Ba, and so qB:Ap -+ Bar defined by x c
x2, is an isomorphism for all P. Therefore, 9, is an HWN extension.
(ii') The presentation of 99, * Q,
shows that 99, * Q has base goand stable letters (so, ...,s,) v (q, q,, ...,
q,). Since x has infinite order, Ap = (x) z (x2) = Ba, and so the maps qa
are isomorphisms, as above; also, the maps qqi are identity maps, where
A, = (x) = Bqj. Thus, B1* Q is an HNN extension with base goand stable
letters (so, ..., s,) u (q, qo, -.-,q,).
(ii) Since Q is free with basis (q, q,, ...,q,), Example 11.10 now shows
that B, *Q is an HNN extension with base B, and stable letters
(4, 40, . .., qN).
(iii) The presentation
shows that B2has base B1* Q and stable letters {r,, i E I). Now, for each i,
the subgroup Ai is (FTqil G,, ssx, all p) and the subgroup 8, is (H:qi,Ki,
S~X-',all p). We claim that both Ai and Bi are free groups with bases the
displayed generating sets. First, use Exercise 11.8 to see that (sax, all p) is .
free with basis (sax, all p): map (sax, all p) onto the free group with basis
{so,..., s,} by setting x = 1; then observe that Ai = (FTqilGi, sax, all /?)z
(F?qil Gi) * (sax, all p) < 39, * Q (because F,bqilGi involves a q-letter and
elements of the free group (sax, all p) do not). A similar argument applies to
Bi, and so there is an isomorphism qi: Ai -+ Bi with qi(K'qi, Gi) = H:qi2Ki
and qi(sax)= sax-' for all p. Thus, B2 is an HNN extension with base
B l * Q.
(iv) Note that B3has base B2and stable letter t:
Since t commutes with the displayed relations, B3 is an HNN extension of
B2, as in Example 11.11.
440 12. The Word Problem The Novikov-Boone-Britton Theorem: Necessity of Boone's Lemma 441

(v) Note that 8 has base a, and stable letter k: If the initial te in the pinch is tej, where j 2 1, then teDt-" r tejqRjq-'tej+l
with qRjq-l = R in B, for some word R on {x, ri, i E I ) . Since B2 I a , , by
Theorem 11.78, we may view this as an equation in 8,:
As in Example 11.1l , B is an HNN extension of 8 , . @
!
t e j q ~ j q -tej+l
l = teqRjq-lt-e = teRt-O in 8,.
Corollary 12.112. But the stable letter t in 9, commutes with x and all ri, so there is an
(i) The subgroup (s, x, ...,S , ~ XI
) 8, is a free group with basis the displayed equation
letters. qRjq-I = R i n 8 , .
(ii) There is an automorphism $ of 8,with $(x) = x-' and $(s,) = sofor all P. Hence, in 8 , ,
Proof. (i) This was proved in part (iii) of the above lemma. (q-l tejq)~j(q-l
tej+lq) = q-l teRtFeq
(ii) The function on the generators sending x r x-' and spHsp for all P = q-' Rq (for t commutes with x, ri)
preserves all the relations.

The reader should view Lemma 12.11 as preparation for the remainder of
the proof; it will allow us to analyze words using Britton's lemma, Theorem
11.81. There is thus a factorization of o in B3having smaller length, contradicting
the choice of n being minimal. Therefore, this case cannot occur.
Lemma 12-13. Let C be a fixed special word satisfying the hypothesis of
Bonne's lemma:
w ( C ) s kC-ltCk-'C-l~-lC = 1 i n 8 .
Then there are freely reduced words L, and L, on {x, ri, i E I ) such that
L,EL2 = q in @.,

Proof. Since 8 is an WINS extension with base 8, and stable letter k,

Britton's lemma applies to the word kZ-ltXk-lC-lt-lE; it says that
kZ-'tCk-l is a pinch and that Z-'tZ = C in 8 , , where C is a word on
(x, q-I tq, r,, i E I). (Since the stable letter k commutes with {x, q-I tq, ri,
i E I ) , we are in the simple case of Example 11.11 when the subgroups A and
B are equal and the isomorphism 9:A -+ B is the identity.) Therefore, there
exist words o of the forin C-'tCC-I = 1 in 8,; in detail,

where the Rj are (possibly empty) freely reduced words on {x, ri, i E I ) and
Figure 12.6
+
ej = 1. We assume co is such a word chosen with n minimal.
Since 8, is an HNN extension with base 8, and stable letter t, Britton's
lemma applies again, showing that w contains a pinch teDt-', and there is a
Geometrically, we have shown that the labeled directed annulus with outer
word R on (x, ri, i E I ) with D = R in 8,.
boundary word w(C) and inner boundary word the octagon w(q) contains a
If the pinch involves the first occurrence of the letter t in o , then teDt-" =
"quadrant" involving C on the outer boundary, q on the inner boundary, and
+
tCRoq-Ite1.Hence e = 1, el = - 1, tZRoq-Ite1= tRt-" and
internal paths L, and L, which are words on {x, ri, i E I ) . Of course, there are
two such quadrants as well as two "mirror images" of these quadrants which
equivalently, involve Z-I on the outer boundary and q-' on the inner boundary. More-
R-'XR0 = q in 8 , , over, the regions subdividing these quadrants are relator polygons corre-
sponding t o the relations A3; that is, they do not involve k-edges or t-edges.
which is of the desired form.
442 12. The Word Problem The Novikov-Boone-Britton Theorem: Necessity of Boone's Lemma

Finally, there is no problem inserting the "border sequences" connecting Yi,

k-edges (and t-edges) on the outer boundary with k-edges (and t-edges) on the

-
inner boundary, for the internal paths of the quadrants involve only x and
r;s, all of which commute with k and with t.
Recall that Z X#qjY, where X and Y are positive s-words and qj E
{q, qO,. ..,q,). We have just shown that

for some freely reduced words L, and L, on (x, ri, i E I]. Rewrite this last
equation as

qi,
Lemma 12.14. Each of the words LIX#qj and qLzl Y-' is ri-reduced for every
i E I. Figure 12.7

Proof. Suppose, on the contrary, that L,X#qj contains a pinch r,"Crce as a

subword. Since X # is an s-word, this pinch is a subword of L,, a word on
(x, ri, i E I). Since L, is freely reduced, C = xmfor some rn # 0. Since B2is an
HNN extension with base a, * Q and stable letters {r,, i E I), Britton's lemma
says that there is a word V in B, * Q, where Q = (q, q , , ...,q,), such that

+
ej = 1, ojis a word on (s,x, ...,sMx)for all j, V is reduced as a word in the
free product, and
xm= V in B, * Q.
Since xmE g l , one of the free factors of B, * Q, we may assume that V
does not involve any q-letters; in particular, V does not involve FTqilGi.
Therefore,
X" = Wo '(splx)f'...(sppx)fp in #,,

where each f, = 1. Since is an HNN extension with base go= (x) and
stable letters (so, ...,s,), another application of Britton's lemma says that
the word x-"o,, which is 1 in B,, contains a pinch of the form s{xcspf7
+
where E = 1. Now inspection of the spelling of o,shows that it contains no
such subword; we conclude that coo = 1, hence xm= 1. But x has infinite
order (since Bo a , ) , and this contradicts rn # 0. We conclude that L,, and
hence L, X"qj, is ri-reduced.
A similar proof shows that qL,' Y-' is also ri-reduced. d
We know that the boundary word of each of the four quadrants is 1, so
that each quadrant is subdivided into relator polygons. The two words in the
lemma are sub-boundary words that do not flank either of the two q-edges;
that is, neither of the q-edges is surrounded by other (boundary) edges on
both sides. As we are working within aA, the octagonal relator polygon is not
inside a quadrant. The only other relator involving a q-letter is the eight-
Figure 12.8
sided "petal" in A; (Figure 12.7). There must be such a petal involving
the q-letter on the quadrant's boundary. The lemma shows that the petal's
boundary must contain edges in Y and edges in X # (Figure 12.8).
The following lemma completes the proof of Boone's lemma and, with
it, the Novikov-Boone-Britton theorem. In view of a further application
of it in the next section, however, we prove slightly more than we need
now.
Lemma 62.15. Let L, and L, be words on (x, ri, i E I ) that are ri-reduced for
all i E I. If X and Y are freely reduced words on {so,..., s,) and if
 12. The Word Problem
then both X and Y are positive and
Remark. In our case, both X and Y are freely reduced, for X and Y are
positive (because C = XiliqjYis special), and positive words are necessarily
freely reduced.
h o o f . The previous lemma shows that LIX#qj = qLil Y-' in 93, and that
both words are ri-reduced. By Corollary 11.82, the number p 2 0 of r-letters
in L, is the same as the number of r-letters in L, (because no r-letters occur
outside of L, or L,); the proof is by induction on p.
If p = 0, then the equation LlX#qjYL2 = q is
This equation involves no r-letters, and so we may regard it as an equation
in B, * Q IB,, where Q = (q, q,, ...,qN). But the normal form theorem for
free products (Theorem 11.52) gives qj = q and xmX#= 1 = Yxnin 3,. Since
B, is an HNN extension with base Bo = (x) and stable letters (so, ...,s,),
it follows from Britton's lemma that rn = 0 = n and that both X and Y are
empty. Thus, X and Y are positive and XqjY = qj = q in T.
Assume now that p > 0. By Lemma 12.14, the words LIX#qj and
qLil Y-' are ri-reduced for all i. Since B2 is an HNN extension with base
g, * Q and stable letters (ri, i E I), Britton's lemma gives subwords L, of L,
and L, of L, such that
where the word in parentheses is a pinch; moreover, either e = - 1 and
o r e = + I and
xmX#qjYxnE Bi = (Hrqi2Ki7sax-', all P).
In the f ~ scase,
t subwords of the form sss;' or s;'ss (if any), that the first surviving letter of
qj = 4il,
for the membership holds in the free product 99, * Q; in the second case,
qj = qi2.We consider only the case
leaving the similar case e = + 1 to the reader. There is a word
where a, = & 1 and the uj are possibly empty words on {sex, all 8). Of all
such words, we assume that o has been chosen with t minimal. We may
The Novikov-Boone-Britton Theorem: Necessity of Boone's Lemma 445
further assume that each uj is a reduced word on (sax, all p), for Corollary
12.12(i) says that this set freely generates its subgroup. Since o = 1 in B, * Q,
the normal form theorem for free products (Theorem 11.52) shows that each
"syllable" of o between consecutive qis is equal to 1 in B,. However, if
one views B, * Q as an HNN extension with base 9, and stable letters
(q, qO,...,qN)(as in Example 11.10, in which case the subgroups A and B are
I), then Britton's lemma says that o contains a pinch q;CqTE as a subword
with C = 1 in B, (of course, this case of Britton's lemma is very easy to see
directly).
If a pinch involves the first occurrence of qj, then - E = a, = - 1 and
We claim that a pinch cannot occur at any other place in o.Otherwise, there
is an index v with a pinch occurring as a subword of (F,#qjGi)%,(Fi#qjGiP1.
If a, = +1, then a,,, = -1, the pinch is qjGiyG;lq;l, and GiuvGrl =
1 in g,; if a, = -1, then a,,, = +1, the pinch is q;'&#-'uVFfqj, and
&#-' uvFi#= 1 in 3,.In either case, we have u, = 1 in 93,. But u, is a reduced
word on the basis (sex: all /I), and so u, = 1, contradicting the minimality of
t. We conclude that t = 1, a, = - 1, and
We have already seen that
and so it follows from o being in the free product that
We rewrite these last two equations, by conjugating, into more convenient
form:
Recall that Gi is a positive s-word. Let us show, after canceling all
GrlY is positive; that is, there is enough cancellation so that the whole of
Gi is eaten by Y). Otherwise, after cancellation, GY ,:' begins with s i l for
some p. Since B, is an HNN extension with base (x) and stable letters
(so, ...,s,), then xnuoGrl Y = 1 in B, implies, by Britton's lemma, that its
post-cancellation version contains a pinch s f ~ s y f= sfxhsyf,where 0 5 .I2
M. Now uo is a reduced word on (sox, ...,s,x), say,
where g, = 1. The pinch is not a subword of xnuo.It follows that the last
letter sif of the pinch sfxhs;* is the first surviving letter SF' of Gyl Y. Thus,
A = p = pp, f = + 1 = gp, and saxhsil = saxsil; that is, h = 1. But x =
 446 12. The Word Problem The Novikov-Boone-Britton Theorem: Necessity of Boone's Lemma 447

sax2s;' in B,, giving x E (x2), a contradiction. The first surviving letter of
G;' Y is thus positive, and so there is a subword Y, of Y beginning with a
positive s-letter for which Y = Gi Y,.
In a similar manner, one sees, after canceling all subwords of the form s s-I
or s;'s, (if any), that the first surviving letter of X#Frw1is negative; that Pis,
there is enough cancellation so that the whole of F:-' is eaten by X"). The
proof is just as above, inverting the original equation X#$#-'u, x m= 1 in
a,. There is thus a subword X, of X with X: ending in a negative letter and
such that X = X, Fi.
We have proved, in B, ,that 1 = xnu0G;' Y = xnuoG;' G, Y, ,and so
must see that X,#H,T and KiY, are freely reduced; that is, they contain no
"forbidden" subwords of the form sps$l or si1se. NOWKi is a positive word
on s-letters, so that it contains no forbidden subwords; further, Y, E G,:' Y is
just a subword of Y (since the whole of Gi is eaten by Y), hence has no
forbidden subwords, by hypothesis. Therefore, a forbidden subword can
occur in KiY, only at the interface. But this is impossible, for we have seen
that Y, begins with a positive letter, namely, "the first surviving letter" above.
-
A similar argument shows that X f H 5 s freely reduced.
By induction, both X, Hi and Ki Y, are positive. Hence, their subwords XI
and Y, are also positive, and hence X X , Fi and Y E Gi Yl are positive. The
inductive hypothesis also gives

Define (Xf Hf qi2KiYl)* = q in T.

vz = r;' uolri. Since (Xf Hi")# = XI Hi, we have
Since uo is a word on sflx's and rilspxri = sflx-' for all p, the element v,' is a
word on (sox-', ...,s,x-'1. But we may also regard ucl and v,' as elements
of B, = (x, so, ...,s,). By Corollary 12.12(ii), there is an automorphism (it is only now that we see why the "sharp" operation # was introduced; had
$ of B, with $(x) = X-I and $(sp) =sfl for all 8. Hence, v;' = $(ZAG') = we used inversion instead, we would now have HiXl qi2KiY, = q in T, and we
$(Y,xn) = Y,x-"; that is, could not finish the proof). Thus,

If one defines v;' = r;'u;'r,, then a similar argument gives

XqjY
- XI FiqjGiYl
X,Fiqi,GiY, = XlHiqi2KiY, i n r .
Combining this with (5) gives
where X, is the subword of X defined above.
Let us return to the induction (remember that we are still in the case
e = - 1 of the beginning equation (1)): as desired. The case e = + 1 at the beginning of the inductive step is entirely
similar, and the proof of Boone's lemma and the Novikov-Boone-Britton
theorem is complete.
There are equations in B2,
Here is some geometric interpretation of the long proof of this last lemma.
At the end of the previous lemma, we had shown that a quadrant involving
L~
= ~,r;'(u;'F;#)q~(G~uo~)r~ Eq. (2)
X = X"qil Y on the outer boundary and a q on the inner boundary must
have a "petal" relator polygon next to qii. Now there is another q-letter on
this petal which is now in the interior of the quadrant. As petals are the only
relator polygons involving q-letters (for we are working in BA),there must be
a sequence of such petals (involving various q-letters) from the outer bound-
ary of the quadrant to the q on the inner boundary (Figure 12.9).
Do any other q-edges occur on interior regions of the quadrant? The only
= (L3x-"X?)H:qi2 Ki(Ylx-"L,) Eqs. (31, (4). other possibility is a flower whose eight-sided petals arise from a petal relator
Therefore, regions (Figure 12.10). We have not drawn the relator polygons that sub-
divide the eye of the flower, but we may assume that the eye contains no relator
L,x-"(XfH:qi2Ki Y,)x-"L, = q in B2. regions having q-edges (otherwise the eye contains a smaller such flower and
Now L3x-" and x-"L, are words on (x, ri, i E I) having at most p - 1 occur- we examine it). The boundary word of the flower's eye involves ri's and sp's,
rences of various r-letters. In order to apply the inductive hypothesis, we and this word is 1 in a,. By Britton's lemma, this word contains a pinch of
 12. The Word Problem The Novikov-Boone-Britton Theorem: Necessity of Boone's Lemma 449

the form rieCrie.There are thus two adjacent petals whose r-edges point in
opposite directions, and this contradicts the orientation of these petals (note
how the geometry of the plane enters).

Figure 12.11

4
Figure 12.9 Now focus on the top portion of the quadrant. The remainder of the proof
shows that the dashed paths comprised of s-edges can be drawn (actually, the
proof shows that the rightward path is X,# (followed by x-", which is incor-
porated into L,)and the leftward path is Y, (followed by xn,which is incorpo-
rated into L,). Induction says that one can repeat this construction, so that
the petals move down to the bottom q; thus the whole quadrant can be
subdivided into relator regions.
Aside from the group-theoretic proof just given (which is a simplification
of Britton's original proof), there are several other proofs of the unsolvability
of the word problem for groups: the original combinatorial proofs of
Woviltov and of Boone; a proof of 6.Higman's, which is a corollary of
his imbedding theorem. The proof of Higman's imbedding theorem that we
shall give in the next section uses our development so far, whereas Higman's
original proof does not depend on the Novikov-Boone-Britton theorem.
We must mention an important result here (see Lyndon and Schupp (1977)
for a proof). W. Magnus (1930) proved the Freiheitsatz. If G is a finitely
generated group having only one defining relation r, say, G = (x,, ...,xnlr),
then any subset of (xi, ...,x,) not containing all the xi involved in r freely
generates its subgroup. As a consequence, he showed (1932) that G has a
solvable word problem.
There are other group-theoretic questions yielding unsolvable problems;
let us consider another such question now.
Definition. A finitely generated group G = (XIA) has a solvable conjagacy
problem if there is a decision process to determine whether an arbitrary pair
Figure 12.10 of words w and co' on X are conjugate elements of G.
450 12. The Word Problem
When G is finitely presented, it can be shown that its having a solvable
conjugacy problem does not depend on the choice of finite presentation. A
group with a solvable conjugacy problem must have solvable word problem,
for one can decide whether an arbitrary word o is (a conjugate of) 1; the
converse is false. We now indicate how this result fits into our account.
Corollary 12.16. The group gAhas solvable word problem and unsolvable
conjugacy problem.
Proof. Recall that gAis Boone's group 9f without the octagonal relation
w(q). A.A. Fridman (1960) and Boone, independently, proved that BAhas
solvable word problem (we will not present this argument).
The following three statements are equivalent for any special word C:
(i) w(f3) = 1 in %
(ii) C* = 1 in r;
(iii) w(C) is conjugate to w(q) in gA.
The necessity of Boone's lemma is (i) => (ii); in geometric terms, we have
already seen that the labeled directed annulus with outer boundary word
w(f3) and inner boundary word w(q) can be subdivided into relator polygons
corresponding to relations other than w(q); that is, using relations of BA.This
proves (ii) => (iii). Finally, (iii) => (i) is obviously true, because w(q) = 1 in a.
The equivalence of (i) and (iii) shows that BAhas a solvable conjugacy prob-
lem if and only if 93 has a solvable word problem. By the Novikov-Boone-
Britton theorem, BAhas an unsolvable conjugacy problem. 811
The Higman Imbedding Theorem
When can a finitely generated group be imbedded in a finitely presented
group? The answer to this purely group-theoretic question reveals a har-
monic interplay of Group Theory and Mathematical Logic. The proof we
present here is due to S. Aanderaa (1970).
The following technical lemma is just a version of the "trick" which allows
an arbitrary word on an alphabet to be viewed as a positive word on a larger
alphabet.
Lemma 12.17. Every group G has a presentation
in which every relation is a positive word on 'I/: If G is finitely generated (or
finitely presented), there is such a presentation in which Y (or both Y and Y) is
finite.
ProoJ If G = (XIA) is a presentation, define a new set X' disjoint from X and
The Higman Imbedding Theorem
in bijective correspondence with it via x ++xf,
X' = {xfIxE X),
and define a new presentation of G:
G = (X u X' /A',
xx', x E X),
where A' consists of all. the words in A rewritten by replacing every occur-
rence of every x-' by xf. sl
Definition. A group R is recursively presented if it has a presentation
R = (ul, ...,urnlco= 1, o E E),
where each co is a positive word on u,, ...,urn,and E is an r.e. set.
The lemma shows that the positivity assumption, convenient for notation,
is no real restriction on R.
12.12. If a group R is recursively presented, then it has a presentation whose relations
form a recursive set of positive words. (Hint. If the given presentation is
R =(ul,..., umlmk= l,k 2 0),
where ( a k = 1, k 2 0) is an r.e. set of positive words, define a new presentation
(ul,..., u,, y l y = 1, y k o k = 1, k 2 O).)
12.13. Every finitely generated subgroup of a finitely presented group is recursively
presented. (Hint. Consider all words that can be obtained from 1 by a finite
number of elementary operations.)
12.14. Every recursively related group can be imbedded in a two-generator recur-
sively presented group. (Hint.Corollary 11.80.)
Theorem 12.18 (6.Higman, 8961). Every recursively presented group R can be
imbedded in a finitely presented group.
With Exercise 12.13, Higman's theorem characterizes finitely generated
subgroups of finitely presented groups.
Assume that R has a presentation
where E is an r.e. set of positive u-words. There is thus a Turing machine T
(with alphabet {so, ..., s,) containing {u,, ..., u,)) enumerating E; more-
,over, by Exercise 12.1l, we may assume that T has stopping state q,. We are
going to use the group g(T), constructed in Boone's lemma, arising from the
semigroup-F'(T).Now the original Markov-Post semigroup y(T) was rewrit-
452 12. The Word Problem The Higman Imbedding Theorem

ten as T(T) for the convenience of the proof of the unsolvability of the word
problem. For Higrnan's theorem, we shall rewrite y(T) another way. Of
course, this will engender changes in the generators and relations of W(T), (i) H3(T) is an HNN extension with base B,(T) and stable letter z.
and so we review the construction. Beginning with a Turing machine T with (ii) B'(7') W(T), and B'(T) is an HNN extension with base g3(T) and sta-
stopping state q,, we constructed y(T) with generators q, h, q,, . ..,q,, so, ..., ble letter K.
s,, and certain relations. The semigroup T(T) renamed h as the last s-letter;
thus, T(T) has generators q's and s's and relations those of y rewritten ac- Proof. As the proof of Lemma 12.11. IU
cordingly. Returning to the original notation (with h no longer an s-letter)
gives a group W(T) with generators: The next lemma shows how #(T) simplifies (6).

Lemma 12.20. If o is a positive word on so, ...,s,, then o E E if and only if

and relations those of the original W(T) but with the relations A, rewritten rc(o-' 70) = ( o - ' z o ) ~in Wf(T).
accordingly:
A',: for all /3 = 0, . ..,M and i E I, Proof. The equation w(h-'q, oh) = 1 in g ( T ) has this simpler form in B'(T)
once t and k are replaced by z and K,respectively. 81
silxsB= x2, h-'xh = x2,

r:', ssxri = XS;', r ~hxri

l = xh-l, Form the free product 9'(T)* R. Recall that R is generated by
{u,, ...,u,). At the outset, the Turing machine T enumerating the relations
ri-' FTqi, Giri = Htqi2Ki. of R was chosen so that its alphabet {so,..., s,) contains {u,, ..., u,). Of
By Corollary 12.9, a positive s-word o lies in E if and only if w(h-' q, oh) = 1 course, the generating sets of the free factors of 9 ( T ) * R must be disjoint.
in W(T); that is, o E E if and only if Let us, therefore, introduce new letters {a,, ..., a,) c {so,..., s,) c #(T)
for the replica of (u,, ...,u,) c R. Henceforth, we will regard the r.e. set E as
(6) k(h-lo-lq;' hth-'q, oh) = (h-'o-lq;' hth-' q, oh)k in g(T). comprised of certain positive words on {a,, ...,a,) c {so,..., s,). Our re-
Let us introduce new notation to simplify this last equation. First, define writing is completed.
W2(T)as the group with the presentation Now define new groups W4, W,, and W6 as follows (these also depend on
IS but we abbreviate notation):
~ 2 ( T ) = ( ~ 7 h , q o , - - - 7 q ~ , S 0 ~ . . . , S~EIIA',).
~,~i,
B4 = (B'(T)*R; b,, . .., b,lb;'ujbi = u,, b;'ajbi = a,,
Now introduce new symbols:
b i l ~ b=
i xu;', all i, j = 1, ... ,m);
z = q~lhth-lq, and lc = hkh-'.
g5= (B4;d Id-' lcd = rc, d-I aibid = ai,i = 1, . .., m);
Define a new group a 3 ( T ) by the presentation
Wk(T) = (g2(T);~lz-'(q;'hr~h-~q,).r = q;lhrih-'ql,
Aanderaa's proof of Higrnan's theorem is in two steps. The first step shows
~-~(q;lhxh-'q,)z = q;lhxh-lq,). that each of these groups is an WNN extension of its predecessor:
Note that Wj(T) is just another presentation of the group
by Theorem 11.78, each group is imbedded in its successor, and so R is a
subgroup of 9 6 . The second step shows that W6 is finitely presented. After the
as can be quickly seen by replacing z by its definition. Similarly, we define
proof is completed, we shall see that the diagram in Plate 4 partially explains
#(T) = (H3(T); ~Irc-'(hr~h-~)rc
= hrih-l, rc-'(hxh-l)rc = hxh-l, how the generators and relations of the groups B4, W,, and W6 arise.
rc-l(hq-lh-'q, t0qy1hqh-')~= hq-'h-lq, toqllhqh-I). Lemma 62.21. The subgroups (a,, ...,a,, rc) and (a,, ...,a,, z) of #(T) *R
Replacing rc by its definition shows that #(T) is another presentation of ,are free groups with respective bases the displayed generating sets.

W(T) = (#; (T); k 1 kri = rik, i E I, kx = xk, k(q-' tq) = (q-I tq)k). Proof. Recall our analysis of 9 ( T ) in Lemma 12.11: Bl is an HNN extension
454 12. The Word Problem
with base Bo and stable letters {so,..., s,); in our present notation, 3, has
stable letters {h, so, .. ., s,). It follows from Lemma 11.76 that (h, so, ...,
s,) is a free group with basis {h, so, .. ., s,). Since B, 5 #(T), this last
statement holds in #(T) * R. But {a,, ..., a,) c {so, . . ., s,), so that
(a,, .. .,a,) is free with basis {a,, ... , a,).
We now show that {a,, . . . , a,, rc) freely generates its subgroup (a similar
argument that {a,, ...,a,, z) freely generates its subgroup is left to the read-
er). Otherwise, there is a word
where e, = f 1 and c, are (possibly empty) freely reduced words on
{a,, ..., a,); we may further assume, of all such words o , that n is chosen
minimal. Since w involves no u-letters, we have w = 1 in #(T). As g ( T ) is
an HNN extension with base 3A(T) and stable letter rc, Britton's lemma says
that either w does not involve rc or u contains a pinch rc"~,rc-~,where c,
is a word on {hrih-', i E I, hxh-', hq-' h-'q, zq;'hqh-' ). But the relations
in #(T) show that K commutes with c,, so that r c " c , ~ -= ~ c, in g ( T ) 5
#(T) * R, and this contradicts the minimality of n. It follows that o does not
involve rc; that is, w is a reduced word on {a,, ...,a,). But we have already
seen that (a,, . ..,a,) is free with basis {a,, ..., a,), so that w = 1.
Lemma 112.22. g4in an H N N extension with base #(T) * R and stable letters
{b,, . -,bm)-
Proof. It suffices to show there are isomorphisms q,: A, -+ B,, where
Bi= (ul, ..., u,, a,, ..., a,, ku,"),
a d cpi(uj)= %, qi(aj) = aj, and cpi(rc) = rcuil. Note that A i= Bin
It is easy to see, in g ( T )* R, that
Ai = (al, ..., a,, rc)*(u,, ..., u,)
= (a,, ..., a,, rc)*R.
By Lemma 12.21, (a,, ..., a,, rc) is freely generated by {a,, ..., a,, rc), so
that qi is a well defined homomorphism. Similarly, the map t,bi: A, -+ A,, giv-
en by $,(uj) = uj, $,(aj) = aj, and $,(rc) = rcu,, is a well defined homomor-
phism. But $, is the inverse of q,, so that q, is an isomorphism and a4is an
HNN extension. W
Lemma 12.23.3, is an H N N extension with base B4 and stable letter d.
Proof. It suffices to show that there is an isomorphism
(P:A = < ~ , a l b,...,
l a,b,)-+B= (rc,a,, ..., a,)
with cp(rc) = rc and q(aibi) = a, for all i.
The Higrnan Imbedding Theorem 455
Since rc-'b,rc = biui in a 4 , the function 8: B4-+#(T)* R, defined by
sending each bi to 1, each ui to 1, and all other generators to themselves, is
a well defined homomorphism (it preserves all the relations: bi = 1 implies
1 = rc-' birc = biui = u,). The map 8 takes each of (K, a,b,, ...,a,b,) and
(K, a,, ...,a,) onto the subgroup (rc, a,, . ..,a,) < #(T) * R which, by the
preceding lemma, is free on the displayed generators. By Exercise 11.8, each
of the two subgroups A and B of 9Y4 is free on the displayed generators, and
so the map q: A -+ B given above is a well defined isomorphism. III
The next lemma will be needed in verifying that B6 is an HNN extension
of a,.
Lemma 12.24. The subgroup A = (rc, a,, .. .,a,, z) 5 #(T) has the presenta-
tion
A = (rc,a,, ..., a , , z I ~ - ~ o - ~ z=wC~U - ~ Z ~ E).
,CUE
Remark. Recall our change in notation: although E was originally given as a
set of positive words on {u,, ... , u,), it is now comprised of positive words
on {a,, ..., a,,,}.
Proof. By Lemma 12.20, the relations rc-lw-'zwic = w-lzo, for all o E E, do
hold in W'(T), and hence they hold in the subgroup A #(T). To see that
no other relations are needed, we shall show that if [ is a freely reduced word
c
on {K,a,, ...,a,, z) with = 1 in A, then [ can be transformed into 1 via
elementary operations using only these relations.
Step 1. [ contains no subword of the form zEwrc7where E = $. 1, q = f I,
and u E E.
It is easy to see that the given relations imply
If [ contains a subword z E w d ,then
[E cl ~ ~ ~ -+ [,
i ) Ol K
~ ~~ ~c - ~ ~ ~ ~ U ~ Z
is an elementary operation. Cancel all subwords (if any) of the form yy-I or
y-' y, where y E z, rc, or some aj. With each such operation, the total number
of occurrences of zEwhich precede some rcQoes down. Therefore, we may
c
assume that is freely reduced and contains no subwords of the form zEoicV.
Step 2. [ involves both K and z.
If [ does not involve rc, then it is a word on (a,, . .., a,, z). But this set
c
freely generates its subgroup, by Lemma 12.21, and so being freely reduced
c
and [ = 1 imply = 1. A similar argument shows that ( involves z as well.
Since 3'(T) is an HNM extension with base B$(T) and stable letter rc,
c
Britton's lemma says that contains a pinch rceVrc-", where e = f 1, and
456 12. The Word Problem The Higman Imbedding Theorem 457

there is a word D on {hrih-l, i E I, hxh-I, hq-lh-lq,zq;lhqh-l) with letter z. Assume now that neither V nor D involves z. Then V is a word on
{a,, .. ., a,) and D is a word on {hrih-', i e I, hxh-I); we may assume that D
has been chosen so that the total number of occurrences of r-letters is mini-
Choose D so that the number of occurrences of z in it is minimal. mal; moreover, we may assume that all adjacent factors equal to hxh-I are
collected as hxmh-'. It follows that the equation V = D holds in the subgroup
Step 3. D is z-reduced. @,(T), which is an FINN extension with base @, * (q, q,, ...,q,) and stable
Now g 3 ( T )is an HNN extension with base @,(T) and stable letter z. Let letters {r,, i E I).
us write Now V is ri-reduced for all i because V; being a word on {a,, ..., a,), does
6 = hq-lhdlq,, not even involve any r-letters. We claim that D is also ri-reduced for all i.
Otherwise, D (a word on {hrih-', i E I, hxh-'1) contains a pinch: there is thus
so that an index i with
hq-I h-l q, zqf hqh-I = 626-I. D = A, hr;h-'A, hr;'h-lA3,
If D, which is now a word on {hrih-l, i E I, hxh-', SzS-'), is not z-reduced,
then it contains a pinch. Since an occurrence of z can only arise from an
occurrence of 6zJw1,it follows that
-
where I = & 1 and A, involves no r-letters (just check the cases I = 1 and
I = - 1 separately); hence, A, hxmh-I (since it is freely reduced), and

D=D , 6 ~ ~ 6 - ~ ~ , 6 z - ~ 6 - ~ ~ , ,
where D, does not involve the stable letter z (just check the cases f = 1 and
f = - 1 separately); moreover, there is a word W on {q;'hrih-'q,, i e I,
qllhxh-I q, ) with
6-I D26 = W in @,(T)
(the subgroups A and B in the HNN extension are here equal, and so we need
not pay attention to the sign off). From the presentation of B2(T),we see
that z and W commute. Therefore,
D = D , ~ ~ ~ W ~ - ~ G - ' D=
, D16W6-ID3 inB,(T),
contradicting our choice of D having the minimal number of occurrences of
z. It follows that D is z-reduced.
Step 4. V is z-reduced.
Otherwise, V contains a pinch zgCz-g, where
and W; a word on {q;'hrih-'q,, i e I, q~lhxh-lq,) (as above), commutes
with z in B,(T). Wow V does not involve rc, so its subword C involves neither
K. nor z. Since (, hence its subword V; is a word on {rc, a,, ...,a,, 23, it follows
that C is a word on {a,, ...,a,). But (z, a,, ...,a,) < &(T) * R is a free
group with basis the displayed generators, by Lemma 12.21, and so C
commutes with z if and only if C -- 1. Therefore, the pinch T ~ C T=- ~
contradicting ( being freely reduced.
Step 5. Both V and D involve z.
Since V = D in W3(T) and both are z-reduced, Corollary 11.82 applies to
show that both of them involve the same number of occurrences of the stable
The pinch in D is thus r;xmryL,and Britton7slemma concludes, depending on
the sign of I, that xm is equal in 2, * (q, qo, ..., q,) either to a word on
{F,XqilG,, sox, ...,sMx) or to a word on {HtqizKi, sox-', ...,sMx-I). As D
does not involve q-letters, xmis equal in B, to a word on either (sox, ...,
sMx)or {sox-', ...,s,x-l), and we have already seen, in the proof of Lemma
12.14, that this forces m = 0. Therefore, the pinch is r;riz, contradicting our
choice of D having the minimal number of r-letters. We conclude that both V
and D are ri-reduced for all i. By Corollary 11.82, D involves no r-letters
(because V involves none), and D is a word on hxh-l; that is, D = hxnh-I in
a,. In this step, V is assumed to be a word on {a,, ..., a,) c {so,.. .,sM),SO
that the equation V = D holds in @, and
Recall that $7, is an HNN extension with base (x) and stable letters
{h, so, .. ., s,). If V involves aj for somej, then Britton's lemma gives a pinch
+
aj"UajV,where v = 1 and U is a power of x (for aj is a stable letter). Now
h # aj, for h & {so, ...,sy), SO that this pinch must be a subword of V. But V
does not involve x, and so U = 1; therefore aj"aJTvis a subword of V; contra-
dicting ( and its subword V being freely reduced. It follows that V involves no
aj; as V is now assumed to be a word on a-letters, we have V = 1. Recall that
V arose in the pinch rce VK-", a subword of [,and this, too, contradicts [ being
freely reduced.
Step 6. V contains a subword zEVp, where VPis a positive a-word lying in the
r.e. set E.
, Since D, a word on {hrih-', i E I, hxh-', hq-I h-'q, zq;' hqh-I ), involves z,
it must involve hq-' h-I q, zqll hqh-'. Write
458 12. The Word Problem The Higman Imbedding Theorem

where a = 1 and the word in parentheses is the final occurrence of the long Pro@$ (i) It suffices to show that there is an isomorphism
word involving z in D; thus, A is a word on (hrihdl, i E I, hxh-I].
Since V involves z, we may write
cp: A = (rc, z, a,, .. ., a,) -+ B = (rc, zd, a,, ..., a,)
with V(K)= K,q(z) = td, and tp(aj) = a, for all j. Since

where E~ = & 1 and each J$ is a freely reduced word on a-letters. By Steps 2

and 3, both D and V are z-reduced, so that Corollary 11.82 says that the subgroup A is precisely the subgroup whose presentation was determined
in the previous lemma:
is a pinch. Thus, there is a word Z on (q;'hrih-'q,, qylhxh-lq,] with
To see that cp is a well defined homomorphism, we must show that it
of course, we may choose preserves all the relations; that is, if o E E, then
. Z Eq ~~hL,h-~~,,
where L, is a word on (x, ri, i E I]; similarly, since A is a word on (hrih-l,
i E I, hxh-I], we may write We shall show that this last equation does hold in B,, and hence it holds in
B I9,.
Let us introduce notation. If o is a word on (a,, ...,a,), write q,to
where L, is a word on (x, ri, i E I). Substituting, we see that denote the word obtained from o by replacing each aj by bj, and let w, denote
the word obtained from o by replacing each aj by uj. If o E E, then o, = 1,
for o, is one of the original defining relations of R. For o E E, each of the
and we rewrite this equation as following equations holds in B,:

L,'h-'q,&hL2 =q inB2(T).
Note that Vph is freely reduced, for T/, is a freely reduced word on a-letters,
and h is not an s-letter, hence not an a-letter. By Lemma 12.15 (with X = h (for daid-' = aibi in B5 and ai and bj commute in g45 B,). Since K and
and Y = Vph),we have Y a positive word, so that its subword $/, is a positive commute in g 5 , we have
word on a-letters; moreover,

By Lemma 12.4, VpE E. Returning to (7), the birthplace of Vp, we see that
2". T/, is a subword of V. Indeed, V E V'z'p Vp, where V' is the initial segment
of V.
(because bi and uj commute and rc-I birc = biui)
Step 7. Conclusion.
Recall that V arose inside the pinch rce Vc-", which is a subword of [. From
c.
the previous step, we see that rce V'z'p T/,ic-" is a subword of In particular, [ (because o, = 1). We have shown that
+ +
contains a subword of the form z"corcq,where E = 1, rl = 1, and o E E.
c
But we showed, in Step 1, that contains no such subword. This completes
the proof. 4

Lemma 12.25. On the other hand,

o-'zdo = w-'(dad-l)d
ji) B, is an HNN extension with base 39, and stable letter a.
(ii) R is imbedded in B,.
 12. The Word Problem The Higman Imbedding Theorem

as we saw above. Therefore, Inserting ~ K K - ' o-l and om-' gives

K-I o-I z ~ o-l zdo in 3,
d o=
and q:A -+ B is a well defined homomorphism. The terms in parentheses are equal (Lemma 12.20 again), so that canceling
To see that q is an isomorphism, we construct a homomorphism $: 98, -+ gives
9, whose restriction $lB is the inverse of 9.Define $ by setting $lgf(T) to (8) K-' W-I do^ = o-' do.
be the identity map and
Now the relations da,d-I = aibi and aibj = bjai, all i, j, give
dud-I = ow,,
Inspection of the various presentations shows that $ is a well defined
homomorphism. Since $(K) = K, $(a,) = a,, and $(zd) = z, we see that $1B is hence
the inverse of q. o = d-'oco,d,
(ii) This follows from several applications of Theorem 11.78. i l for every word o on (a,, ...,a,), and so
(9) a-ldco = o-ldd-'ww,d = o,d.
The following lemma completes the proof of the Higman imbedding
theorem: Substituting (9) into (8) gives
o, = I C - ~( ~ - ~ d o ) ~ =
d -K-'l o,d~d-l.
Lemma 12.26. is finitely presented.
Since K and d commute,
Proof. The original presentation of R is

On the other hand, the relations K-' b i =

~ biuiand biuj = ujbi, all i, j, give
where E is an r.e. set of positive words on (u,, ...,u,). Recall the notation
introduced in the proof of Lemma 12.25: if w is a word on (a,, ...,a,], then
o,and o, are obtained from o by replacing each a, by ui or b,, respectively. This last equation coupled with (10) gives
With this notation, the presentation of W can be rewritten:

and so o, = I, as desired.
Now %(T)* R is a flnaitely generated g o u p having a finite number of reia-
tions occurring in the presentation of 9 ( T )together with the (possibly infi- Let us review the proof of Rigman's theorem to try to understand
nitely many) relations above for R. Each step of the construction of g6from Aanderaa's construction. Certainly, some of the relations of g6are present to
g ( T ) * R contributes only finitely many new generators and relations. Thus, guarantee a chain of HNN extensions, for this gives an imbedding of R into
3 6 is finitely generated, and it is finitely presented if we can show that every g6.The proof of the last lemma, showing that g6is finitely presented,
relation of the form o, = 1, for o E E, is a consequence of the remaining amounts to proving, for o E E, that o, = 1 follows from the other relations;
relations in B6. that is, one can subdivide the labeled directed polygon with boundary word
By Lemma 12.20, ~ - ~ w - ' z w = o-'zw
~ for all o E E. Hence o, into relator polygons corresponding to the other relations in g6.E. Rips
has drawn a diagram (Plate 4) that helps explain the construction of g 6 .
Before we examine Plate 4, let us discuss diagrams in the plane from a
Since a commutes with K and with all a,, this gives different viewpoint, Regard the plane as lying on the surface of a sphere, and
assume that the north pole, denoted by co, lies outside a given diagram.
Otherwise said, we may regard a given planar diagram D having n regions to
actually have n + 1 regions, the new "unbounded" region (containing co)
As a-' zcr = zd, this gives
being the outside of D. We now propose redrawing a diagram so that the
unbounded region is drawn as an interior region. For example, assume that
 12. The Word Problem The Higrnan Imbedding Theorem

h
Figure 12.12
k0
Figure 12.14

Figure 12.13

Figure 12.12 shows that o = aba-lb2 = 1 in some group. Figure 12.13 is

a redrawn version of Figure 12.12 with co marking the old unbounded
region.
To redraw, first number all the vertices, then connect them as they are
connected in the original diagram. Note that all the (bounded) regions are
relator regions corresponding to the inverses of the original relations, with Figure 12.15
the exception of that containing co. The boundary word of the region with co
is sbt, as in the original diagram. In general, every (not necessarily bounded)
region in the redrawn diagram is a relator polygon save the new region
We begin, therefore, with a labeled directed octagon for this word as well
containing co whose boundary word is o.Such a diagram will show that as with a "balloon" region (containing co) inside having boundary word
o = 1 if every region (aside from that containing GO) is a relator polygon and a,.To subdivide, draw a second octagon inside it, and yet a third octagon
the boundary word of the diagram is 1 in the group. perturbed by two d-edges. Now complete this picture, adding a-edges and the
Let us return to 3,.For a word o E E, draw a diagram, new version, subdivision of the bottom, to obtain Plate 4, the diagram showing that o, =
showing that o, = 1 in 3,(using only the other relations of B6). By Lemma 1 follows from the other relations.
12.20, o E E gives Let us indicate, briefly, how the Novikov-Boone-Britton theorem follows
from the Higman theorem. It is not difficult to construct a recursively

464 12. The Word Problem
presented group G having an unsolvable word problem. For example, let G
be a variant of the group in Theorem 11.85: if F is a free group with basis
{a7 b), let
G = (a, b, plp-'o,p = con for all n E E),
where the commutator subgroup F' is free with basis {a,, o , , .. . , a,,, .. .)
and E is an r.e. set in Z that is not recursive. By Higman's theorem, there is a
finitely presented group G* containing G. If G* had a solvable word problem,
then so would all its finitely generated subgroups, by Exercise 12.4, and this
contradicts the choice of G.
Some Applications
Higrnan's theorem characterizes those finitely generated groups G that can
be imbedded in finitely presented groups. Of course, any (perhaps nonfinitely
generated) group G that can be so imbedded must be countable. In Theorem
11.71, we saw that every countable group G can be imbedded in a two-
generator group G".
Lemma 12.27. If G is a countable group for which G" is recursively presented,
then G can be imbedded in a finitely presented group.
Proof. Higman's theorem shows that G" can be imbedded in a finitely
presented group. ll
At this point, we omit some details which essentially require accurate
bookkeeping in order to give an explicit presentation of GIf from a given
presentation of 6. We assert that there is a presentation of the abelian group
where Di z Q @ (Q/Z) for all i, such that GI' is recursively presented.
Theorem 112.28. There exists a finitely presented group containing an isomor-
phic copy of every countable abelian group as a subgroup.
Proof. By Exercise 10.29, every countable abelian group can be imbedded in
G = x&Di, where Di E Q @ (Q/Z) for all i. Lemma 12.27, with the assertion
that GI' is recursively presented, gives the result. Igl
There are only countably many finitely presented groups, and their free
product is a countable group H having a presentation for which H" is
recursively presented.
Some Applications 465
Theorem 12.29. There exists a uniuersalfinitely presented group a ; that is,
is a finitely presented group and a contains an isomorphic copy of every fi-
nitely presented group as a subgroup.
Proo$ The result follows from Lemma 12.27 and our assertion about the
group HZ'.
Groups with a solvable word problem admit an algebraic characterization.
In the course of proving this, we shall encounter groups which are not finitely
generated, yet over whose presentations we still have some control. Let G be
a group with presentation
G = (xi, i 2 01A)
positive integer g(w) = n;=, -
in which each 6 E A is a (not necessarily positive) word on (xi, i 2 0), let Q be
the set of all words on {xi, i 2 O), and let R = { o E Q: o = 1 in G). Encode
52 in N using Godel numbers: associate to the word o x;: ...x c the
p$,pi;,"z, where p, < p, < -..is the sequence of
primes (note that 1 + e, 2 0). The Godel image of this presentation is
g(R) = {g(o):o E R).
Definition. A presentation (xi, i 2 01A) is r.e. if its Godel image g(R) is an r.e.
subset of N; this presentation has a solvable wordproblem if g(R) is recursive.
'-J
Definition. A group G is r.e. or has a solvable wordproblem if it has some
presentation which is either r.e. or has a solvable word problem.
We remarked at the beginning of this chapter that a finitely generated
group G having a solvable word problem relative to one presentation with a
finite number of generators has a solvable word problem relative to any other
such presentation. The analogue of this statement is no longer true when we
allow nonfinitely generated groups. For example, let G be a free group of
infinite rank with basis {xi, i 2 0). Now g(R) is recursive (this is not instantly
obvious, for R is an infinite set of nonreduced words; list its elements lexico-
graphically and according to length), and so this presentation, and hence G,
has a solvable word problem. On the other hand, if E is an r.e. subset of N
that is not recursive, then
(xi, i 2 Olx, = 1 if and only if i E E)
is another presentation of G, but g(R) is not recursive; this second presenta-
tion has an unsolvable word problem.
We wish to avoid some technicalities of Mathematical Logic (this is not the
appropriate book for them), and so we shall shamelessly declare that certain
groups arising in the next proof are either r.e. or have a solvable word prob-
lem; of course, the serious reader cannot be so cavalier.
466 12. The Word Problem
A.V. Kuznetsov (1958) proved that every recursively presented simple
group has a solvable word problem.
Theorem 12.30 (Boone-Higman, 1974). A finitely generated group G has a
solvable word problem if and only if G can be imbedded in a simple subgroup
ofsome finitely presented group.
Sketch of Proof. Assume that
where S is a simple group and H is finitely presented; say,
Let !2' denote the set of all words on (h,, . .., h,), and let R' = {or E N: of=
,
I); let R denote the set of all words on (g ,...,g,), and let R = { o E R: o =
1). Theorem 12.2 shows that R' is r.e. If one writes each gi as a word in the h,,
then one sees that R = R nR'; since the intersection of r.e. sets is r-e. (Exer-
cise 12.8), it follows that R is r.e. We must show that its complement ( a E Ck
o # 1) is also r.e. Choose s E S with s # 1. For each w E Q, define N(w) to be
the normal subgroup of H generated by {a, p,, ...,&I. Since S is a simple
group, the following statements are equivalent for o e $2w # 1 in G;
N(o) n S # 1; S 5 N(o); s = 1 in H/N(o). As H/N(w) is finitely presented,
Theorem 12.2 shows that the set of all words in R which are equal to 1 in
H/N(w) is an r.e. set. A decision process determining whether o = 1 in G thus
consists in checking whether s = 1 in H/N(co).
To prove the converse, assume that G = (g,, ...,g,) has a solvable word
problem. By Exercise 12.10,
a:u # 1 and v # 1)
{(u, V ) E Q x
is a recursive set; enumerate this set (u,, u,), (u,, v,), .. . (each word u or v has
many subscripts in this enumeration). Define Go = G, and define
GI = (G; x,, ti, i 2 01t;'uix;' uixl ti = v i ~ ~ 1 u i 2
~ 10).7
It is plain that G, has base G * (x, ) and stable letters {ti, i 2 0); it is an HNN
extension because, for each i, both Ai = (uix;'uixl) and Bi = (vix;'uixl)
are infinite cyclic. Thus, G 5 G,. One can show that this presentation of GI
has a solvable word problem (note that G, is no longer finitely generated).
We now iterate this construction. For each k, there is an HNN extension G,
,,,
with base Gk-, * (x,), and we define S = U G,; clearly G 5 S. To see that
S is simple, choose y v E S with u # 1 and v # 1. There is an integer k with
u, v E Gk-,. By construction, there is a stable letter p in Gk with
Therefore,
u = (p-' up) (p-' xi1ux,p) (xi1u-I x,)
Some Applications 467
lies in the normal subgroup generated by u. Since u and v are arbitrary
nontrivial elements of S, it follows that S is simple.
It can be shown that S is recursively presented. By Theorem 11.71, there
is a two-generator group S" containing S; moreover, S" is recursively
presented. The Higrnan imbedding theorem shows that S", hence S, and
hence G, can be imbedded in a finitely presented group H.
It is an open question whether a group with a solvable word problem can
be imbedded in a finitely presented simple group (the simple group S in the
proof is unlikely to be finitely generated, let alone finitely presented).
Our final result explains why it is often difficult to extract information
about groups from presentations of them. Before giving the next lemma, let
us explain a phrase occurring in its statement. We will be dealing with a set
of words R on a given alphabet and, for each o E R, we shall construct a
presentation P ( o ) involving the word w. This family of presentations is uni-
form in o if, for each ofE R, the presentation 9 ( a f )is obtained from P(w) by
substituting orfor each occurrence of w. A presentation (XI A) is calledfinite
if both X and A are finite sets; of course, a group is finitely presented if and
only if it has such a presentation.
Lemma 12.31 (Rabin, 1958). Let G = (ZIA) be a finite presentation of a group
and let !2 be the set of all words on Z. There are finite presentations {P(o):
o E Q), uniform in w, such that if R(o) is the group presented by P(u),then
(i) fi o # 1 in G, then G I R(w); and
(ii) if w = 1 in G, then B(o) presents the trivial group 1.
Proof(cC.F. MilIer, Let (x) be an infinite cyclic group; by Corollary
11.80, G * (x) can be imbedded in a two-generator group A = (a,, a,) in
which both generators have infinite order. Moreover, one can argue, as in
Exercise 11.50, that A can be chosen to be finitely presented: there is a finite
s s d of words on (a,, a,) with
A = (~17a21A).
Define
B = (A; b,, b2lb,la,b1 = a:, b,'a2b2 = a;).
It is easy to see that B is an HNN extension with base A and stable letters
(b,, b,), so that G I A I B. Define
C = (B; clc-lblc = b;, cW1b2c= bi).
Clearly C has base B and stable letter c; C is an HNN extension because b,
and b,, being stable letters in B, have infinite order. Thus, G I A I B I C.
If w E Q and o # 1 in G, then the commutator
[a,XI= o ~ o - ~ x - ~
has infinite order in A (because G * (x) 5 A).
468 12. The Word Problem
-
We claim that (c, [o, x]) 5 C is a free group with basis {c, [o, x] 1. Sup-
pose that V is a nontrivial freely reduced word on (c, [w, x]) with V = 1 in
C. If V does not involve the stable letter c, then V [w, xln for some n f 0,
and this contradicts [o, x] having infinite order. If V does involve c, then
Britton's lemma shows that V contains a pinch cefNc-"as a subword, where
e = + 1 and W E (b,, b,). But V involves neither b, nor b,, so that W r 1
and ceWc-" -. cec-", contradicting V being freely reduced. Therefore, V must
be trivial.
We now construct a second tower of HNN extensions. Begin with an
infinite cyclic group (r), define
S = (r, sjs-l rs = r2),
and define
T = (S; t 1 t-'st = s2).
Since both r and s have infinite order, S is an HNN extension with base (r)
and stable letter s, and T is an HNM extension with base S and stable letter
t. Britton's lemma can be used, as above, to show that {r, t) freely generates
its subgroup in T.
Since both (r, t) 5 T and (c, [o, x]) IC are free groups of rank 2, there
is an isomorphism q between them with q(r) = c and q(t) = [a, x]. Form the
amalgam R(w) = T *, C with presentation
P ( 0 ) = (T* Clr = c, t = [o, x]).
We conclude from Theorem 11.67(i) that if o # 1 in G, then G < C I R(o).
If o = 1 in G, the presentation Y(o) is still defined (though it need not be
an amalgam). The presentations g(w) are uniform in o:
P(w) = (a,, a,, b,, b,, c, r, s, tlA, bilaibi = a?, cdlbic = b?,
i = 1,2, s-lrs = r2, L-'s~= s2, r = C,t = [w, XI).
We claim that Y(o) is a presentation of the trivial group if w = 1 in G.Watch
the dominoes fall: o = 1 = [GO,x] = 1 t = 1 s = 1 +-r = 1 * c = 1 3
b1=1=b2*a1=1=a2.
Definition. A property 4 of finitely presented groups is called a Markov
property if:
'
L .
.'
(i) every group isomorphic to a group with property JH also has property
JZ;
(ii) there exists a finitely presented group G, with property JH; and
(iii) there exists a finitely presented group G2 which cannot be imbedded in a
finitely presented group having property d.
Here are some examples of Markov properties: order 1; finite; finite expo-
nent; p-group; abelian; solvable; nilpotent; torsion; torsion-free;free; having a
Some Applications 469
solvable word problem. Being simple is also a Markov property, for the
Boone-Higrnan theorem shows that finitely presented simple groups must
have a solvable word problem (and hence so do all their finitely presented
subgroups). Having a solvable conjugacy problem is also a Markov property:
a finitely presented group G, with an unsolvable word problem cannot be
imbedded in a finitely presented group H having a solvable conjugacy prob-
lem, for Pi and all its finitely presented subgroups have a solvable word
problem. It is fair to say that most interesting group-theoretic properties are
Markov properties.
The following result was proved for semigroups by Markov (1950).
Theorem 12.32 (Adiala-Rabim, 11958). If JH is a Markov property, then there
does not exist a decision process which will determine, for an arbitrary finite
presentation, whether the group presented has property JZ.
ProoJ Let G, and G2 be finitely presented groups as in the definition of
Markov property, and let B be a finitely presented group with an unsolvable
word problem. Define G = G, *B, construct groups R(w) as in Rabin's
lemma, and define (finitely presented) groups 2(0) = G, * R(w).
Restrict attention to words co on the generators of B. If such a word o # 1
in B, then G, < G 2 R(o) 5 2(o). But the defining property of G, implies
that 2(o) does not have property 4.If, on the other hand, o = 1 in @, then
R(o) = 1 and 2(o) 2 G, which does have property JH.Therefore, any deci-
sion process determining whether 2(w) has property 4 can also determine
whether o = 1 in 8; that is, any such decision process would solve the word
problem in 99. ISl
ColroBlarry 12.33. There is no decision process to determine, for an arbitrary
finite presentation, whether the presented grozlp has any of the following
proper~ies:order 1; finite; finite exponent; p-group; abelian; solvable; nil-
potent; simple; torsion; torsion-free; free; solvable word problem; solvable
conjugacy problem.
Pro@$ Each of the listed properties is Markov.
Corollary 12.34. There is no decision process to determine, for an arbitrary pair
of finite presentations, whether the two presented groups are isomorphic.
ProoJ Enumerate the presentations P I , P,, ... and the groups G,, G,, ...
they present. If there were a decision process to determine whether Gi r Gj
for all i and j, then, in particular, there would be a decision process to
determine whether Pnpresents the trivial group.
While a property of finitely presented groups being Markov is sufficient for
the nonexistence of a decision process as in the Adian-Rabin theorem, it is
 12. The Word Problem
not necessary. For example, the property of being infinite is not a Markov
property. However, a decision process that could determine whether the
group given by an arbitrary finite presentation is infinite would obviously
determine whether the group is finite, contradicting Corollary 12.33. Indeed,
this example generalizes to show that the Adian-Rabin theorem also holds
for the "complement" of a Markov property.
Does every finitely presented group have some Markov property?
Theorem 12.35. A finitely presented group H satisfies no Markov property
if and only if it is a universal finitely presented group (i.e., H contains an
isomorphic copy of every finitely presented group as a subgroup).
Proof: Recall that the existence of universal finitely p r e ~ e n t e d . ~ r o was
u~s
proved in Theorem 12.29. -
Let H be a universal finitely presented group, and assume that H has some
Markov property A. There is some finitely presented group G, that cannot
be imbedded in a finitely presented group with property A?.But G, can be
imbedded in H, and this is a contradiction. The converse follows from the
observation that "not universal" is a Markov property. 6111
Epilogue
Any reader wanting to study Group Theory more deeply must first learn
Representation Theory, the analysis of homomorphisms cp: G -+ GL(n, K),
where K is an algebraically closed field. There are two initial approaches to
this subject, and both approaches must eventually be mastered. One ap-
proach, historically the first, is character theory. If cp: G -+ GL(n, C) is a
homomorphism, then q(g) is an n x n complex matrix for each g E G; its
character ~ ( q )G: --+ C is defined to be the trace function g~ tr(cp(g)) (the
values of ~ ( q are ) actually algebraic integers). Of course, if g and gf are
conjugate in 6 , then tr(cp(g)) = tr(q(gf)),so that ~(cp)is really a class function;
that is, ~ ( ( p can
) be regarded as a complex-valued function on the family of
conjugacy classes of 6;. Each character can be uniquely written as a linear
combination of irreducible characters, and the number c of such irreducible
characters is equal to the number of conjugacy classes of 4;. The c x c matrix
containing the values of all the irreducible characters is called the character
table of G. It contains much important information about G, and sufficient
machinery has been developed to allow explicit calculation, in many cases, of
its entries. There are wonderful applications that arise quite early: Burnside's
p'qP-theorem: Every group of order paq5, where p and q are primes, is solv-
able; a theorem of Frobenius: If H is a subgroup of a finite group G such that
H n x H x - ' = 1 for all x $ H , then N = { l ) u ( G - U X e c x ~ x - ' )is a
(normal) subgroup of G (in a Frobenius group G, this shows that the
Frobenius kernel is actually a subgroup). The further one goes into Group
Theory, the more Representation Theory arises, and many of the best
theorems involve some use of representations.
The theory still works when C is replaced by any algebraically closed field
K whose characteristic does not divide /GI; this is the so-called ordinary
representation theory. When the characteristic p of K divides I GI, the study,
472 Epilogue
called modular representation theory, becomes more intricate, but it, too, is an
essential tool.
Let us now discuss the second approach to representations. If K is a field,
then the group algebra KG of a finite group G over K is the vector space over
K having the elements of G as a basis and which is equipped with the multi-
plication (called convolution) determined by the given (group) multiplication
of its basis elements. If q:G -,GL(n, I<) is a homomorphism and if V is an
n-dimensional vector space over K, then one may view V as a KG-module
(and conversely). When K = C, one sees, for example, that ~ ( 9is)irreducible
if and only if V is an indecomposable module. This point of view is quite
valuable; for example, it allows ideas and techniques of Homological Algebra
to be used.
There are many excellent books on Representation Theory. For example:
Alperin, Benson, Curtis and -Reiner, Dornhoff, Feit (1967 and 1982), Isaacs,
James and Liebeck, Puttaswamaiah and Dixon, and Serre (1977).
A Personal Note. If Representation Theory is so important, why have I not
included it in this book? It is not because the beginnings of the subject require
greater sophistication on the part of the reader.
Let me explain with an analogy. I have long felt that many entering univer-
sity students who have seen some Calculus in high school are at a disadvan-
tage. There are, to be sure, good Calculus courses taught in high schools, and
those students who have done well in such courses are well prepared. But, too
often, high school Calculus courses are inadequate, so that, upon completion,
even good students (with good teachers) are poorly prepared. As a conse-
quence, many students must start learning the subject anew when they enter
the university. Their time has been wasted and their enthusiasm has been
dampened.
I feel that one chapter on Representation Theory is necessarily inadequate;
it is like a bad high school Calculus course that leaves one unprepared. After
a longish excursion into Ring Theory (includingthe theorems of Wedderburn
and Maschke), one learns the first properties of characters and how to
compute them, and one proves the theorems of Burnside and Frobenius
mentioned above. However, a group theorist must have a more thorough
course in order to feel comfortable with both characters and modules. Most
likely, a student having read only one chapter in a text like this ohe would
still have to begin the subject anew, and this would be a waste of valuable
time.
Here are some suggestions of other topics in Group Theory that the reader
may wish to pursue. For general group theory, see Huppert, Huppert and
Blackbum (1981 and 1982), Robinson (1982), and Suzuki (1982 and 1986).
Simple Groups. All finite simple groups were classified by the 1980s, and
there is an explicit description of them all. This is the most profound and
Epilogue 473
sophisticated part of Group Theory, using every known technique. Introduc-
tions to this study are Artin (1957), Aschbacher (1994), Borel, Carter (1972
and 1985), Conway et al., Dieudonne, and Gorenstein (1982 and 1983). For
some applications of the classification theorem, see P.J. Cameron, Finite
permutation groups and finite simple groups, Bull. London Math. Soc. 63
(198I), pp. 1-22.
Solvable Groups. See Doerk and Hawkes, Huppert and Blackburn (1981),
and Robinson (1972).
p-Groups. We recommend P. Hall's notes "Nilpotent Groups" in his Col-
lected Works, Dixon and du Sautoy and Mann and Segal, Huppert, Huppert
and Blackburn (1981), Khukhro, and Vaughan-Lee.
Cohomology of Groups. For a general account of Homological Algebra, the
reader may look at Cartan and Eilenberg, Mac Lane, and Rotman (1979).
For Cohomology of Groups, which is Homological Algebra specialized to a
group-theoretic context, see Benson, Brown, Evens, Karpilovsky, and Weiss.
Combinatorial Group Theory. This is the study of presentations of groups.
Suggested books are Coxeter and Moser, Johnson, Lyndon and Schupp,
Magnus and Karrass and Solitar, and Zieschang and Vogt and Coldewey.
There is another aspect involving groups acting on trees; we suggest Dicks
and Dunwoody, and Serre (1980). The Cayley graph of a finitely generated
group can be made into a metric space, and the hyperbolic groups introduced
by Gromov can be found in Gersten.
See Higman for further development of his imbedding theorem, Miller for
group-theoretic decision problems, and Epstein et al. for a treatment of auto-
matic groups.
Abelian Groups. We suggest Fuchs (1970 and 1973),Griffith, and Kaplansky.
Finitely Generated Groups, We suggest Kegel and Wehrfritz, Kurosh, Rob-
inson (1972), and Wehrfritz.
History. We suggest Chandler and Magnus, and Wussing.
There are several computer systems devoted to group theory: for example,
MAGMA (nee CAYLEY) and GAP.
Certainly, there are other valuable books, as well as other valuable areas of
Group Theory (e.g., crystallographic groups, Mobius groups, knot groups,
varieties of groups) that I have not even mentioned. I apologize to their
authors and their practitioners.
Primary Sources. One must always look at the masters. The following
474 Epilogue

books contain extensive bibliographies of journal articles: Carter, Coxeter AIPPENE)Ix I

and Moser, Curtis and Reiner, Fuchs, Gorenstein (1982), Huppert, Huppert
and Blackburn, Lyndon and Schupp, Magnus and Karass and Solitar,
Robinson (1982), Scott, and Suzuki. Both Baumslag, and Gorenstein (1974) Some Major Algebraic Systems
contain reviews of the all the articles on Group Theory written between 1940
and 1970.

Semigroup

Abelian group operator group

Ring

Commutative ring

Domain \
\
\ Division ring

Field Vector space

A ring (always containing 1 # 0) is a set with two operations, addition and

multiplication. It is an abelian group under addition, a semigroup with 1
under multiplication, and the two operations are linked by the distributive
laws.
A commutative ring is a ring in which multiplication is commutative.
476 Appendix I. Some Major Algebraic Systems

A domain (or integral domain) is a commutative ring in which ab = 0 APPENDIX II

implies a = 0 or b = 0; equivalently, the cancellation law holds: if ab = ac and
a # 0, then b = c. Equivalence Relations and
A division ring (or skew field) is a (not necessarily commutative) ring in
which every nonzero element has a multiplicative inverse: if a # 0, then there Equivalence Classes
is b E K with ab = 1 = ba. The set of nonzero elements K x = K - (0) is thus
a multiplicative group.
Afield K is a commutative division ring.
It is a theorem of Wedderburn (1905) that every finite division ring is a
field.

A relation on a set X is a subset = of X x X. One usually writes x E y

instead of (x, y) E -; for example, the relation < on the set of real numbers
IW consists of all points in the plane IW x IW lying above the line with equation
y = x, and one usually writes 2 < 3 instead of (2,3) E <.

-
A relation SE on a set X is reflexive if x = x for all x E X; it is symmetric if,
for all x, y E X, x = y implies y = x; it is trnnsitive if, for all x, y, z E X, x y
and y z imply x r z. A relation = on a set X is an equivalence relation if it
is reflexive, symmetric, and transitive.
If = is an equivalence relation on a set X and if x E X, then the eqai~alence
class of x is

[x] = (y E X: y = X) C X.

Proposition H.P. Let r be an equivalence relation on a set X. If x, a E X, then

[x] = [a] if and only if x = a.

Proof. If [x] = [a], then x E [XI, by reflexivity, and so x E [a] = [x]; that is,
x = a.

-
Conversely, if x = a, then a = x, by symmetry. If y E [x], then y = x. By
transitivity, y E a, y E [a], and [x] c [a]. For the reverse inclusion, if z E [a],
then z a. By transitivity, z r x, so that z E [x] and [a] c [x], as desired.

Apartition of a nonempty set X is a family of nonempty subsets {S,: i E I )

such that X = U i , , S i and the subsets are pairwise disjoint: if i # j, then
Si n Sj = a.
478 Appendix 11. Equivalence Relations and Equivalence Classes

Proposition H.2. If = is an equivalence relation on a nonempty set X, then the APPENDIX 111
family of all equivalence classes is a partition of X.

Proof. For each x E X, reflexivity gives x E 1x1; this shows that the equiva-
Functions
,
lence classes are nonempty and that X = U, [x]. To check pairwise
disjointness, assume that [x] n [y] # @. Therefore, there exists an element
z E [XIn [y]; that is, z = x and z = y. By the first proposition, [z] = [x] and
[z] = [y], so that [x] = [y]. tl

Proposition PL.3. If {Si: i E I) is a partition of a nonempty set X, then there is

an equivalence relation on X whose equivalence classes are the Si.

ProoJ If x, y E X, define x = y to mean that there exists Si containing both x

and y. It is plain that = is reflexive and symmetric. To prove transitivity,
assume that x I y and y r z; that is, x, y E Si and y, z E Sj. Since y E Si n Sj,
pairwise disjointness gives Si = Sj; hence x, z E Si and x E Z.
If x E X, then x E Si for some i. If y E Si, then y, x E Si and y r x; that is,
Si c [x]. For the reverse inclusion, ifz E [XI, then z = x, and so z, x E Si; that
is, [x] c Si. H If X and Y are sets, a relation from X to Y is a subset f of X x Y (if X = Y,
one also says that f is a relation on X). A function from X to Y, denoted by
Proposition 11.1 signals the importance of equivalence relations. If r is an f : X -+ Y, is a relation f from X to Y such that for each x E X, there exists a
equivalence relation on a set X and if E is the family of equivalence classes, unique y E Y with (x, y) E f. If x E X, then the unique element y in the defini-
then x r y in X if and only if [x] = [y] in E; equivalence of elements in X tion is denoted by f (x), and it is called the value off at x or the image of x
becomes equality of elements in E. The construction of the new set E thus under f. With this notation, the relation f consists of all (x, f(x)) E X x Y ;
identifies equivalent elements. that is, a function is what is usually called its graph.

- For example, the fractions i and $ are called equal if the numerators and
denominators satisfy "cross multiplication." In reality, one defines a relation
on X = {(a, b) E Z x Z: b # 0 ) by (a, b) r (c, d) if ad = bc, and a straight-
forward calculation shows that = is an equivalence relation on X. The equiv-
alence class containing (a, b) is denoted by a/b, and the set of all rational
numbers Q is defined as the family of all such equivalence classes. In particu-
lar, (1,2) and (2,4) are identified in Q, because (1,2) = (2,4), and so = $.
The set X is called the domain off and the set Y is called the target off.
One defines two functions f and g to be equal if they have the same domain
X, the same target Y,and f(x) = g(x) for all x E X (this says that their graphs
are the same subset of X x Y).
In practice, one thinks of a function f as something dynamic: f assigns a
value f (x) in Y to each element x in X. For example, the squaring function
f : R --+ R is the parabola consisting of all (x, x2)E R x R, but one usually
thinks off as assigning x2 to x; indeed, we often use a footed arrow to denote
the value of fon a typical element x in the domain: for example, f : x I+ x2.
Most elementary texts define a function as "a rule which assigns, to each x in
X, a unique value f(x) in Y." The idea is correct, but not good enough.
For example, consider the functions f, g: R -,R defined as follows: f(x) =
(x + g(x) = x2 + 2x + 1. Are f and g different functions?They are differ-
ent "rules" in the sense that the procedures involved in computing each of
them are different. However, the definition of equality given above shows
that f = g.
If X is a nonempty set, then a sequence in X is a function f : P -,X, where
P is the set of positive integers. Usually, one writes x, instead of f(n) and
one describes,f by displaying its values: x,, x,, x,, . .. . It follows that two
480 Appendix 111. Functions

sequences x,, x,, x,, ... and y,, y,, y,, . . . are equal if and only if x, = y, for
all n 2 1.
APPENDIX IV
The uniqueness of values in the definition of function deserves more com-
ment: it says that a function is "single valued" or, as we prefer to say, it is well Zorn's Lemma
defined. For example, if x is a nonnegative real number, then f (x) = & is
usually defined so that f(x) 2 0; with no such restriction, it would not be
a function (if $ = 2 and & = -2, then a unique value has not been
assigned to 4). When attempting to define a function, one must take care that
it is well defined lest one define only a relation.
The diagonal ((x, x) E X x X: x E X)is a function X -,X; it is called the
identity function on X , it is denoted by I,, and 1,: x-x for all x E X. If
X is a subset of Y, then the inclusion function i: X -, Y (often denoted by
i: X G Y) is defined by x r-,x for all x E X. The only difference between 1,
and i is that they have different targets, but if X is a proper subset of Y, this
is sufficient to guarantee that 1, # i.
Iff: X -, Y and g: Y -,Z are functions, then their composite g 0 f : X -+ Z
is the function defined by xwg(f(x)). If h: Z -, W is a function, then the
associativity formula h 0 (g 0 f ) = (h 0 g) 0 f holds, for both are functions
X -+ W with x r h(g(f(x))) for all x E X. 1f f : X 4 A and Y is a subset of X A relation Ion a set X is antisymmetric if x I y and y I x imply x = y, for
with inclusion i: Y G X, then the restriction f 1 Y: Y -,A is the composite all x, y E X. A relation < on a nonempty set X is called a partial order if it is
f 0 i ; f o r e a c h y ~Y,onehas(flY)(y)= f(y). reflexive, antisymmetric, and transitive. The best example of a partially or-
A function f : X -, Y is injective (or one-to-one) if distinct elements of X dered set is a family of subsets of a set, where < means c.
have distinct values; equivalently, if f (x) = f (x'), then x = x' (this is the A partial order on X is a simple order (or total order) if, for each x, y E X ,
converse of f being well defined: if x = x', then f(x) = f(xf)). A function either x I y or y I x.
f : X -+ Y is surjective (or onto) if each element of Y is a value; that is, for each If S is a nonempty subset of a partially ordered set X, then an upper bound
y E Y, there exists x E X with y = f (x). (If one did not insist, in the definition of S is an element x E X (not necessarily in S) with s I x for all s E S. Finally,
of equality of functions, that targets must be the same, then every function a maximal element in a partially ordered set X is an element m E X which is
would be surjective!) A function is a bgecfion (or one-to-one correspondence) smaller than nothing else: if x E X and m < x, then x = m. For example, if X
if it is both injective and surjective. A function f : X -+ Y is a bijection if and is the partially ordered set consisting of all the proper subsets of a set A under
only if it has an iaple.sczfun~$iopo: there is a fuaction g: Y + X with g o f = 1, inclusion, then a maximal element is the complement of a point. Thus, a
and f 0 g = 1, (g is usually denoted by f -I). One must have both composites partially ordered set can have many maximal elements. On the other hand, a
identity functions, for g 0 f = 1, implies only that f is injective and g is partially ordered set may have no maximal elements at all.'For example,
surjective. there are no maximal elements in the set of real numbers R regarded as a
partially ordered set under ordinary inequality (indeed, R is a simply ordered
set).

Zorn's Lemma. If X is a partially ordered set in which every simply ordered

subset has an upper bound, then X has a maximal element.

Remember that partially ordered sets are, by definition, nonempty.

Zorn's lemma is equivalent to a much more intuitive statement: the Axiom
of Choice, which states that the cartesian product of nonempty sets is itself
nonempty. (It is easy to prove (by induction) that a cartesian product of a
finite number: of nonempty sets is nonempty, and so the Axiom of Choice
need be invoked only when there are infinitely many factors.) We regard both
482 Appendix IV. Zorn's Lemma

of these statements as axioms of Mathematics, and we will not be ashamed to APPENDIX V

use either of them when convenient.
There is a third statement, equivalent to these, which is also useful. A Countability
partially ordered set X is well-ordered if every nonempty subset contains a
smallest element; that is, if S c X and S # 0, then there is so E S with so _< s
for all s E S. (Well-ordered sets must be simply ordered, for every two-element
subset has a smallest element.) The set of natural numbers N = ( n E 72:
n 2 0) is well-ordered, but the set 72 of all integers is not well-ordered.

Well-Ordering Principle. Given a nonempty set X, there exists a partial order

< on X which is a well-ordering.
Although Z is not well-ordered under the usual ordering, we can well-
order it: 0, 1, - 1,2, -2, ...-.Here is an example of a well-ordered set in
which an element has infinitely many predecessors: define X to be the follow-
ing subset of IW with the usual notion of I:

It is well known that certain "paradoxes" arise if one is not careful about the
foundations of Set Theory. We now sketch some features of the foundations
we accept. Its primitive undefined terms are class, element, and membership,
denoted by E: if X is a member of a class Y, one writes X E Y. A "set" is a
special kind of class, described below. Every usage of "set" in the preceding
appendices can be replaced by the word "class." In particular, one may speak
of cartesian products of classes, so that functions from one class to another
are defined; functions between classes may or may not be injective or
surjective.
Classes X and Y are called equQotent, denoted by 1x1= I YI, if there
exists a bijection f : X -+ X It is easy to see that equipotence is an equiva-
lence relation. Define a relation IX I s I Y I to mean that there is an injection
f : X -+ X It is easy to see that this relation is reflexive and transitive, and
the Cantor-Schroeder-Bernstein theorem shows that it is antisymmetric:
IX I _< I Y / and I Y I s I XI imply 1x1= I Y I; thus, s is a partial order.
The foundations allow one to define the cardinal number of a set; some
classes have a cardinal number and they are called sets; some classes do not
have a cardinal (they are too big) and they are called proper classes. For
example, the class of all abelian groups is a proper class. The notion of
functor in Chapter 10 thus involves a function defined on a proper class.
A set X isfinite if it is empty or if there is a positive integer n and a bijection
f : {1,2, .. ., n ) -+ X (in this case, we write IX I = n); otherwise, X is infinite.
There is an elementary, but useful, result here, sometimes called the pigeon-
hole principle.

Theorem. If X and Y are finite sets with 1x1= I YI, then a function f : X -+ Y
is injective if and only if it is surjective; in either case, therefore, f is a bijection.
484 Appendix V. Countability

A set X is called countable if it is finite or if there is a bijection f : N -t X , APPENDIX VI

where N = (0, 1,2, ...) is the set of natural numbers; otherwise X is uncount-
able. The sets Zand Q are countable, but the set R is uncountable.
If X is any infinite set, then the set Y of all its subsets is an uncountable set; Commutative Rings
moreover, if Z is any set with at least two elements, then the set of all func-
tions X -,Z is uncountable. In particular, the family of all sequences in Z is
uncountable.
Here are some facts about countable sets.
1. If X is any countable set, then the family Y of all its finite subsets is a
countable set.
2. Every subset of a countable set is countable.
3. If X is a countable set and iff: X -, Y is a surjection, then Y is'countable.
4. If X and Y are both countable, then so is X x Y.
5. If A, and A, are countable subsets of a set X, then A, v A, is countable.
More generally, if (A,: n 2 0) is a countable family of countable subsets of
X, then U, A, is countable.

It is assumed that the reader has seen an introduction to commutative rings

before beginning this book, and so this appendix is intended only as a re-
minder of perhaps half-forgot ten ideas.
Let R be a commutative ring. In contrast to some authors, we assume that
R must contain an element 1 # 0 which, under multiplication, behaves like
the number 1: for all r E R, l r = r. If R' is a second commutative ring, then a
(ring) homomorphism is a function f : R 4 R' such that f(1) = 1, f(r s) = +
f (r) + f (s), and f (rs) = f (r)f (s). An ideal I in R is an additive subgroup of R

-
(0 E I and a, b E I implies a - b E I) such that a E I and r E R imply ra E I. If
f : R R' is a homomorphism, then its kernel = { r E R: f (r) = 0) is an ideal.
Here is another important example of an ideal: if a,, . . . , a, E X, then the set
of all their linear combinatioazs is called the ideal genrerated by a,, . . .,a,; it is
denoted by (a,, ...,a,):
(a,, ..., a,) = (r,a, + ... +man: r , ~ R f o all
r i).
Hn the special case n = 1, the ideal generated by a E R, namely,
(a) = (ra: r E R),

is called the principal ideal generated by a. If R = L [XI, the ring of all polyno-
mials with coefficientsin L, one can show that the ideal I = (x, 2), consisting
of all polynomials with coefficients in Z having even constant term, is not a
principal ideal.
The following result, which merely states that long division is possible, can
be proved by the reader. Recall that the zero polynomial is the polynomial all
of whose coefficients are 0; if f(x) is not the zero polynomial, then we write
+
f (4 0-
 Appendix YI. Commutative Rings
Division Algorithm.
(i) If a, b E Z with a + 0, then there exist unique q, r E Z with 0 Ir < la1 and
(ii) If k is a field and a(x), b(x) E k[x] are polynomials with a(x) # 0, then
there exist unique q(x), r(x) E k[x] with either r(x) = 0 or degree r(x) <
degree a(x) and
+
b(x) = q(x)a(x) r(x).
One calls q (or q(x)) the quotient and r (or r(x)) the remainder.
A domain R is a principal ideal domain (or PID) if every ideal in R is a
principal ideal. Our example above shows that Z[x] is not a prjncipal ideal
domain.
Theorem W.1.
(i) Z is a principal ideal domain.
(ii) If k is a field, then k[x] is a principal ideal domain.
Proof. (i) Let I be an ideal in Z. If I consists of 0 alone, then I is principal,
generated by 0. If I # 0, then it contains nonzero elements; indeed, it contains
positive elements (if a E I, then -a = (- 1)a E I also). If a is the smallest
positive integer in I, then (a) c I. For the reverse inclusion, let b E I. By the
+
division algorithm, there are q, r E Z with 0 5 r < a and b = qa r. Hence,
r = b - qa E I. If r # 0, then we contradict a being the smallest element in I.
Therefore, r = 0 and b = qa E (a); hence I = (a).
(ii) The proof for k[x] is virtually the same as for Z. If I is a nonzero ideal
in k[x], then a generator of I is any polynomial a(x) E I whose degree is
smallest among degrees of polynomials in I.
If I is an ideal in R and we forget the multiplication in R, then f is a
subgroup of the abelian additive group R. The quotie~t.ring R/I is the quo-
tient group R/I (whose elements are all cosets r + I) made into a commuta-
tive ring by defining multiplication:
(r + I)(s + I) = rs + I.
Let us show that this definition does not depend on the choice of coset
representative. By Lernma 2.8 in the text, r + I = r' + I if and only if
r - r ' ~ I . I f s + I=s1+I,then L Proof. We may assume that d # 0 (otherwise ai = 0 for all i). If d and d' are
rs - r's' = rs - r's + r's - r's'
= (r - rr)s + r'(s - s').
Since r - r', s - s' E I, we have rs - r's' E I, and so rs + I = r's' + I.
The natural map v: R -+ R/I is the function defined by v(r) = r + I; the
definition of multiplication in the quotient ring shows that v is a (surjective)
Appendix VI. Commutative Rings 487
ring homomorphism with kernel I. If f : R --+ R' is a ring homomorphism,
then the first isomorphism theorem (Theorem 2.23) applies to the additive
groups if one forgets multiplication: if I is the kernel of 5 then I is a subgroup
of R, im f is a subgroup of R', and there is an isomorphism cp: R/I -+ im f
given by r + I Hf (r). If one now remembers the multiplication, then it is easy
to check that I is an ideal in R, im f is a subring of R', and the group
isomorphism cp is also a ring isomorphism. The analogue of the correspon-
I
dence theorem (Theorem 2.27) is valid: If I is an idea in R, then there is a
bijection between all intermediate ideas J with I c J - R and all ideals of
R/I.
If a, b E R, then a divides b in R (or b is a mult@le of a), denoted by a /b, if
there is r E R with ar = b. Note that a divides b if and only if b E (a). If
a,, ...,a, E R, then a common divisor is an element c E R which divides each
a,; a greatest common divisor (gcd), denoted by d = (a,, ... ,a,), is a common
divisor which is divisible by every common divisor.
Theorem V1.2. If R is a PID, then every set of elements a,, ..., a, E R has a
gcd d; moreover, d is a linear combination of a,, .. .,a,.
Proof. The set of all linear combinations of a,, ...,a, is an ideal I in R. Since
R is a PID, I is a principal ideal; let d be a generator. As any element of I, d
is a linear combination of a,, . ..,a,. It follows that any common divisor c
of the a, also divides d. But d itself is a common divisor, for each ai lies in
(4.
We have just shown that (a,, ..., a,) = (d), where d is a gcd of a,, . . . ,a,.
An element u E R is a mi$if u ]1; that is, there exists v E R with uu = 1. Two
elements a, b E R are associafes if there is a unit u with a = bu. An element in
R is irreducible if it is not a unit and its only factors are units and associates.
In Z, the irreducibles have the form +p, where p is a prime; if k is a field,
a polynomial p(x) E k[x] is irreducible if it is not constant and there
is no factorization p(x) = f(x)g(x) with degree f(x) < degree p(x) and
degree g(x) < degree-p(x).
Theorem W.3. If R is a PID and a,, ...,a, E R, then any two gcd's of a,, ...,
a, are associates.
gcd's, then each divides the other: there are u, v E R with d = d'u and d' = dv.
Hence, d = d'u = d v ~and
, the cancellation law gives 1 = uu. Therefore, u is a
unit and d and d' are associates. Ill
If R = Z, then the only units are 1 and - 1, so that two gcd's of a,, .. .,a,
differ only in sign. If R = k[x], where k is a field, then the only units are
nonzero constants; thus, only one gcd d(x) off, (x),...,f,(x) is monic; that is,
488 Appendix VI. Commutative Rings Appendix VI. Commutative Rings 489

it has leading coefficient 1 (i.e., the coefficient of the highest power of x is 1).
Thus, if not all ai(x) = 0, any gcd of a, (x), ...,a,(x) in k[x] is an associate of
a monic polynomial.
In light of Theorem V1.3, we may speak of the gcd of elements in a PID,
and we change the definition, in the special cases of Z and of k[x], so that the
gcd is either positive or monic. The gcd of two elements a and b is denoted
by (a, b); we say that a and b are relatively prime if (a, b) = 1.
The euclidean algorithm shows how to compute gcd's in Zand in k[x], but
we shall not need this here.
Two integers a and b are relatively prime if and only if there are integers s
and t with sa + tb = 1. In other words, a and b are relatively prime if and
only if [a b] is the first row of a matrix having determinant 1 (let the second
row be [- t s]). This result can be generalized. Call a matrix A unimodular if
det A = 1.
Theorem VH.5. Let R be a PID and let p E R be irreducible. If I = (p) is the
principal ideal generated by p, then R/I is a field.
Proof. Iff + I E R/I is not the zero element, then f 4 I = (p); that is, p does
not divide f. Since p is irreducible, its only divisors are units and associates.
Therefore, d = gcd(p, f ) is either a unit or an associate of p. But if d were an
associate of p, then p would divide f, a contradiction. Hence, d = 1, and so
there are elements a, b E R with 1 = ap + bf. Thus, bf - 1 E (p) = I, so that
(b + I)(f + I) = 1 + I; that is, f + I is a unit in R/I, and so R/I is a field.
Corollary W.6.
(i) If p is a prime, then Z,= ZMp) is a field.
(ii) If k is a field and p(x) E k[x] is an irreducible polynomial, then k;Cx]/(p(x))
is a field.

Theorem W.4. A set a,, ...,a, of integers is relatively prime if and only if there When we say that an element a is a product of irreducibles we allow the
is a unimodular n x n matrix A with entries in Z whose first row is [a, ...a,]. possibility that a itself is irreducible (there is only one factor).
A domain R is a uniquefactorization domain (or UFD) if:
Proof. We prove the theorem by induction on n 2 2, the base step being our
observation above. For the inductive step, let d be the gcd of a,, ..., a ,-,, (i) every r E R that is.neither 0 nor a unit is a product of irreducibles; and
(ii) if p, . ..p, = q, .. .q,, where the pi and qj are irreducible, then there is a
and let ai = dbi for i < n - 1. The integers b,, ...,b,-, are relatively prime
(there is no common divisor c > I), and so there exists an (n - 1) x (n - 1) bijection between the sets of factors (so m = n) such that corresponding
unimodular matrix B with entries in Z whose first row is [b,, ...,b,-,I. Now factors are associates.
d and a, are relatively prime (lest there be a common divisor of a,, ...,a,), so We are going to prove that every principal ideal domain R is a unique
there are integers s and t with sd + tan = 1. If C denotes the lower n - 2 factorization domain, and our first task is to show that if r E R is neither 0.
rows of B, define A to be the n x n matrix nor a unit, then it is a product of irreducibles.
db, ... db,-,
Lemma VI.7. If R is a PID, then there is no infitnite strictly increasing sequence
of ideals
-tbl .. . t b s 1, < iT2 < " ' < I, < I,+1< ... .
note that the first row of A is [a, ... a,]. Expanding down the last column,
Proof. If such a sequence exists, then it is easy to check that J = U, I, is an
det A = (- [
lY+'andet -:b] + (- 1)"s det +
d][: ideal. Since R is a PID, there is a E J with J = (a). Now a got into J by being
in some I,. Hence

o w det [d:] = d det B =d and

T r l
det [-:b] = - t det [t]
=
a contradiction. s$B
J = (a) l I, < In+l< J,

(-t)(- I)"+'(- I)"-' = t (because

Li1 is obtained from B by successively
interchanging the top row b with each of the n - 2 rows below it). Hence,
Lemma W.8. If R is a PID and a E R is neither 0 nor a unit, then a is a product
of irreducibles.
+ +
det A = (- l)2n+2ta, sd = tan sd = 1, so that A is unimodular.
Proof. If r E R has a factorization r = bc, where neither b nor c is a unit, then
For the converse, let A be a unimodular n x n matrix with entries in Z
whose first row is [a, ... a,]. Evaluating the determinant by expanding we say that b is a proper factor of a. It is easy to check, using the hypothesis
across the top row gives 1 = det A as a Z-linear combination of a,, ...,a,, that R is a domain, that if b is a proper factor of a, then (a) i(b).
and this shows that a,, ...,a, are relatively prime. IM Call an element a E R good if it is a product of irreducibles; otherwise, a is
 490 Appendix VI. Commutative Rings
bad. If a and b are good, then so is their product ab. Thus, if a is bad, then it
factors (for irreducibles are good), and at least one of its proper factors is bad.
Suppose there is a bad element a = a, which is neither 0 nor a unit. Assume
inductively that there exists a,, a,, . ..,a, such that each a,,, is a proper bad
factor of ai. Since a, is bad, it has a proper bad factor a,,,. By induction,
there exist such a, for all n 2 0. There is thus an infinite strictly increasing
sequence of ideals (a,) < ( a , ) < -. < (a,) < (u,~,,)< .- - , contradicting the
previous lemma. Therefore, every a E R that is neither 0 nor a unit is good;
that is, a is a product of irreducibles. iB
Theorem W.9 (Eucllid's Lemma). Let R be a PID and let p E R be irreducible.
If a, b E R and plab, then pla or p(b.
Proof. If pla, we are done. Otherwise, p does not divide a, and so the
gcd d = (p, a) = 1 (as we saw in the proof of Theorem VI.5). By Theorem
VI.2, there are elements s, t E R with sp + ta = 1. Therefore, b = spb + t(ab).
Since plab, it follows that pl b: i
4 Theorem VI.13. An ideal I in a commutative ring R is maximal if and only
Theorem V1.10 (Fundamental Theorem of Arithmetic). Every principal ideal
domain R is a unique factorization domain.
Proof. By Lemma VI.8, every a E R that is neither 0 nor a unit is a product of
irreducibles. We have only to verify the uniqueness of such a factorization.
If p,. ..pm= q, ...q,, where the pi and the qj are irreducible, then
p, 1 q, ...q,. By Euclid's lemma, p, divides some qj. Since qj is irreducible, p,
and qj are associates: there is a unit u E R with qj = up,. As R is a domain,
we may cancel p, to obtain
and the proof is completed by an induction on ma:;(m, n). @ Theorem VI.14. If R is a PID, then every nonzero prime ideal I is maximal.
A standard proof of Corollary V1.6 uses the notion of prime ideal, which
makes Euclid's lemma into a definition. An ideal I in a commutative ring R
is called a prime ideal if I # R and ab E I implies a E I or b G I . Note that
the ideal 0 is prime if and only if R is a domain.
Theorem VP.11. A nonzero ideal I in a PID R is prime if and only if I = (p),
where p is irreducible.
n
Proof. Since R is a PID, there is an element d with I = jd). Assume that I is
prime. If d is not irreducible, then d = ab, where neither a nor b is a unit. By
hypothesis, either a E I or b s I. But if, say, a E I, then a = rd for some r E R,
and hence d = ab = rdb. Therefore 1 = rb, contradicting b not being a unit.
Conversely, assume that p is irreducible and that ab E (p). Thus, plab, so
that Euclid's lemma gives pl a or p 1 b; that is, a E (p) or b E (p). BFll
49 1
Appendix VI. Commutative Rings
Theorem V1.12. An ideal I in a commutative ring R is prime if and only if R/I
is a domain.
+
Proof. Recall that the zero element of R/I is 0 I = I. Assume that I is
+ +
prime.IfI = ( a I)(b + I ) = ab I,thenab E I, h e n c e a ~ I o r bE I ; thatis,
a + I = I or b + I = I. Hence, R/I is a domain. Conversely, assume that R/I
+ + +
is a domain and that ab E I. Thus, I = ab I = (a I)(b I). By hypothe-
+ +
sis, one of the factors must be zero; that is, a I = I or b I = I. Hence,
a E I or b E I, so that I is prime.
An ideal I in a commutative ring R is a maximal ideal if I # R and there is
no ideal J with I $ J $ R. Note that the ideal 0 is maximal if and only if R is
a field, for if a E R is not zero, then the ideal (a) must be all of R. Hence 1 E (a),
so there is r E R with 1 = ra; that is, a is a unit.
if
RII is a field.
Proof. If I is maximal, then the correspondence theorem for rings shows that
R/I has no proper ideals. Hence the zero ideal is maximal and.R/I is a field.
Conversely, if there is an ideal J with I $ J R, then the correspondence
theorem shows that J/I is a proper nonzero ideal in RII, contradicting the
hypothesis tht RII is a field (the ideal 0 is maximal).
Since every field is a domain, if follows that every maximal ideal in a
commutative ring is a prime ideal. The converse is not always true; for exam-
ple, the ideal I = (x) in Z[x] is prime (for Z[x]/Iz Z is a domain), but it is
not maximal because Zis not a field (or, because (x) $ (x, 2) Z[x]).
Proof. Assume that J is an ideal with I $ J. Since R is a PID, there are
elements a, b E R with I = (a) and J = (b). Now I $ J gives a E J = (b), so
there is r E R with a = rb. But I is prime, so that b E I or r E I. If b E I, then
J = (b) c I, contradicting I 2 J. Therefore, r E I, so there is s E R with r = sa.
Hence, a = rb = sab; as R is a domain, 1 = sb E (b) = J. Therefore, J = R
and I is maximal.
One can now give a second proof of Theorem VI.5.If R is a PID and p E R
is irreducible, then I = (p) is a nonzero prime ideal, by Theorem VI.11. By
Theorem VI.14, I is a maximal ideal, and so Theorem VI.13 shows that R/I
is a field. -
Lemma V1.15. Let k be a field and let p(x) E k[x] be irreducible. Then there is
a field K containing a subfield isomorphic to k and a root of p(x).
492 Appendix VI. Commutative Rings
Proof. Let I = (p(x)) and let K = k[x]/I; since p(x) is irreducible, Theorem
VI.5 shows that K is a field. It is easy to check that the family of all cosets of
the form a + I, where a E k, is a subfield of K isomorphic to k. Let p(x) =
+
Eaixi. We claim that the element x I E K is a root of p(x).
p(x + I) = 1 ai(x + I)'
= C ai(xi + I)
= C (nixi + I)
= ( C aixi) + I
= p(x) + I = I.
The result follows, for I = 0 + I is the zero element of K. N a field F containing Z, (so that F has characteristic p) and all the roots off (x).
Theorem VH.16. If k is a field and f (x) E k[x], then there is a field F containisg
k over which f(x) is a product of linear factors; that is, F contains all the roots
o f f(4.
Proof. The proof is by induction on n, the degree of f(x). If n = 1, then f(x) is
linear and we may set F = k. If n > 1, there is a factorization f (x) = p(x)g(x)
in k [ x ] , where p(x) is irreducible (perhaps g(x) = 1). By Lemma VI.15, there
is a field K containing k and a root /3 of p(x). There is thus a factorization
p(x) = (x - P)h(x) in K [x]. Since degree h(x)g(x) < n, the inductive hypo-
thesis gives a field F containing K over which h(x)g(x), hence f(x) =
(X- P)h(x)g(x), is a product of linear factors.
+
If k is a field and f(x) = anxa an-,xn-' + .-.+ a, E k[x], then its deriva-
the is
f '(x) = nanxn-I + (n - l ) a , - , ~ ~+- ~. + a,.
It is easy to check that the usual formulas of Calculus hold for derivatives
of sums and products of polynomials over arbitrary fields: (f(x) + g(x))' =
f '(x) + g'(x); ( f (x)g(x))' = f (x)gi(x) + f '(x)g(x).
Lemma VH.17. Let k be a field, let f(x) E k[x], and let F be a field containing
k which contains all the roots of f(x). Then f(x) has no repeated roots if and
only iff (x) and f '(x) are relatively prime.
Proof. If f (x) has a repeated root, then f (x) = (x - B)2g(x)in F [x]. ~ e n ~ c
+
fr(x) = 2(x - P)g(x) (x - /?)2g'(~), SO that x - p is a common divisor of
f(x) and fi(x) and they are not relatively prime. Conversely, assume
that x - /? is a common divisor of f(x) and f '(x): say, f(x) = (x - P)g(x)
and f'(x) = (x - P)h(x). By the product formula for derivatives, f'(x) =
+ +
(x - P)gi(x) g(x), so that (x - P)gi(x) g(x) = (x - P)h(x). Therefore,
x - p divides g(x), (x - j?)2 divides f (x), and f (x) has a repeated root. m
Appendix VI. Commutative Rings 493
Lemma '671.18. If k is a field of characteristic p > 0, then for all a, b E k and
for all n 2 1,
+
(a + b)Pn= aPn bPn.
+
Proof. The proof is by induction on n. If n = 1, expand (a b)Pby the bino-
mial theorem. Since p is prime, it divides all the middle binomial coefficients,
and hence each of them is 0 mod p. The inductive step is easy.
Theorem VI.19 (Galois). For every prime p and every n 2 1, there exists a field
having exactly pn elements.
Proof. Let q = pn and let f (x) = xq - x E Zp[x]. By Theorem VI.16, there is
Define K to be the set of all the roots off (x) in F. Since f (x) has degree q, it
follows that J K (_< q, with equality if f(x) has no repeated roots. Now f '(x) =
qxq-I - 1 = - 1, because F has characteristic p and q = pn. Therefore, f(x)
and ff(x) are relatively prime, and Lemma VI.17 shows that f(x) has no
repeated roots.
We now show that K is a subfield of F. Let a and b be roots off (x), so that
aq = a and bq = b. Lemma V1.18 gives (a - b)q = aq - bb"= a - 6, so that
a - b E K; moreover, (ab)4 = aqbq= ab, so that ab E K. Since 1 E K, it
follows that K is a subring of F. Finally, if a # 0, then aq = a implies that
aq-I = 1, so that a-I = aq-2 E K (because K is a subring). Therefore, K is a
field. !B
It is curious that the uniqueness of the finite fields was not established for
more than 60 years after their discovery.
Theorem VVII.20 (E.H. PIoore, 1893). Any ttvo fields having exactly p9' elements
are isomorphic.
PYOO$Let K be a field with exactly q = pn elements. Since the multiplicative
group K x of nonzero elements of K has order q - 1, Lagrange's theorem
gives aq-I = 1 for all nonzero a E K. Therefore, every element of K is a root
of f(x) = xq - x, so that K is a splitting field of f(x). The result follows
from the general fact that any two splitting fields of a given polynomial are
isomorphic (the reader may prove this by induction on the degree of f(x),
using Lemma 5.5 of the text). $I
The following proof is the polynomial version of the fact that every con-
gruence class [a] E Zmis equal to [r], where r is the remainder after dividing
a by m; moreover, any two "remainders," that is, any distinct r and r' between
0 and r - 1 are not congruent.
Theorem Vl.21. Let k be a field and let p(x) E k[x] be an irreducible polyno-
494 Appendix VI. Commutative Rings

rnial of degree n. If k is a subfield of a field E and if a E E is a root of p(x), then

[k(a) :k] = n. Indeed, every element in k(a) has a unique expression of the form

(*I b, + bla + ... + bn_,a"-', Bibliography

where bi E k for all i.

Proof. The map 9:k[x] -+ E, defined by f(x)t-. f(a), is a ring homomor-

phism with im 9 = k(a) and ker q = I # 0 (for p(x) E I). By Theorem VI.11,
I = (q(x)), where q(x) is irreducible. Hence q(x)lp(x), so that q(x) = cp(x) for
some nonzero c E k; that is, I = (p(x)). We claim that X = (1, a, ...,an-') is
a basis of k(a) viewed as a vector space over k. If so, then [k(a) : k] = n, as
desired. If there are b,, not all 0, with
g(x) = 1.':;
x:~;
biai = 0, then a is a root of
bjxi, a nonzero polynomial of degree smaller than n. But g(x) E
ker 9 = (p(x)), so that p(x)lg(x) and the degree of g(x) is at least n, a contra-
diction. Thus, X is linearly independent. To see that X spans k(a), we show,
by induction on m 2 n, that a mE W = (1, a, .. ., an-'), the subspace over k
spanned by X. If p(x) = zy=,cixi, then 0 = p(a) = ~y,,,ciai, so that a n =
-1:;; ciai E W. For the inductive step, assume that a m =
di E k. Then am" = aam = a z : ~ ; diai = X~Z: +
diaifl dnelan E CV.
zy~;
diai, where
Albert, A.A., Introduction to Algebraic Theories, University of Chicago Press, 1941.
Alperin, J., Local Representation Theory, Cambridge University Press, 1986.
Artin, E., Galois Theory, Notre Dame, 1955.
,Geometric Algebra, Interscience, 1957.
Aschbacher, M., Finite Group Theory, Cambridge University Press, 1986.
,Sporadic Groups, Cambridge University Press, 1994,
Babakhanian, A., Cohomological Methods in Group Theory, Dekker, 1972.
Baumslag, G., Reviews on Infinite Groups, Amer. Math. Soc., 1974.
Benson, D., Representations and Cohomology, 2 volumes, Cambridge University
Press, 1991.
Biggs, N.E., Discrete Mathematics, Oxford University Press, 1989.
Biggs, N.L., and White, A.T., Permutation Groups and Combinatorial Structures, Cam-
bridge University Press, 1979.
Birkhoff, G., and Mac Lane, S., A Survey of Modern Algebra, MacmiIlan, 1977.
Borel, A., Linear Algebraic Groups, Benjamin, 1969.
Brown, K., Cohomology of Groups, Springer-Verlag, 1982.
Burnside, W., The Theory of Groups of Finite Order, Cambridge University Press,
1911.
Cameron, P.J., Parallelisms of Complete Designs, Cambridge University Press, 1976.
Carmichael, R., An Introduction to the Theory of Groups, Ginn, 1937.
Cartan, H., and Eilenberg, S., Homological Algebra, Princeton University Press, 1956.
Carter, R.W., Simple Groups of Lie Type, Wiley, 1972.
, Finite Groups of Lie Type: Conjugacy Classes and Complex Characters,
Wiley, 1985.
Chandler, B., and Magnus, W., The History of Combinatorial Group Theory: A Case
Study in the History of Ideas, Springer-Verlag, 1982.
Cohn, P.M., Universal Algebra, Harper & Row, 1965.
,Algebra, 2nd edn., vol. I, Wiley, 1982.
Conway, J.H., Curtis, R.T., Norton, S.P., Parker, R.A.; and Wilson, R.A., Atlas of
Finite Groups, Oxford University Press, 1985.
Conway, J.H., and SIoane, N.J.A., Sphere Packings, Lattices, and Groups, Springer-
Verlag, 1993.
 Bibliography
Coxeter, H.S.M., Projective Geometi.~,Springer-Verlag, 1987.
Coxeter, H.S.M., and Moser, W.O., Generators and Relations for Discrete Groups,
Springer-Yerlag, 1965.
Curtis, C., and Reiner, I., Representation Theory of Finite Groups and Associative
Algebras, Wiley, 1962.
Davis, M., Comwutabilitv and Unsolvabilitv. McGraw-Hill. 1958.
~ i c k s w.;
; anda~unwobdy, M.J., Goup; Acting on ~raphs,Cambridge University
Press, 1989.
Dickson, L.E., Linear Groups, Leipzig, 1901; Dover, 1958.
Dieudonne, J., Sur Ees Groupes Classiques, Hermann, 1958.
Dixon, J.D., Problems in Group Theory, Blaisdell, 1967.
Dixon, J.D., du Sautoy, M.P.F., Mann, A., and Segal, D., Analytic Pro-p-Groups,
Cambridge University Press, 1991.
Doerk, K., and Hawkes, T., Finite Soluble Groups, de Gruyter, 1992.
Dornhoff, L., Group Representation Theory, 2 volumes, Marcel Dekker, -1971.
Epstein, D.B.A., Cannon, J.W., Holt, D.F., Levy, S.V.F., Paterson, M.S., and Thurston,
W.P., Word Processing in Groups, Jones and Bartlett, 1992.
Evens, L., The Cohomology of Groups, Oxford University Press, 1991.
Feit, W., Characters of Finite Groups, Benjamin, 1967.
, The Representation Theory of Finite Groups, North-Holland, 1982.
Fuchs, E., Infinite Abelian Groups, I and TI, Academic Press, 1970 and 1973.
Gersten, S.M., editor, Essays in Group Theory, Springer-Verlag, 1987.
Gorenstein, D., Finite Groups, Harper & Row, 1968.
, Reviews on Infinite Groups, Amer. Math. Soc., 1974.
, Finite Simple Groups, Plenum, 1982.
, The Classification of Finite Simple Groups; Volume 1: Groups of Non-
characteristic 2 type, Plenum, 1983.
Griffith, P.A., Infinite Abelian Group Theory, University of Chicago Press, 1970.
Gruenberg, K.W., Cohomological Topics in Group Theory, Springer SLN 143, 1970.
Gruenberg, K.W., and Weir, A.J., Linear Geometry, Springer-Verlag, 1977.
Hall, M., Jr., The Theory of Groups, Macmillan, 1959.
Hall, P., Collected Works, Oxford University Press, 1987.
Higgins, P.J., Notes on Categories and Groupoids, van Nostrand-Reinhold, 1971.
Higman, G., Existentially Closed Groups, Oxford University Press, 1988.
Humphreys, J.E., Reflection Groups and Coxeter Groups, Cambridge University Press,
1990.
Huppert, B., Endliche Gruppen, Springer-Verlag, 1967.
Huppert, B., and Blacltburn, N., Finite Groups, II and 111, Springer-Verlag, 1981 and
1982.
Isaacs, I.M., Character Theory of Finite Groups, Academic Press, 1976.
Jacobson, M., Basic Algebra, I and 11, Freeman, 1974 and 1979.
James, G., and Liebeck, M., Representations and Characters of Groups, Cambridge,
1993.
Johnson, D.L., Topics in the Theory of Group Presentations, Cambridge University
Press, 1980.
Jordan, C., Trait6 des Substitutions et des ~ ~ u a t i o nAlgkbriques,
s Gauthier-Villars,
1870.
Kaplansky, I., Infinite Abelian Groups, University of Michigan Press, 1969.
, Fields and Rings, University of Chicago Press, 1972.
Karpilovsky, G., The Schur Multiplier, Oxford University Press, 1987.
Kegel, O.M., and Wehrfritz, B.A.F., Locally Finite Groups, North-Holland, 1973.
Khukhro, E.I., Nilpotent Groups and their Automorphisms, de Gruyter, 1993.
Kurosh, A.G., The Theory of Groups, 2 volumes, Chelsea, 1956.
Bibliography 497
Lyndon, R., and Schupp, P., Combinatorial Grottp Theory, Springer-Verlag, 1977.
Mac Lane, S., Homology, Springer-Verlag, 1963.
Magnus, W., Karrass, A., and Solitar, D., Combinatorial Group Theory, Wiley, 1966.
Massey, W.S., Algebraic Topology: An Introduction, Harcourt, Brace, and World,
1967.
Miller, C.F., 111, On Group-Theoretic Decision Problems and Their Classification,
Princeton University Press, 1971.
Miller, G.A., Blichtfeld, H.F., and Dickson, L.E., Theory and Applications of Finite
Groups, Wiley, 1916.
Milnor, J., Introduction to Algebraic K-Theory, Princeton, 1971.
Neumann, H., Varieties of Groups, Springer-Verlag, 1967.
Passman, D.S., Permutation Groups, Benjamin, 1968.
Powell, M.B., and Higrnan, G. (eds), Finite Simple Groups, Academic Press, 1971.
Puttaswamaiah, B.M., and Dixon, J.D., Modular Representations of Finite Groups,
Academic Press, 1977.
Robinson, D.J.S., Finiteness Conditions and Generalized Soluble Groups, 2 volumes,
Springer-Verlag, 1972.
,A Course in the Theory of Groups, Springer-Verlag, 1982.
Rotman, J.J., An Introduction to Homological Algebra, Academic Press, 1979.
,An Introduction to Algebraic Topology, Springer-Verlag, 1988.
, Galois Theory, Springer-Verlag, 1990.
Schenkman, E., Group Theory, van Nostrand, 1965.
Scott, W.R., Group Theory, Prentice-Hall, 1964.
Serre, J.-P., Linear Representations of Finite Groups, Springer-Verlag, 1977.
, Trees, Springer-Verlag, 1980.
Specht, W., Gruppentheorie, Springer-Verlag, 1956.
Speiser, A., Die Theorie der Gruppen von Endlicher Ordnung, Springer-Verlag, 1927.
Stammbach, U., Homology in Group Theory, Springer SLW 359,1980.
Suzuki, M., Group Theory, I and 11, Springer-Verlag, 1982 and 1986.
van der Waerden, B.L., Modern Algebra, Ungar, 1948.
Vaughan-Lee, M., The Restricted Burnside Problem, Second Edition, Oxford Univer-
sity Press, 1993.
Wehrfritz, B.A.F., Infinite Linear Groups, Springer-Verlag, 1973.
Weiss, E., Cohomology of Groups, Academic Press, 1969.
IVeyl, H., Symmetry, Princeton University Press, 1952.
Wielandt, H., Finite Permutation Groups, Academic Press, 1964.
Wussing, H., Die Genesis des abstrakten Grzippenbegri$es, VEB Deutscher Verlag der
Wissenschaften, 1969 (English translation, Genesis of the Abstract Group Concept,
MIT Press, 1984).
Zassenhaus, H., The Theory of Groups, Chelsea, 1956.
Zieschang, H., Vogt, E., and Coldewey, H.-D., Surfaces and Planar Discontinuous
Groups, Springer SLN 835, 1980.
7
 Groups and Subgroups

(A, ( A ) , a) affine space

Pn(K) projective space over K of dimension n
Notation
Groups and Subgroups
A<B A is a subgroup of B
A<B A is a proper subgroup of B
<x> subgroup generated by a subset X
AvB subgroup generated by subgroups A and B
AB (abla E A and b E B)
AaB A is a normal subgroup of B
G/H the family of all left cosets of H in G (it is a group when
H a G)
H char G H is a characteristic subgroup of G
AxB direct product of groups
n
A@B
Ai direct product of groups
direct sum of abelian groups
Set Theory and Algebra direct sum of abelian groups
C Ai
A+B free product of groups
A c B o r B 3 A A is a subset of B A*@B amalgam
0 empty set A >a B or A >a, B semidirect product
AxB cartesian product A B or A 1, B wreath product or regular wreath product
f:A-+B f is a function from A to B aG conjugacy class of element a E G
1, the identity function X -+ X [A, BI subgroup generated by all commutators [a, b] = aba-I b-'
i:A=B i is the inclusion map from the subset A c B G' commutator subgroup = [G,GI
f : a++f(a) f is a function whose value on the element a in its domain center of 6
is f (a) GA-4 centralizer of element a in G
the restriction off to A; that is, i f f : B -+ k: and A c B, C G ( ~ centralizer of subgroup M < 6;
then f l A = f o i , w h e r e i : A c t B NG(H) normalizer of subgroup If < G
the number of elements in a set X Aut (G) automorphism group of group G
Inn(G) all inner automorphisms of G
(7) binomial coefficient m!/i!(m- i)!
the group of units in the ring R
O(x)or Gx
Gx
orbit of element x in a G-set
stabilizer of element x in a G-set
if K is a field, K x = K - (0); the multiplicative group of Hol(G) holomorph of G
K = U(K) Hom(G, B) all homomorphisms from G to B
if G is a group, G# = G - (1) G* character group = Hom(G, C ") (EHom(G, Q/Z) when G
7
equivalence class containing an element x; in particular, is finite)
the congruence class of an integer x E B, U(R) group of units in a ring R
matrix whose entry in row i and column j is aij @(GI Frattini subgroup
(Kroneeker deha) if i # j, then dij = 0;if i =j, then dij = 1. dG maximal divisible subgroup of abelian group G
identity matrix [S,,.] tG torsion subgroup of abelian group G
a and p are homotopic paths ( x 1 A) presentation with generators X and relations A
500 Notation Names of Groups

(G; X' j A') given a presentation of a group G, this is a presentation symplectic group

Assume that K is an abelian group and that 8: Q

(when 8 is trivial, one omits it from the notation).
-
with additional generators X' and additional relations &
K is a homomorphism
nonabelian group of order 12 = Z, x iZ,
circle group = multiplicative group {z E @ 1 lzl = 1)
unitriangular group
4-group
B2(Q,K, 8) coboundaries integers
Z2(e, K, 8) factor sets (cocycles) integers mod n
H2(Q,K7 8) = Z2(Q,K, @)/B2(Q,K,8) all scalar n x n matrices
M(Q) = H2(Q, e x )= Schur multiplier all scalar n x n matrices of determinant 1
Ext(Q, K) = HZ(Q,K) when both Q and K are abelian and 8 is trivial
Der(Q, K, 8) all derivations
PDer(Q, K, 8) all principal derivations
w(Q,K, 0) = Der(Q, K, WP?er(Q, K, 8)

Names of Groups
alternating group
infinite alternating group
affine group
all affine isomorphisms of vector space V
all automorphisms of Steiner system (X, 9 )
binary tetrahedral group
braid group
complex numbers
multiplicative group of nonzero complex numbers
dihedral group of order 2n
infinite dihedral group
finite field having q elements
general linear group of all nonsingular n x n matrices
all nonsingular semilinear transformations
linear fractional transformations over K
Mathieu group i = 10,11,12,22,23,24
orthogonal group
projective linear group
modular group
collineation group
quaternions of order 8
generalized quaternions of order 2"
rational numbers
real numbers
symmetric group on n letters
symmetric group on a set X
all n x n matrices of determinant 1
Index

Aanderaa, S. 450 ascending central series 113

Abbati, P. 1, 57, 58 ascending chain condition 145
Abel, N.H. 1,97 associated graph to directed graph 356
abelian group 13 associated semigroup to Turing machine
ACC 145 426
action 55 associated vector space to affine space
adding a handle 409 265
Adian, S.I. 136 associativity 10
Adian-Rabin theorem 469 generalized 11
adjacency 174 attaching map 398
adjoining roots 92 attaching a 2-cell along cx 398
adjoint transformation 242 automorphism
admissible subgroup 151 field 93
aEne graph 174
group Aff(n, k) 264 group 104,156
group of the plane 71 Steiner system 295
hyperplane 265 tower 163
isomorphism 266 auxiliary table 352
line 265 axiom of choice 481
m-subspace 265
map 70
n-space 265 Baer, R. 111,327,334,350,383
aEnity 264 Baer theorem 320
Alperin-Kuo theorem 210 Baer-Levi theorem 391
alphabet of Turing machine 421 balanced group 365
alternating bilinear form 235 base
alternating group wreathproduct 172
A, 23 HNN extension 407,411
infinite a.g. A, 51 basepoint 370
amalgam 401. basic move of Turing machine 421
Artin, E. 227,347 basic subgroup 326
 Index

basis chain conditions (see ACC, DCC, or tree 372 degree

free abelian group 312 both chain conditions) universal covering 383 field extension 93
free group 343 character group 205, 340 vertices 366 G-set 55
free semigroup 349 characteristic wedge 399 vertex in graph 358
theorem 128,319,447 field 218 components of complex 369 Dehn, M. 430
bijection 480 abelian group of rank 1 332 composition derivation 21 1
bilinear form 235 subgroup 104 factors 101 derived series 104
alternating 235 characteristically simple 106 series 98 derived subgroup = commutator
symmetric 235 Chevalley groups 246 computation of Turing machine subgroup 33
binary tetrahedral group 350 chief series 152 422 Descartes, R. 90
block Chinese remainder theorem for G-sets congruence class 13 descending central series 113
G-set 256 260 congruence on semigroup 349 descending chain condition 146
Steiner system 293 Church's thesis 423 congruent matrices 236 diagram 435
nontrivial 256 circle group T 18 congruent matrices, o- 236 commutative 183
Boone, W.W. 430,450 circuit 371 conjugacy class 43 Dickson, L.E. 232,246
Boone's lemma 431 class equation 74 conjugate elements 31,43 dicyclic group DC, 351
Boone theorem 417 class of nilpotency 115 subgroups 44 dihedral group D,, 68
Boone-Higman theorem 466 classical groups 239 conjugation 18 dilatation 228
both chain conditions 146 classification theorem of finite simple connected complex 368 dimension
boundary word 435 groups 246 continuation 381 afline space 265
bouquet of circles 377 closed path 368 contraction of Steiner system 294 complex 367
braid group 347 coboundary 183 contravariant functor 336 projective space 273
Brandis, A. 210 cocycle 180 convex combination 70 simplex 366
Britton, J.L. 430 identity 180 Corner, A.L.S. 334 direct factor 126
Britton's lemma 414 cohomology group correspondence theorem 38 direct product
Bruck-Ryser theorem 294 first 212 coset 4 infinite 308
Burnside basis theorem 124 second 183 collapse 354 K x Q 40
Burnside lemma 58 Cole, F.N. 225,292 double 31 direct sum
Burnside normal complement the(3rem collineation 274 enumeration 354 infinite 308
196 Collins, D.J. 430 countable 484 matrices 138
Burnside problem 136 coloring 60 counting principle 294 modules 135
Burnside theorem 107,110 common system of representatives covariant functor 335 direct summand 126,308
commutative diagram 153 cover group 208 directed graph 356
commutator 33 covering directed polygon 433
Cameron, P.J. 256 identities 119 complex 377 labeled directed polygon 433
cancellation law 5, 15 subgroup 33 map 382 Dirichlet, P.G.L. 200
Cardano, C. 90 commute 5 Coxeter, H.S.M. 351 disjoint permutations 5
Carmichael, R. 35 companion matrix 138 crossed homomorphism 211 disjoint subcomplexes 368
Carmichael- Witt theorem 297 complement 167 cubic formula 90 divisible by n 309
Cassidy, P.J. 34 complete factorization 6 cycle 3 divisible group 207, 320
Cauchy, A.L. 1,4 complete group 158 index 60 division algorithm 486
Cauchy theorem 102 complete set of coset representatives = structure, same 46 double coset 31
CAYLEY 34 transversal 178 cyclic doubly transitive 250
Cayley graph 357 complete wreath product 175 group 21 dual diagram 323
Cayley theorem 52 complex 366 permutation of word 434 dyadic rationals 331
center 44 components 369 subgroup 21
centerless 45 connected 368 submodule 135
central extension 201 covering 377 cyclically reduced word 434 edge 367
central isomorphism 151 dimension 367 Eilenberg, S. 358
central series 117 edge 367 Eilenberg-Moore 54
centralizer isomorphic 371 data 179 elementary
element 44 pointed 370 DCC 146 abelian p-group 42
subgroup 112 quotient 373 decision process 418 divisors, abelian group 132
centralizes 112 simply connected 372 Dedekind law 37 divisors, matrix 141
 Index

elementary (cont.) Fitting subgroup 118 multiply transitive 250 hamiltonian 87

moves on path 369 Fitting's lemma primitive 256 holomorph 164
operation in semigroup 426 groups 147 rank 249 icosahedral 69
transvection B,(/Z) 220 Q-groups 153 regular 252 indecomposable 145
empty word 344 fixed element of G-set 248 right 55 infinite alternating A, 51
end fixes 3 sharply k-transitive 25 1 infinite dihedral D, 391
path 368 field 93 trivial- 248 integers mod n Z, 13
path class 368 four group = 4-group V 15 Galois, E. 1,493 MI, 284
endomorphism 144 Frattini argument 81 Galois field GF(q) 218 MI, 288
ring 334 Frattini subgroup @(G) 122 Galois group 93 MI2 289
Engel theorem 115 Frattini theorem 123 Galois theorem 96 Mzz 290
enumerates 422 free Gaschutz, W. 123 MZ3 291
equivalence abelian group 312 Gaschutz theorem 191 Mz4 292
characteristics 332 group 343 gcd 487 modular PSL(2, Z) 391
class 477 semigroup 349 general linear group GL(n, k) 13,219 nilpotent 115
extensions 183 free product 388 generalized associativity 11 octahedral 69
forms 238 with amalgamated subgroup = generalized quaternions Q, 87 operator 151
normal series 99 amalgam 401 generates 22 orthogonal O(n, 52) 64
relation 477 freely reduced word 434 generators and relations orthogonal O(n, k) 239
G-invariant 257 Freiheitsatz 449 abelian groups 314 p-group 73
Erlangen program 72 Fridman, A.A. 450 groups 345 p-primary 126
Euclid's lemma 490 Frobenius, G. 58, 132, 199 graph 174 perfect 263,358
Euler q-function 27 Frobenius complement 254 directed 356 periodic 155
Euler-Poincark characteristic 384 Frobenius group 254 Green, J.A. 366 projective unimodular PSL(n, k) 223
evaluation map 341 Frobenius kernel 252 group of motions M ( n R) 64 quaternions Q 83
even permutation 8 Fuchs, L. 334 group of units 13 quotient 32
exact sequence 307 full subcomplex 367 group 12 reduced abelian 322
short 307 fully invariant subgroup 108 abelian 13 rotation 65
exponent 26 function 479 afine Aff(n, k) 264 simple 39
minimal 202 functor affme group of the plane 71 solvable 97
Ext(Q, K) 186 contravariant 336 alternating A, 23 special linear SL(n, k) 23
extension 154 covariant 335 automorphism group 156 supersolvable 107
central 201 left exact 336 balanced 365 symmetric S, 12
extra-special group 124 Fundamental Theorem binary tetrahedral 350 symmetry group of figure 67
Arithmetic 101,490 braid 347 symplectic Sp(2m, k) 238
Combinatorial Group Theory 436 characteristically simple 106 T (order 12) 84
factor group = quotient group 32 Finite Abelian Groups 132 circle T 18 tetrahedral 69
normal series 97 Finitely Generated Abelian Grou classical 239 torsion 308
factor set 180 319 complete 158 torsion-free 155,308
faithful G-set 248 Finitely Generated Modules 142 cyclic 21 unitary U(n, k) 238
Feit-Thompson theorem 107 Projective Geometry 277 dicyclic DC, 351 unitriangular UT(n, k) 82
Fermat theorem 26 fundamental group 370 dihedral D,, 68 groupoid 370
Feynmann, R.P. 84 divisible 320 Gruenberg lemma 214
fiber 380 elementary abelian p- 42 Gruenberg- Wehrfritz 215
finite G-invariant equivalence relation 257 extra-special 124 Grushko theorem 393
field GF(q) 28 G-isomorphism 260 finitely generated 3 14 Grun, 0. 199
module 135 G-map 260 finitely related 400 Godel image of presentation 465
order, module 135 G-set 55 four group = 4-group V 15 Godel number 423
finitely generated group 314 block 256 free 343
module 135 doubly transitive 250 free abelian 312
finitely presented 400 faithful 248 fundamental 370 h-special word 427
finitely related = finitely presented 400 imprimitive 256 Galois 93 Hall, M. 387
first cohomology group 212 isomorphism 282 general linear GL(n, k) 13 Hall, P. 25, 116, 166
first isomorphism theorem 35 k-transitive 250 generalized quaternions Q, 87 Hall subgroup 110
 Index Index

Hall theorem 108,110 terminal 422 Kulikov, L.J. 330 Miller, G.A. 292
Hall-Higman theorem 136 integers mod n Z, 13 Kulikov theorem 327 minimal exponent 202
Hamilton, W.R. 69 intersection of subcomplexes 368 Kurosh theorem 392 minimal generating set 124
hamiltonian group 87 invariant factors Kuznetsov, A.V. 466 minimal normal subgroup 105
height 332 abelian group 129 minimum polynomial 139
sequence 332 matrix 139 modular group PSL(2, Z) 391
hermitian form 235 invariant subspace 135 labeled directed polygon 433 modular law 37
higher centers ci(G) 113 inverse Lagrange, J.-L. 1,57 module 134
higher commutator subgroups G"' element 14 Lagrange's theorem 26 monomial matrix 46,177
104 function 480 Lam, C. 294 Moore, E.H. 493
Higman, G. 86,136,407 word 344 Lam-Leep 162,285 motion 63
Higman theorem 451 inverse image of subcomplex 377 Landau, E. 77 orientation-reversing 65
Higrnan, Neumann, and Neumann involution 68 Latin square 18 moves 3
theorem 404,405 involves 413 least criminal 111 multiplication by m 128
Hirshon R. 150 isometry 237 left exact functor 336 multiplication table 16
HNN extension 407,411 - isomorphism length multiplicator = Schur multiplier 201
holomorph 164 complexes 371 cycle 3 multiplier = Schur multiplier 201
homogeneous coordinates 273 field 93 normal series 97 multiply transitive 250
homomorphism 16 groups 16 path 368
crossed 211 G-sets 282 word 418
homotopic 369 Ivanov, S. 136 Levi, F. 383 N / C lemma 156
Hopf's formula 358 Iwasawa criterion 263 lifting natural map 33,374
Humphreys, J.E. 391 extension 178 Navel, Morris (see Pippik, Moishe)
Holder,O. 101,154 Jacobi identity 118 path 378 Neumann, B.H. 407
Holder theorem 160, 197 linear fractional transformation 281 Neumann, B.H. theorem 406
James, R. 86
hyperbolic plane 240 linear functional 229 Neumann, H. 407
J6nsson, B. 334
hyperplane 66,228 Lodovici Ferrari 90 Neumann, P.M. 86,391
Johnson, D.L. 350
Jordan, C. 101,286,289 lower central series 113 Newman, M.F. 86
block 139 Lyndon, R.C. 433 Nielsen-Schreier theorem 383
icosahedral group 69 nilpotent
canonical form 141
idempotent endomorphism 144 endomorphism 147
decompositions 144
identity element 14 Jordan-Dickson theorem 232 Mac Lane, S. 358 group 115
identity function 1, 480 Magnus, W. 449 Noether, E. 35
Jordan-Holder theorem
image = irn 22 Mann, A. 74 Nombril, kiaurice (see Bmbellico,
groups 100
imbedding 23 Mann, H.B. 31 Mario)
R-groups 152
free product 388 Markov property 468 nondegenerate quadratic form 244
Jordan-Moore theorem 225
imprirnitive juxtaposition 344 Markov-Post Theorem 428 nondegenerate space 236
G-set 256 Mathieu groups nongenerator 123
system 257 MIo 284 nonsingular semilinear transformation
inclusion function 480 k-transitive 250 M I , 288 267
indecomposable group 145 Kaloujnine, L. 176, 187 MIz 289 nontrivial block 256
independence in abelian group 127, Kaplansky, I. 19,330 MZz 290 normal endomorphism 144
3 10 kernel MZ3 291 normal form
index 25 derivation 213 Mz4 292 amalgams 402
infinite alternating group A, 51 group homomorphism 22 matrix of linear transformation free product 390
infinite dihedral group D, 391 ring homomorphism 485 relative to ordered basis 137 HNN extensions 416
injection 480 semigroup homomorphism 350 maximal divisible subgroup 321 normal series 97
injective property 320 Klein, F. 72 maximal normal subgroup 39 normal subgroup 30
inner automorphism 156 Kostrikin, A.L. 136 maximal tree 373 generated by 31
inner automorphism group Inn(G) 156 Kronecker, L. 128 McCormick, G. 200 minimal 105
inner product delta dij 498 McIver, A. 86 normalizer 44
matrix 235 Krull-Schmidt theorem McKay, J.H. 74. normalizer condition 116
space 235 groups 149 McLain, D.H. 115,176 normalizes 112
instantaneous description 421 Q-groups 153 Miller, C.F., III 430,467 Novikov, P.S. 136,430
 Index Index

Novikov-Boone-Britton theorem perfect field 224 pullback 400 permutation 5

43 1 perfect group 263,358 pure subgroup 325 representation (left) 52
periodic 155 pure-independent 326 representation (right) 54
permutation 2 pushout 395 wreath product 175
O'Brien, E.A. 86 even 8 relation table 351
octahedral group 69 matrix 18 relatively prime 488
odd permutation 8 odd 8 quadratic form 244 relator polygon 435
Ol'shanskii, A.Y. 136 regular 5 nondegenerate 244 Remak, R. 144
Ombellico, Mario (see ~ [ ~ J I K M
O Bm, r ~ ) disjoint 5 quadruple of Turing machine 421 repeated roots 95
operation 10 pi-reduced word 415 quartic formula 90 representation
operator group 151 pinch 413 quasi-inverse 19 conjugates 53
orbit 56 Pippik, Moishe (see Nombril, Maurice) quaternions Q 83 cosets 53
order Poincart, H. 54 generalized quaternions Q, 87 theory 471
group 26 pointed complex 370 quotient representative of coset 24
group element 21 Polya theorem 61 complex 373 restricted Burnside problem 136
module 135 polygon 433 group 32 restricted wreath product 175
orientation-reversing motion 65 relator polygon 435 module 134 retract 168
origin polyhedral group 347 semigroup 349 retraction 168
path class 368 Pontrjagin duality 340 Rips, E. 438,461
path 368 positive word 349 rotationS. 65387
Rosset,
Ornstein, D.S. 114 powers 14 r-cycle 3
orthogonal complement 239 presentation r-valued function 1 group 65
orthogonal elements 239 abelian group 314 R-homomorphism 141 Rotman, J.J. 370
orthogonal group group 345 R-isomorphism 141 Ruffmi, P. 1,97
otn, 64 semigroup 350 R-module 134
O(n, k) 239 primary decomposition 126 Rabin theorem 467
orthogonal matrix 65 infinite 311 rank same cycle structure 46
orthogonal transformation 63 primary module 135 abelian group 331 Schenkman, E. 130
orthonormal basis 63,243 prime field 217 free abelian group 315 Schering, E. 128
outer automorphism 156 primitive element 218 free group 348 Schottenfels theorem 233
group 156 primitive G-set 256 G-set 249 Schreier, 0. 100,383
principal derivation 21 1 rational canonical form 139 Schreier refinement theorem 100
principal ideal domain 456 real quaternions 86 Schreier theorem 185
p-complement 110 product formula 30 realizes data 181 Schreier transversal 385
p-group 73 product of paths 369 realizes 6 169 Schupp, P.E. 406,414,417
elementary abelian 42 projection 144 recursive 422 Schur, 1. 39,198,210
p-nilpotent 197 projection of covering complex 377 recursively enumerable = r.e. 422 Schur multiplier 201
p-primary component 126,311 projective r.e. = recursively enumerable 422 Schur theorem 114,198,208,363
group 126 hyperplane 273 r.e. presentation 465 Schur-Zassenhaus lemma 190
partition 477 lifting property 205 r.e. subset of N 422 Scipione del Ferro 89
partition of n 48 line 273 recursively presented 45 1 second cohomology group 183
path 368 m-subspace 273 reduced abelian group 322 second isomorphism theorem 36
class 369 n-space Pn(k) 273 reduced Seifert, H. 396
circuit 37 1 plane of order n 294 path 371 semidirect product 167
closed 368 point 273 word 344 semigroup 12
end 368 property 315,348 in free product 389 free 349
homotopic 369 representation 202 Ree, R. 246 quotient 349
length 368 projective unimodular group refinement of normal series 98 semilinear fractional transformation
lifting 378 PSL(n, k) 223 reflection 66 281
origin 368 projectivity 274 region 435 semilinear transformation 267
product 369 proper subgroup 22 regular nonsingular 267
reduced 371 Prufer, H. 330 G-set 252- sequence 479
trivial 370 Priifer-Baer theorem 327 graph 358 Serre, J.-P. 383
pentad 161 Prufer theorem 328 normal subgroup 259 sgn = signum 8

sharply k-transitive 25 1
sheets of covering complex 377
Shelah, S. 334
Sheu, T.-L. 48
short exact sequence 307
signum = sgn 8
similarity 142
simple group 39
simplex 366
simplicia1map 371
simply connected 372
Sims, C. 86
simultaneous bases 3 19
skeleton 367
Smith normal form 143
solvable by radicals 92
solvable conjugacy problem 449
solvable group 97,102
solvable series 102
solvable word problem
groups 418,425,465
semigroups 425
special linear group SL(n, k) 23,220
special word 43 1
splits 167
splitting field 92
sporadic simple groups 247
stabilizer 56
extension 185
series 165
stable letters in HMN extension 407,
411
standard sane space 265
standard basis 138
star in Steiner system 294
Steinberg, R. 246
Steiner system 293
automorphism 295
block 293
contraction at x 294
star 294
type 293
Stickelberger, L. 132
stopping state 424
subcomplex 367
disjoint 368
full 367
inverse image 377
subgroup generator table 355
subgroup 20
admissible 151
basic 326
center 44
centralizer, element 44
Index
Index
Tietze's theorem 374 Wedderburn, J.M.H. 144,277
centralizer, subgroup 112 Todd, J.A. 351 wedge 399
characteristic 104 torsion group 308 Weichsel, P.M. 34
commutator 33 torsion subgroup 303 well defined 480
cyclic subgroup generated by an torsion theorem for amalgams 404 well-ordering principle 482
element 21 torsion-free 155, 308 Wielandt, H. 124, 163
Frattini @(G) 122 transfer 194 Wielandt's proof 81
fully invariant 108 transgression 205 Wilson's theorem 15
generated by X 22 transitive 58 Witt, E. 297
Hall 110 transitive extension 286 word 23,344
higher commutator 104 translation 15, 63, 264 boundary word 435
maximal divisible subgroup 321 transposition 3 cyclically reduced 434
maximal normal 39 transvection 220,228 empty 344
normal 30 elementary Bij(A) 220 freely reduced 434
normal generated by 31 transversal 178 h-special 427
normalizer 44 tree 372 inverse 344
- proper 22 maximal 373 positive 349
pure 325 triangulated polygon 397 reduced 344
regular normal 259 trivial in free product 389
subnormal 150 action 172 special 431
Sylow 78 G-set 248 wreath product 172
torsion 308 path 370 base 172
trivial 22 subgroup 22 complete 175
n'- 110 Turing machine 421 permutation version 173
z- 110 Turing, A.M. 420 regular 175
submodule 134 type restricted 175
generated by X 135 abelian group 332
subnormal subgroup 150 Steiner system 293
substitution, law of 10
subword 344,413 Yff,P. 34
supersolvable group 107 Ulm, H. 329
surjection 480 unimodular matrix 220
Suzuki, M. 198,246 union of subcomplexes 368 Zassenhaus, W. 39,284,289
Sylow p-subgroup 78 unipotent matrix 144 Zassenhaus lemma 99
Sylow theorem 79 unit 13,487
symmetric Zelmanov, E.7. 136
unitary group U(n,lc) 238 Zorn's lemma 481
bilinear form 235 unitriangular group UT(u, k) 82
function 56 unitriangular matrix 82
groupsn 12 universal central extension 360
symmetry group of figure Y(8) 67 universal covering complex. 383 rliynrco~,MIXAIA
(see Navel, Morris)
symplectic universal finitely related group 465 n'-subgroup 110
basis 241 upper central series 113 n-subgroup 110
group Sp(2m7k) 238 o-congruent 236
syntheme 161 R-composition series 152
van der Waerden trick 344,390 Q-group 151
van Kampen theorem 396 Q-indecomposable 153
Tartaglia (Niccolo Fontana) 90 vertices 174, 366 R-map 151
Tate, J. 199 von Dyck, W. 69 Q-series 152
terminal instantaneous description 422 von Dyck theorem 346 R-simple group 152
tetrahedral group 69
third isomorphism theorem 37
Thomas, S. 163
Thompson, J.G. 107, 199,254,289
three subgroups lemma 118
Thue, A. 430
Graduate Texts in Mathematics
continuedfmrn page ii

65 WELLS.Differential Analysis on Complex 94 WARNER. Foundations of Differentiable

Manifolds. 2nd ed. Manifolds and Lie Groups.
66 WATERHOUSE. Introduction to Affine Group 95 SHIRYAYEV. Probability.
Schemes. 96 CONWAY. A Course in Functional Analysis.
67 SERRE.Local Fields. 2nd ed.
68 WEIDMANN. Linear Operators in Hilbert 97 KOBLITZ. Introduction to Elliptic Curves and
Spaces. Modular Forms. 2nd ed.
69 LANG.Cyclotomic Fields 11. 98 BROCKERITOM DIECK. Representations of
70 MASSEY.Singular Homology Theory. Compact Lie Groups.
71 FARKASKRA. Riemann Surfaces. 2nd ed. 99 GROV~BENSON. Finite Reflection Groups.
72 STILLWELL. Classical Topology and 2nd ed.
Combinatorial Group Theory. 2nd ed. 100 BERG/CHR~STENSEN/~ESSEL. Harmonic
73 HUNGERFORD. Algebra. Analysis on Semigroups: Theory of Positive
74 DAVENPORT. Multiplicative Number Theory. Definite and Related Functions.
2nd ed. 101 EDWARDS. Gdois Theory.
75 HOCHSCHILD. Basic Theory of Algebraic 102 VARADARAJAN. Lie Croups, Lie Algebras and
Groups and Lie Algebras. Their Representations.
76 IITAKA. Algebraic Geometry. 103 LANG.Complex Analysis. 3rd ed.
77 HECKE.Lectures on the Theory of Algebraic 104 D v s ~ o v r ~ / F o ~ ~ ~ ~ o / Modem
Nov~~ov.
Numbers. Geometry-Methods and Applications.
78 BURRIS/SANKAPPANAVAR. A C O U ~inS ~ Part 11.
Universal Algebra. 105 LANG.SL2(R).
79 WALTERS. An Introduction to Ergodic 106 SILVERMAN. The Arithmetic of Elliptic
Theory. Curves.
80 ROBINSON. A Course in the Theory of 107 OLVER.Applications of Lie Groups to
Groups. Differential Equations. 2nd ed.
81 FORSTER. Lectures on Riemann Surfaces. 108 RANGE.Holomorphic Functions and Integral
52 BorrRu. Differential Forms in Algebraic Representations in Several Complex
Topology. Variables.
83 WASHINGTON. Introduction to Cyciotomic 109 LEHTO.Univalent Functions and Teichmiiller
Fields. Spaces.
84 IRELAND~OSEN. A Classical Introduction to 110 LANG.Algebraic Number Theory.
Modern Number Theory. 2nd ed. 111 HUSEMOLLER. Elliptic Curves.
85 EDWARDS. Fourier Series. Vol. TI. 2nd ed. 112 LANG.Elliptic Functions.
86 VAN LINT.Introduction to Coding Theory. 113 KARATZAS~~HREVE. Brownian Motion and
2nd ed. Stochastic Calculus. 2nd ed.
87 BROWN.Cohomology of Groups. 114 K ~ B LA~Course . in Number Theory and
88 PIERCE. Associative Algebras. Cryptography. 2nd ed.
89 LANG.Introduction to Algebraic and Abelian 115 BERGERJG~STJAUX. Differential Geometry:
Functions. 2nd ed. Manifolds, Curves, and Surfaces.
90 BR~~NDSTED. An Introduction to Convex 116 KELLEY/~RINIVASAN. Measure and Integral.
Polytopes. Vol. I.
91 BEARDON. On the Geometry of Discrete 117 SERRE.Algebraic Groups and Class Fields.
Croups. 118 PEDERSEN. Analysis Now.
92 DIESTEL.Sequences and Series in Banach 119 ROTMAN.An Introduction to Algebraic
Spaces. Topology.
93 DUBROVIN/~MENKO/NOVIKOV.Modern 120 ZIEMER.Weakly Differentiable Functions:
Geometry-Methods and Applications. Part Sobolev Spaces and Functions of Bounded
I. 2nd ed. Variation.