Introduction

to
Groups, Invariants
and
Particles
Frank W. K. Firk, Emeritus Professor of Physics, Yale University

2000
 CONTENTS 2

1. INTRODUCTION 3

2. GALOIS GROUPS 8

3. SOME ALGEBRAIC INVARIANTS 20

4. SOME INVARIANTS IN PHYSICS 29

5. GROUPS CONCRETE AND ABSTRACT 43

6. LIE’S DIFFERENTIAL EQUATION, INFINITESIMAL

ROTATIONS, AND ANGULAR MOMENTUM 57

7. LIE’S CONTINUOUS TRANSFORMATION GROUPS 68

8. PROPERTIES OF n-VARIABLE r-PARAMETER

LIE GROUPS 78

9. MATRIX REPRESENTATIONS OF GROUPS 83

10. SOME LIE GROUPS OF TRANSFORMATIONS 94

11. THE GROUP STRUCTURE OF LORENTZ

TRANSFORMATIONS 107

12. ISOSPIN 115

13. LIE GROUPS AND THE STRUCTURE OF MATTER 128

14. LIE GROUPS AND THE CONSERVATION LAWS

OF THE PHYSICAL UNIVERSE 159

15. BIBLIOGRAPHY 163

 1 3

INTRODUC TION

The noti on of geom etrical symmetr y in Art a nd in Natu re is a

familiar one. In Moder n Phy sics, this noti on ha s evo lved to in clude

sym metri es of an a bstra ct ki nd. These new symme tries play an e ssential

par t in the t heories of the microstruc ture of ma tter. The basic sym metri es

found in Natu re se em to orig inate in t he ma thema tical stru cture of t he laws

the mselves  laws th at go vern the m otions of the g alaxies on the one

han d and the motions of quar ks in nucleons on th e oth er.

In the N ewton ian era, t he laws of Natu re we re de duced from a sm all

num ber o f imperfect obs ervat ions by a small numb er of reno wned

scientists an d mat hemat icians. It was not until the Einsteinian era ,

how ever, that the significance of the symme tries associated wit h the laws

was fully app reciated. The discovery of space-time symmet ries has led to

the widely-held belief that the l aws o f Nat ure c an be deri ved f rom

sym metry , or invariance, pri nciples. Our i ncomp lete knowledge of th e

fundamen tal intera ctions mea ns th at we are not y et in a po sition to confirm

this belief. We t herefore u se ar gumen ts ba sed o n emp irically established

laws and rest ricted sym metry prin ciples to guide us i n our sear ch fo r the

fundamen tal symmet ries. Fre quent ly, it is impor tant to un derst and w hy

the symm etry of a system is obser ved t o be broke n.

 In Geome try, an ob ject with a definite shap e, size, location, a nd 4

ori entat ion constitutes a st ate w hose symme try p roper ties, or i nvariants,

are to b e stu died. Any tran sformation that leaves th e sta te un chang ed in

for m is called a s ymmet ry tr ansformati on. The g reate r the numb er of

sym metry tran sformations tha t a s tate can u nderg o, th e higher i ts

sym metry . If the numbe r of conditions that define th e sta te is redu ced

the n the symm etry of th e sta te is incr eased. Fo r exa mple, an o bject

cha racte rized by o blateness alone is symmet ric u nder all transformat ions

exc ept a chan ge of shap e.

In descr ibing the symme try o f a sta te of the most gener al kind (n ot

simply g eomet ric), the algebraic struc ture of th e set of s ymmet ry op erato rs

mus t be given; it is not suf ficient to give the numbe r of opera tions, and

not hing else. The law of c ombin ation of the opera tors must be st ated. It

is the alg ebrai c gro up th at fu lly charac terizes th e sym metry of t he ge neral

sta te.

The The ory o f Gro ups ca me ab out u nexpe ctedly. G alois show ed

tha t an equat ion o f deg ree n , whe re n is an inte ger g reate r tha n or equal to

five can not, in general, be solved by algebraic means. In the cours e of this

gre at wo rk, h e dev eloped the ideas of Lagra nge, Ruffini, and Ab el and

introduc ed th e con cept of a gro up. Galois discussed the func tional

relationships amon g the root s of an eq uation, an d sho wed t hat t he

relationships have symm etries associated with th em un der p ermut ations of

the root s.
 The oper ators that tran sform one funct ional rela tions hip i nto 5

ano ther are e lemen ts of a se t tha t is characteristic of th e equ ation ; the set

of operators is ca lled the G alois grou p of the e quati on.

In the 1 850’s , Cay ley showed that ever y finite g roup is isomorp hic

to a cer tain permu tation gro up. The g eomet rical symm etries of cryst als

are desc ribed in t erms of finite group s. T hese symme tries are discussed in

man y sta ndard work s (see bibliography) and there fore, they will not be

discussed in this book.

In the b rief period bet ween 1924 and 1 928, Quant um Me chanics

was deve loped. Almost immediately, it was r ecogn ized by Weyl, and by

Wigner, that certa in parts o f Gro up Th eory could be u sed a s a p owerful

ana lytical to ol in Quan tum P hysics. T heir ideas have been deve loped over

the deca des in man y are as th at ra nge f rom t he Th eory of So lids to Pa rticle

Phy sics.

The essential role played by grou ps th at ar e cha racte rized by

par amete rs th at va ry co ntinuously in a given ran ge wa s first em phasized

by Wigner. T hese group s are know n as Lie Gro ups. They have becom e

increasingly impor tant in ma ny br anches of conte mpora ry ph ysics,

par ticularly Nuclear an d Par ticle Phys ics. Fifty yea rs af ter G alois had

introduc ed th e con cept of a group in t he Th eory of Eq uations, L ie

introduc ed th e con cept of a continuous tran sformation grou p in the T heory

of Differential Eq uations. Lie’s theo ry un ified many of t he disconnected

met hods of solving different ial equations t hat h ad ev olved over a pe riod of
two hund red y ears. Inf initesimal unit ary t ransformat ions play a key role in 6

discussions o f the fund ament al conserv ation laws of P hysics.

In Classical Dynam ics, the i nvariance of th e equ ations of motion of a

par ticle, or system of particles, unde r the Galilean trans forma tion is a basic

par t of every day r elativity. The sear ch fo r the tran sformation that leaves

Maxwell’s equ ations of Electromag netism unc hange d in form (invariant)

und er a linear tra nsformation of the s pace-time coord inates, led to the

discover y of the L orent z tra nsformation. T he fu ndame ntal impor tance of

this tra nsformation, an d its rela ted i nvariants, cann ot be over stated.

This introduc tion to Gr oup T heory , with its emph asis on Li e Gro ups

and thei r app lication t o the stud y of symme tries of t he fu ndame ntal

con stituents of ma tter, has its origin in a one- semester c ourse that I ta ught

at Yale University for more than ten y ears. The cour se wa s dev eloped for

Sen iors, and advan ced J unior s, ma joring in the P hysical Sciences. T he

stu dents had gener ally completed the c ore c ourse s for thei r maj ors, and

had take n int ermed iate level cour ses in Lin ear A lgebra, Real and Com plex

Ana lysis, Ord inary Line ar Di fferential Equa tions, and some of t he Sp ecial

Fun ctions of Physics. Group Theo ry wa s not a ma thema tical requ irement

for a de gree in th e Phy sical Sciences. The majo rity of ex isting

und ergra duate text books on G roup Theor y and its applications in Phys ics

ten d to be either highly qua litative or hig hly m athem atical. The pu rpose of

this introduc tion is to stee r a m iddle cour se th at pr ovides the stud ent w ith

a s ound mathe matical basis for st udying the symm etry prope rties of t he
fundamen tal p articles. It i s not gene rally appr eciated by Phys icists tha t 7

con tinuous tr ansformati on gr oups originated in t he Th eory of Differential

Equ ations. T he infinitesimal generato rs of Lie Group s the refor e have

forms that involve differential operators and their commutators, and these

operators and their algebraic properties have found, and continue to find, a

natural place in the development of Quantum Physics.

Guilford, CT.

June, 2000
 2 8

GALOIS GROUPS

In the early 19th - century, Abel proved that it is not possible to solve the

general polynomial equation of degree greater than four by algebraic means.

He attempted to characterize all equations that can be solved by radicals.

Abel did not solve this fundamental problem. The problem was taken up and

solved by one of the greatest innovators in Mathematics, namely, Galois.

2.1. Solving cubic equations

The main ideas of the Galois procedure in the Theory of Equations,

and their relationship to later developments in Mathematics and Physics, can

be introduced in a plausible way by considering the standard problem of

solving a cubic equation.

Consider solutions of the general cubic equation

Ax3 + 3Bx 2 + 3Cx + D = 0, where A - D are rational constants.

If the substitution y = Ax + B is made, the equation becomes

y 3 + 3Hy + G = 0

where

H = AC - B2

and

G = A2D - 3ABC + 2B3.

The cubic has three real roots if G2 + 4H 3 < 0 and two imaginary roots if G 2 9

+ 4H 3 > 0. (See any standard work on the Theory of Equations).

If all the roots are real, a trigonometrical method can be used to obtain

the solutions, as follows:

the Fourier series of cos3u is

cos3u = (3/4)cosu + (1/4)cos3u.

Putting

y = scosu in the equation y3 + 3Hy + G = 0

(s > 0),

gives

cos3u + (3H/s2)cosu + G/s3 = 0.

Comparing the Fourier series with this equation leads to

s = 2 √(-H)

and

cos3u = -4G/s3.

If v is any value of u satisfying cos3u = -4G/s3, the general solution is

3u = 2nπ ± 3v, where n is an integer.

Three different values of cosu are given by

u = v, and 2π/3 ± v.

The three solutions of the given cubic equation are then

scosv, and scos(2π/3 ± v).

Consider solutions of the equation

 x 3 - 3x + 1 = 0. 10

In this case,

H = -1 and G2 + 4H 3 = -3.

All the roots are therefore real, and they are given by solving

cos3u = -4G/s3 = -4(1/8) = -1/2

or,

3u = cos-1(-1/2).

The values of u are therefore 2π/9, 4π/9, and 8π/9, and the roots are

x1 = 2cos(2π/9), x2 = 2cos(4π/9), and x3 = 2cos(8π/9).

2.2. Symmetries of the roots

The roots x1, x2, and x3 exhibit a simple pattern. Relationships among

them can be readily found by writing them in the complex form 2cosθ = e iθ +

e-iθ where θ = 2π/9 so that

x1 = e iθ + e -iθ ,

x2 = e 2iθ + e -2iθ ,

and

x3 = e 4iθ + e -4iθ .

Squaring these values gives

x 12 = x2 + 2,

x 22 = x3 + 2,

and
 x 32 = x1 + 2. 11

The relationships among the roots have the functional form f(x1,x2,x3) = 0.

Other relationships exist; for example, by considering the sum of the roots we

find

x 1 + x22 + x2 - 2 = 0

x 2 + x32 + x3 - 2 = 0,

and

x 3 + x12 + x1 - 2 = 0.

Transformations from one root to another can be made by doubling-the-

angle, .

The functional relationships among the roots have an algebraic

symmetry associated with them under interchanges (substitutions) of the

roots. If is the operator that changes f(x1,x2,x3) into f(x2,x3,x1) then

f(x1,x2,x3) → f(x2,x3,x1),
2
f(x1,x2,x3) → f(x3,x1,x2),

and

3
f(x1,x2,x3) → f(x1,x2,x3).
3
The operator = I, is the identity.

In the present case,

(x12 - x2 - 2) = (x22 - x3 - 2) = 0,
and
2
(x12 - x2 - 2) = (x32 - x1 - 2) = 0. 12

2.3. The Galois group of an equation.

2
The set of operators {I, , } introduced above, is called the Galois

group of the equation x3 - 3x + 1 = 0. (It will be shown later that it is

isomorphic to the cyclic group, C3).

The elements of a Galois group are operators that interchange the

roots of an equation in such a way that the transformed functional

relationships are true relationships. For example, if the equation

x 1 + x22 + x2 - 2 = 0

is valid, then so is

(x1 + x22 + x2 - 2 ) = x2 + x32 + x3 - 2 = 0.

True functional relationships are polynomials with rational coefficients.

2.4. Algebraic fields

We now consider the Galois procedure in a more general way. An

algebraic solution of the general nth - degree polynomial

aoxn + a 1xn-1 + ... an = 0

is given in terms of the coefficients ai using a finite number of operations (+,-

,×,÷,√). The term "solution by radicals" is sometimes used because the

operation of extracting a square root is included in the process. If an infinite

number of operations is allowed, solutions of the general polynomial can be

obtained using transcendental functions. The coefficients ai necessarily belong 13

to a field which is closed under the rational operations. If the field is the set

of rational numbers, Q, we need to know whether or not the solutions of a

given equation belong to Q. For example, if

x2 - 3 = 0

we see that the coefficient -3 belongs to Q, whereas the roots of the equation,

xi = ± √3, do not. It is therefore necessary to extend Q to Q', (say) by

adjoining numbers of the form a√3 to Q, where a is in Q.

In discussing the cubic equation x3 - 3x + 1 = 0 in 2.2, we found

certain functions of the roots f(x1,x2,x3) = 0 that are symmetric under

permutations of the roots. The symmetry operators formed the Galois group

of the equation.

For a general polynomial:

x n + a 1xn-1 + a 2xn-2 + .. an = 0,

functional relations of the roots are given in terms of the coefficients in the

standard way

x1 + x2 + x3 .. .. + xn = -a1

x1x2 + x1x3 + .. x 2x3 + x2x4 + ..+ x n-1xn = a2

x1x2x3 + x2x3x4 + .. .. + xn-2xn-1xn = -a3

x1x2x3 .. .. xn-1xn = ±an.

 Other symmetric functions of the roots can be written in terms of these 14

basic symmetric polynomials and, therefore, in terms of the coefficients.

Rational symmetric functions also can be constructed that involve the roots

and the coefficients of a given equation. For example, consider the quartic

x4 + a 2x2 + a 4 = 0.

The roots of this equation satisfy the equations

x 1 + x2 + x3 + x4 = 0

x 1x2 + x1x3 + x1x4 + x2x3 + x2x4 + x3x4 = a 2

x 1x2x3 + x1x2x4 + x1x3x4 + x2x3x4 = 0

x 1x2x3x4 = a 4.

We can form any rational symmetric expression from these basic

equations (for example, (3a43 - 2a2)/2a42 = f(x1,x2,x3,x4)). In general, every

rational symmetric function that belongs to the field F of the coefficients, ai, of

a general polynomial equation can be written rationally in terms of the

coefficients.

The Galois group, Ga, of an equation associated with a field F therefore

has the property that if a rational function of the roots of the equation is

invariant under all permutations of Ga, then it is equal to a quantity in F.

Whether or not an algebraic equation can be broken down into simpler

equations is important in the theory of equations. Consider, for example, the

equation

x 6 = 3.
It can be solved by writing x3 = y, y2 = 3 or 15

x = (√3)1/3.

To solve the equation, it is necessary to calculate square and cube roots

 not sixth roots. The equation x6 = 3 is said to be compound (it can be

broken down into simpler equations), whereas x2 = 3 is said to be atomic.

The atomic properties of the Galois group of an equation reveal

the atomic nature of the equation, itself. (In Chapter 5, it will be seen that a

group is atomic ("simple") if it contains no proper invariant subgroups).

The determination of the Galois groups associated with an arbitrary

polynomial with unknown roots is far from straightforward. We can gain

some insight into the Galois method, however, by studying the group

structure of the quartic

x 4 + a 2x2 + a 4 = 0

with known roots

x 1 = ((-a2 + µ)/2) 1/2 , x 2 = -x 1,

x 3 = ((-a2 - µ)/2) 1/2 , x 4 = -x 3,

where

µ = (a 22 - 4a4)1/2.

The field F of the quartic equation contains the rationals Q, and the

rational expressions formed from the coefficients a2 and a 4.

The relations

x 1 + x2 = x3 + x4 = 0
are in the field F. 16

Only eight of the 4! possible permutations of the roots leave these

relations invariant in F; they are

x1 x2 x3 x4 x1 x2 x3 x4 x1 x2 x3 x4
{ P1 = , P2 = , P3 = ,
x1 x2 x3 x4 x1 x2 x4 x3 x2 x1 x3 x4
x1 x2 x3 x4 x1 x2 x3 x4 x1 x2 x3 x4
P4 = , P5 = , P6 =
x2 x1 x4 x3 x3 x4 x1 x2 x3 x4 x2 x1
x1 x2 x3 x4 x1 x2 x3 x4
P7 = , P8 = }.
x4 x3 x1 x2 x4 x3 x2 x1

The set {P1,...P8} is the Galois group of the quartic in F. It is a subgroup of

the full set of twentyfour permutations. We can form an infinite number of

true relations among the roots in F. If we extend the field F by adjoining

irrational expressions of the coefficients, other true relations among the roots

can be formed in the extended field, F'. Consider, for example, the extended

field formed by adjoining µ (= (a 22 - 4a4)) to F so that the relation

x 12 - x32 = µ is in F'.

We have met the relations

x 1 = -x 2 and x3 = -x 4

so that

x 12 = x22 and x 32 = x42.

Another relation in F' is therefore

 x 22 - x42 = µ. 17

The permutations that leave these relations true in F' are then

{P1, P2, P3, P4}.

This set is the Galois group of the quartic in F'. It is a subgroup of the set

{P 1,...P8}.

If we extend the field F' by adjoining the irrational expression

((-a2 - µ)/2) 1/2 to form the field F'' then the relation

x 3 - x4 = 2((-a2 - µ)/2) 1/2 is in F''.

This relation is invariant under the two permutations

{P1, P3}.

This set is the Galois group of the quartic in F''. It is a subgroup of the set

{P 1, P2, P3, P4}.

If, finally, we extend the field F'' by adjoining the irrational

((-a2 + µ)/2) 1/2 to form the field F''' then the relation

x 1 - x2 = 2((-a2 - µ)/2) 1/2 is in F'''.

This relation is invariant under the identity transformation, P1 , alone; it is

the Galois group of the quartic in F''.

The full group, and the subgroups, associated with the quartic equation

are of order 24, 8, 4, 2, and 1. (The order of a group is the number of

distinct elements that it contains). In 5.4, we shall prove that the order of a

subgroup is always an integral divisor of the order of the full group. The
order of the full group divided by the order of a subgroup is called the index 18

of the subgroup.

Galois introduced the idea of a normal or invariant subgroup: if H is a

normal subgroup of G then

HG - GH = [H,G] = 0.

(H commutes with every element of G, see 5.5).

Normal subgroups are also called either invariant or self-conjugate subgroups.

A normal subgroup H is maximal if no other subgroup of G contains H.

2.5. Solvability of polynomial equations

Galois defined the group of a given polynomial equation to be either

the symmetric group, Sn, or a subgroup of Sn, (see 5.6). The Galois method

therefore involves the following steps:

1. The determination of the Galois group, Ga, of the equation.

2. The choice of a maximal subgroup of Hmax(1) . In the above case, {P1, ...P8}

is a maximal subgroup of Ga = S4.

3. The choice of a maximal subgroup of Hmax(1) from step 2.

In the above case, {P1,..P4} = Hmax(2) is a maximal subgroup of Hmax(1) .

The process is continued until Hmax = {P1} = {I}.

The groups Ga, Hmax(1) , ..,Hmax(k) = I, form a composition series. The

composition indices are given by the ratios of the successive orders of the

groups:
 gn/h(1) , h(1) /h(2) , ...h(k-1)/1. 19

The composition indices of the symmetric groups Sn for n = 2 to 7 are found

to be:

n Composition Indices

2 2

3 2, 3

4 2, 3, 2, 2

5 2, 60

6 2, 360

7 2, 2520

We shall state, without proof, Galois' theorem:

A polynomial equation can be solved algebraically if and only if its

group is solvable.

Galois defined a solvable group as one in which the composition indices are

all prime numbers. Furthermore, he showed that if n > 4, the sequence of

maximal normal subgroups is Sn, An, I, where An is the Alternating Group

with (n!)/2 elements. The composition indices are then 2 and (n!)/2. For n >

4, however, (n!)/2 is not prime, therefore the groups Sn are not solvable for n

> 4. Using Galois' Theorem, we see that it is therefore not possible to solve,

algebraically, a general polynomial equation of degree n > 4.

 3 20

SOME ALGEBRAIC INVARIANTS

Although algebraic invariants first appeared in the works of Lagrange and

Gauss in connection with the Theory of Numbers, the study of algebraic

invariants as an independent branch of Mathematics did not begin until the

work of Boole in 1841. Before discussing this work, it will be convenient to

introduce matrix versions of real bilinear forms, B, defined by

m n
B = ∑ ∑ aijxiyj
I=1 j=1

where

x = [x1,x2,...xm], an m-vector,

y = [y1,y2,...yn], an n-vector,

and aij are real coefficients. The square brackets denote a

column vector.

In matrix notation, the bilinear form is

B = xTAy

where

a 11 . . . a1n

. . . .

A= . . . . .

. . . .

a m1. . . amn

The scalar product of two n-vectors is seen to be a special case of a

bilinear form in which A = I.

 If x = y, the bilinear form becomes a quadratic form, Q: 21

Q = xTAx.

3.1. Invariants of binary quadratic forms

Boole began by considering the properties of the binary

quadratic form

Q(x,y) = ax2 + 2hxy + by2

under a linear transformation of the coordinates

x' = Mx

where

x = [x,y],

x' = [x',y'],

and

|i j|
M = .
|k l|

The matrix M transforms an orthogonal coordinate system into an

oblique coordinate system in which the new x'- axis has a slope (k/i), and the

new y'- axis has a slope (l/j), as shown:

 y 22

y′
[i+j,k+l]

[j,l]
x'

[0,1] [1,1]
x′
[i,k]

[0,0] [1,0] x

Fig. 1. The transformation of a unit square under M.

The transformation is linear, therefore the new function Q'(x',y') is a

binary quadratic:

Q'(x',y') = a'x'2 + 2h'x'y' + b'y'2.

The original function can be written

Q(x,y) = xTDx

where

|a h|
D= ,
|h b|

and the determinant of D is

detD = ab - h2, called the discriminant of Q.

The transformed function can be written

 Q'(x',y') = x' TD'x' 23

where

a' h'
D' = ,
h' b'

and

detD' = a'b' - h'2, the discriminant of Q'.

Now,

Q'(x',y') = (Mx)TD'Mx

= xTMTD'Mx

and this is equal to Q(x,y) if

MTD'M = D.

The invariance of the form Q(x,y) under the coordinate transformation M

therefore leads to the relation

(detM)2detD' = detD

because

detMT = detM.

The explicit form of this equation involving determinants is

(il - jk)2(a'b' - h' 2) = (ab - h2).

The discriminant (ab - h2) of Q is said to be an invariant

of the transformation because it is equal to the discriminant (a'b' - h'2) of Q',

apart from a factor (il - jk)2 that depends on the transformation itself, and not

on the arguments a,b,h of the function Q.

3.2. General algebraic invariants 24

The study of general algebraic invariants is an important branch of

Mathematics.

A binary form in two variables is

f(x1,x2) = aox1n + a 1x1n-1x2 + ...anx2n

= ∑ aix1n-ix2i

If there are three or four variables, we speak of ternary forms or quaternary

forms.

A binary form is transformed under the linear transformation M as

follows

f(x 1,x2) => f'(x1',x2') = a o'x 1' n + a 1'x 1' n-1x2' + ..

The coefficients

a o, a1, a2,..≠ a o', a1', a2' ..

and the roots of the equation

f(x1,x2) = 0

differ from the roots of the equation

f'(x1',x2') = 0.

Any function I(ao,a1,...an) of the coefficients of f that satisfies

r wI(ao',a1',...an') = I(ao,a1,...an)

is said to be an invariant of f if the quantity r depends only on the

transformation matrix M, and not on the coefficients ai of the function being

transformed. The degree of the invariant is the degree of the coefficients, and
the exponent w is called the weight. In the example discussed above, the 25

degree is two, and the weight is two.

Any function, C, of the coefficients and the variables of a form f that is

invariant under the transformation M, except for a multiplicative factor that is

a power of the discriminant of M, is said to be a covariant of f. For binary

forms, C therefore satisfies

r wC(ao',a1',...an'; x1',x2') = C(ao,a1,...an; x1,x2).

It is found that the Jacobian of two binary quadratic forms, f(x1,x2) and

g(x1,x2), namely the determinant

∂f/∂x1 ∂f/∂x2

∂g/∂x1 ∂g/∂x2

where ∂f/∂x1 is the partial derivative of f with respect to x1 etc., is a

simultaneous covariant of weight one of the two forms.

The determinant

∂2f/∂x12 ∂2f/∂x1∂x2
,
∂ g/∂x2∂x1 ∂ g/∂x2
2 2 2

called the Hessian of the binary form f, is found to be a covariant of weight

two. A full discussion of the general problem of algebraic invariants is outside

the scope of this book. The following example will, however, illustrate the 26

method of finding an invariant in a particular case.

Example:

To show that

(aoa2 - a12)(a1a3 - a22) - (aoa3 - a1a2)2/4

is an invariant of the binary cubic

f(x,y) = aox3 + 3a 1x2y + 3a 2xy2 + a 3y3

under a linear transformation of the coordinates.

The cubic may be written

f(x,y) = (aox2+2a1xy+a2y2)x + (a1x2+2a2xy+a3y2)y

= xTDx

where

x = [x,y],

and

a ox + a 1y a 1x + a 2y
D= .
a 1x + a 2y a 2x + a 3y

Let x be transformed to x': x' = Mx, where

i j
M=
k l
then 27

f(x,y) = f'(x',y')

if

D = MTD'M.

Taking determinants, we obtain

detD = (detM)2detD',

an invariant of f(x,y) under the transformation M.

In this case, D is a function of x and y. To emphasize this point, put

detD = φ(x,y)

and

detD'= φ'(x',y')

so that

φ(x,y) = (detM)2φ'(x',y'

= (aox + a 1y)(a2x + a 3y) - (a1x + a 2y)2

= (aoa2 - a12)x2 + (aoa3 - a1a2)xy + (a1a3 - a22)y2

= xTEx,

where

(a oa2 - a12 ) (aoa3 - a1a2)/2

E= .
(a oa3 - a1a2)/2 (a1a3 - a22 )

Also, we have

φ'(x',y') = x' TE'x'

 = xTMTE'Mx 28

therefore

xTEx = (detM)2xTMTE'Mx

so that

E = (detM)2MTE'M.

Taking determinants, we obtain

detE = (detM)4detE'

= (aoa2 - a12)(a1a3 - a22) - (aoa3 -a1a2)2/4

= invariant of the binary cubic f(x,y) under the transformation

x' = Mx.
 4 29

SOM E INV ARIANTS IN PHYS ICS

4.1 . Gal ilean inva rianc e.

Eve nts o f finite extens ion and du ration are part of t he ph ysical

wor ld. It will be conv enient to intro duce the n otion of ide al ev ents th at

hav e neither exten sion nor d uration. Ideal even ts ma y be repre sented as

mat hemat ical points in a spa ce-time ge ometr y. A part icular eve nt, E, is

described by the f our c ompon ents [t,x, y,z] where t is the time of th e eve nt,

and x,y, z, ar e its thre e spa tial coord inates. The tim e and spac e coo rdinates

are refe rred to ar bitrarily chosen ori gins. The spat ial mesh n eed n ot be

Car tesian.

Let an e vent E[t, x], r ecord ed by an o bserv er O at th e ori gin o f an x-

axis, be reco rded as th e eve nt E'[t ',x'] by a seco nd ob serve r O', movi ng at

con stant spee d V a long the x -axis. We supp ose t hat t heir clocks are

syn chron ized at t = t' = 0 w hen t hey c oincide at a co mmon origin, x = x' =

0.

At time t, we writ e the plausible equa tions

t' = t

and

x' = x - Vt,

whe re Vt is t he distance tra velled by O' in a ti me t. The se eq uations ca n

be writt en

E' = GE
whe re 30

1 0
G = .
-V 1

G is the opera tor o f the Galilean trans forma tion.

The inverse e quations a re

t = t'

and

x = x ' + V t'

or

E = G-1E'

whe re G-1 is the inverse Ga lilean ope rator . (It und oes t he effect of G).

If we mu ltiply t a nd t' by t he co nstan ts k and k ', re spectively, whe re

k a nd k' have dime nsions of velocity t hen a ll terms h ave d imensions of

length.

In space-space, we have the Pytha gorea n for m x 2 + y2 = r2, a n

invariant und er ro tations. We are the refor e led to a sk th e que stion: is

(kt )2 + x2 in variant un der t he op erato r G in spac e-time? C alculation gives

(kt)2 + x2 = (k't ')2 + x' 2 + 2Vx't ' + V 2t' 2

= (k't' )2 + x' 2 o nly if V = 0.

We see, there fore, that Galilean space-time is n ot Py thago rean in its

algebraic for m. We not e, ho wever , the key role played by acc eleration in

Galilean-Newtonian phys ics:

 The velocities of the e vents acco rding to O and O' ar e obt ained by 31

differentiating th e equ ation x' = -Vt + x w ith r espect to time, giving

v' = -V + v,

a p lausible r esult, based up on ou r exp erience.

Differentiating v' with resp ect t o tim e gives

d v'/dt ' = a ' = d v/dt = a

whe re a and a ' are the accelerations in the two frames of refer ence. The

classical acceleration is inv arian t un der t he Ga lilean tra nsformation. I f the

relationship v' = v − V is used to desc ribe the m otion of a pulse of light,

mov ing in emp ty sp ace a t v = c ≅ 3 x 10 8 m/ s, it does not fit the facts. All

stu dies of ve ry hi gh sp eed p articles that e mit e lectromagn etic radiation

sho w tha t v' = c f or all va lues of th e rel ative spee d, V.

4.2 . Lor entz invar iance and Einst ein's spac e-tim e

sym metry .

It was E instein, above all others , who adv anced our under stand ing o f

the true natu re of spac e-time and rela tive motion. We sha ll see tha t he

mad e use of a symm etry argum ent t o find the chan ges t hat m ust b e mad e

to the G alilean tr ansformati on if it is to account fo r the rela tive motion of

rap idly moving obj ects and o f beams of light. H e rec ognized an

inconsistency in t he Ga lilean-Newtonian equ ations, based a s the y are , on

eve ryday expe rience. H ere, we sh all restrict th e discussion to non-

accelerating, or s o called inertial, frames

 We have seen that the c lassical equations r elating th e eve nts E an d 32

E' a re E' = GE, a nd th e inverse E = G-1E'

whe re

1 0 1 0
-1
G = an d G = .
−V 1 V 1

These eq uations ar e con necte d by the s ubstitution V ↔ −V; this is an

algebraic sta temen t of the N ewton ian principle of relativity. Einstein

incorpor ated this principle in his the ory. He a lso r etained th e lin earit y of

the classical equa tions in t he ab sence of a ny ev idence to the c ontra ry.

(Eq uispaced interv als of time and distance in on e inertial fram e rem ain

equ ispaced in any other iner tial frame). H e the refor e symmetri zed th e

space-time eq uations as follows:

t' 1 -V t
= .
x' -V 1 x

Not e, ho wever , the inconsistency in th e dim ensions of the time-equat ion

tha t has now been intro duced :

t' = t − Vx.

The term Vx h as dimensions o f [L] 2/[T], an d not [T]. Thi s can be

cor recte d by intro ducing the invariant spee d of light, c  a postu late in

Ein stein's th eory that is consistent w ith e xperiment:

ct' = ct − Vx/c

so that all terms now h ave d imensions of length.

 Ein stein went furt her, and introd uced a dim ensionless quan tity γ 33

instead of th e scaling factor of unity that appe ars in the Galilean equat ions

of space-time. Th is factor must be co nsistent w ith a ll observa tions. Th e

equ ations the n bec ome

ct ' = γct − βγx

x ' = - βγct + γx, where β=V/c.

These can be writt en

E' = LE,

whe re

γ - βγ
L = , and E = [ct,x ]
- βγ γ

L is the opera tor o f the Lore ntz t ransformat ion.

The inverse e quation is

E = L-1E'

whe re
γ βγ
-1
L = .
βγ γ

This is the i nvers e Lor entz trans forma tion, obta ined from L by chan ging

β → −β (o r ,V → −V); it h as th e effect o f und oing the t ransformat ion L.

We can t herefore w rite

LL-1 = I

or
 γ - βγ γ βγ 1 0 34
= .
-βγ γ βγ γ 0 1

Equ ating elements gives

γ2 − β2γ2 = 1

the refor e,

γ = 1/√(1 − β2) (taking the positive root) .

4.3 . The inva riant inte rval.

Pre viously, it was show n tha t the spac e-time of Galileo an d New ton

is not P ythag orean in f orm. We n ow as k the ques tion: is E insteinian spac e-

time Pyt hagor ean in for m? D irect calculation leads t o

( ct)2 + (x)2 = γ2(1 + β2)(ct')2 + 4βγ 2x'c t'

+γ2(1 + β2)x' 2

≠ (c t')2 + (x')2 if β > 0.

Not e, ho wever , tha t the dif ference of squa res is an

invariant und er L:

(ct)2 − (x )2 = (ct')2 − (x ')2

bec ause

γ2(1 − β2) = 1.

Spa ce-time is said to b e pse udo-E uclidean.

The nega tive sign that chara cterizes L orent z invariance ca n be

included in t he th eory in a gener al way as follows.

We intro duce two k inds of 4-vecto rs

 xµ = [x0, x 1, x 2, x 3], a con trava riant ve ctor, 35

and

xµ = [x0, x 1, x 2, x 3], a cov arian t ve ctor, wher e

xµ = [x0,-x1,-x2,-x3].

The scalar pr oduct of t he ve ctors is d efined as

x µT xµ = (x0, x 1, x 2, x 3)[x 0,-x1,-x2,-x3]

= (x 0)2 − ((x1)2 + (x2)2 + (x3)2)

The even t 4-v ector is

E µ = [ct, x, y, z] a nd th e cov ariant for m is

E µ = [ct,-x,-y,-z]

so that the L orent z invariant sca lar p roduc t is

E µT E µ = (ct)2 − (x 2 + y2 + z2).

The 4-ve ctor xµ tr ansforms a s follows:

x' µ = Lxµ

whe re

γ - βγ 0 0
- βγ γ 0 0
L= .
0 0 1 0
0 0 0 1

This is the o perat or of the Loren tz tr ansformati on if the motion of O' is

along th e x-axis of O's fram e of refer ence.

Imp ortan t con sequences of th e Lor entz trans forma tion are t hat

intervals of time measured i n two different iner tial frames are not the s ame

but are related by the equat ion

 ∆t' = γ∆t 36

whe re ∆t i s an interval m easur ed on a cl ock a t res t in O's f rame, and

distance s are given by

∆l' = ∆l/γ

whe re ∆l is a l ength meas ured on a ruler at r est in O's fram e.

4.4 . The ener gy-mo mentu m inv arian t.

A d ifferential time int erval, dt, cann ot be used in a Lore ntz-invariant

way in k inematics. We must use t he pr oper time differential interva l, dτ,

defined by

( cdt)2 - dx2 = (cdt' )2 - dx' 2 ≡ (c dτ)2.

The Newt onian 3-ve locity is

vN = [dx/d t, dy /dt, dz/dt],

and this must be r eplaced by the 4-velocity

Vµ = [d(ct )/dτ, d x/dτ, d y/dτ, d z/dτ]

= [ d(ct)/dt, dx/dt, dy/dt, d z/dt]dt/dτ

= [γc,γvN] .

The scalar pr oduct is t hen

Vµ Vµ = (γc)2 - (γvN)2

= (γc)2(1 - (vN/c)2)

= c 2.

(In form ing t he sc alar produ ct, t he tr anspose is unde rstoo d).

The magn itude of t he 4- velocity is Vµ  = c, th e invariant spe ed of light.

 In Classical Mechanics, the concept of mome ntum is importa nt be cause 37

of its r ole as an invariant in an isolated system. We the refor e int roduc e the

con cept of 4-momen tum i n Relativistic Mechanics in or der t o find

pos sible Lore ntz i nvariants involving this new q uantity. The c ontra variant

4-m oment um is defined a s:

Pµ = mVµ

whe re m is the mas s of the p article. (It is a Lo rentz scalar, t he ma ss

mea sured in t he fr ame in whi ch th e par ticle is a t res t).

The scalar pr oduct is

P µ Pµ = (mc)2.

Now ,

Pµ = [mγc, mγvN]

the refor e,

P µ Pµ = (mγc)2 − (m γvN)2.

Writing

M = γm, the r elativistic mas s, we obta in

P µ Pµ = (Mc)2 − (MvN)2 = (mc)2.

Multiplying t hroug hout by c2 gi ves

M 2c4 − M2vN2c2 = m2c4.

The quan tity Mc2 ha s dimensions of ener gy; w e the refor e wri te

E = Mc2

the tota l energy o f a f reely movi ng pa rticle.

This leads to the fun damen tal i nvari ant o f dyn amics 38

c2Pµ Pµ = E 2 − (p c)2 = E o2

whe re

E o = mc2 is the rest energ y of the p article, and

p is its rel ativi stic 3- momen tum.

The tota l energy c an be writ ten:

E = γE o = E o + T,

whe re

T = Eo(γ − 1) ,

the rela tivistic k inetic ene rgy.

The magn itude of t he 4- momen tum i s a L orent z invariant

Pµ  = mc.

The 4- m oment um tr ansforms a s follows:

P'µ = LPµ .

For rela tive motion along th e x-axis, this equat ion is equ ivalent to the

equ ations

E' = γE − βγcpx

and

cpx = -βγE + γcpx .

Usi ng th e Planck-E instein equations E = hν an d

E = p xc f or ph otons , the ener gy eq uation bec omes

ν' = γν − βγν
 = γν(1 − β) 39

= ν(1 − β)/(1 − β2)1/2

= ν[(1 − β)/(1 + β)]1/2 .

This is the r elativistic Dop pler shift for the f reque ncy ν', measured i n an

inertial fram e (pr imed) in t erms of th e fre quenc y ν me asure d in anoth er

inertial fram e (un primed).

4.5 . The freq uency -wavenumbe r inv arian t

Par ticle-Wave duality, one o f the most prof ound

discover ies in Phy sics, has its origins in Loren tz in variance. It w as

pro posed by d eBrog lie in the earl y 192 0's. He u sed t he fo llowing

arg ument .

The displacement o f a w ave c an be writ ten

y(t ,r) = Acos (ωt - k•r)

whe re ω = 2πν (t he an gular freq uency ), k = 2π/λ (t he wa venum ber),

and r = [x, y , z] (the position v ector ). The ph ase (ωt − k•r) c an be

written ((ω/c)ct − k•r), and t his has th e for m of a Lor entz invariant

obt ained from the 4-vectors

E µ [ct , r], and K µ [ω/c, k]

whe re Eµ is the event 4-ve ctor, and Kµ is the "freq uency -wavenumbe r" 4-

vec tor.

deB roglie not ed th at th e 4-m oment um P µ is conn ected to t he ev ent 4 -

vec tor E µ th rough the 4-velocity Vµ , a nd th e fre quenc y-wavenumb er 4-

vec tor K µ is conn ected to t he ev ent 4 -vector Eµ th rough the Loren tz

invariant pha se of a wa ve ((ω/c)ct − k r). He t herefore p ropos ed th at a 40

dir ect c onnec tion must exist betw een P µ an d Kµ ; it is illustrat ed

in the f ollowing d iagram:

E µ [ct ,r]

( Einstein) Pµ Pµ =inv. E µ Kµ =inv. (deBrog lie)

P µ [E/ c,p] Kµ [ω/c,k]

(deBroglie)

Fig . 2. The coup ling between P µ an d Kµ vi a E µ .

deB roglie pro posed that the conn ection is the

simplest poss ible, name ly, Pµ an d Kµ ar e pro porti onal to ea ch ot her. He

rea lized that ther e cou ld be only one value for the c onsta nt of

pro porti onality if the Planck-Einstein resu lt for pho tons E = h ω/2π is but a

special case of a gener al re sult  it must be h /2π, w here h is Planck’s

con stant . Th erefore, d eBrog lie propos ed th at

Pµ ∝ K µ

or

Pµ = (h/2π)Kµ .

Equ ating the elements o f the 4-ve ctors gives

E = (h/2π)ω

and

p = (h/2π)k .
In these rema rkable equ ations, ou r not ions of pa rticles an d wav es ar e 41

for ever merge d. T he sm allness of the value of P lanck's co nstan t pre vents

us from obser ving the d uality dir ectly; how ever, it is clearly obser ved a t

the mole cular, ato mic, nuclear, a nd pa rticle level.

4.6 . deB rogli e's invari ant.

The invariant form ed fr om th e fre quenc y-wavenumb er 4- vecto r is

K µ Kµ = (ω/c, k)[ω/c,-k]

= (ω/c)2 − k 2 = (ωo/c)2, w here ωo is the prope r

ang ular frequ ency.

Thi s invariant is the w ave v ersion of Einstein's

ene rgy-m oment um in variant; it giv es th e dispersion re lation

ωo2 = ω2 − (k c)2.

The rati o ω/k is the pha se ve locity of the w ave, vφ.

For a wa ve-packet, the group velocity vG is dω/dk; it can b e obt ained by

differentiating th e dispersion eq uation as follows:

ωdω − kc 2dk = 0

the refor e,

v G = dω/dk = kc 2/ω.

The deBr oglie invariant invo lving the produ ct of the phase and group

velocity is t herefore

vφvG = (ω/k).(kc2/ω) = c2.

Thi s is the w ave-equivalent of Einstein's f amous

 E = M c2. 42

We see t hat

v φvG = c2 = E/M

or,

v G = E/Mvφ = Ek/Mω = p/M = v N, t he pa rticle

velocity.

This res ult p layed an i mport ant p art i n the deve lopment of Wave

Mechanics.

We shall find in later chapt ers, that Loren tz tr ansformati ons form a

gro up, a nd th at in variance p rinciples are r elated dir ectly to s ymmet ry

tra nsformations an d the ir associated g roups .

 5 43

GROUPS — CONCRETE AND ABSTRACT

5.1 Some concrete examples

The elements of the set {±1, ±i}, where i = √-1, are the roots of the

equation x4 = 1, the “fourth roots of unity”. They have the following special

properties:

1. The product of any two elements of the set (including the same two

elements) is always an element of the set. (The elements obey closure).

2. The order of combining pairs in the triple product of any elements

of the set does not matter. (The elements obey associativity).

3. A unique element of the set exists such that the product of any

element of the set and the unique element (called the identity) is equal to the

element itself. (An identity element exists).

4. For each element of the set, a corresponding element exists such

that the product of the element and its corresponding element (called the

inverse) is equal to the identity. (An inverse element exists).

The set of elements {±1, ±i} with these four properties is said to form

a GROUP.

In this example, the law of composition of the group is multiplication; this

need not be the case. For example, the set of integers Z = {.., -2, -1, 0, 1, 2,

...} forms a group if the law of composition is addition. In this group, the

identity element is zero, and the inverse of each integer is the integer with the

same magnitude but with opposite sign.

 In a different vein, we consider the set of 4×4 matrices: 44

1000 0001 0010 0100

{M} = 0 1 0 0 , 1 0 0 0 , 0 0 0 1 , 0 0 1 0 .
0010 0100 1000 0001
0001 0010 0100 1000

If the law of composition is matrix multiplication , then {M} is found to obey:

1 --closure

and

2 --associativity,

and to contain:

3 --an identity, diag(1, 1, 1, 1),

and

4 --inverses.

The set {M} forms a group under matrix multilication.

5.2. Abstract groups

The examples given above illustrate the generality of the group

concept. In the first example, the group elements are real and imaginary

numbers, in the second, they are positive and negative integers, and in the

third, they are matrices that represent linear operators (see later discussion).

Cayley, in the mid-19th century, first emphasized this generality, and he

introduced the concept of an abstract group, Gn which is a collection of n

distinct elements (...gi...) for which a law of composition is given. If n is finite,

the group is said to be a group of order n. The collection of elements must

obey the four rules:

1. If gi, gj ∈ G then gn = gj•gi ∈ G ∀ g i, gj ∈ G (closure) 45

2. g k (gjgi) = (gk gj)gi [leaving out the composition symbol•] (associativity)

3. ∃ e ∈ G such that gie = eg i = gi ∀ g i ∈ G (an identity exists)

4. If gi ∈ G then ∃ g i-1 ∈ G such that gi-1gi = gigi-1 = e (an inverse exists).

For finite groups, the group structure is given by listing all

compositions of pairs of elements in a group table, as follows:

e . gi gj . ←(1st symbol, or operation, in pair)

e . . . .
. . . . .
g i . . gigi g igj .
gj . gjgi g jgj .
g k . gkgi g kgj .
.
.

If gjgi = gigj ∀ g i, gj ∈ G, then G is said to be a commutative or abelian

group. The group table of an abelian group is symmetric under reflection in

the diagonal.
A group of elements that has the same structure as an abstract group is

a realization of the group.

5.3 The dihedral group, D3

The set of operations that leaves an equilateral triangle invariant under

rotations in the plane about its center, and under reflections in the three

planes through the vertices, perpendicular to the opposite sides, forms a

group of six elements. A study of the structure of this group (called the 46

dihedral group, D3) illustrates the typical group-theoretical approach.

The geometric operations that leave the triangle invariant are:

Rotations about the z-axis (anticlockwise rotations are positive)

Rz (0) (123) → (123) = e, the identity

Rz (2π/3)(123) → (312) = a

Rz (4π/3)(123) → (231) = a 2

and reflections in the planes I, II, and III:

RI (123) → (123) = b

RII (123) → (321) = c

RIII (123) → (213) = d

This set of operators is D3 = {e, a, a 2, b, c, d}.

Positive rotations are in an anticlockwise sense and the inverse rotations are in

a clockwise sense., so that the inverse of e, a, a2 are

e-1 = e, a -1 = a 2, and (a2)-1 = a.

The inverses of the reflection operators are the operators themselves:

b-1 = b, c -1 = c, and d-1 = d.

We therefore see that the set D3 forms a group. The group

multiplication table is:

e a a2 b c d
e e a a2 b c d
a a a2 e d b c
a2 a 2 e a c d b
b b c d e a a2
c c d b a2 e a
d d b c a a2 e
In reading the table, we follow the rule that the first operation is written on 47

the right: for example, ca2 = b. A feature of the group D3 is that it can be

subdivided into sets of either rotations involving {e, a, a2} or reflections

involving {b, c, d}. The set {e, a, a2} forms a group called the cyclic group

of order three, C 3. A group is cyclic if all the elements of the group are

powers of a single element. The cyclic group of order n, Cn , is

Cn = {e, a, a 2, a3, .....,an-1 },

where n is the smallest integer such that an = e, the identity. Since

ak an-k = a n = e,

an inverse an-k exists. All cyclic groups are abelian.

The group D 3 can be broken down into a part that is a group C3, and a

part that is the product of one of the remaining elements and the elements of

C3. For example, we can write

D3 = C 3 + bC 3 , b ∉ C3

= {e, a, a2} + {b, ba, ba2}

= {e, a, a2} + {b, c, d}

= cC3 = dC 3.

This decomposition is a special case of an important theorem known as

Lagrange’s theorem. (Lagrange had considered permutations of roots of

equations before Cauchy and Galois).

5.4 Lagrange’s theorem

 The order m of a subgroup H m of a finite group Gn of order n is a 48

factor (an integral divisor) of n.

Let

Gn = {g1=e, g2, g3, ...gn } be a group of order n, and let

Hm = {h1=e, h2, h3, ...hm } be a subgroup of Gn of order m.

If we take any element gk of Gn which is not in Hm , we can form the set of

elements

{gkh1, gkh2, gkh3, ...gk hm } ≡ g k Hm .

This is called the left coset of Hm with respect to g k . We note the important

facts that all the elements of gk hj, j=1 to m are distinct, and that none of the

elements gk hj belongs to Hm .

Every element gk that belongs to G n but does not belong to Hm

belongs to some coset gk Hm so that G n forms the union of Hm and a number

of distinct (non-overlapping) cosets. (There are (n - m) such distinct cosets).

Each coset has m different elements and therefore the order n of Gn is

divisible by m, hence n = Km, where the integer K is called the index of the

subgroup Hm under the group Gn . We therefore write

Gn = g1 Hm + gj2 Hm + gk3 Hm + ....gνKHm

where
 gj2 ∈ Gn ∉ Hm , 49

gk3 ∈ Gn ∉ Hm , gj2 Hm

gnK ∈ Gn ∉ Hm , gj2 Hm , gk3 Hm , ...gn-1 , K-1 Hm .

The subscripts 2, 3, 4, ..K are the indices of the group.

As an example, consider the permutations of three objects 1, 2, 3 ( the

group S3) and let Hm = C 3 = {123, 312, 231}, the cyclic group of order

three. The elements of S3 that are not in H3 are {132, 213, 321}. Choosing

gk = 132, we obtain

gk H3 = {132, 321, 213},

and therefore

S 3 = C 3 + gk2 C3 ,K = 2.

This is the result obtained in the decomposition of the group D3 , if we make

the substitutions e = 123, a = 312, a2 = 231, b = 132, c = 321, and d = 213.

The groups D3 and S 3 are said to be isomorphic. Isomorphic groups have

the same group multiplication table. Isomorphism is a special case of

homomorphism that involves a many-to-one correspondence.

5.5 Conjugate classes and invariant subgroups
 If there exists an element v ∈ Gn such that two elements a, b ∈ Gn are 50

related by vav-1 = b, then b is said to be conjugate to a. A finite group can

be separated into sets that are conjugate to each other.

The class of Gn is defined as the set of conjugates of an element a ∈

Gn . The element itself belongs to this set. If a is conjugate to b, the class

conjugate to a and the class conjugate to b are the same. If a is not conjugate

to b, these classes have no common elements. Gn can be decomposed into

classes because each element of Gn belongs to a class.

An element of Gn that commutes with all elements of Gn forms a class

by itself.

The elements of an abelian group are such that

bab-1 = a for all a, b ∈ Gn ,

so that

ba = ab.
If Hm is a subgroup of Gn, we can form the set

{aea-1, ah2a-1, ....ahma-1} = aHm a-1 where a ∈ Gn .

Now, aHm a-1 is another subgroup of Hm in Gn . Different subgroups may be

found by choosing different elements a of Gn. If, for all values of a ∈ Gn

aHm a-1 = H m
(all conjugate subgroups of Hm in Gn are identical to Hm ), 51

then Hm is said to be an invariant subgroup in Gn .

Alternatively, Hm is an invariant in Gn if the left- and right-cosets

formed with any a ∈ Gn are equal, i. e. ahi = hk a.

An invariant subgroup Hm of Gn commutes with all elements of Gn .

Furthermore, if hi ∈ Hm then all elements ahia-1 ∈ Hm so that H m is an

invariant subgroup of Gn if it contains elements of Gn in complete classes.

Every group Gn contains two trivial invariant subgroups, Hm = G n and

Hm = e. A group with no proper (non-trivail) invariant subgroups is said to

be simple (atomic). If none of the proper invariant subgroups of a group is

abelian, the group is said to be semisimple.

An invariant subgroup Hm and its cosets form a group under

multiplication called the factor group (written Gn /Hm ) of Hm in Gn .

These formal aspects of Group Theory can be illustrated by considering

the following example:

The group D 3 = {e, a, a 2, b, c, d} ~ S3 = {123, 312, 231, 132, 321, 213}.

C3 is a subgroup of S3 : C3 = H 3 = {e, a, a 2} = {123, 312, 231}.

Now,

bH3 = {132, 321, 213} = H3b

 cH3 = {321, 213, 132} = H3c 52

and

dH3 = {213,132, 321} = H3d.

Since H3 is a proper invariant subgroup of S3, we see that S 3 is not simple.

H3 is abelian therefore S3 is not semisimple.

The decomposition of S3 is

S3 = H 3 + bH 3 = H 3 + H 3b.

and, in this case we have

H 3b = H 3c = H 3d.

(Since the index of H3 is 2, H3 must be invariant).

The conjugate classes are

e=e

eae-1 = ea = a

aaa-1 = ae = a

a 2a(a2)-1 = a2a2 = a

bab-1 = bab = a2

cac-1 = cac = a2

dad-1 = dad = a2

The class conjugate to a is therefore {a, a2}.

The class conjugate to b is found to be {b, c, d}. The group S3 can be

decomposed by classes:

S 3 = {e} + {a, a2} + {b, c, d}.

S3 contains three conjugate classes.

 If we now consider Hm = {e, b} an abelian subgroup, we find 53

aHm = {a,d}, H m a = {a.c},

a2Hm = {a2,c}, Hm a2 = {a2, d}, etc.

All left and right cosets are not equal: Hm = {e, b} is therefore not an

invariant subgroup of S3. We can therefore write

S 3 = {e, b} + {a, d} + {a2, c}

= Hm + aHm + a2Hm .

Applying Lagrange’s theorem to S 3 gives the orders of the possible

subgroups: they are

order 1: {e}

order 2: {e, d}; {e, c}: {e, d}

order 3: {e, a, a2} (abelian and invariant)

order 6: S 3 .

5.6 Permutations
A permutation of the set {1, 2, 3, ....,n} of n distinct elements is an

ordered arrangement of the n elements. If the order is changed then the

permutation is changed. the number of permutations of n distinct elements is

n!

We begin with a familiar example: the permutations of three distinct

objects labelled 1, 2, 3. There are six possible arrangements; they are

123, 312, 231, 132, 321, 213.

These arrangements can be written conveniently in matrix form: 54

123 123 123

π1 = , π2 = , π3 = ,
123 312 231

123 123 123

π4 = , π5 = , π6 = .
132 321 213

The product of two permutations is the result of performing one arrangement

after another. We then find

π2π3 = π1

and

π3π2 = π1

whereas

π4π5 = π3

and

π5π4 = π2.

The permutations π1, π2, π3 commute in pairs (they correspond to the

rotations of the dihedral group) whereas the permutations do not commute

(they correspond to the reflections).

A general product of permutations can be written

s 1 s2 . . .s n 1 2 . . n 1 2 . . n
= .
t1 t2 . . .tn s 1 s 2 . . sn t 1 t 2 . . tn

The permutations are found to have the following properties:

1. The product of two permutations of the set {1, 2, 3, ...} is itself a

permutation of the set. (Closure)

2. The product obeys associativity: 55

(πkπj)πi = πk(πjπi), (not generally commutative).

3. An identity permutation exists.

4. An inverse permutation exists:

s 1 s 2 . . . sn
π-1 = | |
1 2 . . . n

such that ππ -1 = π-1π = identity permutation.

The set of permutations therefore forms a group

5.7 Cayley’s theorem:

Every finite group is isomorphic to a certain permutation group.

Let G n ={g 1, g2, g3, . . .g n} be a finite group of order n. We choose any

element gi in Gn, and we form the products that belong to Gn:

gig1, gig2, gig3, . . . gign.

These products are the n-elements of Gn rearranged. The permutation πi,

associated with gi is therefore

g1 g2 . . gn
πi = .
g ig1 gig2 . . gign

If the permutation πj associated with gj is

g1 g2 . . gn
πj = ,
g jg1 g jg2 . . gjgn

where gi ≠ g j, then

g1 g2 . . gn
πjπi = .
(g jgi)gi (gjgi)g2 . . (gjgi)gn
This is the permutation that corresponds to the element gjgi of Gn. 56

There is a direct correspondence between the elements of Gn and the n-

permutations {π1, π2, . . .πn}. The group of permutations is a subgroup of

the full symmetric group of order n! that contains all the permutations of the

elements g1, g2, . . g n.

Cayley’s theorem is important not only in the theory of finite groups

but also in those quantum systems in which the indistinguishability of the

fundamental particles means that certain quantities must be invariant under

the exchange or permutation of the particles.

 6 57

LIE’S DIFFERENTIAL EQUATION, INFINITESIMAL ROTATIONS

AND ANGULAR MOMENTUM OPERATORS

Although the field of continuous transformation groups (Lie groups)

has its origin in the theory of differential equations, we shall introduce the

subject using geometrical ideas.

6.1 Coordinate and vector rotations

A 3-vector v = [vx, vy, vz ] transforms into v´ = [v x´, vy´, vz ´] under a

general coordinate rotation R about the origin of an orthogonal coordinate

system as follows:

v´ = R v,

where

i.i´ j.i´ k.i´

R = i.j´ j.j´ k.j´
i.k´ j.k´ k.k´

cosθii´ . .
= cosθij´ . . 
cosθik´ . cosθkk´

where i, j, k, i´, j´, k´ are orthogonal unit vectors, along the axes, before and

after the transformation, and the cosθii´’s are direction cosines.

The simplest case involves rotations in the x-y plane:

v  x´ =  cosθ ii´ cosθji´ vx 

v y´ cosθij´ cosθjj´ vy
 =  cosφ sinφ  = Rc (φ)v 58
-sinφ cosφ

where Rc (φ) is the coordinate rotation operator. If the vector is rotated in a

fixed coordinate system, we have φ → −φ so that

v´ = Rv(φ)v,
where

Rv(φ) =  cosφ -sinφ  .

sinφ cosφ

6.2 Lie’s differential equation

The main features of Lie’s Theory of Continuous Transformation

Groups can best be introduced by discussing the properties of the rotation

operator Rv(φ) when the angle of rotation is an infinitesimal. In general,

Rv(φ) transforms a point P[x, y] in the plane into a “new” point P´[x´, y´]:

P´ = Rv(φ)P. Let the angle of rotation be sufficiently small for us to put

cos(φ) ≅ 1 and sin(φ) ≅ δφ, in which case, we have

Rv(δφ) = 1 −δφ 
δφ 1

and

x´ = x.1 - yδφ = x - yδφ

y´ = xδφ + y.1 = xδφ + y

Let the corresponding changes x → x´ and y → y´ be written 59

x´ = x + δx and y´ = y +δy

so that

δx = -yδφ and δy = xδφ.

We note that

Rv(δφ) = 1 0 + 0 -1  δφ
0 1 1 0

= I + iδφ
where
i= 0 -1  = Rv(π/2).
1 0

Lie introduced another important way to interpret the operator

i = Rv(π/2), that involves the derivative of R v(φ) evaluated at the identity

value of the parameter, φ = 0:

dRv(φ)/dφ = −sinφ −cosφ  = 0 -1  = i

φ =0
cosφ −sinφ 1 0
φ=0

so that

Rv(δφ) = I + dRv(φ)/dφ  .δφ,

φ=0

a quantity that differs from the identity I by a term that involves the

infinitesimal, δφ: this is an infinitesimal transformation.

Lie was concerned with Differential Equations and not Geometry. He

was therefore motivated to discover the key equation

 dRv(φ)/dφ =  0 -1 cosφ -sinφ  60
1 0 sinφ cosφ

= iRv(φ) .
This is Lie’s differential equation.

Integrating between φ = 0 and φ = φ, we obtain

Rv(φ) φ
∫ dR (φ)/R (φ) = i ∫ dφ
v v
I 0
so that
ln(Rv(φ)/I) = iφ,
or
Rv(φ) = Ieiφ , the solution of Lie’s equation.

Previously, we obtained

R v(φ) = Icosφ + isinφ.

We have, therefore
Ieiφ = Icosφ + isinφ .

This is an independent proof of the famous Cotes-Euler equation.

We introduce an operator of the form

O = g(x, y, ∂/∂x, ∂/∂y),

and ask the question: does

δx = Of(x, y; δφ) ?

Lie answered the question in the affirmative; he found

δx = O(xδφ) = (x∂/∂y − y∂/∂x)xδφ = −yδφ

and
 δy = O(yδφ) = (x∂/∂y − y∂/∂x)y∂φ = xδφ . 61

Putting x = x1 and y = x2, we obtain

δxi = Xxiδφ , i = 1, 2

where

X = O = (x 1∂/∂x2 − x 2∂/∂x1), the “generator of rotations” in the plane.

6.3 Exponentiation of infinitesimal rotations

We have seen that

Rv(φ) = eiφ,

and therefore

Rv(δφ) = I + iδφ, for an infinitesimal rotation, δφ

Performing two infinitesimal rotations in succession, we have

Rv2(δφ) = (I + iδφ)2

= I + 2iδφ to first order,

= Rv(2δφ).

Applying Rv(δφ) n-times gives

Rvn(δφ) = Rv(nδφ) = einδφ = e iφ

= Rv(φ) (as n → ∞ and δφ → 0, the

product nδφ → φ).

This result agrees, as it should, with the exact solution of Lie’s differential

equation.
 A finite rotation can be built up by exponentiation of infinitesimal 62

rotations, each one being close to the identity. In general, this approach has

the advantage that the infinitesimal form of a transformation can often be

found in a straightforward way, whereas the finite form is often intractable.

6.4 Infinitesimal rotations and angular momentum operators

In Classical Mechanics, the angular momentum of a mass m, moving in

the plane about the origin of a cartesian reference frame with a momentum p

is

Lz = r × p = rpsinφnz

where nz is a unit vector normal to the plane, and φ is the angle between r

and p.

In component form, we have

L z cl = xpy − yp x, where px and p y are the cartesian

components of p.

The transition between Classical and Quantum Mechanics is made by

replacing

p x by −i(h/2π)∂/∂x (a differential operator)

and p y by −i(h/2π)∂/∂y (a differential operator),where h

is Planck’s constant.

We can therefore write the quantum operator as

Lz Q = −i(h/2π)(x∂/∂y − y∂/∂x) = −i(h/2π)X

and therefore
 X = iLz Q/(h/2π), 63

and

δxi = Xxi δφ = (2πiLz Q/h)xi δφ, i = 1,2.

Let an arbitrary, continuous, differentiable function f(x, y) be

transformed under the infinitesimal changes

x´ = x - yδφ

and

y´ = y + xδφ .

Using Taylor’s theorem, we can write

f(x´, y´) = f(x + δx, y + δy)

= f(x − yδφ, y + xδφ)

= f(x, y) + ((∂f/∂x)δx + ((∂f/∂y)δy)

= f(x, y) + δφ(−y(∂/∂x) + x(∂/∂y))f(x, y)

= I + 2πiδφL z /h)f(x, y)

= e2πiδφLz/h f(x, y)

= Rv(2πL z δφ/h) f(x, y).

The invatriance of length under rotations follows at once from this result:

If f(x, y) = x2 + y2 then
∂f/∂x = 2x and ∂f/∂y = 2y, and therefore

f(x´, y´) = f(x, y) + 2xδx + 2yδy

= f(x, y) - 2x(yδφ) + 2y(xδφ)

= f(x, y) = x2 + y2 = invariant.
This is the only form that leads to the invariance of length under rotations. 64

6.5 3-dimensional rotations

Consider three successive counterclockwise rotations about the x, y´,

and z´´ axes through angles µ, θ, and φ, respectively:

z
z′ y′
µ about x
y y

x x, x′

z′ y′ z′′ y′, y′′

θ about y´

x′ x′′ x′
z′′ z′′′
y′′ y′′′
φ about z´´

x′′ x′′ x′′′

The total transformation is

Rc (µ, θ, φ) = Rc (φ)Rc (θ)Rc (µ)

cosφcosθ cosφsinθsinµ + sinφcosµ −cosφsinθcosµ + sinφsinµ

= −sinφcosθ −sinφsinθsinµ + cosφcosµ sinφsinθcosµ + sinφsinµ 
sinθ −cosθsinµ cosθcosµ

For infinitesimal rotations, the total rotation matrix is, to 1st-order in the δ’s:

1 δφ −δθ
Rc (δµ, δθ, δφ) = −δφ 1 δµ  .
δθ −δµ 1
The infinitesimal form can be written as follows: 65

1 δφ 0 1 0 −δθ 1 0 0
Rc (δµ, δθ, δφ) = −δφ 1 0 0 1 0 0 1 δµ
0 0 1 δθ 0 1 0 −δµ 1

=  I + Y3δφ  I + Y2δθ  I + Y1δµ

where

0 0 0 0 0 -1 0 1 0
Y1 =  0 0 1 , Y2 =  0 0 0 , Y3 =  -1 0 0 .
0 -1 0 1 0 0 0 0 0
To 1st-order in the δ’s, we have

Rc (δµ, δθ, δφ) = I + Y1δµ + Y2δθ + Y3δφ .

6.6 Algebra of the angular momentum operators

The algebraic properties of the Y’s are important. For example, we find

that their commutators are:

0 0 0 0 0 -1 0 0 -1 0 0 0
[Y1, Y2] = | 0 0 1|| 0 0 0| − | 0 0 0|| 0 0 1|
0 -1 0 1 0 0 1 0 0 0 -1 0

= −Y3 ,

[Y1, Y3] = Y2 ,

and

[Y2, Y3] = -Y1 .

These relations define the algebra of the Y’s. In general, we have

[Yj, Yk] = ± Yl = εjkl Yl ,

where εjkl is the anti-symmetric Levi-Civita symbol. It is equal to +1 if jkl is 66

an even permutation, -1 if jkl is an odd permutation, and it is equal to zero if

two indices are the same.

Motivated by the relationship between Lz and X in 2-dimensions, we

introduce the operators

Jk = -i(2π/h)Yk , k = 1, 2, 3.

Their commutators are obtained from those of the Y’s, for example

[Y1, Y2] = -Y3 → [2πiJ1/h, 2πiJ2/h] = -2πiJ3/h

or

-[J1, J2](2π/h)2 = -2πiJ3/h

and therefore

[J1, J2] = ihJ3/2π .

These operators obey the general commutation relation

[Jj, Jk] = ihεjkl Jl /2π .

The angular momentum operators form a “Lie Algebra”.

The basic algebraic properties of the angular momentum operators in

Quantum Mechanics stem directly from this relation.

Another approach involves the use of the differential operators in 3-

dimensions. A point P[x, y, z] transforms under an infinitesimal rotation of

the coordinates as follows

P´[x´, y´, z´] = Rc (δµ, δθ, δφ]P[x, y, z]

Substituting the infinitesimal form of Rc in this equation gives

 δx = x´ - x = yδφ - zδθ 67

δy = y´ - y = -xδφ + zδµ

δz = z´ - z = xδθ - yδµ .

Introducing the classical angular momentum operators: Licl, we find that

these small changes can be written

3
δxi = ∑ δαk Xkxi
k=1

For example, if i = 1

δx1 = δx = δµ(z∂/∂y - y∂/∂z)x

+ δθ(-z∂/∂x + x∂/∂z)x

+ δφ(y∂/∂x - x∂/∂y)x = -zδθ + yδφ .

Extending Lie’s method to three dimensions, the infinitesimal form

of the rotation operator is readily shown to be

3
Rc (δµ, δθ, δφ) = I + ∑ (∂Rc /∂αi)| ⋅ δαi .
i= 1 All αi’s = 0
 7 68

LIE’S CO NTINUOUS T RANSF ORMATION G ROUPS

In the p revious ch apter , we discussed the p roper ties of infinitesimal

rot ations in 2- an d 3-d imensions, and we fo und t hat t hey a re re lated

dir ectly to t he an gular mome ntum opera tors of Qu antum Mech anics.

Imp ortan t al gebra ic pr opert ies of the matr ix re prese ntations o f the

ope rator s also wer e int roduc ed. In th is chapter , we shall cons ider the

sub ject in general term s.

Let x i, i = 1 to n be a set o f n v ariables. They may be co nsidered t o

be the c oordi nates of a poin t in an n- dimensional vector s pace, Vn. A set

of equat ions involving the x i’s is obtained by the t ransformat ions

x i´ = f i(x1, x 2, . ..xn: a1, a 2, . ...ar), i = 1 to n

in which the set a 1, a 2, . ..ar co ntains r-indepe ndent para meter s. T he se t Ta ,

of trans forma tions maps x → x´ . We shall write

x´ = f(x; a) or x´ = T a x

for the set o f functions.

It is assumed that the functions fi ar e differentiable with res pect to

the x’s and t he a’ s to any r equired or der. These fun ctions nec essarily

dep end o n the essential para meter s, a. Thi s mea ns th at no two

tra nsformations with di fferent nu mbers of p arame ters are t he sa me. r is

the smallest numbe r req uired to c harac terize the tran sformation,

com pletely.

The set of fu nctio ns fi fo rms a fini te co ntinu ous g roup if:

1. The result of two s uccessive trans forma tions x → x´ → x´ ´ is equivalent 69

to a single t ransformat ion x → x´ ´:

x´ = f( x´; b) = f(f(x; a); b)

= f(x ; c)

= f(x ; χ(a; b))

whe re c is the set of p arame ters

cλ = χλ (a ; b) , λ = 1 to r,

and

2. To e very trans forma tion there corr esponds a unique inverse that

belongs to th e set :

∃ a such that x = f (x´; a) = f(x´; a)

We have

Ta Ta -1 = T a -1Ta = I, the i dentity.

We shall see that 1) is a hi ghly restr ictive req uirement.

The tran sformation x = f(x; a0) is the identity. Without loss of

gen erality, w e can take a0 = 0. T he es sential point o f Lie’s th eory of

con tinuous tr ansformati on gr oups is to cons ider that part of th e gro up th at

is close to t he identity, an d not to c onsider th e gro up as a wh ole.

Suc cessive infinitesimal changes can b e use d to build up t he finite chang e.

7.1 One- param eter group s

Con sider the trans forma tion x → x´ unde r a f inite chan ge in a si ngle

par amete r a, and t hen a chan ge x´ + dx ´. There are t wo pa ths f rom x →

x´ + dx´ ; the y are as s hown:

 x´ 70
an “infinitesimal”
δa
a , a f inite para meter chan ge

x´ + dx´
a + da
a “di fferential”
x (a = 0)

We have

x ´ + d x´ = f(x; a + d a)

= f (f(x; a); δa) = f(x´; δa)

The 1st- order Taylor ex pansion is

dx´ = ∂f(x´; a)/∂a δa ≡ u( x´) δa

a= 0

The lie group cond itions the n dem and

a + da = χ(a; δa).

But

χ(a; 0) = a, (b = 0)

the refor e

a + da = a + ∂χ(a; b)/∂ b δa
b= 0

so that
d a = ∂ χ(a; b)/∂ b δa
b= 0

or

δa = A(a)da.

Therefor e

dx ´ = u (x´)A(a)da,

leading to
 dx ´/u(x´) = A(a )da 71

so that

∫ dx´/u(x´) = ∫A(a)da ≡ s, (s = 0 → th e identity).

x´ a

x 0

We there fore obtain

U (x´) - U(x) = s .

A t ransformat ion o f coo rdinates (new v ariables) there fore trans fers all

elements of t he gr oup b y the same tran sformation: a o ne-pa rameter group

is equiv alent to a grou p of translatio ns.

Two cont inuous tra nsformation gro ups a re sa id to be s imilar whe n

the y can be o btained fr om on e ano ther by a chang e of variable. For

exa mple, cons ider the g roup defined by

x 1´ a 0 x1
x 2´ = 0 a 2 x2

The identity coprr esponds to a = 1. T he infinitesimal tra nsformation is

the refor e

x 1´ ( 1 + δa) 0 x1
x 2´ = 0 (1 + δa) x2 .
2

To 1st-order in δa w e hav e

x 1´ = x1 + x1δa

and

x 2´ = x2 + 2x2δa

or

δx1 = x 1δa
and

δx2 = 2x 2δa. 72

In the l imit, thes e equ ations give

dx 1/x1 = dx 2/2x2 = da.

These ar e the different ial equations t hat c orres pond to th e infinitesimal

equ ations abo ve.

Int egrat ing, we ha ve

∫ dx /x ∫ da ∫ dx /2x
x1´ a x2´ a

1 1 = and 2 2 = da ,
x1 0 x2 0

so that

lnx 1´ - ln x1 = a = ln (x1´/x1)

and

ln (x2´/x2) = 2a = 2ln(x1´/x1)

or

U´ = (x2´/x1´2) = U = ( x2/x12) .

Put ting V = lnx1, w e obt ain

V´ = V + a and U ´ = U, th e tra nslat ion g roup.

7.2 Det ermin ation of t he fi nite equat ions from the i nfini tesim al

for ms

Let the finite equ ations of a one -parameter grou p G(1) be

x 1´ = φ(x1, x 2 ; a)

and

x 2´ = ψ(x1, x 2 ; a),
and let the i dentity co rresp ond t o a = 0. 73

We consider t he tr ansformati on of f(x1, x 2) t o f(x1´, x2´). We expan d

f(x1´, x2´) in a Maclaurin series in the p arame ter a (at definite v alues of x 1

and x 2):

f(x 1´, x2´) = f (0) + f´ (0)a + f ´´(0)a2/2! + ...

whe re

f(0) = f(x1´, x2´)| a=0 = f(x 1, x 2),

and

f´(0) = (df(x1´, x2´)/da| a=0

= {(∂f/∂x1´)(dx1´/da) + (∂f/∂x2´)(dx2´/da)}| a= 0

= {(∂f/∂x1´)u(x1´, x2´) + ( ∂f/∂x 2´)v(x1´, x2´)}|a=0

the refor e

f ´(0) = {( u(∂/∂ x1) + v( ∂/∂x2))f}|a=0

= Xf(x1, x 2).
Con tinuing in this way, we h ave

f´´(0) = {d 2f(x1´, x2´)/da2}| a=0 = X2f(x1, x 2), etc....

The func tion f(x1´, x2´) can b e exp anded in t he se ries

f(x1´, x2´) = f (0) + af ´(0) + ( a2/2!)f´´(0) + ...

= f(x1, x 2) + aXf + (a 2/2!)X2f + .. .

Xnf is the symb ol for ope rating n-t imes in succession o f f w ith X.

The finite eq uations of the group are there fore

x 1´ = x1 + aXx1 + (a2/2!)X2x1 + ...

and
x 2´ = x2 + aXx2 + (a2/2!)X2x2 + = ...
If x1 a nd x 2 ar e definite values to which x 1´and x 2´ r educe for the i dentity 74

a=0 , the n the se eq uations ar e the series solutions of the differential

equ ations

dx1´/u(x1´, x2´) = d x2´/v(x1´, x2´) = d a.

The grou p is referred t o as the g roup Xf.

For exam ple, let

Xf = (x1∂/∂ x1 + x 2∂/∂ x2)f

the n

x 1´ = x1 + aXx1 + (a2/2!)X2f ...

= x 1 + a(x 1∂/∂ x1 + x 2∂/∂ x2)x1 + ...

= x 1 + ax1 + (a2/2!)(x1∂/∂ x1 + x 2∂/∂ x2)x1 +

= x 1 + ax 1 + (a2/2!)x1 + ...

= x1(1 + a + a2/2! + ...)

= x 1ea .

Also, we find

x 2´ = x2ea .

Put ting b = e a , w e hav e

x 1´ = bx 1, a nd x 2´ = bx 2.

The fini te gr oup i s the grou p of magni ficat ions.

If X = (x∂/∂ y - y∂/∂x ) we find, for examp le, that t he finite group is t he

gro up of 2-dimensional rotat ions.

7.3 Inv arian t fun ction s of a gro up 75

Let

Xf = (u∂/∂x1 + v∂/∂x2)f define a o ne-paramet er

gro up, a nd let a=0 give the identity. A fu nction F(x 1, x 2) is ter med a n

inv arian t un der t he tr ansformati on gr oup G (1) if

F( x1´, x2´) = F (x1, x 2)

for all values of the p arame ter, a.

The func tion F(x1´, x2´) can b e exp anded as a series in a:

F(x 1´, x2´) = F(x 1, x 2) + aXF + (a 2/2!)X(XF) + . ..

If

F(x1´, x2´) = F (x1, x 2) = in variant fo r all values of a,

it is necessary fo r

XF = 0,

and this mean s tha t

{u( x1, x 2)∂/∂x1 + v(x 1, x 2)∂/∂x2}F = 0 .

Con sequently,

F(x1, x 2) = co nstan t

is a solution of

dx 1/u(x1, x 2) = dx 2/v(x1, x 2) .

This equ ation has one s olution th at de pends on o ne ar bitrary co nstan t, an d

the refor e G(1) ha s only one basic invariant, an d all othe r pos sible invariants

can be g iven in te rms o f the basic invariant.

For exam ple, we no w rec onsider th e the invariants of rotat ions:

 The infinitesimal trans forma tions are given by 76

Xf = (x 1∂/∂ x2 - x 2∂/∂ x1),

and the differential equation tha t giv es th e invariant fun ction F of the

gro up is obta ined by so lving the chara cteristic differential equations

dx 1/x2 = dφ, a nd dx 2/x1 = -dφ,

so that

dx1/x2 + dx 2/x1 = 0.

The solution of th is equation is

x 12 + x 22 = cons tant,

and ther efore the invariant function is

F (x1, x 2) = x12 + x 22.

All func tions of x 12 + x 22 ar e the refor e invariants of the 2-dimensional

rot ation grou p.

This met hod c an be gene ralized. A gro up G(1) in n-va riables de fined

by the e quation

xi´ = φ(x1, x 2, x 3, . ..xn; a), i = 1 to n ,

is equivalent to a uniq ue infinitesimal tra nsformation

Xf = u1(x1, x 2, x 3, . ..xn)∂f/∂x1 + ... un(x1, x 2, x 3, . ..xn)∂f/∂xn .

If a is the g roup param eter then the i nfinitesimal tr ansformati on is

x i´ = xi + u i(x1, x 2, . ..xn)δa (i = 1 t o n),

the n, if E(x1, x 2, . ..xn) is a f unction th at ca n be differentiated n-times with

res pect to it s arg ument s, we have

 E (x1´, x2´, ...xn´) = E (x1, x 2, . ..xn) + aXE + (a 2/2!)X2E + . 77

Let (x1, x 2, . ..xn) b e the coor dinates of a po int in n-space and let a be a

par amete r, in depen dent of th e xi’s. As a var ies, the p oint (x1, x 2, . ..xn) w ill

describe a tr ajectory, start ing from t he initial poin t (x1, x 2, . ..xn). A

nec essary and sufficient con dition tha t F(x 1, x 2, . ..xn) b e an invariant

function is t hat XF = 0. A cur ve F = 0 i s a t rajectory and t herefore a n

invariant cur ve if

XF(x 1, x 2, x 3, . ..xn) = 0.
 8 78

PROPERTIES OF n-VARIABLE, r-PARAMETER LIE GROUPS

The change of an n-variable function F(x) produced by the

infinitesimal transformations associated with r-essential parameters is:

n
dF = ∑ (∂F/∂xi)dxi
i=1

where
r
dx i = ∑ u iλ(x)δaλ , the Lie form.
λ=1

The parameters are independent of the xi’s therefore we can write

r n
dF = ∑ δaλ{∑ u iλ(x)(∂/∂xi)F}
λ= 1 i=1

r
= ∑ δaλ Xλ F
λ=1

where the infinitesimal generators of the group are

n
Xλ ≡ ∑ u iλ(x)(∂/∂xi) , λ= 1 to r.
i=1

The operator
r
I + ∑ Xλδaλ
λ=1

differs infinitesimally from the identity.

The generators Xλ have algebraic properties of basic importance in the

Theory of Lie Groups. The Xλ’s are differential operators. The problem is

therefore one of obtaining the algebraic structure of differential operators.

This problem has its origin in the work of Poisson (1807); he

introduced the following ideas:

The two expressions

X1f = (u 11∂/∂x1 + u12∂/∂x2)f

and
 X2f = (u 21∂/∂x1 + u22∂/∂x2)f 79

where the coefficients uiλ are functions of the variables x1, x2, and f(x1, x2)

is an arbitrary differentiable function of the two variables, are termed

linear differential operators.

The “product” in the order X2 followed by X1 is defined as

X1X2f = (u 11∂/∂x1 + u12∂/∂x2)(u21∂f/∂x1 + u22∂f/∂x2)

The product in the reverse order is defined as

X2X1f = (u 21∂/∂x1 + u22∂/∂x2)(u11∂f/∂x1 + u12∂f/∂x2).

The difference is

X1X2f - X2X1f = X1u21∂f/∂x1 + X1u22∂f/∂x2

- X2u11∂f/∂x1 - X2u12∂f/∂x2.

= (X1u21 - X2u11)∂f/∂x1 + (X1u22 - X2u12)∂f/∂x2

≡ [X1, X2]f.

This quantity is called the Poisson operator or the commutator of the

operators X1f and X2f.

The method can be generalized to include λ = 1 to r essential parameters

and i = 1 to n variables. The ath-linear operator is then

Xa = uia∂f/∂xi
n
= ∑ u ia∂f/∂xi , ( a sum over repeated indices).
i=1

Lie’s differential equations have the form

∂xi/∂aλ = uik(x)Akλ(a) , i = 1 to n, λ = 1 to r.

Lie showed that

(∂ckτσ /∂aρ)uik = 0
in which 80

u jσ∂uiτ /∂xj - ujτ ∂uiσ/∂xj = ckτσ (a)uik(x),

so that the c kτσ ’s are constants. Furthermore, the commutators can be

written

[Xρ, Xσ] = ( ckρσujk)∂/∂xj

= c kρσXk.

The commutators are linear combinations of the Xk’s. (Recall the earlier

discussion of the angular momentum operators and their commutators).

The ckρσ’s are called the structure constants of the group. They have the

properties

c kρσ = -ckσρ ,

cµρσcνµτ + cµστcνµρ + cµτρ cνµσ = 0.

Lie made the remarkable discovery that, given these structure constants,

the functions that satisfy

∂xi/∂aλ = uikAkλ(a) can be found.

(Proofs of all the above important statements, together with proofs of

Lie’s three fundamental theorems, are given in Eisenhart’s

standard work Continuous Groups of Transformations, Dover Publications,

1961).
8.1 The rank of a group

Let A be an operator that is a linear combination of the generators

of a group, Xi:

A = αiXi (sum over i),

and let 81
X = xjXj .

The rank of the group is defined as the minimum number of commuting,

linearly independent operators of the form A.

We therefore require all solutions of

[A, X] = 0.
For example, consider the orthogonal group, O+ (3); here

A = αiXi i = 1 to 3,
and
X = xjXj j = 1 to 3
so that
[A, X] = αixj[Xi, Xj] i, j = 1 to 3

= αixjεijkXk .

The elements of the sets of generators are linearly independent, therefore

αixjεijk = 0 (sum over i, j,, k = 1, 2, 3)

This equation represents the equations

-α2 α1 0 x1 0
α3 0 -α2x2 = 0 .
0 -α3 α2 x3 0

The determinant of is zero, therefore a non-trivial solution of the xj’s

exists. The solution is given by

x j = αj (j = 1, 2, 3)

so that

A = X.

O+ (3) is a group of rank one.

8.2 The Casimir operator of O+ (3)

The generators of the rotation group O + (3) are the operators. Yk’s, 82

discussed previously. They are directly related to the angular momentum

operators, Jk:

Jk = -i(h/2π)Yk (k = 1, 2, 3).

The matrix representations of the Yk’s are

0 0 0 0 0 -1 0 1 0
Y1 =  0 0 1, Y2 =  0 0 0 , Y3 =  -1 0 0 .
0 -1 0 1 0 0 0 0 0
The square of the total angular momentum, J is
3
J2 = ∑ Ji2
1

= (h/2π)2 (Y12 + Y22 + Y32)

= (h/2π)2(-2I).

Schur’s lemma states that an operator that is a constant multiple of I

commutes with all matrix irreps of a group, so that

[Jk, J2] = 0 , k = 1,2 ,3.

The operator J2 with this property is called the Casimir operator of the

group O+ (3).

In general, the set of operators {Ci} in which the elements commute

with the elements of the set of irreps of a given group, forms the set of

Casimir operators of the group. All Casimir operators are constant multiples

of the unit matrix:

Ci = aiI.

The constants ai are characteristic of a particular representation of a group.

 9 83

MATRIX R EPRES ENTATIONS OF GR OUPS

Mat rix r epres entat ions of linear opera tors are i mport ant i n Lin ear

Algebra; we s hall see t hat t hey a re eq ually impo rtant in G roup Theor y.

If a gro up of m × m matri ces

Dn(m) = {D1(m) (g1),...Dk(m) (gk), ...Dn(m) (gn)}

can be f ound in wh ich each e lement is associated with the corre spond ing

element gk of a gr oup o f ord er n

G n = {g1,...gk,....gn},

and the matri ces o bey

Dj(m) (gj)Di(m) (gi) = Dji(m) (gjgi),

and

D1(m) (g1) = I, t he identity,

the n the matr ices Dk(m) (gk) a re sa id to form an m -dimensional

rep resen tation of Gn. If the association is one -to-one we have an

isomorph ism and th e rep resen tation is said to be fai thful .

The subject o f Gro up Re prese ntations forms a ver y lar ge bra nch o f

Gro up Th eory. The re ar e man y sta ndard work s on this topic (see the

bibliography), each one cont aining num erous definitions, lemmas and

the orems . He re, a rath er br ief accoun t is given of s ome o f the more

importan t res ults. The read er sh ould delve into the deeper asp ects of th e

sub ject as th e nee d ari ses. The subject will be intr oduce d by considering
rep resen tations of the rotat ion g roups , and thei r cor respo nding cyclic 84

gro ups.

9.1 The 3-di mensi onal representa tion of ro tatio ns in the plane

The rota tion of a vecto r thr ough an an gle φ in the plane is

cha racte rized by t he 2 x 2 m atrix

c osφ -sinφ
Rv(φ) = .
s inφ cos φ

The grou p of symme try t ransformat ions that leaves an equilatera l

tri angle invariant unde r rot ations in the p lane is of orde r thr ee, and ea ch

element of th e gro up is of d imension t wo

Gn ~ R3(2) = {R(0), R(2π /3), R(4π /3)}

= 1 0 , -1 /2 -√3/2 , -1/2 √3 /2 .
0 1 √3/2 -1/ 2 , -√3 /2 -1 /2

≈ { 123, 312, 231} = C 3.

These ma trices for m a 2 -dimensional re prese ntation of C3 .

A 3 -dimensional re prese ntation of C3 ca n be obtained a s follows:

Con sider an e quilateral tria ngle located in the plane and let t he

coo rdinates o f the thre e ver tices P 1[x, y], P 2[x´ , y´] , and P 3[x´ ´, y´´] be

written as a 3-vector P 13 = [P 1, P 2, P 3], in no rmal order . We intr oduce

3 × 3 matri x ope rator s Di(3) th at ch ange the o rder of th e elements of P 13,

cyc lically. The identity is

P 13 = D1(3) P 13, w here D1(3) = diag(1, 1, 1).

The rear range ment 85

P 13 → P 23[P 3, P 1, P 2] i s given by

P 23 = D2(3) P 13,

whe re

0 0 1
(3)
D2 = 1 0 0 ,
0 1 0

and the rearr angem ent

P 13 → P 33[P 2, P 3, P 1] i s given by

P 33 = D3(3) P 13

whe re

0 1 0
(3)
D3 = 0 0 1 .
1 0 0

The set of ma trices {Di(3) } = {D1(3) , D2(3) , D3(3) } i s said to form a 3-

dim ensio nal r epresentat ion of the original 2-dimensional re prese ntation

{R3(2) }. The elements Di(3) ha ve th e sam e gro up mu ltiplication ta ble as

tha t ass ociated with C3.

9.2 The m-di mensi onal representa tion of sy mmetr y

tra nsfor matio ns in d-di mensi ons

Con sider the case in wh ich a grou p of order n

G n = {g1, g 2, . ..gk, . ..gn}

is repre sented by
 Rn(m) = {R1(m) , R2(m) , . ....Rn(m) 86

whe re

Rn(m) ~ Gn,

and Rk(m) is an m × m matri x rep resen tation of gk. Let P 1d be a ve ctor in

d-dimensional spac e, wr itten in n ormal orde r:

P 1d = [P 1, P 2, . ..P d],
and let

P1m = [P 1d, P 2d, . ...P md]

be an m- vecto r, wr itten in n ormal orde r, in which the comp onent s are each

d-vector s. Introd uce t he m × m matri x ope rator Dk(m) (gk) such t hat

P1m = D1(m) (g1)P1m

P2m = D2(m) (g2)P1m

Pkm = D k(m) (gk)P1m , k = 1 to m , the numb er of

symme try o perat ions,

whe re Pkm is the kth ( cyclic) permuta tion of P1m , and Dk(m) (gk) is called

the “m-d imensional repr esentation of g k”.

 Infinitely ma ny re prese ntations o f a g iven repre sentation can b e 87

found, for, i f S is a ma trix repre sentation, and M is any definite m atrix

with an inverse, we can form T(x) = MS(x)M-1, ∀ x ∈ G. Sin ce

T(xy) = MS(xy)M-1 = MS(x)S(y)M-1 = MS(x)M-1MS(y)M-1

= T(x)T(y),

T is a re prese ntation of G. The n ew re prese ntation simply involves a

cha nge o f var iable in t he co rresp onding sub stitutions. Re prese ntations

related in th e man ner o f S an d T ar e equ ivale nt , and a re no t reg arded as

different rep resen tations. All repres entat ions that are e quivalent to S ar e

equ ivalent to each othe r, an d the y for m an infinite class. Two equivalent

rep resen tations will be writ ten S ~ T.

9.3 Dir ect s ums

If S is a re prese ntation of dime nsion s, a nd T is a re prese ntation of

dimension t o f a g roup G, th e mat rix

S(g) 0
P = , (g ∈ G)
0 T(g)

of dimension s + t is c alled the dir ect s um of the matri ces S(g) and T(g),

written P = S ⊕ T. Therefore, given two repr esentations (th ey ca n be the

sam e), we can obta in a third by a dding them dire ctly. Alt ernat ively, let P

be a rep resen tation of dimension s + t ; we suppo se th at, for al l x ∈ G, the

mat rix P(x) is o f the form

A(x) 0

0 B(x)
whe re A(x) and B(x) are s × s and t × t matri ces, respe ctively. (The 0’s 88

are s × t and t × s zero matri ces). Def ine t he ma trices S an d T as follows:

S(x) ≡ A(x) and T(x) ≡ B(x), ∀ x ∈ G.

Since, b y the grou p pro perty , P(xy) = P(x)P(y),

A(xy) 0 A(x) 0 A(y) 0

=
0 B(xy) 0 B(x) 0 B(y)

A(x)A(y) 0
= .
0 B(x)B(y)

Therefor e, S(xy) = S(x)S(y) and T(xy) = T(x)T(y), so that S an d T ar e

rep resen tations. The r epres entat ion P is said to b e dec ompos able, w ith

com ponen ts S an d T. A rep resen tation is indecompos able if it cann ot be

dec ompos ed.

If a com ponen t of a dec ompos able repre sentation is itself

dec ompos able, we c an co ntinue in this manne r to decom pose any

rep resen tation int o a f inite numb er of indecompo sable comp onent s. (It

sho uld b e not ed th at th e pro perty of i ndeco mposa blity depe nds o n the field

of the r epres entat ion; the r eal field must somet imes be ex tende d to the

com plex field to c heck for indeco mposa bility).

A w eaker form of d ecomp osability arises whe n we consider a

mat rix o f the form

A(x) 0
P(x) =
E(x) B(x)
whe re A(x), and B(x) are matri ces o f dimensions s × s and t × t 89

res pectively and E(x) is a matr ix th at de pends on x , and 0 is the s × t zero

mat rix. The matri x P, a nd an y equ ivalent fo rm, i s said to be reducibl e.

An irreduci ble rep resen tation is one t hat c annot be r educed. E very

dec ompos able matri x is reduc ible (E(x) = 0), where as a reduc ible

rep resen tation nee d not be d ecomp osable.

If S an d T ar e red ucible, we can continue in thi s way to o btain a se t

of irred ucible com ponen ts. The c ompon ents are d eterm ined uniquely, up

to an eq uivalence. The set of distinct irr educible r epres entat ions of a finite

gro up is (in a given field) an inv arian t of the group . Th e com ponen ts fo rm

the bui lding bloc ks of a re prese ntation of a gr oup.

In Physics, dec ompos able re prese ntations a re ge nerally re ferre d to as

reducibl e representatio ns (re ps).

9.4 Sim ilari ty an d uni tary trans forma tions and matri x

dia gonal izati on

Before d iscussing the q uestion of the possibility of reduc ing t he

dimension of a given re prese ntation, it will be useful to consider s ome

importan t res ults in th eTheory of Matr ices. The proo fs of thes e sta temen ts

are given in the s tanda rd wo rks o n Mat rix Theory . (S ee bibliography ).

If there exists a matri x Q su ch th at

Q-1AQ = B ,

the n the matr ices A an d B ar e rel ated by a sim ilari ty tra nsformatio n.
 If Q is unit ary ( QQ† = I: Q† = (Q*)T , t he he rmitian co njugate) 90

the n A an d B ar e rel ated by a uni tary transforma tion.

If A´ = Q-1AQ; B´ = Q-1BQ; C´ = Q-1CQ..then a ny algebra ic

relation amon g A, B, C...is also satisfied by A´, B´, C´ ...

If a similarity tr ansformati on pr oduce s a d iagonal matrix then the

pro cess is called dia gonal izati on.

If A an d B ca n be diagonalized by the same matri x the n A an d B

com mute.

If V is form ed fr om th e eigenvectors of A th en th e similarity

tra nsformation V-1AV wi ll produce a di agona l mat rix w hose elements a re

the eigenvalues of A.

If A is herm itian then V wi ll be unit ary a nd th erefore an herm itian

mat rix c an always be diagona lized by a unit ary t ransformat ion. A re al

sym metri c mat rix c an always be diagona lized by a n ort hogon al

tra nsformation.

9.5 The Schu r-Auerbach theo rem

This the orem states

Eve ry ma trix representa tion of a finit e gro up is equi valen t to a

uni tary matri x rep resentatio n

Let Gn = {D1, D2, . ...Dn} b e a m atrix grou p, an d let D be the matri x

for med b y tak ing t he su m of pairs of e lements

n
D = ∑ DiDi†
i= 1

whe re Di† is the hermitian conjugate of Di.

Since Di is non- singular, each term in th e sum is p ositive de finite. 91

Therefor e D it self is positive definite. Le t Ld be a di agona l mat rix t hat i s

equ ivalent to D, a nd let Ld1/2 be the positive definite ma trix forme d by

rep lacing the elements of Ld by their positive squar e roo ts. Let U be a

unitary matri x wit h the prop erty that

Ld = UDU -1.

In trodu ce th e mat rix

S = Ld-1/ 2U,

th en SDiS-1 is uni tary. (This prop erty can b e dem onstr ated by co nsidering

(SDiS-1)(SDiS-1)†, a nd sh owing that it is equ al to the identity.). S wi ll

tra nsform the orig inal matri x rep resen tation Gn in to di agona l form. E very

unitary matri x is diagonalizable, and there fore every matr ix in ever y finite

mat rix r epres entat ion can be diagonalized.

9.6 Sch ur’s lemma s

A m atrix repr esentation is r educible if every el ement of t he

rep resen tation can be p ut in block-diagonal form by a single similarity

tra nsformation. I nvoking th e res ult o f the prev ious section, w e nee d onl y

discuss unitary re prese ntations.

If Gn = {D(ν)(R)} is an irreduci ble rep resen tation of dimension ν of

a g roup Gn, a nd {D(µ)(R)} is an irreduci ble re prese ntation of dime nsion µ

of the s ame g roup, Gn, a nd if ther e exists a matr ix A su ch th at

D(ν)(R)A = AD(µ)(R) ∀ R ∈ Gn

the n eit her

 i) A = 0 92

or

ii) A is a sq uare non-singular ma trix (so t hat ν = µ)

Let the µ co lumns of A be writ ten c1, c2, . ..cµ, t hen, for a ny ma trices

D(ν) an d D(µ) we have

D(ν)A = (D(ν)c1, D(ν)c2, . ..D(ν)cn)

an
µ µ µ
AD(µ) = ( ∑ D(µ)k1ck, ∑ D(µ)k2ck, . ..∑D(µ)kµck).
k= 1 k= 1 k= 1

the refor e µ
D(ν)cj = ∑ D(µ)kjck
k= 1

and ther efore the µ c-vectors span a sp ace t hat i s invariant und er th e

irr educible set of ν-dimensional matri ces { D(ν)}. The c-vectors are

the refor e the null-vector or th ey sp an a ν-dimensional vecto r spa ce. The

first ca se co rresp onds to A = 0, a nd th e second t o µ ≥ ν an d A ≠ 0.

In the s econd case, the hermi tian conjugates D(ν)1†, . ..D(ν)n† an d D(µ)1†,

...D(µ)n† also ar e irreduci ble . Furt hermo re, s ince D(ν)i(R)A = AD(µ)i(R)

D(µ)i†A† = A†D(ν)i† ,

and ther efore, following the meth od ab ove, we find th at ν ≥ µ. We must

the refor e hav e ν = µ, s o tha t A is squa re.. Sinc e the ν-columns of A sp an

a ν-dimensional space, the matr ix A is nece ssarily no n-singular.

As a cor ollary, a matri x D th at com mutes w ith a n irr educible set of

mat rices must be a scalar ma trix.

9.7 Cha racte rs 93

If D(ν)(R) and D(µ)(R) are related by a si milarity t ransformat ion t hen

D(ν)(R) gives a r epres entat ion o f G t hat i s equ ivalent to D(µ)(R). Th ese t wo

set s of matri ces are ge nerally different, w herea s the ir st ructu re is the same.

We wish, ther efore, to answer the ques tion: what intr insic prop erties of the

mat rix r epres entat ions are i nvariant u nder coord inate tran sformations?

Con sider

∑ [ CD(R)C-1]ii = ∑ CikDkl(R)Cli-1
i ikl

= ∑ δklDkl(R)
kl

= ∑ Dkk(R) , th e tra ce of D(R).

k

We see t hat t he tr ace, or ch aract er, is an invariant under a ch ange of

coo rdinate ax es. We write t he ch aract er as

χ(R) = ∑ Dii(R)
i

Equ ivalent re prese ntations h ave t he sa me se t of chara cters . Th e

ch aract er of R in the repre sentation µ is writ ten

χ(µ)(R) or [ µ; R].

Now , the conjugate elements of G have the f orm S = URU -1, a nd th en

D(R) = D(U)D(R)[D(R)]-1

the refor e

χ(S) = χ(R).

We can d escribe G by gi ving its charac ters in a particular repr esentation;

all elements in a class have the same χ.

 10 94

SOM E LIE GROU PS OF TRAN SFORM ATIONS

We shall cons ider those Lie group s tha t can be d escribed b y a f inite

set of c ontinuously var ying essential param eters a1,...ar:

x i´ = f i(x1,...xn; a1,...ar) = f( x; a) .

A s et of para meter s a e xists that is a ssociated with the i nvers e

tra nsformations:

x = f( x´; a).

These eq uations mu st be solvable to gi ve th e xi’s in te rms o f the x i´’s.

10. 1 Li near group s

The gen eral linea r gro up GL (n) in n-d imensions is given b y the set

of equat ions
n
x i´ = ∑ aijxj , i = 1 to n ,
j= 1

in which det |aij| ≠ 0.

The grou p con tains n 2 pa ramet ers t hat h ave v alues cove ring an infinite

ran ge. The g roup GL(n) is said t o be not clos ed.

All linear gr oups with n > 1 are non-abelian. The group GL(n ) is

isomorph ic to the group of n × n matri ces; the l aw of comp osition is

the refor e mat rix m ultiplication.

The special linea r gro up of tran sformations SL( n) in n-dimensions is

obt ained from GL(n ) by imposing t he co ndition de t|aij| = 1. A fun ctional

relation ther efore exists am ong t he n2- p arame ters so th at th e num ber o f

req uired para meter s is reduc ed to (n2 - 1).

10. 2 Or thogo nal g roups 95

If the t ransformat ions of th e gen eral linear gro up GL (n) are su ch

tha t
n
∑ xi2 → in variant ,
i= 1

the n the rest ricted gro up is called th e ort hogon al gr oup, O (n), in n-

dimensions. There are [n + n(n - 1)/2] con ditions imposed on t he n2

par amete rs of GL(n ), and the refor e the re ar e n(n - 1) /2 essential

par amete rs of O(n).

For exam ple, in th ree d imensions

x´ = Ox ; O ≡ { O3×3: OOT = I, d etO = 1, aij ∈ R}

whe re

a11 a 12 a 13
O = a21 a 22 a 23 .
a 31 a32 a 33

We have

x1´2 +x 2´2 + x3´2 = x 12 +x 22 +x 32 → in variant un der O (3).

This invariance imposes six conditions on t he or iginal nine par amete rs, a nd

the refor e O(3 ) is a thr ee-paramet er gr oup.

10. 3 Un itary grou ps

If the x i’s and t he aij’s of th e gen eral linear gro up GL (n) are

com plex, and the t ransformat ions are r equired to leave xx† in variant in the

com plex space, the n we obtain the unit ary g roup U(n) in n-dimensions:

U( n) ≡ { Un×n: UU† = I, d etU ≠ 0, uij ∈ C}.

There ar e 2n2 in depen dent real param eters ( th e rea l and imag inary part s of

the aij’s), and the unitary co ndition im poses n + n(n - 1) c onditions on

the m so the g roup has n 2 re al paramet ers. The unitary co ndition me ans 96

tha t

∑j |a ij|2 = 1,

and ther efore

|aij|2 ≤ 1 fo r all i, j.

The para meter s are limited t o a f inite rang e of values, and the refor e the

gro up U( n) is said to b e clo sed.

10. 4 Sp ecial unit ary g roups

If we impose the r estriction detU = +1 on the unitary gr oup U (n)

the n we obtain the special unita ry group SU (n) in n-d imensions. We have

SU( n) ≡ { Un×n: UU† = I, d etU = +1, u ij ∈ C}.

The dete rminantal condition reduc es th e num ber o f req uired real

par amete rs to (n2 - 1). We shall see th at th e gro ups S U(2) and S U(3) play

an impor tant part in Modern Physics.

10. 5 Th e gro up SU (2), the i nfini tesim al fo rm of SU(2 ), an d the

Pau li sp in ma trice s

The spec ial unitary gro up in 2-dimensions, SU(2), is defined as

SU (2) ≡ { U2×2: UU† = I, d etU = +1, u ij ∈ C}.

It is a three -parameter grou p.

The defining conditions can be us ed to obta in th e mat rix

rep resen tation in its simplest fo rm; let

a b
U =
c d
whe re a, b, c , d ∈ C. 97

The herm itian conjugate is

a* c*
†
U = ,
b* d*
and ther efore

| a|2 + |b|2 ac* + bd *

†
UU = .
a* c + b *d |c |2 + |d|2
The unit ary c ondition g ives

|a| 2 + |b|2 = |c| 2 + |d|2 = 1,

and the deter minantal condition g ives

ad - bc = 1 .

Solving these equa tions , we obta in

c = - b*, a nd d = a*.

The gene ral form o f SU(2) is ther efore

a b
U= .
-b* a*

We now s tudy the i nfinitesimal form of SU(2 ); it must have the

str uctur e

1 0 δa δb 1 + δa δb
Uinf = + = .
0 1 -δb* δa* -δb* 1 + δa*

The dete rminantal condition there fore gives

d etUinf = (1 + δa)(1 + δa*) +δbδb* = 1 .

To first orde r in the δ’s, we o btain

1 + δa* + δa = 1,
or 98
δa = -δa*.

so that

1 + δa δb
Uinf = .
-δb* 1 - δa

The matr ix element s can be w ritte n in their comp lex forms:

δa = iδα/2 , δb = δβ/2 + iδγ/2.

(The factor o f two has been intro duced for later conv enience).

1 + iδα/2 δβ/2 + iδγ/2

Uinf = .
-δβ/2 + iδγ/2 1 - i δα/2

Now , any 2×2 m atrix can be wr itten as a linear co mbination of th e

mat rices

1 0 0 1 0 -i 1 0
, , , .
0 1 1 0 i 0 0 -1

as follows

a b 1 0 0 1 0 -i 1 0
=A +B +C +D ,
c d 0 1 1 0 i 0 0 -1

whe re
a = A + D, b = B - iC, c = B + iC, and d = A - D.

We then have

a b (a + d) 1 0 (b + c) 0 1 i(b - c) 0 -i (a - d) 1 0
= + + + .
c d 2 0 1 2 1 0 2 i 0 2 0 -1

The infinitesimal form of SU (2) can th erefore be writ ten

 Uinf = I + (i δγ/2) 1 + ( iδβ/2) 2 + (i δα/2) 3 , 99
or
Uinf = I + (i/2)∑ δτj j . j = 1 to 3 .
This is the L ie form.

The ’s are t he Pau li sp in-ma trices:; they are t he ge nerat ors o f the grou p
j

SU( 2):

0 1 0 -i 1 0
1 = , 2 = , 3 = .
1 0 i 0 0 -1

They play a f undam ental role in t he de scription of spin-1/2 par ticles in

Qua ntum mecha nics. (See later dis cussions).

10. 6 Co mmuta tors of th e spi n mat rices and struc ture const ants

We have previously intr oduce d the comm utato rs of the infinitesimal

gen erato rs of a Li e gro up in conn ection wit h the ir Li e Algebra. In this

section, we c onsider th e com mutat ors o f the gene rator s of SU(2); the y are

found to have the symme tric forms

[ 1, 2 ] = 2 i 3, [ 2, 1 ] = -2i 3,

[ 1, 3 ] = -2 i 2, [ 3, 1] = 2 i 2,

[ 2, 3 ] = 2 i 1, [ 3, 2 ] = -2 i 1.

We see t hat t he co mmuta tor o f any pair of t he th ree m atrices g ives a

con stant mult iplied by the v alue of th e rem aining mat rix, thus

[ j, k] = εjkl2i l .

whe re th e qua ntity εjkl = ±1, d epend ing o n the perm utations o f the indices.

(ε(xy )z = +1, ε(yx )z = -1 ..etc...).

The quan tities 2iεjkl ar e the structur e con stant s as sociated w ith t he gr oup. 100

Oth er pr opert ies of the spin matr ices are f ound to be

2 2 2
1 = 2 = 3 = I; 1 2 = i 3, 2 3 = i 1, 3 1 = i 2.

10. 7 Ho momor phism of S U(2) and O + (3)

We can form t he ma trix

P = xT = x j j, j = 1, 2, 3

fro m the matr ices

x = [x1, x 2, x 3] a nd = [ 1, 2, 3]:

the refor e

x3 x 1 - ix2
P = .
x 1 + ix2 -x3

We see t hat

x3 x 1 - ix2
† T
P = (P*) = = P,
x 1 + ix2 -x3

so that P is herm itian.

Fur therm ore,

TrP = 0,
and

detP = -(x 12 + x22 + x32).

Ano ther matri x, P´, can b e for med b y car rying out a similarity

tra nsformation, th us

P´ = UPU †, (U ∈ SU (2)).

A s imilarity trans forma tion leaves bot h the trac e and the deter minant
unc hange d, th erefore 101

T rP = TrP´,
and
d etP = det P´.

How ever, the condition d etP = detP´ means that

xxT = x´x´T,
or
x 12 + x22 + x32 = x 1´2 + x2´2 + x3´2 .

The tran sformation P´ = UPU † is ther efore equivalent to a thr ee-

dim ensio nal or thogo nal t ransformat ion t hat l eaves xxT in variant.

10. 8 Ir reduc ible representa tions of S U(2)

We have seen that the b asic form of th e 2×2 m atrix repr esentation
of

th e gro up SU (2) is

a b
U = , a, b ∈ C; | a|2 + |b| 2 =1 .
-b * a*

Le t the basis vec tors of th is space b e

1 0
x1 = and x2 = .
0 1

We then have
a
x1´ = Ux1 = = ax1 - b*x2 ,
-b *
and
b
x2´ = Ux2 = = bx1 + a*x2 ,
a*
and ther efore
x´ = Utx.

If we wr ite a 2-dimensional vecto r in this complex sp ace a s c = [u, v ]

the n the comp onent s tra nsform und er SU (2) as 102

u ´ = au + bv

and

v ´ = -b*u + a*v ,

and ther efore

c´ = Uc .

We see t hat t he co mpone nts o f the vect or c tr ansform dif ferently

fro m tho se of the basis vect or x — the t ransformat ion m atrices ar e the

tra nsposes of each othe r. T he ve ctor c = [u, v ] in this complex sp ace is

called a spinor (C artan , 191 3).

To find an ir reduc ible repre sentation of SU (2) in a 3 -dimensional

space, w e nee d a s et of thre e linearly independe nt ba sis functions.

Following Wigner ( see b ibliograph y), we can choo se th e polynomials

u2, u v, an d v 2,

and intr oduce the polynomials defined by

1+ m 1- m
j= 1 u v
f =
m
√ { (1 + m)! (1 + m )!}

whe re

j = n/2 (the dimension of th e spa ce is n + 1) .

and

m = j, j - 1, ...-j .

In the p resen t cas e, n = 2, j = 1 , and m = 0, ±1 .

(The factor 1 /√{(1 + m) ! (1 - m)!} is chosen to make the r epres entat ive
mat rix u nitary). 103

We have, ther efore

f11 = u2/√2 , f01 = uv, a nd f-11 = v2/√2 .

A 3 ×3 r epres entat ion o f an element U ∈ SU (2) in thi s space ca n be found

by defining t he tr ansformati on

Ufm1(u, v) = fm1(u´, v´).

We then obtain

Ufm1(u, v) = (au + bv )1 + m(-b*u + a*v)1 - m , m = 0 , ±1,

√{(1 + m) !(1 - m)!}

so that

Uf11(u, v) = (au + bv) 2/√ 2

= (a2u2 + 2abuv + b2v2)/√ 2 ,

Uf01(u, v) = (au + bv) (-b*u + a* v)

= -ab* u2 + (|a|2 - |b|2)uv + a* bv2 ,

and

Uf-11(u, v) = (-b*u + a *v)2/√ 2

= (b* 2u2 - 2a*b* uv + a*2v2)/√ 2 .

We then have

a2 √ 2ab b2 f 11 f11´
-√ 2ab* |a| - |b| √2a*b f 0 = f 01´
2 2 1

b*2 -√2a *b* a* 2 f -11 f-11´

or
UF = F´.

We find that UU † = I an d the refor e U is, indeed, unitary.

This pro cedur e can be g enera lized to a n (n + 1)-dimensional space as

follows

Let
 f mj(u, v) = u j + mvj - m , m = j, j - 1, ...-j. 104
√{(j + m) !(j - m)!}

(Note th at j = n/2 = 1/ 2, 1/1, 3/2, 2/1, ..).

For a gi ven v alue of j, ther e are 2j + 1 linearly ind epend ent p olynomials,

and ther efore we c an fo rm a (2j + 1) × (2 j + 1 ) rep resen tative mat rix o f an

element U of SU(2 ):

Ufmj(u, v) = f mj(u´, v´).

The deta ils of this gen eral case are g iven in Wigner’ s classic text. He

dem onstr ates the i rredu cibility o f the (2j + 1)-dimensional rep resen tation

by showing th at an y mat rix M wh ich commut es with Uj fo r all a, b such

tha t |a| 2 + |b|2 = 1 mus t nec essarily be a con stant ma trix, and there fore, by

Sch ur’s lemma, Uj is an i rredu cible repr esentation.

10. 9 Re presentati ons o f rot ation s and the conce pt of tens ors

We have discussed 2- a nd 3- dimensional rep resen tations of the

ort hogon al gr oup O (3) and th eir c onnec tion to an gular mome ntum

ope rator s. H igher-dimensional re prese ntations o f the orth ogona l gro up ca n

be obtained b y con sidering a 2-index q uantity , Tij — a ten sor — that

con sists of a set of 9 elements t hat t ransform u nder a rot ation of t he

coo rdinates a s follows:

Tij → Tij´ = R ilR jmTlm (sum ov er re peate d ind ices 1, 2, 3).

If Tij = Tji (Tij is symm etric), th en th is symmetr y is an inv arian t un der

rot ations; we have
 Tji´ = R jlR imTlm = R jmR ilTml = R ilR jmTlm = T ij´ . 105

If TrTij = 0, then so is TrTij´, for

Tii´ = R ilR imTlm = (R TR)lmTlm = δlmTlm = T ll = 0.

The comp onent s of a sym metri c tra celess 2-index tenso r con tains 5

mem bers so th at th e tra nsformation Tij → Tij´ = R ilR jmTlm de fines a ne w

rep resen tation of them of dimension 5.

Any tens or Tij ca n be writt en

Tij = (Tij + Tji)/2 + (T ij - Tji)/2 ,

and we h ave

Tij = (Tij + Tji)/2 = (T ij - (δijTll)/3) + ( δijTll)/3 .

The deco mposition of th e ten sor T ij gi ves a ny 2- index tens or in term s of a

sum of a single compone nt, p ropor tional to the i dentity, a set of 3

independ ent q uantities combined in an anti-symme tric tenso r (Tij - Tji)/2,

and a se t of 5 ind epend ent c ompon ents of a symme tric trace less tenso r.

We write the dim ensio nal eq uation

9=1 ⊕3 ⊕5 .

This is as far as it is poss ible to go in t he pr ocess of d ecomp osition: n o

oth er su bsets of 2 -index ten sors can b e fou nd th at pr eserv e the ir identities

und er th e defining tran sformation of t he co ordin ates. Rep resentatio ns wi th

no subsets of tens ors t hat p reserve th eir i denti ties under the defin ing

rot ation s of tenso rs are irr educi ble r epresentat ions.

 We shall see that the d ecomp osition of tens or pr oduct s into 106

sym metri c and anti -symmetric part s is impor tant in th e Qua rk Mo del of

elementa ry pa rticles.

The repr esentations of the o rthog onal group O(3) are found to b e

importan t in defining t he int rinsic spi n of a pa rticle. T he dy namics of a

par ticle of f inite mass can always be descibed in its rest fram e (all inertial

fra mes a re eq uivalent!), and ther efore the particle can be char acter ized by

rot ations. A ll known p articles have d ynamical states that can be de scribed

in terms of t he te nsors of s ome i rredu cible repr esentation of O (3). If t he

dimension of the i rrep is (2j + 1 ) the n the part icle spin is found t o be

pro porti onal to j. In Parti cle Physics, irreps with values of j = 0 , 1, 2,... and

with j = 1/2, 3/2, ... are f ound that corre spond to t he fu ndame ntal boson s

and ferm ions, resp ectively.

The thr ee dimensional ortho gonal grou p SO( 3) (det = +1) a nd th e

two dime nsional gr oup S U(2) have the s ame L ie algebra . In the case of

the grou p SU( 2), t he (2 j + 1 )-dimensional r epres entat ions are a llowed for

bot h int eger and h alf -integer va lues of j, wher eas, the r epres entat ions of

the grou p SO( 3) ar e limited to in teger values of j. Since all the

rep resen tations ar e allowed in SU (2), it is called th e cov ering gr oup. We

not e tha t rot ations thr ough φ a nd φ + 2π ha ve different ef fects on t he 1/2-

integer repre sentations, and ther efore they are (spinor) t ransfomations

associated with SU (2).

 11 107

THE GROU P STR UCTURE OF LORENTZ TRANSFO RMATIONS

The squa re of the invariant interval s, bet ween the o rigin [0, 0, 0, 0]

of a spa cetime coo rdinate sy stem and a n arb itrar y eve nt x µ = [x0, x 1, x 2,

x3] i s, in index not ation

s2 = xµxµ = x´µx´µ , (sum over µ = 0, 1, 2, 3 ).

The lower ind ices can b e rai sed u sing the m etric tens or

ηµν = diag(1, –1 , –1, –1),

so that

s2 = ηµνxµxν = ηµνx´µx´v , (sum over µ an d ν).

The vect ors n ow ha ve co ntrav ariant for ms.

In matri x not ation, the invariant is

s2 = xT x = x´T x´ .

(The tra nspose mus t be writt en ex plicitly).

The prim ed an d unp rimed column ma trices (contrav ariant vec tors) are

related by th e Lor entz matri x ope rator , L

x´ = Lx .

We there fore have

xT x = (Lx)T (Lx)

= xTLT Lx .

The x’s are a rbitr ary, there fore

LT L = .
This is the def ining pr opert y of the L orent z tra nsformations. 108

The set of all Lor entz trans forma tions is t he se t L of all 4 × 4

mat rices that satisfies the defining p roper ty

L = {L: LT L = ; L: all 4 × 4 real matri ces;

= diag(1, –1 , –1, –1}.

(Note th at ea ch L ha s 16 (independe nt) r eal matrix elements, and there fore

belongs to th e 16- dimensional space, R16).

11. 1 Th e gro up st ructu re of L

Con sider the result of two s uccessive Loren tz tr ansformati ons L1

and L2 tha t tra nsform a 4 -vector x as follows

x → x´ → x´´

whe re

x´ = L1x ,

and

x´´ = L2x´.

The resu ltant vect or x´´ is given b y

x´´ = L2(L1x)

= L2L1x

= Lc x

whe re

Lc = L2L1 (L1 fo llowed by L2).

If the c ombined op eration Lc is always a Loren tz tr ansformati on th en it

mus t sat isfy

 Lc T Lc = . 109

We must there fore have

(L2L1)T (L2L1) =

or

L1T(L2T L2)L1 =

so that

L1T L1 = , (L1, L2 ∈ L)

the refor e

Lc = L2L1 ∈ L .

Any numb er of succ essive Lor entz trans forma tions may be ca rried out to

give a r esultant t hat i s itself a Lore ntz t ransformat ion.

If we ta ke th e det erminant o f the defining equat ion o f L,

d et(LT L) = det

we obtain

(detL)2 = 1 (d etL = detLT)

so that

de tL = ±1.

Si nce t he de termi nant of L is not zero, an i nvers e tra nsformation L–1

exists, and t he eq uation L–1L = I, t he identity, is always va lid.

Con sider the inverse of the defining equation

(LT L)–1 = –1
,

or
 L–1 –1
(LT)–1 = –1
. 110
–1
Using = , a nd re arran ging, gives

L–1 (L–1)T = .

This res ult shows that the i nvers e L–1 is always a membe r of the s et L.

We there fore see t hat

1. If L1 an d L2 ∈ L , then L2 L1 ∈ L

2. If L ∈ L , then L–1 ∈ L

3. The identity I = diag(1, 1, 1, 1 ) ∈ L

and

4. The m atrix oper ators L ob ey as sociativity.

The set of all Lor entz trans forma tions ther efore form s a gro up.

11. 2 Th e rot ation grou p, re visit ed

Spa tial rotat ions in tw o and thre e dim ensions ar e Lor entz

tra nsformations in which the time -component rema ins unchan ged.

Let R be a re al 3×3 m atrix that is p art o f a L orent z tra nsformation

with a c onsta nt ti me-compone nt. In th is case, t he de fining pro perty of t he

Lor entz trans forma tions leads to

RTR = I , t he identity mat rix, diag(1,1,1).

This is the d efining pr opert y of a thr ee-dimensional ortho gonal matr ix

If x = [x1, x 2, x 3] i s a t hree-vecto r tha t is trans forme d und er R to

give x´ t hen
 x´Tx´ = xTRTRx 111

= xTx = x12 + x22 + x32

= invariant unde r R.

The action of R on any three -vector pr eserv es length. The set of all 3×3

ort hogon al matrices is denot ed by O(3),

O(3) = {R: RTR = I, r ij ∈ R}.

The elements of th is set sat isfy the f our g roup axioms.

The grou p O(3) can be sp lit into t wo pa rts t hat a re sa id to be

dis conne cted:: one w ith d etR = +1 an d the othe r wit h det R = -1. The

two part s are writ ten

O+ (3) = {R: d etR = +1}

and

O-(3) = {R: d etR = -1} .

If we de fine the par ity o perat or , P, t o be the o perat or th at re flects

all poin ts in a 3- dimensional car tesian sys tem t hroug h the orig in th en

-1 0 0
P = 0 -1 0 .
0 0 -1

The two parts of O(3) are related by the opera tor P:

if R ∈ O+ (3) then PR ∈ O-(3),

and
 if R´ ∈ O-(3) then PR´ ∈ O+ (3). 112

We can t herefore c onsider on ly th at pa rt of O(3) that is a grou p, na mely

O+ (3), tog ether with the opera tor P.

11. 3 Co nnect ed an d dis conne cted parts of t he Lo rentz grou p

We have shown , pre viously, that e very L orent z tra nsformation, L,

has a de termi nant equal to ± 1. T he ma trix elements o f L ch ange

con tinuously as th e rel ative velocity chang es co ntinuously. It is n ot

pos sible, how ever, to m ove c ontinuously in such a way that we c an go

fro m the set of tr ansformati ons w ith d etL = +1 to thos e with det L = -1; we

say that the set { L: d etL = +1} i s dis conne cted fr om th e set {L: d etL = -

1}.

If we wr ite t he Lo rentz tran sformation in its co mpone nt fo rm

L → L µν

whe re µ = 0,1,2,3 labels the r ows, and ν = 0,1,2,3 labels the c olumns the n

the time comp onent L 00 ha s the values

L 00 ≥ +1 or L 00 ≤ -1.

The set of tr ansformati ons c an th erefore be split int o fou r

disconne cted parts , labelled as f ollows:

{ L↑+ } = {L: d etL = +1, L 00 ≥ +1}

{L↑-} = {L: d etL = -1, L 00 ≥ +1}

{L↓+ } = {L: d etL = +1, L 00 ≤ -1},

and
 {L↓-} = {L: d etL = -1, L 00 ≤ -1}. 113

The identity is in {L↑+ }.

11. 4 Pa rity, time -reversal and o rthoc hrono us tr ansfo rmati ons

Two discrete Loren tz tr ansformati ons a re

i) the p arity tran sformation

P = {P: r → -r, t → t}

= di ag(1, -1, -1, -1),

and

ii) the time-rever sal transfprmat ion

T = {T: r → r, t → -t }

= di ag(-1, 1, 1, 1} .

The disconnec ted p arts of {L} a re re lated by t he tr ansformati ons

tha t inv olve P, T, a nd PT, a s sho wn:

PT
↑
L + L↓-

P T

L↑- L↓-

Fig . 3 Con necti ons b etween the disc onnec ted p arts of th e Lor entz

tra nsformatio ns
 The prop er or thoch ronou s tra nsformations ar e in the g roup L↑+ . We 114

see that it is not nece ssary to c onsider th e com plete set {L} o f Lor entz

tra nsformations — we ne ed co nsider onl y tha t sub set { L↑+ } t hat f orms a

gro up by itself, and either P, T, o r PT co mbined. E xperiments have

sho wn clear v iolations under the parity tra nsformation, P an d vio lations

und er T ha ve be en inferre d fro m exp eriment an d the ory, combined.

How ever, not a single experiment has b een c arried out that show s a

violation of the p roper orth ochro nous trans forma tions, {L↑+ }.
 12 115

ISO SPIN

Par ticles can be d istinguished fr om on e ano ther by th eir intrin sic

pro perties: mass, charg e, spin, parity, and thei r ele ctric and magne tic

mom ents. In o ur on -going que st fo r an under stand ing o f the true natu re of

the fund ament al particles, and th eir intera ctions, ot her i ntrin sic proper ties,

with nam es such as “iso spin” and “stra ngene ss”, have been discovered .

The intr insic prop erties are defined b y qua ntum numbe rs; for ex ample, the

qua ntum numbe r a is defined b y the eigenvalue equ ation

Aφ = a φ

whe re A is a linear opera tor, φ is the wavefunction of the system in the

zer o-mom entum fram e, an d a is an e igenvalue of A.

In this chapt er, w e sha ll discuss the first of t hese new p roper ties to

be intro duced , nam ely, iso spin.

The building b locks of n uclei are proto ns (positively charg ed) and

neu trons (neu tral). Nu merou s exp eriments o n the scat tering of proto ns by
pro tons, and proto ns by neut rons, have show n tha t the nuclear f orces

bet ween pairs have the same stren gth, provi ded t he an gular mome ntum

and spin stat es ar e the same . Th ese o bserv ations for m the basis of an

importan t con cept — the cha rge-i ndepe ndenc e of the n ucleo n-nuc leon

for ce. (Corr ections for the coulomb e ffects in proto n-pro ton s catte ring

mus t be made). Th e ori gin o f this con cept is found i n a n ew symmetry

pri ncipl e. In 19 32, C hadwick no t onl y identified th e neu tron in studying

the inte raction of alpha-particles on beryllium nuclei but also show ed th at 116

its mass is a lmost equa l to the m ass of the prot on. (Recent me asure ments

give

m ass of pro ton = 938⋅272 31(28) MeV/c2

and

ma ss of neut ron = 939⋅565 63(28) MeV/c2)

Within a few month s of Chadwick’s discovery , Heisenberg in trodu ced a

the ory o f nuc lear forces in which he c onsidered the n eutro n and the proto n

to be tw o “st ates” of t he sa me ob ject — the nucleon. He i ntrod uced an

intrinsic var iable, later ca lled isospin, that p ermit s the char ge st ates (+, 0 ) of

the nucleons to be distinguished. Thi s new vari able is needed (in addition

to the t raditional spac e-spin var iables) in the descr iption of nucleon-

nuc leon scattering.

In nuclei, pr otons and neutr ons b ehave in a rema rkably sym metri cal

way : the bind ing energy of a nucleus is closely propo rtion al to the numbe r

of neutr ons a nd pr otons , and in light nuclei (mass nu mber <40), the

num ber o f neu trons can be eq ual t o the numb er of prot ons.

Before d iscussing the i sospin of particles and n uclei, it is necessary to

introduc e an ext ended Pa uli Exclusion P rinciple. In i ts or iginal form, th e

Pau li Exclusion Pr inciple was introduc ed to acco unt f or fe ature s in the

obs erved spec tra o f ato ms th at co uld n ot be unde rstoo d using th e the n

cur rent models of atomic str uctur e:

no two e lectrons i n an atom can e xist in th e sam e qua ntum state defi ned 117

by the q uantu m num bers n, , m , m s wh ere n is t he pr incip al qu antum

num ber, is the orbit al an gular mome ntum quant um nu mber, m is the

mag netic quan tum n umber, and ms is the spin quant um nu mber.

For a sy stem of N particles, the complete w avefunction is writt en as

a p roduc t of single-particle wave functions

Ψ(1, 2, . ..N) = ψ(1)ψ(2)...ψ(N).

Con sider this form in t he simplest cas e — f or tw o identical par ticles. L et

one be i n a s tate labelled Ψa an d the othe r in a sta te Ψb. For i dentical

par ticles, it make s no difference to t he pr obability density |Ψ|2 of the 2-

par ticle syst em if the particles are e xchan ged:

|Ψ(1, 2)|2 = |Ψ(2, 1)|2 , (the Ψ’s are n ot me asura ble)

so that, eith er

Ψ(2, 1) = Ψ(1, 2) (symmetric)

or

Ψ(2, 1) = -Ψ(1, 2) (anti-symmetric).

Let

ΨI = ψa (1)ψb(2) (1 an a, 2 in b)

and

ΨII = ψa (2)ψ(1) (2 in a, 1 in b).

The two particles are i ndistinguishable, th erefore we have no w ay of 118

kno wing wheth er ΨI or ΨII de scribes th e system; we po stulate th at th e

system s pends 50% of its time in ΨI an d 50% of i ts time in ΨII . The t wo-

par ticle syst em is cons idered to be a lin ear c ombin ation of ΨI an d ΨII :

We have, ther efore, either

Ψsym m = (1 /√2){ ψa (1)ψb(2) + ψa (2)ψb(1)} (BOS ONS)

or

Ψant isymm = (1/√2 ){ψa (1)ψb(2) - ψa (2)ψb(1)} (FER MIONS) .

(The coe fficient ( 1/√2) norm alizes the sum of th e squ ares to be 1).

Exc hanging 1↔ 2 leaves Ψsym m un chang ed, w herea s exc hanging pa rticles

1↔ 2 rever ses t he sign of Ψant isymm .

If two p articles are in ΨS, b oth p articles can ex ist in the same stat e with

a = b. If two par ticles are in ΨAS , and a = b, we h ave ΨAS = 0 — t hey

can not e xist in th e sam e qua ntum state. El ectro ns (fermions, spin = (1/2)h)

are desc ribed by a nti-symmet ric w avefunctions.

We can n ow in trodu ce a more gener al Pauli Exclusion P rinciple.

Write th e nuc leon wavefunction as a pr oduct :

Ψ(χ, q ) = ψ(χ)φN(q) ,

whe re

χ = χ(r, s )

in which r is the space vect or, s is t he sp in, and q is a charg e or isospin

label.
 For two nucleons, we wr ite 119

Ψ(χ1, q 1; χ2, q 2),

for two proto ns:

Ψ2p = ψ1(χ1, χ2)φN(p1)φN(p2),

for two neutr ons:

Ψ2n = ψ2(χ1, χ2)φN(n1)φN(n2),

and for an n- p pair:

Ψnp = ψ3(χ1, χ2)φN(p1)φN(n2)

or

= ψ4(χ1, χ2)φN(n1)φN(p2).

If we re gard the p roton and neutr on as different stat es of the same object,

labelled by t he “c harge or i sospin coo rdinate”, q, we must exte nd th e Pau li

pri nciple to cover the new c oordi nate: the total wave function is the n

Ψ(χ1, q 1; χ2, q 2) = -Ψ(χ2, q 2; χ1, q 1) .

It must be an ti-sy mmetric un der t he fu ll ex chang e.

For a 2p - or a 2n- pair, the exchange q 1↔ q 2 is symm etrical, and th erefore

the spac e-spin par t mus t be anti-symme trical.

For an n -p pa ir, t he sy mmetr ic (S) and anti -symmetric (AS)

“is ospin” wav efunctions are

I) ΦS = (1/ √2){φN(p1)φN(n2) + φN(n1)φN(p2)}

(symmetr ic under q 1 ↔ q 2),

and ther efore the space-spin part is a nti-symmet rical,

 II) ΦAS = (1/ √2){φN(p1)φN(n2) - φN(n1)φN(p2)} 120

(anti-symmetr ic under q 1 ↔ q 2),

and ther efore the space-spin part is symmet rical.

We shall need thes e res ults in later d iscussions of t he sy mmetr ic and ant i-

sym metri c pro perties of quar k sys tems.

12. 1 Nu clear -decay

Nuc lei are bo und s tates of n eutro ns an d pro tons. If the n umber of

pro tons in a nucleus is Z an d the numb er of neut rons is N then the m ass

num ber o f the nucleus is A = N + Z. S ome n uclei are natur ally unsta ble.

A p ossible mo de of deca y is by th e emission of a n electron ( th is is

β-decay — a pr ocess that typi fies the f undam ental “wea k int eract ion”) .

We write the decay as

A
Z XN → A
Z+1 XN-1 + e–1 + νe ( β–-decay)

or, we c an ha ve
A
Z XN → A
Z-1 XN-1 + e+ + νe ( β+ - decay).

A r elated pro cess is that of electron captu re of an o rbital electron that is

sufficiently close to t he po sitively charge d nuc leus:

e– + A
Z XN → A
Z+1 XN+1 + νe .

Oth er re lated proc esses are

νe + A
Z XN → A
Z-1 XN-1 + e+

and

νe + A
Z XN → A
Z+1 XN-1 + e– .
The deca y of the fre e pr oton has n ot be en ob serve d at the p resen t tim e. 121

The expe rimental limit on th e half-life of the p roton is > 10 31 ye ars! Many

cur rent theor ies of the micr ostru cture of m atter pred ict t hat t he pr oton

dec ays. If, howev er, t he life-time is > 10 32 - 1 033 ye ars t hen t here is no

rea listic pos sibility o f observing the deca y dir ectly (The limit is set b y

Avo gadro ’s nu mber and t he finite numbe r of proto ns th at ca n be

assembled in a suitable expe rimental appara tus).

The fund ament al β-decay is tha t of the f ree n eutro n, first o bserv ed in

194 6. T he pr ocess is

n 0 → p + + e– + νe 0 , t1/2 = 10⋅37 + 0⋅19 minutes.

This mea sured life-time is o f fundamen tal import ance in Pa rticle Phy sics

and in C osmology.

Let us s et up an a lgebraic description of t he β-decay p roces s, re cognizing

tha t we have a 2-state system in which the trans forma tion p ↔ n occur s:

In the β–-decay o f a f ree n eutro n

n → p + + e– + νe ,

and in t he β+ -decay o f a p roton , bou nd in a nu cleus,

p → n + e+ + νe .

12. 2 Is ospin of t he nu cleon

The spon taneo us tr ansformati ons p ↔ n obser ved in β-decay lead u s

to intro duce the o perat ors ± th at tr ansform p ↔ n:

φ = φp ,
+ n φ = 0, (eliminates a p roton )
+ p
and 122

φ = φn ,
- p φ = 0, (eliminates a n eutro n).
- n

Since we are dealing wi th a two-state system, we choo se th e “is ospin”

par ts of the proto n and neut ron w avefunctions to be

1 0
φ(p) = a nd φ(n) = ,
0 1

in which case the opera tors must have the f orms:

0 1 0 0
+ = and - = .
0 0 1 0

They are singular and n on-hermitian.

We have, for examp le

0 1 0 1
+ n φ = = , φn → φp,
0 0 1 0

and

0 1 1 0
φ =
+ p = ( + re moves a pr oton) .
0 0 0 0

To make the p resen t alg ebraic des cription analogous t o the two- state

sy stem of th e int rinsic spin of the e lectron, w e int roduc e linear

com binations of th e ± :

0 1
1 = + + - = = 1, a Paul i mat rix,
1 0

and
 0 -i 123
2 = i( - - + )= = 2, a Paul i mat rix.
i 0

A t hird opera tor t hat i s diagonal is, as ex pecte d

1 0
3 = = 3 , a Paul i mat rix.
0 1

The thre e ope rator s { 1, 2, 3} t herefore o bey t he co mmuta tion

relations

[ j/2, k/2] = i εjkl l/2 ,

whe re th e factor o f(1/2) is intro duced beca use o f the 2:1 homom orphi sm

bet ween SU(2) and O+ (3): the vect or op erato r

t = /2

is called the iso spin operator o f the nucl eon.

To classify the isospin stat es of the nucleon we may use t he

pro jection of t on the 3rd a xis, t3. The e igenvalues, t3, o f t3 co rresp ond t o

the prot on (t 3 = +1/2) and neutr on (t 3 = -1/2) stat es. The n ucleon is said to

be an iso spin doubl et wi th isospin quan tum n umber t = 1/2. (The numb er

of states in the m ultiplet is 2t + 1 = 2 fo r t = 1/2).

The char ge, Q N of the nucleon ca n be writt en in term s of the i sospin

qua ntum numbe rs:

Q N = q(t 3 +( 1/2)) = q or 0,

whe re q is the pro ton c harge . (It is one o f the grea t uns olved prob lems of

Par ticle Phys ics to und ersta nd wh y the char ge on the proto n is equal to t he

cha rge o n the electron) .

12. 3 Is ospin in n uclei . 124

The conc ept o f isospin, and of ro tations in isospin space, associated

with ind ividual nucleons can be a pplied to nuclei — s ystem s of many

nuc leons in a boun d sta te.

Let the isospin of the ith-nucleon be ti, a nd let ti = i /2 . Th e

ope rator of a syst em of A nu cleons is defined as

T = ∑Ai=1 ti = ∑Ai=1 i/2 .

The eigenvalue of T3 of the isospin op erato r T3 is the sum o f the individual

com ponen ts

T3 = ∑Ai=1 t 3i = ∑Ai=1 τ3i/2

= (Z – N)/2 .

The char ge, Q N of a nu cleus can be wr itten

QN = q∑Ai=1 (τ3i + 1)/2

= q( T3 + A/2) .

For a gi ven e igenvalue T of the o perat or T, t he st ate is (2T + 1) -fold

deg enera te. The e igenvalues T3 of T3 are

T3 = −T, −T + 1,...0,...T + 1, T .

If the H amiltonian H of the nucleus is char ge-indepen dent then

[H, T] = 0.

and T is said to b e a g ood q uantu m num ber. In l ight nuclei, where t he

isospin-violating coulomb in terac tion betwe en pa irs o f pro tons is a small

effect, the c oncep t of isospin is part icularly u seful. Th e stu dy of isospin

effects in nu clei was first applied to the obser ved p roper ties of th e lowest-
lying st ates in th e thr ee nu clei with mass numbe r A = 14: 14C, 14N, and 14O. 125

The rela tive energ ies of the stat es ar e sho wn in the following diagram:

Ene rgy ( MeV)

6
0+ T = 1, T3 = 1

0+ T = 1 , T3 = 0

0+ T = 1 , T3 = -1 1+ T = 0 , T3 = 0
0

Fig 4. An isospin si nglet (T = 0) a nd an isos pin t riple t (T = 1) in

th e A = 14 s ystem. In the absence of the coulo mb in teraction, the three

T = 1 s tates woul d be degen erate.

The spin and parity of the g round stat e of 14C, the f irst excited st ate o f 14N

and the groun d sta te of 14O a re me asure d to be 0+ ; t hese three stat es ar e

cha racte rized by T = 1. The grou nd st ate o f 14N h as spin and par ity 1 + ; it

is an isospin singlet (T = 0 ).

12. 4 Is ospin and meson s

We have seen that it is poss ible to cl assify the char ge st ates of

nuc leons and nuclear isobars using the conc ept o f isospin, and the a lgebra

of SU(2). It will be u seful to c lassify ot her p articles, including field

par ticles (quanta) in t erms of th eir isospin.

Yukawa (1935), first pr oposed tha t the stro ng nu clear forc e bet ween

a p air o f nuc leons is c arried by mas sive field particles called mesons.
Yukawa’s meth od wa s a m aster ful development of t he th eory of th e 126

electrom agnet ic field t o include the c ase o f a m assive field particle. I f ψπ is

the “mes on wa vefunction” the n the Yukawa differential equa tion for t he

mes on is

∂µ∂µ ψπ + (E0/hc)2ψπ = 0.

whe re

∂µ∂µ = (1/ c2)∂2/∂t 2 - ∇2 .

The r-de pende nt (s patial) form of ∇2 is

∇2 → (1 /r2)d/dr(r2d/dr)

The stat ic (time-indepe ndent ) solution of t his equation is read ily checked to

be

Ψ(r) = (-g2/r)exp(-r/rN)

whe re

rN = h/mπc = hc/mπc2 = hc/Eπ0,

so that

1 /rN2 = (E π0/hc)2

The “ran ge of the nuclear fo rce” is defined by t he co ndition

r = r N = h/mπc ≈ 2 ×10-13 cm .

This gives th e mas s of the m eson to be close to the m easur ed va lue. It i s

importan t to note that the “ range of t he fo rce” ∝ 1/ (mass of t he field

qua ntum) . In the case of th e electrom agnet ic field, the m ass of the field

qua ntum (the photo n) is zero , and ther efore the force has an infinite ran ge.
 The meso ns co me in thre e cha rge s tates: +, -, and 0. The meson s 127

hav e int rinsic spins eq ual t o zer o (th ey ar e field particles an d the refor e the y

are boso ns), and t heir rest energ ies are me asure d to be

E π±0 = 139 ⋅5 M eV, and E π00 = 135 ⋅6 M eV.

They are ther efore cons idered to be me mbers of a n isospin triplet:

t = 1, t3 = ±1, 0.

In Parti cle Physics, it is t he cu stom to de signate th e isospin quant um

num ber b y I, we sh all follow this conv ention fro m now on.

The thir d com ponen t of the i sospin is an ad ditive qua ntum numbe r.

The comb ined values of the i sospin pro jections o f the two particles, one

with isospin projection I3(1) , and t he ot her w ith I3(2) , is

I 3(1+ 2) = I3(1) + I3(2) .

Their isospins com bine to gi ve st ates with different numbe rs in each

multiplet. F or ex ample, in pion (meson)-nucleon scat tering

π + N → st ates with I3(1 + 2) = (3/2) or ( 1/2).

These va lues are o btained by noti ng th at

Iπ(1) = 1, an d IN(2) = 1/2, so th at

I 3π(1) + I3N(2) = (±1 , 0) + ( ±1/2)

= (3/2), an isospin q uarte t, or (1/2), an

isospin doublet.

Sym bolically , we write

3 ⊗ 2 = 4 ⊕ 2.

(This is the rule for f ormin g the prod uct ( 2I3(1) + 1)⊗(2I3(2) + 1).
 13 128

GRO UPS A ND TH E STR UCTURE OF MATTER

13. 1 St range ness

In the e arly 1950’ s, ou r und ersta nding of t he ultimate str uctur e of

mat ter s eemed to b e com plete. We requ ired neutr ons, proto ns, electrons

and neut rinos, and meso ns an d pho tons. Our opti mism was short- lived.

By 1953, excited s tates of t he nu cleons, and mor e mas sive meson s, had

bee n discover ed. Some of th e new part icles had completely unex pecte d

pro perties; for ex ample, in the i ntera ction betw een p roton s and π-mesons

(pions) the f ollowing d ecay mode was o bserv ed:

Pr oton (p+ )
Sig ma (∑ + ) Pion
(π0 )
❊ Kao n (K+ ) ❊
Pion
(π + )
Pi on (π + )
⇑ ⇑
Initial interaction Fina l decay
lasts ~10 -23 se conds ta kes ~ 10-10 se conds

( Str ong fo rce a cting) (Wea k fo rce a cting)

Gell-Mann, an d ind epend ently Nishijima, pro posed that the kaons (heavy

mes ons) were endow ed with a new intrin sic proper ty no t affected by t he

str ong f orce. Gel l-Mann called t his proper ty “s trang eness”. S trang eness

is cons erved in t he st rong interactions bu t cha nges in th e wea k

interact ions. The Gell-Mann - Nishijima in terpr etation of the stran geness- 129

cha nging invo lved in th e pro ton-p ion intera ction is

p + (S = 0) ∑ + (S = –1 )
π0 (S = 0)

❊ K + (S = +1 ) ❊
π+ (S = 0)

π+ (S = 0)
⇑ ⇑
∆S = 0 ∆S = 1

In the s trong part of t he interac tion, ther e is no ch ange in th e num ber

defining the stran geness, wh ereas in t he we ak pa rt, t he st range ness chang es

by one u nit. Having de fined the values of S for the particles in th is

interact ion, they are d efined for ever. All subs equen t exp eriments involving

the se ob jects have been cons istent wit h the orig inal assignment s.

13. 2 Pa rticl e pat terns

In 1961, Gell-Mann, and independe ntly Ne’em an, introd uced a

scheme t hat c lassified the s trong ly interac ting particles into family gro ups

(Mendeleev ag ain?). Th ey we re no t, at this poin t in the d evelopment of t he

the ory, concerned with the n ature of p ossible building blo cks o f the

nuc leons and meson s. T hey w ere c oncer ned, howev er, w ith t he inclusion

of “stra ngene ss” i n the theo ry, a nd th erefore th ey we re le d to study the

arr angem ents of pa rticles in an abs tract spac e de fined by t heir electric

cha rge a nd st range ness. The comm on fe ature of e ach family was chosen
to be th eir intrin sic spin; the f amily of s pin-1/2 baryons (str ongly 130

interact ing p articles) has e ight membe rs: n 0, p + ,∑± ,∑ 0 ,Ξ– ,Ξ0 , and Λ0 .

Their st range ness quant um nu mbers are:

S = 0: n 0, p + ; S = – 1: ∑± ,∑ 0 , and Λ0 ; and S = –2 : Ξ0,– . (No n egatively-

cha rged proto n exists, and n o pos itively-charged Ξ ex ists). If the positions

of these eight par ticles are given in charg e-strangen ess space, a re marka ble

pat tern emerg es:

Str angen ess

0 +
n p ⇓
0

Λ0
∑– ∑+ –1
∑ 0

+1

–2
Ξ –
Ξ 0

Cha rge –1

Fig . 5. The fami ly po rtrait of spin- 1/2 b aryons in

cha rge- strangeness spa ce

There ar e two part icles at t he ce nter, each with zero char ge an d zer o

str angen ess; they are t he ∑0 an d the Λ0. ( They have different rest masses).
 They stu died the s truct ure o f oth er fa milies. A part icularly 131

importan t set of p articles consists of all baryo ns with sp in 3/2. A t the time ,

the re we re ni ne kn own p articles in thi s cat egory : ∆ 0, ∆ ±1, ∆ +2 , ∑ *0, ∑ *±1,

Ξ0, a nd Ξ-1 . They have the following patte rn in char ge-strange ness space:

Cha rge: – 1 0 +1 +2 Str angen ess

⇓
0
∆- ∆0 ∆+ ∆ ++

–1
∑*– ∑ *0 ∑+

–2
Ξ* –
Ξ* 0

–3
Ω –

The symm etry patte rn of the family of spin-3/2 b aryon s, shown b y the

kno wn ni ne ob jects was sufficiently compelling for Ge ll-Mann, in 196 2, to

sug gest that a ten th me mber of th e fam ily should exist. F urthe rmore , if

the symm etry has a phys ical basis, the tent h mem ber s hould have spin-3/2,

cha rge – 1, st range ness –3, a nd it s mas s sho uld b e abo ut 15 0MeV great er

tha n the mass of t he Ξ0 pa rticle. T wo ye ars a fter this suggestion, the tent h

mem ber o f the family was identified in high ener gy pa rticle collisions; it

dec ayed via weak intera ctions, and pos sessed the pred icted prop erties.

This cou ld no t hav e bee n by chance. T he discovery of the Ω– pa rticle was

cru cial in helping to e stablish the co ncept of t he Ge ll-Mann – Ne’em an 132

sym metry mode l.

In addition t o the symm etries of baryo ns, g roupe d by their spins, th e

mod el was used to obtain sym metri es of meso ns, also groupe d by their

spins.

13. 3 Th e spe cial unita ry gr oup S U(3) and p artic le st ructu re

Sev eral years befo re th e wor k of Gell-Mann and N e’ema n, Sa kata

had atte mpted to b uild-up th e kno wn pa rticles fr om {n eutro n- pr oton-

lambda0} t riplets. The lambda part icle was r equired to “car ry th e

str angen ess”. Alt hough the model was shown not to be valid, Ikeda e t al.

(1959) introd uced an im porta nt ma thema tical analysis of th e thr ee-state

system t hat i nvolved th e gro up SU (3). The notion tha t an under lying

gro up st ructu re of elementar y par ticles might ex ist was po pular in t he

ear ly 19 60’s. (Sp ecial Unit ary G roups were used by J . P. Elliott in the

late1950 ’s to desc ribe symme try p roper ties of light n uclei).

The prob lem facing Part icle Physicists, at the t ime, was t o find the

app ropri ate g roup and its fu ndame ntal repre sentation, and to co nstru ct

higher-d imensional repr esentations tha t wou ld account for the w ide variety

of symme tries obse rved in charge-stran geness space. We have se en th at

the char ge of a pa rticle can be w ritte n in terms of i ts isospin, a c oncep t tha t

has its origin in the c harge -independe nce o f the nucleon-nucleon for ce.

When app ropri ate, we sh all discuss the symm etry prope rties of p articles in

iso spin- strangeness spa ce.

 Pre viously, we discussed the prop erties of the L ie gr oup S U(2). It is 133

a g roup chara cterized b y its thre e gen erato rs, t he Pa uli spin m atrices.

Two-state systems, such as t he electro n wit h its quan tized spin-up a nd sp in-

dow n, an d the isospin states of n ucleons an d nuc lei, can b e tre ated

qua ntitatively using th is gr oup. The symme tries of n ucleon and meso n

families discovere d by Gell-Mann and N e’ema n, implied an u nderlying

str uctur e of nucleons a nd me sons. It could not be a struc ture simply

associated with a two-state system bec ause the o bserv ed pa rticles we re

end owed not o nly w ith p ositive, n egative, a nd ze ro ch arge but a lso with

str angen ess. A th ree-state system was ther efore cons idered nec essary, at

the very least; th e mos t pro mising can didate was the group SU(3 ). We

shall discuss the infinitesimal form o f this gro up, a nd we shall find a

suitable set of ge nerat ors.

13. 3.1 The a lgebr a of SU(3)

The grou p of special unitary tran sformations in a 3-d imensional

com plex space is d efined as

SU( 3) ≡ {U3×3 : UU† = I, d etU = +1, u ij ∈ C}.

The infinitesimal form of SU (3) is

SU(3 )inf = I + iδαj j/2 , j = 1 to 8.

(There a re n 2 - 1 = 8 gene rator s).

The quan tities δαj ar e rea l and infinitesimal, and the 3 ×3 m atrices j ar e

the linearly independen t gen erato rs of the group . Th e rep eated index, j,

mea ns th at a sum o ver j is t aken.

 The defining prope rties of t he gr oup r estrict th e for m of the 134

gen erato rs. For e xample, th e unitary condition is

UU† = (I + iδαj j/2)(I – iδαj †j/2)

= I – iδαj j†/2 + iδαj j/2 to 1s t-ord er,

†
= I if j = j .

The gene rator s mus t be hermitian.

The dete rminantal condition is

de t = + 1; and the refor e Tr j = 0.

The gene rator s mus t be trace less.

The finite fo rm of U is obta ined by ex ponen tiation:

U = exp{iαj j/2}.

We can find a suitable set o f 8 g enera tors by ex tending th e met hod

used in our d iscussion of isospin, thu s:

Let thre e fun damen tal states of t he sy stem be ch osen in th e simplest

way , nam ely:

1 0 0

u = 0 , v = 1 , an d w = 0 .

0 0 1

If we wish to tran sform v → u, we can d o so by de fining the oper ator A+ :

0 1 0 0 1

A+ v = u, 0 0 0 1 = 0 .

0 0 0 0 0
We can introd uce o ther opera tors that trans form the s tates in p airs, thus 135

0 0 0

A– = 1 0 0 ,

0 0 0

0 0 0 0 0 0

B+ = 0 0 1 , B– = 0 0 0 ,

0 0 0 0 1 0

0 0 0 0 0 1

C+ = 0 0 0 , C– = 0 0 0 .

1 0 0 0 0 0

These ma trices are singular and n on-hermitian. In th e discussion of isospin

and the group SU(2 ), th e non -singular, trac eless, her mitian mat rices 1, a nd

2 ar e for med f rom t he ra ising and lowering o perat ors ± ma trices by

introduc ing t he co mplex linear co mbinations

1 = + + – = 1 an d 2 = i( 1 – 2) = 2.

The gene rator s of SU(3) are forme d fro m the oper ators A±, B±, C± by

con struc ting complex linear combinations. For e xample:

the isospin o perat or 1 = 1 = + + –, a gene rator of S U(2) becom es

0
1
0 = A+ + A– ≡ 1, a gene rator of S U(3).

0 0 0

Con tinuing in this way, we o btain

 A+ = 1/2 + i 2/2 , 136

whe re

0
2
2 = 0 ,

0 0 0

and

C+ + C– = 4 , C+ – C– = –i 5,

B+ + B– = 6 a nd B + – B– = i 7 .

The rema ining gene rator s, 3 an d 8 ar e tra celess, diagona l, 3×3 m atrices:

0 1 0 0

3 = 3 0 , 8 = 0 1 0 .

0 0 0 0 0 -2

The set of ma trices { 1, . .... 8} a re ca lled the G ell-Mann m atrices,

introduc ed in 1961 . Th ey ar e nor malized so that

Tr ( j k ) = 2δjk.

The norm alized for m of 8 is ther efore

1 0 0

8 = (1/ √3) 0 1 0 .

0 0 –2

If we pu t F i = /2. we f ind
i

A± = F 1 ± iF 2 ,

B± = F 6 ± iF 7,
and 137

C± = F 4 + iF 5 .

Let A3 = F 3, B3 = –F 3/2 + (√3 /4)F 8 , and C3 = (–1/2)F 3 - (√3/4)F 8., so th at

A3 + B3 + C3 = 0.

The last cond ition mean s tha t onl y eight of the nine opera tors are

independ ent.

The gene rator s of the g roup are r eadily sho wn to obey the Lie

com mutat ion r elations

[F i, F j] = ifijk F k , i,j,k = 1 to 8.

whe re th e qua ntities fijk ar e the non- zero struc ture const ants of th e gro up;

the y are foun d to obey

fijk = –fjik ,

an d the Jaco bi identity.

The comm utation re lations [F i, F j] c an be writ ten i n ter ms of the opera tors

A±, . ..Some typ ical results ar e

[ A+ , A-] = 2A3, [ A+ , A3] = -A+ , [ A-, A3] = +A-,

[A3, B3] = 0, [A3, C3] = 0, [B3, C3] = 0

[B+ , B-] = 2B3, [ B+ , B3] = -B-, [B-, B3] = +B-, e tc.

The two diagonal operat ors c ommut e:

[F 3, F 8] = 0 .

Now , F 1, F 2, a nd F 3 co ntain the 2×2 i sospin ope rator s (Pauli matrices),

each wit h zer os in the third row and c olumn; the y obe y the comm utation

relations of isospin. We therefore ma ke th e identifications

 F 1 = I1, F 2 = I2, a nd F 3 = I3 138

whe re th e Ij’s are t he co mpone nts o f the isospin.

Par ticles tha t exp erience th e str ong n uclear int eract ion are ca lled

had rons; t hey a re se parat ed into tw o set s: the bar yons, w ith h alf-intege r

spins, and th e mes ons w ith z ero o r int eger spins. Pa rticles th at do not

exp erience th e str ong i ntera ction are called lep tons. In or der t o qua ntify

the differenc e bet ween baryo ns an d leptons, it h as be en fo und n ecessary t o

introduc e the bary on nu mber B = + 1 to denot e a b aryon , B = –1 t o

den ote a n ant i-baryon a nd B = 0 f or al l oth er pa rticles. Lepto ns ar e

cha racte rized by t he lepton numbe r L = +1, anti-leptons ar e assigned L =

–1, and all other particles are a ssigned L = 0. It i s a p resen t-day fact ,

based up on nu merou s obs ervat ions, that the total bary on an d lepton

num ber i n any inte raction is cons erved . Fo r exa mple, in t he de cay o f the

fre e neu tron we find

n 0 = p+ + e– + νe 0

B = +1 = +1 + 0 + 0

L = 0 = 0 + 1 + (–1) .

The fund ament al symmetr ies in Nat ure r esponsible for these cons ervat ion

laws are not known at t his time. These con serva tion laws may, in all

likelihood, b e bro ken.

In discussing the patte rns o f bar yon f amilies in char ge-strange ness

space, w e wish to incorporat e the fact that we a re de aling with bar yons

tha t int eract via the s trong nuclear f orce in wh ich iso spin an d strangen ess
are cons erved . We ther efore choo se to desc ribe their patt erns in isospin- 139

hyp ercha rge s pace, wher e the hype rchar ge Y is defined to i nclude bot h the

str angen ess and th e bar yon a ttrib ute o f the part icle in an additive way:

Y = B + S.

The diagonal opera tor F 8 i s the refor e assumed to be dire ctly associated

with the hype rchar ge op erato r,

F 8 = (√3/2)Y.

Because I3 an d Y co mmute , sta tes c an be chos en th at ar e

simultaneous eigenstates of the o perat ors F 3 an d F 8. Since no o ther SU(3)

ope rator s com mute with I3 an d Y, n o oth er ad ditive qua ntum numbe rs ar e

associated with th e SU( 3) symmetr y. T he op erato rs F 4,...F 8 ar e con sidered

to be ne w con stant s-of-the-motion of t he st rong interaction ham iltonian.

13. 4 Irr educi ble r epresentat ions of SU (3)

In an ea rlier discussion of the i rredu cible repr esentations of SU(2),

we found that the commu tation rel ations of the g enera tors of th e gro up

wer e sat isfied not only by t he fu ndame ntal 2×2 m atrices bu t also by

mat rices of h igher dime nsion [(2J + 1) ⊗ (2 J + 1 )], w here J can have the

values 1 /2, 1, 3/2, 2, ....The J-values cor respo nd to the spin of th e par ticle

who se st ate is given by a sp inor (a column vecto r wit h spe cial

tra nsformation pro perties). In t he 2×2 r epres entat ion, both cov arian t an d

con trava riant sp inors are allowed:

i) cov ariant spinors (with lower ind ices) are writt en as 2-co mpone nt 140

columns that trans form under U ∈ SU (2) as

´ = Uij
i j ,

whe re

a1
= ,
a2

and

ii) con trava riant spinors (with uppe r ind ices) are writt en as

2-compon ent r ows t hat t ransform a s:

j i
´= Uij † ,

whe re

= (b1, b 2).

The co- and c ontra -variant s pinor s are tran sformed with th e aid of t he an ti-
ij
sym metri c ten sors ij an d . For e xample,
j
i = ij

tra nsforms as a co variant sp inor with the f orm

b2
i = .
- b1

The high er-dimensional repre sentations are built up f rom t he fu ndame ntal
j
for m by taking ten sor p roduc ts of the fundamenta l spinors i , , or i

and by s ymmet rizing and anti -symmetrizing t he re sult. We state, without

pro of, t he th eorem that is u sed in thi s met hod:

whe n a t ensor prod uct o f spi nors has b een b roken down into its symmetric
and anti -symm etric part s, it has been decom posed into irre ducib le 141

representatio ns of the SU(n) . (S ee Wigner’ s sta ndard work for the

ori ginal discussion of the m ethod , and de S wart in Rev. Mo d. Ph ys. 35,

(1963) for a detailed discussion of te nsor analysis in the stud y of the i rreps

of SU(n))

As an ex ample, we wr ite t he te nsor produ ct of two covar iant spinors

i an d j in the following way

i⊗ j= i j = ( i j + )/2 + (
j i i j - )/2
j i

There ar e fou r ele ments associated wit h the prod uct ( i,j can ha ve va lues 1

and 2).

The symm etric part of t he pr oduct has three independe nt el ement s,

and tran sforms as an ob ject that has spin J=1. (There are 2J + 1 me mbers

of the s ymmet ric set). The anti-symme tric part has o ne element , and

the refor e tra nsforms as an o bject with spin J = 0. T his r esult is familiar in

the theo ry of angu lar m oment um in Quan tum M echanics. The explicit

for ms of the four elements a re:

J3 = +1: 1 1

J=1 J3 = 0 : (1/√2 )( 1 2 + 2 1 )

J 3 = –1 : 2 1

and

J=0 J 3 = 0 : ( 1/√2)( 1 2 – 2 1).

Higher-d imensional repr esentations are built up from the t ensor prod ucts 142

of covar iant and c ontra variant 3- spinors, an d re spectively. Th e

pro ducts are then writt en in term s of their symm etric and anti-symme tric

par ts in orde r to obtain the irre ducible re prese ntations. For examp le, the
j
pro duct i , i,j = 1,2,3, can be w ritte n
j j
i = ( i - (1/ 3)δij k
k
) + (1/3)δij k
k
,

in which the trace has been separ ated out. The trace is a zero -rank tens or

with a sin gle co mpone nt. The o ther tenso r is a tra celess, symmetr ic tensor

with eig ht in depen dent compo nents . Th e dec ompos ition is w ritte n

sym bolically as:

3 ⊗ 3 = 8 ⊕ 1.

We can form t he te nsor produ ct of two covar iant 3-spinors, i j as

follows:

i j = (1/ 2)( i j + ) + (1/2)(

j i i j – j i ), i,j = 1,2, 3.

Sym bolically, we h ave

3⊗3 = 6⊕3,

in which the symme tric tenso r has six compo nents and the a nti-symmet ric

ten sor h as th ree c ompon ents.

Oth er te nsor produ cts t hat w ill be of interest a re

3 ⊗ 3 ⊗ 3 = 10 ⊕ 8 ⊕ 8 ⊕ 1 ,

and

8 ⊗ 8 = 27 ⊕ 10 ⊕ 10 ⊕ 8 ⊕ 8´ ⊕ 1 .
 The appe aranc e of the o ctet “8” i n the 3 ⊗ 3 de compo sition (re call 143

the obse rved octet of s pin-1/2 baryons ), and the decu plet “10” in th e tri ple

pro duct 3 ⊗ 3 ⊗ 3 de compo sition (re call the o bserv ed de cuplet of spin-3/2

bar yons), was of p rime impor tance in t he de velopment of th e gro up th eory

of “elem entar y” pa rticles.

13. 4.1 Weigh t dia grams

Two of t he Ge ll-Mann ma trices, 3 an d 8, a re di agona l. We can

write th e eigenvalue eq uations:

3 u = αuu, 3 v = αvv, a nd 3 w = αww,

an d

8 u = βuu, 8v = βvv, a nd 8w = β ww ,

wh ere αi an d βi ar e the eigenvalues.

Let a an d b b e nor malization factor s associated with th e ope rator s 3

and 8, r epect ively, so that

a 0 0 b 0 0
N N
3 = 0 – a 0 , an d 8 = 0 b 0 .
0 0 0 0 0 –2 b

If

u = [1, 0 , 0], v = [0, 1 , 0], and w = [0, 0 , 1] (columns), we f ind

N N
3 u = au , 8 u = bu,
N N
3 v = –av , 8 v = bv ,

and
N N
3 w = 0w , 8 w = –2bw.

The wei ght v ectors ar e for med f rom t he pa irs o f eigenvalues:

 [αu, βu] = [a, b], 144

[αv, βv] = [-a, b],

and

[αw, βw] = [0, -2b].

A w eight diagram i s obt ained by p lotting th ese v ector s in the α–β

space, t hus:

β
2b

–2a –a a 2a α
−b

–2 b

This weight d iagram for the fundamenta l “3” repr esentation of S U(3) was

well-known to Math ematicians at t he ti me of the first use of SU (3)

sym metry in P article Ph ysics. It was to pl ay a key r ole in the deve lopment

of the q uark model.

13. 5 Th e 3-q uark model of m atter

Although the octet and decup let p atter ns of hadr ons o f a g iven spin

and pari ty em erge as irreduc ible repre sentations of t he gr oup S U(3),

major pr oblems rem ained that resu lted in a great deal of s cepticism

con cerning th e validity of t he SU (3) model of fundame ntal particles. The

mos t pre ssing prob lem was: why ar e the re no know n par ticles associated 145

with the fund ament al tr iplets 3, 3 of SU(3 ) tha t exh ibit the s ymmet ry of

the weight di agram discussed in t he last se ction? In 1964 , Gell-Mann, an d

independ ently, Zwe ig, propos ed th at th ree f undam ental enti ties do ex ist

tha t cor respo nd to the base states of SU(3), and that they form boun d

sta tes o f the hadr ons. That such enti ties have not b een o bserv ed in the

fre e sta te is simply re lated to t heir enorm ous b inding ene rgy. The three

ent ities were called qua rks by Gell-Mann, and ace s by Zwei g. T he Ge ll-

Mann ter m has surv ived. The anti -quar ks ar e associated with th e

con jugate 3 re prese ntation. The t hree quark s, denoted by u , d, and s (u

and d fo r the up- and d own-isospin sta tes, and s for stran geness) have

highly u nusual pro perties; they a re

Lab el B Y I I3 Q= I3 +Y /2 S = Y − B

u 1/3 1/3 1/2 +1/2 +2 /3 0

d 1/3 1/3 1/2 –1/ 2 –1 /3 0

s 1/3 –2/3 0 0 –1/3 –1

s –1 /3 2/3 0 0 + 1/3 +1

d –1 /3 –1/3 1 /2 +1 /2 + 1/3 0

u –1 /3 –1/3 1/2 –1/ 2 – 2/3 0

The quar ks oc cupy the f ollowing p ositions in I3 - Y spac e

 Y Y 146
s
d u

I3
I3

u d

s
These diagram s hav e the same rela tive forms as t he 3 an d 3 we ight

diagrams of S U(3).

The bary ons a re ma de up of q uark triplets, and t he me sons are m ade

up of th e simplest poss ible struc tures , nam ely quark– anti-quark pair s. T he

cov ariant and cont ravar iant 3-spinors intro duced in t he pr evious section

are now given phys ical significance:

= [u, d , s], a co variant co lumn 3-spinor,

and

= (u, d, s), a co ntrav ariant row 3-spinor.

whe re u = [1, 0, 0 ], d = [0, 1, 0 ], an d s = [0, 0, 1] repr esent the unitary

sym metry part of t he to tal wavefunctions of the three quar ks.

The form al operato rs A±, B±, a nd C±, introd uced in section 13. 3.1,

are now viewed as opera tors that trans form one fla vor (t ype)of qua rk in to

ano ther flavor (th ey ar e shi ft op erato rs):

A± ≡ I±(I3) → I3 ± 1 ,

B± ≡ U±(U3) → U3 ± 1, ca lled the U -spin oper ator,

and
 C± ≡ V±(V3) → V3 ± 1, ca lled the V-spin oper ator. 147

Exp licitly, we hav e

I+ (–1/2) → 1 /2 : d → u

I–(+1 /2) → –1 /2 : u → d

U+ (–1/2) → 1/2 : s → d

U–(+1 /2) → –1 /2 : d → s

V+ (–1/2) → 1/2 : u → s

and

V–(+1 /2) → - 1/2 : s → u.

The quar ks ca n be chara cterized b y the thre e qua ntum numbe rs I3, U 3, V3.

Their po sitions in the I3-U3-V3 - space again sho w the unde rlying

sym metry :

U3 -V 3

+1/2

d (-1/2, 1/2, 0) u(1/2, 0, -1/2)

-I 3 - 1/2 + 1/2 I3

+1/ 2

s( 0, -1/2, 1/2)
V3 -Y -U 3
The memb ers o f the octe t of meson s with JP = 0– ar e for med f rom q q-pairs 148

tha t bel ong t o the fund ament al 3, 3 re prese ntation of the quark s. T he π0

and η0 me sons are l inear comb inations o f the q q(bar) s tates, thu s

K0 d s Y K+ us

d u

π– d u π0 π+ ud

-1 η0 +1 I3

u d
s

K– s u K0 s d

The none t for med f rom t he te nsor produ ct 3 ⊗ 3 i s split into an octe t

tha t is even under the label exch ange of tw o par ticles, and a s inglet tha t is

odd unde r lab el exchang e :

3⊗3 = 8⊕1

whe re th e “1” is

η0´ = (1/√ 3)(uu + dd + ss ),

an d the two membe rs of the octet at t he ce nter are:

π0 = ( 1/√2)(uu – dd) and η0 = (1/√6 )(uu + dd - 2ss).

The action of I– on π+ is to t ransform i t int o a π 0. This opera tion has t he

following mea ning in te rms o f I– ac ting on th e ten sor p roduc t, u ⊗ d:

 I –(u ⊗ d) ≡ (I–u) ⊗ d + u ⊗ (I–d) (c.f. der ivative ru le) 149
↓ ↓ ↓
I– ( π ) = d ⊗ d + u
+
u

→ π0
Omitting the tenso r pro duct sign, norm alizing th e amp litudes, and ch oosing

the phas es in the gener ally accepted w ay, w e hav e:

π0 = (1/√2 )(uu – dd) .

The singlet η0´ is said to b e ort hogon al to π0 an d η0 at the origin.

If the s ymmet ry of the octet were exac t, th e eight me mbers of t he

oct et wo uld h ave t he sa me ma ss. This is not qui te th e case; th e sym metry

is broke n by the d ifference in eff ectiv e ma ss between the u- an d d-q uark

(essentially the s ame e ffective m asses: ~ 3 00 Me V/c2) a nd th e s-quark

(effective ma ss ~ 500 M eV/c2). (It should be noted that the effective

mas ses of the quar ks, d erived fro m the mass differenc es of hadr on-pairs, is

not the same as th e “cu rrent -quar k” ma sses that appear in the

fundamen tal t heory . Th e discrepa ncy b etween the effective mass es an d the

fundamen tal m asses is n ot fu lly unders tood at th is time).

The deco mposition of 3 ⊗ 3 ⊗ 3 is

3 ⊗ 3 ⊗ 3 = (6 ⊕ 3) ⊗ 3

= 10 ⊕ 8 ⊕ 8´ ⊕ 1

in which the states of the 10 ar e sym metri c, th e 1 is anti symme tric, and the

8, 8´ states are of mixed s ymmet ry. The d ecuplet th at ap pears in t his

dec ompos ition is a ssociated with the o bserv ed de cuplet of spin-3/2 b aryon s.

In terms of t he th ree f undam ental quar ks — u, d, and s, th e mak e -up of

the individual mem bers of th e dec uplet is shown schematically in the 150

following diagram:

d dd ~ du d ~ u ud uuu

~ dds ~ dus ~ uus

~ sds ~ sus

sss

The prec ise make-up of each state, labelled by ( Y, I, I3,) is given in the

following tab le:

(++ )
(1, 3/2, +3/2 ) = uuu
(1, 3/2, +1/2 ) = (1/ √3)(udu + duu + uud)
(1, 3/2, –1/2 ) = (1/ √3)(ddu + udd + dud)
(–)
(1, 3/2, –3/2 ) = ddd

(0, 1, + 1) = (1 /√3)(usu + suu + uus)

(0, 1, 0) = ( 1/√6)(uds + dsu + su d + d us + sdu + usd)
(0, 1, – 1) = (1 /√3)(dsd + sdd + dds )

( –1, 1 /2, +1/2) = (1/√3 )(ssu + us s + s us)

( –1, 1 /2, –1/2) = (1/√3 )(ssd + ds s + s ds)
(–)
(-2, 0, 0) = sss

The gene ral t heory of t he pe rmuta tion group of n enti ties, and its

rep resen tations, is out side the s cope of th is introdu ction. Th e use of t he

Young ta bleaux in obtaining the m ixed symme try s tates is t reate d in

Ham ermes h (19 62).

The char ges o f t he ∆ ++ , ∆ –, a nd Ω– pa rticles fix the frac tional values

of the q uarks , nam ely

 quar k flavor c harge (in units of t he electro n cha rge) 151

u +2/3

d –1/3

s –1/3

The char ges o f the anti -quar ks ar e opp osite in s ign t o the se va lues.

Ext ensive rev iews of th e 3-q uark model and its application to t he

phy sics of th e low-energy pa rt of the hadro n spe ctrum can be fo und i n

Gasiorowicz (1966) and Gibson and Pollard ( 1976).

13. 6 Th e nee d for a ne w qua ntum numbe r: hi dden color

Im mediately after the intro duction of th e 3-q uark model by

Gell-Mann and Zwei g, it was recog nized that the model was not c onsistent

with the exte nded Pauli prin ciple when applied t o bou nd st ates of th ree

qua rks. For examp le, the st ructu re of the spin-3/2 ∆ + st ate is such tha t, if

each qua rk is assigned a spin sq = 1/2, the t hree spins must be a ligned ↑↑↑

to give a net spin of 3 /2. (It is assumed that the r elative or bital angu lar

mom entum of t he qu arks in th e ∆ + is zero (a s ymmet ric s-state) — a

rea sonable assumpt ion t o mak e, as it c orres ponds to m inimum kin etic

ene rgy, and t herefore t o a s tate of lowest total ener gy). The quark s are

fer mions, and ther efore they must obey the gener alized Pau li Principle;

the y can not e xist in a completely aligned s pin state when they are i n an s-

sta te th at is symm etric unde r par ticle (qua rk) e xchan ge. The uni tary sp in

com ponen t of the t otal wavefunction mu st be anti -symmetric. Gr eenberg

(1964) p ropos ed th at a new d egree of f reedo m mus t be assigned t o the

qua rks i f the Paul i Principle is not t o be violated. The new p roper ty wa s 152

later ca lled “colo r”, a prop erty with profo und c onsequences. A quar k

with a c ertain flavor p ossesses color (red, blue, gre en, say) t hat

cor respo nds t o the trip let r epres entat ion o f ano ther form of SU (3) —

nam ely S U(3)C, w here the s ubscr ipt C different iates the group from that

introduc ed by Gell-Mann and Zweig — th e flavor g roup SU(3)F. The a nti-

qua rks ( that possess anti-color) have a tri plet repre sentation in SU (3)C th at

is the c onjugate r epres entat ion (the 3). Alth ough the S U(3)F sy mmetr y is

kno wn no t to be ex act, we ha ve ev idence tha t the SU( 3)C sy mmetry is an

exa ct sy mmetry of Nature. Baryo ns an d mes ons a re fo und t o be colorless;

the color sin glet of a baryo n occ urs i n the deco mposition

S U(3)C = 3 ⊗ 3 ⊗ 3 = 10 + 8 + 8´ + 1 .

The meso n singlets cons ist of linear c ombinations of the f orm

1 = (RR + BB + GG)/√3 .

Although the hadro ns ar e colorless, certain obse rvable qua ntities are

dir ectly rela ted t o the numb er of colors in the model. Fo r exa mple, the

pur ely electromagn etic decay of t he ne utral pion , π0, into t wo ph otons

π0 = γ + γ,

has a lifetime tha t is found to b e closely propo rtion l to the squ are o f the

num ber o f col ors. (A dler (1970) gives Γ = h/τ = 1(eV) (no.of co lors)2 .

The meas ureme nts o f the lifetime give a value of Γ ~8 eV, consistent with

Ncol s = 3. S ince these earl y mea surem ents, refined e xperiments have
dem onstr ated that there are three , and only thre e, co lors associated with 153

the quar ks.

In studies of electron- positron i ntera ctions in the G eV-region, the

rat io of cros s sections:

R = σ(e+ e– → ha drons )/σ(e+ e– → µ+ µ–)

is found to d epend linearly on th e num ber o f colors. Good agre ement

bet ween the t heore tical mode l and the measured v alue of R, over a wi de

ran ge of ener gy, is obt ained for thr ee co lors.

The color att ribut e of the q uarks has been respo nsible for the

dev elopment o f a t heory of t he st rongl y int eract ing p articles, called

qua ntum chrom odyna mics. It is a field theo ry in which the quar ks ar e

gen erato rs of a ne w typ e of field — th e color field. The mediators of th e

field ar e called glu ons; t hey p ossess col or, t he at tribu te of the sourc e of the

field. Consequent ly, they c an interac t wit h eac h oth er th rough the color

field. This is a field quit e unlike t he electro dynam ic field of classical

electrom agnet ism, in wh ich t he field quanta do n ot ca rry t he at tribu te of

the sour ce of the field, nam ely electric charge. The phot ons, there fore, do

not inte ract with each other .

The gluo ns tr ansform a quark of a part icular col or in to a quark of a

different col or. For e xample, in the interaction bet ween a red quar k and a

blue qua rk, t he co lors are e xchan ged. This requ ires that the e xchan ged

gluon ca rry c olor and a nti-color, as s hown:

 qb qr 154

glu on, g rb(bar) -ca rries red and a nti-blue:

th e color lines a re con tinuo us.
qr qb

Thr ee different co lors permit nin e different way s of coupling q uarks

and gluo ns. Three of t hese are r ed-red, blue-blue, a nd gr een-green that do

not chan ge th e colors. A linear combination ~(R →R + B→B + G→G) is

sym metri c in the c olor labels, and thi s com bination is the singlet state of

the grou p SU( 3)C. Eight gluo ns, each w ith t wo co lor indices, ar e the refor e

req uired in t he 3- color theo ry of quar ks.

13. 7 Mo re ma ssive quar ks

In 1974, the results of two independen t ex periments, one a stu dy of

the reac tion p + B e → e+ + e– .. (Ting et al.) and t he ot her a stud y of

e+ + e– → ha drons ..(Richter et al) — show ed th e pre sence of a shar p

res onanc e at a cen ter-o f-mass ene rgy o f 3.1 GeV. The lifetime of th e

res onant stat e was foun d to be ~1 0–20 se conds — mo re th an 10 3 se conds

longer t han e xpect ed fo r a s tate forme d in the s trong inte raction. The

res onant stat e is called the J/ψ. It wa s quickly realized t hat t he st ate

cor respo nds t o the grou nd st ate o f a n ew qu ark–a nti-quark system, a

bou nd st ate c c, wh ere c is a four th, m assive, qu ark e ndowe d wit h one unit

of a new quan tum n umber c, c alled “cha rm”. The quant um nu mbers

assigned to t he c-quark are

 JP = 1/2+ , c = 1, Q/e = +2/ 3, an d B = 1/3. 155

Sou nd th eoret ical argum ents for a four th qu ark, carry ing a new

qua ntum numbe r, ha d bee n put forw ard s evera l years be fore the

exp erimental obser vation of the J /ψ st ate. Sinc e 197 4, a complex se t of

sta tes o f the “cha rmoni um” s ystem has been obser ved, and t heir decay

pro perties st udied. De tailed com parisons h ave b een m ade w ith

sop histicated theo retical mo dels of th e system.

The inclusion of a char med q uark in th e set of q uarks mean s tha t the

gro up SU (4)F mu st be used in p lace of th e ori ginal Gell-Mann-Zweig gro up

SU( 3)F. Although t he SU (4)F sy mmetr y is badly brok en be cause the

effective mas s of the c harme d qua rk is ~ 1. 8 GeV/c2, s ome u seful

app lications have been made using the model. Th e fun damen tal

rep resen tations ar e

[ u, d, s, c ], a covar iant column spinor,

an d

( u, d, s, c ), a contr avariant r ow sp inor.

The irre ps ar e con struc ted i n a w ay th at is analogous to t hat u sed in

SU( 3)F, n amely, by finding th e sym metri c and anti -symmetric

dec ompos itions of the v arious ten sor p roduc ts. The m ost u seful are:

4 ⊗ 4 = 15 ⊕ 1,

4 ⊗ 4 = 10 ⊕ 6,

4 ⊗ 4 ⊗ 4 = 20sym ⊕ 20mix ⊕ 20´mix ⊕ 4ant i,

an d
 15 ⊗ 15 = 1 ⊕ 15sym ⊕ 15ant i ⊕ 20sym ⊕ 45 ⊕ 45 ⊕ 84. 156

The “15” includes the n on-charmed (JP = 0– ) m esons and the f ollowing

cha rmed meson s:

D0 = cu, D 0 = cu, m ass = 1863 MeV/c2 ,

D+ = cd, D – = cd, m ass = 1868 MeV/c2,

F + = cs, F – = cs, mass = 2.0 4 MeV/c2.

In order to d iscuss the bary ons, it is nece ssary to i nclude the quar k spin,

and ther efore the group must be e xtend ed to SU(8 )F. Relatively few

bar yons have been studied in deta il in this exte nded framework.

In 1977, well-defined r esonant st ates were obser ved a t ene rgies of

9.4 , 10. 01, a nd 10 .4 Ge V, and wer e int erpre ted a s bou nd st ates of an other

qua rk, t he “b ottom ” qua rk, b , and its anti-partn er, t he b. Mes ons c an be

for med t hat i nclude the b-qu ark, thus

Bu = bu, B d0 = bd, B s0 = bs, and Bc = bc .

The stud y of the w eak d ecay modes of t hese states is curre ntly fashionable.

In 1994, definitive evidence was obtained f or th e existence of a sixth

qua rk, c alled the “top” quar k, t. It is a massive en tity with a mas s almost

200 time s the mass of t he pr oton!

We have seen that the q uarks inte ract stron gly v ia gluon e xchan ge.

They also tak e par t in the w eak intera ction. In an e arlier dis cussion of

isospin, the group gene rator s wer e int roduc ed by cons idering th e β-decay

of the f ree n eutro n:

n 0 → p + + e– + ν0 .
We now k now t hat, at th e microsco pic level, this proc ess involves th e 157

tra nsformation of a d-q uark into a u-q uark, and the p roduc tion of th e

car rier of th e wea k for ce, t he ma ssive W– pa rticle. T he W– bo son (spin 1)

dec ays instan tly into a n electron –anti-neutrino pair, as s hown:

ν0
– 1
W e–

d u
neu tron, n 0 d(– 1/3) → u( +2/3) prot on, p +
u u

d d

The carr iers of th e Weak For ce, W± , Z 0, were first identified in p-p

collisions at high cent er-of-mass ener gy. The p roces ses involve

qua rk–an ti-quark i ntera ctions, and the dete ction of t he de cay e lectrons a nd

pos itron s.

e+ e–
0
Z
u( +2/3) u ( –2/3)
ν 0

W+ e+
u( +2/3) d (+1/3)
ν 0

W– e–
d( −1/3) u (−2/3)

The char ge is cons erved at e ach v ertex .

 The carr iers have very large meas ured masses: 158

mas s W± ~ 8 1 GeV/c2, a nd ma ss Z0 ~ 93 Ge V/c2.

(R ecall that the range of a forc e ∝ 1/ (mass of c arrier); the W and Z mass es

cor respo nd to a ve ry sh ort r ange,~10 -18 m, for the Weak F orce).

Any qua ntitative discussion of c urren t wor k using Gr oup T heory to

tac kle Grand Unified Th eories, re quires a k nowledge o f Qua ntum Field

Theory t hat i s not expe cted of re aders of t his introd uctor y boo k.
 14 159

LIE GROU PS AN D THE CONS ERVATION L AWS O F THE

PHY SICAL UNIV ERSE

14. 1 Po isson and Dirac Brac kets

The Poi sson Brack et of two differentiable functions

A(p1, p 2, . ..pn, q 1, q 2, . ..qn)

and

B(p1, p 2, . ..pn, q 1, q 2, . ..qn)

of two s ets o f var iables (p1, p 2, . ..pn) a nd (q 1, q 2, . ..qn) is defined as

n
{A, B} ≡ ∑1 (∂ A/∂qi)(∂ B/∂pi) – (∂A/∂pi)(∂ B/∂qi) .

If A ≡ (pi, q i), a dyn amical var iable, and

B ≡ H(pi, q i), the h amiltonian of a dyna mical syst em,

whe re p i is the (canonical) mom entum and qi is a (g enera lized) coo rdinate,

the n
n
{ , H} = ∑1 (∂ /∂q i)(∂ H/∂p i) – (∂ /∂p i)(∂ H/∂q i) .

(n is the”num ber o f deg rees of fr eedom ” of the s ystem ).

Ham ilton’s eq uations ar e

∂H/∂p i = dqi/dt and ∂H/∂q i = – dp i/dt ,

and ther efore

n
{ , H} = ∑1 (∂ /∂q i)(dqi/dt) + ( ∂ /∂q i)(dpi/dt) .

The tota l differential of (pi, q i) is

 d = ∑1n (∂ /∂q i)dqi + (∂ /∂p i)dpi. 160

and its time derivative is

( d /dt) = ∑ 1n (∂ /∂q i)(dqi/dt) + ( ∂ /∂p i)(dpi/dt)

•
= { , H} = .

If the P oisson Bra cket is ze ro, t he ph ysical qua ntity is a co nstan t

of the m otion .

In Quant um Me chanics, t he re lation

(d /dt) = { , H}

is replaced b y

(d /dt) = - (i/h))[ , H],

Heisenberg’s equat ion o f mot ion. It i s the cust om to refe r to the

com mutat or [ , H] a s the Dira c Bra cket.

If the D irac Brack et is zero, the quan tum m echan ical quant ity is

a c onsta nt of the motio n..

(Dirac p roved that the classical Poisson Br acket { , H} c an be

identified with th e Heisenberg co mmuta tor – (i/h)[ , H] b y mak ing a

suitable choice of the ord er of the q’s a nd p’ s in the P oisson Bra cket).

14. 2 In finit esima l uni tary trans forma tions in Q uantu m Mec hanic s

The Lie form of an infinitesimal unitary tr ansformati on is

U = I + iδαX/h ,

whe re δα ia real infinitesimal param eter, and X is an h ermit ian operat or.

(It is straightfor ward to sh ow th at th is form of U is, indeed, unitary).

 Let a dy namical operato r ch ange under an i nfinitesimal unitary 161

tra nsformation:

→ ´ = U U–1

= (I + iδaX/h) (I – iδaX/h)

= – iδa X/h + iδaX /h to 1st- order

= + i(δaX – δaX)/h

= + i(F – F)/h.

whe re

F = δaX.

The infinitesimal chang e in is ther efore

δ = ´–

= i[F, ]/h

If we identify F wi th –Hδt ( the c lassical form f or a purely tem poral chan ge

in the s ystem ) the n

δ = i[−Hδt, ]/h,

or

–δ = i[H, ]δt/h ,

so that

–δ /δt = i[H, ]/h.

For a te mpora l change in the syst em, δ /δt = – d /dt.

The fund ament al He isenb erg e quati on of moti on 162

d /dt = i[ , ]/h

is therefore deduc ed fr om th e uni tary infin itesi mal t ransformat ion o f the

ope rator .

This app roach was taken by S chwinger i n his form ulation of Quan tum

Mechanics.

|F| = Hδt i s dir ectly rela ted t o the gene rator , X, o f a Q uantu m

Mechanical infinitesimal tra nsformation, an d the refor e we can a ssociate

with eve ry sy mmetr y tra nsformation of the s ystem an h ermit ian operat or

F th at is a co nstan t of the m otion - it s eigenvalues d o not chan ge with

time. T his is an examp le of Noe ther’ s The orem:

A c onservatio n law is a ssociated with every symm etry of th e

equ ation s of motio n. I f the equa tions of m otion are uncha nged by th e

tra nsformations of a Gr oup t hen a prop erty of th e system w ill remain

con stant as t he sy stem evolves with ti me. As a well-known exam ple, if the
equ ations of motion of an ob ject are i nvariant u nder trans lations is spac e,

the linear mo mentu m of the s ystem is c onser ved.

 15 163

BIBLIOGRAPHY

The following book s are typ ical of th ose t hat a re su itable for

Und ergra duate s:

Arm stron g, M. A., Gro ups a nd Sy mmetry, S pring er-Verlag, New York,

198 8.

Bur ns, G erald, Int roduction to Gr oup T heory, A cadem ic Pr ess, New York,

197 7.

Fri tzsch, Har ald, Qua rks: the S tuff of Ma tter, B asic Books, New York ,

198 3.

Jon es, H. F., Gro ups, Repre senta tions and Physi cs, A dam H ilger, Bristol,

199 0.

The following book s are of a spec ialized na ture; t hey a re ty pical of

wha t lies bey ond t he pr esent intr oduct ion.

Car ter, Roger ; Seg al, Graeme ; and Macd onald, Ian, Lec tures on L ie

Gro ups a nd Li e Alg ebras, C ambridge U niversity Press , Cam bridg e, 19 95.

Com mins, E. D ., and Buc ksbaum, P. H., Wea k Int eractions of Le ptons and

Qua rks, C ambridge U niversity Press , Cam bridg e, 19 83

Dickson, L. H ., Lin ear G roups, D over, New York, 1960 .

Eisenhar t, L. P., Con tinuo us Gr oups of Tr ansfo rmati ons, D over, New

York, 19 61.

Elliott, J. P ., and Daw ber, P. G. , Sym metry in P hysics, Vol . 1,

Oxford U niversity Press , New York , 197 9.

Gell-Mann, Mu rray, and Ne’em an, Yuval, The Eigh tfold Way, 164

Ben jamin, New York , 196 4.

Gibson, W. M., and Pollard, B. R., Sym metry Prin ciple s in Eleme ntary

Par ticle Phys ics, C ambridge U niversity Press , Cam bridg e, 19 76.

Ham ermes h, Mo rton, Gro up Th eory and i ts Ap plica tions to P hysical

Pro blems, D over, New York, 1989 .

Lichtenb erg, D. B., Uni tary Symmetry a nd El ement ary P artic les,

Aca demic Pres s, New Yor k, 19 78.

Lip kin, Harry J., Lie Grou ps fo r Ped estrians, N orth- Holland, A mster dam,

196 6.

Lom ont, J. S., Ap plica tions of F inite Grou ps, D over, New York, 1993 .

Racah, G ., Gro up Th eory and S pectroscopy, R eprinted i n CER N(61-68),

196 1.

Wigner, E. P. , Gro up Th eory and i ts Ap plica tions to t he Qu antum

Mec hanic s of Atomi c Spe ctra, A cadem ic Pr ess, New York, 1959.