You are on page 1of 65

Algebra (Advanced)

Math 2968
Lecture notes by
Andrew Mathas
with revisions and additions by
Alexander Molev
S
I
D
ERE ME
N
S EAD
E
M
MUTA
T
O
School of Mathematics and Statistics
University of Sydney
2006
2 Algebra (Advanced)
Algebra (Advanced)
This course is concerned with inner product spaces and with group theory.
An inner product spaces is a special type of vector space which comes equipped with an
inner product; this leads to notions of length, distance and angles.
A group is a set equipped with a single binary operation which satises some simple axioms;
you should think of this operation as multiplication (or addition). The group axioms are an ab-
straction of the essential properties of multiplication (or addition) in a general context. Groups
are ubiquitous in mathematics and have a wide range of applications throughout the sciences
and beyond. Whenever symmetry is present there is a group in the background and group theory
can be used to help solve the problem.
There are bound to be typos in these notes; I would be very pleased to hear of any that you
spot.
Contents
1. Inner product spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Length and the CauchySchwartz inequality . . . . . . . . . . . . . . . . . . . . . . 4
3. Orthogonality and projection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4. Application to linear regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5. Complex inner product spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
6. Isometries and inner product spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
7. Normal, Hermitian and unitary matrices . . . . . . . . . . . . . . . . . . . . . . . . . 14
8. The denition of a group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
9. Subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
10. Generators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
11. Cosets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
12. Equivalence relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
13. Homomorphisms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
14. Normal subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
15. Quotient groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
16. The isomorphism theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
17. The structure of groups I: the existence of pelements. . . . . . . . . . . . 40
18. Group actions on sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
19. Conjugacy classes and groups of ppower order . . . . . . . . . . . . . . . . . 46
20. Direct and semidirect products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
21. The structure of groups II: Sylows rst theorem . . . . . . . . . . . . . . . . 51
22. Sylows second theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
23. Groups of order pq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
24. Simple groups of small order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
25. Free groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
26. Generators and relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Inner product spaces 3
PART I: INNER PRODUCT SPACES
1. Inner product spaces
Let V be a real vector space. Recall that this means that V is a set equipped with two opera-
tions, vector addition ((x, y) x + y for x, y V ) and scalar multiplication ((, x) x
for x V and R), such that for all x, y, z V and , R the following hold:
a) x + y = y + x;
b) (x + y) + z = x + (y + z);
c) there is an element 0 V with 0 + x = x + 0;
d) there exists an element x

V with x + x

= 0 = x

+ x (usually we write x

= x);
e) 1 x = x (where 1 R is the multiplicative identity of R);
f) (x) = ()x;
g) (x + y) = x + y;
h) ( + )x = x + x.
1.1 Denition Let V be a real vector space. An inner product on V is a map
, ) : V V R
such that for all x, y, z V and , R the following hold:
a) (symmetric) x, y) = y, x);
b) (linear) x + y, z) = x, z) + y, z); and,
c) (positive denite) x, x) > 0 whenever x ,= 0.
A real vector space V is an inner product space if it has an inner product.
Notice that by combining (a) and (b) we also have
x, y + z) = y + z, x) = y, x) + z, x) = x, y) + x, z).
Hence, , ) is linear in both variables; we say that , ) is bilinear.
1.2 Example We should start with some examples of inner product spaces.
a) Let V = R
n
. An element x R
n
can be thought of as a row vector x = (x
1
, . . . , x
n
).
(We could also use column vectors; however, this is slight harder typographically.) Given
y = (y
1
, . . . , y
n
) V dene
x, y) = x y = (x
1
, . . . , x
n
) (y
1
, . . . , y
n
) = x
1
y
1
+ x
2
y
2
+ + x
n
y
n
,
where x y is the standard dot product on R
n
. It is straightforward to check that this is an
inner product.
b) Suppose that a < b are two real numbers and let
V = ([a, b] = f : [a, b] R [ f is continuous .
It is easy to check that V is a real vector space and that
f, g) =
_
b
a
f(x)g(x) dx, f, g V,
denes an inner product on V . Notice that this inner product is just the continuous ana-
logue of (a).
4 Algebra (Advanced)
1.3 Lemma Suppose that W is a vector subspace of V . Then W is an inner product space.
Proof See tutorials.
1.4 Lemma Suppose that x V . Then x, 0) = 0. Consequently, x, x) 0 for all x V .
Proof By linearity, x, 0) = x, 0 + 0) = x, 0) + x, 0); hence, x, 0) = 0 as claimed. In
particular, 0, 0) = 0 so x, x) 0 for all x V (since, by assumption, x, x) > 0 if x ,= 0).
2. Length and the CauchySchwartz inequality
Suppose that V is an inner product space. Then the length (or norm) of a vector x V is
dened to be |x| =
_
x, x). Note that |x| 0 for all x V with equality if and only
if x = 0 by Lemma 1.4. It is also easy to check that |x| = [[ |x| for all R and x V
(see tutorials).
The key property relating lengths and inner products is the following.
2.1 Theorem (CauchySchwartz inequality) Suppose that x, y V . Then
[x, y)[ |x| |y|.
Proof If x = 0 then x, y) = 0 and |x| = 0 so [x, y)[ = |x| |y| and were done. Suppose
now that x ,= 0. Then |x|
2
= x, x) > 0. For any real we have by linearity,
0 y x, y x) = y, y x) x, y x)
= y, y) y, x) x, y) +
2
x, x)
= |y|
2
2x, y) +
2
|x|
2
.
Hence, the discriminant of this quadratic polynomial must be non-positive. Therefore, we have
x, y)
2
|x|
2
|y|
2
0 and hence x, y)
2
|x|
2
|y|
2
. Taking square roots gives the result.
2.2 Corollary Suppose that x, y V . Then |x + y| |x| +|y|.
Proof Using the CauchySchwartz inequality on the second line we have
|x + y|
2
= x + y, x + y) = x, x) + 2x, y) +y, y) = |x|
2
+ 2x, y) +|y|
2
|x|
2
+ 2|x| |y| +|y|
2
= (|x| +|y|)
2
,
Hence, |x + y| |x| + |y| as claimed.
Orthogonality and projection 5
3. Orthogonality and projection
Rearranging the CauchySchwartz inequality (Theorem 2.1) shows that
1
x, y)
|x| |y|
1,
for all nonzero x, y V . Therefore, there is a unique angle [0, ] such that
cos =
x, y)
|x| |y|
.
By denition is the angle between x and y.
3.1 Denition Two vectors x, y V are orthogonal if x, y) = 0. An orthogonal basis of V is a
basis f
1
, . . . , f
n
such that f
i
and f
j
are orthogonal whenever i ,= j.
Orthogonality gives an easy way to test for linear independence.
3.2 Lemma Suppose that f
1
, f
2
, . . . is a set of pairwise orthogonal nonzero vectors in V
(that is, f
i
, f
j
) = 0 if i ,= j). Then f
1
, f
2
, . . . is linearly independent.
Proof See tutorials.
Note that the following results also apply in the case W = V .
3.3 Proposition Suppose that W is a subspace of an inner product space V and that f
1
, . . . , f
m
)
is an orthogonal basis of W and let w W. Then
w =
m

i=1
w, f
i
)
f
i
, f
i
)
f
i
.
Proof As f
1
, . . . , f
m
is a basis of W we can write w uniquely in the form w =

m
i=1

i
f
i
for
some
i
R. Therefore, if 1 j m then by linearity
w, f
j
) =
m

i=1

i
f
i
, f
j
) =
m

i=1

i
f
i
, f
j
) =
j
f
j
, f
j
),
since f
i
, f
j
) = 0 by orthogonality. Hence,
j
= w, f
j
)/f
j
, f
j
) as claimed. (Note that
f
j
, f
j
) > 0 because f
j
,= 0.)
A vector v V is normal if |v| = 1. An orthonormal basis of V is an orthogonal basis
f
1
, . . . , f
n
such that each f
i
is normal; equivalently,
f
i
, f
j
) =
ij
=
_
1, if i = j,
0, otherwise.
If v is any nonzero vector in V then we can normalize it by setting v =
1
v
v. Consequently, we
can always replace a basis of orthogonal vectors with a basis of orthonormal vectors.
Using an orthonormal basis gives a slightly nicer reformulation of the last result.
6 Algebra (Advanced)
3.4 Corollary Suppose that W is a subspace of V and that f
1
, . . . , f
m
) is an orthonormal
basis of W and let w W. Then
w =
m

i=1
w, f
i
)f
i
.
These last two results are important because they have the following consequence.
3.5 Theorem Assume that W has an orthonormal basis f
1
, . . . , f
m
and let v V . Then
there exists a unique vector w W such that
v, x) = w, x), for all x W;
indeed, w =
m

i=1
v, f
i
)f
i
. Moreover, |v w| < |v x|, whenever x W and x ,= w.
Proof Let w be an arbitrary element of W and suppose that v, x) = w, x) for all x
W. Then, in particular, v, f
i
) = w, f
i
) for 1 i m; therefore, w =

m
i=1
v, f
i
)f
i
by
Corollary 3.4. This shows that if such a w exists then it is unique (every element is uniquely
determined by its expansion with respect to a basis). Conversely, if x is arbitrary element of W
then we need to show that w, x) = v, x). Write x =

m
i=1

i
f
i
, for
i
R. Then
w, x) = w,
m

i=1

i
f
i
) =
m

i=1

i
w, f
i
) =
m

i=1

i
v, f
i
) = v,
m

i=1

i
f
i
) = v, x).
It remains to show that |v w| < |v x| whenever x W and x ,= w. Again we compute
|v x|
2
= v x, v x) = (v w) + (w x), (v w) + (w x))
= v w, v w) + 2v w, w x) +w x, w x)
= |v w|
2
+ 2v w, w x) +|w x|
2
.
Now, x

= w x W so v, x

) = w, x

) by the dening property of w; hence, 0 =


v w, x

) = v x, w x). Therefore,
|v x|
2
= |v w|
2
+|w x|
2
|v w|
2
,
with equality if and only if w = x by Lemma 1.4. Hence, |v x| > |v w| if x ,= w, as
claimed.
In Theorem 3.5 we have assumed that W has an orthonormal basis; it turns out that this is
always true so this assumption can be dropped. Before we prove this we make a denition
which will be useful in the proof.
3.6 Denition Suppose that f
1
, . . . , f
m
is an orthonormal basis of W, a subspace of V and
that v V . The orthogonal projection of v onto W is the vector

W
(v) =
m

i=1
v, f
i
)f
i
.
Orthogonality and projection 7
The point of Theorem 3.5 is that |v
W
(v)| |v x| for all x W with equality if and
only if x =
W
(v). In other words,
W
(v) is the element of W which is closest to v and
there is a unique such element (shortly we will see how this can be applied to linear regression
problems). As a consequence, if w W then
W
(w) = w since W is certainly the element
of W which is closest to itself! (This also follows by comparing the formulae of Corollary 3.4
and Theorem 3.5.) It follows that
W
is a surjective map from V onto W.
Note that we could equally well dene
W
(v) in terms of an orthogonal basis; more precisely,
it is easy to check that if g
1
, . . . , g
m
is an orthogonal basis of W then

W
(v) =
m

i=1
v, g
i
)
g
i
, g
i
)
g
i
.
(To see this set f
i
=
1
g
i

g
i
, for each i; then f
1
, . . . , f
m
is an orthonormal basis of W and,
moreover, v, f
i
)f
i
= v,
1
g
i

g
i
)
1
g
i

g
i
=
1
g
i

2
v, g
i
)g
i
=
v,g
i

g
i
,g
i

g
i
.)
3.7 Example In Tutorial 1, question 6, it is shown that the functions
f
n
(x) =
_
sin(nx), if n > 0,
cos(nx), if n 0,
for n Z, are pairwise orthogonal elements of ([, ]. Furthermore, f
0
, f
0
) = 2 and
f
n
, f
n
) = when n ,= 0. Let (
N
be the subspace of ([, ] spanned by the functions
f
n
[ N n N and let f be a function in ([, ]. Then the projection of f onto (
N
is

C
N
(f) =
N

n=N
f, f
n
)
f
n
, f
n
)
f
n
=
1
2
a
0
+
1

n=1
_
a
n
cos(nx) + b
n
sin(nx)
_
,
where
a
n
= f, f
n
) =
_

f(x) cos(nx) dx and b


n
= f, f
n
) =
_

f(x) sin(nx) dx.


If f is 2periodic then it is possible to show that
C
N
(f) is a very good approximation to f
(when N 0). This example is the start of Fourier analysis.
We now prove that every vector space has an orthogonal basis. The proof is constructive.
3.8 Theorem (GramSchmidt orthogonalization) Suppose that V is a nite dimensional in-
ner product space. Then V has an orthonormal basis.
Proof Choose a basis v
1
, . . . , v
n
of V and for m = 1, . . . , n let V
m
be the subspace of V with
basis v
1
, . . . , v
m
. We show by induction on m that V
m
has an orthonormal basis. If m = 1
then f
1
is an orthonormal basis of V
1
, where we set f
1
=
1
v
1

v
1
. Assume, by way of induc-
tion, that f
1
, . . . , f
m
is an orthonormal basis of V
m
. Then Theorem 3.5 can be applied to V
m
,
so we can set g
m+1
= v
m+1

Vm
(v
m+1
). Note that g
m+1
is nonzero because v
1
, . . . , v
m+1

is linearly independent. Moreover, if 1 k m then


g
m+1
, f
k
) = v
m+1

Vm
(v
m+1
), f
k
) = v
m+1
, f
k
)
Vm
(v
m+1
), f
k
) = 0
8 Algebra (Advanced)
by Theorem 3.5. Set f
m+1
=
1
g
m+1

g
m+1
. Then f
1
, . . . , f
m+1
is an orthonormal basis
of V
m+1
, completing the proof of the inductive step. As V = V
n
the theorem follows.
The proof of Theorem 3.8 gives us a way of constructing an orthonormal basis of V given
a basis; namely, if v
1
, . . . , v
n
is a basis of V then g
1
, . . . , g
n
is an orthogonal basis of V
where
g
1
= v
1
g
2
= v
2

V
1
(v
2
) = v
2

v
2
, g
1
)
g
1
, g
1
)
g
1
g
3
= v
3

V
2
(v
3
) = v
3

v
3
, g
1
)
g
1
, g
1
)
g
1

v
3
, g
2
)
g
2
, g
2
)
g
2
.
.
.
.
.
.
.
.
.
g
n
= v
n

V
n1
(v
n
) = v
n

v
n
, g
1
)
g
1
, g
1
)
g
1

v
n
, g
n1
)
g
n1
, g
n1
)
g
n1
By normalizing g
1
. . . . , g
n
we obtain an orthonormal basis of V . Incidentally, in carrying
out the GramSchmidt algorithm we do not need to assume that v
1
, . . . , v
n
is a basis of V ;
a spanning set is enough. To see this assume that v
1
, . . . , v
m
is linearly independent and let
V
m
be the space spanned by these vectors. By Theorem 3.5, v
m+1
V
m
if and only if v
m+1
=

Vm
(v
m+1
); in the notation above, this is inequivalent to saying that g
m+1
= 0. Consequently,
the GramSchmidt algorithm will rene a spanning set to an orthogonal basis.
Finally, we remark that the orthonormal basis of V produced by GramSchmidt orthogonal-
ization is not unique (that is, V has many different orthonormal bases). Rather it depends on the
choice of initial basis v
1
, . . . , v
n
and the order in which these basis elements are listed. For
example,
1
v
1

v
1
always belongs to the resulting orthonormal basis.
4. Application to linear regression
As an example we now address the following problem: given a collection of random vari-
ables X and Y how can we determine their line Y = aX + b of best t? At rst sight this
does not relate to inner product spaces; however we will see that it is a direct application of
Theorem 3.5. In principle, the technique that we describe can be used to t any type of curve to
a collection of data; cf. Example 3.7.
Consider a collection of random variables (x
1
, y
1
), . . . , (x
n
, y
n
); for example, coming from
an experiment. We want to determine the line Y = aX + b which best describes this data.
Let x = (x
1
, . . . , x
n
), y = (y
1
, . . . , y
n
) and z = (1, . . . , 1). We want to nd a, b R
such that y ax + bz. In other words, we want to nd the projection
W
(y) of y onto
the subspace W of R
n
spanned by x and z. Applying the GramSchmidt algorithm to the
basis z, x of W yields the orthogonal basis g
1
, g
x
, where g
1
= z and g
x
= x xz and
x =
1
n

n
i=1
x
i
, the mean of the x
i
. Hence, the projection of y onto W is

W
(y) =
y, g
1
)
z, z)
g
1
+
y, g
x
)
g
x
, g
x
)
g
x
= yg
1
+
y, g
x
)
S
2
x
g
x
,
Complex inner product spaces 9
where y =
1
n

n
i=1
y
i
and S
2
x
=

n
i=1
(x
i
x)
2
. Noting that

n
i=1
(x
i
x) = 0 we have
y, g
x
) = y, x xz) =
n

i=1
_
y
i
x
i
xy
i
_
=
n

i=1
_
y
i
x
i
xy
i
y(x
i
x)
_
=
n

i=1
(x
i
x)(y
i
y);
call this last quantity S
xy
(it is the covariance of X and Y ). Then

W
(y) = yg
1
+
S
xy
S
2
x
g
x
= yz +
S
xy
S
2
x
_
x xz) =
S
xy
S
2
x
x + (y
S
xy
S
2
x
x)z.
This gives us the coefcients a =
Sxy
S
2
x
and b = (y
Sxy
S
2
x
x) and hence the line of best t. With
a little more work it is possible to show that the correlation coefcient r = S
xy
/S
x
S
y
measures
how well the data is described by a linear model. It is always true that [r[ 1 and the closer [r[
is to 1 the better the t.
5. Complex inner product spaces
Now we extend the denition of inner product spaces to complex vector spaces. We need to
relax the assumption that the inner product is symmetric.
Throughout this section we will assume that V is a complex inner product space.
5.1 Denition Let V be a complex vector space. An inner product on V is a map
, ) : V V C
such that for all x, y, z V and , C the following hold:
a) x, y) = y, x);
b) x + y, z) = x, z) + y, z); and,
c) x, x) > 0 whenever x ,= 0.
A complex vector space V is a complex inner product space if it has an inner product.
Here we write for the complex conjugate of the complex number C.
A complex inner product is not bilinear; but it is close to being so. This time by combining
(a) and (b) we obtain
x, y + z) = y + z, x) = y, x) + z, x) = x, y) + x, z).
More generally, we have
5.2 w + x, y + z) = w, y) + w, z) + x, y) + y, z),
for vectors w, x, y, z V and , , , C. Notice that if and are both real numbers
then = and = and , ) is honestly bilinear; thus, the complex inner product is a true
generalization of the real inner product. In general, complex inner products are linear in the rst
variable and conjugate linear in the second.
5.3 Example As with the real case, we have two natural examples of complex inner product
spaces.
10 Algebra (Advanced)
a) Let V = C
n
. An element x C
n
can be thought of as a row vector x = (x
1
, . . . , x
n
).
Given y = (y
1
, . . . , y
n
) V dene
x, y) = x y = (x
1
, . . . , x
n
) (y
1
, . . . , y
n
) = x
1
y
1
+ x
2
y
2
+ + x
n
y
n
,
where x y is the standard dot product on C
n
. It is straightforward to check that this is an
inner product.
b) Suppose that D C and let V = ((D) = f : DC [ f is continuous . If you knew
some complex analysis then it would be easy to check that V is a complex vector space
and that
f, g) =
_
D
f(z)g(z) dz, f, g V,
denes an inner product on V .
Exactly as before (see Lemma 1.4) we obtain:
5.4 Lemma Suppose that x V . Then x, 0) = 0. Consequently, x, x) 0 for all x V .
Note that if x V then x, x) = x, x), so x, x) is real. Therefore, we dene the length
of x to be the real number |x| =
_
x, x). We could prove the CauchySchwartz inequality
by modifying the proof of Theorem 2.1; you might like to try to do this. Rather than do this at
the end of this section we will give a more conceptual proof using orthogonal projections. This
will also give us another proof in the real case.
Recall that if = a + ib C then [[ =

a
2
+ b
2
is the modulus of . Note that [[
2
= .
5.5 Lemma Suppose that x V and C. Then |x| = [[ |x|.
Proof By direct calculation, using (5.2), we see that
|x|
2
= x, x) = x, x) = [[
2
|x|
2
.
Now take square roots of both sides.
As before, say that two vectors x, y V are orthogonal if x, y) = 0. As before, an orthogonal
basis of V is a basis consisting of pairwise orthogonal vectors and an orthonormal basis is a basis
f
1
, . . . , f
n
such that f
i
, f
j
) =
ij
for 1 i, j n
Repeating the argument of Proposition 3.3 word for word (for variation well prove Corol-
lary 3.4 directly, the corresponding result using an orthonormal basis) we obtain:
5.6 Proposition Suppose that W is a subspace of V and that f
1
, . . . , f
m
) is an orthonormal
basis of W and let x W. Then
w =
m

i=1
w, f
i
)f
i
.
Proof As f
1
, . . . , f
m
is a basis of W we can write w uniquely in the form w =

m
i=1

i
f
i
for
some
i
R. Therefore, if 1 j m then by linearity
w, f
j
) =
m

i=1

i
f
i
, f
j
) =
m

i=1

i
f
i
, f
j
) =
j
,
Complex inner product spaces 11
since f
i
, f
j
) =
ij
. Hence,
j
= w, f
j
) as claimed.
With this result in hand it is reasonably straightforward to generalize Theorem 3.5; you will
nd the proof when you do the rst assignment.
5.7 Theorem Assume that W has an orthonormal basis f
1
, . . . , f
m
and let v V . Then
there exists a unique vector w W such that
v, x) = w, x), for all x W;
indeed, w =
m

i=1
v, f
i
)f
i
. Moreover, |v w| < |v x|, whenever x W and x ,= w.
Proof See assignment 1.
Again, we dene the orthogonal projection of v V onto W to be the vector

W
(v) =
m

i=1
v, f
i
)f
i
.
Repeating the proof of Theorem 3.8 we nd that every nite dimensional complex inner product
space also has an orthogonal basis.
5.8 Corollary Every nite dimensional complex inner product space has an orthogonal basis.
Nowwe do something different and use orthogonal projections to prove the CauchySchwartz
inequality for complex inner product spaces. The key fact that we need is obvious from a geo-
metrical viewpoint (at least in the real case). For the proof note that if C then = [[
2
.
5.9 Proposition Suppose that W is a subspace of V . Then |v| |
W
(v)| for all v V .
Proof By the results above we can choose an orthonormal basis f
1
, . . . , f
m
for W. Extend
f
1
, . . . , f
m
to a basis f
1
, . . . , f
m
, v
m+1
, . . . , v
n
of V . By applying the GramSchmidt algo-
rithm to this basis we obtain an orthonormal basis of V of the form f
1
, . . . , f
m
, f
m+1
, . . . , f
n
;
in particular, the rst m elements of this basis are our original orthonormal basis of W.
Let
i
= v, f
i
) for 1 i n. Then v =

n
i=1
v, f
i
)f
i
and
W
(v) =

m
i=1
v, f
i
)f
i
by
Proposition 5.6. Therefore,
|
W
(v)|
2
=
W
(v),
W
(v)) =
m

i=1

i
f
i
,
m

j=1

j
f
j
) =
m

i=1

i
=
m

i=1
[
i
[
2
.
Similarly, |v|
2
=

n
i=1
[
i
[
2

m
i=1
[
i
[
2
= |
W
(v)|
2
, completing the proof.
As promised, this gives us a different proof of the CauchySchwartz inequality for complex
inner product spaces.
5.10 Corollary Suppose that x, y V . Then [x, y)[ |x| |y|.
12 Algebra (Advanced)
Proof If x = 0 then there is nothing to prove, so suppose that x ,= 0. Let W = Cx be the one
dimensional subspace of V spanned by x. Then, by Proposition 5.9 and Lemma 5.5,
|y| |
W
(y)| =
_
_
_
y, x)
x, x)
x
_
_
_ =
[y, x)[
|x|
2
|x| =
[x, y)[
|x|
.
Rearranging this equation gives the result.
Of course, this gives another proof of Theorem 2.1 when V is a real inner product space.
Notice that unlike the case of real inner product spaces, Corollary 5.10 does not lead to a natural
denition of the angle between two vectors x, y V because, in general,
x,y
xy
will be a
complex number.
6. Isometries and inner product spaces
Linear transformations (or, equivalently, matrices) are the maps between vector spaces which
preserve the vector space structure. We now consider those homomorphisms of inner product
spaces which preserve the inner product. Throughout we will consider inner product spaces
which are either real or complex vector spaces.
6.1 Denition Suppose that V and W are (real or complex) inner product spaces. A linear
transformation T : V W is an isometry if it preserves lengths; that is, |T(x)| = |x| for all
x V .
Strictly speaking an isometry should be any map which preserves lengths; however, as far as
were concerned, at least when it comes to vector spaces, all maps are linear.
In fact, isometries are precisely those linear transformations which preserve inner products.
6.2 Proposition Suppose that V and W are inner product spaces and that T : V W is linear
transformation. Then T is an isometry if and only if T(x), T(y)) = x, y) for all x, y V .
Proof One direction is trivial. Suppose that T(x), T(y)) = x, y), for all x, y V . Then, in
particular, |T(x)|
2
= T(x), T(x)) = x, x) = |x|
2
for all x V . Hence, T is an isometry.
Conversely, suppose that |T(x)| = |x| for all x V . Then for any x, y V we have
T(x), T(y)) +T(y), T(x)) = |T(x + y)|
2
|T(x)|
2
|T(y)|
2
= |x + y|
2
|x|
2
|y|
2
= x, y) +y, x).
If V and W are real inner product spaces then this shows that T(x), T(y)) = x, y) so were
done. If V and W are complex inner product spaces then this argument only shows that the
real parts of T(x), T(y)) and x, y) are equal. The same argument applied to T(ix), T(y)) =
iT(x), T(y)) shows that the imaginary parts of T(x), T(y)) and x, y) are equal (here, i =

1). Hence, T(x), T(y)) = x, y) as claimed.


It is often convenient to use that fact that isometries preserve inner products (rather than just
lengths); we will use this fact freely from now on. (By freely what I really mean is that I will
apply Proposition 6.2 automatically from now on without explicitly mentioning that I am doing
so!)
Recall from the Linear Algebra course that if T : V W is a linear transformation then the
kernel of T is ker T = v V [ T(v) = 0 , a subspace of V , and the image of T is imT =
T(v) [ v V , a subspace of W.
Isometries and inner product spaces 13
6.3 Lemma Suppose that T : V W is an isometry. Then ker T = 0 and so V

= imT.
Proof Suppose that x ker T; that is, T(x) = 0. Then 0 = |T(x)| = |x|; so x = 0 by
Lemma 5.4. Hence. ker T = 0.
For the second claim, the RankNullity theorem says that dimV = dimker T + dimimT.
(A more familiar statement for you is probably that dimV = rank T +nullity T; by denition,
rank T is the dimension of the image of T and nullity T is the dimension of the kernel.) There-
fore, dimV = dimimT since ker T = 0. However, two vector spaces are isomorphic if and
only if they have the same dimension, so V

= imT as claimed. (Later we will see that this is a
special case of the rst isomorphism theorem.)
In light of this result we may as well restrict our attention to the isometries T : V V which
map V to V ; by the Lemma such isometries are isomorphisms.
6.4 Theorem Suppose that V is an inner product space and that T : V V is a linear trans-
formation. Then T is an isometry if and only if T maps every orthonormal basis of V to an
orthonormal basis of V .
Proof For the proof, x an orthonormal basis f
1
, . . . , f
n
of V ; then f
i
, f
j
) =
ij
for all
1 i, j n.
Suppose rst that T is an isometry. Then we have T(f
i
), T(f
j
)) = f
i
, f
j
) =
ij
, so that
T(f
1
), . . . , T(f
n
) is also an orthonormal basis of V .
The harder part is the converse. Let x, y V and write x =

n
i=1

i
f
i
and y =

n
j=1

j
f
j
.
Then, by bilinearity and (5.2),
x, y) =
n

i=1

i
f
i
,
n

j=1

j
f
j
) =
n

i=1
n

j=1

j
f
i
, f
j
) =
n

i=1

i
.
Moreover, since T is linear T(x) =

n
i=1

i
T(f
i
) and T(y) =

n
j=1

j
T(f
j
). Therefore,
T(x), T(y)) =
n

i=1

i
T(f
i
),
n

j=1

j
T(f
j
)) =
n

i=1
n

j=1

j
T(f
i
), T(f
j
)) =
n

i=1

i
,
the last equality following because T(f
1
), . . . , T(f
n
) is also an orthonormal basis of V .
Hence, comparing these two equations, T(x), T(y)) = x, y) as required.
Fix an orthonormal basis of V ; in fact, we may as well assume that our basis is the standard
basis of column vectors e
1
, . . . , e
n
for V . Then a linear transformation T corresponds to left
multiplication by the matrix A
T
= (a
ij
) where
T(e
j
) =
n

i=1
a
ij
e
i
;
so the j
th
column of A
T
is the column vector describing the vector T(e
j
) in terms of the standard
basis e
1
, . . . , e
n
of V . (In the Linear Algebra course the matrix A
T
was usually denoted by
[T]
B
B
, where B is the basis e
1
, . . . , e
n
.) This gives a correspondence, T A
T
, between the
set of linear transformations on V and the set of nn matrices. Since we have xed a basis we
can (and do) identify vectors in V with column vectors (relative to our xed orthonormal basis).
We now characterize isometries in terms of matrices.
14 Algebra (Advanced)
6.5 Corollary Suppose that T : V V is a linear transformation and dene the matrix A
T
as
above. Then T is an isometry if and only if the columns of A
T
give an orthonormal basis of V .
Proof With respect to the basis e
1
, . . . , e
n
, the vector T(e
j
) corresponds to the column vector
(a
1j
, . . . , a
nj
)
t
, which is column j of the matrix A
T
.
In other words, the different isometries of V correspond to (permutations of) the orthonormal
bases of V .
A similar argument establishes the same result for the rows of A
T
.
6.6 Corollary Suppose that T is an isometry. Then the rows of A
T
correspond to an orthonor-
mal basis of V .
6.7 Denition Suppose that A = (a
ij
) is a matrix (with real or complex entries). Then the
conjugate transpose of A is the matrix A

= (a
ji
).
In particular, if all of the entries of A are real then A

= A
t
is just the transpose of A.
In general, A

= (A)
t
= A
t
where A = (a
ij
) is the matrix whose entries are the complex
conjugates of the entries of A. Using standard facts about transposes of matrices it follows that
(A

= A, (AB)

= B

, (A+ B)

= A

+ B

and so on; see Tutorials.


6.8 Corollary Suppose that T : V V is a linear transformation. Then T is an isometry if
and only if A

T
A
T
= 1.
Proof As in the proof of Corollary 6.5, the j
th
column of A
T
is the vector T(e
j
). Also, by
Theorem 6.4 the basis T(e
1
), . . . , T(e
n
) is orthonormal. Therefore, T(e
i
), T(e
j
)) =
ij
.
Rewriting this equation in terms of the matrix A
T
= (a
ij
) this becomes

ij
= T(e
i
), T(e
j
)) =
n

k=1
a
ki
e
k
,
n

l=1
a
lj
e
l
) =
n

k=1
a
ki
a
kj
=
n

k=1
a

jk
a
ki
.
In other words, the (j, i)
th
entry of A

T
A
T
is
ij
. Hence, A

T
A
T
= 1.
We can rephrase this argument so as to give a clearer explanation for what is really happening.
If x = (x
1
, . . . , x
n
)
t
and y = (y
1
, . . . , y
n
)
t
are two column vectors when x, y) = y

x
remember that we are now identifying elements of V with columns vectors (with respect to the
basis e
1
, . . . , e
n
). In the proof above, the column vector A
T
e
i
is the i
th
column of A
T
and the
row vector (A
T
e
j
)

= e

j
A

T
is the j
th
row of A

T
; so,

ij
= e
i
, e
j
) = T(e
i
), T(e
j
)) = (A
T
e
j
)

(A
T
e
i
) = e

j
A

T
A
T
e
i
is the (j, i)
th
entry of the matrix A

T
A
T
. This is precisely the claim that A

T
A
T
= 1.
7. Normal, Hermitian and unitary matrices
Now that we have a matrix theoretic characterization of isometries we concentrate upon un-
derstanding the corresponding matrices. It turns out to be easier to study a larger class of
matrices and that the additional matrices which arise are also important from the point of
harmonic analysis and quadratic forms (not that well talk about these subjects in this course).
Normal, Hermitian and unitary matrices 15
7.1 Denition Let A be a matrix with complex entries. Then
a) A is normal if AA

= A

A;
b) A is Hermitian if A = A

; and,
c) A is unitary if AA

= 1.
Let A be a matrix with real entries. Then
a) A is normal if AA
t
= A
t
A;
b) A is symmetric if A = A
t
; and,
c) A is orthogonal if AA
t
= 1.
Now, T : V V is an isometry if and only if A
T
is a unitary matrix by Corollary 6.8; so
these are the matrices which we care about the most. Notice that every unitary matrix is normal
(if A is unitary then A
1
= A

so AA

= 1 = A

A). Similarly, every Hermitian matrix is also


normal. Finally, observe that if A is normal then AA

= A

A is Hermitian.
The main result that we want to prove is that if N is a normal matrix then there exists a unitary
matrix T such that T
1
NT = T

NT is a diagonal matrix; in particular, this result will apply


to unitary matrices and hence to isometries. Now we know that every matrix is conjugate to its
Jordan canonical form (since we are working over the complex eld C which is algebraically
closed), so what we really need to show is that V has an orthogonal basis which consists of
eigenvectors for N.
Before we start the proof we backtrack and explain the real signicance of the conjugate
transpose of a matrix.
7.2 Lemma Suppose that A is any n n matrix. Then Ax, y) = x, A

y) for all x, y V .
Proof Recalling our identication of elements of V with columns vectors, if x, y V then
Ax, y) = y

(Ax) = (A

y)

x = x, A

y).
7.3 Proposition Suppose that N is a normal matrix and that v V is an eigenvector of N with
eigenvalue . Then N

v = v; that is, v is an eigenvector of N

with eigenvalue .
Proof First consider the case when = 0; that is, Nv = 0. Then N

Nv = 0 so v is also an
eigenvector of N

N = NN

(since N is normal). Therefore, by Lemma 7.2,


0 = NN

v, v) = N

v, N

v) = |N

v|
2
;
whence N

v = 0 by Lemma 5.4.
Nowconsider the general case where Nv = v (and is not necessarily 0). Let

N = NI
n
,
where I
n
is the n n identity matrix (and n = dimV ). Then

N

= N

I
n
and

N

N

= (N I
n
)(N

I
n
) = NN

N N

+ I
n
= N

N N N

+ I
n
= (N

I
n
)(N I
n
) =

N


N.
So,

N is also normal. Now,

Nv = Nv v = 0 since v is a eigenvector of N. Therefore,

v = 0 by the rst paragraph of the proof. Expanding this equation we nd that N

v = v as
we wanted to show.
7.4 Corollary a) Suppose that A is a Hermitian matrix and that is an eigenvalue of A.
Then R.
16 Algebra (Advanced)
b) Suppose that A is a unitary matrix and that is an eigenvalue of A. Then [[ = 1.
Proof See tutorials.
In particular, notice that this says that all of the eigenvalues of a symmetric matrix are real.
Notice that part (b) also follows from the fact left multiplication by a unitary matrix is an
isometry.
7.5 Lemma Suppose that N is a normal matrix and that v and w are two eigenvectors of N
with eigenvalues and respectively. Then v, w) = 0 unless = .
Proof Suppose that ,= . Then using Lemma 7.2, once again, and Proposition 7.3 we have
v, w) = Nv, w) = v, N

w) = v, w) = v, w).
As ,= this gives the result.
Finally, we need a more technical Lemma.
7.6 Lemma Suppose that N is a normal matrix and that (N I
n
)
k
v = 0 for some nonzero
vector v V , some C (or R) and an integer k 1. Then (N I
n
)v = 0; that is, v is
a eigenvector of N.
Proof We rst make a series of reductions. As in the proof of Proposition 7.3 we may assume
that = 0 by replacing N with

N = N I
n
if necessary; thus we have that N
k
v = 0.
Next, since N
k
v = 0 we certainly have (N

)
k
N
k
v = 0. However, N

N = NN

so
(N

)
k
N
k
= (N

N)
k
and we have (N

N)
k
v = 0. Let A = N

N; then A is Hermitian.
Assume, for the moment that we know that Av = 0. Then
0 = Av, v) = N

Nv, v) = Nv, Nv) = |Nv|


2
by Lemma 7.2; so Nv = 0 by Lemma 5.4.
Hence, it is enough to prove that if A is an Hermitian matrix such that A
k
v = 0 then Av = 0.
Choose m large enough so that 2
m
k. Then A
2
m
v = 0. Now, A is Hermitian so A = A

and, consequently, A
2
m
= (A
2
m1
)
2
= (A

)
2
m1
A
2
m1
= (A
2
m1
)

A
2
m1
. Therefore, by
Lemma 7.2 once again,
0 = A
2
m
v, v) = (A
2
m1
)

A
2
m1
v, v) = A
2
m1
v, A
2
m1
v).
So, A
2
m1
v = 0 by Lemma 5.4. If m > 1 then we can repeat this argument and eventually we
will nd that Av = 0, as we wished to show.
We are now ready to prove the main result of this section.
7.7 Theorem Suppose that N is a normal matrix. Then there exists a unitary matrix T such
that T

NT = T
1
NT is a diagonal matrix.
Normal, Hermitian and unitary matrices 17
Proof Let J = P
1
NP be the Jordan canonical form of N and let v
1
, . . . , v
n
be the cor-
responding Jordan basis. (Thus, v
i
= Pe
i
, for all i, and J is the matrix which describes the
linear transformation determined by N relative to the basis v
1
, . . . , v
n
.) Now suppose that J
contains a Jordan block of the form
_
_
_
_
_
_
_
1 0 0
0 1 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 1
0 0
_
_
_
_
_
_
_
and let v
i
1
, . . . , v
im
be the corresponding basis elements (so Nv
ia
= v
ia
+ v
i
a1
, with the
understanding that v
i
0
= 0). Then (N I
n
)
m
v = 0 for all v v
i
1
, . . . , v
im
). Therefore,
(N I
n
)v = 0 by Lemma 7.6; that is, Nv = v and V is an eigenvector of N. However, this
means that every vector in v
i
1
, . . . , v
im
) is an eigenvector of N; so m = 1 and this is a Jordan
block of size 1. Consequently, all of the Jordan blocks in J have size one; in other words, J is
a diagonal matrix and so V has a basis which consists of eigenvectors for N.
For each let V

be the subspace of v consisting of eigenvectors; that is,


V

= v V [ Nv = v .
Then V =

, where runs over the eigenvalues of N. By Lemma 7.5 if v V

and
w V

, for ,= , then v, w) = 0; so the eigenspaces V

and V

are automatically orthogonal


to each other. Moreover, using GramSchmidt we can nd an orthonormal basis for each V

.
Hence, we can nd an orthonormal basis f
1
, . . . , f
n
for V which is built up from the orthonor-
mal bases of the V

. By construction, Nf
i
=
i
f
i
for some
i
C. Let T be the matrix such
that f
i
= Te
i
; in other words the i
th
column of T is the column vector for f
i
(with respect to the
standard basis e
1
, . . . , e
n
). Then T
1
NT is the diagonal matrix diag(
1
, . . . ,
n
).
Finally, it remains to observe that T is a unitary matrix by Theorem 6.4 since it maps the
orthonormal basis e
1
, . . . , e
n
to the orthonormal basis f
1
, . . . , f
n
. Consequently, we also
know that T
1
= T

by Corollary 6.8. Hence, T


1
NT = T

NT is diagonal as we wished to
show.
As the proof shows, the diagonal entries of T

NT are just the eigenvalues of N.


If we now specialize to the case of isometries we see that a linear transformation T : V V
is an isometry if and only if there exists an orthonormal basis f
1
, . . . , f
n
of V together with
some (complex) numbers
1
, . . . ,
n
of norm 1 such that T(v
i
) =
i
v
i
, for i = 1, . . . , n.
Application: quadratic surfaces in R
n
.
A general quadratic surface in R
n
is given by the equation
n

i,j=1
a
ij
x
i
x
j
+
n

i=1
b
i
x
i
+ c = 0,
where a
ij
, b
i
, c are real constants. We can certainly assume that a
ij
= a
ji
for all i and j. This
means that the corresponding matrix A = [a
ij
] is symmetric, A = A
t
. In particular, if n = 2
then taking x = x
1
and y = x
2
we get the equation of a quadratic curve in R
2
,
a
11
x
2
+ 2a
12
xy + a
22
y
2
+ b
1
x + b
2
y + c = 0.
18 Algebra (Advanced)
By Corollary 7.4(a), all the eigenvalues
1
, . . . ,
n
of the matrix A are real. Furthermore, by
the real version of Theorem 7.7, there exists an orthogonal matrix T with real entries such that
the matrix D := T
1
AT is diagonal with the diagonal entries
1
, . . . ,
n
.
Denote by x the vector-column with the entries x
1
, . . . , x
n
. Then the quadratic part of the
equation of the surface can be written as
n

i,j=1
a
ij
x
i
x
j
= x
t
Ax.
Introduce new coordinates y
1
, . . . , y
n
in R
n
by setting y = T
1
x, where y denotes the vector-
column with the entries y
1
, . . . , y
n
. Then x = Ty, and in the new coordinates the quadratic part
of the surface takes the form
x
t
Ax = y
t
T
t
ATy = y
t
Dy =
n

i=1

i
y
2
i
.
This brings the equation of the surface to the simpler form
n

i=1

i
y
2
i
+
n

i=1

i
y
i
+ c = 0,
where the
i
are real constants. The surface is said to be nondegenerate if
i
,= 0 for all i. In
this case, using the shifts y
i
:= y
i
+
i
and a possible renumbering of the coordinates we can
bring the equation of the surface to the canonical form
p

i=1
y
2
i
a
2
i

i=p+1
y
2
i
a
2
i
= 1, or
p

i=1
y
2
i
a
2
i

i=p+1
y
2
i
a
2
i
= 0,
where the a
i
are positive numbers and p 0, 1, . . . , n. For the rst equation, if p = 0 then the
surface is empty. If p = n the surface is an ellipsoid and for the remaining values of p the surface
is a hyperboloid. For the second equation, the corresponding surface is a cone which degenerates
into a point if p = 0 or p = n.
The degenerate surfaces (i.e. where
i
= 0 for certain i) include paraboloids, cylinders of
various kinds as well as plains or pairs of plains.
In particular, for n = 2 the quadratic equations determine ellipses, hyperbolas, parabolas,
lines, pairs of lines, points or the emptyset. The detailed analysis of possible canonical forms
in the cases n = 2 and n = 3 is left to the reader.
The denition of a group 19
PART II: GROUP THEORY
8. The denition of a group
A group is nothing more than a set which is equipped with one operation, such as addition or
multiplication. Groups are ubiquitous in mathematics and also in the real world. They were rst
studied in detail by the French mathematician Evariste Galois who used them to show that there
is no general formula for the solution of a polynomial equation of degree 5 or higher which uses
only surds.
The denition of a group is very abstract but, as we shall see, we already know may exam-
ples. Many new examples also arise when we look at the symmetry properties of objects from
geometry.
8.1 Denition A binary operation on a set X is a map : X XX; (x, y)x y.
For example, addition, multiplication and division are usually binary operations. Note, how-
ever, that we do have to take some care here: if X = R then addition and multiplication are
both binary operations on X but division is not because we cannot divide by zero!
A group is a set which comes with a special type of binary operation.
8.2 Denition A group is a set G together with a binary operation,
: GGG; (g, h)g h,
such that:
a) (associative) If g, h, k G then (g h) k = g (h k);
b) (identity) there exists an element e G such that e g = g = g e, for all g G; and,
c) (inverses) for each g G there is an element g

G such that g g

= e = g

g.
Note that implicit in the denition is the assumption that g h G, for all g, h G.
An element e G which satises property (b) is called an identity element of G. If g G
then an element g

G such that g g

= e = g

g is called an inverse of g. Note that we


are assuming only that G has at least one identity and that each element of G has at least one
inverse; we will shortly see that these elements are unique.
The following examples (and others) were discussed in more detail in lectures. You should
check that all of these examples are groups; in particular, you need to ask yourself what the
identity element is in each group and what the inverses are.
8.3 Denition A group G is abelian (or commutative) if g h = h g for all g, h G.
Abelian groups are the simplest sorts of groups around; however, even here there are still
some nontrivial questions to be answered.
8.4 Examples a) Let G = Z with being multiplication. Then G is not a group because,
for example, 0 does not have an inverse. Note, however, that 1 is an identity element.
b) Let G = Z with the operation being addition. This time 0 is an identity element (since
0 + n = n = n + 0 for all n Z) and the inverse of n Z is n (since n + (n) =
0 = (n) +n). As addition is associative this means that Z is a group (when we take the
operation to be addition). Notice that Z is abelian.
c) Let G = Q

be the set of nonzero rational numbers with the operation of multiplication.


This time G is an abelian group.
20 Algebra (Advanced)
d) Let G be the set of all n n matrices with entries in a eld F under addition. Again G is
an abelian group.
e) Let G be the set of all n n matrices with entries in a eld F under multiplication. This
time Gis not a group for the same reason as in example (a): the zero matrix does not have
an inverse.
f) Let G = GL
n
(F) = A M
n
(F) [ det A ,= 0 be the set of invertible n n matrices
under multiplication (of matrices). Then G is a group. It is also our rst example of a
nonabelian group.
g) Let G = O
n
(R) be the set of all n n orthogonal matrices, with the operation of multi-
plication. Then G is a nonabelian group.
h) Let V be a vector space, with the operation of addition (of vectors) so we are forget-
ting about scalar multiplication. Then V is a group.
i) Let V be a vector spaces and let G = GL(V ) be the set of all isomorphisms from V
to V . Then G becomes a group under composition of maps: f g = f g. Actually, as
an isomorphism from V to V corresponds to an invertible n n matrix, where n is the
dimension of V , this the same group as in (f) above.
j) For a positive integer n, let Z
n
= 0, 1, . . . , n 1 with the operation being addition
modulo n. (Recall that for any integer a there is a unique integer r such that a = kn + r
and 0 r < n. Dene a = r; it is also common to write a r (mod n). Then the
operation on Z
n
is a b = a + b.) Then Z
n
is the cyclic group of order n.
k) For a positive integer n, let = exp(2i/n) be a primitive n
th
root of unity in C; that
is, =
n

1. Let C
n
=
a
[ a Z = 1, ,
2
, . . . ,
n1
, with the operation being
multiplication (of complex numbers). Then C
n
is (also?) the cyclic group of order n.
If you compare the multiplication tables for Z
n
and C
n
you will see that they are the
same. Later, we will see that these two groups are isomorphic (via the map a
a
).
l) Fix a positive integer n and let Sym(n) be the set of all permutations of the set n =
1, 2, . . . , n. A permutation of n is nothing more than a bijection from n to n: to each
integer i n we assign another integer j n. In particular, Sym(n) contains n! elements.
The group operation on Sym(n) is just composition of maps; however, in order to make
the multiplication of permutations read more naturally I want to dene composition as
follows: if , Sym(n) then is the permutation of n given by
()(i) = ((i)), for all i n.
(Usually, composition works from the left: (f g)(x) = f(g(x)). For permutations many
authors dene composition from the right. In fact, we should really write the maps on the
right as well: (i)() = ((i)), but we wont do this.) As before, it is easy to check that
this operation makes Sym(n) into a group.
The most obvious way of specifying the elements of Sym(n) is to use two line notation;
for example, let be the element of Sym(5) given by (1) = 2, (2) = 3, (3) = 1,
(4) = 5 and (5) = 4. Then can be described more compactly as =
_
1 2 3 4 5
2 3 1 5 4
_
.
The advantage of the convention that we are using for the operation in Sym(n) is that
in order to calculate the product of two permutations we just read the equations from left
to right. For example,
_
1 2 3 4 5
2 3 1 5 4
_

_
1 2 3 4 5
5 4 3 2 1
_
=
_
1 2 3 4 5
4 3 5 1 2
_
.
There is also a more compact notation for permutations known as the cycle notation;
using this notation we would write as = (1, 2, 3)(4, 5). You read this, cyclically,
The denition of a group 21
from left to right: the rst cycle (1, 2, 3) says that sends 1 to 2, 2 to 3, and 3 back to
1; the second cycle (4, 5) says that interchanges 4 and 5. Note that in cycle notation
(1, 2, 3) = (2, 3, 1) = (3, 1, 2); all that matters is the order of the numbers up to a cyclic
permutation. Using cycle notation the product of the two permutations above becomes
(1, 2, 3)(4, 5) (1, 5)(2, 4) = (1, 4)(3, 5, 2); again to work this out we read from left to
right.
Finally, notice that in cycle notation
_
1 2 3 4 5
5 4 3 2 1
_
= (1, 5)(2, 4). Strictly speaking we
should write (1, 5)(2, 4)(3), where the (3) indicates that 3 is sent to 3 (i.e. it is xed);
however, we normally just omit the xed points of the permutations.
Warning: Armstrong reads his permutations differently.
m) Generalizing the last example, let X be any set and let Sym(X) be the set of bijections
from X to X permutations of X. The argument of the last example shows that
Sym(X) is a group.
n) Let be a graph; in other words, consists of a set of vertices 1 and a set of edges
c = (x, y) [ x, y 1 . A graph automorphism of is a bijection f : 1 1 which
preserves the edges of ; that is, if (x, y) c then (f(x), f(y)) c. The symmetry group
of is the set G

of all graph automorphisms, where the group operation is composition


of maps. Notice that if X has no edges then G

= Sym(X).
o) Suppose that n 3 and let P
n
be a regular ngon; that is, P
n
is the graph with vertices
1, 2, . . . , n and edges joining n and 1, and i and i + 1 for 1 i < n. (For P
n
to be
a regular ngon we also require that all of the edges in P
n
have the same length.) For
example, P
3
is an equilateral triangle, P
4
is a square, P
5
is a pentagon, P
6
is a hexagon
and so on. See Example 10.7 for P
8
and a detailed analysis of D
8
.
The symmetry group of P
n
is known as the dihedral group D
n
(of order 2n). An element
of D
n
is completely determined by where it sends the vertices labelled 1 and 2. If 1 is
mapped to the vertex i then 2 must be sent to either i+1 or i1 (interpret i1 modulo n);
hence, the group D
n
contains exactly 2n elements.
Notice that the subset of D
n
consisting of the rotations is also a cyclic group of
order n.
p) Consider the following four matrices with complex entries
1 =
_
1 0
0 1
_
, I =
_
i 0
0 i
_
, J =
_
0 1
1 0
_
, K =
_
0 i
i 0
_
.
One easily veries the relations
I
2
= J
2
= K
2
= 1, IJK = 1
which imply that the set of 8 matrices Q
8
= 1, I, J, K is closed under matrix
multiplication and forms a group called the quaternion group or Hamilton group. (The lower
case letters i, j, k are often used instead of I, J, K).
As the course unfolds it is a good idea to ask what the theorems we prove say about the
various examples above.
The examples show that we need to be a little careful with our notation: the operation is
reminiscent of multiplication; however in many examples the operation is nothing like multi-
plication. Shortly we will drop this dot notation and simply write gh rather than g h. This is
a matter of convenience only: when we are talking about an abstract group (or, if you prefer,
an arbitrary group) groups we need a notation for our operation. When we are talking about
specic examples then the operation could be addition, multiplication, or possibly something
quite different (as in the example of braid groups).
22 Algebra (Advanced)
8.5 Proposition Suppose that G is a group. Then:
a) The identity element of G is unique; that is, if e and e

are elements of G such that


g e = g = e g and g e

= g = e

g, for all g G, then e = e

.
b) Each element of G has a unique inverse; that is, if g G and there exist elements
g

, g

G such that g g

= e = g

g and g g

= e = g

g then g

= g

.
Proof
a) Suppose that there exist elements e, e

G as the statement of the Proposition. Then


e = e e

= e

, the rst equality following because e

is a (right) identity element and the


second one because e is a (left) identity.
b) Suppose that we have elements g

and g

as above. Then
g

= g

e = g

(g g

) = (g

g) g

= e g

= g

,
where we have used the facts that the binary operation is associative, e is an identity
element, g

is a (right) inverse to g, and that g

is a (left) inverse of g.
Now that we know that identity elements and inverses are unique we adopt the following
convention.
8.6 Notation If G is a group we let 1
G
(or simply 1 when G is understood), denote the identity
element of G. If g G then we write g
1
for the inverse of g.
With this notation we can now rewrite the group axioms for identity elements and inverses as
the familiar looking equations 1 g = g = g 1 and g g
1
= 1 = g
1
g, respectively.
We are really using a multiplicative (and exponential) notation for our group operation: we
could equally well use an additive one. The only reason for preferring a multiplicative notation
over an additive one is that addition tends to be commutative whereas multiplication is often
not commutative (for example consider matrix multiplication).
A word of warning here: we use this shorthand notation because it is very convenient; how-
ever, you should not forget that this notation is shorthand. In particular, 1
G
= 1 is the identity
element of G and not the number one; indeed, 1
G
could well be the number zero (for example,
this would be the case if G = Z and the operation were addition). Similarly, g
1
is the inverse
of g (and not
1
g
, even when this does make sense). Exactly how g
1
is described will depend on
the group in question (for example, if G = Z then g
1
is the negative g of g G).
From this point on we will also (mostly) drop the notation; so rather than gh we will simply
write gh.
8.7 Lemma Suppose that g, h G. Then (g
1
)
1
= g and (gh)
1
= h
1
g
1
.
Proof As gg
1
= 1 = g
1
g, the rst equality is obvious (as (g
1
)
1
is the unique element of G
which satises this equation). For the second claim note that
(gh)(h
1
g
1
) = g(hh
1
)g
1
= g 1 g
1
= gg
1
= 1.
Similarly, (h
1
g
1
)(gh) = 1. As inverses are uniquely determined, (gh)
1
= h
1
g
1
as
claimed.
We end this section with some more notation.
Subgroups 23
8.8 Denition Suppose that G is a group.
a) If g G then the order [g[ of g is the smallest positive integer such that g
n
= e, if such n
exists. Otherwise, g is of innite order.
b) The order [G[ of G is the number of elements in G.
c) We say that G is a nite group if it has nite order.
Given the nomenclature, it is natural to ask whether there is a relationship between the order
of a group and the possible orders of its elements. As a challenge, try and work out what the
connection is (see what happens in the examples above with [G[ nite).
9. Subgroups
Now that we know that the inverse of an element is uniquely determined for any integer n we
also dene
g
n
=
_

_
g g . . . g, (n times), if n > 0,
1, if n = 0,
g
1
g
1
. . . g
1
, (n times), if n < 0.
At the risk of boring you, once again this is just a convenient shorthand and the meaning of g
n
will depend upon the particular example we have in mind (for example, if G = Z then g
n
is
actually ng = g + g + + g). We now have the easy Lemma.
9.1 Lemma Suppose that G is a group, m and n are integers and that g G. Then g
m+n
=
g
m
g
n
and (g
m
)
n
= g
mn
.
Proof See tutorials.
If g G let g) = g
n
[ n Z; so g) is a subset of G. In fact, g) is also a group
in its own right. First note that the operation on G gives a operation on g) by restriction
since g
m
g
n
= g
m+n
by the Lemma. Also, 1 = g
0
g), so g) has an identity element.
Finally, if g
n
g) then g
n
g), so every element of g) has an inverse in g) (since
g
n
g
n
= 1 = g
n
g
n
, by the Lemma once again). We say that g) is the subgroup of G
generated by g.
9.2 Denition Suppose that G is a group. A subgroup of G is any nonempty subset of G which
is itself a group, where the operation on H is the restriction of the operation on G to H. If H is
a subgroup of G then we write H G.
Every group always has at least two subgroups; namely, 1
G
and G itself. A subgroup H
of G is nontrivial if H ,= 1
G
and it is proper if H ,= G. The interesting subgroups are the
nontrivial proper subgroups.
We saw above that g) is subgroup of G whenever g G. Further, if g ,= 1
G
then g) is a
nontrivial subgroup of G. If G = g) then we say that G is a cyclic group; cyclic groups are
the simplest types of all possible groups. For example, Z and Z
n
are both cyclic groups; in fact,
later we will see that every cyclic group is isomorphic to one of these.
The rst problem that arises when considering subgroups is that a priori if a, b H then
there is no reason to expect that a b H. The next problem is that the identity element of H
may not be the same element as the identity element of G. Certainly, if 1
G
H then it must
be true that 1
G
= 1
H
(since identity elements are unique); however, we have not assumed that
24 Algebra (Advanced)
1
G
H. Finally, the same problem arises when we consider inverses: the inverse of a H
inside H could be different to the inverse of a in G.
In the statement of the next result for an element h H we write h
1
H
and h
1
G
for the inverse
of h when considered as an element of H and of G respectively.
9.3 Proposition Suppose that H is a subgroup of G. Then 1
H
= 1
G
and h
1
H
= h
1
G
for all
h H.
Proof First consider 1
H
. We know that 1
H
1
H
= 1
H
; therefore, 1
1
H
= 1
H
(in either G or H
since both have the same operation). Therefore, 1
G
= 1
H
1
1
H
= 1
H
1
H
= 1
H
.
The result for the inverses is now clear: inside H we have hh
1
H
= 1
H
= h
1
H
h. However, by
the rst paragraph 1
H
= 1
G
, so we can rewrite this equation as hh
1
H
= 1
G
= h
1
H
h. Therefore,
h
1
H
= h
1
G
because inverses are unique by Proposition 8.5.
Given this result we can now drop the subscripts on the identity and inverse elements of
subgroups and unambiguously write 1 and h
1
for the identity element and inverse elements.
Our immediate goal is to understand when a subset of a group is actually a subgroup.
9.4 Lemma Suppose that H is a nonempty subset of G. Then H is a subgroup of G if and
only if the following two conditions hold:
a) if a, b H then ab H; and,
b) if a H then a
1
H
Proof If H is a subgroup of G then both of these conditions are satised by denition.
Conversely, suppose that (a) and (b) are true. First, by (a) the binary operation on H does
restrict to give an operation on H. Further, as the operation on G is associative, it is still
associative when considered as an operation on H. Next, because H ,= we can nd an
element a H. By condition (b), we know that a
1
H; in particular, element of H has an
inverse in H. Finally, since H is closed under multiplication, 1 = a a
1
is also an element
of H; so H has an identity element. Hence, H is a subgroup of G.
When condition (a) holds we say that H is closed under multiplication. Similarly, H is closed
under the taking of inverses if it satises (b).
9.5 Proposition Suppose that G is a nite group. Then a nonempty subset H is G is a sub-
group of G if and only if H is closed under multiplication.
Proof Again, if H is a subgroup of H then it is closed under multiplication, so there is nothing
to prove here.
Conversely, suppose that H is closed under multiplication. By the Lemma, in order to show
that H is a subgroup of G it is enough to show that a
1
H whenever a H. By assumption
if a H then a
n
= a a . . . a H whenever n 1. However, G is a nite group so it must
be true that a
m
= a
n
for some m > n (otherwise a, a
2
, a
3
, . . . would be an innite subset of
G). Therefore, multiplying by a
n1
(inside G), we see that a
mn1
= a
1
. However, m > n
so that means that a
1
= a
mn1
H. (Note that if mn 1 = 0 then a
1
= 1 so a = 1; so
if a ,= 1 then mn 1 1.) This is what we needed to show, so H is a subgroup as claimed.
We now come to the main result which characterizes when a subset of a group is a subgroup.
Generators 25
9.6 Theorem (The Subgroup Criterion) Suppose that Gis a group and that H is a nonempty
subset of G. Then H is a subgroup of G if and only if ab
1
H for all a, b H.
Proof If H is a subgroup of G then, in particular, it is a group so ab
1
H whenever a, b H.
Conversely, suppose that ab
1
H for all a, b H. Then 1 = bb
1
H, taking a = b.
Consequently, if b H then b
1
= 1 b
1
is also in H; hence H is closed under inverses.
Finally, since b
1
H we see that ab = a(b
1
)
1
H; so H is closed under multiplication.
Thus, H is a subgroup by Lemma 9.4.
10. Generators
Our current goal, which will occupy us for most of the rest of this course, is to understand
the structure of groups; that is, how groups are built up out of smaller groups.
10.1 Proposition Suppose that G is a group. Then the intersection of an arbitrary collection
of subgroups of G is again a subgroup of G.
Proof Let H
i
[ i I be a family of subgroups of G, for some indexing set I. We need to
show that H
I
=

iI
H
i
is also a subgroup of G. Suppose that a, b H
I
. Then a, b H
i
for
all i I. Therefore, by the subgroup criterion (Theorem 9.6), ab
1
H
i
for all i I. Hence,
ab
1
H
I
, so H
I
is a subgroup by Theorem 9.6.
The proof of the last result is straightforward because we have the subgroup criterion to work
with. Notice, however, that the result is very general because we are not assuming anything
about the set I which indexes the subgroups. This is crucial for the next denition, which
otherwise would not make sense.
10.2 Denition Suppose that X G. Then the subgroup X) of G generated by X is the
intersection of all of the subgroups of G which contain X; that is,
X) =

XHG
H.
Recall that we write H G to indicate that H is a subgroup of G.
10.3 Corollary Suppose that X G. Then X) is the smallest subgroup of G which con-
tains X.
Proof By denition, X) is contained in every subgroup of G which contains X.
This Corollary is really just a restatement of the denition. The content of the result is
that there is a (unique) smallest subgroup of G which contains X. A priori, it is not clear
that there is a (unique) smallest subgroup of G which contains X; however, this follows from
Proposition 10.1 because if H
1
and H
2
are two subgroups containing X then H
1
H
2
is another
such subgroup.
In the last section we dened g) = g
n
[ n Z, for g G. It is not immediately apparent
that this notation agrees with Denition 10.2; however, it does.
26 Algebra (Advanced)
10.4 Proposition Let G be a group.
a) If g G then the subgroup of G generated by g consists of the elements g
n
[ n Z.
b) If X G then X) = x

1
1
x

2
2
. . . x

k
k
[ x
i
X and
i
= 1 for 1 i k, k 1
Proof We prove only part (a) and leave (b) to the tutorials.
Let H be the subgroup of G generated by g. We have already seen that g) is a subgroup,
and certainly g g), so H g) by denition. Conversely, g H so g
n
H, for all n 1,
since H is closed under multiplication. Also, g
1
H because H is closed under the taking of
inverses: so g
n
H, for all n 1. Finally, 1 = g
0
H. Hence, g) H and so H = g) as
required.
10.5 Denition A group G is cyclic if G = g).
For example Z = 1) and Z
n
= 1) are both cyclic groups.
10.6 Proposition Suppose that G = g) is a cyclic group and that g has nite order m = [g[.
Then [G[ = [g[ and G = 1, g, . . . , g
m1
.
Proof By the proposition G = g) = g
n
[ n Z. Further, if n Z then n = km + r for a
unique integer r with 0 r < m; therefore,
g
n
= g
km+r
= g
km
g
r
= (g
m
)
k
g
r
= g
r
by Lemma 9.1. Hence, g) = 1, g . . . , g
m1
.
It remains to show that [G[ = m; that is, that the elements 1, g, . . . , g
m1
are all distinct.
Suppose that g
r
= g
s
for some integers r and s with 0 r < s < m. Then g
sr
= 1; however,
this is nonsense because 0 s r < m and [g[ = m. Hence, the result.
10.7 Example We close this section by looking at the example of the dihedral group of order 16
in detail; this is the symmetry group G of the octagon:
1 2
3
4
5 6
7
8
Notice that G has exactly 16 elements because the vertex labelled 1 can be sent to any of the 8
vertices; after this the vertex 2 must be connected to 1, so it can only be sent to one of the two
adjacent vertices. Once we have specied where 1 and 2 go the permutation of the vertices is
completely determined; so [G[ = 2 8 = 16 as claimed.
The last paragraph gave an explicit description of the elements of G; we now describe them
in terms of two generators of G. Let r be a clockwise rotation through 2/8 radians; so r =
(1, 2, 3, 4, 5, 6, 7, 8) as a permutation of the vertices. Then r has order 8 this is clear whether
we think of r as a permutation or geometrically; therefore,
r) = 1, r, r
2
, r
3
, r
4
, r
5
, r
6
, r
7

Generators 27
is a subgroup of G of order 8. Next let s be the reection in the line that bisects the edges
joining 1 and 2, and 5 and 6; as a permutation of the vertices s = (1, 2)(3, 8)(4, 7)(5, 6).
Then s
2
= 1 so s is an element of order 2. Therefore, s) = 1, s is a subgroup of G of
order 2.
I claim that G = r, s). (Strictly speaking I should write G = r, s); as a general rule Ill
omit such extraneous brackets.) Certainly,
r, s) 1, r, r
2
, r
3
, r
4
, r
5
, r
6
, r
7
s, sr, sr
2
, sr
3
, sr
4
, sr
5
, sr
6
, sr
7
.
If we can show that these elements are all distinct then well be done because [G[ = 16
and [r, s)[ 16 with equality if and only if G = r, s). Geometrically it is clear that all
of these elements are different because the powers of r are precisely the rotations of the oc-
tagon, whereas the elements of the form sr
b
the rotations followed by a twist. (It is perhaps not
immediately obvious that every automorphismof the octagon is of one of these forms; however,
this must be the case because [G[ = 16).
We can also see algebraically that these elements are distinct. First, the powers of r are
distinct since r has order 8. Next, if r
a
= sr
b
for some a and b with 0 a, b < 8 then s = r
ba
;
so 1 = s
2
= r
2(ba)
so b a = 4 and s = r
4
, which is nonsense! Finally, if sr
a
= sr
b
, for a
and b as before, then r
a
= r
b
so a = b.
We have now shown that G = r, s); however, this is a little curious because r, s) also
contains elements like r
5
s
3
r
2
s
7
r which does not seem to appear in the list above. The reason
for this is the following: I claim that srs = r
1
. Again, we can see this geometrically or by
multiplying out the permutations; Ill leave the geometrical argument to you. The permutation
calculation is the following:
srs = (1, 2)(3, 8)(4, 7)(5, 6) (1, 2, 3, 4, 5, 6, 7, 8) (1, 2)(3, 8)(4, 7)(5, 6)
= (1, 8, 7, 6, 5, 4, 3, 2) = r
1
.
(Recall that we read permutations from left to right.) Hence, srs = r
1
; or, equivalently (by
multiplying on the left by s and using the fact that s
2
= 1), rs = sr
1
. In other words, whenever
we have an s to the right of an r we can move it to the left by changing r into r
1
. It is now easy
to see by induction on b that r
b
s = sr
b
, for all b with 0 b < 8. Therefore, the expression
above becomes
r
5
s
3
r
2
s
7
r = (r
5
s)(r
6
s)r = (sr
5
)(sr
6
)r = s(r
3
s)r
2
r = s(sr
3
)r
3
= 1,
noting that s
2
= 1 and r
8
= 1 so r
a
= r
8k+a
and s = s
2k+1
, for any a, k Z.
In fact, the multiplication in G is completely determined by the relations r
8
= 1, s
2
= 1
and rsr = r
1
. (In general, a relation says that some word in the generators is equal to 1.)
Because of this we write
G = r, s [ r
8
= 1, s
2
= 1 and srs = r
1
) .
This is called dening a group by generators and relations which we will come to towards
the end of the course. To see that the relations determine the multiplication in G rst suppose
that b > 0. Then we have
sr
b
s = (srs)(sr
b1
s) = r
1
(sr
b1
s) = r
1
r
(b1)
= r
b
,
where the second last equality follows by induction on b. Therefore, just as before, the relations
in G (i.e. r
8
= 1, s
2
= 1 and srs = r
1
) imply that
G = s
a
r
b
[ a = 0 or 1, and 0 b < 8 .
28 Algebra (Advanced)
(With a small amount of effort it is possible to show that the relations also force all of these
elements to be distinct.) Using the last formula we can show how to multiply two arbitrary
elements of G: suppose that 0 a, c < 2 and 0 b, d < 8; then
(s
a
r
b
)(s
c
r
d
) = s
a
(r
b
s
c
)r
d
=
_
s
a
r
b+d
, if c = 0
s
a+1
r
db
, if c = 1.
So, as claimed, the multiplication in G is completely determined by the relations.
Finally, we remark that it is not very hard to generalize this example to an arbitrary dihedral
group D
n
, the symmetry group of the regular ngon. The crucial point in all of the calcu-
lations above was that srs = r
1
; this still true in D
n
if we take r to be a rotation through
2/n and take s to be any reection. Repeating these arguments for D
n
we nd that D
n
=
s
a
r
b
[ a = 0 or 1, and 0 b < n and that D
n
= r, s [ r
n
= 1, s
2
= 1 and srs = r
1
).
11. Cosets
The key to understanding how a subgroup sits within a group is the following denition.
11.1 Denition Suppose that H is a subgroup of G and that g G. Then
a) gH = gh [ h H is the left coset of H in G which contains g;
b) Hg = hg [ h H is the right coset of H in G which contains g.
Most of the time we will work with left cosets; however, any result about left cosets can be
translated into a result about right cosets because
(aH)
1
def
= (ah)
1
[ h H = h
1
a
1
[ h H = ha
1
H = Ha
1
.
The following properties of (left) cosets are both elementary and fundamental.
11.2 Lemma Let H be a subgroup of H.
a) Suppose that h H. Then hH = H.
b) Suppose that a G. Then [H[ = [aH[.
Proof (a) Certainly hH H since H is closed under multiplication. Conversely, if h

H
then h

= h(h
1
h

) hH, so H hH.
(b) Let : HaH be the map (h) = ah. By denition is surjective; it is also injective
because (h
1
) = (h
2
) if and only if ah
1
= ah
2
, so h
1
= h
2
(multiply on the left by a
1
).
Hence, is a bijection and [H[ = [aH[.
11.3 Proposition Suppose that H is a subgroup of G and a, b G. Then the following are
equivalent:
a) aH = bH;
b) aH bH ,= ;
c) a bH;
d) b aH;
e) a
1
b H; and,
f) b
1
a H.
Proof If aH = bH then aH bH ,= ; so (a) implies (b).
Next, suppose that aH bH ,= . Then we can nd an element x aH bH; so, x = ah
a
and x = bh
b
, for some h
a
, h
b
H. Then ah
a
= bh
b
, so a = bh
b
h
1
a
h. Therefore, (b)
implies (c).
Cosets 29
Now suppose that (c) holds; that is, a = bh for some h H. But then b = ah
1
aH so (c)
implies (d).
Next, if (d) is true then b = ah for some h H. Therefore, a
1
b = h H, showing that (d)
implies (e). In turn, (e) implies (f) because b
1
a = (a
1
b)
1
and H is closed under the taking
of inverses.
Finally, suppose that (f) is true. Then h = b
1
a H, so a = bh and
aH = ah

[ h

H = bhh

[ h

H = bh

[ h

H = bH;
the second last equality following because hH = H by Lemma 11.2(a). Hence, (f) implies (a).
An important consequence of the Proposition is that if aH and bH are two cosets then either
aH = bH or aH bH = . The Proposition also gives a way of deciding when two cosets are
equal; in particular, aH = H if and only if a H
If A and B are subsets of a set X we write X = A

B if X = A B and A B = .
11.4 Corollary Suppose that H is a subgroup of G. Then there exist elements g
i
[ i I
in G such that
G =

iI
g
i
H.
Proof As x = x 1 xH we certainly have G =

xG
xH. For each coset xH pick a
representative g
i
xH; then xH = g
i
H and g
i
H g
j
H = if i ,= j. Hence, G =

iI
g
i
H.
11.5 Denition If G =

iI
g
i
H we say that g
i
[ i I is a (complete) set of left coset
representatives of H in G. Similarly, g

i
[ i I is a (complete) set of right coset representatives
for H in G if G =

iI
Hg

i
.
These sets are often called left and right transversal, respectively.
Corollary 11.4 says that if H is a subgroup of G then we can always nd a set of coset
representatives. More than this, if we x a set of left coset representatives g
i
[ i I for H
in G then every element g G can be written uniquely in the form g = g
i
h for some i I and
h H.
Notice that if g
i
[ i I is a set of left coset representatives for H in G then g
1
i
[ i I
is a set of right coset representatives for H in G.
Combining the last few results we obtain our rst important structural theorem.
11.6 Theorem (Lagranges Theorem) Suppose that G is a nite group and that H is a sub-
group of G. Then [H[ divides [G[ and
|G|
|H|
is equal to the number of cosets of H in G.
Proof As G is a nite group, we can nd a nite set g
1
, . . . , g
k
of coset representatives for H
in G; then G =

k
i=1
g
i
H. Therefore, by Lemma 11.2(b),
[G[ =

iI
[g
i
H[ =
k

i=1
[H[ = k[H[.
Hence, k =
|G|
|H|
.
30 Algebra (Advanced)
11.7 Denition The index [G : H] of H in G is the number of cosets of H in G. If G is nite
then [G : H] =
|G|
|H|
.
11.8 Corollary Suppose that G is a nite group and that g G. Then the order of g divides
the order of G.
Proof Let H = g) = g
n
[ n Z. Now G is a nite group, so g must have nite order; say
m = [g[ < . Then g) = 1, g, . . . , g
m1
and g) = [g[ = m by Proposition 10.6. Hence,
the result follows by applying Lagranges theorem to the subgroup g).
11.9 Corollary Suppose that G is a group of prime order. Then G is cyclic.
Proof Suppose that [G[ = p, where p is prime. Pick any element g G such that g ,= 1. Then
g) is a nontrivial subgroup of G (that is, g) , = 1), so [g)[ divides p = [G[. Therefore,
either [g)[ = 1 or [g)[ = p; however, [g)[ 2, so we must have [g)[ = p. Hence, G = g)
and G is cyclic.
As every cyclic group is abelian, we have also shown that the groups of prime order are
abelian.
12. Equivalence relations
The key point about cosets is that even though a coset is a set all of its properties are deter-
mined by one of its coset representatives: if aH is a coset then aH = xH, for any x aH, and
it does not really matter which representative x we choose.
It will be useful to formalise (or abstract) what is going on here.
If X is a set then, formally, a relation on X is subset 1 of X X; we write x y if
(x, y) 1. Informally, you should think of relationships in the usual sense of the word (for
example, as a deep and meaningful connection between various elements of X).
12.1 Denition Let X be a set. An equivalence relation on X is a relation on X with the
following properties:
a) (reexive) if x X then x x;
b) (symmetric) if x, y X and x y then y x; and
c) (transitive) if x, y, z X, x y and y z then x z.
If is an equivalence relation on X and x X then x = y X [ x y X is the
equivalence class of x in X.
12.2 Examples a) If H is a subgroup of G dene a relation on G by a b if aH = bH.
Then is an equivalence relation and the equivalence class of a G is the coset aH.
b) Let X be the set (class) of all nite dimensional vector spaces, and let be the relation
on X given by V W if V

= W. Then

= is an equivalence relation on X and the
equivalence class of a vector space V consists of all those vector spaces isomorphic to V .
Notice that the isomorphism class of V is completely determined by its dimension (this
is by no means obvious; but this was proved in Math2902).
c) Suppose that G is a group. Dene a relation on G by a b if a = g
1
bg for some
g G; we say that a and b are conjugate in G. Then is an equivalence relation on G
and the equivalence classes a
G
= g
1
ag [ g G are the conjugacy classes of G. Of
these three examples, conjugacy is the one we currently understand the least about; for
example, if g G how big is the conjugacy class of g?
Homomorphisms 31
12.3 Proposition Suppose that is an equivalence relation on X. Then X is a disjoint union
of its equivalence classes.
Proof The proof is much the same as that of Corollary 11.4. As is reexive, x x so
X =

xX
x. We just need to check that different equivalence classes do not overlap.
Suppose that x z ,= . Then there exists some y x z; so x y and z y. By
symmetry, y z; so x z by transitivity. Hence, z x and z x by transitivity. Applying
symmetry once again, z x; so x z and x z. It follows that xz ,= if and only if x = z,
which is what we needed to show.
Note that if is an equivalence class on a set X then, in general, the equivalence classes
in X will have different sizes (i.e. there can exist x, y X with [x[ , = [y[). In this sense
the equivalence classes corresponding to the cosets of a subgroup H in a group G are special
because [aH[ = [H[, for all a G.
13. Homomorphisms
When studying vector spaces it is natural to consider maps f : V W which preserve
the vector space structure; that is linear maps. If V and W are inner product spaces, then we
also want f to preserve inner products. When studying groups we consider those maps which
preserve the group structure.
13.1 Denition Suppose that G and H are groups.
a) A group homomorphism fromGto H is a function : GH such that (ab) = (a)(b),
for all a, b G.
b) A group isomorphism is a group homomorphism which is also a bijection.
c) Two groups G and H are isomorphic if there exists a group isomorphism from G to H; in
this case we write G

= H.
There is a very important subtlety in this denition: on the left hand side, in (ab), we
compute the product of a and b inside G; on the right hand side, with (a)(b) we compute
the product of (a) and (b) inside H. Thus, a group homomorphism (or, more simply, a
homomorphism), is a map : GH which is compatible with the different operations in G
and H.
13.2 Examples The following maps are examples of group homomorphisms.
a) Let G be any group and let : GG be the identity map (so (g) = g for all g G).
Then is an isomorphism.
b) Let G = GL
n
(F), for any eld F, and dene : GL
n
(F) F

by (A) = det A.
Then is a group homomorphism because (AB) = det(AB) = det(A) det(B) =
(A)(B), for all A, B GL
n
(F).
c) Let G = GL
n
(C) and let : G G be the map given by (A) = (A
1
)

. Then
is a homomorphism since (AB) = (B
1
A
1
)

= (A
1
)

(B
1
)

= (A)(B), for all


A, B GL
n
(C). Since is bijective, it is an isomorphism.
d) Let be the map from the braid group to the symmetric group which forgets over
crossings and under crossings.
e) Let V and W be vector spaces, which we think of as additive (abelian) groups. Then
any linear transformation T : V W is a group homomorphism because (x + y) =
(x) + (y), for all x, y V .
32 Algebra (Advanced)
f) Let n be a positive integer and let Z
n
and C
n
be the cyclic groups of order n; see (j)
and (k) in Example 8.4. Dene : Z
n
C
n
by (a) =
a
. Then is an isomorphism.
To see this let a be equal to a modulo n; so a is the unique integer such that 0 a < n
and a = a + qn for some q Z. Then the operation in Z
n
is a + b = a + b. Write
a + b = a + b + kn, for k Z. Then
(a + b) = (a + b) =
a+b
=
a+b+kn
=
a

kn
=
a

b
= (a)(b),
since
n
= 1. Hence, is a group homomorphism. What is really happening here is that
(kn) =
kn
= 1, so (a) =
a
, for all a Z.
g) Let n be a positive integer and dene : Z C
n
by (a) =
a
. The is a group
homomorphism since (a + b) =
a+b
=
a

b
= (a)(b). Notice that is also
surjective.
h) Suppose that m and n are positive integers with m dividing n. Let = e
2i/n
and =
e
2i/m
; then C
n

= ) and C
m

= ). Let : C
n
C
m
be the map (
a
) =
a
. To
see that is a group homomorphism rst observe that, for any k Z, (
kn
) =
kn
= 1
since m divides n. Therefore, for any a, b Z,
(
a

b
) = (
a+b
) =
a+b
=
a

b
= (
a
)(
b
).
Hence, is a surjective group homomorphism.
13.3 Proposition Suppose that Gand H are groups and that : GH is a group homomor-
phism. Then (1
G
) = 1
H
and (g
1
) =
_
(g)
_
1
, for all g G.
Proof The proof of this is basically the same as Proposition 9.3. We have
(1
G
) = (1
G
1
G
) = (1
G
)(1
G
).
Multiplying this equation on the left by
_
(1
G
)
_
1
shows that
1
H
= (1
G
)
_
(1
G
)
_
1
=
_
(1
G
)(1
G
)
__
(1
G
)
_
1
= (1
G
)
_
(1
G
)(1
G
)
1
_
= (1
G
).
For the second claim, by what we have just shown, (g)(g
1
) = (gg
1
) = (1
G
) = 1
H
and, similarly, (g
1
)(g) = 1
H
. Hence, (g
1
) =
_
(g)
_
1
by Proposition 8.5(b).
13.4 Denition Suppose that : GH is a group homomorphism.
a) The image of is im = (g) [ g G, a subset of H.
b) The kernel of is ker = g G [ (g) = 1
H
, a subset of G.
If H is an additive group then the kernel of is the set of elements which are sent to zero
by ; this is how you should think of the kernel in general: it is the set of elements which are
killed by .
13.5 Corollary Suppose that : GH is a group homomorphism. Then im is a subgroup
of H and ker is a subgroup of G.
Proof If x, y im then x = (a) and y = (b) for some a, b G. Therefore,
xy
1
= (a)
_
(b)
_
1
= (a)(b
1
) = (ab
1
) im
by the Proposition. Hence, im is a subgroup of H by the Subgroup Criterion (Theorem 9.6).
Similarly, if a, b ker then (ab
1
) = (a)(b
1
) = 1
H
_
(b)
_
1
= 1
H
. Hence, ab
1
also belongs to the kernel; so ker is also a subgroup of G by the Subgroup Criterion.
Normal subgroups 33
14. Normal subgroups
Corollary 13.5 is slightly misleading because it seems to say that image and kernel of a
(group) homomorphism are pretty much the same; in fact, the kernel of a homomorphism is a
very special type of subgroup.
For any subgroup H of G we can always form the setwise product
(aH)(bH) = xy [ x aH and y bH
of two cosets of H. Since (aH)(bH)

hH
ahbH we can write (aH)(bH) as a (disjoint)
union of Hcosets (by Corollary 11.4). For most subgroups H the setwise product will be a
union of two or more cosets rather than just a single coset; however, for some subgroups, for
example the kernel of a homomorphism, we obtain just a single coset.
Consider a group homomorphism : GH and let K = ker = g G [ (g) = 1
H
.
Claim: If a G then aK = g G [ (g) = (a) = Ka.
One direction is easy: if g aK then g = ak, for some k K, so (g) = (a)(k) = (a)
since (k) = 1
H
. Hence, aK g G [ (g) = (a) . Conversely, if (g) = (a) then
1
H
= (g)
1
(a) = (g
1
)(a) = (g
1
a).
so g
1
a K = x G [ (x) = 1
H
. Therefore, aK = gK by Proposition 11.3.
To prove the claim for the right cosets of K we proceed in much the same way. If (g) =
(a) then 1
H
= (g)(a)
1
= (ga
1
), so ga
1
K; this implies that Kg = Ka by the
right handed version of Proposition 11.3(e). Hence, g G [ (g) = (a) Ka. The other
inclusion is virtually identical to the previous case.
Now suppose that a, b G and that x aK = Ka and y bK = Kb. Then
(xy) = (x)(y) = (a)(b) = (ab).
Therefore, (aK)(bK) = xy [ (x) = (a) and (y) = (b) g G [ (g) = (ab) =
abK. In fact, (aK)(bK) = abK because (aK)(bK) is a union of left cosets whereas the right
hand side is just a single coset. A quicker way to see this is to note that
(aK)(bK) = a(Kb)K = a(bK)K = abK
2
= abK,
the second equality following because bK = Kb by the claim.
As you are not used to working with cosets we should perhaps be a little more careful here.
We have
(aK)(bK) = axby [ x, y K = a(bx

)y [ y, x

K = abK
2
= abK,
where the second equality follows because xb = bx

for some x

K (since Kb = bK), and


the last equality follows because K
2
= K. Notice that (aK)(bK) =

xK
axbK is a union of
left cosets of K; what is surprising is that we get a single coset.
14.1 Denition Suppose that G is a group. A normal subgroup of G is any subgroup N such
that gN = Ng for any g G. If N is a normal subgroup of G we write N G.
34 Algebra (Advanced)
From the discussion above, the kernel of a group homomorphism is always normal.
It is important to note that N is normal if and only if gN = Ng as sets, for all g G. That
is, we require that gN = gn [ n N = ng [ n N = Ng. So if N is normal, g G
and n N then gn = n

g for some n

N; in general, n

will be genuinely different from n


and gn ,= ng.
14.2 Proposition Suppose that N is a subgroup of G. Then the following are equivalent:
a) g
1
Ng = N for all g G.
b) N is a normal subgroup of G;
c) every left coset of N is also a right coset (i.e. if g G then gN = Ng

for some g

G);
d) the product of any two left cosets of G is a single left coset (i.e. if a, b G then
(aN)(bN) = cN, for some c G);
e) (aN)(bN) = abN for all a, b G;
Proof If g
1
Ng = N for all g G then gN = Ng by multiplying on the right by g. Hence,
(a) implies (b).
Next suppose that (b) is true. Then gN = Ng for all g G so (c) holds.
If (c) is true then if b G there exists a b

G such that Nb = b

N. Therefore, for any


a G we have (aN)(bN) = a(Nb)N = a(b

N)N = ab

N. Hence, the product of two cosets


is again a coset; so (c) implies (d).
Now assume that (d) is true: that is, (aN)(bN) = cN for some c G. Now, ab = a 1 b 1
(aN)(bN), so ab cN. Therefore, cN = abN by Proposition 11.3. Hence, (d) implies (e).
Finally, suppose that (e) is true. Then, in particular, (g
1
N)(gN) = g
1
gN = N. Therefore,
if n N then g
1
ng = g
1
ng 1 g
1
NgN = N; so g
1
Ng N. By the same argument
(using g
1
in place of g), gNg
1
N; therefore, N g
1
Ng multiply on the left by g
1
and the right by g. Hence, g
1
Ng = N, for all g G; so (a) holds.
We will meet some more characterizations of normal subgroups in the next few sections.
14.3 Examples a) Suppose that Gis an abelian group. Then every subgroup of Gis normal.
b) Consider GL
n
(F), for some eld F, and let SL
n
(F) = a GL
n
(F) [ det(a) = 1 . It
is straightforward to check that SL
n
(F) is a subgroup of GL
n
(F); this group is called the
special linear group. Then SL
n
(F) is a normal subgroup of GL
n
(F). Looking at cosets this
is by no means obvious; however, if g GL
n
(F) and a SL
n
(F) then
det(gag
1
) = det(g) det(a) det(g
1
) = det(g) 1 det(g)
1
= 1;
so gag
1
SL
n
(F).
c) Once again take D
8
= r, s [ r
8
= 1, s
2
= 1 and srs = r
1
) to be the dihedral group of
order 16; see Example 10.7. Then the following subgroups of D
8
are normal:
H
1
= 1) = 1
H
2
= r
4
) = 1, r
4

H
3
= r
2
) = 1, r
2
, r
4
, r
6

H
4
= r) = 1, r, r
2
, r
3
, r
4
, r
5
, r
6
, r
7

H
5
= s, r
4
) = 1, r
4
, s, sr
4

H
6
= s, r
2
) = 1, r
2
, r
4
, r
6
, s, sr
2
, sr
4
, sr
6

H
7
= s, r) = G.
Quotient groups 35
To check this because r and s generate D
8
it is enough to note that r
1
H
i
r H
i
and that
sH
i
s H
i
, for each i. This follows easily once you remember that r
a
s = sr
a
for all a.
Finally, a subgroup of D
8
which is not normal is H
8
= s) = 1, s. To see this
notice that rH
8
= r, rs = r, sr
7
; whereas, H
8
r = r, sr. Similarly, if 0 b < 8
then sr
b
) = 1, sr
b
is not normal unless b = 4 (since r
b
s = sr
b
is equal to sr
b
if and
only if b = 4).
d) Let G = Sym(3) = 1, (1, 2), (2, 3), (1, 3), (1, 2, 3), , (1, 3, 2). Let H = (1, 2)) is not
normal in G whereas K = (1, 2, 3)) is a normal subgroup.
In general, only a small number of the subgroups of a (non Abelian) group will be normal.
We close this section with another characterization of normal subgroups. First, recall from
Example 12.2(c) that if a G then the conjugacy class of a is a
G
= g
1
ag [ g G.
14.4 Proposition Suppose that N is a subgroup of G. Then N is normal if and only if N is a
union of conjugacy classes.
Proof By Proposition 14.2 N is normal in G if and only if N = g
1
Ng for all g G.
Therefore, if a N then a
G
N, so N is a union of conjugacy classes.
Conversely, if N is a union of conjugacy classes and n N then g
1
ng N, for all g G.
Hence, N = g
1
Ng, for all g G; so N is a normal subgroup of G.
15. Quotient groups
The main reason why we care about normal subgroups is that we can use them to construct
new groups.
Suppose that N is a normal subgroup of Gand let G/N = aN [ a G be the set of cosets
of N of N in G. By Proposition 14.2 the (setwise) product of two cosets of N is again a coset
of N; in other words, the operation on G induces an operation on G/N; namely,
(aN)(bN) = abN
for all aN, bN G/N.
15.1 Theorem Suppose that N is a normal subgroup of G. Then G/N is a group with operation
(aN)(bN) = abN,
for all aN, bN G.
Proof The multiplication in G is associative so we have
_
(aN)(bN)
_
(cN) = (abN)(cN) = (ab)cN = a(bc)N = (aN)(bcN) = (aN)
_
(bN)(cN)
_
;
therefore the operation on G/N is also associative. The set G/N also has an identity element,
namely N = 1 N, because
(1 N)(aN) = aN = (aN)(1 N),
for all aN G/N. Finally, G/N also has inverses since
(aN)(a
1
N) = N = (a
1
N)(aN),
for all aN G/N.
36 Algebra (Advanced)
15.2 Denition Suppose that N is a normal subgroup of G. Then G/N is the quotient group (or
factor group) of G by N.
Before giving some examples we backtrack a little and point out a hidden subtlety of the
denition of the operation on G/N. Namely, given two cosets we have dened (aN)(bN) =
abN. However, suppose that we had chosen different coset representatives x and y, respectively,
so that xN = aN and yN = bN. Then we would have (aN)(bN) = (xN)(yN) = xyN. So,
in order for our denition to make sense, we require that abN = xyN. In fact, this is already
guaranteed because we have checked that (aN)(bN) = abN as a product of sets, so the choice
of coset representative does not matter (that is, the choice of representative does not change the
underlying set).
As the denition of the product does not depend on the choice of coset representatives we
say that operation on G/N is welldened; the term welldened means that the denition is
unambiguous and does not depend on the choices implicit in the denition. We will soon come
across more situations where we need to check that our denitions are welldened. It will not
always be as obvious that our denitions are independent of the choices made.
15.3 Examples a) The best example that I know of is the following (it is best in the sense
that, when we work it out, we will see that the quotient group is already familiar). Let
G = Z (under addition) and x a positive integer n. Then N = nZ = kn [ k Z
is a normal subgroup of Z indeed, since Z is abelian all subgroups of Z are normal.
Therefore, 0, 1, . . . , n 1 is a complete set of coset representatives of nZ in Z, so
Z/nZ = 0 + nZ, 1 + nZ, . . . , n 1 + nZ = 0, 1, . . . , n 1,
where, by denition, a = a + nZ for a = 0, 1, . . . , n 1. Extend this notation to the
whole of Z: for any integer k dene k to be the unique integer in 0, 1, . . . , n 1 such
that k + nZ = k + nZ. Then
a = k k + nZ = a + nZ k a nZ k a mod n.
Applying the denitions we can now work out the operation on Z/nZ:
a + b = (a + nZ) + (b + nZ) = (a + b) + nZ = a + b.
It is evident that Z/nZ

= Z
n
via the group homomorphism : Z/nZ Z
n
given
by (a+nZ) = a, for all a Z/nZ. Thus, quotienting out by nZ is the same as working
modulo n! In other words, working inside Z/nZ is the same as working in Z providing
we set n (or, equivalently, nZ) equal to zero. Note that 0 is the identity element of Z.
b) For a second example suppose that V = R
n
is an ndimensional real inner product space
under addition. Let e
1
, . . . , e
n
be the standard (orthonormal) basis of row vectors of R
n
.
Fix an integer m with 1 m n and let W be the vector subspace of V spanned by
e
1
, . . . , e
m
. Then W is a subgroup of V . (Note, however, that as a subgroup W is
not generated by e
1
, . . . , e
m
: the subgroup of V generated by e
1
, . . . , e
m
consists of all
vectors of the form

m
i=1

i
e
i
, for
i
Z rather than
i
R.)
Now

n
i=m+1

j
e
j
[
i
R is a complete set of coset representatives for W in V .
Therefore, the elements of V/W are all of the form
n

i=m+1

j
e
j
+ W =
n

j=m+1

j
e
j
, for
j
R,
The isomorphism theorems 37
where e
j
= e
j
+W. In other words, V/W

= R
nm
. In terms of cosets what is happening
here is quite straightforward: the coset

n
i=m+1

j
e
j
+ W consists of all row vectors
of the form (, . . . , ,
m+1
, . . . ,
n
), where the indicates an arbitrary real number. The
elements of W are precisely the vectors of the form (, . . . , , 0, . . . , 0) and in the quotient
space V/W we just set all of the s equal to zero. Finally, notice that V/W

= W

.
To appreciate the true signicance of normal subgroups and quotient groups we need to
return to homomorphisms. As an appetiser of the next section we observe that the following
result holds.
15.4 Proposition Suppose that N is a normal subgroup of G. Then the function

N
: GG/N; aaN, for all a G,
is a group homomorphism. Moreover, N = ker
N
.
Proof If a, b G then
N
(ab) = abN = (aN)(bN) =
N
(a)
N
(b), which shows that
N
is a
group homomorphism. Also, if a G then a ker
N
if and only if
N
(a) = 1
G/N
; that is, if
and only if aN = N. Therefore, ker
N
= N as claimed.
The homomorphism
N
is often called the canonical homomorphism.
15.5 Corollary Suppose that N is a subgroup of G. Then N is normal if and only if N = ker
for some group homomorphism .
Proof If N is normal then it is the kernel of the homomorphism
N
from the Proposition.
Conversely, at the beginning of the last section we saw that the kernel of a homomorphism is
normal.
16. The isomorphism theorems
Recall that if : G H is a group homomorphism then im = (g) [ g G is the
image of and ker = g G [ (g) = 1
H
is the kernel of . The intimate relationship
between these two groups is one of the most fundamental results in group theory (and algebra).
16.1 Theorem (The rst isomorphism theorem) Suppose that : GH is a group homo-
morphism. Then im

= G/ ker .
Proof Let K = ker . Then, by denition, the elements of the quotient group G/K = G/ ker
are precisely the cosets aK [ a G. Dene a function : G/Kim by
(aK) = (a), for all a G.
We rst need to check that is welldened: namely, if aK = bK then we need to check that
(aK) = (bK). If aK = bK then a
1
b K, so
1
H
= (a
1
b) = (a
1
)(b) = (a)
1
(b).
Hence, (a) = (b) and so (aK) = (a) = (b) = (bK). Therefore, the denition of
does not depend on the choice of coset representative.
38 Algebra (Advanced)
Next we check that is a group homomorphism. If aK, bK G/K then

_
(aK)(bK)
_
= (abK) = (ab) = (a)(b) = (aK)(bK),
where the second last equality follows because is a group homomorphism.
Finally, we show that is an isomorphism. It is certainly surjective because if x im then
x = (a), for some a G; so, x = (aK). On the other hand, if aK, bK G/K then
(aK) = (bK) (a) = (b) (b)
1
(a) = 1
H
(b
1
)(a) = 1
H
(b
1
a) = 1
H
b
1
a K aK = bK, by Proposition 11.3.
Hence is injective.
The most surprising consequence of the rst isomorphism theorem is that the possibilities
for the groups which are homomorphic images of a group G are completely determined by the
(normal) subgroups of G.
16.2 Corollary Suppose that : GH is a surjective group homomorphism. Then
H

= G/ ker .
Surjective homomorphisms are often called epimorphisms.
16.3 Examples a) Given a positive integer n dene : ZZ
n
by (k) = k. Then is a
group homomorphism (check) and ker = nZ. Thus, Z
n

= Z/nZ; see Example 15.3(a).
b) Let : G G be the identity map; so (g) = g for all g G. Then ker = 1
G

and im = G; so G

= G/ ker . In fact, we could have seen this before we proved the
rst isomorphism theorem as the cosets of the kernel are just the sets g1
G
= g, for
all g G.
c) Suppose that V and U are vector spaces and that : V U is a vector space homomor-
phism. Then we can also consider to be a homomorphism of additive groups; so, by
the rst isomorphism theorem, im

= V/W where W = ker . By Example 15.3(b),
if dimV = n and dimW = m then V/W is a vector space of dimension n m. There-
fore, dimim = dimV/W = n m = dimV dimW or, equivalently,
dimV = dimim + dimker .
This is just the ranknullity theorem! The difference is that in addition to having worked
out the relationship between the dimensions of these vector spaces we have also con-
structed a new vector space V/W which explains this otherwise purely numerical result.
d) Let : GL
n
(F) F

be the homomorphism (g) = det g, for all g GL


n
(F); here,
F

= F0 is the multiplicative group of the eld F. Then is surjective and g ker


if and only if det(g) = 1; thus, ker = SL
n
(F) and GL
n
(F)/ SL
n
(F)

= F

.
The following lemma provides a useful criterion for determining when a group homomor-
phism is injective.
16.4 Corollary Suppose that : GH is a group homomorphism. Then is injective if and
only if ker = 1
G
; in this case we also have that G

= im.
The isomorphism theorems 39
Proof See tutorials.
In particular, G is isomorphic to a subgroup of H if ker = 1
G
.
Injective homomorphisms are often called monomorphisms.
As our rst real application of the rst isomorphism theorem we will prove the second iso-
morphism theorem. Before we do this however, we state the following result for contrast.
16.5 Lemma Suppose that H and K are subgroups of G. Then
HK = hk [ h H and k K
is a subgroup of G if and only if HK = KH.
Proof See tutorials.
The lemma is relevant to the result because HN must be a subgroup of G. The statement
of the second isomorphism theorem is quite convoluted; all it really says is that HN/N

=
H/(H N) whenever H and N are subgroups of G with N being normal. The problem is that
before this statement makes sense we require that (i) HN is a subgroup of G; (ii) that N is a
normal subgroup of HNl and (iii) that H N is a subgroup of H.
16.6 Theorem (The second isomorphism theorem) Suppose that H and N are subgroups of G
with N being normal. Then HN is a subgroup of G, N is a normal subgroup of HN, H N
is a normal subgroup of H, and HN/N

= H/(H N).
Proof First suppose that hn, h

HN, for h, h

H and n, n

N. Then
(h

)(hn)
1
= h

(n

n
1
)h
1
= h

h
1
n

HN,
the last equality following because N is a normal subgroup of G, so Nh
1
= h
1
N. Therefore,
HN is a subgroup of G by the subgroup criterion (Theorem 9.6). Next, since N is normal in G
it is certainly also normal in HN. We could prove directly that H N is a normal subgroup of
H but we will do this more efciently.
Dene a map : HHN/N by (h) = hN. This makes sense because
HN/N = hnN [ h H and n N = hN [ h H .
In particular, is a surjective map. The map is a group homomorphism because if h, h

H
then (hh

) = hh

N = (hN)(h

N) = (h)(h

). Consequently, by the rst isomorphism


theorem, H/ ker

= HN/N. Now, h ker if and only if (h) = 1
HN/N
; that is, if and only
if hN = N, or h H N. Hence, ker = H N. Therefore, H N is a normal subgroup
of H (it is the kernel of a homomorphism) and HN/N

= H/(H N).
There is one more isomorphism theorem, the proof of which we leave to the tutorials.
16.7 Theorem (The third isomorphism theorem) Suppose that N and M are normal sub-
groups of G with N a subgroup of M. Then M/N is a normal subgroup of G/N and
(G/N)/(M/N)

= G/M.
Proof See tutorials.
Finally, we remark that the proofs of the all three isomorphismtheorems are fairly straightfor-
ward: it is largely a matter of writing down what the elements of the various (quotient) groups
in each result and then dening obvious maps between them (and checking that this works).
40 Algebra (Advanced)
17. The structure of groups I: the existence of pelements
By Lagranges theorem (Theorem 11.6) we know that if Gis a nite group and g Gthen [g[
divides [G[. In general, the converse to this statement is not true; that is, if m divides [G[ then
there is no guarantee that G will contain an element of order m. For example, Sym(4) has
order 4! = 24 but only contains elements of order 1, 2, 3 and 4.
A partial converse to Lagranges theorem (Theorem 11.6), is Theorem 17.2 below. It is our
rst serious theorem. The proof uses quotient groups in a crucial way. Before we can begin we
need a simple lemma.
17.1 Lemma Suppose that G is a nite group. Then g
|G|
= 1 for all g G.
Proof Suppose that g G and let m = [g[. Then m divides n so we can nd an integer k such
that [G[ = km. Therefore, g
|G|
= g
km
= (g
m
)
k
= 1 by Lagranges Theorem.
17.2 Theorem Suppose that G is a nite abelian group and that p is a prime dividing the order
of G. Then G contains an element of order p.
Proof We argue by induction on the order of G. If [G[ = 1 then there is nothing to prove (as
the theorem is claiming nothing), so suppose that [G[ > 1.
If G has no proper nontrivial subgroups (that is, no subgroups except 1 and G), then G
is cyclic and we may write G = g) for some g ,= 1. Since p divides [G[, we have [G[ = p,
otherwise g
|G|/p
) would be a proper subgroup of G. So, g is an element of order p.
Suppose now that G contains a proper nontrivial subgroup N. If p divides [N[ then N
contains an element of order p by induction (since [N[ < [G[). Hence, G contains an element
of order p since N G.
Thus we are now reduced to considering the case where G contains a proper nontrivial
subgroup N of order not divisible by p. Now N is a normal subgroup of G, since G is abelian;
so we can consider the quotient group G/N. Now[G/N[ = [G[/[N[ so p divides [G/N[ since p
divides [G[ and does not divide [N[. Moreover, [G/N[ < [G[; so, by induction, G/N contains
an element gN of order p. Consequently, N = (gN)
p
= g
p
N; so g
p
N. Let x = g
|N|
. Then
x
p
= (g
|N|
)
p
= (g
p
)
|N|
= 1,
the last equality following from the Lemma since g
p
N. We are now almost done: we only
need to show that x ,= 1. By way of contradiction suppose that x = 1. Since p does not divide
[N[ we can nd integers a and b such that 1 = ap + b[N[; therefore, in G/N we have
gN = (gN)
1
= (gN)
ap+b|N|
= (gN)
ap
(gN)
b|N|
=
_
(gN)
p
_
a
(g
|N|
N)
b
= x
b
N = N.
Hence, if x = 1 then gN = 1
G/N
. However, this cannot be true because gN has order p > 1.
Therefore, x is an element in G of order p.
17.3 Corollary Suppose that Gis a nite abelian group and that p is a prime dividing the order
of G. Then G contains a subgroup of order p.
Group actions on sets 41
Proof Any element of order p generates a subgroup of order p.
We could extend the argument of Theorem 17.2 to prove that whenever p

divides the order


of a nite abelian group G then G has a subgroup of order p

. Instead of doing this we pursue


the much stronger result that says that whenever p

divides the order of a nite group G (which


is not necessarily abelian) then Ghas a subgroup of order p

. Before we can tackle this we need


to understand group actions on sets.
18. Group actions on sets
In Example 8.4(n) we dened the symmetry group of a graph . We now turn this denition
around and ask when a given group captures some of the symmetry in a graph.
18.1 Denition Suppose that G is a group and that X is a set. Then G acts on X (from the
right) if there is a function
X G X; (x, g)x
g
such that x
1
G
= x and x
gh
= (x
g
)
h
, for all x X and all g, h G.
The key component of the denition of a group action is that the action respects the operation
in the group.
We could also consider groups acting on the left, say (g, x)
g
x; for left actions the second
condition gets replaced with the requirement that
gh
x =
g
(
h
x) for all x X and g, h G. The
main reason why we use right actions is that our convention for the symmetric group means that
it is acts on 1, 2, . . . , n from the right. Warning: Armstrong considers left actions.
Before we discuss some examples lets take the following denitions on board.
18.2 Denition Suppose that G acts on X (from the right). If x G then the orbit of G on x
is x
G
= x
g
[ g G and the stabilizer of x is G
x
= g G [ x
g
= x. The action is transitive
if X = x
G
for some x X.
Note that x
G
X and G
x
G. We will frequently refer to x
G
as a Gorbit. Sometimes we
will also write Stab
G
(x) = G
x
for the stabilizer of x in G.
18.3 Examples a) The most important example of a group action is the following. Recall
that the symmetric group G = Sym(n) is the group of permutations of 1, 2 . . . , n. The
group Sym(n) acts on n = 1, 2, . . . , n from the right: if i n and Sym(n)
then i

= j if sends i to j. This rule denes an action because: (i) i


1
G
= i, for
all i n, and (ii) if , Sym(n) and i n then i

= (i

because earlier we
adopted the convention that the permutation is the map dened by ()(i) = ((i));
see Example 8.4(l). We adopted this convention because it makes the multiplication in
Sym(n) more natural: we just read permutations from left to right; but, in fact, this
convention amounts to saying that Sym(n) acts on the set n from the right.
If i n then the orbit i
G
of i under Sym(n) is the whole of n because, for example,
if j n then j = i
(i,j)
i
G
. Also, the stabilizer G
i
of i is the set of permutations which
x i; for example, G
n
= Sym(n 1).
For the reason why this is the most important example of a group action see Proposi-
tion 18.8 below.
42 Algebra (Advanced)
b) Lets make the last example more explicit. Let = (1, 8, 2, 5)(3, 7, 6)(4, 9), an element
of Sym(n) for n 9. Let H = ) be the subgroup of Sym(n) generated by . Then
[H[ = [[ = 12 and H also acts on n: we just restrict the action of Sym(n) to H. If i n
then i
H
= i
h
[ h H = i

k
[ k Z so
i
H
=
_

_
1, 2, 5, 8, if i = 1, 2, 5, 8
3, 6, 7, if i = 3, 6, 7,
4, 9, if i = 4, 9,
i, otherwise.
So the orbits of H on n correspond exactly to the disjoint cycles in . In particular, the
orbits of H on n coincide with the orbits of the subgroup H

= (1, 2, 5, 8)(3, 6, 7)(4, 9))


on n even though these are different subgroups of Sym(n). On the other hand, the
orbits of the subgroup H

= (1, 2, 3, 4)(5, 6, 7)(8, 9)) on n are different, but not in any


essential way (they have the same sizes, for example). One of the things we want to do is
understand in what sense these orbits are the same.
Working out the stabilizers takes more effort. First note that
H =
_
1, (1, 8, 2, 5)(3, 7, 6)(4, 9), (1, 2)(3, 6, 7)(5, 8), (1, 5, 2, 8)(4, 9), (3, 7, 6),
(1, 8, 2, 5)(3, 6, 7)(4, 9), (1, 2)(5, 8), (1, 5, 2, 8)(3, 7, 6)(4, 9), (3, 6, 7),
(1, 8, 2, 5)(4, 9), (1, 2)(3, 7, 6)(5, 8), (1, 5, 2, 8)(3, 6, 7)(4, 9)
_
.
Using this list we can explicitly write down the elements of H
i
for each i n. For
example, H
i
= 1, (3, 6, 7), (3, 7, 6) = (3, 6, 7)), when i 1
H
= 1, 2, 5, 8. When
you do this you will notice that in each case H
i
is a subgroup of G: moreover, H
i
= H
j
whenever i and j are in the same orbit and
[H
i
[ =
_

_
3, if i 1
H
= 1, 2, 5, 8,
4, if i 3
H
= 3, 6, 7,
6, if i 4
H
= 4, 9,
12, otherwise.
In particular, notice that in each case [i
H
[ [H
i
[ = [H[.
c) Suppose that G = G

is the symmetry group of a graph . Then G acts on .


d) Let n be a positive integer and let D
n
be the symmetry group of the regular ngon P
n
;
see Example 8.4(o). Then D
n
acts on P
n
, by the last example. The group C
n
= )
also acts on P
n
, where acts as a rotation through 2/n (so
a
acts as a rotation through
2a/n). For any vertex i in P
n
we have i
Dn
= i
Cn
= n and for the stabilizers we have
[ Stab
Cn
(i)[ = 1 and [ Stab
Dn
(i)[ = 2, as there is a unique reection which xes i. For
both groups (i.e. G = D
n
or G = C
n
), we have that [G
i
[ [i
G
[ = [G[, for all i.
e) Let H be a subgroup of G and let X = Hx [ x G be the set of right cosets of H
in G. Then G acts on X via (Hx)
g
= Hxg. Then
Stab
G
(Hx) = g G [ Hxg = Hx = g G [ xgx
1
H
= g G [ g x
1
Hx = x
1
Hx.
It is easy to check that x
1
Hx is a subgroup of G. Further, the orbit of the coset Hx
under G is the whole of X since Hy = (Hx)(x
1
y) = (Hx)
x
1
y
for any Hy X. So,
the action is transitive.
Group actions on sets 43
f) By taking H = 1
G
in the last example we obtain an action of G on itself. This is called
the regular action of G on itself: (x, g)xg. There is another more interesting action of
G on itself; namely, an action by conjugation: if x G and g G set x
g
= g
1
xg. This
denes a Gaction because x
1
= x and
x
gh
= (gh)
1
x(gh) = h
1
g
1
xgh = (g
1
xg)
h
= (x
g
)
h
,
for all x, g, h G. The Gorbit of x G is the conjugacy class x
G
= g
1
xg [ g G
of x (see Example 12.2(c)). Also, the stabilizer of x is
Stab
G
(x) = g G [ g
1
xg = x = g G [ xg = gx .
Hence, Stab
G
(x) = C
G
(x) is the centralizer of x in G, which we have (or soon will) come
across in tutorials.
Notice that group actions, together with orbits and stabilizers, encapsulates (and generalizes)
our previous constructions of cosets and conjugacy classes. As cosets and conjugacy classes are
both equivalence classes the next result is no surprise.
18.4 Lemma Suppose that G acts on a set X. Then the orbits of G on X determine an equiva-
lence relation on X. In particular, X is a disjoint union of Gorbits.
Proof Dene a relation of X by x y if y x
G
= x
g
[ g G. We need to check that
is an equivalence relation on X. First, x x because x = x
1
G
x
G
; so is reexive. Next,
if x y then y = x
g
for some g G. Therefore, y
g
1
= (x
g
)
g
1
= x
gg
1
= x, so y x and
is symmetric. Finally, if x y and y z then y = x
g
and z = y
h
, for some g, h G. Hence,
z = y
h
= (x
g
)
h
= x
gh
x
G
and x z, establishing transitivity.
That X is a disjoint union of equivalence classes now follows from Proposition 12.3.
Next we show that G
x
is always a subgroup of G.
18.5 Lemma Suppose that G acts on X and let x X. Then the stabilizer G
x
of x is a
subgroup of G.
Proof If a, b G
x
then x
a
= x and x
b
= x; so x
ab
= (x
a
)
b
= x
b
= x. Hence, ab G
x
and G
x
is closed under multiplication. Similarly, x
a
1
= (x
a
)
a
1
= x
aa
1
= x. Therefore, G
x
is closed
under the taking of inverses, so G
x
is a subgroup of G by Lemma 9.4.
This brings us to the main result concerning group actions, which we have already observed in
the examples above. The orbitstabilizer theorem is very easy to prove; however, it is important
because it gives a way of counting the size of an orbit or the size of a stabilizer.
18.6 Theorem (The orbitstabilizer theorem) Suppose that Gacts on X and let x X. Then
there is a onetoone correspondence, G
x
g x
g
, between the right cosets of G
x
in G
and x
G
, the Gorbit of x. Consequently, [x
G
[ = [G : G
x
].
Proof We need to show that x
a
= x
b
if and only if G
x
a = G
x
b; it turns out that there is
not much to do. Suppose rst that x
a
= x
b
. Then x = (x
a
)
b
1
= x
ab
1
; so ab
1
G
x
and,
consequently, G
x
a = G
x
b. Conversely, if G
x
a = G
x
b then b = ga for some g G
x
. Therefore,
x
b
= x
ga
= (x
g
)
a
= x
a
(since x
g
= x whenever g G
x
).
44 Algebra (Advanced)
18.7 Example Let G = Sym(n) and let X be the set of all polynomials in the independent
variables x
1
, x
2
, . . . , x
n
. Then Sym(n) acts on X by permuting the variables; explicitly, if
f = f(x
1
, . . . , x
n
) X and Sym(n) then f

= f(x
1
, . . . , x
n
). This denes an action of
Sym(n) on X because Sym(n) acts on 1, 2, . . . , n.
Let =

1i<jn
(x
i
x
j
). Then X and

= for all Sym(n). We say


that Sym(n) is an even permutation if

= and is an odd permutation if

= .
Let Alt(n) be the set of even permutations in Sym(n); then Alt(n) is a subgroup of Sym(n)
since Alt(n) = Sym(n)

is the stabilizer of . In fact, Alt(n) is a normal subgroup of Sym(n)


because it is the kernel of the group homomorphism(which by abuse of notation we also call )
: Sym(n)C
2
;
_
1, if is even,
1, if is odd.
The group Alt(n) is called the alternating group of degree n. Since Alt(n) is the stabilizer of ,
it follows from the orbit stabilizer theorem that Alt(n) has order
1
2
n!.
Corollary (Burnside formula). Let a nite group G act on a nite set X with the orbits
X
1
, . . . , X
k
. Then the number of orbits k can be found from the formula
k =
1
[G[

gG
[ Fix
g
(X)[,
where Fix
g
(X) = x X [ x
g
= x.
Proof We have the relation

gG
[ Fix
g
(X)[ =

xX
[G
x
[,
because both sides count the number of pairs (g, x) G X such that x
g
= x. Applying the
orbitstabilizer theorem and Lemma 18.4, we get

xX
[G
x
[ =

xX
[G[
[x
G
[
=
k

i=1

xX
i
[G[
[X
i
[
=
k

i=1
[G[ = k [G[.
The Burnside formula has a number of applications in combinatorics. Various objects can be
counted in the case where they can be regarded as orbits for the action of a certain nite group.
Example Let us count in how many ways we can colour the faces of a cube so that each face
is coloured red, blue or yellow.
Here G = C is the group of rotations of the cube with [G[ = 24 and X is the set of xed
cubes whose faces are coloured red, blue or yellow. For instance, the cube whose front face is
red while the remaining faces are blue and the cube whose top face is red while the remaining
faces are blue are two different elements of X.
The number in question is precisely the number of orbits for the action of G on X. Calculate
[ Fix
g
(X)[ for various elements g G.
If g = 1 then [ Fix
g
(X)[ = [X[ = 3
6
.
If g is a rotation around a diagonal of the cube through 120

or 240

then [ Fix
g
(X)[ = 3
2
.
Indeed, if the diagonal passes through the opposite vertices Aand H then the three faces having
Group actions on sets 45
A as a common vertex must have the same colour. The same holds for the three faces having H
as a common vertex.
Similarly, if g is a rotation through 180

around the line connecting the middles of two oppo-


site edges then [ Fix
g
(X)[ = 3
3
.
If g is a rotation through 90

or 270

around the line connecting the middles of two opposite


faces then [ Fix
g
(X)[ = 3
3
.
If g is a rotation through 180

around the line connecting the middles of two opposite faces


then [ Fix
g
(X)[ = 3
4
.
Taking into account the number of elements of each type we get the required number from
the Burnside formula:
1
24
(3
6
+ 8 3
2
+ 6 3
3
+ 6 3
3
+ 3 3
4
) = 57.
Finally, we show how to recast group actions in terms of permutation groups. Recall from
Example 8.4(m) that Sym(X) is the group of bijections from X to X; if X is a nite set and
X = x
1
, x
2
, . . . , x
n
then Sym(X)

= Sym(n), where the isomorphism identies a bijection
f : XX with the permutation
f
Sym(n) such that (i) = j if f(x
i
) = x
j
.
Nowsuppose that Gacts on X. Then every element g Gdetermines a permutation
g
of the
elements of X; explicitly,
g
(x) = x
g
. Moreover, the map g
g
is a group homomorphism
because

gh
(x) = x
gh
= (x
g
)
h
=
_

g
(x)
_
h
=
h
_

g
(x)
_
= (
g

h
)(x);
that is, gh
g

h
=
gh
. (Recall that in Example 8.4(o) we dened the product of two
permutations and to be the permutation given by ()(i) = ((i)) for all i.) Hence, we
have proved the following.
18.8 Proposition Suppose that G acts on X. Then the map
: GSym(X); g
g
,
is a group homomorphism.
Conversely, given a homomorphism : GSym(X) we can dene an action of G on X
by x
g
= (x), for g G and x X.
As a special case of this observation we have the following striking result due to Cayley.
18.9 Theorem (Cayleys theorem) Suppose that G is a nite group of order n. Then G is
isomorphic to a subgroup of Sym(n).
Proof Let G act on itself by right multiplication: x
g
= xg this is an action because it is
a special case of the action of G on the right cosets of a subgroup. By the Proposition this
action induces a homomorphism : GSym(n), where we identify Sym(G) with Sym(n)
as above. Finally, it remains to observe that is injective because 1
G
is the only element of G
which acts as the identity permutation (see Corollary 16.4).
46 Algebra (Advanced)
19. Conjugacy classes and groups of ppower order
The orbit stabilizer theorem is very easy to prove; however, it is actually quite a powerful
result. In the few sections we will apply the orbit stabilizer theorem to prove some strong
results about the internal structure of groups.
First recall that a group Gacts on itself by conjugation: x
g
= g
1
xg, for x, g G. The orbits
x
G
= g
1
xg [ g G of this action are the conjugacy classes of G. From tutorials we know
that C
G
(x) = g G [ xg = gx is the stabilizer of x (when G acts on itself by conjugation).
Hence, by the orbit stabilizer theorem we have the following.
19.1 Corollary Suppose that x G and let x
G
= g
1
xg [ g G be the conjugacy class
of x in G. Then [x
G
[ = [G : C
G
(x)].
The centre of G is the set
Z(G) = z G [ zg = gz for all g G .
We saw in tutorials that Z(G) is a subgroup of G; in fact, this is clear because Z(G) =

xG
C
G
(x). Further, Z(G) is a normal subgroup of G.
Let p be a prime. A pgroup is any group G of ppower order; that is, [G[ = p

, for some
0. (It is convenient to call the trivial group 1
G
a pgroup.)
19.2 Theorem Suppose that p is a prime and that G is a nontrivial pgroup. Then G has a
nontrivial centre; that is, Z(G) ,= 1
G
.
Proof Let G act on itself by conjugation and for each x G let x
G
= g
1
xg [ g G be
the conjugacy class of x. Observe that z G belongs to Z(G) if and only if gz = zg or,
equivalently, z = g
1
zg, for all g G. Therefore, z Z(G) if and only if z
G
= z. In
particular, the conjugacy class x
G
intersects Z(G) if and only if x
G
= x (and x Z(G)).
Let x
G
1
, . . . , x
G
m
be the noncentral conjugacy classes of G. Then for each i there exists an
integer a
i
0 such that [x
G
i
[ = [G : G
x
i
] = p
a
i
by Corollary 19.1 (and Lagranges theo-
rem). In fact, by the last paragraph a
i
1 for all i since x
i
/ Z(G). By Lemma 18.4 (or
Proposition 12.3), G is a disjoint union of its conjugacy classes, so
p

= [G[ = [Z(G)[ +
m

i=1
[x
G
i
[ = [Z(G)[ +
m

i=1
p
a
i
.
Reading this equation modulo p shows that [Z(G)[ 0 (mod p). Consequently, [Z(G)[ p
and G has a nontrivial centre as required.
Notice that by Lagranges theorem the order of Z(G) must be a power of p.
19.3 Corollary Suppose that p is a prime and that G is a group of order p
2
. Then G is abelian.
Proof Notice that G is abelian if and only if Z(G) = G. By way of contradiction suppose that
Z(G) ,= G. By Theorem 19.2 the center of G is nontrivial; so Z(G) must contain exactly p
elements by Lagranges theorem. Now, Z(G) is a normal subgroup of G and the quotient group
Direct and semidirect products 47
G/Z(G) contains p elements. By Corollary 11.9, the group G/Z(G) is cyclic, G/Z(G) =
gZ(G)). We have
G = Z(G) gZ(G) g
2
Z(G) g
p1
Z(G).
Then any two elements h and h

of G commute. Indeed, h = g
k
z and h

= g
l
z

for some k, l
and some z, z

Z(G). Therefore, hh

= g
k
zg
l
z

= g
k
g
l
zz

= g
l
g
k
z

z = g
l
z

g
k
z = h

h. This
implies that every element of Gbelongs to Z(G), contradiction. So Z(G) = Gand Gis abelian
as claimed.
There are two obvious groups of order p
2
; namely Z
p
2 and Z
p
Z
p
. Using the last result it
is not so hard to see that, up to isomorphism, these are the only groups of order p
2
. (The phrase
up to isomorphism means that if G is a group of order p
2
then either G

= Z
p
2 or G

= Z
p
Z
p
.)
20. Direct and semidirect products
In tutorials we have come across the direct product A N = (a, n) [ a A and n N
of two groups A and N. As our second application of group actions we now look at the more
general notion of semidirect products. We also ask when a group is a (semi)direct product of
two of its subgroups.
Given any set X we know that the set of bijections from X to X forms a group under com-
position of maps. Now, consider the case where X = G is a group and rather than all bijections
consider just the group isomorphisms : GG; again, these maps form a group under com-
position.
20.1 Denition Let G be a group.
a) An automorphism of G is a group isomorphism : GG.
b) The automorphism group Aut(G) of G is the group of all automorphisms of G under com-
position of maps.
Recall that by convention (see Example 8.4(l)) we are composing maps on the right: if
and are automorphisms of G then is the map which sends g G to ((g)); the picture
is the following.
G G G

This is important because the reason why we care about automorphisms is that Aut(G) acts
on G (from the right). (If we dened ()(g) = ((g)) then we would get a left action
of Aut(G) on G.)
20.2 Proposition Suppose that G is a group. Then Aut(G) acts on G via
g

= (g), for all Aut(G) and g G.


Proof The identity of Aut(G) is the identity map id
G
on G; by denition, g
id
G
= id
G
(g) = g,
for all g G. Similarly, if , Aut(G) then (using our convention)
g

= ()(g) = ((g)) = ((g))

= (g

,
for all g G. Hence, Aut(G) acts on G.
When one group acts on another group we want to have more than just an ordinary action:
the action should respect the operations in both groups. For a general action there is no reason
to expect that (gh)

= g

; however, this is true for the action of Aut(G) on G.


48 Algebra (Advanced)
20.3 Lemma Let G be a group. Then
1

G
= 1
G
, (g
1
)

= (g

)
1
and (gh)

= g

,
for all , Aut(G) and all g, h G.
Proof This is really just a special case of Proposition 13.3. First, 1

G
= (1
G
) = 1
G
. Next,
applying the denitions,
(g
1
)

= (g
1
) =
_
(g)
_
1
=
_
g

)
1
.
Similarly, (gh)

= (gh) = (g)(h) = g

.
Therefore, in order to make a group A act nicely on a group N we need to nd a homo-
morphism from A into Aut(N); this observation is the basis of the semidirect product of two
groups.
20.4 Denition Suppose that A and N are groups and that : A Aut(N); b
b
is a
group homomorphism. Then the semidirect product of A and N is the group
AN = (a, g) [ a A and g N
with operation (a, g) (b, h) = (ab, g
b
h), for all a, b A and g, h N and where g
b
=
b
(g).
This looks more complicated than it really is. The point is that A acts on N and this is a very
nice action in the sense that
20.5 1
a
N
= 1
N
, (g
1
)
a
= (g
a
)
1
and (gh)
a
= g
a
h
a
, for all a A and g, h N;
this follows from Lemma 20.3 as the action comes from the homomorphismA Aut(N). It
is more important to remember that the action of A satises these equations than to remember
that the Aaction comes from the action of Aut(N).
In terms of notation, the symbol is meant to remind you that A N is almost a direct
product; indeed, the underlying set is just the product of Aand N and (as we will see shortly), N
is (isomorphic to) a normal subgroup of A N. It is not uncommon for people to write the
semidirect product the other way around as N A.
There is one major problem with Denition 20.4; namely, we do not yet know that AN is
actually a group.
20.6 Proposition Suppose that A and N are groups and that : AAut(N); a
a
is a
group homomorphism. Then AN is a group.
Proof First, we need to show that multiplication in A N is associative; if (a, g), (b, h)
and (c, k) are elements of AN then
(a, g)
_
(b, h)(c, k)
_
= (a, g)(bc, h
c
k) = (abc, g
bc
h
c
k)
= (abc, (g
b
h)
c
k) = (ab, g
b
h)(c, k) =
_
(a, g)(b, h)
_
(c, k),
where the rst equality on the second line uses (20.5).
Direct and semidirect products 49
Next, observe that (1
A
, 1
N
) is an identity element in A N because if (a, g) AN then
(a, g)(1
A
, 1
N
) = (a, g
1
A
) = (a, g) = (1
A
, 1
N
)(a, g);
for the last equality note that 1
a
N
= 1
N
since every automorphism of N maps 1
N
to 1
N
. Finally,
the inverse of (a, g) is (a
1
, (g
1
)
a
1
) = (a
1
, (g
a
1
)
1
); again, these are equal by (20.5). To
check this we just multiply:
(a, g)
_
a
1
, (g
1
)
a
1
_
=
_
aa
1
, g
a
1
(g
1
)
a
1
_
=
_
1
A
, (gg
1
)
a
1
_
= (1
A
, 1
N
)
and
(a
1
, (g
1
)
a
1
)(a, g) =
_
a
1
a, ((g
1
)
a
1
)
a
g
_
=
_
1
A
, (g
1
)
aa
1
g
_
= (1
A
, g
1
g) = (1
A
, 1
N
).
Hence, AN is a group as claimed.
20.7 Examples a) Let D
n
be the dihedral group of order 2n; thus, D
n
is the symmetry
group of the regular ngon (see Example 8.4(o)). In Example 10.7 we saw that D
n
is
generated by elements r and s, where r is rotation through 2/n and s is a reection,
and that r
n
= s
2
= 1 and srs = r
1
. Actually, this example looked specically at
D
8
; however, as we remarked then, the argument applies to all of the dihedral groups.
Let A = s) = 1, s and N = r) = 1, r, r
2
, . . . , r
7
; we claim that D
n

= A N.
In order for this to make sense we need to have an action of A upon N. Luckily, we
already have such an action because srs = r
1
, so we can dene (r
b
)
s
:= sr
b
s = r
b
,
for 0 b < n; so A acts on N by conjugation.
Notice that conjugation by s denes an action on N only because N is a normal sub-
group of D
n
. Also, the map A Aut(N) sends 1
A
to id
N
and s to the homomorphism
: N N given by (g) = g
1
, for all g N. Note that is a group homomorphism
because N = r) is an abelian group (if g, h N then (gh) = (gh)
1
= h
1
g
1
=
g
1
h
1
= (g)(h)).
We have now checked that the semidirect product A N makes sense; we have still
to establish that D
n

= A N. Let : GA N be the map (s
a
r
b
) = (s
a
, r
b
), for
a = 0, 1 and 0 b < n. By denition is a bijection so we need only check that it is a
group homomorphism. By denition, in AN multiplication is given by
(s
a
, r
b
)(s
c
, r
d
) = (s
a
s
c
, (r
b
)
s
c
r
d
) =
_
(s
a+c
, r
d+b
), if c = 0,
(s
a+c
, r
db
), if c = 1,
where the last equality follows because (r
b
)
s
= r
b
. Comparing this with the multipli-
cation in D
n
(see the end of Example 10.7), we see that does indeed dene an isomor-
phism. Hence, D
n

= AN.
b) Let A act on N trivially; that is, g
a
= g for all a A and all g N. Then the multiplica-
tion in A N becomes (a, g)(b, h) = (ab, g
b
h) = (ab, gh); hence A N = A N. In
other words, the direct product of two groups is the same as the semidirect product with
a trivial action. Generally, when we write H

= A N we implicitly mean that A acts
nontrivially on N, so H ,

= A N.
50 Algebra (Advanced)
Given two groups A and N (and a homomorphism : AAut(N)), we have articially
constructed a new group A N. We now ask how we can recognise when a group H is a
semidirect product; more precisely, if H contains subgroups A and N when is H

= AN?
First consider this question for A N itself. The semidirect product has two obvious sub-
groups

A = (a, 1) [ a A and

N = (1, g) [ g N and it is easy to check that A

=

A
and N

=

N. Moreover, (a) A N =

A

N, (b)

A

N = (1, 1) = 1
AN
and (c)

N is a normal
subgroup of A N because
(a, g)(1, h)(a, g)
1
= (a, g)(1, h)
_
a
1
, (g
1
)
a
1
_
= (a, gh)
_
a
1
, (g
1
)
a
1
_
=
_
1, (gh)
a
1
(g
1
)
a
1
_


N.
(In fact, this last expression is equal to
_
1, (ghg
1
)
a
1
_
, but we dont need to know this.) It is
also easy to see that (A N)/

N

= A. In general, these properties of the subgroups

A and

N
determine when a group can be written as a semidirect product of two of its subgroups.
20.8 Proposition Suppose that Gis a group which contains two subgroups Aand N such that
a) G = AN;
b) A N = 1
G
; and,
c) N G.
Then G

= AN, where A acts on N via conjugation: g
a
= a
1
ga, for all a A and g N.
Proof First note that if g N and a A then a
1
ga N since N is a normal subgroup of G;
therefore, the denition g
a
= a
1
ga does give an action of A on N.
Next, if x G then x = ag for some a A and g N by (a). We claim that, in fact, a
and g are uniquely determined by x; that is, if x = ag and x = bh then a = b and g = h, for
a, b A and g, h N. To see this suppose that ag = bh, for some a, b A and g, h N.
Then b
1
a = hg
1
A N = 1
N
; so b
1
a = 1 and hg
1
= 1, whence a = b and g = h as
claimed. Therefore, we can dene a bijection
: GAN; x = ag (a, g).
To complete the proof we need to show that is a group homomorphism. Let x = ag and
y = bh be two elements of G. Then xy = a(gb)h = a(bg

)h, for some g N since N G.


Notice that gb = bg

, so g

= b
1
gb. Therefore,
(xy) = (abg

h) = (ab, g

h) =
_
ab, (b
1
gb)h
_
= (ab, g
b
h) = (a, g)(b, g) = (x)(y).
Hence, is a homomorphism and the Proposition follows.
In light of this result, the example above which shows that D
n
is a semidirect product of two
of its subgroups becomes much clearer.
20.9 Example Consider G = Alt(4), the alternating group of degree 4. In Question 4 of
Tutorial 7 we saw that N = 1, (1, 2)(3, 4), (1, 3)(2, 4), (1, 4)(2, 3) is the unique subgroup
of Alt(4) of order 4; hence, N is normal. Let A = (1, 2, 3)) = 1, (1, 2, 3), (1, 3, 2). Then
H = AN, A N = 1
G
and N G; therefore, Alt(4)

= AN

= Z
3
(Z
2
Z
2
).
20.10 Corollary Suppose that H is a group which contains two subgroups A and N such that
a) H = AN;
The structure of groups II: Sylows rst theorem 51
b) A N = 1
H
; and,
c) N H and A H.
Then H

= AN.
Proof By the Proposition H

= A N, where A acts on N by conjugation. Therefore, it is
enough to show that A acts on N trivially; in other words, we need to show that a
1
ga = g
for all a A and all g N. Now, as in the proof of the Proposition, if a A and g N
then ga = ag

for some g

N since N is a normal subgroup; however, now A is also normal


so ga = a

g for some a

A. Therefore, ag

= a

g, so a = a

and g = g

since every element


of H can be expressed uniquely in the form bh, for some b A and h N. Hence, ga = ag,
for all a A and g N; so a
1
ga = g and H

= AN as required.
21. The structure of groups II: Sylows rst theorem
In Theorem 17.2 we proved that if G is a nite abelian group of order divisible by a prime p
then G contains an element of order p (and hence also a subgroup of order p). As our third
application of group actions we now prove a series of much stronger results which hold for all
nite groups, and not just the abelian ones.
As usual, we start with a (combinatorial) lemma.
21.1 Lemma Let p be a prime and m a positive integer. Suppose that p
r
is the largest power
of p dividing m. Then p
r
is also the largest power of p dividing
_
mp

_
for any 0.
Proof We can factorize
_
mp

_
as
_
mp

_
=
(mp

)!
p

!
_
(m1)p

_
!
=
mp

(mp

1)
(p

1)

(mp

i)
(p

i)

(mp

+ 1)
(p

+ 1)
.
As the same power of p divides mp

i and p

i, for 0 i p

1, the lemma follows.


For any integer n let n
p
be the ppart of n; that is n
p
= p

is the largest power of p which


divides n. Thus, n = kn
p
for some integer k which is not divisible by p. In particular, if X is a
set we write [X[
p
for the ppart of the order of X.
21.2 Theorem Suppose that G is a nite group and that p

divides [G[, where p is a prime


and 0. Then G has a subgroup of order p

.
Proof Let ( = X G [ p

= [X[ be the set of subsets of G of size p

. Now, [([ =
_
|G|
p

_
.
Since p

divides [G[ we can write [G[ = mp

, for some m; then [([ =


_
mp

_
. Fix r 0 such
that p
r
is that largest power of p which divides m (so [G[
p
= p
r+
). By the Lemma, p
r
is also
the largest power of p which divides [([.
Let G act on ( by right multiplication; if X ( then X
g
:= Xg, for all g G. Then (
is a disjoint union of Gorbits by Lemma 18.4. We claim that there exists an X ( such
that [X
G
[
p
p
r
; thus, p
r+1
[X
G
[. If this were not true then p
r+1
would divide [X
G
[ for all
X (, so p
r+1
would also divide [([, contrary to the rst paragraph.
Fix X ( such that [X
G
[
p
p
r
. Replacing X with Xx
1
(for some x X), if necessary,
we may assume that 1 X. Now consider the stabilizer G
X
= g G [ Xg = X of X.
52 Algebra (Advanced)
Then G
X
is a subgroup of G and G
X
X since 1 X. Therefore, [G
X
[ p

= [X[. On the
other hand, by the orbit stabilizer theorem, [G : G
X
] = [X
G
[; so
[G
X
[
p
= [G[
p
/[X
G
[
p
p
r+
/p
r
= p

.
Hence, [G
X
[ p

. Therefore, [G
X
[ = p

, and G
X
is a subgroup of G of order p

.
The proof actually shows that if 1 X then G
X
is a subgroup of order p

whenever p
r+1
does not divide [X
G
[; moreover, by the orbit stabilizer theorem, the Gorbit X
G
consists of the
cosets of G
X
in G.
As an important special case the theorem yields the rst Sylow theorem.
21.3 Corollary (Sylows rst theorem) Let G be a nite group and let p be a prime. Then G
has a subgroup of order [G[
p
.
21.4 Denition Let G be a nite group and p a prime. Then a Sylow psubgroup of G is any
subgroup of order [G[
p
. Let Syl
p
(G) be the set of Sylow psubgroups of G.
Observe that if P is a Sylow psubgroup then so is P
g
= g
1
Pg, for all g G. Indeed we
saw in tutorials that P
g

= P; so, in particular, [P
g
[ = [P[ = [G[
p
. We will see in the next
section that, in fact, every Sylow psubgroup is equal to P
g
for some g G.
21.5 Examples a) Suppose that Gis a nite abelian group. Then G
p
= g G [ g
|G|p
= 1
is the unique Sylow p-subgroup of G. See tutorials.
b) Let G = Sym(4). Then the order of G is 24 = 2
3
3 so G has a Sylow 2subgroup of
order 8 and a Sylow 3subgroup of order 3. There are four Sylow 3subgroups; namely
the subgroups G
abc
= (a, b, c)) = 1, (a, b, c), (a, c, b), where a, b, c is any three
element subset of 1, 2, 3, 4. There are three Sylow 2subgroups; however, they are
more difcult to track down because they have order p
3
. For example, it is tempting to
look at subgroups like (1, 2), (1, 2, 3, 4)); but you can (and should) check that this is the
whole of Sym(4). The Sylow 2subgroups are the subgroups
(a, b), (c, d), (a, c)(b, d)) = (a, b), (a, d, b, c)),
for a, b, c, d = 1, 2, 3, 4. The easiest way to see that this is a group of order 8 is to
observe that it is the symmetry group of the following graph.
a c
d b
You can check that the Sylow 2subgroups are isomorphic to C
2
(C
2
C
2
); the action
of C
2
on C
2
C
2
interchanges the two factors.
Notice that in all of these examples the different Sylow psubgroups were always isomorphic
and that the number of Sylow psubgroups was always congruent to 1 modulo p.
Sylows second theorem 53
22. Sylows second theorem
Recall that if H is a subgroup of H then H
g
= g
1
Hg = g
1
hg [ h H is a subgroup
of G and that H

= H
g
(the isomorphism is the obvious one: hg
1
hg, for all h H). Two
subgroups H and K are conjugate if K = H
g
, for some g G.
22.1 Theorem (The second Sylow theorem)
Suppose that G is a nite group and that p is a prime. Then
a) any two Sylow psubgroups of G are conjugate; and,
b) the number [ Syl
p
(G)[ of Sylow psubgroups divides
|G|
|G|p
and [ Syl
p
(G)[ 1 (mod p).
In particular, part (a) shows that different Sylow psubgroups are always isomorphic since
H

= H
g
, for any subgroup H and any g G. (Part (a) is often referred to as the second
Sylow theorem, and part (b) as the third; as we prove both parts simultaneously we will call
Theorem 22.1 the second Sylow theorem.)
The basic idea of the proof is to let G act on the set of Sylow psubgroups by conjugation.
Part (a) then says that Syl
p
(G) is a single Gorbit. To prove (b) we look at the action of a xed
Sylow psubgroup P on Syl
p
(G) and show that if Q Syl
p
(G) and Q
a
= Q, for all a P,
then P = Q.
Before we can prove Theorem 22.1 we need a few lemmas. Recall that if H and K are any
two subsets of a group G then HK = hk [ h H and k K is dened to be their setwise
product.
22.2 Lemma Suppose that H and K are subgroups of a nite group G. Then [HK[ =
|H||K|
|HK|
.
Proof First note that H K is a subgroup of both H and K. Therefore, we can nd a
set k
1
, . . . .k
m
of right coset representatives of H K in K. Hence, K =

m
i=1
(H K)k
i
and
HK =
m
_
i=1
H(H K)k
i
=
m
_
i=1
Hk
i
,
since HK is also a subgroup of H. In fact, HK is a disjoint union of the cosets Hk
i
because
if Hk
i
= Hk
j
, for some i and j, then k
i
k
1
j
H; therefore, k
i
k
1
j
HK (since k
i
, k
j
K),
so (H K)k
i
= (H K)k
j
and i = j. Hence, [HK[ = m[H[ = [K : H K] [H[ =
|H||K|
|HK|
.
If H is a subgroup of G then its normalizer is
N
G
(H) = g G [ g
1
Hg = H .
In other words, N
G
(H) is the stabilizer of H when G acts on the set of its subgroups by conju-
gation; in particular, N
G
(H) is a subgroup of G.
In general if H and K are subgroups of Gthen HK is not a subgroup of G. The next Lemma
says that HK is a subgroup of G whenever we choose K carefully enough.
22.3 Lemma Suppose that H is a subgroup of G and that K N
G
(H). Then HK is a
subgroup of G.
54 Algebra (Advanced)
Proof See tutorials.
Note in particular that if H is a normal subgroup then N
G
(H) = G so HK is always a
subgroup of G.
We are almost ready. If G acts on a set X let
Fix
G
(X) = x X [ x
g
= x for all g G
be the set of Gxed points of X. Recall that P is a pgroup if P is a group of ppower order;
that is, [P[ = p
n
for some n.
22.4 Lemma Let p be a prime and suppose P is a pgroup which acts on a set X. Then
[X[ [ Fix
P
(X)[ (mod p).
Proof Let x X. Then x Fix
P
(X) if and only if x
P
= x; hence, by the orbit stabilizer
theorem, x Fix
P
(X) if and only if P
x
= P, so [P : P
x
] = 1. Therefore, if y / Fix
P
(X)
then P
y
is a proper subgroup of P and [P : P
y
] 0 (mod p). Hence, if we write X =
x
P
1

x
P
k
as a disjoint union of orbits then [X[ = [x
P
1
[+ +[x
P
k
[ [ Fix
P
(X)[ (mod p),
as claimed.
Notice that we used exactly this argument in the proof of Theorem 19.2.
We can now prove the second Sylow theorem. We prove both statements simultaneously.
Proof of Theorem 22.1 Let o = Syl
p
(G) be the set of Sylow psubgroups. Then G acts on o
by conjugation: if Q o then Q
g
= g
1
Qg o. Let P be a Sylow subgroup P o; then P
also acts on o by conjugation.
Claim: If P is a Sylow psubgroup of G then Fix
P
(o) = P.
Suppose that Q Fix
P
(o); that is, Q
a
= Qfor all a P. Then P is contained in the normalizer
N
G
(Q) = g G [ Q
g
= Q of Q. Therefore, PQ is a subgroup of G, by Lemma 22.3, and
[PQ[ =
|P||Q|
|PQ|
[P[ by Lemma 22.2. However, P and Q are subgroups of G of maximal
ppower order, so we must have [PQ[ = [P[ = [Q[. Hence, Q = PQ = P, which establishes
the claim.
Claim: o is a single Gorbit.
Write o = P
G
o

, where o

= o P
G
is a union of Gorbits; we want to show that
o

= . As both P
G
and o

are a union of orbits, P acts on P


G
and P also acts on o

. By the
previous claim, Fix
P
(P
G
) = P and Fix
P
(o

) = (since P P
G
and P / o

). Therefore,
[P
G
[ [ Fix
P
(P
G
)[ = 1 (mod p) and [o

[ [ Fix
P
(o

)[ = 0 (mod p), by Lemma 22.4.


Now suppose, if possible, that o

,= . Then there is a Sylow psubgroup Q o

. As
above, Q acts on o

and Fix
Q
(o

) = Q by the rst claim. Therefore, [o

[ [ Fix
Q
(o

)[ = 1
(mod p) by Lemma 22.4. This contradicts the last paragraph, so o

= and o = P
G
as
claimed.
Groups of order pq 55
We have now proved almost everything in Theorem 22.1; part (a) follows because o = P
G
is a single Gorbit, and the last claim in part (b) follows because [o[ = [P
G
[ 1 (mod p).
Finally, by the orbit stabilizer theorem
[o[ = [P
G
[ = [G : N
G
(P)] =
[G[
[N
G
(P)[
=
[G[
[P[
_
[N
G
(P)[
[P[
=
[G[
[G[
p
1
[N
G
(P) : P]
.
Hence, [ Syl
P
(G)[ divides
|G|
|G|p
.
The Sylow theorems actually tell us quite a lot about the structure of nite groups.
22.5 Example We claim that any group of order 45 is abelian. To see this let G be a group of
order 45 and note that 45 = 3
2
5. Therefore, G contains Sylow 3subgroups (of order 9) and
Sylow 5 subgroups (of order 5). Let s
p
= [ Syl
p
(G)[ be number of Sylow psubgroups of G, for
p = 3, 5. By Theorem 22.1, s
3
divides 5 and s
3
1 (mod 3); so s
3
= 1. Similarly, s
5
divides 9
and s
5
1 (mod 5); so s
5
= 1 as well. Hence, G has a is a unique Sylow 3subgroup S
3
and a
unique Sylow 5subgroup S
5
. It follows that S
3
and S
5
are both normal subgroups of G (since
S
g
p
= S
p
for all g G). Hence, S
3
S
5
is a subgroup of G (by Lemma 22.3), so G = S
3
S
5
since
[S
3
S
5
[ = [S
3
[ [S
5
[/[S
3
S
5
[ = 9 5 = [G[. Therefore, G

= S
3
S
5
by Corollary 20.10.
Finally, G is abelian because S
5

= Z
5
is abelian, and S
3
is abelian by Corollary 19.3 and
the direct product of abelian groups is abelian. In fact, we have shown that, up to isomorphism,
there are exactly two groups of order 45; namely, the groups Z
3
Z
3
Z
5
and Z
9
Z
5
.
As we saw in the last example, we can also tell when a Sylow psubgroup is normal.
22.6 Corollary Let Gbe a nite group. Then a Sylow psubgroup P of G is normal if and only
if [ Syl
p
(G)[ = 1.
Proof Any group P is normal if and only if P
g
= P for all g G. By the second Sylow
theorem (Theorem 22.1), Syl
p
(G) = P
g
[ g G; so P is a normal subgroup of G if and
only if Syl
p
(G) = P.
22.7 Corollary Let G be a nite abelian group. Then [ Syl
p
(G)[ = 1 for any prime p.
Proof This is immediate from the previous corollary because any subgroup of an abelian group
is normal.
23. Groups of order pq
We now want to explicitly describe the groups of order pq, for primes p and q. It turns out that
there are two cases; we illustrate this with two examples. Throughout we let s
p
= [ Syl
p
(G)[ be
the number of Sylow psubgroups of G and let S
p
be a xed Sylow psubgroup.
23.1 Examples a) Suppose that G is a group of order 15. Then G has non-trivial Sylow 3
and Sylow 5 subgroups. By the second Sylow theorem s
3
[5 and s
3
1 (mod 3); so
s
3
= 1. Similarly, s
5
= 1. Therefore, G contains unique Sylow psubgroups S
3
and S
5
.
Hence, both Sylow psubgroups are normal (by Corollary 22.6). Evidently, S
3
S
5
= 1
and G = S
3
S
5
. Therefore, by Corollary 20.10, G

= S
3
S
5

= Z
3
Z
5

= Z
15
.
56 Algebra (Advanced)
b) Suppose now that G is a group of order 6; perhaps surprisingly, this is harder than the
rst example. Then s
2
[3 and s
2
1 (mod 2), so s
2
= 1 or s
2
= 3. Similarly, s
3
= 1.
Therefore, there is a unique (hence normal) Sylow 3subgroup S
3

= Z
3
and 1 or 3 Sylow
2subgroups S
2

= Z
2
. Note also that S
2
S
3
= 1 and that G = S
2
S
3
for any S
2
.
If s
2
= 1 then S
2
is also normal in G, so G

= S
2
S
3

= Z
2
Z
3

= Z
6
, as in the rst
example. In particular, note that G is abelian in this case,
If s
2
= 3 then gS
2
,= S
2
g for some g G. So, if S
2
= x) = 1, x then gx ,= xg; in
particular, Gis not abelian. Hence, G

= S
2
S
3
by Proposition 20.8 since S
3
is a normal
subgroup. Note that if S
3
= y) = 1, y, y
1
then xyx = x
1
yx S
3
so either xyx = y
or xyx = y
1
; however, xyx = y if and only if xy = yx, which cannot happen since G
is not abelian. Hence, xyx = y
1
and G

= D
3
.
We actually know three groups of order 6; namely, Z
6

= Z
2
Z
3
, Sym(3) and D
3
.
As Sym(3) is not abelian we have just shown that Sym(3)

= D
3
; this is easy to check
directly.
We now classify all groups of order pq, for p and q prime. If p = q then G is a group of
order p
2
and we already know that Gis isomorphic to either Z
p
Z
p
or Z
p
2 by the remarks after
Corollary 19.3. It remains then to consider the cases where p > q.
23.2 Theorem Suppose that G is a group of order pq where p and q are primes with p > q.
Then G is abelian unless p 1 (mod q); further
G

=
_
C
pq
, if G is abelian,
C
q
C
p
, if G is not abelian (and p 1 (mod q)).
In the nonabelian case we can choose generators x and y of C
q
and C
p
, respectively, so that
the action of C
q
on C
p
is given by y
x
= y
a
for some value of a 2, 3, . . . , p 1 satisfying
a
q
1 (mod p). Moreover, all nonabelian groups of order pq are isomorphic to each other.
Proof As above, let S
q
and S
p
be xed Sylow q and psubgroups, respectively, and let s
q
=
[ Syl
q
(G)[ and s
p
= [ Syl
p
(G)[. Then S
q

= C
q
and S
p

= C
p
. By Theorem 22.1(b), s
p
divides q
and s
p
1 (mod p); as p > q this forces s
p
= 1. Hence, S
p
is the unique Sylow p-subgroup
of G and S
p
is necessarily normal in G (by Corollary 22.6). Similarly, s
q
divides p and s
q
1
(mod q); therefore, s
q
= 1 or s
q
= p, the second case only being possible if p 1 (mod q).
Next observe that S
p
S
q
= 1 since S
p
S
q
is a subgroup of both S
p
and S
q
; so the order
of S
p
S
q
divides both p and q by Lagranges theorem. Further, G = S
q
S
p
since [S
q
S
p
[ = pq
by Lemma 22.2. It remains to determine how S
q
acts on S
p
.
Suppose rst that s
q
= 1. Then S
q
is the unique Sylow qsubgroup of G, so S
q
is normal.
Hence, G

= S
q
S
p
by Corollary 20.10; thus, G

= C
q
C
p

= C
pq
.
Finally, suppose that s
q
= p (so p 1 (mod q)). Write S
q
= x) and S
p
= y). Now
G = S
q
S
p
, so G is abelian if and only if xy = yx since S
p
= y); however, if xy = yx then
gS
q
g
1
= S
q
for all g G, so s
q
= 1. Hence, G is not abelian when s
q
= p, and S
q
acts on S
p
nontrivially in this case, that is, x
1
yx = y
a
for some a 2, 3, . . . , p1. However, the map
x
x
with
x
: y y
a
denes a homomorphism S
q
Aut S
p
. Hence, (
x
)
q
=
x
q = id
is the identity homomorphism S
p
S
p
. On the other hand, (
x
)
q
: y y
a
q
. Thus, we must
have a
q
1 (mod p). The proof of the nal claim we leave to the tutorials.
We have written C
q
and C
p
(rather than Z
q
and Z
p
) in the statement and the proof only
because throughout the group operation is written multiplicatively; as C
n

= Z
n
we could just
as easily have used Z
q
and Z
p
.
Simple groups of small order 57
23.3 Example Suppose that G is a nonabelian group of order 2p where p > 2 is a prime.
We claim that G

= D
p
, the dihedral group of order 2p. Notice that p 1 (mod 2), so, by
Theorem 23.2, G

= C
2
C
p
with C
2
= x) and C
p
= y), where the action of C
2
on C
p
is
given by xyx = y
a
with a
2
1 (mod p). Hence, p divides (a1)(a+1). The only admissible
value of a with this property is a = p 1. Thus, xyx = y
p1
= y
1
. Consequently, G

= D
p
since
D
p
= r, s [ r
p
= 1 = s
2
and srs = r
1
) ;
an isomorphism is given by x
n
y
m
s
n
r
m
.
23.4 Example Let us nd all nonabelian groups of order 21. We have 21 = 7 3. So, taking
p = 7 and q = 3 notice that p 1 (mod q). By Theorem 23.2, the value of a 2, 3, 4, 5, 6
is found from the relation a
3
1 (mod 7). Hence, the possible values are a = 2 and a = 4.
The groups corresponding to these values are isomorphic to each other. In order to see this, note
that in the rst group we have x
1
yx = y
2
. This implies x
2
yx
2
= y
4
. However, x = x
2
is also
a generator of C
3
and we have x
1
y x = y
4
.
24. Simple groups of small order
A group G is simple if it has no proper nontrivial normal subgroups. The simple groups are
the basic building blocks of group theory: if a group G is not simple then it has a nontrivial
proper subgroup N and questions about G can often be reduced to corresponding questions
about the smaller groups G/N and N we sawsuch an argument in the proof of Theorem17.2.
Quite amazingly, all of the nite simple groups are known. It turns out that there are several
innite families of nite simple groups (such as PSL
n
(F) = SL
n
(F)/T, where T is the normal
subgroup of SL
n
(F) consisting of scalar matrices; here, F is a nite eld), together with 26
sporadic groups which dont t any nice pattern. This result was one of the major highlights of
twentieth century mathematics; its proof runs to over 10,000 pages of mathematics and was the
work of several hundred mathematicians. We wont give the proof here. Rather, we will use the
results that we have developed to classify the simple groups of order less than 60.
24.1 Proposition Suppose that G is a group of order less than 60. Then G is simple if and only
if the order of G is prime; in which case G

= C
p
, for some prime p.
Proof The basic strategy is to use the Sylow theorems to nd a normal subgroup of G. Essen-
tially, we need to consider the groups on a casebycase basis using only their order; however,
we can be a little bit more sophisticated.
First, suppose [G[ = p, where p is a prime. If g G and g ,= 1 then the order of g must be p
so [g)[ = [g[ = p = [G[. Therefore, G = g)

= C
p
is simple.
Next, suppose that [G[ = pq where p > q are primes. Then the Sylow psubgroup of G is
normal by Theorem 23.2; hence, G is not simple.
Now suppose that [G[ = p
a
where p is prime and a > 1. If Gis abelian then it is not simple as
it contains a proper (and hence normal) subgroup order p by Theorem 21.2. If G is not abelian
then the centre Z(G) of G is nontrivial by Theorem 19.2; as Z(G) is a normal subgroup this
shows that G is not simple.
We have now reduced to considering those groups Gof order less than 60 such that [G[ is not
a prime power and is not the product of two distinct primes. That is, it remains to consider the
groups of order 12, 18, 20, 24, 28, 30, 36, 40, 42, 44, 45, 48, 50, 52, 54 and 56. We consider
each of these in turn. The following lemma will be useful for some cases.
Lemma Let s
p
> 1 and s
p
! < [G[. Then G has a nontrivial proper normal subgroup.
58 Algebra (Advanced)
Proof Let P
1
, . . . , P
sp
be the Sylow psubgroups. Each element g G determines a permuta-
tion
g
of these subgroups by
(P
g
1
, . . . , P
g
sp
) = (P
g(1)
, . . . , P
g(sp)
).
The map g
g
denes a homomorphism : G Sym(s
p
). Its kernel ker is a normal
subgroup of G. We have ker ,= 1 since [G[ > s
p
! and ker ,= G since [ im[ > 1.
If G has order 12 then [S
2
[ = 4 and s
2
1 (mod 2) and s
2
divides 3 by Theorem 22.1;
so s
2
= 1 or s
2
= 3. Similarly, s
3
= 1 or s
3
= 4. If s
3
= 1 then the Sylow 3subgroup of
G is normal, so G is not simple. Suppose, if possible, that s
3
= 4. Then there are four Sylow
3subgroups; say P
1
, P
2
, P
3
and P
4
. For i ,= j we have P
i
P
j
= 1
G
(since [P
i
[ = 3). As
each P
i
contains two elements of order 3, it follows that G has exactly 8 = 2 4 elements of
order 3. Hence, G has 4 = 12 8 elements of order not equal to 3. As each Sylow 2subgroup
contains 4 elements it follows that there must be a unique Sylow 2subgroup. Therefore, the
Sylow 2subgroup of G is normal and G is not simple. (In fact, if s
3
= 4 then G

= Alt(4).)
If G has order 18 then s
3
= 1, so the Sylow 3subgroup is normal and G is not simple.
If G has order 20 then s
5
= 1, so the Sylow 5subgroup is normal and G is not simple.
If G has order 24 then s
2
= 1, 3 and s
3
= 1, 4. If s
2
= 1 or s
3
= 1 then the corresponding
Sylow subgroup is normal showing that G is not simple; so suppose that s
2
= 3 and s
3
= 4.
Let P
1
, P
2
, P
3
be the three Sylow 2subgroups. Then [P
i
[ = 8 for each i. Therefore, for i ,= j
24 = [G[ [P
i
P
j
[ =
[P
i
[ [P
j
[
[P
i
P
j
[
=
64
[P
i
P
j
[
by Lemma 22.2; so [P
i
P
j
[
64
24
> 2. By Lagranges theorem [P
i
P
j
[ = 1, 2, 4 since P
i
P
j
is a proper subgroup of P
i
; so [P
i
P
j
[ = 4. Similarly,
24 = [G[ [P
1
(P
2
P
3
)[ =
[P
1
[ [P
2
P
3
[
[P
1
P
2
P
3
[
=
32
[P
1
P
2
P
3
[
;
so [P
1
P
2
P
3
[ = 2 or 4. However, if g G then for each i there is a unique j such that
P
g
i
= g
1
P
i
g = P
j
; therefore, P
1
P
2
P
3
is a normal subgroup of G. Hence, G is not simple.
(Note that the symmetric group Sym(4) is an example of such a group.)
If G has order 28 then s
7
= 1, so the Sylow 7subgroup is normal and G is not simple.
We leave the case [G[ = 30 as an exercise.
If G has order 36 then s
3
= 1 or 4. If s
3
= 1, then the Sylow 3subgroup is normal and G is
not simple. If s
3
= 4 then s
3
! = 24 < 36 = [G[, so the Lemma applies, G is not simple.
If G has order 40 then there is a unique normal Sylow 5subgroup, so G is not simple.
If G has order 42 then there is a unique normal Sylow 7subgroup, so G is not simple.
If G has order 44 then there is a unique normal Sylow 11subgroup, so G is not simple.
If G has order 45 then there is a unique normal Sylow 3subgroup, so G is not simple.
If G has order 48 then s
2
= 1 or 3. If s
2
= 1, then the Sylow 2subgroup is normal and G is
not simple. If s
2
= 3 then s
2
! = 6 < 48 = [G[, so the Lemma applies, G is not simple.
If G has order 50 then there is a unique normal Sylow 5subgroup, so G is not simple.
If G has order 52 then there is a unique normal Sylow 13subgroup, so G is not simple.
If G has order 54 then there is a unique normal Sylow 3subgroup, so G is not simple.
If G has order 56 then s
2
= 1 or 7 and s
7
= 1 or 8. Suppose, if possible, that there are 8
Sylow 7 subgroups. Each of these subgroups has order 7 and so contains 6 elements of order 7;
Free groups 59
therefore, G contains exactly 48 = 8 6 elements of order 7 (exactly, because each element
of order 7 generates a Sylow subgroup). This means that there are 8 = 56 48 elements of G
of order different from 7. However, each Sylow 2subgroup contains 8 elements, so there can
only be one Sylow 2 subgroup. Hence, G is not simple.
The reason why we have considered only those groups G with [G[ < 60 is not that 60 is a
big number; rather, the alternating group Alt(5) is a simple group of order 60. In fact, Alt(n)
is a simple group for n 5. The proof of this is not particularly hard; however, it is a fairly
grotty and uninspiring calculation so we wont give it.
25. Free groups
In Example 10.7 we described the dihedral group D
n
as the group
D
n
= r, s [ r
n
= 1 = s
2
and srs = r
1
) ,
and we saw that these relations completely determined the multiplication in D
n
. We now put
this notion on a more rigorous footing. First, we consider the groups without any relations.
Intuitively, if such a group is generated by certain elements x and y, it must contain all products
of the elements x and y as well as their inverses, x
1
and y
1
. This motivated the following
denition.
25.1 Denition Let X be a set and let X
1
= x
1
[ x X be a set of formal inverses.
a) A word in X is a nite sequence w = x
1
x
2
. . . x
n
, where n 0, x
i
X X
1
. The word
w has length n and we write (w) = n.
b) A word x
1
x
2
. . . x
n
is reduced if x
i
,= x
1
i+1
, for 1 i < n. Let F
X
be the set of reduced
words in X.
c) Dene the product on F
X
as follows: if x = x
n
. . . x
1
and y = y
1
. . . y
m
are reduced
words then x y is the reduced word obtained by consecutive cancellations in the word
x
n
. . . x
1
y
1
. . . y
m
. More precisely, if k is the maximum index such that x
i
= y
1
i
for all
i = 1, . . . , k then x y = x
n
. . . x
k+1
y
k+1
. . . y
m
.
d) The inverse of a word x = x
1
x
2
. . . x
n
is the word x
1
= x
1
n
. . . x
1
2
x
1
1
, where we dene
(x
1
)
1
= x for all x X.
The set X is often called an alphabet. If we take X = a, b, c . . . , z then cat and dog
are typical words in X; however, we also have other less familiar words like q
1
zwcaa
1
x.
We emphasize that (at this stage) x
1
is only the formal inverse of x. We do not yet have a
group structure to play with. The aim is to dene a group structure on the set of reduced words
in such a way that the inverse of the word w actually is the word w
1
.
We include in this denition the empty word ; this is the unique word of length 0. The empty
word will end up being the identity element in the free group. Notice also that if w is a word
then w = (w
1
)
1
.
Now, a word is reduced if and only if it does not contain xx
1
or x
1
x as a subword, for some
x X. In other words, reduced words are those words in which no cancellation can (morally)
occur. Note also that applying a sequence of cancellations we can bring every word to a reduced
form.
25.2 Example Lets agree to write x
3
for the word xxx and so on. Suppose that X = x, y, z
and let w = x
3
yy
4
z
1
zy
3
x
2
x
3
z
2
y
1
yz
3
. Then w is not reduced; however, applying cancel-
lations we nd that
w = x
3
yy
4
(z
1
z)y
3
(x
2
x
3
)z
2
(y
1
y)z
3
x
3
y(y
4
y
3
)x(z
2
z
3
) x
3
(yy
1
)xz
5
x
4
z
5
,
60 Algebra (Advanced)
and x
4
z
5
is reduced. However, we could have done our reductions differently; for example,
w = x
3
(yy
4
)(z
1
z)y
3
(x
2
x
3
)z
2
(y
1
y)z
3
x
3
(y
3
y
3
)x(z
2
z
3
) x
4
z
5
.
With both sequences of reductions we do get the same reduced word. It is not hard to show that
every word in a set X is equivalent to a unique reduced word.
25.3 Theorem Let F
X
be the set of reduced words in X. Then F
X
is a group with the operation
(x, y)x y.
Proof We have already observed that the empty word is an identity element for this operation.
Furthermore, if x F
X
then clearly x
1
F
X
and xx
1
= x
1
x = ; hence, every element of
F
X
has an inverse. It remains to verify that the operation is associative, that is, (xy)z = x(yz)
for any x, y, z F
X
. We need to distinguish three cases. Suppose rst, that the cancellations in
the products x y and y z do not overlap so that the reduced words x, y and z can be written
as
x = xy
1
k
y
1
1
, y = y
1
y
k
y y
m+1
y
l
, z = y
1
l
y
1
m+1
z
with y
i
XX
1
. Here x, y and z are subwords of x, y and z, respectively, such that ( y) 1
and the words x y and y z are reduced. Then both products (x y) z and x (y z) are equal to
the reduced word x y z.
Suppose now that the reduced words x, y and z have the form
x = xy
1
m
y
1
k
y
1
1
, y = y
1
y
k
y
k+1
y
m
y
m+1
y
l
, z = y
1
l
y
1
m+1
y
1
k+1
z,
where m > k and the words xy
m+1
y
l
and y
1
y
k
z are reduced. Then both products
(x y) z and x (y z) are equal to the reduced word xy
1
m
y
1
k+1
z.
Finally, consider the case where the reduced words x, y and z have the form
x = xy
1
k
y
1
1
, y = y
1
y
k
y
k+1
y
l
, z = y
1
l
y
1
k+1
z
and the words xy
k+1
y
l
and y
1
y
k
z are reduced. Then both products (x y) z and x (y z)
are equal to the reduced word obtained by calculating the product x z.
We are nowin a position to showthat the group F
X
possesses important freeness properties.
We start with the denition of free groups.
25.4 Denition Suppose that F is a group and that X is a subset of F. Then F is freely
generated by X if whenever f : XG is a map from X into a group G there exists a unique
group homomorphism
f
: F G such that
f
(x) = f(x), for all x X.
If F is freely generated then we say that F is a free group.
The denition is best understood in terms of the following commutative diagram:
X F x x
G f(x) =
f
(x)
f !
f
i i
f
f
where i : X F is the inclusion map. We say that
f
is an extension of f.
Notice that the denition does not guarantee that a free group on a given set X actually exists;
however, we will show later that free groups are ubiquitous and that they are essentially unique.
For now we will content ourselves with showing that at least one free group does exist.
Free groups 61
25.5 Example The group F = Z of integers is freely generated by X = 1. To see this
suppose that G is any group and let f : X G be any map. Write f(1) = g and dene a
function
f
: ZG by
f
(n) = g
n
. Then
f
is a group homomorphism. By now this should
be almost second nature; if not see Lemma 9.1
This example motivates our rst result about free groups.
25.6 Lemma Suppose that F is freely generated by a nonempty set X. Then F is an innite
group.
Proof Let f : X Z be the map f(x) = 1, for all x. By assumption, f extends to a group
homomorphism
f
: F Z and im
f
is a subgroup of Z containing 1; so, im
f
= Z.
Therefore, by the rst isomorphism theorem, F/ ker
f

= Z. Hence, F is innite.
The next problem with Denition 25.4 is that even though we say that F is freely generated
by X the denition does not say that F is generated by X. We now show that F = X).
25.7 Lemma Suppose that F is freely generated by X. Then F is generated by X.
Proof Let f : X X) be the map f(x) = x, for all x X. Then f extends to a group
homomorphism
f
: F X) such that
f
(x) = x, for all x X. As X) is a subgroup of F
we can think of
f
as a map from F to F with image X). Hence, we have a map
f
: F F
such that
f
(x) = x, for all x X, and im
f
= X).
On the other hand, the identity map id
F
: F F is also a group homomorphism with the
property that id
F
(x) = x, for all x X. Therefore,
f
= id
F
because
f
is the unique such
group homomorphism. Hence, F = imid
F
= im
f
= X) as required.
The same idea also shows that free groups are determined by the size of their generating sets.
25.8 Proposition Suppose that F is freely generated by X, that G is freely generated by Y ,
where X and Y are nite sets. Then F

= G if and only if [X[ = [Y [.
Proof We prove only that if [X[ = [Y [ then F

= G. The converse will be proved in tutorials.
Let f : X Y be any bijection and let g : Y X be the inverse map. By assumption
there exists unique group homomorphisms
f
: F G and
g
: GF extending f and g,
respectively. Therefore, there are group homomorphisms
f

g
: F F and
g

f
: GG.
Moreover, if x X and y Y then

g
(x) = g(f(x)) = x and
g

f
(y) = f(g(y)) = y
since f and g are mutually inverse functions. Now the identity maps id
F
: F F and
id
G
: G G are the unique group homomorphism with the property that id
F
(x) = x, for
all x X, and id
G
(y) = y, for all y Y . Therefore,
f

g
= id
F
and
g

f
= id
G
.
As
f

g
= id
F
, the map
f
is injective and because
g

f
= id
G
, the map
f
is also sur-
jective. Hence,
f
is a group isomorphism (with inverse
g
), and F

= G as required.
25.9 Theorem Let X be a set. The group F
X
is freely generated by X.
62 Algebra (Advanced)
Proof Suppose that G is a group and that f : XG is any map. Suppose that : F
X
G
is a group homomorphism such that (x) = f(x), for all x X. Now, if x X then
(x
1
) = (x)
1
= f(x)
1
; for convenience we set f(x
1
) = f(x)
1
. Let r = x
1
. . . x
n
be a
reduced word. Then r = x
1
x
2
. . . x
n
; so
(r) = (x
1
. . . x
n
) = (x
1
) . . . (x
n
) = f(x
1
) . . . f(x
n
);
by Proposition 13.3. Hence, is uniquely determined by f.
We are not quite done as we still need to check that the formula (r) = f(x
1
) . . . f(x
n
)
actually does dene a group homomorphism; that is, we need to check that respects the
multiplication in F
X
. Let r = x
1
. . . x
n
as above. As X generates F
X
, it is enough to show that
(rx) = (r)(x), for x X X
1
. Now,
r x =
_
x
1
. . . x
n
x, if x
n
,= x
1
,
x
1
. . . x
n1
, if x
n
= x
1
.
Therefore,
(r x) =
_
(x
1
) . . . (x
n
)(x), if x
n
,= x
1
,
(x
1
) . . . (x
n1
), if x
n
= x
1
However, if x
n
= x
1
then (x
n
)(x) = (x
1
)(x) = (x)
1
(x) = 1
G
. Hence, (r x) =
(r)(x) in all cases and the proof is complete.
Notice, in particular, that every element of the free group F
X
can be written as a reduced
word in X in a unique way.
By Proposition 25.8 we know that a free group is uniquely determined by the size of its
generating set (up to isomorphism). Theorem 25.9 tells us how to construct a free group on a
given set X. So we now know that there is a unique free group on any set X.
25.10 Examples a) Let X = x. Then the free group on X is
F
X
= 1, x, x
1
, x
2
, x
2
, . . . = x
n
[ n Z ,
where the operation is given by x
n
x
m
= x
n+m
, for n, m Z. Thus F
X

= Z (via the
map x
n
n). In fact, we already know this as it is a consequence of Example 25.5 and
Proposition 25.8.
b) In contract the group C
n

= 1, x, x
2
. . . , x
n1
, is not free because the set of not all words
in C
n
are distinct. For example, 1 = x
n
= x
2n
= . . . ; more generally, x
i
= x
i+kn
, for
k Z.
c) Let X = x, y. Then the free group on X contains the elements
1, x, x
1
, y, y
1
, x
2
, xy, xy
1
, x
2
, x
1
y, x
1
y
1
, y
2
, yx, yx
1
, y
2
, y
1
x, y
1
x
1
, x
3
, . . .
and where the group operation is given by concatenation followed by reduction. In
other words, the elements of F
X
are the monomials in the noncommuting variables
x, y, x
1
, y
1
.
Generators and relations 63
26. Generators and relations
In the last section we constructed a free group F
X
on a set X; we now use free groups to
describe arbitrary groups. The key observation is the following.
26.1 Lemma Let G be any group. Then G is a quotient of a free group.
Proof Choose a set of generators Y for G (for example, take Y = G). Let X be a set in
bijection with Y and x a bijection f : XY . Then, because F
X
is free, the map f extends
(uniquely) to a group homomorphism
f
: F
X
G. By construction im
f
= Y ) = G, so
G

= F
X
/ ker
f
. Thus, G is a quotient of F
X
as claimed.
The lemma motivates the denition below. First we need a Lemma.
26.2 Lemma Suppose that F is a group and R F. Let
^(R) =

NF
RN
N.
Then ^(R) is the smallest normal subgroup of F which contains R.
Proof See tutorials (and compare with Denition 10.2).
The group ^(R) is called the normal closure of R in F.
26.3 Denition Let X be a set and let R be a set of words in X. Then the group generated by X
subject to the relations R is the group
X [ R) = F
X
/^(R).
If G = X [ R) then X [ R) is a presentation of G.
26.4 Examples a) Z

= x [ )

= x, y [ y )

= x, y, z [ y, z )

= . . . .
b) More generally, if X is any set then F
X
= X [ ).
c) If n > 0 then C
n

= x [ x
n
). To see this note that if X = x then F
X
= x
k
[ k Z
is an abelian group. Therefore, every subgroup of F
X
is normal and
^(x
n
) = (x
n
)
k
[ k Z = x
kn
[ k Z .
Hence, x [ x
n
) = 1, y, . . . , y
n1
, where y = x +x
n
).
As all of the relations r R are equal to 1 in the group X [ R), it is also quite common to
write
X [ R) = X [ r = 1, for all r R) .
For example, C
n

= x [ x
n
= 1 ). More generally, we also write such expressions as
D
n
= r, s [ r
n
= 1, s
2
= 1, srs = r
1
)
where the last relation is to be interpreted as saying that srsr = 1.
64 Algebra (Advanced)
26.5 Proposition Let X be a set, R a set of words in X and set G = X [ R). Suppose that H
is a group and that f : XH is a function such that if r = x
1
. . . x
n
R then
f(x
1
) . . . f(x
n
) = 1
H
.
Then f extends to a unique homomorphism
f
: GH such that
f
(x) = f(x), for all x X.
Proof Let F
X
be the free group on X and write G = F
X
/^, where ^ = ^(R) is the normal
closure of R in F
X
. Then the function f : XH extends to a unique group homomorphism

f
: F
X
H such that
f
(x) = f(x), for all x XX
1
. Now if r R then
f
(r) = 1; so,
R ker
f
. Therefore, ker
f
is a normal subgroup of F
X
which contains R, so ^ ker
f
.
Therefore, we can dene a group homomorphism
f
: GH by
f
(g^) =
f
(g), for all g
G. Notice that this is well dened because if g^ = g

^ then g
1
g

^, so
f
(g) =
f
(g

)
since ^ ker
f
. Further, if x X then
f
(x^) =
f
(x) = f(x). Finally, as in the proof of
Theorem 25.9,
f
(x
1
. . . x
n
^) = f(x
1
) . . . f(x
n
), so
f
is uniquely determined.
So this is not quite as strong as the universal property which denes free groups: a map
f : X H extends if and only if f respects the relations in G. Note that if we know a
presentation for a group then we now have an easier way of checking whether is given map is a
group homomorphism: we look at what the map does to the generators and then check whether
or not the relations are preserved.
26.6 Proposition Let G be any group and let X = x
g
[ g G be a set in bijection with the
elements of G. Then
G

= X [ x
g
x
h
= x
gh
for all g, h G)
is a presentation of G. In particular, every group has at least one presentation.
Proof Let F
X
be the free group on X. As F
X
is a free group we can extend the map
f : F
X
G; x
g
g
to a group homomorphism
f
. If g, h G then

f
(x
g
x
h
x
1
gh
) = gh(gh)
1
= 1.
So (x
g
x
h
)x
1
gh
ker
f
and hence ^(R) ker
f
.
We claim that ^(R) = ker
f
. Let v = x
g
1
. . . x
g
k
be any reduced word in ker
f
, where
each x
g
i
X and k > 1. As k > 1 we may write v = x
g
1
x
g
2
u for some reduced word u. Then
x
g
1
g
2
(x
g
1
x
g
2
)
1
^(R), so u ker
f
. As (u) < (v) the claim now follows by induction.
We conclude these notes by giving two examples.
26.7 Proposition Let D
n
be the dihedral group of order 2n. Then
D
n

= x, y [ x
n
= 1, y
2
= 1, yxy = x
1
) .
Generators and relations 65
Proof Let G = x, y [ x
n
= 1, y
2
= 1, yxy = x
1
). We already know that these relations
are satised by elements in D
n
; so by Proposition 26.5 we have a surjective homomorphism
: G D
n
such that x r and y s. Therefore, D
n
is a quotient of G; consequently
[G[ [D
n
[ = 2n. To showthat D
n

= Git is therefore enough to showthat [G[ 2n. However,
it is easy to derive from the relations in G that every element of G is contained in the set
1, x, . . . , x
n1
, y, yx, . . . , yx
n1
,
thus completing the proof.
26.8 Proposition Let Sym(n) be the symmetric group of degree n. Then
Sym(n)

= s
1
, . . . , s
n1
[ s
2
i
= 1, (s
i
s
i+1
)
3
= 1, (s
i
s
j
)
2
= 1 if [i j[>1 ) .
Proof Again, let G be the group with this presentation.
We know from tutorials that the transpositions (1, 2), (2, 3), . . . , (n 1, n) generate Sym(n)
and it is easy to see that these elements satisfy the relations above. Hence, there is a surjective
group homomorphism : G Sym(n) with (s
i
) = (i, i + 1), for i = 1, . . . , n 1 by
Proposition 26.5. In particular, [G[ n!.
We show that [G[ = n! by induction on n. When n = 1 then G = 1 = Sym(1); so the
claim is true in this case. Suppose then that n > 1. Let H be the subgroup of G generated
by s
1
, . . . , s
n2
. Then H is a quotient of Sym(n 1), so [H[ (n 1)! note that H is not
necessarily isomorphic to Sym(n 1) because the additional relations in G might lead to extra
relations in H. Therefore, [H[ (n 1)!.
We claim that 1, s
n1
, s
n1
s
n2
, . . . , s
n1
. . . s
1
is a complete set of right coset representa-
tives for H in G; that is, that
() G = H

Hs
n1

Hs
n1
s
n2

Hs
n1
. . . s
1
.
In fact, it is enough to prove that G is contained in the union of these cosets because this would
imply that [G[ n[H[ n(n 1)! = n!; however, by the rst paragraph [G[ n!, so we have
equality throughout and [G[ = n!.
To prove our claim it is enough to see that () is closed under right multiplication by elements
of G; indeed, since s
1
, . . . , s
n1
generate G it is enough to check that () is closed under right
multiplication by s
1
, . . . , s
n1
. Multiplying on the right by s
n1
interchanges the rst two
cosets. For the remaining cosets suppose that1 j n2; then, using the relation s
a
s
b
= s
b
s
a
when [a b[ > 1, we have
Hs
n1
. . . s
j
s
n1
= Hs
n1
. . . s
j+1
s
n1
s
j
= = Hs
n1
s
n2
s
n1
s
n3
. . . s
j
= Hs
n2
s
n1
s
n2
s
n3
. . . s
j
= Hs
n1
s
n2
s
n3
. . . s
j
,
where the second last line uses the relation s
n2
s
n1
s
n2
= s
n1
s
n2
s
n1
, and the last one the
fact that s
n2
H. Hence, () is closed under right multiplication by s
n1
. The proof for the
remaining generators s
i
is similar and we leave it as an exercise.

You might also like