CHAPTER 1

Group Actions and Sylows Theorem

1.1. Group Actions
Definition 1.1.1. Let G be a group and T a set. An action of G on T is a
function : GT T such that, denoting (g, t) by g t,
(1) For all g, h G, t T, g (h t) = (gh) t.
(2) For all t T, e
G
t = t.
Given such an action we may dene maps
g
: T T for all g G by

g
(t) = g t. Denote by T(T) the set of bijections (or permutations) from T to T.
This of course is a group under composition of functions.
Theorem 1.1.2. Let be an action of a group G on a set T. For g G dene

g
: T T by
g
(t) = g t. Then
(1)
g
T(T) for all g G
(2) the map : G T(T) given by (g) =
g
is a homomorphism of groups.
Conversely, given any group homomorphism : G T(T), the map : GT T
given by (g, t) = (g)(t) is a group action.
Proof. First observe that for any g, h G and t T

g

h
(t) =
g
(
h
(t)) = g (h t)) = (gh) t =
gh
(t)
so (
g

h
) =
gh
. This proves that
g

g
1 = I
T
=
g
1
g
. Hence
g
T(T).
Obviously the same result now implies that is a homomorphism. The proof of
the converse is similar.
Example 1.1.3. The classic example of a group action is the action of a matrix
group on the corresponding vector space. For instance, for any eld F, take G =
GL
n
(F) and T = F
n
. It is clear that the usual matrix multiplication satises the
axioms above.
Example 1.1.4. The group operation denes an action of G on itself by left
multiplication. Let
g
T(G) be the associated permutation of G given by
g
(h) =
gh and let : G T(G) be the map (g) =
g
. The homomorphism is called
the left regular representation of G.
Example 1.1.5. A group can also act on itself by conjugation: g h = ghg
1
.
Theorem 1.1.6 (Cayleys theorem). Any nite group is isomorphic to a sub-
group of S
n
.
Proof. Let G be a nite group. Consider the left regular representation
: G T(G). If g ker , then
g
is the trivial permutation; that is gh =
g
(h) =
h for all h G. But this implies that g = e by cancellation. Thus ker = e
1
1.2. THE COUNTING FORMULA 2
and is injective. By the First Isomorphism theorem, G is isomorphic to its image
which is a subgroup of T(G). Since T(G)

= S
|G|
, the theorem follows.
1.2. The Counting Formula
Definition 1.2.1. Let G be a group acting on a set T. Let t T.
(1) The stabilizer of t in G is dened to be Stab
G
(t) = g G [ g t = t
(2) The orbit of t under G is dened to be O(t) = g t [ g G.
We leave it as an exercise to the reader to verify that Stab
G
(Y ) is a subgroup
of G. For instance, let T be the real plane and G the group of rigid motions of the
plane. Let Y be a gure in the plane (such as a hexagon or the letter H). Then
Stab
G
(Y ) is the usual symmetry group of the gure Y . Thus, we see again how the
theory of groups captures and generalizes the notion of symmetry. One can think
of the stabilizer Stab
G
(Y ) of the subset Y as an abstract version of the concept of
symmetry of an object.
Theorem 1.2.2 (The Counting Formula). Let G be a group acting on a set T.
Let t T.
(1) Stab
G
(t) is a subgroup of G
(2) If G is a nite group, then [G[ = [Stab
G
(t)[[O(t)[
Proof. Part (1) has already been discussed.
(2) Let H = Stab
G
(t). Then by Lagranges Theorem, [G[ = [H[[G : H].
Hence it suces to establish a bijection between G/H and O(t). Dene a function
: G/H O(t) by (gH) = gt. We claim that is a bijection. First we need
to check that is well-dened. Suppose that gH = g

H. Since g

gH, there
exists an h H such that g

= gh. But then g

t = (gh) t = g (h t) = g t, as
required. If t

O(t), then there exists a g G such that t

= g t. In this case
t

= g t = (gH) so is surjective. Finally,

(g) = (g

) = g t = g

t = g
1
g

t = t = g
1
g

H = gH = g

H
so is also one-to-one.
Notice that the G-orbits inside T form a partition of T (that is, T is the
disjoint union of the orbits). Recall that in this case the orbits must arise from
an equivalence relation on T. We can recreate the partition of T by dening this
relation.
Let G be a group acting on a set T. Dene a relation on T by s t if and
only if there exists a g G such that t = gs. It is a routine exercise to verify that
this denes an equivalence relation. The equivalence class of t T is clearly just
the orbit O(t).
Example 1.2.3. Consider the group D
8
of symmetries of the octagon. Each
element of D
8
induces a permutation of the set V of vertices of the octagon. Let
v be one such vertex. The stabilizer of v is the set of all symmetries that x v.
clearly this is just the identity and the reection through v and the opposite vertex.
On the other hand the orbit O(v) is just the set of dierent vertices to which v
can be sent by a symmetry. Since the octagon is regular, this is just the set of all
8 vertices. Thus [D
8
[ = 16. Naturally the same argument shows that the set of
symmetries D
n
of a regular n-gon has order 2n.
1.3. ACTION OF A GROUP ON ITSELF - THE CLASS EQUATION 3
Example 1.2.4. More interesting examples involve the symmetries of the reg-
ular solids. Consider for example the dodecahedron and let D be its group of
symmetries. Recall that the dodecahedron has 20 vertices, 30 edges and 12 pentag-
onal faces. Three faces meet at each vertex. Again x a vertex v. The stabilizer of
v is he set of symmetries that x v. These consist of the cyclic group of order three
of rotations and the three reections through the edges meeting at the vertex. So
Stab
D
(v)

= S
3
and has order 6. On the other hand the orbit of v is just the set
of 20 vertices. So [D[ = 20 6 = 120. The reader will probably easily agree that
this is an easier method of nding the order of D than trying to identify all 120
symmetries.
Another way to approach the problem is to notice that the group D acts on
the set of faces. If we x a particular face f, then Stab
D
(f)

= D
5
, the group of
symmetries of the pentagon, which has order 10. Since the orbit of f has size 12
we again get 120 for the order of the group.
Example 1.2.5. We can also derive the usual formula for combinations using
the counting formula. Recall that the number of ways of choosing a set of k elements
from a set of n elements is given by the binomial coecient
_
n
k
_
=
n!
k!(n k)!
We can derive this formula in the following way. Let T = 1, 2, . . . , n and let T(T)
be the power set of T - the set of all subsets. Then clearly S
n
acts on T(T) in the
natural way and the set of subsets of size k is precisely the orbit of the element
1, 2, . . . , k T(T). An element of S
n
will be in the stabilizer if it separately
permutes the sets 1, 2, . . . , k and k, k + 1, . . . , n. Thus
Stab
S
n
(1, 2, . . . , k) = /(1, 2, . . . , k) /(k, k + 1, . . . , n)

= S
k
S
nk
hence the counting formula states that
n! = [S
k
S
nk
[[O(1, 2, . . . , k)[ = k!(n k)!
_
n
k
_
which yields the formula above.
1.3. Action of a group on itself - the class equation
Now we study in more detail the action of a group on itself via conjugation:
g h = ghg
1
. In this case the stabilizer of an element h is the set:
C
G
(h) = g G [ ghg
1
= h
= g G [ gh = hg
called the centralizer of h; it is the set of elements of G that commute with h. Note
that C
G
(h) contains the center Z(G) and the subgroup h), so that if h ,= e then
C
G
(h) is a non-trivial subgroup. Note also that C
G
(h) = G if and only if h Z(G).
The orbit of h is called the conjugacy class of h and is the set of elements of G
which are conjugate to h:
((h) = ghg
1
[ g G
Interpreting the counting formula in this context yields:
1.3. ACTION OF A GROUP ON ITSELF - THE CLASS EQUATION 4
Corollary 1.3.1. Let G be a group and h G. Then
[((g)[ = [G[/[C
G
(h)[ = [G: C
G
(h)]
In particular, the cardinality of a conjugacy class must divide the order of the
group. Of course, the conjugacy class ((h) must contain h. We say the conjugacy
class is trivial if ((h) = h.
Theorem 1.3.2 (The class equation). Let G be a nite group and let g
1
, . . . , g
t
be representatives of the non-trivial conjugacy classes of G. Then
[G[ = [Z(G)[ +
t

i=1
[G: C
G
(g
i
)].
Proof. Since being conjugate to is an equivalence relation, the conjugacy
classes of G partition G. For elements in the center, the conjugacy classes are all
trivial. Thus these can be lumped together to yield the partition
G = Z(G)
t
_
i=1
((g
i
).
Hence
[G[ = [Z(G)[ +
t

i=1
[((g
i
)[
and the theorem follows from the corollary above.
The reader should pause to take in the signicance of this theorem. the sum-
mands on the right all divide [G[ and at most one (the rst) can be equal to 1.
Example 1.3.3. Consider the case when G = S
3
. The conjugacy classes are
e, (12), (13), (23) and (123), (132), so the class equation is 6 = 1 + 3 + 2.
Suppose that [G[ = 9, then the class equation could only be 9 = 3+3+3 or 9 = 9
(the last case of course is the class equation of an Abelian group). In particular, it
is not possible to have a class equation that begins 1 +. . . , so the center of G must
be non-trivial. This is a general phenomenon which occurs whenever the order of
G is a power of a prime.
Corollary 1.3.4. Let p be a prime and let G be a nite group of order p
t
for
some positive integer t. Then Z(G) ,= e.
Proof. All the summands of the form[((g
i
)[ on the right hand side are divisors
of p
n
and are not equal to 1. Therefore they are divisible by p. Since the left hand
side is divisible by p, the remaining summand, [Z(G)[ must also be divisible by p.
Hence Z(G) ,= e.
Corollary 1.3.5. Let p be a prime and let G be a nite group of order p
2
.
Then G is Abelian.
Proof. Exercise.
Note that the example of S
3
shows that a group of order pq with p ,= q does
not need to be Abelian.
1.4. SYLOWS THEOREM 5
1.4. Sylows Theorem
Sylows theorem represents one of the high points of elementary group theory,
combining simple ideas to prove a complex and surprising result. The theorem
consists of three parts, the rst of which is a partial converse to Lagranges Theorem.
Recall that Lagranges Theorem states that if G is a nite group of order n and if
H is a subgroup of order k, then k[n. We turn this around to ask whether for any
k dividing n whether there exists a subgroup of order k. In general the answer to
this question is No. Sylows Theorem answers this question when k is a largest
possible power of a prime.
Let G be a group of order n and let p be a prime divisor of n. Then we can
write n = p
e
m where gcd(p, m) = 1. In its simplest from Sylows Theorem has
three parts:
(1) There exists subgroups of G of order p
e
.
(2) All such subgroups are conjugate.
(3) The number of such subgroups divides m and is congruent to 1 modulo p.
To prove Sylows Theorem we shall make use of some further group actions on
sets connected to G. Let S be the set of subgroups of G. We know that if H is
a subgroup, then so is any conjugate gHg
1
. Hence G acts on S by conjugation.
That is, we may dene an action of G on S by:
for any g G, g H = gHg
1
Then the stabilizer of a subgroup under this action is the normalizer of the subgroup,
^
G
(H) = g G [ gHg
1
= H
The orbit is just the set of subgroups conjugate to H, so part (2) above can be
restated as saying the subgroups of order p
e
from a single orbit under this action.
Definition 1.4.1. Let G be a nite group and let p be a prime dividing [G[.
Suppose that p
e
is the highest power of p dividing n. A subgroup of G of order p
e
is called a Sylow p-subgroup of G.
Lemma 1.4.2. Let G be a nite Abelian group of order n and let p be a prime
dividing n. Then G contains a subgroup of order p.
Proof. We work by induction on n = [G[. The case where n = 2 is trivial.
Now suppose that the result is true for all Abelian groups of order less than n.
Clearly it suces to nd an element of G of order p. We may certainly pick a
non-trivial element g G. Let o(g) = m. If p [ m, then we may write m = pm

and o(g
m

) = p.
On the other hand suppose that p m. Let N = g). Then N is a (normal)
subgroup of order m and the quotient group G/N has order n/m, which is smaller
than n but still divisible by p (since p m). By induction G/N contains an element
of order p, say hN. This means that h
p
N = (hN)
p
= N and hence that h
p
N.
Since h , N, this implies that h
p
) is strictly contained in h) and hence that
o(h
p
) < o(h). But we know that
o(h
p
) =
o(h)
gcd(p, o(h))
This implies that gcd(p, o(h)) ,= 1 and hence, since p is prime that p[o(h). Returning
to the argument above, this yields an element in G of order p.
1.4. SYLOWS THEOREM 6
Theorem 1.4.3. Let G be a nite group of order n and let p be a prime divisor
of n. Then Sylow p-subgroups exist.
Proof. We prove the result by complete induction on n = [G[. Clearly the
result is true when n = 2. Assume that the result is true for all groups of order
k < n.
Let p be a prime dividing n. Suppose rst that p Z(G). Looking at the class
equation we see that p cannot divide all the remaining terms on the right. So there
must exist a g G such that [G: C
G
(g)] is not divisible by p. But in this case
[C
G
(g)[ = p
e
m

where gcd(p, m

) = 1 and m

< m. By induction C
G
(g) has a
subgroup H of order p
e
and H is clearly a Sylow p-subgroup of G.
Now assume that p[Z(G). By the lemma, Z(G) contains a subgroup N of
order p. Since N is contained in the center of G it is a normal subgroup of G. Now
[G/N[ = p
e1
m, so by induction G/N contains a Sylow p-subgroup, K, which has
order p
e1
. Let : G G/N be the natural surjection and let H =
1
(K). Then
K

= H/N by the rst isomorphism theorem, so [H[ = p.p
e1
= p
e
. Hence H is a
Sylow p-subgroup of G.
The next lemma touches on one of the crucial properties of Sylow subgroups:
no element of one Sylow p-subgroup can normalize non-trivially another one. That
is if P and P

are two Sylow p-subgroups and x P satises xP

x
1
= P

, then
x P

.
Lemma 1.4.4. Let P and P

be two Sylow p-subgroups. Then P ^

G
(P

) =
P P

. In particular, P ^
G
(P

) if and only if P = P

.
Proof. Let N = ^
G
(P

) and let Q = P N. Then applying the second

isomorphism theorem to N and the subgroups Q and P

, yields
QP

=
Q
Q P

This implies that the order of QP

/P

is of the form p
f
for some non-negative
integer f and that [QP

[ = [QP

/P

[[P

[ = p
e+f
, which contradicts Lagranges
Theorem unless f = 0. Hence Q P

= Q. So P N P

which implies that

P N = P P

.
Theorem 1.4.5. Let G be a nite group of order n and let p be a prime dividing
n.
(1) All Sylow p-subgroups are conjugate to each other.
(2) Let n
p
be the number of Sylow p-subgroups of G. Then
n
p
[m and n
p
1 (mod p).
Proof. Let P be a Sylow p-subgroup. Let T = gPg
1
[ g G be the
set of conjugates of P; lets denote these conjugates by P = P
1
, . . . , P
s
, so that
T = P
1
, . . . , P
s
. Then P itself acts on T by conjugation. An orbit will be a
singleton set P
i
if and only if P ^
G
(P
i
). But this can only happen if P = P
i
by the lemma. Hence P is the only trivial orbit; all the other orbits must have
more than one element and hence, because [P[ = p
e
, the size of these orbits is
divisible by p. This proves that [T[, the number of conjugates of P, is congruent
to 1 modulo p.
Now let P

be any other Sylow p-subgroup. Again we let P

act on T and
consider the orbit decomposition. The size of the orbits must be a power of p.
1.4. SYLOWS THEOREM 7
Thus at least one orbit must be trivial; otherwise we would conclude that p [
T, contradicting the conclusion of the previous paragraph. Suppose this orbit is
P
i
. Then P

^
G
(P
i
). By the lemma, this implies that P

= P
i
. Hence
every Sylow p-subgroup is conjugate to P, proving (1). Thus n
p
is equal to the
number of conjugates of P; hence n
p
1 (mod p). Finally, applying the Counting
Theorem to the action of G on subgroups, we see that n
p
= [G: ^
G
(P)]. But
[G: ^
G
(P)][^
G
(P): P] = [G: P] = m, so n
p
[m.
Example 1.4.6. As an easy illustrative example, consider the group D
5
of
symmetries of the regular pentagon. Recall that D
5
has order 10 = 2.5 and its
elements are the identity, four non-trivial rotations and ve reections. The theorem
states that the number of Sylow 5-subgroups must divide 2 and be congruent to
1 modulo 5. Thus there is a single normal subgroup of order 5; of course this
is the group consisting of the identity and the four non-trivial rotations. On the
other hand the theorem states that in a group of order 10 the number of Sylow
2-subgroups must divide 5 and be congruent to 1 modulo 2. Hence it must be 1
or 5. Clearly in the case of D
5
, the answer is 5 and these groups are the 5 cyclic
groups of order 2 generated by the reections.
As a rst application we see that most groups of order pq are Abelian (of
course S
3
is an example that shows that this is not always true).
Theorem 1.4.7. Let G be a group of order pq where p and q are primes and
p < q. If p does not divide q 1, then G is Abelian.
Proof. Consider the number n
q
of Sylow q-subgroups. Then n
q
[p and n
q
1
(mod q). hence n
q
= 1 and there is only one Sylow q-subgroup, say H, which must
then be normal.
Now consider the number n
p
of Sylow p-subgroups. We know by Sylows The-
orem that n
p
[q and n
p
1 (mod p). If n
p
= q, this would imply that p[(q 1),
which is not possible. Hence n
p
= 1 also. Let the unique Sylow p-subgroup be
K. By Lagranges Theorem, H K = e. Now HK must be a subgroup that is
properly bigger than K, so its order is greater than q and by Lagranges Theorem
divides pq. Hence HK = G. Thus by Lemma ??, G

= H K

= C
q
C
p
and in
particular, G is Abelian.
If there is only Sylow p-subgroup for some p, then we know that this subgroup
must be normal. Thus the Sylow theorems provide us with a method of proving
that a group has a proper non-trivial normal subgroup. This turns out to be an
extremely important question. Groups that do not have proper non-trivial normal
subgroups are called simple groups and form the fundamental building blocks of
nite group theory as we shall see in the next chapter.
Definition 1.4.8. A group is said to be simple if it has no non-trivial proper
normal subgroups.
Cyclic groups of prime order are simple because they have no non-trivial proper
subgroups at all. Conversely, an Abelian simple group must be cyclic of prime order
(why?). At this stage it is not at all obvious whether it is possible for a groups
to have non-trivial proper subgroups yet still be simple (because none of them are
normal). We shall answer this question in the next section. Our next result says
that if there are any such groups, they have to have at least 60 elements. Its proof
1.5. THE ALTERNATING GROUPS 8
is one of the most satisfying exercises in elementary group theory, so I wont spoil
the fun by providing the proof.
Theorem 1.4.9. There are no non-Abelian simple groups of order less than 60.
Proof. Exercise.
1.5. The Alternating Groups
Continuing with our theme of group actions, we consider a canonical action of
the symmetric group S
n
on the vector space R
n
. this allows us easily to identify
the alternating group, the subgroups of S
n
consisting of permutations that can
be represented as a product of an even number of transpositions. the alternating
groups are among the most important families of nite groups and their structure
plays a key role in explaining the non-existence of a formula for the solutions of a
quintic equations.
Proposition 1.5.1. Let S
n
and let

i
e
i
R
n
, where
i
R. Dene

: R
n
R
n
by

i
e
i
_
=

i
e
(i)
Then

is an invertible linear transformation and

. Dene : S
n

GL
n
(R) by () =

. Then is an injective homomorphism.

Proof. We leave it to the reader to verify that

is a linear transformation.
Now

i
e
i
_
=

i
e

(i)
_
=

i
e

(i)
=

i
e
i
_
Since
e
= I
R
n, the above proves that

is invertible. It also proves that is a

homomorphism.
Let : S
n
GL
n
(R) be the map () =

. Then denes an action of S

n
on R
n
. Note that since

(e
i
) = e
(i)
,

is the matrix with a 1 in the ((i), i)-

th position and zeros everywhere else. These matrices are known as permutation
matrices.
Example 1.5.2. Consider explicitly the case when n = 3. Then
((12)) =
_
_
0 1 0
1 0 0
0 0 1
_
_
, ((13)) =
_
_
0 0 1
0 1 0
1 0 0
_
_
, ((23)) =
_
_
1 0 0
0 0 1
0 1 0
_
_
and
((123)) =
_
_
0 0 1
1 0 0
0 1 0
_
_
, ((132)) =
_
_
0 1 0
0 0 1
1 0 0
_
_
.
Notice that the determinant of any permutation matrix is 1 since there is
exactly one term in the usual expansion of the determinant that is non-zero for
such matrices and this term is 1. The composition sgn = det : S
n
1 is
a homomorphism and is known as the sign representation.
1.5. THE ALTERNATING GROUPS 9
Definition 1.5.3. The kernel of the map sgn is called the alternating group
and is denoted by A
n
.
Note that A
n
must have n!/2 elements by the rst isomorphism theorem since
S
n
has n! elements and S
n
/A
n

= C
2
.
Example 1.5.4. The alternating group A
4
is the subgroup consisting of all
three cycles and products of disjoint transpositions:
e, (123), (132), (134), (143), (124), (142), (234), (243), (12)(34), (13)(24), (14)(23)
Sylows theorem predicts that the number of Sylow 3-subgroups must be 1 or 4;
clearly in this case the answer is 4, the eight 3-cycles splitting into 4 mutually
inverse pairs. On the other hand the number of Sylow 2-subgroups should be either
1 or 3; clearly the answer here is 1. The group V = e, (12)(34), (13)(24), (14)(23)
is the unique Sylow 2-subgroup.
We now look more closely at the group A
5
and prove that it is simple. Up to
this point we know that the cyclic groups of prime order are simple but we know
of no other examples. We also know that no non-Abelian group of order less than
60 can be simple. Thus A
5
is the smallest non-Abelian simple group. Its simplicity
is intimately connected with the Abel-Runi theorem that there is no algebraic
formula for the roots of a quintic polynomial.
The group A
5
has 60 elements, too many to write out explicitly. One can
however look at the number of elements of dierent types. In addition to the
identity element we see, using some straightforward counting arguments, that we
have:
20 3-cycles such as (123).
24 5-cycles such as (12345).
15 products of two disjoint 2-cycles such as (12)(34).
Now the orders of elements of A
5
that are possible from Lagranges Theorem
are the divisors of 60, namely 1, 2, 3, 4, 5, 6, 10, 12, 15, 20 and 30. But only 1, 2, 3
and 5 actually occur. This is an important illustration of the fact that k divides [G[
does not necessarily imply that there exist an element of order k.
Lets look at the Sylow subgroups. Note that 60 = 2
2
.3.5. The Sylow 3-
subgroups have order 3, so they must be cyclic. The number of them should be
congruent to 1 modulo 3 and must divide 20, so the options are 1, 4 and 10. Since
each of the 3-cycles above generates a cyclic subgroup of order 3 which contains
it and one other 3-cycle, there must be 10 such Sylow 3-subgroups. Similarly the
Sylow 5-subgroups are cyclic of order 5, and the number of them must divide 12 and
be congruent to 1 module 5. So there must be six Sylow 5-subgroups. Finally the
Sylow 2-subgroups are of order 4 and the number of them is congruent to 1 modulo
4 and divides 15. The possibilities are 1 or 5. But there are 5 subgroups similar to
the Klein group e, (12)(34), (13)(24), (14)(23). so the answer is obviously ve.
Now lets review the class equation for A
5
. Recall that the class equation comes
from the partition of G into conjugacy classes. The equation itself describes the
order of G as the sum of (a) the order of the center, which is the union of the trivial
conjugacy classes; and (b) the orders of the non-trivial conjugacy classes.
First recall the following basic fact about conjugation inside S
n
.
Lemma 1.5.5. Let S
n
and let (a
1
a
2
. . . a
t
) be a t-cycle. Then
(a
1
a
2
. . . a
t
)
1
= ((a
1
)(a
2
) . . . (a
t
))
1.5. THE ALTERNATING GROUPS 10
hence all t-cycles are conjugate in S
n
.
Thus the three cycles in S
5
form a single conjugacy class. Of course its not
completely clear that they are still conjugate inside the subgroup A
5
. Recall that
[G[ = [((g)[.[C
G
(h)[. Since [(
S
5
((123))[ = 20, the size of the centralizer of (123)
must be 6. But the centralizer contains (123) and (45) and together these generate
the group of order 6, so we must have:
C
S
5
((123)) = e, (123), (132), (45), (123)(45), (132)(45)
But then
C
A
5
((123)) = C
S
5
((123)) A
5
= e, (123), (132)
So the order of the conjugacy class of (123) in A
5
must be 60/3 = 12 and therefore
all of the 3-cycles are still conjugate in A
5
.
A similar argument can be applied to the 5-cycles but with a dierent result.
the 24 5-cycles are conjugate in S
5
. therefore [C
S
5
((12345))[ = 120/24 = 5; hence
C
S
5
((12345)) = (12345)). Thus,
C
A
5
((12345)) = C
S
5
((12345)) A
5
= (12345))
Hence, the order of the conjugacy class of (12345) in A
5
must be 60/5 = 12. Thus
inside A
5
the 5-cycles fall into two distinct conjugacy classes of size 12.
A similar argument applied to the products of disjoint transpositions yields a
single conjugacy class of size 15. hence the class equation of A
5
reads:
60 = 1 + 12 + 12 + 15 + 20
Interestingly, the simplicity of A
5
is an easy consequence of this.
Theorem 1.5.6. The group A
5
is simple
Proof. The key observation is that any normal subgroup must be a union of
conjugacy classes; for the denition of normality can be interpreted as the statement
that a group is normal if and only if contains all conjugates of any of its elements.
Thus if K is a proper non-trivial normal subgroup, its order must be a subsum
of the right hand side of the class equation that includes the rst term. The only
possibilities less than 30 are:
13, 16, 21, 25, 28
and clearly none of these numbers divide 60. Hence no proper non-trivial normal
subgroup exists.
Theorem 1.5.7. The group A
n
is simple for n 5.
Proof. We prove the result by induction on n. The base case n = 5 was
proved in Theorem 1.5.6. So suppose that n 6. To simplify notation, let G = A
n
.
Consider the groups G
i
= Stab
G
(i). The elements of G
i
are all even permutations
of the set 1, 2, , . . . , i 1, i +1, . . . , n so G
i

= A
n1
. Assume that G is not simple
and let N be a proper, non-trivial normal subgroup of N. Then N G
i
G
i
. Since
by induction G
i
is simple, NG
i
= G
i
or e. Suppose that NG
i
= G
i
. Let j be
dierent from i. Pick k and l also distinct from i and j. Since (ij)(kl)(i) = j, we
know (from a homework exercise) that Stab
G
(i) = (ij)(kl)Stab
G
(j)(kl)(ij) (note
that we cant just use (ij) here because it does not belong to G.) Since N is normal,
it must therefore contain all the groups G
1
, . . . , G
n
. let G and write as a
product of transpositions,
= (a
1
a
2
)(a
3
a
4
) . . . (a
r1
a
r
)
1.6. FREE GROUPS AND RELATIONS 11
where the a
i
are not necessarily distinct. Note that the number of transposi-
tions must be even so we can split them into consecutive pairs beginning with
(a
1
a
2
)(a
3
a
4
). Each of these products must belong to some G
i
(since n 5), so the
product belongs to N. We have proved that G = N contradicting the assumption
that N was proper.
Hence N G
i
= e for all i. This means that if N, then for all i, (i) ,= i.
It also implies that if ,

N and (i) =

(i) for any i, then =

. Let N
and consider the decomposition of into disjoint cycles. We consider two cases.
(1) The decomposition involves at least one cycle of length at least three.
So we can write = (a
1
a
2
a
3
. . . )
0
. Choose such that (a
1
) = a
1
,
(a
2
) = a
2
and (a
3
) ,= a
3
. (Note that must involve at least three
numbers distinct from a
1
, a
2
, so we are using the fact that n 5 at this
point.) Then

=
1
= (a
1
a
2
(a
3
) . . . )
0

1
Hence

(a
1
) = (a
1
) and

,= , a contradiction.
(2) Otherwise the decomposition must involve only transpositions. Since all
n numbers must be involved and is even, we must have n 8 and
= (a
1
a
2
)(a
3
a
4
)(a
5
a
6
)(a
7
a
8
)
0
Take = (a
3
a
5
)(a
7
a
8
). Then,

=
1
= (a
1
a
2
)(a
5
a
4
)(a
3
a
6
)(a
7
a
8
)
0

1
So again,

(a
1
) = (a
1
) and

,= , a contradiction.
Thus we contradicted the choice of a non-trivial element of N. Hence N = e.
This contradicts the assumption that N is non-trivial. We have contradicted the
assumption that G is not simple. Hence G must be simple.
1.6. Free Groups and Relations
Let S be a set and dene W(S) to be the set of words in S; that is the set of
all nite ordered strings of the form
a
1
a
2
. . . a
n
where a
i
S. For instance if S = a, b, c then the following are examples of words:
abaac, ababa, aaa, bbbcbab
Note that we include in W(S) the empty word which we will denote by 1. We
can dene a binary operation on S by concatenation
(a
1
a
2
. . . a
n
).(b
1
. . . b
m
) = a
1
a
2
. . . a
n
b
1
. . . b
m
This operation is associative and has an identity operation (hence it makes W(S)
into a semi-group). We want to make W(S) into a group. In order to do this we
rst introduce elements that will serve as the inverses of the elements of S. Let

S be a copy of S whose elements consists of s [ s S and let S

= S

S. Let
W = W(S

). Inside W we want to dene some additional equality relations such

as
a a = 1, bc

bbc cabc = bc

bbabc
To do this we will dene an equivalence relation on W so that these statements
are true in W/ . We dene a reduction to be a pair of words of W (w, w

) where
1.6. FREE GROUPS AND RELATIONS 12
w

= a
1
a
2
. . . a
t
and w = a
1
a
2
. . . a
i
b

ba
i+1
. . . a
t
for some a
1
, . . . a
t
, b S. We say
that a word w can be reduced to another word w

if there is a series of reductions

(w, w
1
), (w
1
, w
2
), . . . (w
s
, w

). A reduced word is a word that cannot be reduced to

another word of shorter length. A word is reduced if and only if it contains no
subwords of the form a a.
Lemma 1.6.1. Every word has a unique reduced form.
Proof.
We dene two words to be equivalent if they have the same reduced form.
Proposition 1.6.2. If w w

and v v

, then wv w

.
Proof. Let the reduced form of w and w

be w
0
and that of v and v

be v
o
. By
a series of reductions we can reduce w to its reduced form w
0
. The same reductions
will reduce wv to w
0
v. Similarly we may reduce w
0
v to w
0
v
0
. Analogously we can
reduce w

to w
0
v

and then to w
0
v
0
. Hence the reduced form of both wv and w

is w
0
v
0
. Thus wv w

.
Proposition 1.6.3. The operation of concatenation induces a group structure
on F(S) = W(S

)/ .
Proof. By the previous proposition the operation factors through to an op-
eration on F(S). It remains associative and the empty word is till an identity.
However we now have that every element has an inverse.
The group F(S) is called thefree group on the set S. It has the following
important universal property.
Theorem 1.6.4. Let G be a group, let S be a set and let : S G be a
function. Then there is a group homomorphism

: F(S) G such that

[
S
= .
Proof. Extend to

S by dening ( s) = (s)
1
. Dene

: W(S

) G by

(a
1
. . . a
t
) = (a
1
) . . . (a
t
). If w

= a
1
a
2
. . . a
t
and w = a
1
a
2
. . . a
i
b

ba
i+1
. . . a
t
then

(w) =

(a
1
a
2
. . . a
i
b

ba
i+1
. . . a
t
)
= (a
1
) . . . (a
i
)(b)(b)
1
a
i+1
. . . (a
n
)
= (a
1
) . . . (a
i
)(a
i+1
) . . . (a
n
)
=

(w)
So

is invariant under reductions. Hence if w w

then

(w) =

(w

). Thus

induces a map on F(S) by

([w]) =

(w).
Let R F(S), and denote by R) the normal closure of R (the intersection of
all the normal subgroups containing R).
Definition 1.6.5. The group generated by a set S subject to the relations R
is dened to be
S [ R) = F(S)/ R)
For instance S
3
can be show to be isomorphic to

a, b [ a
2
, b
3
, abab
_
.
1.6. FREE GROUPS AND RELATIONS 13
Theorem 1.6.6. Let G be a group and let S G which satisfy the relations
R. Then there exists a homomorphism : S [ R) G such that (s) = s for all
s S.
Proof. let

: F(S) G be the homomorphism guaranteed by theorem 1.6.4.
Then R ker(

and hence R) ker(

) also. Thus

induces a map on
F(S)/ R) = S [ R) such that (s) = s.
CHAPTER 2
Composition series and solvable groups
2.1. Composition series
One of the fundamental theorems in number theory is the unique factorization
theorem which states that every number can be expressed in a unique way as a
product of primes. Interestingly an analogous result holds for groups. If N G,
then we can think of N and G/N as being factors of G. We can then try and
factorize N and G/N. If G is nite this process must stop and at this point we
have reduced down to simple groups, which play the role of primes in group theory.
Definition 2.1.1. A normal series for a group G is a nite sequence of sub-
groups, G
0
= e G
1
G
2
. . . G
t
= G such that G
i
G
i+1
. The length of the
series is dened to be t.
Definition 2.1.2. A composition series is a normal series in which all the
factor groups are simple
Example 2.1.3. Let G be S
4
. Recall the denition of the Klein 4-group V and
let C = e, (12)(34). Then
e C V A
4
S
4
is a composition series for S
4
. Note that the denition of normal series does not
require that the terms of the series be normal in the group G. In the case above V
and A
4
are normal in G but C is not.
The unique factorization theorem for integers has two parts. One is the exis-
tence part which states that every integer can be factored as a product of primes.
The second is the uniqueness part which states that there is only one way of fac-
toring a number into a product of primes. The analogous result for groups has two
similar parts: (1) composition series always exist for nite groups; and (2) the set
of composition factors is the same for any composition series.
Proposition 2.1.4. Any nite group has a composition series.
Proof. We use induction on n = [G[. The case n = 1 is trivial. Consider
the set of all proper normal subgroups of [G[ and pick one, say N, whose order
is as large as possible. By the fourth isomorphism theorem G/N is simple. Now
[N[ < n, so by induction, N has a composition series,
N
0
= e N
1
N
2
. . . N
s
= N
Then clearly N
0
N
1
N
2
. . . N
s
G is a composition series for G.
14
2.2. THE JORDAN HOLDER THEOREM 15
2.2. The Jordan Holder Theorem
Suppose that G is a nite Abelian group of order n and let G
0
= eG
1
G
2

. . .G
t
= G be a composition series for G. Then the composition factors must all be
Abelian simple groups; that is, they are isomorphic to a cyclic group of prime order,
say G
i
/G
i1

= C
p
i
. By Lagranges Theorem, n = p
1
. . . p
t
. Thus while there may
be many dierent composition series for G, the length of any composition series is
equal to t, the number of primes in the prime factorization of n; and the collection
of composition factors is the same in the sense that the types of composition factors
that occur and the number of times they occur is invariant. For instance, if G = )
is an Abelian group of order 60 then
e
12
)
6
)
2
) G
and
e
20
)
10
)
5
) G
are both composition series with composition factors C
5
, C
2
, C
3
, C
2
and C
3
, C
2
, C
2
, C
5
respectively. We can thus think of a composition series in some way as a prime
factorization for a nite group. The Jordan Holder Theorem is then the analog of
the unique factorization theorem for integers. Its states that composition series are
unique in the sense that they have the same length and that the set of composition
factors occurring, counted with multiplicity is the same for any such series.
Definition 2.2.1. Let G be a group and let G
0
= e G
1
G
2
. . . G
s
= G
and H
0
= e H
1
H
2
. . . H
t
= G be two normal series for G. We say that the
series are equivalent if s = t and there exists a bijection between the sets of factor
groups for which corresponding factors are isomorphic.
Theorem 2.2.2 (Jordan-Holder). Any two composition series of a nite group
are equivalent.
Proof. We shall prove the theorem by induction on the s length of the shorter
series. If s = 1, then G is simple, so the result is trivially true. Now assume that
G
0
= e G
1
G
2
. . . G
s
= G and H
0
= e H
1
H
2
. . . H
t
= G are two
composition series with s t. If G
s1
= H
t1
then the result from the induction
hypothesis applied to G
s1
. Otherwise G
s1
H
t1
is a normal subgroup of G strictly
bigger than G
s1
. Since G/G
s1
is simple, the fourth isomorphism theorem tells
us that G
s1
H
t1
= G. Therefore,
G
s1
G
s1
H
t1

=
G
s1
H
t1
H
t1
=
G
H
t1
and
H
t1
G
s1
H
t1

=
G
s1
H
t1
G
s1
=
G
G
s1
Now let
e K
1
K
2
. . . K
u
= G
s1
H
t1
be a composition series for G
s1
H
t1
. Then
e K
1
K
2
. . . K
u
G
s1
is a composition series for G
s1
, as is
e G
1
G
2
. . . G
s1
2.3. SOLVABLE GROUPS 16
Thus, by induction, u = s2 and the two series are equivalent. A similar argument
shows that
e K
1
K
2
. . . K
u
H
t1
is equivalent to
e H
1
H
2
. . . H
t1
and u = t 2. Hence s = t and by combining all the above information we see that
the two original series are equivalent.
2.3. Solvable groups
If Gis an Abelian group, its composition factors must also be Abelian. However,
the converse is far from true, as the examples of S
3
and S
4
illustrate. It is quite
possible for a non-Abelian group to have Abelian composition factors. These kind
of groups turn out to be extremely important.
Definition 2.3.1. A group G is said to be solvable if it has a normal series
G
0
= e G
1
G
2
. . . G
t
= G
for which the factors G
i
/G
i1
are Abelian.
This denition works for both nite and innite groups. For nite groups one
can equivalently dene a solvable group to be one whose composition factors are
all Abelian. The symmetric groups S
3
and S
4
are solvable but S
5
is not (since its
composition factors are A
5
and C
2
). More generally, the fact that there are no
non-Abelian simple groups of order less than 60 implies that all groups of order less
than 60 are solvable and that A
5
is the smallest group that is not solvable.
The following result about solvable groups is of fundamental importance.
Theorem 2.3.2. Let G be a group and H a normal subgroup. Then G is
solvable if and only if both H and G/H are solvable.
Proof. Suppose that G is solvable. Then there exists a normal series
G
0
= e G
1
G
2
. . . G
t
= G
for which the factors G
i
/G
i1
are Abelian. Let H
i
= G
i
H. Then for any k H
i1
and h H
i
, then hkh
1
G
i1
H = H
i1
, so the H
i
form a normal series for
H. Moreover, using the second isomorphism theorem,
H
i
H
i1
=
G
1
H
G
i1
H
=
G
i
H
G
i1
(G
i
H)

=
G
i1
(G
i
H)
G
i1

G
i
G
i1
Hence H
i
/H
i1
is Abelian, and H is solvable. Similarly, let K
i
= G
i
H/H. Then
K
0
= e K
1
K
2
. . . K
t
= G/H is a normal series for G/H. Using both the
second and third isomorphism theorems we see that
K
i
K
i1
=
G
i
H/H
G
i1
H/H

=
G
i
H
G
i1
H
=
G
i
(G
i1
H)
G
i1
H

=
G
i
G
i1
H G
i

=
G
i
/G
i1
G
i1
H G
i
/G
i1
Thus K
i
/K
i1
is a homomorphic image of G
i
/G
i1
and hence is Abelian. Thus
G/H is solvable.
The converse we leave as an exercise.
2.4. P-GROUPS AND NILPOTENT GROUPS 17
2.4. p-Groups and Nilpotent Groups
Definition 2.4.1. Let p be a prime. A p-group is a group of order p
r
for some
r.
Theorem 2.4.2. Any p-group is solvable.
Proof. Let G be a p-group. We know that G has a composition series. By
Lagranges Theorem, the composition factors of a p-group must all be p-groups
themselves. Thus the composition factors are simple p-groups. Since the center of
a p-group is a non-trivial normal subgroup by 1.3.4, a simple p-grop must be equal
to its center and hence abelian.
In fact p-groups have many other very special properties.
Theorem 2.4.3. Let G be a group of order p
r
where p is prime.
(1) If N is a normal subgroup of G, then N Z(G) ,= e.
(2) G contains a composition series consisting of normal subgroups. In par-
ticular G contains a normal subgroup of order p
s
for all s = 1, . . . , n.
(3) If H is a proper subgroup of G, then H < N
G
(H).
Proof. (1) Notice that N is closed under conjugation by elements of G. There-
fore we have an action of G on N by conjugation. Just as in the proof of 1.3.4, we
look at the orbits under this action and deduce that the number of trivial orbits
must be divisible by p. The trivial orbits are precisely the intersection of N with
Z(G).
(2) We prove that G contains a composition series consisting of normal sub-
groups by induction on r. If r = 1, the assertion is trivial. If r > 1, then G contains
a non-trivial center and hence an element of order p by 1.4.2. Thus, G contains
a normal subgroup of order p, say N. By induction, G/N contains a composition
series consisting of normal subgroups, say
N/N = N
1
/N N
2
/N . . . N
t
/N = G/N
By the fourth isomorphism theorem, the N
i
must be normal in G, so
e N = N
1
N
2
. . . N
t
= G
is a composition series for G consisting of normal subgroups of G. Since N
i
/N
i1

=
C
p
, we must have that [N
i
[ = p
i
, proving the second assertion.
(3) Again we use induction on the order of G, the base case being trivial.
Consider HZ(G). If HZ(G) = e, then the elements of the center are in N
G
(H)
but not in H. If HZ(G) ,= e, then H contains a normal subgroup M of order p.
Now N
G
(H)/M = N
G
(H/M) (exercise). By induction N
G
(H/M) strictly contains
H/M. By the fourth isomorphism theorem, N
G
(H) strictly contains H.
Groups that have composition series consisting of normal subgroups are called
supersolvable groups. However p-groups actually have an even stronger property
they have a normal series where each G
i
/G
i1
Z(G/G
i1
). Such groups are
called nilpotent. the standard deition is slightly dierent.
Definition 2.4.4. Let G be a group. we dene iteratively the upper central
series by
Z
0
(G) = e, Z
1
(G) = Z(G), Z
i
(G)/Z
i1
(G) = Z(G/G
i1
)
The group Z
i
(G is called the i-th center of the group G.
2.4. P-GROUPS AND NILPOTENT GROUPS 18
Definition 2.4.5. The commutator of two elements x, y in a group G is the
element
[x, y] = xyx
1
y
1
Lemma 2.4.6. The i-th center Z
i
(G) is the set of elements x G [ [x, y]
Z
i1
(G) for all y G.
Proof. Let x G. Set Z
i1
= Z
i1
(G). Then
x Z
i
xZ
i1
Z(G/Z
i1
)
xZ
i1
yZ
i1
= yZ
i1
xZ
i1
for all y G
xZ
i1
yZ
i1
[xZ
i1
]
1
y[Z
i1
]
1
for all y G
xyx
1
y
1
Z
i1
= Z
i1
for all y G
[x, y] Z
i1
for all y G

Definition 2.4.7. A group is called nilpotent if there exists an integer t such

that Z
t
(G) = G.
Theorem 2.4.8. All p-groups are nilpotent.
Proof. It suces to note that if Z
i
(G) ,= G, then Z
i
(G) Z
i+1
(G).
Recall that if K and N are normal subgroups of a group G such that H K =
e, then HK

= H K .
Lemma 2.4.9. Let G be a nite group and let p
1
, . . . p
r
be the prime divisors of
[G[. If G has a unique Sylow p
i
-subgroup P
i
, for each i, then G

= P
1
P
r
.
Proof. We prove by induction that P
1
. . . P
i

= P
1
P
i
. The case i = 1
is trivial. By induction P
1
. . . P
i

= P
1
P
i
, so [P
1
. . . P
i
[ = [P
1
[ . . . [P
i
[ and
p
i+1
[P
1
. . . P
i
[. Hence P
i+1
P
1
. . . P
i
= e. So
P
1
. . . P
i+1
= (P
1
. . . P
i
)P
i+1

= (P
1
. . . P
i
) P
i+1

= P
1
P
i+1
The result then follows.
Lemma 2.4.10. Let G be a nite nilpotent group. If H < G, then H < N
G
(H).
Proof. As for p-groups. Either H Z(G) in which case we can apply induc-
tion, or the elements of the center not in H are in N
G
(H).
Theorem 2.4.11. A nite group is nilpotent if and only if it is a product of
p-groups.
Proof. It suces to show from Lemma 2.4.9 that any Sylow p-subgroup is
normal. Let P be a Sylow p-subgroup and let N = N
G
(P). Since P N, it must
be the unique Sylow p-subgroup of N. Consider M = N
G
(N). Let m M. Then
mPm
1
mNm
1
= N and [mPm
1
[ = [P[ so mPm
1
is a Sylow p-subgroup of
N. Hence mPm
1
= P. Thus P M. Since N is the maximal subgroup in which
P is normal we must have M = N. By 2.4.10, if N were a proper subgroup of G,
it would have to be a proper subgroup of M. hence N = G and P G.
2.5. THE DERIVED SERIES 19
Definition 2.4.12. For two subgroups H and K of a group G, we dene
[H, K] = [h, k] [ h H, k K)
This group is called the commutator of the groups H and K.
Definition 2.4.13. The lower central series is dened by
G
0
= G, G
1
= [G, G], G
i
= [G, G
i1
]
Lemma 2.4.14. Let G be a group and let N M be normal subgroups. Then
M/N ZG/N if and only if N [G, M]. In particular G/N is abelian if and only
if N [G, G]
Proof.
M/N Z(G/N) xNyN = yNxN for all x G, y M
xyx
1
y
1
N = N for all x G, y M
[x, y] N for all x G, y M
[M, G] G

Theorem 2.4.15. The groups G

i
are all normal in G. A group G is nilpotent
if and only if G
t
= e for some t.
Proof. Exercise.
2.5. The Derived Series
We now return to solvable groups and give an alternative characterization of a
solvable group.
Definition 2.5.1. The derived series of a group G is dened by
G
(0)
= G, G
(1)
= [G, G], G
(i)
= [G
(i1)
, G
(i1)
]
Theorem 2.5.2. A group G is solvable if and only if G
t
= e for some positive
integer t.
Proof. Suppose that G
(t)
= e for some t. The commutator of a pair of
normal subgroups is again normal, so the derived series is a normal series for G:
G
(t)
G
(t1)
. . . G
(1)
G
Moreover G
(i1)
/G
(i)
= G
(i1)
/[G
(i1)
, G
(i1)
] is abelian by Lemma 2.4.14 so G is
solvable.
Conversely, suppose that G is solvable and let
N
0
= e N
1
N
2
. . . N
t
= G
be a normal series with abelian factors. Then by Lemma 2.4.14, [N
j
, N
j
] N
j1
for
all j = 1, . . . , t. In particular, G
(1)
= [N
t
, N
t
] N
t1
. We prove that G
(i)
N
ti
by induction, the base case having just been proved. Suppose that G
(i)
N
ti
.
Then
G
(i+1)
= [G
(i)
, G
(i)
] [N
ti
, N
ti
] N
ti1
= N
t(i+1)
Thus G
(t)
N
0
= e, as required.
2.6. AUTOMORPHISMS AND THE SEMI-DIRECT PRODUCT 20
2.6. Automorphisms and the Semi-Direct product
Definition 2.6.1. An automorphism of a group G is a group isomorphism
from G to G. The set of all automorphisms of G is a subgroup of T(G) called the
automorphism group and denoted by Aut(G).
We leave it to the reader to verify that Aut(G) is indeed a subgroup of T(G).
Theorem 2.6.2. Let g G. Denote by
g
: G G the conjugation map

g
(h) = ghg
1
. Then
g
Aut(G). Dene : G G by (g) =
g
. Then is a
homomorphism with kernel ker = Z(G).
Proof. Exercise for the reader.
Definition 2.6.3. The automorphisms
g
are known as inner automorphisms
and the image of is called the inner automorphism group. It is denoted Inn(G) =
=
g
[ g G.
Proposition 2.6.4. let Aut(G) and let g G. Then

1
=
(g)
.
Hence Inn(G) is a normal subgroup of Aut(G).
Proof. Exercise for the reader.
Definition 2.6.5. Let H and K be groups and let : K Aut(H) be a
homomorphism. Denote the induced action of k K on h H by kh = (K)(h).
Dene an operation on H K by
(h
1
, k
1
) (h
2
, k
2
) = (h
1
(k
1
h
1
), k
1
k
2
)
Theorem 2.6.6. The operation dened above endows the set H K with a
group structure.
Proof. Associativity: Let (h
1
, k
1
), (h
2
, k
2
), (h
3
, k
3
) H K. Then
(h
1
, k
1
)[(h
2
, k
2
)(h
3
, k
3
)] = (h
1
, k
1
)[(h
2
(k
2
h
3
), k
2
k
3
)]
= (h
1
(k
1
(h
2
(k
2
h
3
))), k
1
(k
2
k
3
))
= (h
1
(k
1
h
2
)(k
1
(k
2
h
3
)), (k
1
k
2
)k
3
)
= (h
1
(k
1
h
2
)((k
1
k
2
) h
3
), (k
1
k
2
)k
3
)
= (h
1
(k
1
h
2
), k
1
k
2
)(h
3
, k
3
)
= [(h
1
, k
1
)(h
2
, k
2
)](h
3
, k
3
)
It is easily seen that (e
H
, e
K
) is an identity element. Observe that
(k
1
h
1
, k
1
)(h, k) = ((k
1
h
1
)(k
1
h), k
1
k) = (k
1
e
H
, e
K
) = (e
h
, e
K
)
and that
(h, k)(k
1
h
1
, k
1
) = (h(k (k
1
h
1
)), kk
1
) = (h(e
K
h
1
), e
K
) = (e
H
, e
K
)
so (h, k)
1
= (k
1
h
1
, k
1
). Thus all the group axioms are satised for this
operation.
Definition 2.6.7. The group described in the above theorem is called the
semidirect product of H and K with respect to . It is denoted by H

K.
2.7. LINEAR GROUPS 21
Example 2.6.8. It is easy to see that S
3
is isomorphic to the semidirect product
of C
3

C
2
where is the (unique) isomorphism from C
2
to Aut(C
3
).
Theorem 2.6.9. Let G be a group and let H and K be subgroups. Suppose that
(1) H G
(2) HK = G
(3) H K = e
Then G

= H

K where : K Aut(H) is the map induced from conjugation on

H.
Proof. Dene f : H

K G by f(h, k) = hk. Then

f((h
1
, k
1
))f((h
2
, k
2
)) = h
1
k
1
h
2
k
2
= h
1
(k
1
h
2
k
1
1
)k
1
k
2
= h
1
(k
1
)(h
2
)k
1
k
2
= f((h
1
(k
1
)(h
2
), k
1
k
2
))
= f((h
1
, k
1
)(h
2
, k
2
))
So f is a homomorphism. It is clearly surjective by assumption (2). To verify
injectivity note that
f(h, k) = e = hk = e = h = k
1
H K = h = k = e
by part (3).
Example 2.6.10. Groups with two Sylow subgroups. Suppose [G[ = p
e
q
f
for e
and f distinct primes. Let P and Q be a Sylow p-subgroup and a Sylow q-subgroup
respectively. If P and Q are both normal, then G

= P Q (as we saw earlier in
the discussion of nilpotent groups). If P is normal but Q isnt then we are in the
situation of the Theorem 2.6.9 and G

= P Q. In particular this will always
happen when G is a non-abelian group of order pq where q < p.
Example 2.6.11. Dihedral groups. Consider the dihedral groups D
2n
, the
symmetries of the regular n-gon. let H be the subgroup of rotations through
2k/n and let K be the subgroup generated by a reection . If H, then

1
=
1
. Again the hypotheses of the theorem are satised and G

= H K.
Example 2.6.12. Ane transformations. let V be a vector space. An ane
transformation is function f : V V of the form f(v) = (v) + w where is
linear and w V . Such a transformation is invertible if and only if is. the
set of invertible ane transformations forms a group A. Let L be the subgroup
of invertible linear transformations and T the group of translations (v v + w).
Then T is a normal subgroup and A = T L.
2.7. Linear Groups
2.7.1. General and Special Linear Groups. Let F be a eld. The general
linear group is dened to be the set of n n matrices over F,
GL
n
(F) = A M
n
(F) [ det(A) ,= 0
It is easily seen to be a group under matrix multiplication. The special linear group
is the subgroup of matrices of determinant 1.
SL
n
(F) = A GL
n
(F) [ det(A) = 1
If F is a nite eld then these groups are both of course nite.
2.7. LINEAR GROUPS 22
Proposition 2.7.1. Let F be a eld of order q. Then
[GL
n
(F)[ = (q
n
1)(q
n
q) . . . (q
n
q
n1
)
Proof. A matrix is invertible if and only if its columns are independent.
We can form such a matrix by choosing successive columns so that they are not
contained in the vector subspace of F
n
spanned by the previous columns. Since
[F
n
[ = q
n
and an i1-dimensional subspace contains q
i1
vectors, there are q
n
q
i1
choices for the i-th column.
The special linear groups provide us with one of the simplest innite families of
simple nite groups of Lie type. The center of SL
n
(F) is just the scalar matrices
of determinant 1 and so is isomorphic to the subgroup of n-th roots of unity in F.
The quotient is called the projective special linear group,
PSL
n
(F) = SL
n
(F)/(Z(SL
n
(F))
If F is a nite eld then PSL
n
(F) is simple with two exceptions: the groups PSL
2
(F)
when [F[ = 2 or 3.
2.7.2. Orthogonal groups. The orthogonal group is the group
O
n
(F) = A GL
n
(F) [ A

A = I
It is easily veried that O
n
(F) is a subgroup of GL
n
(F). Note that since det A

=
det A, we must have (det A)
2
= 1. Moreover there exist orthogonal matrices of
determinant 1 (note that if characteristic of F is 2, then 1 = 1 so these statements
are true but vacuous). The determinant map det: O
n
(F) F

is a homomorphism
with image 1, 1 and kernel
SO
n
(F) = O
n
(F) SL
n
(F).
Hence by the rst isomorphism theorem, if the characteristic of F is not 2,
O
n
(F)/SO
n
(F)

= Z
2
.
2.7.3. Symplectic groups. Let J be the 2n 2n matrix:
J =
_
0 I
I 0
_
.
We dene the symplectic group Sp
2n
(R) by
Sp
2n
(R) = A GL
2n
(R) [ A

JA = J.
2.7.4. Lorentz group. Another interesting group arises as the symmetries of
space-time. Note that the denition of the symplectic group in terms of the matrix
J would dene a subgroup of GL
n
(R) for any matrix J. Dene
J =
_
_
_
_
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
_
_
_
_
and set
L = A GL
4
(R) [ A

JA = J
2.7. LINEAR GROUPS 23
2.7.5. Unitary groups. When working over the complex numbers, the natu-
ral bilinear form is now the Hermitian form x, y) = x

y where the bar denotes the

complex conjugate. the symmetries of Hermitian space are the unitary matrices
U
n
= A GL
n
(C) [

A

A = I
Again we can look at the determinant map det: U
n
R

. If A U
n
and det A = z,
then it follows from the multiplicativity of the determinant that [[z[[ = zz = 1.
Hence the image of the determinant is contained inside the unit circle S
1
of complex
numbers of modulus one. Again one can nd suitable matrices A of any such
determinant, so that the image of det is the whole of S
1
. We dene
SU
n
= U
n
SL
n
(C)
Thus the kernel of det is SU
n
and the rst isomorphism gives
U
n
/SU
n

= S
1
APPENDIX A
The Isomorphism Theorems
Theorem A.0.2 (The First Isomorphism Theorem). Let G be a group and let
: G H be a homomorphism. then G/ ker

= (G).
Proof. Let K = ker . Dene a map : G/K (G) by (gK) = (g).
We must rst verify that this map is well-dened: that is, if gK = g

K, then
(g) = (g

). But if gK = g

K, then g

= gk for some k K and so (g

) =
(gk) = (g)(k) = (g)e = (g). Thus is well-dened.
For any gK, g

K G/K, (gKg

K) = (gg

K) = (gg

) = (g)(g

) =
(gK)(g

K). So is a homomorphism.
Suppose that (gK) = e
H
, then (g) = e
H
and g K; hence gK = K = e
G/K
,
and so is injective.
Finally, let h (G). Then h = (g) for some g G. Hence h = (gK),
proving that is surjective. Thus is an isomorphism, as required.
Lemma A.0.3. Let G be a group, let K be a normal subgroup of G and let H
be a subgroup of G. Then
(1) HK = KH and this is a subgroup of G
(2) H K is a normal subgroup of H
Proof. Exercise
Theorem A.0.4 (The Second Isomorphism Theorem). Let G be a group, let
K be a normal subgroup of G and let H be a subgroup of G. Then HK/K

=
H/(H K).
Proof. Dene a map : H HK/K by (h) = hK. Then is clearly
a homomorphism because (hh

) = hh

K = hKh

K = (h)(h

). Now (h) =
e
HK/K
(h) = K hK = K h K. So ker = H K. Finally
is clearly surjective since an arbitrary element of HK/K is of the form hkK for
some h H, k K. Since kK = K, hkK = hK and we see that every element of
HK/K is of the form hK for some h H. The result then follows from the rst
isomorphism theorem.
Theorem A.0.5 (The Third Isomorphism Theorem). Let G be a group, let
H, K be normal subgroups of G with H K. Then K/H is a normal subgroup of
G/H and
(G/H)/(K/H)

= G/K.
Proof. Dene : G/H G/K by (gH) = gK. One shows that is a well-
dened homomorphism with ker = K/H. The result then follows again from the
rst isomorphism theorem. The details are left as an exercise for the reader.
24
A. THE ISOMORPHISM THEOREMS 25
Theorem A.0.6 (The Fourth Isomorphism Theorem). Let G be a group and let
N be a normal subgroup of G. Then there is a one-to-one correspondence between
subgroups of G containing N and subgroups of G/N given by K K/N. The
correspondence also is a one-to-one correspondence between normal subgroups of G
containing N and normal subgroups of G/N. The correspondence also preserves
inclusions and intersections.
Proof. Let K

be a subgroup of G/N and let K = g G [ gN K/N.

Clearly K N. We claim it is a subgroup. let g, g

K. Then gg

N =
(gN)(g

N) K

since K

is a subgroup of G/N. Similarly if g K, then

gN G/N. But (gN)
1
= g
1
N G/N so g
1
K. In the other direction, if K
is a subgroup of G containing N, then it is easily seen that K/N = kN [ k K
is as subgroup of G/N.
We leave it as an exercise to prove that the correspondence takes normal sub-
groups to normal subgroups and that it preserves inclusions and intersections.