B2bGroupTheoryHT11 Notes PDF

Group Theory
Notes to accompany the course B2b Group Theory given in by Prof. Michael Collins
Hilary Term 2011, Mathematical Institute, University of Oxford.
Jan E. Grabowski
Keble College, Oxford
jan.grabowski@maths.ox.ac.uk
Version of January 12, 2011
Contents
1 Introduction 3
References 5
2 Preliminaries 6
2.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Recollections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3 Some basic tools 16
3.1 Generating sets in groups and their subgroups . . . . . . . . . . . . . . 16
3.2 Presentations of groups . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3 Direct and semi-direct products. Extensions . . . . . . . . . . . . . . . 22
3.4 Composition series. The JordanHolder theorem . . . . . . . . . . . . 29
4 Abelian groups 34
4.1 Finite Abelian groups . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2 Finitely generated Abelian groups . . . . . . . . . . . . . . . . . . . . 38
5 Prime numbers in nite group theory 40
5.1 Cauchys theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.2 Finite p-groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.3 Sylows theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
6 Soluble groups 49
6.1 Polycyclic groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6.2 Soluble groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
7 Character theory 56
7.1 Group representations . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
7.2 Characters and character tables . . . . . . . . . . . . . . . . . . . . . . 61
7.3 An application: Burnsides p
theorem . . . . . . . . . . . . . . . . 71
8 Simple groups 75
8.1 Non-Abelian nite simple groups . . . . . . . . . . . . . . . . . . . . . 75
8.2 The simplicity of A
n
, n 5 . . . . . . . . . . . . . . . . . . . . . . . . 78
8.3 Projective special linear groups . . . . . . . . . . . . . . . . . . . . . . 80
8.4 The classication of nite simple groups . . . . . . . . . . . . . . . . . 84
A The groups of order less than 16 86
Index 91
2
Chapter 1
Introduction
This course studies the structure of groups and their representations. As well
as developing general theory, applicable to a large number of problems in group
theory, we will be guided by a desire to understand nite simple groups. These
groups are the building blocks from which all nite groups are made. A complete
classication is beyond this course but we shall see several methods to help
understand the structure of nite groups.
The methods we use will allow us to deduce key properties of a group from
as little information as just the prime decomposition of its order and hence we
will also exclude many possibilities for orders of simple groups. For example,
we will show by solely group-theoretic methods that there are no non-Abelian
simple groups of order strictly less than 60 and we will identify two innite
families of non-Abelian simple groups, namely the alternating groups of degree
ve or more and the projective special linear groups over nite elds.
We also introduce a key notion in the study of group representations, namely
characters of a group, and study important properties of characters such as their
orthogonality relations. Using character theory in a crucial way, we will show
that every group having order divisible by at most two distinct primes is soluble
(a theorem of Burnside) and hence there are no non-Abelian simple groups
having order of this type. For a long time, this result had no proof that did not
use character theory. Although such a proof is now known, this theorem and
its original proof illustrate the power of representation theory in understanding
the structure of groups.
These notes will dier from the course as lectured in 2011 in both organisa-
tion and content and what follows is a summary of these notes. In particular,
there will be separate notes to cover Alperins Fusion Theorem.
We start by recalling basic concepts of group theory and some important
examples. We will mostly concentrate on nite groups but much of what we do
has relevance for innite groups too.
It is important to have good ways of describing and dening groups. Our rst
example is that of group presentationslists of generators and dening relations.
We will also study ways of extending one group by another, particularly by
forming direct and semi-direct products. We then justify our claim that the
nite simple groups are the building blocks of all nite groups, using composition
series and the JordanHolder theorem.
The rst step towards understanding simple groups is the classication of the
3
4 CHAPTER 1. INTRODUCTION
nitely generated Abelian groups. Several of the techniques we use in proving
this classication are more generally applicable, however.
Next we focus on the role of the primes in group theory, where we use
Cauchys theorem on the existence of elements of prime order to help understand
groups of prime power order. These groups are called p-groups and we will prove
Sylows theorems about p-subgroups of groups.
From composition series we are naturally led to the denition and study of
soluble groups and we introduce important advanced tools for studying soluble
groups, such as the derived series.
We next move to studying group representations by means of characters,
which are certain maps from the group to the complex numbers. The value
on a given group element of the character associated to a representation is the
trace of the representing matrix. Although this seems at rst sight to discard
a lot of information about the representation, in fact much remains and we
will construct the character table of a group and extract a lot of structural
information about the group from this. In special cases, one can even show that
the character table determines the group. One surprising feature is that we can
get information about nite groups from working with complex representations,
i.e. over an innite eld.
Also, we will use the orthogonality relations for characters to see that even
if one has only part of the character table, it is often possible to reconstruct the
whole table. We will give several examples but our main application of character
theory will be a proof of Burnsides p
theorem, which states that all groups

whose orders have this form are soluble and hence cannot be non-Abelian simple
groups.
Using all these tools, we will show that there are actually no non-Abelian
simple groups of order strictly less than 60 or strictly between 60 and 168. We
prove that the alternating groups A
n
are non-Abelian simple groups when the
degree n is ve or more and study another class of such groups: the projective
special linear groups, which arise from matrix groups over nite elds, one of
which has order 168.
Classication problems are a large part of what abstract algebra is about
and we will gain some insight into one of the largest and most impressive such
projects ever undertaken in mathematics.
Acknowledgements
The author would like to thank Dr Peter Neumann for his very helpful comments
on these notes during their preparation as well as a number of colleagues and
students who have pointed out corrections during the three years these notes
have been used. Any remaining errors or omissions are solely the responsibility
of the author. Should you spot any of these, I would be grateful if you could
email both me and Prof. Collins (mjc@maths.ox.ac.uk), so that these may be
corrected and announced in lectures.
References
[AB] J. L. Alperin and Rowen B. Bell. Groups and representations, volume 162
of Graduate Texts in Mathematics. Springer-Verlag, New York, 1995. ISBN
0-387-94525-3; 0-387-94526-1.
[B2a] B2a. Introduction to Representation Theory. Course lectured by Dr M. Wemyss
in the Mathematical Institute, Oxford, Michaelmas Term 2009.
[Col] M. J. Collins. Representations and characters of nite groups, volume 22 of
Cambridge Studies in Advanced Mathematics. Cambridge University Press,
Cambridge, 1990. 0-521-23440-9.
[Der] John Derbyshire. Unknown Quantity: A Real and Imaginary History of Alge-
bra. Atlantic Books, 2008. ISBN 978-1-84354-570-5.
[dS] Marcus du Sautoy. Finding Moonshine: A Mathematicians Journey through
Symmetry. Fourth Estate, 2008. ISBN 978-0007214617.
[Gor] Daniel Gorenstein. Finite groups. Harper & Row Publishers, New York, 1968.
Reissued Chelsea, 1980.
[Her] I. N. Herstein. Topics in algebra. Blaisdell Publishing Co. Ginn and Co. New
York-Toronto-London, 1964.
[JL] Gordon James and Martin Liebeck. Representations and characters of groups.
Cambridge Mathematical Textbooks. Cambridge University Press, Cambridge,
1993. 0-521-44024-6; 0-521-44590-6.
[Mac] Ian D. Macdonald. The theory of groups. Oxford University Press, Oxford,
1968. ISBN 0-19-853138-9.
[Neu] Peter M. Neumann. A Hundred Years of Finite Group Theory. The Mathe-
matical Gazette, 80(487):106118, March 1996.
[NST] Peter M. Neumann, Gabrielle A. Stoy, and Edward C. Thompson. Groups and
geometry. Oxford Science Publications. Oxford University Press, New York,
1994. ISBN 0-19-853452-3; 0-19-853451-5.
[Ron] Mark Ronan. Symmetry and the monster. Oxford University Press, Oxford,
2006. ISBN 978-0-19-280722-9; 0-19-280722-6.
[Ros] John S. Rose. A course on group theory. Cambridge University Press,
Cambridge-New York-Melbourne, 1978. ISBN 0-521-21409-2; 0-521-29412-9.
[ST] Geo Smith and Olga Tabachnikova. Topics in group theory. Springer Un-
dergraduate Mathematics Series. Springer-Verlag London Ltd., London, 2000.
ISBN 1-85233-235-2.
Above, we list the books referred to in these notes or suggested by the
lecturer. The recommended texts for this course are [ST] and [JL], with [AB],
[Col] and [Gor] recommended for reference. However, there are many books on
group theory available in Oxford libraries or to buy. Any of the books above
will provide an alternative perspective to the recommended text and you will
nd others not on this list that are equally suitable.
5
Chapter 2
Preliminaries
2.1 History
We start by recalling the key denition for the whole course:
Denition 2.1. A group is a pair (G, ) consisting of a set G and binary
operation : GG G such that
i) g (h k) = (g h) k g, h, k G (associativity)
ii) e G such that e g = g e = g g G (identity element)
iii) g G, g
1
G such that g g
1
= g
1
g = e (inverses)
However, for much of the history of group theory, its practitioners would
not have recognised this denition. Groups began life in a much more concrete
way and the abstract denition above was not really settled until early in the
twentieth century. (Even still, there is some minor debate about the inclusion
of an axiom stating closure: if g and h are in G, then g h should be in G.) The
following very condensed account of the history of the denition of a group is
drawn from [Neu] and [Der].
The idea of a group, or really of groups of permutations, dates back at least
to Lagrange (1736-1813) and his study of roots of polynomials, though this itself
was built on earlier work from the mid-17th century. Also, Lagrange does not
compose his permutations and so he doesnt really have a group in the sense we
would understand.
Typically, credit is given to Galois (1811-1832) for getting group theory
going, not least because in his nal notes on his papers written the night before
he died, he did include a closure axiom and so he did see that he had an
object or closed system. However even Galois did not give an explicit
denition of a group, with sucient axioms. But in some ways one cannot
blame Galois for this: the (occasionally severe) axiomatic approach to algebra
we know today is a much later phenomenon.
In parallel to Galois, Runi (1765-1822) was working on the question of the
solubility of the general quintic equation by radicals. Runi also does not give
an explicit denition of a group but certainly had a good idea of the object
he had. Aware of Runis work and some of Galois earlier work too, Cauchy
6
2.2. RECOLLECTIONS 7
(1789-1857) gave in 1845 a denition of something he called a conjugate system
of substitutions. Jordan (1838-1922) promoted Galois term, a group, and
this then became the standard name.
The next major attempt at a fully abstract denition was made by Cayley
(1821-1895), in a paper dating to 1854. However, it was not quite right, although
some authors credit Cayley with using the term group in its modern abstract
sense for the rst time. In particular, the associative law automatically held in
his situation but he still stated it as an axiom. But one notable feature was
that the denition did not assume the niteness of the underlying set.
Kronecker (1823-1891) gave another almost-denition in 1870, but did not
relate his idea to group theory. In 1882, von Dyck (1856-1934) and Weber
(1842-1913) both gave denitions, again close to the modern one.
In his famous book of 1897 Burnside (1852-1927) gave a denition involving
closure, associativity (though one has the same caveat as with Cayleys) and
inverses (though not explicitly an identity, though this of course follows). At
around the same time, Frobenius (1849-1917), H older (1859-1937) and Weber
(again) made progress towards the modern denition, independently of Burn-
side.
A diagram summarizing the above relationships may be found in Figure 2.1.
So in one sense, group theory is a pretty much a twentieth-century phe-
nomenon, as this is when the denition was really settled, but this understates
how well-developed the subject was even by 1897. Many theorems, particu-
larly Galois, Cauchys and Sylows (1832-1918) (the latter two theorems we
will meet in this course), were established even without the precise form of the
denition we know today. Ideas such as normal subgroups (due to Galois), quo-
tient groups (H older, 1889), composition series and the Jordan-Holder theorem
(we meet these later too) and solubility (ditto) were all present before 1900.
Now we have reached the stage of the development of group theory where
this course sits, putting it in context, I hope. We will always be looking towards
the classication of the nite simple groups and at the end of the course, I hope
to be able to say some more about how the history of that theorem developed.
But for now, I will turn to recalling the denitions and ideas you have met in
previous years, to remind ourselves of these and to set our notations.
2.2 Recollections
2.2.1 Basic denitions and theorems
In Denition 2.1, we have already given the denition of a group, (G, ). We
rst give some names to some basic concepts associated with groups, as well as
some key theorems.
Denition 2.2. The order of a group (G, ) is the cardinality of the set G,
written [G[.
From now on, (G, ) will be written simply as G and the operation denoted
by concatenation.
Denition 2.3. Let G be a group. A subset H G is said to be a subgroup
if for all h, k H, hk
1
H. We write H G.
8 CHAPTER 2. PRELIMINARIES
Frobenius (18491917)
Holder (18591937)
Weber
17th century mathematicians
Lagrange (17361813)
Galois (18111832) Ruffini (17651822)
Cauchy (17891857)
Jordan (18381922)
Cayley (18211895) Kronecker (18231891)
Weber (18421913)
Modern definition
von Dyck (18561934)
Burnside (18521927)
?
?
Liouville (18091882)
1846
1845
1870
1882 1882
1897
1854
1887
1889
1895
Figure 2.1: Diagram showing inuences on the mathematicians working towards
the modern denition of a group.
A subgroup H is said to be proper if H is a proper subset of G (i.e. H ,= G)
and we write H < G.
Denition 2.4. Let G be a group and N a subgroup of G. Then N is said to
be a normal subgroup of G if for all n N and g G, g
1
ng N. We will
write N G. Equivalently, N is normal in G if gN = Ng for all g G.
A proper normal subgroup will be denoted N G.
Subgroups are one way of getting a new group from an old one. Another is
via the quotient construction.
Denition 2.5. Let G be a group and H a subgroup of G. A right coset of H
is a subset of G of the form Hg = hg [ h H. The index of H in G, written
[G : H[, is the number of distinct right cosets of H in G.
Theorem 2.6. (Lagranges theorem) Let G be a group and H a subgroup of G.
Then [G[ = [H[[G : H[.
Denition 2.7. Let G be a group and N a normal subgroup of G. The quotient
group G/N is the group with underlying set cos(G : N)
def
= Ng [ g G (the
set of right cosets of N in G) and group operation given by (Ng)(Nh) = N(gh)
for all g, h G. The identity element is the coset Ne = N and the inverse of
Ng is Ng
1
.
Whenever we have an algebraic structure, we are interested in objects having
that structure and maps (or morphisms) preserving that structure. (Considering
objects and morphisms together, we have the notion of a category.)
Denition 2.8. Let G and H be groups. A group homomorphism from G to
H is a function f : G H such that f(g
1
g
2
) = f(g
1
)f(g
2
) for all g
1
, g
2
G.
Note that it follows immediately from the denition that f(e) = e and
f(g
1
) = f(g)
1
.
The kernel of a homomorphism f : G H is the set
Ker f
def
= g G [ f(g) = e
H
.
The image of f is the set
Imf
def
= h H [ g G such that f(g) = h.
A (group) homomorphism is called a (group) isomorphism if it is bijective.
So, a homomorphism f : G H is an isomorphism if Ker f = e.
The isomorphism theorems are key in identifying isomorphisms of groups.
Theorem 2.9. (First Isomorphism Theorem) Let G and H be groups and let
f : G H be a homomorphism. Then
i) the kernel of f, Ker f, is a normal subgroup of G,
ii) the image of f, Imf, is a subgroup of H and
iii) the quotient group G/Ker f is isomorphic to Imf.
Theorem 2.10. (Second Isomorphism Theorem) Let G be a group, H a sub-
group of G and K a normal subgroup of G. Then HK
def
= hk [ h H, k K is
a subgroup of G and K is a normal subgroup of HK. Furthermore, H K is a
normal subgroup of H and the quotients HK/K and H/H K are isomorphic.
Theorem 2.11. (Third Isomorphism Theorem) Let G be a group and let H and
K be normal subgroups such that H K. Then the quotients (G/H)/(K/H)
and G/K are isomorphic.
The proofs of the second and third isomorphism theorems are left as an
exercise.
2.2.2 Group actions
Groups arose as objects that permuted sets and this idea, generalised a little
to that of a group action, is a very useful one. It will be used often in what
follows, particularly in the guise of a group acting on the set of its subgroups.
Denition 2.12. Let G be a group and a set. A (right) G-action on is a
function : G such that
i) ( g) h) = (gh) for all g, h G, and
ii) e = for all .
The set equipped with the G-action is said to be a G-space.
Denition 2.13. Assume that G acts on a set .
The orbit of an element is the set
Orb()
def
= g [ g G .
The stabilizer of an element is
Stab()
def
= g G [ g = G.
The stabilizer is a subgroup of G.
An action of a group G on a set is said to be transitive if for any distinct
elements , , there exists g G such that g = . Note that the action
of G restricted to an orbit is transitive.
As a geometric example, consider the symmetries of a tetrahedron as an
action on the vertices. Recall that the full symmetry group of the tetrahedron
is S
4
and the rotation group is A
4
.
Denition 2.14. Let G be a group and H a subgroup. The coset space of H
in G is the set cos(G : H) of right cosets of H in G equipped with the action of
G by right translation, namely Hx g = H(xg).
Theorem 2.15. Every G-space (, ) whose action is transitive is isomorphic
to a coset space, for some subgroup H of G. (One takes H = Stab() for some
.)
When one has a subgroup H that is not normal in a larger group G, one
would like to know if there is a subgroup in between H and G in which H is
normal. There is: the normalizer of H. Similarly, given an element h G that
does not necessarily commute with all the other elements of G, one would like
to know if there is a subgroup of G, all of whose elements do commute with h.
Again, there is: the centralizer of h.
Denition 2.16. Let G be a group and H a subgroup of G. We dene the
normalizer of H to be N
G
(H) = g G [ Hg = gH (so H is normal in its
normalizer).
Denition 2.17. Let G be a group. The centralizer of the element h G is
dened to be C
G
(h)
def
= g G [ hg = gh.
Both the normalizer and centralizer can be obtained via group actions.
We conclude with a counting lemma we will utilize later.
Lemma 2.18. Let G be a nite group having order [G[ = p
r
for some prime p
and suppose G acts on a set . Dene Fix
G
() = [ g = g G
(the set of global xed points for G acting on ). Then [Fix
G
()[ [[ (mod p).
Proof: Exercise.
2.2.3 Some examples of groups
In Table 2.1, we recall some of the examples of groups we have met before. Then
in Figure 2.2, we have illustrated how these examples fall into various classes,
namely nite versus innite and Abelian versus non-Abelian. We also highlight
the examples that are simple.
Group Description
C
n
cyclic group of order n, C
n
= e, x, x
2
, x
3
, . . . , x
n1
V
4
group of symmetries of a rectangle (Kleins Vierergruppe)
S
n
(n > 2) symmetric group of degree n
= group of permutations of n symbols
A
n
(n > 4) alternating group of degree n
= group of even permutations of n symbols
A
f
group of even nitary permutations of N
( is nitary if it moves only nitely many points)
D
2n
dihedral group of order 2n
= group of symmetries of a regular n-gon
D
innite dihedral group

C
= Z
+
innite cyclic group = group of integers under addition
C
2
C
3
direct product of C
2
and C
3
C
2
C
direct product of C
2
and C
direct product of C
with itself
Q quaternion group (of order 8)
GL(n, k), group of n n matrices with non-zero determinant,
k nite eld with entries in k
GL(n, k), group of n n matrices with non-zero determinant,
k innite eld with entries in k
SL(n, k), group of n n matrices with determinant 1,
k nite eld with entries in k
SL(n, k), group of n n matrices with determinant 1,
k innite eld with entries in k
isometries
of cube
Q
+
group of rational numbers under addition
Q
group of non-zero rational numbers under multiplication

R
+
group of real numbers under addition
R
group of non-zero real numbers under multiplication

C
+
group of complex numbers under addition
C
group of non-zero complex numbers under multiplication

Table 2.1: List of examples of groups.
2
.
2
.
R
E
C
O
L
L
E
C
T
I
O
N
S
1
3
Infinite
NonAbelian
Infinite
Abelian
Finite
NonAbelian
Finite
Abelian
Finitely generated Infinitely generated
GROUPS
D
8
V
4
2
C x C
3
C
n
C x C
2
8
C
8
C x C
88
S
n
(n>2) D
2n
(n>2)
Q
+
Q
*
R
+
R
*
C
+
C
*
A
f
C
p
(p prime)
A
n
(n>4)
Q
GL(n,k), k infinite field
SL(n,k), k infinite field
GL(n,k), k finite field
SL(n,k), k finite field
{isometries of cube}
SIMPLE
F
i
g
u
r
e
2
.
2
:
D
i
a
g
r
a
m
i
l
l
u
s
t
r
a
t
i
n
g
v
a
r
i
o
u
s
g
r
o
u
p
s
a
n
d
t
h
e
c
l
a
s
s
e
s
t
h
e
y
f
a
l
l
i
n
t
o
.
2.2.4 Cyclic groups
We will see later that the nite cyclic groups are the building blocks for general
nite Abelian groups. You were originally introduced to the nite cyclic groups
before, along the following lines:
Denition 2.19. The nite cyclic groups are the groups that are powers of one
element, a, say, so that the elements of the group consist of e, a, a
2
, . . . , a
n1
for some n. The group operation is given by a

i
a
j
= a
i+j (mod n)
.
The better denition of cyclic groups is that they are precisely the groups
generated by a single element. You have met generators before and we will
discuss them more in Chapter 3, and from that point the better denition will
apply. For now, the above suces.
Examples 2.20 ([ST, Ex. 1.36]).
i) the subgroup 1, 1 of Q
(Q
= Q 0 under multiplication). Observe

that 1 is a generator but 1 is not.
ii) the subgroup 2
i
[ i Z of Q
. The set underlying this subgroup is

. . . ,
1
8
,
1
4
,
1
2
, 1, 2, 4, 8, . . . . Here both 2 and
1
2
generate the subgroup; no
other elements do.
iii) the subset !
12
of C comprising the roots of x
12
1, with group structure
coming from multiplication in C. More explicitly, this set is
!
12
= e
2ki
12
[ 0 k 11.
Set = e
2i
12
and observe that ,
5
,
7
and
11
each generate !
12
but
no other powers of do.
Even without many tools to hand, we can completely describe the structure
of cyclic groups, which we shall do in the following results. First, we recall the
denition of the order of an element.
Denition 2.21. Let G be a group and let x G. Dene o(x), the order of x,
to be [ x[.
The phrasing of this denition may surprise you a little: you might have
been expecting o(x) to be dened to be the least positive integer n such that
x
n
= e, or innite if no such n exists. Our new version improves the old one
slightly, by not having to separate the two cases. Any concerns you may have
about our new denitions of cyclic groups and the order of an element should
be assuaged by the following proposition.
Proposition 2.22 ([ST, Prop. 1.40]). Let G = x be a cyclic group. Then
exactly one of the following two statements is true:
i) 1 [G[ = n < , and the elements e = x
0
, x
1
, x
2
, . . . , x
n1
are distinct,
G = x
0
, x
1
, x
2
, . . . , x
n1
and x
n
= e. For n 2, so G is non-trivial,
the generic cyclic group will be denoted C
n
, the cyclic group of order n.
ii) [G[ is (countably) innite, in which case the elements x
i
, i Z, are dis-
tinct. This group will be denoted C
.
Proof: Exercise. In particular, verify the implicit claim that up to isomorphism
there is only one cyclic group of each nite order n 2 and one countably
innite cyclic group.
Corollary 2.23 ([ST, Prop. 1.38]). Let a be a cyclic group. Then a is
Abelian.
Now we look at the subgroup structure of the cyclic groups.
Proposition 2.24. Let G = x be cyclic.
i) ([ST, Prop. 1.43]) If H is a subgroup of G then H is cyclic. Indeed
H = x
n
for some n 0.
ii) ([ST, Prop. 1.45]) If G is nite of order n, then G has exactly one subgroup
of each order k 1 dividing n, and no others.
iii) ([ST, Prop. 1.45]) If G is innite, then G has exactly one subgroup of
index r for each r 1 and exactly one subgroup of innite index.
Proof:
i) Proved in Moderations course; alternatively, an easy warm-up exercise.
ii) Ditto.
iii) Let r 1 be a positive integer. Then x
r
= x
r
has index r in
G = x since the r cosets x
i
x
r
for 0 i < r are distinct and their
union is G. By (i), since all subgroups of a cyclic group are cyclic, the
only other subgroup of G is x
0
= e, which has (countably) innite
index in G.
We conclude our study of cyclic groups with the determination of the orders
of their elements.
Proposition 2.25 ([ST, Prop. 1.46]). Let G = x be cyclic.
i) If [G[ = n < and j Z then o(x
j
) =
n
(n,j)
where (n, j) = hcf(n, j).
ii) If [G[ = and j Z 0 then o(x
j
) = .
Proof: Exercise. (Part (i) is essentially a number-theoretic result. It also justi-
es the observation at the end of Example 2.20(iii).)
Chapter 3
Some basic tools for
creating and analyzing
groups
In this chapter, we introduce two ways in which groups can be created and hence
described. The rst is presentation. Generators are special elements of a group,
distinguished by the property that from a set of generators we may build the
whole of the group. A presentation is a list of generators together with a list
of dening relations between the generators. This gives a much more compact
description of a group than its Cayley table, where one has to list all elements
and their products.
Secondly we look at various products of groups. We focus on two sorts,
direct and semi-direct products. In fact, each comes in two avours, internal
and external, though these are very closely related. The external direct or
semi-direct product is a way of putting a group structure on the Cartesian
product of the sets underlying two (or more) groups. The internal direct or
semi-direct products are really statements about certain groups decomposing
into subgroups, which are glued back together by the corresponding external
product construction.
Finally, we have another way to try to untangle the structure of a group,
by looking at composition series, which are certain chains of normal subgroups
within our group. The JordanHolder Theorem tells us that for a nite group,
composition series have a strong invariance property and so give us signicant
information about the group.
3.1 Generating sets in groups and their sub-
groups
The concept of generators of a group is fundamental for advanced group theory.
The use of the word generator might also trigger memories of generators of
ideals in the context of ring theory. Although there are some dierences between
that idea and that of generators for groups, the core of the idea is the same. We
start with a slightly informal denition, which hopefully captures the essence of
16
3.1. GENERATING SETS IN GROUPS AND THEIR SUBGROUPS 17
the idea, then formalise it.
Denition 3.1. Let G be a group and let X be a subset of G. The subgroup
of G generated by X is the smallest subgroup of G containing X.
The informality is in the use of the word smallest: what if two dierent
subgroups of the same order (i.e. size as sets) both contain X? In fact this
cannot happen and by improving our denition, we can see this immediately.
First, remember that the intersection of two subgroups is a subgroupin fact,
for any non-empty set I (possibly uncountable), if G
i
[ i I is a set of
subgroups of G then
iI
G
i
is again a subgroup. (Check you can prove this:
note that induction from the [I[ = 2 case wont work in general!) Now we can
give a formal denition.
Denition 3.2 ([ST, Def. 1.34]). Let G be a group and let X be a subset of
G. First dene o
X
= H [ H subgroup of G, X H, the set of subgroups of
G containing X. (Note that G o
X
so o
X
,= .) The subgroup of G generated
by X, written X, is dened to be X =
HS
X
H.
This denition takes care of both the existence and uniqueness of X.
Notice that H = H for any subgroup H of G, and of course G = G.
Indeed H = H for a subset H G implies H is a subgroup. Since the
identity element is the only element in every subgroup of G (certainly it is the
only element in the subgroup e!) we have e =
HS
{e}
H = e. Even
more trivially, = e.
Examples 3.3.
i) in S
3
(the symmetric group on three symbols, 1, 2, 3), we have
(1 2) = , (1 2)
(1 2 3) = , (1 2 3), (1 3 2)
(1 2), (1 2 3) = S
3
(1 2), (2 3) = S
3
(1 2), (2 3), (1 3) = S
3
(Notice that we drop the , s: rather than writing a, b we use
a, b . Also, as the last of our examples here shows, generating sets need
not be minimal, though often we are interested in the minimal ones.)
ii) in the Abelian group Z under addition, 2 = 2Z. You may be tempted
to think of ideals here, in particular principal ideals. However here is a
warning. In a general ring R, a principal ideal is aR = ar [ r R, but
in a group, gG = gh [ h G = G for all elements g G. The point is
that g is invertible (G is a group!) but a typically is not in R. Indeed,
ideals are supposed to be like normal subgroups, so that one may take
quotients, so the analogy between generators of subgroups and of ideals is
not exact.
We placed no restrictions on the cardinality of G or of Xthe above deni-
tion works ne in all cases. However, since this course will largely concern nite
groups, it should be of no surprise that we are interested in X and/or G being
nite.
18 CHAPTER 3. SOME BASIC TOOLS
Denition 3.4. Let G be a group. We say that G is nitely generated if there
exists a nite subset X G such that X = G.
Denition 3.5. Let G be a group and n N. We say G is an n-generator
group if there exists a set of generators X for G with [X[ n. If there ex-
ists a countable set X such that X = G, we say G is countably generated.
Otherwise, we call G uncountably generated.
We also apply the corresponding terminology to subgroups. A crucial special
case is n = 1, a class of groups you already know well.
Denition 3.6. A group G is said to be cyclic if it is 1-generator. That is, G
is generated by one element: there exists a G such that G = a .
Importantly, this denition applies to G innite as well as nite. The lan-
guage of generators also gave us the sophisticated denition of the order of an
element, similarly not distinguishing between nite and innite cases:
Denition 3.7 ([ST, Def. 1.39]). Let G be a group and let x G. Dene o(x),
the order of x, to be [ x[.
Examples 3.8.
i) For example, S
3
is nitely generatedin fact, 2-generator. Indeed for
n 1, S
n
is a 2-generator group.
ii) By our denition, a nitely generated group is countably generated.
iii) Clearly if G is nite, it is nitely generated (by itself) and so are all of its
subgroups. However as soon as G is not nite, the situation can become
much more complicated, as we see in the next proposition.
Proposition 3.9. Let G be an innite group.
i) if G is nitely generated then G is countable ([G[ = [N[).
ii) there exists a countable group that is not nitely generated.
iii) if G is countably generated then G is countable.
iv) if G is nitely generated and H is a subgroup of G such that the index
[G : H[ is nite then H is nitely generated.
v) if H is a subgroup of G and H is nitely generated and the index [G : H[
is nite then G is nitely generated.
vi) there exists a nitely generated group W (in fact, 2-generator) having a
subgroup that is not nitely generated.
Proof: Since (iii) implies (i), we prove (iii) rst. Let X be a (countable) gener-
ating set for G. Dene the set X
1
= x
1
[ x X and then let
X
= e y
1
y
2
y
n
[ n 1, y
i
X X
1
.
We call expressions y = y
1
y
2
y
n
words in X X
1
; note that words have
nite length. Observe that certainly X X
but also X
is a subgroup of
3.2. PRESENTATIONS OF GROUPS 19
G: X
is non-empty and if y = y
1
y
2
y
m
and z = z
1
z
2
z
n
are in X
,
then so is yz
1
= y
1
y
m
(z
1
z
n
)
1
= y
1
y
m
z
1
n
z
1
1
. So by denition
X X
.
Now consider y X
. If y = e then y X
. If y = y
1
y
n
for
y
i
XX
1
then since X X and X is a subgrouphence closed under
products and inverseswe have y X. (This can be made more formal by
an induction on n.) Hence X
X and thus X = X
. This is the precise

formulation of the idea that one should get all of X by combining generators
and their inverses. (This proof is based on that of [Mac, Thm. 3.10].)
We now simply need to see that if X is countable, then so is X
: for each n
there are only countably many words y = y
1
y
n
(y
i
X X
1
) and X
is
the union of countably many such sets of words. Hence G is countable.
For (ii), we turn to a familiar example: Q
+
, the Abelian group of rational
numbers under addition. We claim that Q
+
, which we know to be countable,
is not nitely generated. We do this by contradiction, in a fairly explicit way.
Suppose that Q
+
is generated by nitely many rational numbers q
1
, . . . , q
m
.
Now rationalise denominators: choose n Z, n 1, such that nq
i
Z for all
i. Let A =
1
n
. Then q
i
A for each i so Q
+
= q
1
, . . . , q
m
A Q
+
and
hence Q
+
= A. But this cannot be:
1
2n
/ A, for example. Hence Q
+
cannot be
nitely generated. (This part is based on [ST, Prop. 1.44].)
We omit the proof of (iv) and leave (v) as an exercise. For (vi) we refer to
[Ros, Section 3.37].
3.2 Presentations of groups
In this section, we meet the idea of a presentation of a group. We have met
generating sets already and have seen that they are very useful things to know
but we would like to describe a group completely and to do this, we need to know
if any relations hold between the generators. A presentation gives us exactly
this information but the trade-o is that one has to accept that a group will not
have a unique presentation.
We begin with the idea of a free group: this is a group without relations
between its generators.
Denition 3.10. A group F is said to be free on a set X if
i) X F and
ii) for any group G and map : X G, extends to a unique homomorphism
F G.
Set rank(F) = [X[.
Two key properties of free groups are the following.
Proposition 3.11.
i) If F is free on a set X then F = X.
ii) If F
is also free on the set X then F

= F. (Thus it is reasonable to talk
of the free group on X.)
Proof:
i) Let H = X F and let i : X F, : X H and j : H F be the
natural injective set maps. Since F is free on X, by denition extends
to a homomorphism
: F H. The set map j : X F extends

to the homomorphism j
: F F. But i : X F extends to the

homomorphism id
F
: F F (the identity homomorphism on F) and i
and j have the same image in F. So by uniqueness, j
= id
F
, so
Imj
= Im id
F
= F. Thus Imj = F and H = X = F.
ii) To show F

= F if F
is also free on X, let i : X F and i
: X F
be
the natural injective set maps. Then since F is free on X, i
extends to
a homomorphism f
: F F
. Similarly, i extends to a homomorphism

f : F
F. We wish to prove that f f
= id
F
and f
f = id
F
.
Let x X. Since f extends i, f[
X
= i so f(x) = i(x) = x for all x X;
similarly for f
. Then (ff
)(x) = f(i
(x)) = f(x) = x so (ff
)[
X
= id
X
.
But then (f f
)[
X
extends to both f f
: F F and id
F
: F F, hence
by uniqueness f f
= id
F
. Similarly, f
f = id
F
and so F

= F.
Free groups exist: given a set X, use the construction in the proof of Proposi-
tion 3.9 to build the set of words X
in X and (formal) inverses of elements,

X
1
. Then take the underlying set for F to be the set X
of reduced words
in X
: a word is reduced if it contains no adjacent pair of mutually inverse

elements (e.g. x
1
x
2
x
1
1
is reduced but x
1
x
1
1
x
2
is not).
The group structure on F is the obvious one, concatenation of words followed
by reduction if necessary. For example, x
1
(x
1
1
x
2
) = x
1
x
1
1
x
2
= x
2
. The empty
word (denoted ) is the identity element. Of course, there are some things still
to check to see that F is indeed a groupnotably associativitybut we will
assume these (and you can convince yourself at your leisure).
We claimed that a free group is without relations (other than that inverses
multiply to the identity, as required by denition). In particular, no power of a
non-identity element can be the identity.
Proposition 3.12. A free group F is torsion-free, that is, no non-identity
element of F has nite order.
Proof: We say a word w = x
1
x
2
x
l
is cyclically reduced if it is reduced (in
the sense above) and x
1
,= x
1
l
. We claim that any non-trivial word w may be
written as w = u
1
vu, a reduced expression for w with v ,= cyclically reduced.
If w is cyclically reduced, take u = , v = w. If w = x
1
x
2
x
l
is not
cyclically reduced, let r be the largest integer such that x
s
= x
1
ls+1
for all
s r. Then take u = x
lr+1
x
lr+2
x
l
and v = x
r+1
x
r+2
x
lr
, so we see
that
w = x
1
x
2
x
r
x
r+1
x
r+2
x
lr
x
lr+1
x
lr+2
x
l1
x
l
= x
1
l
x
1
l1
x
1
lr+1
. .
u
1
x
r+1
x
r+2
x
lr
. .
v
x
lr+1
x
lr+2
x
l1
x
l
. .
u
.
Since v is cyclically reduced by the choice of r, we have the desired factorisation.
Then w
n
= u
1
v
n
u ,= for all n 1, so F is torsion-free.
3.2. PRESENTATIONS OF GROUPS 21
Remark 3.13. An alternative proof, not using cyclically reduced words, may be
found in [ST, Proposition 6.14, p. 201].
The signicance of the class of free groups among all groups is indicated by
the following theorem.
Theorem 3.14. Any group is isomorphic to a quotient of a free group.
Proof: Let G be a group and X a generating set for G. Let F be the free group
on X and let : X G be the identity set map. By the denition of a free
group, extends to a homomorphism : F G. Then F/Ker

= Im = G.
Alternatively, we may dene : F G to be the evaluation map: given a
word in F, map this to the corresponding product of elements of G. Then is
clearly a surjective homomorphism and so G

= F/Ker .
Next we need to consider how to impose relations.
Denition 3.15. If G is a group and R is a subset of G then the normal closure
R
G
of R in G is dened to be the intersection of all normal subgroups of G
containing R. In general, this is larger than R.
The normal closure has a nice set of generators, as follows.
Lemma 3.16 ([ST, Prop. 2.15]). We have R
G
= g
1
rg [ g G, r R,
the subgroup of G generated by the elements g
1
rg.
Proof: Let N = R
G
, S = g
1
rg [ g G, r R and T = S . Note that
S N since N is normal in G, N being an intersection of normal subgroups.
Since S M for any M such that R M G, we see that S N implies
T N. Conversely, R S so R T. Also S is invariant under conjugation so
T G and so N T by the denition of normal closure. Hence N = T.
Now we can dene a presentation properly:
Denition 3.17. Let X be a set and let R X
be a subset of the set of

reduced words in X
. Dene the group X [ R to be the quotient group

F/ R
F
where F is the free group on X and R
F
is the normal closure of R
in F.
If G is a group, X G and R X
such that G = X [ R, we say X [ R

is a presentation for G.
It is not clear from the denition that an arbitrary group has a presentation,
but indeed it does.
Theorem 3.18. Let G be a group. Then G has a presentation X [ R.
Furthermore, if G is nite, we can choose X and R to be nite.
Proof: By the proof of Theorem 3.14, we have a free group F on a set X that
generates G and a normal subgroup N of F such that F/N

= G. Let R be any
generating set for N. Then we have G = X [ R, since N is normal in F and
so N = R
F
, and thus G has a presentation.
If G is nite, we can choose X = G to get X nite. Since F/N

= G is nite,
N must have nite index in F and so is nitely generated by Proposition 3.9(iv),
giving R nite.
The usual notation for presentation follows the pattern in the examples
that follow. To see how one might prove that a group is given by a particular
presentation, we give an example. We claim that x [ x
n

= C
n
for n 2
(clearly, x [ x

= e). First we note that x
def
= x [

= C
(the
torsion-free Abelian group of rank one). The group x [ x
n
is then seen to be
isomorphic to C
/ y
n
(where y is a generator of C
). Since y
n
has index
n in C
, we see that [C
/ y
n
[ = n. Now C
n
= z , generated by the element
z, is of order n and z satises the relation z
n
= e. Hence there is a surjective
homomorphism x [ x
n
C
n
but these sets have the same cardinality, n, so
there is an isomorphism of these groups.
Examples 3.19.
i) x, y [ xy = yx

= C
.
ii) x [ x
a
= x
b
= e

= e if (a, b) = 1.
iii) x, y [ x
3
= y
2
= (xy)
2
= e

= S
3
.
3.3 Direct products, semi-direct products and
extensions
In the next chapter, we will see that cyclic groups are the building blocks for the
nite and nitely generated Abelian groups. Of course, a brick wall is no good
without mortar to hold it together so we need to learn to how glue our groups
together. That is the aim of this section: to see how to build larger groups from
smaller ones. In fact there are many ways to do this but we will concentrate
on two of the simplest, the direct product and the semi-direct product, which
give you ways of putting a group structure on the Cartesian product of the sets
underlying two (or more) groups. We will conclude this chapter with a brief
look at the more general framework of extensions of groups.
Both types of group product come in two avours and we start with a re-
minder of the external direct product, since this is more familiar to you.
Denition 3.20. Let G
1
, G
2
be groups. The external direct product of G
1
and G
2
, denoted G
1
G
2
, is the group with underlying set G
1
G
2
and group
operation (g
1
, g
2
) (h
1
, h
2
) = (g
1
h
1
, g
2
h
2
).
Remark 3.21. There are several claims implicit in this denition regarding as-
sociativity, an identity element and inverses, and you should convince yourself
of the truth of these.
Remark 3.22. Strictly speaking, G
1
and G
2
are not subgroups of G
1
G
2
.
However it is clear that

G
1
= (g
1
, e
2
) [ g
1
G
1
and

G
2
= (e
1
, g
2
) [ g
2
G
2
are isomorphic to G
1
and G
2
respectively (where e
i
is the identity element of
G
i
). In fact,

G
1
and

G
2
are normal subgroups of G
1
G
2
and G
1
G
2
/
G
1

= G
2
and vice versa. Also, G
1
G
2

= G
2
G
1
.
Now we turn to the internal direct product. First we need some general
ideas about products of subgroups.
Denition 3.23. Let A, B be subsets of a group G (not necessarily subgroups).
We dene AB = ab [ a A, b B.
3.3. DIRECT AND SEMI-DIRECT PRODUCTS. EXTENSIONS 23
Lemma 3.24. Let G be a nite group and let A and B be subgroups of G. Then
[AB[ = [A[[B[/[A B[.
Proof: Exercise.
Lemma 3.25. Let G be a group and let A, B G. Then AB G if and only
if AB = BA.
Proof: Exercise.
Let A, B be subgroups of a group G such that AB = G. Notice that if
A B ,= e, so that there exists c ,= e in A B, we do not have unique
factorisation of elements of G into a product of elements of A and B. For
example, ee = cc
1
= e are two dierent factorisations of e.
However, if A B = e it is easy to see that factorisations are unique. It
is not true in general that elements of A commute with elements of B, though.
If b B and a A, then ba need not have the factorisation ab, even though
as sets AB = BA. This problem does not occur if A and B are normal, which
leads to the following denition and lemma.
Denition 3.26 ([ST, Def. 2.36]). Let G be a group with normal subgroups
A and B such that AB = G and A B = e. Then we say G is the internal
direct product of A and B and write G = A B (= B A by the preceding
lemma).
Lemma 3.27. If G is the internal direct product of normal subgroups A and B
then ab = ba for all a A, b B.
Proof: Recall the notation g
h
= h
1
gh for g, h G. Let a A and b B and
consider the element a
1
b
1
ab (this element is called the commutator of a and
b). Now a
1
b
1
ab = a
1
(a
b
) = (b
1
)
a
b. Since A is normal in G, a
b
A so
a
1
b
1
ab A. Similarly, B is normal in G so (b
1
)
a
B and so a
1
b
1
ab B.
So a
1
b
1
ab A B = e and hence ab = ba.
Then if elements g and h of G have factorisations g = a
1
b
1
, h = a
2
b
2
then
gh = (a
1
b
1
)(a
2
b
2
) = (a
1
a
2
)(b
1
b
2
) by the lemma. This should be sucient to
convince you that the external direct product G
1
G
2
of groups G
1
and G
2
is
the internal direct product

G
1

G
2
, where

G
i
is as previously.
Remark 3.28. If G = A B is Abelian, some authors write A B (the direct
sum). However we will continue to write A B in all cases (and will not dis-
tinguish internal and external products, since context will make the distinction
clear).
Remark 3.29. The above can be extended to nitely many groups or subgroups.
For the internal version, one takes nitely many normal subgroups H
i
G
(1 i n) such that G = H
1
H
2
H
n
and asks that
H
i
(H
1
H
2
H
i1
H
i+1
H
n
) = e.
(Compare with the denition of the direct sum of nitely many subspaces of a
vector space.)
It is not sucient to ask that H
i
H
j
= e for all i ,= j. For example, take
G = C
2
C
2
(also known as Kleins Vierergruppe, V
4
). Since G is Abelian, all
subgroups are normal and there are three subgroups of order 2, H
i
(i = 1, 2, 3),
consisting of one of the non-trivial elements of C
2
C
2
and the identity element.
Now of course H
i
H
j
= e for i ,= j but G ,= H
1
H
2
H
3

= C
2
C
2
C
2
(a group of order 8!). Indeed we see that H
j
H
k
= G for all j ,= k and hence
H
i
H
j
H
k
= H
i
for all i ,= j ,= k ,= i.
Example 3.30. C
6
= x with x
6
= e, so C
6
= e, x, x
2
, x
3
, x
4
, x
5
. Since C
6
is
Abelian all subgroups are normal so we need only nd subgroups whose product
is all of C
6
. We see that
x
2
x
3
= e, x
2
, x
4
e, x
3
= ee, ex
3
, x
2
e, x
2
x
3
, x
4
e, x
4
x
3
= e, x
3
, x
2
, x
5
, x
4
, x
= C
6
so C
6
= x
2
x
3
. We can see that x
2

= C
3
and x
3

= C
2
so that
C
6

= C
3
C
2

= C
2
C
3
.
Next we study a modication of the construction of the direct product,
the semi-direct product. Semi-direct products, as with direct products, can be
thought of and constructed in two dierent ways. This time, we shall consider
the internal version rst. (This terminology of internal semi-direct product
is not standard but we use it for the moment to highlight the relationship with
the corresponding notion for direct products. In Lemma 3.51, we see why the
name split extension is more often used.)
Denition 3.31 ([ST, Def. 2.51]). Let G be a group with normal subgroup A
and (not necessarily normal) subgroup B such that AB = G and A B = e.
Then we say G is the internal semi-direct product of A by B and write AB.
Remark 3.32. If B is normal in G then by our previous denition (Deni-
tion 3.26) G is the internal direct product of A and B.
Remark 3.33. Since A and B are no longer assumed to have the same properties,
they cannot simply be interchanged. Indeed, if B is not normal in G, B A
does not make sense. This is emphasised by saying semi-direct product of A
by B, not of A and B. Although we have weakened the denition of a direct
product, we have not lost everything.
Lemma 3.34. Assume that G is the internal semi-direct product of A by B, so
G = A B. Then for all g G there exist unique elements a A and b B
such that g = ab.
Proof: If g = ab = cd with a, c A, b, d B then c
1
a = db
1
A B = e
so a = c and b = d.
Remark 3.35. For G = A B, elements of A and B commute: ab = ba for all
a A, b B. This need not hold in a semi-direct product.
Example 3.36. Consider G = S
3
, A = (1 2 3) and B = (1 2) . Clearly
A B = and it is straightforward to check that AB = G. Therefore
S
3
= (1 2 3) (1 2) . Notice that (1 2) and (1 2 3) do not commute and
S
3
,= (1 2 3) (1 2) (the latter is Abelian, the former is not).
Lemma 3.37. Let G be a nite group such that G is the semi-direct product of
A by B. Then [G[ = [A[[B[.
Proof: A straightforward application of Lemma 3.34.
It is not true, however, that if A G then there exists a suitable B G
so that G = A B. For example, the Abelian group C
4
= a has a normal
subgroup A = e, a
2

= C
2
but the only subgroups of C
4
are e, A and C
4
and none satisfy the required conditions (A A ,= e). So C
4
does not have a
subgroup B such that C
4
= AB.
We can actually be more explicit about the multiplication in a semi-direct
product G = AB than one might expect. Elements from the two factors will
not commute in general, as we have seen, but consider the following equations:
(a
1
b
1
)(a
2
b
2
) = a
1
b
1
a
2
b
1
1
b
1
b
2
= (a
1
(
b
1
a
2
))(b
1
b
2
), a
1
, a
2
A, b
1
, b
2
B.
Here
b
1
a
2
= b
1
a
2
b
1
1
is a
2
left conjugated by b
1
. Until now, we have used the
right conjugate, a
b
= b
1
ab but our convention for G = AB requires us to use
left conjugation
1
instead. So for any g, h G, since we have unique expressions
g = a
1
b
1
and h = a
2
b
2
, we know how to multiply g by h: gh = (a
1
(
b
1
a
2
))(b
1
b
2
)
and the right-hand side is the unique expression for gh G = A B. Notice
that if elements of A commute with elements of B then we just have the usual
formula for multiplication in a direct product.
In order to justify our denition of an external semi-direct product, we for-
malise this a little more. Let b B. Since A is normal in G, left conjugation by
b sends an element of A to another element of A. So we can write down a map
b
: A A,
b
(a) =
b
a = bab
1
. It is easy to show that
b
is a bijective homo-
morphism (that is, an automorphism of A) and that
b

d
=
bd
(Exercise).
Then we may write a
1
(
b
1
a
2
)b
1
b
2
= a
1
b
1
(a
2
)b
1
b
2
.
We can collect together all the maps
b
into one map: : B Aut(A),
dened by b
b
, where Aut(A) is the group of all automorphisms of A (i.e.
bijective homomorphisms A A). The product in Aut(A) is composition.
Then the map is a homomorphism: (bc) =
bc
=
b

c
; such maps are
precisely the group actions of B on A. Now we can write
gh = (a
1
b
1
)(a
2
b
2
) = a
1
[(b
1
)(a
2
)]b
1
b
2
.
However, this notation is somewhat clumsy, so we will adopt the following. For
(b)(a), we write simply
b
a. We see that the semi-direct product structure
is completely determined by the data (A, B, ). Knowing this, we may build
external semi-direct products.
Denition 3.38. Let A and B be groups and let : B Aut(A) be a homo-
morphism. Denote the element (b)(a) A by
b
a. On the set A B dene a
group structure by
(a
1
, b
1
)(a
2
, b
2
) = (a
1
(
b
1
a
2
), b
1
b
2
),
with identity element (e
A
, e
B
). Denote this group by A
B, the external
semi-direct product of A by B with action .
1
The terms left conjugation and right conjugation are not entirely standard, though
the notations
b
a and a
b
are more so. If we say simply conjugation, we will mean right
conjugation.
Remark 3.39. There are implicit claims regarding the associativity of the prod-
uct and existence of inversesthese should be veried. Indeed, one can write
an explicit expression for (a, b)
1
(Exercise).
Remark 3.40. If : B Aut(A) is the trivial homomorphism (b) = id
A
for
all b B, where id
A
: A A, a a, then A
B = AB. If is non-trivial
then A
B is non-Abelian, since there must exist some a A, b B such

that
b
a ,= a and then (e
A
, b)(a, e
B
) = (
b
a, b) ,= (a, b) = (a, e
B
)(e
A
, b).
Remark 3.41. The subgroups

A = (a, e
B
) [ a A,

B = (e
A
, b) [ b B
are isomorphic to A and B respectively and A
B is the internal semi-direct

product of

A by

B.
Note that if and are distinct homomorphisms from B to Aut(A), there
is no reason to expect that A
B and A
B are isomorphic and indeed in

general they are not. However in certain special cases, they are.
Proposition 3.42 ([AB, Prop. 11, p. 23]). Let G and H be groups. If and
are injective homomorphisms from G to Aut(H) with the same image in
Aut(H), then H
G is isomorphic to H
G.
Proof: (non-examinable) We rst claim that there is an automorphism of G,
Aut(G), such that = . Let A = Im = Im. Then since and are
injective, they are isomorphisms between G and their images, i.e. are bijective
homomorphisms G A. Dene : G G by =
1
. Then since and
are bijective homomorphisms, so is . By denition, = .
Now dene : H
G H
G by ((h, g)) = (h, (g)). To distinguish the

actions of and , we will write
b
a and
b
a for (b)(a) and (b)(a), respectively.

Then
((h
1
, g
1
)(h
2
, g
2
)) = ((h
1
(
g
1
h
2
), g
1
g
2
))
= (h
1
(
g
1
h
2
), (g
1
g
2
))
= (h
1
(
(g
1
)
h
2
), (g
1
)(g
2
))
= (h
1
, (g
1
))(h
2
, (g
2
))
= ((h
1
, g
1
))((h
2
, g
2
))
so that is a homomorphism.
Similarly, the map : H
G H
G, ((h, g)) = (h,

1
(g)) is a
homomorphism. It is clear that and are mutually inverse homomorphisms.
Indeed, one sees from this proof that the key requirement on and is that
there exists Aut(G) having = the conditions on the kernels and
images we have given just guarantee the existence of a .
So far, our only example of a semi-direct product has been S
3
, so you might
question the value of the construction at this point. However, now we can give
one of the most important and well-known examples of a semi-direct product,
namely the dihedral groups.
Example 3.43. Let A = C
n
= a for some n N and let B = C
2
= b .
Let : B Aut(A) be the map determined by
b
(a
r
) = a
r
,for r Z, i.e.
the non-trivial element of B acts by inverting elements of A. Then the group
A
B = C
n

C
2
is isomorphic to D
2n
, the dihedral group of order 2n. If
n > 2, C
n
C
2

= D
2n
is non-Abelian; C
2
C
2
= C
2
C
2

= D
4
(A = C
2
= a
has a
1
= a so is trivial).
The group D
2n
is dened to be the symmetry group of a regular n-gon, so
what we have done is to give a realisation of the dihedral group as a semi-direct
product. The generator a of C
n
corresponds to rotation by 2/n radians and b
corresponds to reection about some xed axis. As a result, D
2n
is sometimes
called D
n
, but we shall stick to the convention that the index gives the order of
the group.
Example 3.44. A generalisation is obtained by taking A = C
, B = C
2
and
as above: this gives D
def
= C
C
2
, the innite dihedral group.
Example 3.45. This generalises further by letting A be any Abelian group and
taking B = C
2
and as above (the generalised dihedral groups A
C
2
).
More complicated examples are the following.
Example 3.46. Let H be any group. Set A = H
r
=
r times
..
H H and observe
that S
r
(the symmetric group of degree r) acts on A by permuting the factors.
Setting B = S
r
and letting be this action, we can form H
r
S
r
, the wreath
product of H by S
r
, often written H S
r
. This group has order [H[
r
r!, of course.
More generally, replace S
r
by any group K acting on a set of size r, to get H K.
Semi-direct products are of use in classifying groups, too, as we will see
later. Now, however, we discuss the general extension problem. An extension
is dened as follows.
Denition 3.47. Let A and B be groups. Then an extension of A by B is a
group G together with a normal subgroup K such that K

= A and G/K

= B.
The extension problem is: given groups A and B, can one classify all exten-
sions of A by B? This problem is hard but sophisticated methods to tackle it
have been developed. Notice that some extensions certainly exist: A B is an
extension of A by B (and of B by A). Semi-direct products are also extensions,
since A AB and AB/A

= B.
Another way to view an extension is as a short exact sequence of groups:
Denition 3.48. Let M, N, P be groups and (temporarily) let 1 denote the
trivial group, e. A short exact sequence involving M, N and P is a sequence
1
//
M
i
//
N

//
P
//
1
with i injective, surjective and Imi = Ker .
(The last of these is exactness at N; exactness at M is equivalent to i
being injective and exactness at P is equivalent to being surjective, since i
and are maps from and to the trivial group, respectively.)
Observe that i(M) = Ker N and P

= N/M. So in our previous notation
(Denition 3.47) an extension of A by B is a short exact sequence
1
//
A
i
//
G

//
B
//
1
where K = Imi.
Examples 3.49. (including some not given in lectures)
i) extensions of C
2
= a by C
2
= b :
(a)
1
//
C
2
i
1
//
C
2
C
2
1
//
C
2
//
1 ,
i
1
(a) = (a, e),
1
(e, b) = b
(b)
1
//
C
2
i
2
//
C
4
2
//
C
2
//
1 ,
C
4
= c , i
2
(a) = c
2
,
2
(c
2
) = e
ii) extensions of C
3
= a by C
2
= b :
(a)
1
//
C
3
i
1
//
C
6
1
//
C
2
//
1 ,
C
6
= c , i
1
(a) = c
2
(b)
1
//
C
3
i
2
//
S
3
2
//
C
2
//
1 ,
i
2
(a) = (1 2 3)
iii) extensions of C
n
= a by C
2
= b :
(a)
1
//
C
n
i
1
//
C
2n
1
//
C
2
//
1 ,
C
2n
= c , i
1
(a) = c
2
(b)
1
//
C
n
i
2
//
D
2n
2
//
C
2
//
1 ,
determined by D
2n
= C
n
C
2
as above (a a)
iv) extensions of A
n
, the alternating group, by C
2
3.4. COMPOSITION SERIES. THE JORDANH
OLDER THEOREM 29
(a)
1
//
A
n
//
A
n
C
2
//
C
2
//
1
(b)
1
//
A
n
//
S
n
//
C
2
//
1
Several of these examples are trivial: we know direct and semi-direct prod-
ucts are extensions. But how do we know when an extension gives us something
trivial?
Denition 3.50. An extension of a group A by a group B described by the
short exact sequence
1
//
A
i
//
G

//
B
//
1
is said to split if there exists a group homomorphism j : B G such that
j = id
B
.
The importance of the splitting of an extension (that is, a split extension)
is given by the following.
Lemma 3.51. There exists homomorphisms i and making the sequence
1
//
A
i
//
G

//
B
//
1
split if and only if G

= A
B for some : B Aut(A).

Proof: (A reasonably challenging) Exercise.
[Hint for (): consider
b
a = i
1
(j(b)i(a)j(b
1
)).]
Examples 3.52.
S
3
as an extension of C
3
by C
2
splits (3.49(ii)(b))
S
n
as an extension of A
n
by C
2
splits (3.49(iv)(b))
C
4
as an extension of C
2
by C
2
does not split (3.49(i)(b)). So is C
4
more
interesting than S
n
. . . ?
3.4 Composition series and the JordanHolder
theorem
Informally, a series for a group G is a chain of subgroups of G. In general, a
group has many dierent series. However, if we ask for certain special properties
of our series then perhaps we will be lucky and nd that there is a property
that is invariant among all such series. In this rst part, we study composition
series, for which the JordanHolder theorem tells us we are lucky. This theorem
should be thought of as a unique factorization theorem. It means that if we
want to know if every composition series of a group has a specied property
(has Abelian factors, for example), we need only nd one composition series
and check the property for that one, then it will automatically be true for all
others.
Denition 3.53 ([ST, Def. 5.8]). Suppose that G is a group with subgroups
e = G
0
G
1
G
2
G
n
= G
for some n N. We say that such a list of inclusions is a series for G. If G
i
G
for all i, we call this a normal series. If we only have that G
i
G
i+1
for all i,
we call the series a subnormal series. The factors of a subnormal (or normal)
series are the groups G
i+1
/G
i
for 0 i n 1 and the length of the series is
n. (Note that series have nite length.)
Composition series are particular sorts of subnormal series.
Denition 3.54 ([ST, Def. 5.14]). A composition series for a group G is a
subnormal series
e G
1
G
n1
G
n
= G
with G
i
a maximal proper normal subgroup of G
i+1
. That is, for all i, there
does not exist N
i
G
i+1
with G
i
N
i
G
i+1
.
The factors G
i+1
/G
i
are called composition factors.
Note that composition factors are simple: by the third isomorphism theorem, if
N is normal in G then the set of subgroups H such that N H G is in one-
to-one correspondence with the set of subgroups H/N of G/N. Furthermore,
H G if and only if H/N G/N. Therefore N is a maximal normal subgroup
of G if and only if the only proper normal subgroup of G/N is N/N, i.e. is
trivial.
Remark 3.55. A composition series is a particular kind of subnormal series. If
one replaces subnormal with normal in the preceding denition, one obtains
the denition of a chief series. Chief factors (i.e. factors of a chief series) need
not be simple. However one can show that for a group having a composition
series, a chief factor is a direct product of mutually isomorphic simple groups
(i.e. is isomorphic to S S S for some simple group S). For example,
e C
2
C
2
A
4
with factors C
2
C
2
and C
3
is a chief series for A
4
.
Proposition 3.56.
i) Any nite group has a composition series.
ii) The innite cyclic group C
does not have a composition series.

Proof: Exercise.
Examples 3.57.
i) C
12
has several composition series, for example
e C
2
C
4
C
12
with factors C
2
, C
2
and C
3
e C
2
C
6
C
12
with factors C
2
, C
3
and C
2
e C
3
C
6
C
12
with factors C
3
, C
2
and C
2
OLDER THEOREM 31
ii) A
4
has three composition series, all of the form e C
2
C
2
C
2
A
4
with factors C
2
, C
2
and C
3
. (There are three choices of subgroups of
C
2
C
2
isomorphic to C
2
.)
We emphasise again the dierence between a normal and a subnormal
series: in this example, C
2
C
2
C
2
and C
2
C
2
A
4
but C
2
A
4
.
Example (i) is suggestive of a special property of composition series: the set
of composition factors of C
12
is the same for all three series. Another group-
theoretic miracle is that this is true not just for cyclic groups, or even just for
nite Abelian groups but in fact for any nite group. This is the celebrated
JordanHolder theorem.
Theorem 3.58. (JordanHolder theorem) Let G be a nite group. Then all
composition series of G have the same composition factors including multiplic-
ities, up to isomorphism.
Let us introduce some terminology, to help simplify the proof of the theorem.
We will say two composition series of a group G are equivalent if they have the
same composition factors, including multiplicities (up to isomorphism). So we
want to prove that any two composition series for a nite group G are equivalent.
Proof of the JordanHolder theorem: Let
e G
1
G
2
G
r1
G
r
= G (3.1)
and
e H
1
H
2
H
s1
H
s
= G (3.2)
be two composition series of G. We will proceed by induction on r. If r = 1 then
G/e = G is simple and e G is its only composition series. So, let r > 1
and assume that the theorem holds for any group having some composition
series of length less than r.
If G
r1
= H
s1
then G
r1
has two composition series
e G
1
G
2
G
r2
G
r1
and
e H
1
H
2
H
s2
H
s1
= G
r1
,
of length r 1 and s 1 respectively. By the inductive hypothesis, r 1 = s 1
and these two composition series of G
r1
are equivalent. Hence r = s and
G/G
r1

= G/H
s1
so (3.1) and (3.2) are equivalent.
If G
r1
,= H
s1
, since G
r1
G and H
s1
G, we have G
r1
H
s1
G.
But G/G
r1
is simple so we cannot have G
r1
H
s1
, else H
s1
/G
r1
is
a non-trivial proper normal subgroup of G/G
r1
. Therefore we must have
H
s1
< G
r1
H
s1
, but since G/H
s1
is simple, we must have that G
r1
H
s1
is equal to G. Let K = G
r1
H
s1
G. By the second isomorphism theorem,
we have G/G
r1

= H
s1
/K and G/H
s1

= G
r1
/K and so G
r1
/K and
H
s1
/K are simple.
Now K is nite so certainly has a composition series
e K
1
K
2
K
t1
K
t
= K.
Then
e G
1
G
2
G
r2
G
r1
(3.3)
and
e K
1
K
2
K
t1
K G
r1
(3.4)
are composition series of G
r1
of length r 1 and t + 1 respectively. By the
inductive hypothesis, t = r 2 and (3.3) is equivalent to (3.4). Similarly we
have composition series
e H
1
H
2
H
s2
H
s1
(3.5)
and
e K
1
K
2
K
t1
K H
s1
(3.6)
of H
s1
of length s 1 and t +1 = r 1 respectively, so again by the inductive
hypothesis r = s and (3.5) and (3.6) are equivalent.
Finally, since G
r1
/K

= G/H
s1
= G/H
r1
and H
r1
/K

= G/G
r1
, we
see that the composition series
e K
1
K
2
K
t1
K G
r1
G (3.7)
and
e K
1
K
2
K
t1
K H
s1
G (3.8)
are equivalent. So since (3.3) and (3.4) are equivalent, and (3.5) and (3.6) are
equivalent, then (3.1) and (3.7) are equivalent and (3.2) and (3.8) are equivalent.
Now since (3.7) and (3.8) are equivalent, (3.1) and (3.2) are equivalent.
Remarks 3.59. Now that we have proved this theorem, we can talk about the
composition factors of G, rather than of a particular composition series for G.
Jordan saw that the indices [G
i+1
: G
i
[ = [G
i+1
[/[G
i
[ were, up to re-ordering,
the same for any composition series of G. H older, having introduced quotient
groups, observed that these quotients G
i+1
/G
i
were the same, not just their
orders.
The theorem also has direct counterparts for other mathematical structures,
for example modules over rings, where the proof is essentially identical. One can
also consider the result that the cardinality of a nite set is well-dened to be
a JordanHolder type result: elements of a nite set correspond to composition
factors, the cardinality to the number of factors (no multiplicities arise here,
because we are working with sets).
A vector space analogue says that if V is an n-dimensional vector space then
a descending sequence of subspaces V V
, each of dimension
exactly one less than the last, terminates in the zero subspace and has length
n + 1. The factors are all copies of the eld (so here there is one factor, up to
isomorphism, but it occurs with maximum multiplicity).
Remarks 3.60.
i) A collection of simple groups with multiplicities does not determine a
group G up to isomorphism. More precisely, there exist non-isomorphic
groups having the same composition factors including multiplicities.
(Exercise. [Hint: examples may be found even among small nite Abelian
groups.])
OLDER THEOREM 33
ii) Let G be a nite group and let H G. If X
1
, . . . , X
r
are the composition
factors of H and Y
1
, . . . , Y
s
are the composition factors of G/H, then the
composition factors of G are X
1
, . . . , X
r
, Y
1
, . . . Y
s
.
iii) Let X and Y be non-Abelian nite simple groups, and let G = X Y .
Then the only two composition series of G are e X e G and
e e Y G.
Chapter 4
Abelian groups
Now that we have our basic tools, we can begin to really do some group theory.
We start with what might reasonably be called the simplest class of group
although remember that the word simple means something very dierent in
a group-theoretical context from this casual usage! This class we study rst is
that of Abelian groups and we will be able to say a lot about the structure of
Abelian groups.
Those who took B2a Introduction to Representation Theory have already
seen the punchline, namely the classication of nitely generated Abelian groups.
In that course, this was deduced from a very general and powerful theorem. We
will not use that route here but instead show an alternative path towards the
proof, one that is more group-theoretic in spirit and that indicates how one
might apply the result to an actual given Abelian group. Also, some of the
methods we use will be reused or generalized to the non-Abelian setting: one
can consider this chapter as a warm-up for classication results in the general
setting.
4.1 Finite Abelian groups
We already have all the tools we need to classify the nite Abelian groups. By
classify, we mean that we will prove a theorem that says that every nite
Abelian group is isomorphic to one of a list of explicitly described groups. The
list will be innite but this is not surprisingthere is a cyclic group having any
order you care to name and all of these are non-isomorphic. On the other hand,
the explicit description is really very explicit and given an Abelian group, one
can nd out which group on the list it is isomorphic to relatively easily.
Our strategy will be the following:
i) show that a nite Abelian group A has a direct product decomposition
into subgroups having prime-power order, for dierent primes dividing [A[,
then
ii) show that each of these subgroups is a direct product of cyclic groups.
Hence A will be a direct product of cyclic groups. We see here a hint of how
important prime numbers are in nite group theory (a topic that will be the
theme of the next chapter).
34
4.1. FINITE ABELIAN GROUPS 35
We begin with a general denition, even though we will only be interested
in the Abelian case for now.
Denition 4.1. Let G be a group and p a prime. We say G is a p-group if the
order of every element of G is a power of p. That is, for all g G o(g) = p
k
for
some k depending on g.
Clearly a non-trivial group G can be a p-group for at most one p but lots
of groups are not p-groups (e.g. C
2
C
3
, S
3
). A very useful characterisation of
nite Abelian p-groups is the following.
Lemma 4.2. Let A be a non-trivial nite Abelian group and let p be a prime.
Then A is a p-group if and only if A has p-power order, i.e. [A[ = p
r
for some
r 1.
Proof:
() If [A[ = p
r
then A is a p-group, by Lagranges theorem: a A, o(a) [ [A[.
() If A has no proper non-trivial subgroup then A is cyclic of prime order.
This is the base case for an induction on [A[. If A has a non-trivial
proper subgroup B then [B[ and [A/B[ are both strictly less than [A[,
hence by the inductive hypothesis B and A/B have p-power order. Since
[A[ = [B[[A/B[, A has p-power order.
Remark 4.3. This result is true for non-Abelian groups but is much harder to
prove in generality. We will prove it, once we have Cauchys theorem available
to us, in Chapter 5.
Another name for Abelian p-groups is primary groups, and so the following
theorem is often called the Primary Decomposition Theorem for nite Abelian
groups.
Theorem 4.4. Let A be a nite Abelian group and let T = p
1
, . . . , p
n
be the
set of primes occurring as orders of elements of A. Then the elements of A of
p
i
-power order form a subgroup A(p
i
) and
A

= A(p
1
) A(p
2
) A(p
n
).
Proof: We rst prove that A(p
i
) is a subgroup of A, in the usual direct way. Let
p = p
i
(for notational convenience) and let a, b A(p) with o(a) = p
r
, o(b) = p
s
for some r, s. Let t = maxr, s. Then
(ab
1
)
p
t
= a
p
t
(b
1
)
p
t
=
_
a
p
r
_
p
tr
_
_
b
p
s
_
p
ts
_
1
= e.
Hence o(ab
1
) divides p
t
so ab
1
has p-power order. Since A(p
i
) is non-empty
(by denition of T), A(p) A.
Following Remark 3.29, we need to show that A(p
i
) A for all i, that
A = A(p
1
)A(p
2
) A(p
n
) and A(p
i
) A(p
1
) A(p
i1
)A(p
i+1
) A(p
n
) = e
for all i. Since A is Abelian, A(p
i
) is certainly normal for all i. So now we
show that A = A(p
1
)A(p
2
) A(p
n
). We want to show that each a A has the
normal form a = a
1
a
2
a
n
with a
i
A(p
i
).
36 CHAPTER 4. ABELIAN GROUPS
Let a A have o(a) = m and suppose the prime q divides m. Then a
m/q
A
has order q so q T. Therefore m = p
1
1
p
2
2
p
n
n
for some non-negative
integers
i
. By the Euclidean algorithm, there exist integers
1
, . . . ,
n
such
that
1
_
m
p
1
1
_
+ +
n
_
m
p
n
n
_
= 1
because the integers
m
p
1
1
, . . . ,
m
p
n
n
are coprime.
Thus
a = a
1
m
p
1
1
a
2
m
p
2
2
a
n
m
p
n
n
so setting a
i
= a
i
m
p
i
i
, we see that (a
i
)
p
i
i
= a
i
m
= (a
m
)
i
= e. Therefore a
i
has
p
i
-power order, so a
i
A(p
i
), and we have a decomposition of a = a
1
a
2
a
n
as a product of elements a
i
A(p
i
).
Finally, we need that A(p
i
) A(p
1
) A(p
i1
)A(p
i+1
) A(p
n
) = e. But
this is clear: by the denition of the A(p
i
), the only element having p
i
-power
order in the product A(p
1
) A(p
i1
)A(p
i+1
) A(p
n
) is the identity element.
Hence A is the direct product A(p
1
) A(p
n
) as required.
The following corollary also looks forward to Chapter 5.
Corollary 4.5. If the order of a nite Abelian group is divisible by p
but not
p
+1
for some prime p, then A has a subgroup of order p
.
Proof: Using Lemma 4.2, we have in fact shown that [A(p
i
)[ = p
i
i
where
i
is
such that p
i
i
[ [A[ but p
i
+1
i
[A[.
This establishes part (i) of the strategy, the reduction to the case of nite
Abelian p-groups. Now we prove part (ii).
Theorem 4.6. A nite Abelian p-group A is isomorphic to a direct product of
cyclic p-groups.
Proof: Since by Lemma 4.2, [A[ = p
n
for some n, we have a way in for an
inductive proof and this is what we do: we work by induction on n. If n = 0,
there is nothing to prove but this certainly gives a base case for the induction.
Now let n > 0 and let b be an element of maximal order in A (that is
o(b) o(a) a A) and let B = b . We show that B is a direct factor of A,
i.e. there exists a subgroup C A such that A = BC and BC = e, so that
A = B C. (We call C a direct complement for B in A.)
If B = A, then A is cyclic and we are done. Otherwise, [B[, [A/B[ < [A[ and
so by the inductive hypothesis A/B is a direct product of cyclic p-subgroups.
Let the generators of these subgroups be a
1
B, . . . , a
r
B and let their orders (in
A/B) be p
1
, . . . , p
r
(where
i
1). Thus a
p
i
i
B ((a
i
B)
p
i
= a
p
i
i
B = B)
but a
p
i
1
i
/ B, by the denition of order. Hence a
p
i
i
= b
i
for some integer
i
, for each i.
We need to show that p
i
divides
i
, to be able to nd c
i
A having order
p
i
(p
i
was the order of an element of the quotient, only). In order to derive
a contradiction, suppose that
i
=
i
p
i
where p
i
and 0
i
<
i
. Thus
a
p
i
i
= b
i
p
i
for 1 i r. Then the order of a
i
exceeds that of b, as we now
show.
4.1. FINITE ABELIAN GROUPS 37
Let o(a
i
) = p
t
for some t 1. Then since a
p
i
1
i
/ B, a
p
i
1
i
,= e so t >
i
1
and s
def
= t (
i
1) 1. Hence
_
a
p
i
i
_
p
s1
= a
p
i
+s1
i
= a
p
i
+t
i
+11
i
= e
and so
_
b
i
p
i
_
p
s1
=
_
b
i
_
p
i
+s1
= e.
Thus o(b
i
) [ p
i
+s1
. Now (p,
i
) = 1 so o(b
i
) = o(b) (by Proposition 2.25 and
since o(b) [ p
n
), and so o(b) [ p
i
+s1
. But since
i
<
i
,
i
+s1 <
i
+s1 = t
and so o(b) p
i
+s1
< p
t
= o(a
i
). This is a contradiction: o(b) was assumed
maximal. Hence p
i
divides
i
.
Consequently, for each i there is an element b
i
B such that a
p
i
i
= b
p
i
i
(take b
i
= b
i
/p
i
) and hence c
i
def
= a
i
b
1
i
has
c
p
i
i
= a
p
i
i
_
b
p
i
i
_
1
= a
p
i
i
_
a
p
i
i
_
1
= e.
Dene C = c
1
, . . . , c
r
. We will prove that A = B C.
If x A then xB A/B has an expression (a
1
B)
m
1
(a
2
B)
m
2
(a
r
B)
m
r
since the a
i
Bs generate A/B. Thus x = a
m
1
1
a
m
2
2
a
m
r
r
y for some y B and
since a
i
= b
i
c
i
, x = c
m
1
1
c
m
2
2
c
m
r
r
z for z = b
m
1
1
b
m
r
r
y B, so x CB = BC
since A is Abelian. Thus A = BC.
Now suppose x BC. Since x C, x = c
m
1
1
c
m
2
2
c
m
r
r
for suitable m
i
Z,
so xB = c
m
1
1
c
m
r
r
B. Since x B, xB = B and, using c
i
= a
i
b
1
i
, we see that
B = xB = (a
m
1
1
B)(a
m
2
2
B) (a
m
r
r
B). This implies that a
m
i
i
B = B for each i,
since by the inductive hypothesis, A/B is a direct product of cyclic subgroups,
each generated by an a
i
B. Therefore a
m
i
i
B so p
i
divides m
i
. Since c
p
i
i
= e,
then c
m
i
i
= e and hence x = c
m
1
1
c
m
r
r
= e. Therefore B C = e.
We have shown that A = BC with B cyclic. By the inductive hypothesis,
C is a direct product of cyclic p-subgroups, and thus A is too: indeed,
A

= C
p
e
1 C
p
e
2 C
p
e
l
with e
1
e
2
e
l
and
l
i=1
e
i
= n.
The two theorems of this section then yield the classication of nite Abelian
groups.
Theorem 4.7. Every nite Abelian group A is a direct product of cyclic groups.
Proof: Theorem 4.4 tells us that a nite Abelian group is the direct product of
its primary components, each of which is a nite Abelian p-group (for some p)
and thus a direct product of cyclic groups by Theorem 4.6.
This theorem may be improved, to give a recipe for identifying a standard
decomposition of A into cyclic groups.
Theorem 4.8 ([ST, Thm. 2.44]). Let A be a nite Abelian group. Then
A

= C
d
1
C
d
2
C
d
t
such that 1 < d
i
[ d
i+1
for 1 i < t and t and the sequence (d
i
) are uniquely
determined by A.
38 CHAPTER 4. ABELIAN GROUPS
Proof: Omitted, though the results we have proved give strong pointers as to
how to obtain (d
i
).
This standard decomposition might not be exactly what you expected: C
2
C
3
would have the associated data t = 1, d
1
= 6 (corresponding to its isomorphic
form C
6
), rather than t = 2, (d
1
, d
2
) = (2, 3), since 2 3.
Example 4.9. If A is Abelian of order 8 then A is isomorphic to exactly one of
C
8
, C
2
C
4
and C
2
C
2
C
2
.
Exercise 4.1. Classify the nite Abelian groups having order 100, order 128 and
order 74,088.
Remark 4.10. Since Abelian groups and Z-modules are the same thing, we have
classied the nite Z-modules (i.e. nite as sets!). More substantially, in the
next section we see the classication of the nitely generated Z-modules.
4.2 Finitely generated Abelian groups
In this section, we will extend the classication of nite Abelian groups to that
of the nitely generated Abelian groups. Since nitely generated Abelian groups
need not be niteC
being the rst obvious exampleit is no surprise that

we need some innite groups to use in the classication. What may be surprising
is that we only need this innite cyclic group, C
. No other innite Abelian

groups are needed. Or to put it another way, nitely generated Abelian groups
are mostly just lots of copies of Z. The same is certainly not true of innitely
generated Abelian groups in general, though.
The proofs of any results in this section are non-examinable. The statements
are examinable. All proofs may be found in [ST, Section 2.8] or other texts.
We begin with a result bounding the number of generators of a subgroup of a
nitely generated Abelian group. Indeed the bound is a strong one. The result
is even more remarkable when one considers that there are groups, such as that
mentioned in Proposition 3.9(vi), which are 2-generator groups with innitely
generated subgroups.
Proposition 4.11 ([ST, Prop. 2.45]). Suppose that A = a
1
, . . . , a
n
is a nitely
generated Abelian group and that B is a subgroup of A. Then B can be generated
by n elements (or fewer).
Proof: Unsurprisingly, we work by induction on n. If n = 0, A = B = e
so there is nothing to do. Assume n 1 and let C = a
1
, . . . , a
n1
A
(normal since A is Abelian). Then the subgroup B C C can be generated
by n 1 elements by the inductive hypothesis applied to C and by the second
isomorphism theorem B/B C

= BC/C, the latter being a subgroup of A/C.
Since A is Abelian, elements of A have the normal form a
1
1
a
2
2
a
n
n
with
i
Z. Each element a
1
1
a
2
2
a
n1
n1
C so every right coset of C in A is of
the form Ca
n
n
. Moreover, as
n
varies in Z, each Ca
n
n
is a right coset of C in
A. Hence A/C = Ca
n
is cyclic. We have shown that BC/C A/C = Ca
n
so BC/C is also cyclic. Hence B/B C is cyclic and of the form (B C)y
for some y A. Then every element of B is an element of B C multiplied on
4.2. FINITELY GENERATED ABELIAN GROUPS 39
the right by a power of y. So adjoin y to a generating set of B C, which has
size at most n 1, to obtain a generating set of size at most n for B.
This proposition will in particular tell us about a very special subgroup of
a nitely generated Abelian group, one that is very natural to consider in the
context of classication. In an innite but nitely generated Abelian group,
while some element must have innite order, there may still be other elements
of nite order. These should relate to a nite Abelian groupand indeed they
doand getting rid of them should leave a simpler structure. The following
denition and proposition make this precise.
Denition 4.12 ([ST, p. 80]). Let A be an Abelian group. Dene the torsion
subgroup of A to be the subgroup (A) consisting of all elements of A of nite
order. If (A) = e we say A is torsion-free.
Remark 4.13. You should verify the implicit claim.
If A is a nitely generated Abelian group then by the previous proposition
(A) must also be nitely generated. But then (A) must in fact be nite (think
about normal forms) and hence is subject to the classication of the previous
section. So, we understand (A) if A is a nitely generated Abelian group: can
we understand A/(A) and hence A? The answer is yes.
Proposition 4.14 ([ST, Prop. 2.46]). Let A be a nitely generated Abelian
group. Then A/(A) is a torsion-free nitely generated Abelian group.
Proof: Exercise.
Theorem 4.15 ([ST, Thm. 2.47]). Let A be a torsion-free nitely generated
Abelian group. Then there exists n N such that
A

=
n copies
..
C
def
= C
n
.
Also if B is a subgroup of nite index in A then B

= A.
Proof: Omitted.
Exercise 4.2. Prove that C
m
= C
n
if and only if m = n.
Now we understand both (A) and A/(A). Normally this would not quite
be sucient to tell us the whole structure of A but in this special situation, it is
and one can prove the following, the main theorem classifying nitely generated
Abelian groups.
Theorem 4.16 ([ST, Thm. 2.50]). Let A be a nitely generated Abelian group.
Then
A

= C
d
1
C
d
2
C
d
t
C
m
such that 1 < d

i
[ d
i+1
for 1 i < t and t, m and the sequence (d
i
) are uniquely
determined by A.
Proof: Omitted.
Remark 4.17. This statement includes the classication of nite Abelian groups
(m = 0) but relies on that classication to understand (A).
Remark 4.18. The classication of nite non-Abelian groups is hopeless, let
alone that of nitely generated non-Abelian groups. Even the classication of
simple nite groups is extraordinarily hardbut it was possible.
Chapter 5
Prime numbers in nite
group theory
We now turn to looking at how much information about the structure of a nite
group we can wring out of the prime decomposition of its order. The answer is a
surprisingly large amount! One can nd out about orders of elements (Cauchys
theorem), and hence cyclic subgroups, but also about certain non-cyclic sub-
groups having prime power order (Sylows theorems). This information can be
used to prove results about properties of groups, particularly about whether or
not they are simple.
5.1 Cauchys theorem
Consider a nite group G and an element g G. Recall that the order of g
is o(g) = [ g [, the order of the subgroup of G that g generates. Then by
Lagranges theorem, o(g) divides [G[. However, if n divides [G[, this does not
mean that there is an element of order n: for example in S
3
there is no element
of order [S
3
[ = 6. You might speculate that perhaps the problem here is that
6 is not prime, noticing that there certainly are elements of orders 2 and 3. It
would be a wild speculation with such a small amount of evidence but nothing
ventured, nothing gained. In fact, this would be a great guess at the following
theorem, due to Cauchy. This theorem acts as a partial converse to Lagranges
theorem.
Theorem 5.1 ([NST, Thm. 8.1]). (Cauchys theorem) Let G be a nite group
and p a prime dividing [G[. Then there exists an element of G of order p.
Proof: It is natural to work by induction on [G[, so we shall. If [G[ = 1, so
G = e, there is nothing to prove so let [G[ > 1 and p a prime divisor of [G[.
First let us tighten up our inductive hypothesis somewhat, as follows: suppose
that any group G
having order [G
[ < [G[ also divisible by p does contain an

element of order p.
Then if G has a proper subgroup H such that p divides [H[, then by the
inductive hypothesis there is an element a H of order p. But a H G so
a is an element of order p in G, and we are done. So now suppose that for any
proper subgroup H of G, p does not divide [H[.
40
5.2. FINITE P-GROUPS 41
In any group, we have the distinguished subgroup Z(G), the centre of G,
dened to be the set a G [ ag = ga for all g G. The elements of Z(G)
are precisely those having conjugacy class of size 1: a
g
= a for all g G.
Assume Z(G) ,= G and let h GZ(G). Then the centralizer of the element h,
C
G
(h)
def
= g G [ hg = gh, is not all of G, so p [C
G
(h)[ by our assumption on
proper subgroups. Hence p must divide [G[/[C
G
(h)[ = [G : C
G
(h)[, the latter
being the size of the conjugacy class containing h. So if a conjugacy class is not
of size 1, its size is a multiple of p. But G is the disjoint union of its conjugacy
classes (realised as orbits under the conjugation action) and we have the class
equation
[G[ =
|Z(G)| times
..
1 + 1 + + 1 +[(
|Z(G)|+1
[ +[(
|Z(G)|+2
[ + +[(
r
[
where (
i
([Z(G)[ + 1 i r) are conjugacy classes of size at least 2. We have
seen that p divides [(
i
[ for these (
i
, [Z(G)[ + 1 i r, and of course p divides
[G[. We therefore deduce that p divides [Z(G)[. Since we have supposed that p
does not divide the order of any proper subgroup of G, Z(G) cannot be proper,
i.e. Z(G) = G and so G is Abelian.
This is a major reduction: we have shown that either the problem reduces
to a subgroup or G is Abelian. However, Cauchys theorem is true for Abelian
groups. We could try to use the classication but this is a bit excessive so we
give a direct proof here. Let a ,= e be an element of G (now assumed Abelian),
and let m = o(a). If p [ m then m = pm
0
and a
m
0
has order p. If p m,
let A = a and

G = G/A. We must then have p [ [
G[ but [
G[ < [G[ so
by induction there is an element

b

G of order p. Now

G is not a subgroup
of G so we cannot simply use

b but we can lift it to G:

b = Ab for some
b G A and we have

b
p
= (Ab)
p
= Ab
p
= A (the trivial element in G/A =

G)
so b
p
A. Hence o(b) is divisible by p (since A is cyclic) so letting k = o(b)/p,
b
k
is an element of order p in G, as required. So the Abelian case is done and
the reduction to subgroups is given by induction.
Remark 5.2. An alternative proof of Cauchys theorem may be found in [NST,
p. 84], a reproduction of a proof by J.H. McKay from 1959. The proof we have
given follows an argument originating with Frobenius, from around 1886. It
is rather shorter than Cauchys 100 page original proof! McKays makes even
more use of group actions and has the advantages of being direct (i.e. not by
induction) and not requiring restriction to a special case (G Abelian is treated
equally with G non-Abelian).
5.2 Finite p-groups
Let p be a prime. Associated to p is a class of groups called p-groups, dened
as follows.
Denition 5.3. Let G be a group and p a prime. We say G is a p-group if the
order of every element of G is a power of p. That is, for all g G o(g) = p
k
for
some k depending on g.
Clearly G can be a p-group for at most one p but lots of groups are not
p-groups (e.g. S
3
). Next we have the following important corollary to Cauchys
theorem.
42 CHAPTER 5. PRIME NUMBERS IN FINITE GROUP THEORY
Corollary 5.4. A nite group G is a p-group for a prime p if and only if [G[
is a power of p.
Proof: Assume G is a nite p-group and [G[ is not a power of p. Then there
exists q ,= p, q prime, such that q [ [G[. Then by Cauchys theorem, there exists
an element of G of order q: contradiction with G being a p-group. Conversely,
if [G[ is a power of p then for all g G, o(g) [ [G[ (Lagrange) so o(g) is a power
of p and G is a p-group.
We note some elementary properties of p-groups.
Proposition 5.5 ([ST, Prop. 3.38]). Let G be a non-trivial nite p-group. Then
the centre of G, Z(G), is non-trivial.
Proof: The method was set out in the course of the proof of Cauchys theorem.
In a p-group, for g G Z(G) we have [C
G
(g)[ and hence [G : C
G
(g)[ being
powers of p, since [G[ is. Then use the class equation to see that [Z(G)[ p (it
is non-zero, certainly) so Z(G) is non-trivial.
Corollary 5.6 ([ST, Cor. 3.39]). If G is a group of order p
2
then G is Abelian.
Proof: Exercise.
Recall that if H is a subgroup of G we dene the normalizer of H to be
N
G
(H) = g G [ Hg = gH (so H is normal in its normalizer). The following
proposition is useful, relating to normalizers in p-groups.
Proposition 5.7 ([ST, Prop. 3.40]). Let G be a nite p-group and suppose that
H is a (possibly trivial) proper subgroup of G. Then H is properly contained in
its normalizer N
G
(H).
Proof: Exercise.
If G is any nite group, some of its subgroups may be p-groupscall these
p-subgroups. Certain p-subgroups are particularly important, so much so that
their denition and the key theorems about them deserve a whole new section.
5.3 Sylows theorems
Very often, we want to focus our attention on the largest objects of a particular
sort. However one must be careful exactly what one means by largest. In
algebra, the correct idea is usually that of maximality and we make a precise
(meta-)denition for subgroups of groups now.
Denition 5.8. Let G be a group and P be a property
1
of groups. Then we
say that H G is a maximal subgroup of G with property P (or a maximal P
subgroup) if H is maximal in the class of all subgroups of G having the property
P, that is, for every subgroup K with property P we have H K G implies
K = H.
1
For those concerned about exactly what constitutes a property of a group we take the
following denition: P is a property of a group G if P is an attribute of G and for any group
H isomorphic to G, H has the attribute P.
5.3. SYLOWS THEOREMS 43
Remark 5.9. If G has property P then G itself is the unique maximal P subgroup
of G. If G does not have property P then any maximal P subgroup is proper.
It is not true in general that maximal P subgroups are unique. For example
if G = C
2
C
2
and P is the property is a proper subgroup of G, then in
C
2
C
2
= (e, e), (a, e), (e, b), (a, b), the two subgroups H
1
= (e, e), (a, e)
and H
2
= (e, e), (e, b) have property P (i.e. are proper). They can easily seen
to be non-trivial proper subgroups of C
2
C
2
and hence are maximal proper
subgroups but H
1
,= H
2
.
Our interest now moves to P = is a p-group for a prime p, and we use the
terminology maximal p-subgroup for maximal is a p-group subgroup.
Denition 5.10. Let p be a prime and G a group. Then we call a subgroup
H of G a Sylow p-subgroup if H is a maximal p-subgroup of G, i.e. if K is a
p-subgroup of G and H K then H = K. Denote by Syl
p
(G) the set of Sylow
p-subgroups of G.
Remark 5.11. This denition does not require G or H to be nite. Sylows
original paper of 1872 treated p-subgroups of greatest order in a nite group G,
so our denition is the natural generalization of this. However it diers from
the denition given in many books and our theorems about Sylow p-subgroups
will correspondingly be slightly dierent.
Remark 5.12. This denition guarantees the existence of Sylow p-subgroups for
any nite group G and prime p. For innite groups, existence relies on Zorns
lemma (or equivalently, the axiom of choice).
We concentrate on Sylow p-subgroups of nite groups: their key properties
are established in Sylows theorems, which we shall present as a single theorem.
Theorem 5.13. (Sylows theorems) Let G be a nite group and p a prime. We
may write uniquely [G[ = p
a
m with a 0, m N and p m. Then
i) every Sylow p-subgroup of G has order p
a
;
ii) if P
1
, P
2
Syl
p
(G) then there exists g G such that P
2
= P
g
1
= g
1
P
1
g;
iii) the number of Sylow p-subgroups [Syl
p
(G)[ is of the form kp +1 for some
k Z, k 0. Furthermore, [Syl
p
(G)[ divides m and hence [Syl
p
(G)[
divides [G[.
Remark 5.14. Thus any p-subgroup of G is contained in a subgroup of order p
a
,
and any subgroup of order p
a
is a Sylow p-subgroup.
Also, Syl
p
(G) is an orbit for the action of G on its set of subgroups by
conjugation. Therefore any two Sylow p-subgroups are isomorphic.
Finally, if P Syl
p
(G) then [Syl
p
(G)[ = [G : N
G
(P)[ 1 (mod p).
Remark 5.15. We comment on the case p [G[, so [G[ = p
0
[G[. Then every
Sylow p-subgroup has order 1, i.e. Syl
p
(G) =
_
e
_
, all of whose members
are certainly conjugate and isomorphic, and [Syl
p
(G)[ = 0p + 1 divides [G[.
Furthermore, e Syl
p
(G) if and only if p [G[.
We return to the general case and the proof of Sylows theorems.
Proof: Let P Syl
p
(G) so that [P[ = p
b
for some b and there does not exist a
subgroup Q of G with P Q and [Q[ = p
c
for c > b.
Let H = N
G
(P) = g G [ g
1
Pg = P, and let m
0
= [H : P[ = [H/P[.
If p divides m
0
then by Cauchys theorem there exists a subgroup

A H/P of
order p. But by the third isomorphism theorem, then there exists A H such
that

A = A/P and P A H. Now [A[ = [A : P[[P[ = [
A[[P[ = p
b+1
but this
contradicts the maximality of P. So p cannot divide m
0
and [H[ = p
b
m
0
with
p m
0
.
Now P is a p-subgroup of G and hence of H, so that every element of P
has order a power of p. In fact, P must be exactly the elements of H having
order a power of p, as any such element in H but not in P would give rise to a
subgroup of H/P with order divisible by p, but p [H/P[ = m
0
. Therefore if Q
is a p-subgroup of H we must have Q P.
Now let = cos(G : H) = Hg [ g G with the action of G on by right
multiplication, (Hg) x
def
= Hgx. Let m
1
= [[ = [G : H[. Denote by the coset
He = H cos(G : H) so that
Stab() = g G [ g = = g G [ Hg = H = H.
Consider the action of P on . Since P H, P xes , i.e. there is at least
one P-orbit of length 1, namely . Let be any other P-orbit of length 1.
Then since the action of any group on one of its coset spaces is transitive, there
exists x G such that = x. Thus P Stab() = x
1
Stab()x = x
1
Hx.
So, xPx
1
H. Now for all xgx
1
G with o(g) a power of p, p
t
say,
(xgx
1
)
p
t
= xg
p
t
x
1
= e so xPx
1
is a p-subgroup of H and as discussed
above we therefore have xPx
1
P. Then xPx
1
must be equal to P (they
have the same order) and so x normalizes P, i.e. x N
G
(P) = H. But then
x = H x = H so = x = , and we have shown that there is a unique
P-orbit of length 1, namely , for the action of P on .
Recall Lemma 2.18 from Chapter 2.
Lemma. Let G be a nite group having order [G[ = p
r
for some prime p and
suppose G acts on a set . Dene Fix
G
() = [ g = g G (the
set of global xed points for G acting on ). Then [Fix
G
()[ [[ (mod p).
Using this lemma for the p-group P acting on , we see that
[[ = m
1
[Fix
P
()[ 1 (mod p).
But then
p
a
m = [G[ = [H[[G : H[ = [H[m
1
= [P[m
0
m
1
= p
b
m
0
m
1
.
Furthermore p m
0
m
1
so we must have a = b and m = m
0
m
1
. This proves (i).
For (ii), we show the stronger claim that for any P Syl
p
(G) and any
(not necessarily maximal) p-subgroup Q of G, there exists an element x G
such that Q x
1
Px. For if Q = P
is in fact maximal (i.e. Sylow) then

P
= Q = x
1
Px, hence (ii).
We consider the action of Q by right multiplication on = cos(G : H)
with H = N
G
(P). By the above lemma again, [[ [Fix
Q
()[ (mod p) and
[Fix
Q
()[ is the number of Q-orbits of size 1. But p [G : H[ = [[ (P is Sylow
so of maximal p-power order) so [Fix
Q
()[ , 0 (mod p) and there exists at
least one Q-orbit of size 1, say. The action of G on cos(G : H) is transitive
so as before there exists x G such that x = H (H cos(G : H)). Thus
Stab() = x
1
Hx, also as before. Since is a Q-orbit of size 1, we see that
Q Stab() = x
1
Hx so xQx
1
is a p-subgroup of H (and therefore of P) so
Q x
1
Px as required. This proves our claim, from which we deduced (ii).
Finally, for (iii), we now know that Syl
p
(G) is the set of all subgroups of G of
order p
a
and all of these subgroups are conjugate. But the conjugates of a given
Sylow p-subgroup P are in one-to-one correspondence with the cosets of the
normalizer of P in G, N
G
(P) = H. By the above, [G : H[ = [cos(G : H)[ 1
(mod p) and [cos(G : H)[ (= m
1
) divides [G[ = p
a
m
0
m
1
. This proves (iii) and
so the theorem.
Remark 5.16. This proof follows [NST, p. 85]; the same authors give an alter-
native proof ([NST, p. 86]), due originally to Wielandt.
Examples 5.17.
i) the Sylow subgroups of S
4
[S
4
[ = 4! = 2
3
3 hence there are non-trivial Sylow p-subgroups for
p = 2 and 3.
Sylow 3-subgroups:
any Sylow 3-subgroup has order 3, hence is cyclic.
the cyclic subgroups of S
4
of order 3 are
P
1
= (1 2 3) ,
P
2
= (1 3 4) ,
P
3
= (1 2 4) ,
P
4
= (2 3 4)
and
P
2
= (2 4)P
1
(2 4),
P
3
= (3 4)P
1
(3 4),
P
4
= (1 4)P
1
(1 4).
[Syl
3
(S
4
)[ = 4 = 1 3 + 1 1 (mod 3) and [Syl
3
(S
4
)[ = 4 divides
2
3
= [S
4
[/3.
Sylow 2-subgroups:
any Sylow 2-subgroup has order 8. There is no element of order 8 in
S
4
, so Sylow 2-subgroups are not cyclic.
by Sylows theorems, [Syl
2
(S
4
)[ 1 (mod 2), i.e. [Syl
2
(S
4
)[ is odd,
and [Syl
2
(S
4
)[ divides [S
4
[/2
3
= 3 so we must have [Syl
2
(S
4
)[ = 1 or
3.
there are three subgroups
Q
1
= (1 2 3 4), (1 3) ,
Q
2
= (1 2 4 3), (1 4) ,
Q
3
= (1 3 2 4), (1 2)
of order 8. Hence these are the Sylow 2-subgroups; they are all
conjugate.
ii) the Sylow subgroups of A
5
[A
5
[ = [S
5
[/2 = 5!/2 = 60 = 2
2
3 5 hence there are non-trivial
Sylow p-subgroups for p = 2, 3 and 5.
Sylow 5-subgroups:
are cyclic of order 5.
[Syl
5
(A
5
)[ 1 (mod 5) and divides 12, hence [Syl
5
(A
5
)[ is 1 or 6.
there are 4 3 2 = 24 5-cycles in S
5
(all even, hence in A
5
) giving
rise to 6 subgroups of A
5
of order 5, so [Syl
5
(A
5
)[ = 6.
one is (1 2 3 4 5) ; the others are all conjugate to this one.
Sylow 3-subgroups:
are cyclic of order 3.
[Syl
3
(A
5
)[ 1 (mod 3) and divides 20, hence [Syl
5
(A
5
)[ is 1, 4 or 10.
there are 20 3-cycles in A
5
giving rise to 10 subgroups of order 3, so
[Syl
3
(A
5
)[ = 10.
one is (1 2 3) ; the others are all conjugate to this one.
Sylow 2-subgroups:
are of order 4 and, since A
5
has no elements of order 4, are not cyclic,
so must be isomorphic to C
2
C
2
.
[Syl
2
(A
5
)[ 1 (mod 2), and divides 15 so [Syl
2
(A
5
)[ is 1, 3, 5 or 15.
A
5
has 15 elements of order 2, these being the (2, 2)-cycles (2-cycles
are odd, so not in A
5
), giving rise to 5 subgroups of order 4 (deter-
mined by the xed point), so [Syl
2
(A
5
)[ = 5.
one of these is , (1)(2 3)(4 5), (1)(2 4)(3 5), (1)(2 5)(3 4); all others
are conjugate to this one.
The following consequence of Sylows theorems is very useful, transmitting
information about Sylow p-subgroups of a nite group G to normal subgroups
and quotients of G.
Proposition 5.18 ([ST, Prop. 3.45]). Let G be a nite group and let N G.
Choose P Syl
p
(G) for some prime p. Then PN/N Syl
p
(G/N) and NP
Syl
p
(N).
Proof: P is a Sylow p-subgroup so [G : P[ = [G[/[P[ is coprime to p and
hence so is [G : PN[, since [G : PN[[PN : P[ = [G : P[. By the second
isomorphism theorem, PN/N

= P/P N which has order a power of p. Now
[G/N : PN/N[ = [G : PN[ by the third isomorphism theorem, so PN/N has
p-power order but index in G/N coprime to p. Hence PN/N Syl
p
(G/N) by
part (i) of Sylows theorems.
Certainly P N has p-power order since P N P but we also have
[PN[ =
|P||N|
|PN|
= [P[[N : P N[. If p divides [N : P N[ then p[P[ divides [PN[
which itself divides [G[, but P Syl
p
(G) so this cannot happen. Therefore PN
has p-power order but index in N coprime to p, hence P N Syl
p
(N).
Two useful classication results are the following, which are typical of results
making use of Sylows theorems.
Proposition 5.19 (PlanetMath.org
2
). Let G be a group of order pq with p, q
prime and p > q. Then G is a semi-direct product of C
p
by C
q
and moreover
i) if q p 1 then G

= C
pq
ii) if q [ p 1 then G

= C
pq
or G

= C
p
C
q
where C
p
C
q
is the semi-direct
product of C
p
by C
q
with action
b
a = a
r
, where b = C
q
, a = C
p
and
r N, 1 < r < p such that r
q
1 (mod p).
In particular, up to isomorphism there are at most two groups of order pq.
Proof: By Cauchys theorem, G has subgroups of orders p and q. Consider
the subgroup of order p, i.e. for the larger prime. If P
1
and P
2
were distinct
subgroups of order p then their intersection would be trivial, so P
1
P
2
would
contain p
2
elements (by Lemma 3.24). This cannot happen, since [G[ = pq < p
2
.
Hence there is a unique subgroup of order p, which must therefore be normal.
Denoting the p-subgroup by P and letting Q be any q-subgroup, P Q = e
and PQ = G so G is a semi-direct product of P by Q.
(The remainder of this proof is not examinable.)
Observe that P

= C
p
and Q

= C
q
. If there is only one q-subgroup Q, then
Q is also normal and G is a direct product of P and Q and is therefore cyclic
of order pq.
Given G = P Q (possibly a direct product), it remains to determine the
action of Q on P by conjugation. We have two cases:
i) If q p 1 then since since any q-subgroup is a Sylow q-subgroup (q is the
highest power of q dividing pq), [Syl
q
(G)[ = 1 + mq ,= p so we have only
one q-subgroup and G = C
p
C
q

= C
pq
.
ii) If q [ p 1 then Aut(P) = Aut(C
p
)

= C
p1
has a unique subgroup Q
of order q, where Q
=
s
: P P, x x
s
[ s Z/pZ and s
q
= 1.
Let P = a , Q = b and suppose b acting on P by left conjugation
is by
r
, r Z/pZ, r ,= 1. Then G = a, b is as described in (ii).
Since Q
is unique, any action of Q on P by left conjugation is an injective

homomorphism with image Q
. Thus by Proposition 3.42, all the resulting

semi-direct products of P by Q are isomorphic. (Choosing a dierent r
amounts to choosing a dierent generator b for Q, so does not change the
isomorphism class.)
We will not prove that Aut(C
p
)

= C
p1
: this may be found in [AB, Proposi-
tion 2.2, p. 15], for example.
Proposition 5.20 ([Ros, Thm. 5.20]). Let G be a group of order pqr with p, q
and r primes and p > q > r. Then G is not simple.
2
http://planetmath.org/encyclopedia/ApplicationsOfSylowsTheoremsToGroupsOfOrderPq.
html
Proof: Suppose for a contradiction that there is a simple group G of order pqr
with p, q and r primes and p > q > r. Let n
p
, n
q
and n
r
be the numbers
of Sylow p-, q- and r-subgroups of G, respectively. Since G is simple, each
of these numbers is strictly larger than 1. For if, without loss of generality,
n
p
were equal to 1, then the unique Sylow p-subgroup P would be a proper
non-trivial normal subgroup of G (normal since conjugation preserves the set of
Sylow p-subgroups), contradicting the simplicity of G.
Since all Sylow p-subgroups of G are cyclic of order p, any two distinct
Sylow p-subgroups intersect trivially. Hence the n
p
Sylow p-subgroups contain
n
p
(p 1) distinct elements of order p. Similarly, the n
q
Sylow q-subgroups
contain n
q
(q 1) distinct elements of order q and the n
r
Sylow r-subgroups
contain n
r
(r 1) distinct elements of order r.
Therefore
[G[ = pqr 1 +n
p
(p 1) +n
q
(q 1) +n
r
(r 1).
By Sylows theorems, n
p
divides qr and n
p
1 (mod p). Since n
p
> 1 and
p > q > r, it follows that n
p
= qr. By similar arguments, n
q
p and n
r
q so
that we have
pqr 1 +qr(p 1) +p(q 1) +q(r 1).
But then 0 (p 1)(q 1) which is plainly false. So there is no simple group
having order of the form stated above.
Chapter 6
Soluble groups
Thus far, we have been collecting tools for use towards our ultimate aim, under-
standing simple groups. In this section, we do something that looks slightly odd:
we study a class of groups whose members are almost never simple. These are
the soluble groups and one should think of solubility as diametrically opposed
to simplicity. We will see that soluble groups have lots of normal subgroups; of
course, simple groups should have essentially none.
6.1 Polycyclic groups
Before looking at soluble groups, we consider another meta-denition, similar
to that at the start of our section on Sylows theorems, and a resulting class of
groups, polycyclic groups.
Denition 6.1 ([ST, p. 165]). Let G be a group and P a property of groups.
We say G is a poly-P group if G has a subnormal series whose factors all have
the property P.
Examples 6.2.
i) we have just been studying poly-simple groups, these being the groups
having a composition series.
ii) a group is poly-nite if and only if it is nite.
Remark 6.3. Since e G is a subnormal series for any group, if G has the
property P then G is certainly a poly-P group. However, in general the class of
poly-P groups is strictly bigger than the class of P-groups, as we see now.
Examples 6.2. (continued)
iii) the trivial group is poly-innitethe subnormal series of e with no terms
has no factors!
iv) S
3
is poly-cyclic but not cyclic (from now, polycyclic).
v) S
3
is poly-Abelian but not Abelian (from now, polyabelian).
49
50 CHAPTER 6. SOLUBLE GROUPS
The last two classes, polycyclic and polyabelian groups, merit further study.
Denition 6.3 ([ST, Def. 5.10]). A group G is said to be soluble if it is poly-
abelian, that is, G has a subnormal series with Abelian factors. (American
usage is solvable, rather than solublethe concept is the same, however.)
We will give some equivalent characterizations of solubility later on but we
rst study polycyclic groups. The rationale for this is that since cyclic groups
are the most straightforward sort of Abelian groups, we ought to be able to say
more about polycyclic groups. Several important classes of groups are polycyclic
but we will not get anything remotely as comprehensive as our understanding
of cyclic groups, or even nite Abelian groups.
However polycyclic groups do enjoy one remarkable property: the class of
polycyclic groups is closed under taking subgroups, quotients and nite direct
products. This property is very useful, as it helps us construct lots of exam-
ples and also argue by induction in various ways. We prove this property for
polycyclic groups.
Proposition 6.4 ([ST, Prop. 5.42]). Let G be a polycyclic group.
i) If H G then H is polycyclic.
ii) If N G then G/N is polycyclic.
iii) If H is polycyclic then GH is polycyclic.
Proof: Throughout, let e G
1
G
n1
G
n
= G be a subnormal
series with cyclic factors.
i) Let H
i
= G
i
H, for 1 i n. Consider the maps j : H
i+1
G
i+1
(the natural inclusion) and : G
i+1
G
i+1
/G
i
. Let = j be the
composition, so : H
i+1
G
i+1
/G
i
. We have Ker H
i+1
and we see
that in fact Ker = H
i
so H
i
H
i+1
. Also, the second isomorphism the-
orem yields H
i+1
/H
i

= H
i+1
G
i
/G
i
. But H
i+1
G
i
/G
i+1
G
i+1
/G
i
which
is cyclic: all subgroups of a cyclic group are cyclic (Proposition 2.24(i)) so
H
i+1
/H
i
is cyclic for all i. Hence e H
1
H
n1
H
n
= H is a
subnormal series for H with cyclic factors, so H is polycyclic.
ii) First, we show that G
i
N is a normal subgroup of G
i+1
N for all i. Let
x G
i
, y G
i+1
and m, n N. Then
(xm)
yn
= (x
y
)
n
m
yn
= x
y
[ x
y
, n]m
yn
G
i
N
where a
b
def
= b
1
ab (the conjugation of a by b) and [ a, b ]
def
= a
1
b
1
ab (the
commutator of a and b). (NB. Verify this equation.) The reasoning that
the expression x[ x, y ][ x
y
, n]m
yn
is in G
i
N is as follows:
(a) x
y
G
i
since G
i
G
i+1
(b) [ x
y
, n] = (n
1
)
x
y
n and N G so (n
1
)
x
y
N, hence [ x
y
, n] N
(c) N G so m
yn
N
6.2. SOLUBLE GROUPS 51
So (xm)
yn
G
i
N so G
i
N G
i+1
N.
Now we have a natural inclusion j : G
i+1
G
i+1
N and a surjection
: G
i+1
N G
i+1
N/G
i
N. The composite = j is again surjective
(: G
i+1
G
i+1
N/G
i
N) and G
i
Ker . By the third isomorphism
theorem, we have an induced surjection : G
i+1
/G
i
G
i+1
N/G
i
N and
in fact : xG
i
xG
i
N. Now G
i+1
/G
i
is cyclic and so Im is cyclic, i.e.
G
i+1
N/G
i
N is cyclic. Then the series
e = eN/N G
1
N/N G
n1
N/N G
n
N/N = G/N
is a subnormal series for G/N and (G
i+1
N/N)/(G
i
N/N)

= G
i+1
N/G
i
N
is cyclic. Hence G/N is polycyclic.
iii) Note that e H G H and e H

= H is polycyclic. Also
G H/e H

= G is polycyclic. Hence there is a subnormal series of
GH whose factors are cyclic. (Compare with the remark 3.60(ii) after
the JordanHolder theorem.)
Exercise 6.1. Construct a subnormal series of S
3
S
3
that demonstrates that
this group is polycyclic.
Of course, a proposition such as this is only of any use as a machine for
producing examples if we have some to feed in in the rst place. At the moment,
we have only identied S
3
as polycyclic but a little thought will make us realise
that we already know a large class of examples:
Proposition 6.5 ([ST, Prop. 5.41]). Every nitely generated Abelian group is
polycyclic.
Proof: This is immediate from the classication and the preceding proposition,
as any nitely generated Abelian group is a nite direct product of cyclic groups
and these are polycyclic. For a direct proof, see [ST, Prop. 5.41, p. 184].
6.2 Soluble groups
Now we turn to soluble groups: we dened these to be the polyabelian groups,
so the factors of a subnormal series are now assumed only to be Abelian, not
specically cyclic. Some alternative characterizations are useful and we give
these now. First we construct another type of series associated to a group, the
derived series, and commutators make a reappearance.
Denition 6.6. Let G be a group. The commutator subgroup (or derived
group) of G is the subgroup G
= [ G, G]
def
= [ a, b ] [ a, b G.
Some elementary properties of commutator subgroups are useful:
Lemma 6.7 ([AB, Prop. 6, p. 18]). Let G be a group.
i) If H G and G/H is Abelian then G
H.
ii) Conversely, if G
H G then H G and G/H is Abelian.

iii) If H G then H
G.
Proof: Exercise. [Hint: observe that G is Abelian if (and only if ) G
= e.]
The derived series arises out of repeated taking of commutator subgroups.
Denition 6.8. Let G be a group. Set G
(0)
= G and for k 1, dene
G
(k)
def
= (G
(k1)
)
. The series G = G
(0)
G
(1)
(= G
) G
(2)
is called
the derived series of G. By the preceding lemma, each G
(k)
is normal in G
(including G
G) and G
(k1)
/G
(k)
is Abelian so the derived series is a normal
series with Abelian factors.
Technically, our denition of series stipulated that the series had nite
length, i.e. terminated in e after nitely many steps, but I hope you will agree
that this extension of our denition to non-terminating descending sequences of
subgroups is reasonable. We can now prove the following theorem.
Theorem 6.9 ([AB, Prop. 2, p. 96]). Let G be a group. Then the following are
equivalent:
i) there exists k N such that G
(k)
= e;
ii) G has a normal series with Abelian factors;
iii) G has a subnormal series with Abelian factors, i.e. G is polyabelian, i.e.
G is soluble.
Proof: We have (i) (ii) (iii) by our observation that the derived series
is a normal series with Abelian factors. Thus it remains to prove (iii) (i).
Suppose G = G
0
G
1
G
r
= e is a subnormal series for G with
Abelian factors (note the numbering is the reverse of our usual one). To show
the existence of k with G
(k)
= e, it suces to show that G
(i)
G
i
for each i,
as then G
(r)
G
r
= e. We work by induction on i.
For i = 0, G
(0)
= G = G
0
, so this our base case. Now let i 1 and assume by
the inductive hypothesis that G
(i1)
G
i1
. Then G
(i)
= (G
(i1)
)
(G
i1
)
and (G
i1
)
G
i
(by Lemma 6.7(i)), since G
i1
/G
i
is Abelian.
Denition 6.10. For a soluble group G, the least k N such that G
(k)
= e
is called the derived length of G and is denoted (G).
Thus any non-trivial Abelian group has derived length 1 and S
3
has derived
length 2. As you might expect, we can nd groups having any derived length
we like.
Proposition 6.11 ([Ros, Cor. 9.23]). For every natural number n, there are
soluble groups of derived length n.
Proof: Let A be any non-trivial Abelian group. Dene groups G
i
, i N, re-
cursively as follows. Let G
0
= e, G
1
= A and for each integer n > 1 let
G
n
= G
n1
C
2
. Certainly G
0
has derived length 0 and as we just observed G
1
has derived length 1.
We proceed by induction and assume that G
n
is soluble of derived length n.
Let C
2
= h be cyclic of order 2, generated by h, and let G
n+1
= G
n
C
2
be the
wreath product of G
n
by C
2
. That is, G
n+1
is equal as a set to (G
n
G
n
) C
2
and the action of C
2
on G
n
G
n
is (g
1
, g
2
)
h
= (g
2
, g
1
). Since G
n
is soluble, so
is G
n
G
n
and (G
n
G
n
) = n, as is easily seen. Now [G
n+1
/(G
n
G
n
)[ = 2
so this quotient is Abelian and by Lemma 6.7(i) (G
n+1
)
G
n
G
n
e and
G
n+1
is soluble.
The semi-direct product structure of G
n+1
= G
n
C
2
implies that for each
g G
n
we have (e, e, h
1
)(e, g, e)(e, e, h) = (g, e, e) (since (e, g)
h
= (g, e)), and
so
(g
1
, e, e)(e, g, e) = (g
1
, e, e)(e, e, h)(g, e, e)(e, e, h)
= [ (g, e, e), (e, e, h) ]
(G
n+1
)
.
Letting
2
denote the projection of G
n
G
n
e onto the second factor and
2
its restriction to (G
n+1
)
, we then have a homomorphism (G

n+1
)
G
n
that is
surjective since
2
_
(g
1
, e, e)(e, g, e)
_
= g.
Since (G
n
) = n, it follows that ((G
n+1
)
) n. However we showed that

(G
n+1
)
G
n
G
n
e and (G
n
G
n
e) = n, so ((G
n+1
)
) n. Then
((G
n+1
)
) = n and therefore (G
n+1
) = n + 1. This proves our inductive step
and so we conclude the result.
We claimed before that soluble and simple are opposite concepts: that there
are very few soluble simple groups is given by the following.
Proposition 6.12 ([AB, Prop. 1, p. 96]). A soluble simple group has prime
order.
Proof: If G is soluble and simple then G
,= G (else G
(k)
= G ,= e for all k).
Since G is simple and G
G we deduce that G
= e and so G is Abelian.
Since every subgroup of an Abelian group is normal, we must have x = G for
any x G e so G is cyclic. But G simple and cyclic implies G is nite of
prime order.
Notice, though, that S
5
is neither simple nor soluble. We also want to
know about subgroups, quotients and direct products of soluble groups, as for
polycyclic groups.
Proposition 6.13 ([AB, Prop. 3, p. 97]).
i) If G is soluble and H G then H is soluble.
ii) If G is soluble and N G then G/N is soluble.
iii) If N G and both N and G/N are soluble then G is soluble.
iv) If G and H are soluble then GH is soluble.
Proof: Exercise. [Hint: look carefully at the proof of Proposition 6.4.]
Part (i) of Theorem 6.9 (characterizing solubility) gives, in principle at least,
a concrete way of checking if a group is soluble. However if G is nite, we can
do even better.
Proposition 6.14 ([AB, Prop. 4, p. 97]). A group having a composition series
is soluble if and only if every composition factor has prime order. (Such a group
must be nite.)
Proof: A composition seriesif one existsis a subnormal series with simple
factors. If every factor has prime order then this is a subnormal series with
Abelian factors (every group of prime order is cyclic) and so the group is soluble.
Suppose G has a composition series and is soluble; let H/K be a composition
factor of G, so K H G. Then by Proposition 6.13, H/K is soluble. Then
by Proposition 6.12, H/K, as a simple soluble group, has prime order.
Corollary 6.15. Finite p-groups are soluble.
Proof: This follows from the preceding proposition and the fact that the com-
position factors of nite p-groups are nite simple p-groups and hence are of
prime order.
Taken together with our previous work on polycyclic groupswhich are
certainly solublethis corollary tells us that most of the groups we have studied
so far (nitely generated Abelian groups and nite p-groups) are soluble.
We conclude our study of soluble groups by mentioning two extremely sig-
nicant theorems on nite soluble groups. We will prove the rst once we have
some character-theoretic techniques available but the proof of the second is far
beyond our scope, unfortunately.
Theorem 6.16 (Burnsides p
theorem). If p and q are primes then any

group of order p
, , N, is soluble.
However, there exist groups whose orders are divisible by only three distinct
primes that are not soluble (A
5
is an example, as we will see later).
Theorem 6.17. (FeitThompson) All nite groups of odd order are soluble.
(Take a minute to appreciate how phenomenal this theorem is in its scope.
Its proof is similarly phenomenal: taking 255 pages and an entire issue of the
journal in which it was published.) The signicance of the result is that this
theorem tells us that the only nite simple groups of odd order are cyclic of
prime order. So non-Abelian nite simple groups have even order and Cauchys
theorem tells us that these have elements of order 2.
Elements of order 2, also called involutions, became critical to the enterprise
of establishing the classication of the nite simple groups. For subsequent work
on the so-called minimal simple nite groups, Thompson won a Fields medal;
he has also been awarded an honorary degree, the Doctor of Science (DSc), by
this university.
This point may seem rather late in the day to address the question some
of you may have had regarding the name soluble for the class of polyabelian
groups. I will now outline the origins of the terminology: those of you who have
taken Part A Fields will hopefully recognise aspectsthose who took B9a Galois
Theory certainly should. Those who took neither should either not worry too
much about the details or refer to Chapter 5 of Hersteins Topics in Algebra
([Her]).
Given a eld F and f(x) F[x], we have the splitting eld K for f over
F: the eld K is a nite-dimensional vector space over F such that f can be
written as a product of linear factors over K and K is the smallest such eld.
Associated to F and K is a group, the Galois group Gal(K, F), the group of eld
automorphisms of K leaving F xed. We call Gal(K, F) the Galois group of f;
it can be thought of as the group of permutations of the roots of f. Subelds
of K can be shown to correspond to subgroups of Gal(K, F).
Informally, a polynomial f(x) is soluble by radicals (or solvable by radi-
cals) if we can reach the splitting eld K from F by adjoining nitely many
roots of elements of F (that is, radicals,
n
as) of varying degrees. Then roots

of f(x) can be expressed by combinations of radicals of rational functions in the
coecients of f.
These are tied together by
Theorem 6.18. Let F be a eld of characteristic zero. Then f(x) F[x] is
soluble by radicals if and only if the Galois group of f over F is soluble.
In particular, if the Galois group is not soluble then f(x) is not soluble by
radicals. It can be shown that for n 5, S
n
is not soluble and hence we obtain
Abels great theorem:
Theorem 6.19. The general polynomial of degree n 5 is not soluble by radi-
cals.
Chapter 7
Character theory
In this chapter we move from a structural type of group theory over to rep-
resentation theory, the study of how groups may be realized as symmetries of
various types of object. We mainly will restrict ourselves to groups acting on
vector spaces and use the techniques of character theory to analyze these linear
representations. Characters themselves are a way of picking out key parts of
the information contained in a representation, although necessarily some infor-
mation is lost.
After a brief outline of the major features of group representations, we ex-
amine group characters. These are typically recorded in a character table and
these tables have some combinatorial properties that often allow one to con-
struct the whole table from only partial data, and hence we can obtain insight
into representations we did not know by using information about others we did.
We conclude with the main application, Burnsides p
theorem, which we will

make considerable use of in the nal chapter.
7.1 Group representations
As we said above, we will focus on groups acting on vector spaces. In this
chapter, unless otherwise specied, our groups will all be nite. We begin with
a linearization of a group, the group algebra.
Denition 7.1. Let k be a eld and G a group. The group algebra kG of G
over k is the k-algebra having basis g [ g G and product
_
_
gG
g
g
_
_
_
hG
h
h
_
=
g,hG
h
(gh).
The unit for this algebra is the identity element of the group.
It is straightforward to check that kG is an algebra. Note that
g,hG
h
(gh) =
gG
hG
(
h
h
1
g
)g
and that dimkG = [G[.
56
7.1. GROUP REPRESENTATIONS 57
Any k-algebra A has an associated category of right A-modules; that is,
k-vector spaces admitting an action of A. More precisely, (V, ) is a right A-
module if : V
k
A V satises (v, 1
A
) = v and ((v, a), b) = (v, ab)
for all v V and a, b A. (The denition of as a map from V
k
A A
encodes linearity of .) We often suppress , denoting (v, a) by va, and talk
only of a right A-module V . We write Mod-A for the category with objects
the right A-modules and morphisms being module homomorphisms, i.e. linear
maps compatible with the action.
So we have a category Mod-kG, the category of right kG-modules. We
will mostly restrict ourselves to considering mod-kG, the category of nite-
dimensional right kG-modules. Modules for the group algebra give us one
approach to considering actions of groups but there are others. We have al-
ready made use of the most general: groups acting on sets. If V is a k-vector
space, it makes sense to only consider G-actions on V that are linear, i.e.
(v+w)g = (vg)+(wg). (We say in such a case that G is acting by linear
transformations.) Such a linear action of a group can be encoded as a linear
representation, that is a group homomorphism : G GL(V ), where GL(V ) is
the group of invertible linear transformations from V to itself (invertible since
group elements are). If V is nite-dimensional then choosing a basis sets up a
correspondence between linear representations and matrix representations, i.e.
group homomorphisms : G GL(n, k). Moreover,
Lemma 7.2. Let k be a eld and G a nite group. There is a bijective cor-
respondence between linear representations of G on nite-dimensional k-vector
spaces and nite-dimensional kG-modules.
Remark 7.3. We could replace nite-dimensional kG-modules by nitely-
generated kG-modulesgiven our assumption on the niteness of G, these are
the same.
We have more structure than just the modules, though. We also have maps
between them, as we noted above.
Denition 7.4. let A be an algebra and let V , W be right A-modules. A (right)
A-module homomorphism is a linear map : V W such that (va) = (v)a
for all v V and a A. We denote by Hom
A
(V, W) the vector space of
A-module homomorphisms from V to W.
An invertible A-module homomorphism is called an A-module isomorphism,
as usual. For matrix representations, equivalence is dened as follows: matrix
representations , are equivalent if there exists T GL(n, k) such that (g) =
T
1
(g)T for all g G. Equivalence of matrix representations of G over k
corresponds to isomorphism of the associated kG-modules.
As usual for algebraic structures we have sub- and quotient structures, sub-
modules being subspaces preserved by the action and quotients constructed in
the usual way. We also have kernels and images of homomorphisms as well as
the associated isomorphism theorems.
Denition 7.5. A kG-module is simple if it has no proper non-trivial submod-
ules.
Note also that the JordanHolder theorem holds for nite-dimensional mod-
ules: composition series are dened in an entirely analogous way to those for
groups, with maximal submodule replacing maximal normal subgroup.
58 CHAPTER 7. CHARACTER THEORY
We say that a submodule U of V has a complement if there exists a sub-
module W of V such that V = U W (as modules). A module V is com-
pletely reducible if every submodule has a complement; equivalently for V nite-
dimensional, V is a direct sum of simple submodules. We have
Theorem 7.6 (Maschkes theorem, [B2a, 13.1]). Let k = R or k = C and
let G be a nite group. Then every nite-dimensional kG-module is completely
reducible.
Proof: Exercise.
In fact, the theorem remains true for any nite group G and any eld whose
characteristic does not divide [G[. Thus, at least for nite groups, real and
complex representation theory can be studied by concentrating on simple mod-
ules. It is therefore very useful to know about homomorphisms between simple
modules.
Lemma 7.7 (Schurs lemma; general form, [B2a, 8.6]). Any non-zero homo-
morphism
between simple modules is an isomorphism.
Over C there is a much stronger form, confusingly also called Schurs lemma.
Lemma 7.8 (Schurs lemma for CG-modules, [B2a, 8.6]). If V and W are
simple CG-modules then
Hom
CG
(V, W) =
_
C if V

= W
0 if V ,
= W
This is often expressed as every CG-module homomorphism between simple
modules is a scalar multiple of the identity, the scalar being 0 if V ,
= W.
The proof is straightforward: the scalar multiple is the eigenvalue of the map
(eigenvectors exists since we are working over C).
Any algebra has a centre, Z(A) = z A [ za = az a A. The centre of
a group algebra has a particularly nice description.
Proposition 7.9 ([JL, Prop. 12.22]). Let G be a nite group and enumerate
the conjugacy classes of G as (
i
[ 1 i r. Dene (
i
def
=
gC
i
g kG, the
class sum. Then (
i
[ 1 i r is a basis for Z(kG), which in particular has
dimension r.
Proof: We rst show that the class sums are central. Fixing g (
i
, we may
write (
i
= y
1
j
gy
j
[ 1 j m, so (
i
=
m
j=1
y
1
j
gy
j
. Then for all h G,
h
1
(
i
h =
r
j=1
h
1
y
1
j
gy
j
h =
m
j=1
(y
j
h)
1
g(y
j
h) = (
i
since (
i
is a conjugacy class, so as j runs from 1 to m, (y
j
h)
1
g(y
j
h) will run
through (
i
, since y
1
j
gy
j
does. So (
i
h = h(
i
and (
i
is central in kG: by linearity,
it suces for (
i
to commute with elements of G.
Now (
i
[ 1 i r is a linearly independent set, by the disjointness of
conjugacy classes. This set also spans Z(kG): let z =
gG
g
g kG. Then
7.1. GROUP REPRESENTATIONS 59
for any h G, h
1
zh = z so
gG
g
h
1
gh =
gG
g
g. Hence for every
h G the coecient
g
of g in z is the same as
h
1
gh
, i.e. that of h
1
gh. So
the function g
g
is constant on conjugacy classes and so z =
r
i=1
i
(
i
,
with
i
the coecient of any g (
i
.
Example 7.10. A basis for Z(CS
3
) is , (1 2) + (1 3) + (2 3), (1 2 3) + (1 3 2)
and dimZ(CS
3
) = 3 (compared to dimCS
3
= [S
3
[ = 6).
Remark 7.11. It follows from Schurs lemma that an element of Z(CG) must
act on an irreducible CG-module by multiplication by a scalar.
One particularly important module for any algebra A is the regular right
module A
A
, given by A as a vector space and A-action by right multiplication.
This module is nitely-generated, by the unit. The corresponding representation
is called, unsurprisingly, the regular representation. The regular representation
is faithful: the kernel of the corresponding homomorphism is trivial. (Faithful-
ness for an A-module V is the condition that a A [ va = v v V = 1
A
.)
That the regular module is faithful for kG can also be stated as the existence
of a faithful module of dimension [G[. Since G is isomorphic to a subgroup of
S
|G|
by Cayleys theorem, this is the permutation module. The reason for the
importance of the regular module is contained in the following theorem.
Theorem 7.12 ([JL, Thm. 10.5],[B2a, 10.3]). Let G be a nite group and let
CG
CG
be the regular right CG-module. By Maschkes theorem, we may write
CG
CG
= U
1
U
t
with U
i
simple CG-modules. Then any simple CG-module
is isomorphic to one of the modules U
i
. That is, every simple module occurs as
a composition factor of the regular module.
Proof: Let W be a simple CG-module and let w W, w ,= 0. Then w generates
W, since W is simple, and so W = wa [ a CG. Now dene : CG
CG
W
by (a) = wa for all a CG. This map is clearly linear, surjective and a
CG-module homomorphism. By complete reducibility there exists a submodule
U of CG
CG
so that CG

= Ker U, but then U

= Im

= W. Hence W is
isomorphic to a submodule of CG
CG
and in fact to a simple submodule, since
W is simple.
A very important corollary of this is the following.
Corollary 7.13 ([JL, Cor. 10.7]). A nite group G has nitely many non-
isomorphic simple CG-modules.
We can even understand the multiplicities:
Theorem 7.14. Let G be a nite group and V
i
[ 1 i s a complete set of
non-isomorphic simple submodules of the regular CG=module. Let d
i
= dimV
i
and m
i
the multiplicity of V
i
in the decomposition CG
CG
= V
m
1
1
V
m
s
s
.
Then
i) m
i
= d
i
for all 1 i s;
ii)
s
i=1
d
2
i
= [G[;
iii) setting B
i
= V
d
i
i
, we have CG
CG
= B
1
B
s
and there exist elements
e
i
B
i
for each 1 i s such that
(a) B
i
is a two-sided ideal of CG (after identifying CG
CG
with CG), of
dimension d
2
i
,
(b) Z(B
i
) = span
C
(e
i
) and so dimZ(B
i
) = 1, and
(c) e
i
[ 1 i s is a complete set of central orthogonal idempotents.
Proof: We rst show that dimV = dimHom
CG
(CG
CG
, V ) for any CG-module
V . For if v
i
[ 1 i d is a basis for V then for each 1 i d the map
i
: CG
CG
V ,
i
(a) = v
i
a is an element of Hom
CG
(CG
CG
, V ), as is easily
checked. Furthermore
i
[ 1 i d is a basis for Hom
CG
(CG
CG
, V ):
i) if Hom
CG
(CG
CG
, V ) then (e) V so (e) =
d
i=1
i
v
i
for some
i
C. Since is a CG-module homomorphism, if a CG
CG
then
(a) = (ea) = (e)a =
_
d
i=1
i
v
i
_
a =
d
i=1
i
(a),
so =
d
i=1
i
and the
i
span.
ii) if
d
i=1
i
= 0 then evaluating both sides at the identity gives
0 =
_
d
i=1
i
_
(e) =
d
i=1
i
v
i
and hence
i
= 0 for all i.
Next, if V has the decomposition V = U
1
U
t
with each U
i
sim-
ple and if W is any simple CG-module, then dimHom
CG
(V, W) is equal to
the number of U
i
such that U
i

= W. This is true because for general rea-
sons dimHom
CG
(V, W) =
t
i=1
dimHom
CG
(U
i
, W) and by the application of
Schurs lemma. Hence d
i
= dimV
i
= dimHom
CG
(CG
CG
, V
i
) = m
i
and the rst
part is proved.
So then
[G[ = dimCG
CG
= dim
_
s
i=1
V
m
i
i
_
=
s
i=1
m
i
d
i
=
s
i=1
d
2
i
and we have proved the second part too.
The stated decomposition of CG
CG
into a direct sum of the B
i
gives elements
e
i
B
i
such that 1
CG
= e = e
1
+ +e
s
. For any algebra, a submodule of the
regular right module is a right ideal of the algebra, so certainly the B
i
are right
ideals. Hence for any i and j, e
i
e
j
B
i
and
e
j
= ee
j
= (e
1
+ +e
s
)e
j
= e
1
e
j
+ +e
s
e
j
.
But CG is a direct sum of the B
i
so e
i
e
j
= 0 for i ,= j and e
2
j
= e
j
and the e
i
are orthogonal idempotents. (The condition 1
CG
= e = e
1
+ + e
s
is exactly
the completeness of this set.)
Now let a CG and consider its action by left multiplication on CG. Since,
as actions, left and right multiplication commute (this is associativity!), this
gives a module endomorphism L
a
: CG
CG
CG
CG
. Let
j
: CG
CG
B
j
be
7.2. CHARACTERS AND CHARACTER TABLES 61
the module homomorphism with kernel
k=j
B
k
. Since B
j
has no composition
factors isomorphic to V
i
unless i = j, we see that
j
(L
a
(B
i
)) = 0 if i ,= j. So
L
a
(B
i
)
j=i
Ker
j
= B
i
. So B
i
is closed under left multiplication and hence
is a two-sided ideal.
Finally, the orthogonality of the idempotents e
i
tells us that for b B
i
,
b = eb = be implies b = e
i
b = be
i
and so e
i
is the unit for B
i
, considered as
an algebra. So the e
i
are central in CG. Then s r, the number of conjugacy
classes of G, since dimZ(CG) = r. Since any z Z(CG) acts on V
i
by a scalar
i
, say, z does so on B
i
and in particular e
i
z =
i
e
i
. Then
z = ez = (
s
i=1
e
i
)z =
s
i=1
e
i
z =
s
i=1
i
e
i
and so z span(e
1
, . . . , e
s
). Thus dimZ(CG) s and so r = s and we must
have dimZ(B
i
) = 1, with Z(B
i
) = span(e
i
).
Corollary 7.15. A nite group G has precisely as many non-isomorphic simple
CG-modules as it does distinct conjugacy classes, namely r = s = dimZ(CG)
many.
7.2 Characters and character tables
From this point, unless otherwise specied, in this chapter G will be a nite
group, all vector spaces will be C-vector spaces and all CG-modules will be
nite-dimensional right CG-modules.
A complex matrix representation of a nite group : G GL(n, C) contains
a large amount of data: [G[n
2
complex numbers, in fact. Passing from a repre-
sentation to its character discards the majority of this but retains enough to be
useful, as we will see. The character of is the map : G C given by tr ,
taking each group element to the trace of the representing matrix, and so keeps
only [G[ pieces of data. One may alternatively prefer to dene the character of
a module:
Denition 7.16. Let V be a CG-module with basis B. The character of V is the
function : G C dened by (g) = tr(M
B
B
([g]), where [g] is the endomorphism
of V given by [g](v) = vg.
Since tr(P
1
AP) = tr(A) for any invertible matrix P, we see that does
not depend on the choice of basis B. It also follows that isomorphic CG-modules
have the same character.
Denition 7.17. We say that a map : G C is a character of G if it is the
character of some CG-module. We say is irreducible (resp. reducible) if is
the character of some irreducible (resp. reducible) CG-module.
Examples 7.18.
i) Let C be the trivial CG-module, with zg = z for all z C and g G.
The character of this module,
1
: G C, is given by
1
(g) = 1 for all
g G and is called the trivial (or principal) character of G. It is (almost)
always enumerated as
1
, both since its values are all 1 and since it is the
rst, most natural, character in some sense.
ii) If is a G-set and C the corresponding permutation module, then the
permutation character
: G C has
(g) = [Fix
(g)[ for all g G.

iii) The right regular CG-module CG
CG
has character
reg
=
_
[G[ if g = e
0 if g ,= e
and this character is called the regular character. (This formula for
reg
is a consequence of (ii) with = G, as no element other than e xes an
element of G: hg = h g = e.)
Characters are examples of a special type of function from a group to the
complex numbers.
Denition 7.19. A function : G C is a class function if (g) = (h
1
gh)
for all g, h G. That is, a class function is constant on conjugacy classes.
Proposition 7.20. Let be a character of G and thus the character of some
CG-module V . Then
i) (e) = dimV ,
ii) is a class function, and
iii) (g
1
) = (g) for all g G.
Proof: Let n = dimV .
i) The endomorphism [e] is just the identity map so M
B
B
([e]) = I
n
and hence
(e) = n = dimV .
ii) Let g, h G. As endomorphisms, [h
1
gh] = [h]
1
[g][h] so
tr(M
B
B
([h
1
gh])) = tr
_
M
B
B
([h])
1
M
B
B
([g])M
B
B
([h])
_
= tr(M
B
B
([g])
and (g) = (h
1
gh).
iii) Since G is nite, g G has some nite order m. Then M = M
B
B
([g])
satises x
m
1, which splits as distinct linear factors over C, hence the
minimum polynomial of M does and so M is diagonalisable. This implies
that (g) = tr(M) is the sum of the eigenvalues of M and since these are
roots of x
m
1, they are mth roots of unity.
Now the eigenvalues of M
1
= M
B
B
([g
1
]) are inverses of roots of unity,
but if is a root of unity then
1
= and hence (g
1
) = (g).
The proof of (iii) also tells us that [(g)[ (e) = n and if g has order 2 then
(g) Z and (g) (e) (mod 2).
Denition 7.21. The degree of a character of G is (e). Characters of degree
1 are called linear characters.
It is easy to see that a linear character : G C is a group homomorphism
: G C
, where C
is the multiplicative group of the eld C. The trivial

character is linear; the regular character has degree [G[. The map sign: S
n
C
given by taking the sign of a permutation is a linear character of S

n
.
Exercise 7.1. The only non-zero characters of a group G that are group homo-
morphisms G C
are the linear characters.

Exercise 7.2. Let : G GL(n, C) be a representation of G and : G C
its character. Then [(g)[ = (e) for all g G if and only if g = I
n
for
some C. Hence the kernel of is equal to the set g G [ (g) = (e),
so that the character retains full information on the kernel of its originating
representation. We then set Ker
def
= g G [ (g) = (e) = Ker .
Denition 7.22. Let be a character of G. Dene : G C by (g) = (g)
for all g G.
Lemma 7.23. If is a character of G then so is .
Proof: Dening A GL(n, C) in the obvious way, (AB) = AB, so if is a
representation having character we have g
def
= (g) a representation of G with
character .
Furthermore, is irreducible if and only if is.
We illustrate some of the above results by looking at characters of CC
4
.
There are precisely four non-isomorphic 1-dimensional CC
4
-modules and they
form a complete set of irreducible CC
4
-modules. The corresponding represen-
tations are
1
,
1
,
i
and
i
, given by a
k
z
= z
k
for C
4
= a , z 1, i.
Since C
4
is Abelian, its conjugacy classes are in bijection with its elements and
the characters corresponding to the above representations are as follows:
e a a
2
a
3
1
1 1 1 1
1
1 1 1 1
i
1 i 1 i
i
1 i 1 i
Note that
i
(a
1
) =
i
(a
3
) = i =
i
(a) and moreover
i
=
i
.
We have actually constructed the character table of C
4
. Before we see more
character tables, we will prove a few more properties of characters, to help
us build our examples. These are the orthogonality relations, for which we of
course need an inner product. The set of functions from G to C, Fun(G, C), has
a natural vector space structure and the subset of class functions ClassFun(G, C)
is a subspace of dimension r = dimZ(CG). (A basis for ClassFun(G, C) is given
by
i
[ 1 i r,
i
[
C
j
=
ij
.)
Denition 7.24. There is a complex inner product , on Fun(G, C) dened
by
, =
1
[G[
gG
(g)(g)
for all , Fun(G, C).
Since for characters, (g) = (g
1
) and g
1
[ g G = G, it is easy to see
that when we restrict this inner product to the set of characters of G, we have
, = , , so , is in fact real and , is symmetric, rather than
just conjugate-symmetric.
Recall that we may write CG
CG
=
r
i=1
B
i
where no pair of submodules
B
i
and B
j
have a common composition factor. Recall also that dimZ(B
i
) = 1,
yielding e
i
[ 1 i r a complete set of orthogonal central idempotents, with
each e
i
spanning Z(B
i
).
Proposition 7.25. Let
i
denote the character of the CG
CG
-submodule B
i
.
Then
i
,
i
=
i
(e) = dimB
i
.
Proof: We claim that e
i
=
1
|G|
gG
i
(g
1
)g. To see this, x h G and
consider the CG-endomorphism : CG CG, (k) = ke
i
h
1
. Now for b
j
B
j
we have b
j
e
i
=
ij
b
j
so [
B
i
is a B
i
-endomorphism with trace
i
(h
1
) (the map
[
B
i
acts on B
i
by right multiplication by h
1
) and [
B
j
= 0 for i ,= j. Hence
tr() =
i
(h
1
).
Since e
i
CG, there exist scalars
g
, g G such that e
i
=
gG
g
G.
From our analysis of the regular character, the endomorphism k kgh
1
has
trace zero if gh
1
,= e and trace [G[ if gh
1
= e. So the map , which has
(k) = k
gG
g
gh
1
, must have tr() =
h
[G[. Hence
i
(h
1
) =
h
[G[ so
for any h G,
h
=
i
(h
1
)/[G[. Thus e
i
=
1
|G|
gG
i
(g
1
)g, as claimed.
It follows that the coecient of the identity element e in e
2
i
is equal to both
1
|G|
2
gG
i
(g
1
)
i
(g) =
1
|G|

i
,
i
and
i
(e)/[G[ (since e
2
i
= e
i
) and so

i
,
i
=
i
(e) = dimB
i
.
From this, we derive our rst orthogonality theorem.
Theorem 7.26 (Row orthogonality relations, [JL, Thm. 14.12]).
Let Irr(G) =
i
[ 1 i r be the set of irreducible characters of G. Then

i
,
j
=
ij
.
Proof: Let V
i
and V
j
be irreducible CG-submodules of CG
CG
with characters
i
and
j
. Set d
k
= dimV
k
(k i, j). Then since B
i
= V
d
i
i
, the character
of B
i

i
is equal to d
i
i
and so by the previous proposition
d
i
i
, d
i
i
= dimB
i
= (dimV
i
)
2
= d
2
i
.
Hence
i
,
i
= 1.
For the case i ,= j, set Y = B
i
B
j
and Z =
k=i,j
B
k
, so Y and Z have
no common composition factors and CG
CG
= Y Z. Now letting e
i
be as
above, we see from the proof of the previous proposition that
e
i
+e
j
=
1
[G[
gG
_
i
(g
1
) +
j
(g
1
)
_
g
and the coecient of e in e
i
+ e
j
is (
i
(e) +
j
(e))/[G[. Denote by
g
the
coecient of g in [G[(e
i
+e
j
), so
g
=
i
(g
1
) +
j
(g
1
). Then
(e
i
+e
j
)
2
=
1
[G[
2
_
_
gG
g
g
_
_
2
=
1
[G[
2
gG
hG
(
h
h
1
g
)g
has coecient of e equal to
hG
h
1 =
hG
_
i
(h
1
) +
j
(h
1
)
_ _
i
(h) +
j
(h)
_
.
Also

i
+
j
,
i
+
j
=
1
[G[
hG
_
i
(h) +
j
(h)
_ _
i
(h
1
) +
j
(h
1
)
_
and so we see that
i
+
j
,
i
+
j
=
i
(e) +
j
(e). Since
i
= d
i
i
and
j
= d
j
j
, and
i
(e) = d
2
i
and
j
(e) = d
2
j
, we see that on the one hand
d
i
i
, d
j
j
= d
2
i
i
,
i
+ 2d
i
d
j
i
,
j
+d
2
j
j
,
= d
2
i
+ 2d
i
d
j
i
,
j
+d
2
j
and on the other
d
i
i
, d
j
j
= d
i
i
(e) +d
j
j
(e)
= d
2
i
+d
2
j
.
Hence 2d
i
d
j

i
,
j
= 0 and so
i
,
j
= 0.
Many important results follow from this.
Theorem 7.27. The set Irr(G) of irreducible characters is an orthonormal
basis for ClassFun(G, C).
Proof: Orthonormality is the content of the preceding theorem and linear in-
dependence is immediate from this. Since the number of irreducible char-
acters is equal to the number of conjugacy classes and this is also equal to
dimClassFun(G, C) as noted above, we have the result.
Indeed, =
r
i=1
,
i
i
for any character .
Theorem 7.28 ([JL, Thm. 14.20]). A CG-module V is irreducible if and only
if its character satises , = 1.
Proof: By the above, if V is irreducible, , = 1. Conversely, if , = 1
then since =
r
i=1
m
i
i
for some m
i
, we have 1 = , =
r
i=1
m
2
i
and
so exactly one m
i
is non-zero, m
k
say, and m
k
= 1. So =
k
and V

= V
k
is
irreducible.
Theorem 7.29 ([JL, Thm. 14.21]). Two CG-modules are isomorphic if and
only if their characters are equal.
Proof: We discussed the forward direction earlier. If V , W are CG-modules with
equal character then V

=
r
i=1
V
i
i
, W

=
r
i=1
V
i
i
and
i
= ,
i
=
i
,
so V

= W.
Theorem 7.30 ([JL, Thm. 14.24]). Let V and W be CG-modules with charac-
ters and . Then , = dimHom
CG
(V, W).
Proof: This follows straightforwardly from complete reducibility, orthogonality
and Schurs lemma: dimHom
CG
(V
i
, V
j
) =
ij
.
There is a second orthogonality result, called column orthogonality for rea-
sons we will see very shortly.
Theorem 7.31 ([JL, Thm. 16.4]). Let G be a group and g
i
[ 1 i r a
complete set of conjugacy class representatives. Let Irr(G) =
i
[ 1 i r
be the irreducible characters of G. Then
i)
r
k=1
i
(g
k
)
j
(g
k
)
|C
G
(g
k
)|
=
ij
(row orthogonality relations);
ii)
r
k=1
k
(g
i
)
k
(g
j
) =
ij
[C
G
(g
i
)[ (column orthogonality relations).
Proof: The rst claim is simply a restatement of Theorem 7.26, making use of
the fact that characters are class functions and that [(
i
[[C
G
(g
i
)[ = [G[.
For the second part, let
i
[ 1 i r be the aforementioned basis for
ClassFun(G, C), with
i
taking value 1 on (
i
and 0 elsewhere. Since Irr(G) is a
basis for ClassFun(G, C),
j
=
r
k=1
j
k
k
for some
j
k
C. Then
j
k
=
j
,
k
=
1
[G[
gG
j
(g)
k
(g) =
1
[G[
[(
j
[
j
(g
j
)
k
(g
j
) =
k
(g
j
)
[C
G
(g
j
)[
.
Hence
ij
=
j
(g
i
) =
r
k=1
j
k
k
(g
i
) =
r
k=1
k
(g
i
)
k
(g
j
)
|C
G
(g
j
)|
.
We can now turn properly to character tables.
Denition 7.32. Let Irr(G) =
i
[ 1 i r be the irreducible characters
of G and g
i
[ 1 i r a complete set of conjugacy class representatives,
with
1
the trivial character and g
1
= e (by convention). The r r matrix with
(i, j)-entry
i
(g
j
) is called the character table of G.
Note that the character table is an invertible matrix: by orthogonality its
rows, the irreducible characters, are linearly independent. Also since
r
i=1
i
(e)
2
=
r
i=1
(dimV
i
)
2
= [G[
we see that the sum of the squares of the entries in the rst column equals [G[.
We will refer to this as the degree sum property.
Examples 7.33.
i) G = C
2
= a [ a
2
; (
1
= e, (
2
= a so set g
1
= e, g
2
= a.
g
i
e a
[C
G
(g
i
)[ 2 2
1
1 1
2
1 1

1
is the trivial character

2
(e) = 1 since [G[ = 2 = 1
2
+ 1
2

2
(a) = 1 by row orthogonality: 1 1 +1
2
(a) = 0
2
(a) = 1
Note that
reg
=
1
+
2
.
ii) G = S
3
g
i
(1 2 3) (1 2)
[C
G
(g
i
)[ 6 3 2
1
1 1 1
2
1 1 1
3
2 1 0

1
is the trivial character;
2
is the sign character

3
() is obtained from the degree sum property: 2
2
= 6 1
2
1
2

3
((1 2 3)) is obtained by column orthogonality:
1 1 + 1 1 + 2
3
((1 2 3)) = 0

3
((1 2 3)) = 1

3
((1 2)) is obtained by row orthogonality:
1 2
6
+
1 1
3
+
1
3
((1 2))
2
= 0

3
((1 2)) = 0
For the natural permutation module on = 1, 2, 3,
=
1
+
3
(recall
that
(g) = [Fix
(g)[). This is a general phenomenon: for any subgroup

H of S
n
with permutation character
, =
1
is a character of H.
Here
reg
=
1
+
2
+ 2
3
.
iii) G = C
2
C
2
= a, b [ a
2
, b
2
, ab = ba
g
i
e a b ab
[C
G
(g
i
)[ 4 4 4 4
1
1 1 1 1
2
1 1 1 1
3
1 1 1 1
4
1 1 1 1
Here
reg
=
1
+
2
+
3
+
4
.
In order to further aid our search for characters of a group, we return to
linear characters; that is, characters of degree 1.
Proposition 7.34 ([JL, Prop. 17.14]). Let be a linear character of G and
any character. Then the product : G C dened by ()(g) = (g)(g) for
all g G is a character of G. Furthermore, is irreducible if and only if is
irreducible.
Proof: Exercise.
Corollary 7.35. Let

G be the set of linear characters of G. Dene a binary
operation on

G by (
1

2
)(g) =
1
(g)
2
(g) for all g G. Then (
G, ) is an
Abelian group, called the dual group, with identity element the trivial character.
Proof: Exercise.
Now G is Abelian if and only if every irreducible character is linear, as follows
easily from the theorem that the number of characters equals the number of
conjugacy classes and the degree sum property. In fact, if A is a nite Abelian
group then

A

= A.
To nd linear characters, we use a special case of a process called lifting,
which produces characters of G from those of quotients of G. Proper quotients
are smaller, so nding their characters is in principle easier. We will also see that
if we know all characters of G, we may identify the possible normal subgroups
of G and so we can use character theory as another way to see if a group is
simple.
Proposition 7.36 ([JL, Prop. 17.1]). Let N be a proper normal subgroup of a
group G and let be a character of G/N. Dene : G C by (g) = (Ng)
for all g G. Then is a character of G of the same degree as and is
irreducible if and only if is. The character is called the lift of to G.
Proof: Let : G/N GL(n, C) be a representation with character and let
: G G/N be the natural surjective homomorphism then their composition
is clearly a representation of G with character values
(g) = tr(g ) = tr(Ng ) =

(Ng),
and thus character . Then (e) = (Ne) = (e
G/N
) so the degrees are equal.
The rest of the proof is deferred to the following theorem, which shows that
a slightly stronger statement than this proposition in fact holds.
Theorem 7.37 ([JL, Thm. 17.3]). Let N be a proper normal subgroup of G.
There is a bijection between the set of characters of G/N and the set of charac-
ters of G having N in their kernel. This bijection preserves irreducibility.
Proof: Exercise.
Corollary 7.38 ([JL, Thm. 17.11]). The number of linear characters of G is
equal to [G/G
[ where G
is the derived group of G.

Proof: Let m = [G/G
[. Since G/G
is Abelian, G/G
has m linear characters

1
, . . . ,
m
, all of which are irreducible. Their lifts
1
, . . . ,
m
have degree 1
and are the irreducible linear characters of G having G
in their kernel. Now if

is any linear character of G, then (g
1
h
1
gh) = (g)
1
(h)
1
(g)(h) = 1
so G
Ker . Thus
1
, . . . ,
m
are precisely the linear characters of G.
So we may easily nd the linear characters of S
n
: S
n
= A
n
so there are two
linear characters,
1
(trivial) and
2
. We know the sign character is linear, so
this is
2
, and there are no others.
For the other direction, nding normal subgroups from character tables,
recall that Ker = g G [ (g) = (e) G. So any intersection of kernels of
characters is a normal subgroup. In fact, every normal subgroup arises in this
way.
Proposition 7.39 ([JL, Prop. 17.5]). If N G then there exist irreducible
characters
i
1
, . . . ,
i
s
of G such that N =
s
k=1
Ker
i
k
.
Proof: If g Ker for all Irr(G) then g Ker
1
(where
1
is the class
function with value 1 on (
1
= e and 0 otherwise), since Irr(G) is a basis for
the space of class functions. So g = e and
Irr(G)
Ker = e.
Let Irr(G/N) =
i
1
, . . . ,
i
s
, so N =
s
k=1
Ker
i
k
by the above. Let
i
k
be the lift of
i
k
to G and note that g Ker
i
k
implies Ng Ker
i
k
.
So if g

s
k=1
Ker
i
k
then Ng

s
k=1
Ker
i
k
= N and so g N. Hence
N =
s
k=1
Ker
i
k
.
From this we see how to detect simplicity.
Proposition 7.40 ([JL, Prop. 17.6]). The group G is simple if and only if G
is non-trivial and Ker
i
= e for all
i
Irr(G)
1
.
Proof: We prove that G is not simple if and only if Ker
i
,= e for some non-
trivial irreducible character
i
. If N is a proper non-trivial normal subgroup
of G, then by the previous proposition there exists an irreducible character
i
whose kernel is neither e nor G, as required. Conversely, given a non-trivial
i
with non-trivial kernel, this kernel is a proper non-trivial normal subgroup.
We see from its character table that S
3
is not simple:
2
((1 2 3)) = 1 =
2
().
Indeed the sign character shows that S
n
is not simple for all n 3.
We return to giving some more examples, beginning with A
4
.
Examples 7.33. (continued)
iv) G = A
4
g
i
(1 2)(3 4) (1 2 3) (1 3 2)
[C
G
(g
i
)[ 12 4 3 3
1
1 1 1 1
2
1 1
2
3
1 1
2
4
3 1 0 0
A
4
= V
4

= C
2
C
2
and A
4
/V
4

= C
3
so there are three linear
characters, lifting those of C
3
. Set = e
2i/3
. Then the linear
characters are
1
,
2
,
3
as below.
one irreducible character remains: =

1
is a character and
, =
9
12
+
1
4
+ 0 + 0 = 1 so
4
= is irreducible.
v) G = S
4
g
i
(1 2) (1 2 3) (1 2)(3 4) (1 2 3 4)
[C
G
(g
i
)[ 24 4 3 8 4
1
1 1 1 1 1
2
1 1 1 1 1
3
2 0 1 2 0
4
3 1 0 1 1
5
3 1 0 1 1
the linear characters are
1
(trivial) and
2
(sign) and so there are
three irreducible non-linear characters to nd.

4
=

1
is a character and
4
,
4
= 1, as is easily checked,
so
4
is irreducible.
hence
5
=
4
2
is another irreducible character.
V
4
S
4
and S
4
/V
4

= S
3
. The trivial and sign characters of S
3
lift
to those of S
4
. The lift
3
of the degree 2 character
3
of S
3
is as
in the table, since (1 2)(3 4) V
4
implies
3
((1 2)(3 4)) =
3
() = 2
and V
4
(1 2 3 4) = V
4
(1 3) implies
3
((1 2 3 4)) =

3
((1 3)) =

3
((1 2)) = 0.
vi) G = D
14
= a, b [ a
7
, b
2
, b
1
ab = a
1
g
i
e a a
2
a
3
b
[C
G
(g
i
)[ 14 7 7 7 2
1
1 1 1 1 1
2
1 1 1 1 1
3
2 +
1
2
+
2
3
+
3
0
3
2
2
+
2
4
+
4
6
+
6
0
3
2
3
+
3
6
+
6
2
+
2
0
a set of conjugacy class representatives is e, a, a
2
, a
3
, b so we need
to nd ve irreducible characters.
D
14
= a and D
14
/ a

= C
2
so we obtain
1
and
2
by lifting the
(linear) characters of C
2
.
the remaining characters of D
14
all have degree 2 and are obtained
from the following natural representations of D
14
. Set = e
2i/7
and
for each integer 1 j < 7/2 dene
A
j
=
_
j
0
0
j
_
and B
j
=
_
0 1
1 0
_
.
It is straightforward to check that A
7
j
= B
2
j
= I and B
1
j
A
j
B
j
= A
1
j
so
j
: D
14
GL(2, C), (a
r
b
s
)
j
= A
r
j
B
s
j
is a representation of D
14
for each j. If i ,= j, A
i
has dierent eigenvalues from A
j
so
i
and
j
cannot be equivalent. For 1 j 3 dene
j+2
to be the character
of
j
. An easy calculation shows that
i
,
i
= 1 for 3 i 5,
so these characters are irreducible. Alternatively, it is clear from the
denitions that the representations
j
are irreducible.
So we see that character tables encode a lot of information about groups.
However they do not capture everything.
Proposition 7.41 ([JL, Ex. 17.1]). Dene Q = a, b [ a
4
, a
2
= b
2
, b
1
ab =
a
1
, the quaternion group of order 8. (A concrete construction of this group
and explanation of its name may be found in Appendix A on page 87.) The
character tables of Q and D
8
, the dihedral group of order 8, are equal but Q ,
=
D
8
, so non-isomorphic groups may have the same character table.
Proof: Exercise.
There are many other important and interesting constructions to help nd
characters but we cannot describe them all here. Some of the most signicant
include tensor products, restriction and induction and Frobenius reciprocity
but we next turn our attention towards our main application, Burnsides p
theorem.
7.3. AN APPLICATION: BURNSIDES P
THEOREM 71
7.3 An application: Burnsides p
theorem
We will begin with some properties of algebraic integers necessary for our proof
of Burnsides p
theorem and, along the way, demonstrate a pair of very

useful theorems, namely that rational character values are necessarily integers
and that the degree of any irreducible character divides the group order.
Denition 7.42. A complex number is said to be an algebraic integer if it
is the eigenvalue of an integer matrix.
Equivalently, is an algebraic integer if it is the root of some monic polynomial
in Z[x]. So every integer is an algebraic integer, as is every positive square root
m of a natural number m (consider (

0 1
m 0
)). If is an algebraic integer so are
and . Also every root of unity is an algebraic integer (consider x
n
1).
Indeed, if and are algebraic integers then so are and +, so the set of
algebraic integers A is a subring of C. It follows that every character value (g)
is an algebraic integer, as we showed earlier that character values are sums of
roots of unity.
The next theorem tells us even more about character values.
Theorem 7.43 ([JL, Prop. 22.5]). Any rational algebraic integer is an integer:
Q A = Z.
Proof: Suppose Q Z. Write = r/s with (r, s) = 1 and s ,= 1. Let p
be a prime dividing s. Then if A M
n
(Z), the o-diagonal entries of sA rI
are divisible by s and hence by p. So det(sA rI) = (r)
n
+ mp for some
m Z. Since p r we conclude that det(sA rI) ,= 0. Then det(A I) =
_
1
s
_
n
det(sArI) ,= 0 for all A M
n
(Z) and is not an algebraic integer.
Corollary 7.44 ([JL, Cor. 22.6]). Every rational character value is an integer.
Corollary 7.45.
2 is irrational.
The following lemmas yield algebraic integers naturally coming from characters.
Lemma 7.46 ([JL, Lem. 22.8]). Let r =
gG
g
g CG with
g
Z for all
g G. If there exists a non-zero u CG such that ur = u with C then
is an algebraic integer.
Proof: The existence of u gives us that is an eigenvalue of the integer matrix
A = (
g
1
i
g
j
) where g
1
, . . . , g
n
is an enumeration of the elements of G.
Lemma 7.47 ([JL, Cor. 22.10]). If is an irreducible character of G and g G
then
|G|
|C
G
(g)|
(g)
(e)
is an algebraic integer.
Proof: Let V be an irreducible CG-module with character , ( be the conjugacy
class containing g and ( =
hC
h the class sum. We claim that u( = u for
all u V , where =
|G|
|C
G
(g)|
(g)
(e)
.
Since ( Z(CG) there exists some C such that u( = u for all u V . If
B is a basis for V then
hC
M
B
B
([h]) = I so taking traces,
hC
(h) = (e).
But is constant on conjugacy classes, hence on (, so [([(g) = (e) and hence
=
|G|
|C
G
(g)|
(g)
(e)
. We conclude that is an algebraic integer, by the previous
lemma.
Therefore we have:
Theorem 7.48 ([JL, Thm. 22.11]). If is an irreducible character of G then
(e) divides [G[.
Proof: Let g
i
[ 1 i r be a complete set of conjugacy class repre-
sentatives. Then for all i,
|G|
|C
G
(g
i
)|
(g
i
)
(e)
and (g
i
) are algebraic integers, so
r
i=1
|G|
|C
G
(g
i
)|
(g
i
)(g
i
)
(e)
is an algebraic integer. But this algebraic integer is equal
to [G[/(e) by the row orthogonality relations. So the rational algebraic integer
[G[/(e) is an integer and hence (e) [ [G[.
Hence, for example, the degree of every irreducible character of a p-group is a
power of p (and is not equal to [G[ by the degree sum property). So groups of
order p
2
have all irreducible characters of degree 1 or p. But the degree cannot
be p for any irreducible character, since there is the trivial character of degree
1 and a degree p irreducible character would violate the degree sum property.
Hence all irreducible characters of a group of order p
2
are linear and we recover
the fact that such groups are Abelian (compare with Proposition 5.6).
Corollary 7.49. No simple group has an irreducible character of degree two.
Proof: Exercise.
In fact, we can tell when a character value is an integer if we fully understand
conjugacy in our group.
Theorem 7.50 ([JL, Thm. 22.15]). Let g G have order n and suppose that
for all 1 i n with (i, n) = 1 g is conjugate to g
i
. Then (g) is an integer
for any character of G. Conversely, if (g
i
) Z for all characters of G,
then g is conjugate to g
i
whenever (i, n) = 1.
Proof: Omitted. The converse requires Galois theory.
Corollary 7.51 ([JL, Cor. 22.16]). All character values for symmetric groups
are integers.
Now we can turn directly towards our main goal. For this, we need an
extension of the notion of an algebraic integer, namely algebraic numbers. A
complex number is said to be algebraic (or an algebraic number) if it is the root
of some non-zero polynomial in Q[x]. An algebraic number has a minimum
polynomial m(x) and we will refer to the elements of the set of roots of m(x)
as conjugates of . We will need the following facts about algebraic numbers.
Lemma 7.52. Let and be algebraic numbers. Then every conjugate of +
is of the form
where
(respectively,
) is a conjugate of (respectively,
). If r Q then every conjugate of r is of the form r
with
conjugate to
.
Next, we can see which character values give algebraic integers.
Lemma 7.53 ([JL, Lem. 29.2]). Let be a character of G and let g G. Then
[(g)/(e)[ 1 and if 0 < [(g)/(e)[ < 1 then (g)/(e) is not an algebraic
integer.
7.3. AN APPLICATION: BURNSIDES P
THEOREM 73
Proof: Let (e) = d. Then (g) is a sum of d roots of unity
1
, . . . ,
d
so
(g)/(e) =
1
d
d
i=1
i
. Since [(g)[
d
i=1
[
i
[ = d, we have [(g)/(e)[ 1.
Suppose (g)/(e) is an algebraic integer with [(g)/(e)[ < 1. We claim
that (g) = 0. Set = (g)/(e) and let m(x) be the minimum polynomial of
, so m(x) =
n
i=1
i
x
i
for some n with
n
= 1 and
i
Z for all i. By the
previous lemma, every conjugate of is of the form
1
d
d
i=1
i
with
i
conjugate
to
i
and hence also a root of unity. So every conjugate of also has modulus
at most 1. Then the product of all the conjugates of , including itself,
has modulus strictly less than 1. But the product of the roots of m(x) is equal
to
0
and since
0
Z and [[ < 1, we have
0
= 0. Since m(x) is the
minimum polynomial it is irreducible, so we must have m(x) = x and = 0.
Hence (g) = 0.
Theorem 7.54 ([JL, Thm. 29.3]). Let G be a group with a conjugacy class (
of size p
r
with p prime and r 1. Then G is not simple.
Proof: Let g (. Since [([ > 1 by assumption, G is not Abelian and g ,= e. Let
Irr(G) =
i
[ 1 i r with
1
the trivial character. From the column or-
thogonality relations we see that 1 +
r
i=2
i
(g)
i
(e) = 0. So
r
i=2
i
(g)
i
(e)
p
=
1
p
/ A. So for some i 2,

i
(g)
i
(e)
p
is not an algebraic integer and hence
i
(e)/p is not, since
i
(g) is. That is, p
i
(e). Since [([ = p
r
,
i
(e) and [([
are coprime so there exist integers a and b such that a[([ + b
i
(e) = 1 and so
a|G|
|C
G
(g)|
i
(g)
i
(e)
+ b
i
(g) =

i
(g)
i
(e)
. Now we have seen that the left-hand side is an
algebraic integer and
i
(g) ,= 0 so it is non-zero. So
i
(g)/
i
(e) A and hence
[
i
(g)/
i
(e)[ = 1.
Let be a representation of G with character
i
. Then by Exercise 7.2,
there exists C such that g = I. Let K = Ker G, a proper normal
subgroup since
i
,=
1
. If K ,= e then G is not simple. If K = e then
g = I commutes with h for all h G and so g commutes with every h G,
since K = e means that is faithful. So g Z(G) G and Z(G) ,= e, so
G is not simple.
Theorem 6.16 (Burnsides p
theorem, [JL, Thm. 29.4]). If p and q are

primes then any group of order p
, , N, is soluble.
Proof: We have dealt with a number of special cases already:
= = 0: G = e is soluble
+ = 1: G is cyclic of prime order and is soluble (and simple)
p = q, + 2: Gis a non-Abelian p-group and is soluble (Corollary 6.15)
p ,= q, = = 1: G has order pq and is soluble (Proposition 5.19)
We consider the remaining cases: p ,= q, 2, 1. We may assume
G is not Abelian, as nitely generated Abelian groups are polycyclic and so
are certainly soluble (Proposition 6.5). By Sylows theorems, G has a Sylow
q-subgroup Q of order q
and Z(Q) ,= e, since Q is a non-trivial q-group. Let

g Z(Q) e, so Q C
G
(g) and if ( is the conjugacy class of G containing
g then [([ = [G : C
G
(g)[ = p
c
for some c. If p
c
= 1 then g Z(G) and Z(G)
is a non-trivial proper normal subgroup, so G is not simple. If p
c
> 1 then the
previous theorem applies and G is again not simple.
Hence G has a proper non-trivial normal subgroup N and both [N[ and
[G/N[ divide p
. By induction on +, N and G/N are soluble and hence

G is soluble.
Chapter 8
Simple groups
Finally, we come to the culmination of all our hard work. We use all the tools we
have collected to study simple groups. We begin by recording the classication
of groups of order less than 16, up to isomorphism, in Table 8.1. The details of
this classication may be found in Appendix A. One might well think that 16
isnt a very impressive number to be able to get to, given all the results we have
at our disposal, but this simply illustrates how hard the isomorphism problem,
that of determining all groups of a given order up to isomorphism, is. We will
see that even order 8 is pretty hard (the alert among you will have spotted that
both 8 and 16 are powers of 2 and indeed this is not a coincidence).
Having given up rather quickly on understanding every group, we try to
get a handle on the nite simple groups, or rather, for which orders there are
simple groups and what those groups are when they do exist. This has been a
huge endeavour for group theorists and it is now commonly held to have been
completed. The whole classication is beyond our reachit would take long
enough just to dene all of the groups appearing in the classicationbut we
can make a start.
If one focuses on the part of the problem which asks for which orders there
are no simple groups, then we can consider ourselves to have made some decent
progress: we will rule out innitely many possible orders. This leaves innitely
many still to be taken care of, but its a lot better than only getting nitely
many!
One might legitimately ask why we single out the nite simple groups for
classication. This is because if we knew all nite simple groups and if we
could solve the extension problem for nite groups, then we would know all
nite groups. As hard as this approach may appear, it is considerably easier
than the order-by-order approach.
8.1 Non-Abelian nite simple groups
Rather than work one possible group order at a time, we will use the machinery
we have developed to strike out many possibilities in one go, based on the prime
factorisation of the order. To start we recall what it means to be simple:
Denition 8.1. A group is simple if it is non-trivial and has no non-trivial
proper normal subgroups.
75
76 CHAPTER 8. SIMPLE GROUPS
n Isomorphism types of groups of order n
1 e
2 C
2
3 C
3
4 C
4
, C
2
C
2
5 C
5
6 C
6
, D
6
7 C
7
8 C
8
, C
2
C
4
, C
2
C
2
C
2
, D
8
, Q
9 C
9
, C
3
C
3
10 C
10
, D
10
11 C
11
12 C
12
, C
2
C
6
, A
4
, D
12
, C
3
C
4
13 C
13
14 C
14
, D
14
15 C
15
Table 8.1: List of groups of order less than 16, up to isomorphism.
It is of course sucient to restrict our attention to non-Abelian nite simple
groups, as we have already classied the Abelian ones, these being cyclic of prime
order. We begin with a number of very useful tools for our analysis. Many, but
not all, are superseded by Burnsides p
theorem but we still include them

here to show that one can do much of the analysis without resorting to that
substantial theorem.
We have already used the following corollary to Sylows theorems several
times.
Corollary 8.2. If G is a nite group whose order is divisible by a prime p such
that [Syl
p
(G)[ = 1 then G is not simple.
If the order of our group has a special form, we can often see whether it can be
simple.
Proposition 8.3 ([ST, Prop. 4.1]). There is no non-Abelian simple group of
order
i) 2p with p prime, or
ii) pq with p ,= q primes, or
iii) p
n
with p prime, n 0.
Proof: Clearly (i) is a special case of (ii) if p ,= 2 or (iii) if p = 2. Now (ii)
is dealt with by Proposition 5.19. We need only the start of the proof of that
result, showing that there is a unique and hence normal subgroup of order p.
8.1. NON-ABELIAN FINITE SIMPLE GROUPS 77
For (iii), if n = 0 we have the trivial group which is not simple by denition.
For n 1, let G be a group of order p
n
with p prime. Then Proposition 5.5
tells us that the centre of G, Z(G) is non-trivial. If G is simple then since Z(G)
is normal in G we must have G = Z(G), i.e. G Abelian. Hence there is no
non-Abelian simple group of order p
n
.
We can actually improve on (ii) in the previous proposition a little.
Proposition 8.4 ([ST, Prop. 4.2]). Let G be a nite group such that [G[ = pr
with p prime and 1 < r < p. Then G is not a non-Abelian simple group.
Proof: Exercise.
We may use Cauchys theorem to improve on the case [G[ = 2p, p ,= 2, as
follows.
Proposition 8.5. Let G be a nite group of order [G[ = 2m for m odd. Then
G has a normal subgroup of order m.
Proof: Exercise. [Consider the action of G on itself by right multiplication. As
a coset space, this is cos(G : e).]
So in particular if G is a non-Abelian group of order [G[ = 2p
a
for p ,= 2 prime,
a 1, then G is not simple.
Two very useful results are the following proposition and its corollary.
Proposition 8.6. Let G be a nite group and H a subgroup of index n in G.
Recall that core
G
(H)
def
=
gG
H
g
is the largest normal subgroup of G contained
in H. Then [G : core
G
(H)[ divides n!.
Proof: Exercise. This uses only ideas from Groups in Action. The key is to iden-
tify an appropriate set for G to act on and hence construct a homomorphism
G Sym().
Corollary 8.7. If G is a nite simple group and H is a subgroup of index n
then [G[ divides n!/2.
Proof: Exercise.
We now have enough machinery to attempt the following theoremwhich
is really a corollary of all our results thus far.
Theorem 8.8. There is no non-Abelian simple group G having order [G[ < 60
or 60 < [G[ 167.
Proof: (cf. [ST, p. 131-134]) By means of the results of this section together with
Proposition 5.20 and Burnsides p
theorem, all of the orders in the specied

range are excluded as orders of non-Abelian simple groups.
You should be wondering about the exclusion of [G[ = 60 from the statement
of the above theorem. This is because there is a non-Abelian simple group of
order 60, as we will prove in the next section, whose title rather gives the game
away! Similarly, there is a good reason for stopping at 167, but we will have to
wait for that too.
8.2 The simplicity of the alternating groups of
degree at least 5
The symmetric group S
n
of degree n is one of the rst examples of a group that
one meets, as the group of permutations of n objects. One learns quickly that
S
n
is interesting partly because it is not Abelian. Unfortunately for us, S
n
is
not simple: it has a subgroup of index 2, which is of course non-trivial, proper
and normal. This subgroup is also well-known to us as the subgroup of even
permutations and is called the alternating group A
n
of degree n. In fact, A
n
is
the only index 2 subgroup of S
n
(see Lemma 8.11 below).
For small n, these groups A
n
are easily identied: both A
1
and A
2
are trivial
and A
3
is cyclic of order 3; of these, only A
3
is simple (but it is Abelian). The
group A
4
is a little more interesting: it has order 4!/2 = 12 and is not Abelian.
As we have already seen, there is no non-Abelian simple group of order 12
since a Sylow 2-subgroup has index 3 but 12 3!/2 = 3 (Corollary 8.7). Thus
A
4
is not simple. We can even explicitly identify a non-trivial proper normal
subgroup, namely V
4
= , (1 2)(3 4), (1 3)(2 4), (1 4)(2 3) (which is isomorphic
to C
2
C
2
). This is a little disappointing but A
4
is really rather an oddity, since,
as we have suggested, the alternating groups of degree at least 5 are simple. We
now prove this by induction on the degree n, starting with a separate theorem
to establish the base case.
Theorem 8.9 ([ST, Prop. 4.4]). The alternating group of degree 5, A
5
, is a
simple group, of order 60.
Proof: (non-examinable) We begin by considering 5-cycles and 5-subgroups
(60 = 2
2
3 5). By a straightforward counting argument, there are 24 5-
cycles in S
5
, all of which are even and hence in A
5
. Hence there must be exactly
6 cyclic subgroups of order 5. If a normal subgroup contains a 5-cycle, it must
contain every power and conjugate of that 5-cycle and therefore all 5-subgroups,
and hence all 24 5-cycles. So the only candidate order of a proper normal sub-
group of A
5
that is a multiple of 5 is 30 (recall that by Cauchys theorem, if 5
divides the order of such a subgroup then it contains an element of order 5, i.e.
a 5-cycle). But 3 [ 30 too so the subgroup would also have an element of order
3 and so, by normality, all 20 of these. Then since 24 + 20 > 30 this is absurd.
We may make a similar argument for the candidate orders 3, 6 and 15 and
so the only possible orders of proper normal subgroups of A
5
are 2 and 4. Now
A
5
contains no elements of order 4 (they are 4-cycles in S
5
, hence odd). So any
non-trivial element of a 2-subgroup of A
5
is of order 2 and hence is a (2, 2)-cycle.
Any pair of (2, 2)-cycles are conjugate in S
5
but it is straightforward to check
that they are also conjugate in A
5
. So any normal 2-subgroup would have to
contain every conjugate of its (2, 2)-cycles and hence every (2, 2)-cycle. But the
number of (2, 2)-cycles is certainly greater than 4, so this is not possible.
Hence A
5
is simple.
With the base case established, we can now tackle the inductive step and
hence the theorem.
Theorem 8.10 ([Ros, Thm. 5.28]). The alternating group of degree n, A
n
, is
a simple group, of order n!/2.
8.2. THE SIMPLICITY OF A
N
, N 5 79
Proof: We argue by induction on n. The assertion is true for n = 5, by the
preceding theorem. Assume now that n > 5 and, inductively, that A
n1
is
simple.
Let G = A
n
and consider the natural action of Gon the set X = 1, 2, . . . , n.
For each i = 1, . . . , n, let H
i
= Stab
G
(i). Note that G acts transitively on X, for
if i, j, k X are all distinct, (i j k) G = A
n
moves i to j. Since stabilizers of
points in the same orbit are always conjugate, the subgroups H
i
, i = 1, . . . , n,
are all conjugate and hence isomorphic. Furthermore, it is easy to see that
H
i

= A
n1
for all i and so H
i

= H
j

= A
n1
for all i, j = 1, . . . , n. By the
inductive hypothesis, H
i
is simple for each i, therefore.
Now assume, contrary to what we wish to show, that G has a non-trivial
proper normal subgroup K. Then, for each i = 1, . . . , n, H
i
K H
i
so since
H
i
is simple, H
i
K is either trivial or H
i
. Suppose that there exists a j such
that H
j
K = H
j
, i.e. H
j
K. Then, as we saw above, there is an element
G such that H
i
= H
j
and so, since K G, H
i
K
= K. Thus K would
contain every H
i
but then since [X[ > 3 K = G, a contradiction. Hence
H
i
K = e for all i.
Let K, ,= . Then for every i, / H
i
, that is, xes no point of
X. Let a X and let b = a, so b ,= a. Since [X[ > 3, there is a point
c X such that c ,= a, c ,= b and c ,= a
1
. Let d = c: then, since is a
permutation of X which does not x c, we see that d ,= c, d ,= b (else a = c) and
d ,= a (else c = a
1
). Since in fact [X[ 6, we can choose two more distinct
points e, f X, both distinct from a, b, c and d. Now let = (a b)(c d e f);
G = A
n
, as is easily veried. Then since moves a to b and c to d,
moves b to a and d to e. Moreover,
K since K G. Hence
K
but
xes a and moves c to e (,= c). Thus
H
a
K but
,= , a
contradiction. Hence G = A
n
is simple and by induction we are done.
To tidy up the one remaining loose end, we show that A
5
is in fact the only
simple group of order 60, up to isomorphism. Three simple lemmas will see us
home.
The proofs of the remaining results in this section are not examinable.
Lemma 8.11 ([ST, Lem. 4.11]). Let n 2 and suppose that A is a subgroup
of S
n
of index 2. Then A = A
n
.
Proof: Since A has index 2 in S
n
, it is certainly normal in S
n
. Furthermore,
if S
n
is any transposition (i.e. cycle of length 2) and A then every
transposition is in A, because all transpositions are conjugate in S
n
. However,
S
n
is generated by the set of all transpositions so we would have A = S
n
, which
is not the case. Therefore for every transposition S
n
, A ,= A.
Now the group S
n
/A is cyclic of order 2. If A
n
, then is the product
of an even number of transpositions and so A S
n
/A is an even power of the
generator of S
n
/A. Hence A = A and so A. Since A
n
was arbitrary,
A
n
A. However these two subgroups of S
n
have the same order, n!/2, hence
are equal.
Lemma 8.12. Suppose G is a simple group of order 60. Then G has a subgroup
H of order 12.
Proof: Exercise.
Lemma 8.13 ([ST, Lem. 4.10]). Suppose that G is a simple group of order 60.
Then G is isomorphic to a subgroup

G of S
5
.
Proof: By the previous lemma, G has a subgroup H of order 12 and thus index
5. Let = cos(G : H), so [[ = 5. The action of G on = cos(G : H)
by right multiplication induces a group homomorphism : G Sym() and
hence

: G S
5
, since Sym()

= S
5
. Since Ker G, Ker H and G is
simple, Ker = e and is injective and so is

. Thus G

= Im

S
5
so take
G = Im

.
Then we deduce:
Proposition 8.14 ([ST, Prop. 4.12]). Let G be a simple group of order 60.
Then G

= A
5
.
Proof: We have assembled the tools to do this quickly. By the preceding lemma,
G

= A < S
5
. Now [S
5
: A[ = 2 and so by Lemma 8.11, A

= A
5
, so G

= A
5
.
In fact, one may push the arguments of this section and the previous one a
little further to show that there is a unique non-Abelian simple group of order
168 (and we have shown that there are none having intermediate order). This
simple group is not A
6
, of course: [A
6
[ = 360. We will show that it is a group
called PSL(2, 7) and that is the story of the next section. However, the next
simple group after PSL(2, 7) has order 360 and is indeed A
6
.
8.3 The projective special linear groups and their
simplicity
We start by introducing the various linear groups we will be concerned with.
Denition 8.15. Let F be a eld.
i) The general linear group of degree n over F, GL(n, F), is the group of
n n matrices with entries from F and non-zero determinant.
ii) The special linear group of degree n over F, SL(n, F), is the subgroup of
GL(n, F) of matrices of determinant one.
iii) The projective general linear group of degree n over F, PGL(n, F), is the
quotient group of GL(n, F) by its centre Z(GL(n, F)).
iv) The projective special linear group of degree n over F, PSL(n, F), is the
quotient group of SL(n, F) by its centre Z(SL(n, F)).
The denition of the special linear group is concrete enough for us to know
what its elements are, but this is not the case with the projective versions. So
we start by identifying Z(GL(n, F)) and Z(SL(n, F)).
Lemma 8.16. Let F be a eld.
i) Z(GL(n, F)) = I
n
[ F 0 where I
n
is the nn identity matrix.
ii) Z(SL(n, F)) = Z(GL(n, F)) SL(n, F) = I
n
[ F 0,
n
= 1.
8.3. PROJECTIVE SPECIAL LINEAR GROUPS 81
Proof: The rst part is linear algebra; the second follows immediately.
Notice that although SL(n, F) is a subgroup of GL(n, F), PSL(n, F) is not
a subgroup of PGL(n, F). However, there is an obvious natural isomorphism of
PSL(n, F) with a normal subgroup of PGL(n, F).
Of particular importance to us will be the case of F a nite eld. Recall that
nite elds have prime power order and that any two nite elds of the same
order are isomorphic. If [F[ = q, we will write GL(n, q) instead of GL(n, F),
and similarly for the other matrix groups above. If F is a nite eld of order q
then GL(n, q) is nite (and hence the other matrix groups are too) and we can
give formul for the orders of these groups.
Lemma 8.17. Let F be a eld of order q and let t be the number of roots of
the polynomial x
n
1 over F. Then t = (n, q 1), the highest common factor
of n and q 1.
i) [GL(n, q)[ =
n1
i=0
(q
n
q
i
),
ii) [SL(n, q)[ = [GL(n, q)[/(q 1),
iii) [PGL(n, q)[ = [GL(n, q)[/(q 1) and
iv) [PSL(n, q)[ = [SL(n, q)[/t =
1
(n,q1)
q
n(n1)/2
n
i=2
(q
i
1).
Proof: Omitted.
These groups are more familiar to you than they might appear at rst sight.
Examples 8.18.
i) GL(1, F)

= F
= F 0 and so PGL(1, F)

= SL(1, F)

= PSL(1, F)

=
e.
ii) Z(SL(2, C)) = I
2
, I
2
, since the only two square roots of unity in C (or
indeed, any eld) are 1 and 1. In fact, one can show that PSL(2, C)

=
PGL(2, C). This group is known to you already: it is the group of Mobius
transformations of the Riemann sphere (the extended complex plane).
iii) Taking F = F
2
to be the eld of two elements (i.e. Z
2
with addition
and multiplication modulo 2), we have the nite group PSL(2, 2) of order
((2
2
1)(2
2
2
1
)/(2 1))/(1) = 6. One can easily show that this group
is not Abelian hence, by our previous classication, is isomorphic to S
3
.
iv) With F = F
2
again, q 1 = 1 and (n, q 1) = (n, 1) = 1, so all four of
GL(n, 2), SL(n, 2), PGL(n, 2) and PSL(n, 2) have the same order. In fact,
we see that Z(GL(n, 2)) = Z(SL(n, 2)) = e so GL(n, 2) = SL(n, 2),
PGL(n, 2) = PSL(n, 2) and there is a canonical isomorphism between
these pairs (the isomorphism G/e

= G).
v) Taking F = F
3
, the eld of three elements, [PSL(2, 3)[ = 12 and one may
show that PSL(2, 3) is isomorphic to A
4
.
The terminology projective . . . refers to the fact that these groups act on
projective spaces. Spaces such as C
n
and R
n
are called ane spaces and pro-
jective spaces are quotients of these. One major geometric reason for studying
projective space rather than ane space is that projective spaces are compact
topological spaces. The general denition of projective n-space is as follows.
Figure 8.1: The Fano plane.
Denition 8.19. Let F be a eld. Then the projective n-space over F, FP
n
,
is the quotient of F
n+1
0 by the equivalence relation given by
v w F 0 such that v = w.
Equivalently, projective n-space is the space of lines in F
n+1
passing through
the origin.
Then, just as GL(n, F) is the group of automorphisms of F
n
, PGL(n+1, F)
is the group of automorphisms of FP
n
. In particular, PGL(2, C) is the group of
automorphisms of CP
1
(but we knew this already: CP
1
is the Riemann sphere
and PGL(2, C) is Mobius transformations).
The smallest projective plane is F
2
P
2
and, from the above, its automorphism
group is PGL(3, 2) (= GL(3, 2) = . . . ). The projective plane F
2
P
2
is well-known
and usually goes by the name of the Fano plane. Its points are in bijection
with the elements of F
3
2
0 so there are seven of them. The lines are as in
Figure 8.1. The action of GL(3, 2) on the points of the Fano plane is simply by
matrix multiplication on the vector indexing the point. We also note that by
a coincidence, GL(3, 2)

= PSL(2, 7) but the presence of the 7 should now
not be such a surprise!
For us, the primary interest is in the projective special linear groups over
nite elds, as nite groups. We note that the isomorphisms PSL(2, 2)

= S
3
and
PSL(2, 3)

= A
4
show that these groups are not simple. However, the following
isomorphisms suggest that this is not typical: PSL(2, 4)

= PSL(2, 5)

= A
5
,
PSL(2, 9)

= A
6
and PSL(4, 2)

= A
8
, so these groups are all simple.
We now look at the rst PSL(2, q) not already discussed: this is PSL(2, 7).
There is no PSL(2, 6), as 6 is not a prime power so there is no eld of order 6.
Theorem 8.20 ([ST, Section 4.1]). The group PSL(2, 7) of order 168 is simple.
8.3. PROJECTIVE SPECIAL LINEAR GROUPS 83
Proof: First note that 168 = 2
3
3 7. Let N be a maximal non-trivial proper
normal subgroup of G, so G/N is simple and of order at most 168/2 = 84. Since
the only non-Abelian simple group of order at most 84 is A
5
, of order 60, but
60 168, G/N must be an Abelian simple group, hence cyclic of order 2, 3 or 7.
If M SL(2, 7), let [M] represent the corresponding coset Z(SL(2, 7))M in
PSL(2, 7). Let
x =
__
1 1
0 1
__
and y =
__
1 0
1 1
__
.
Then x and y have order 7, by an easy calculation, and if we set u = xy
2
x, so
u =
__
3 4
2 3
__
,
then we have an element of order 3 (another easy calculation). Now since
[ x[ = [ y [ = 7 and y / x, G has at least 2 Sylow 7-subgroups.
If [G/N[ = 2, N would be a normal subgroup of order 84 and index 2. By
Sylows theorems, a group of order 84 has only one Sylow 7-subgroup so, since
this would be a Sylow 7-subgroup of G conjugate (in G, since N G) only to
itself, G would have a unique Sylow 7-subgroup. This contradiction tells us that
[G/N[ ,= 2.
If [G/N[ = 3, then x, y N so u Ncontradiction since 3 [N[ = 56 in
this case. Hence [G/N[ = 7. Now by Sylows theorems, N has either 1 or 3
Sylow 2-subgroups, and these are all the Sylow 2-subgroups of G, since N G.
If G has a unique Sylow 2-subgroup S, then S G and G/S, of order 21, has
a unique Sylow 7-subgroup, of index 3. Such a subgroup of G/S would give rise
to a normal subgroup of G of index 3, but we have discounted this possibility
already.
Therefore G has 3 Sylow 2-subgroups and if we let H be the normalizer of
one of these, H has index 3 in G, since [G : N
G
(P)[ = [Syl
p
(G)[ if P Syl
p
(G).
Since we have ruled out the existence of normal index 3 subgroups, H is not
normal so the action of G on cos(G : H) must yield a homomorphism from G
to S
3
whose kernel has index strictly more than 3. So G maps onto S
3
by this
homomorphism. But S
3
has a subgroup of index 2 which would give rise to a
subgroup of index 2 in G. Since all subgroups of index 2 are normal, we would
have a normal subgroup of order 84, but we eliminated this possibility already.
Hence G has no non-trivial proper normal subgroup, so it is simple.
As noted at the end of the previous section, the list of non-Abelian nite sim-
ple groups begins A
5
, PSL(2, 7), A
6
(where we prefer to label the rst and third
of these by their alternating group name, rather than (one of) their projective
special linear group name(s)). Their orders are 60, 168 and 360 respectively.
We know that the alternating groups of degree at least 5 continue to contribute
to this sequence, but what about the projective special linear groups? Do these
stop being simple at some point, or are they all alternating groups? Well, no
and no (the latter by looking at their orders).
Theorem 8.21. For any n 2 and q a prime power, the group PSL(n, q) is
simple, of order given by the formula in Lemma 8.17, unless (n, q) = (2, 2) or
(2, 3).
We remark that the proof does not proceed in the way we did for PSL(2, 7).
Instead, one identies certain elements with nice properties and uses linear
algebra and geometrical arguments.
The projective special linear groups are part of a family called the groups
of Lie type and the nite simple members of this family make up the third and
nal innite class of nite simple groups. Groups of Lie type are matrix groups
over nite elds (including orthogonal and unitary groups, among others) or
twisted versions of these. We conclude this course by looking at the result we
have been aiming towards, the classication of nite simple groups.
8.4 The classication of nite simple groups
The full classication theorem would take a signicant amount of time and space
to state in detail but one gets an idea of the avour of the theorem from the
following summary.
Theorem 8.22. Every nite simple group is isomorphic to one of the following:
i) a cyclic group of prime order,
ii) an alternating group A
n
, for n 5,
iii) a simple group of Lie type over F
q
, or
iv) the 26 sporadic simple groups.
The rst two parts of this theorem have been covered in detail already and
the third was introduced in the previous section. In particular, the groups
PSL(n, q) are simple groups of Lie type. Some other Lie type groups can be
constructed as matrix groups preserving inner products of various sorts (sym-
metric inner products give orthogonal groups, skew-symmetric ones symplectic
groups, Hermitian ones unitary groups).
There is another approach, due to Chevalley, involving Lie algebras. A Lie
algebra over a eld F is an F-vector space L with a bilinear map L L L
called the Lie bracket, satisfying
i) [ x, x] = 0 and
ii) [ [ x, y ], z ] + [ [ y, z ], x] + [ [ z, x], y ] = 0.
Then one can dene an ideal of a Lie algebra L to be a subspace I such that
i I, x L implies [ i, x] I, and we say L is simple if it contains no proper
ideals. The nite-dimensional simple Lie algebras over C were classied by
Cartan and Killing and these are labelled as follows: A
n
(n 1), B
n
(n 2),
C
n
(n 3), D
n
(n 4), E
6
, E
7
, E
8
, F
4
, G
2
.
Now these do not give nite simple groups directly but one can choose a
good basis for a simple complex Lie algebra L and then consider the Lie
algebra L(q) over F
q
given by the F
q
-span of this basis. Then Aut(L(q)) has
one non-cyclic composition factor, a nite simple group.
We have A
n
(q)

= PSL(n + 1, q), B
n
(q)

= P
2n+1
(q) (an orthogonal-type
group), C
n
(q)

= PSp
2n
(q) (a symplectic group) and D
n
(q) = P
2n
(q), and
groups associated to the E, F and G types. These are called the Chevalley
groups.
8.4. THE CLASSIFICATION OF FINITE SIMPLE GROUPS 85
These are not quite all the simple groups of Lie type: the remainder are
constructed as xed-point subgroups under certain automorphisms of the groups
above. These are called twisted groups and these together with the Chevalley
groups make up the simple groups of Lie type.
The 26 sporadic groups are:
the Mathieu groups, M
11
, M
12
, M
22
, M
23
and M
24
;
the Janko groups, J
1
, J
2
, J
3
and J
4
;
the Conway groups, Co
1
, Co
2
and Co
3
;
the Fischer groups, Fi
22
, Fi
23
and Fi
24
;
the HigmanSims group, HS;
the McLaughlin group, McL;
the Held group, He;
the Rudvalis group, Ru;
the Suzuki sporadic group, Suz;
the ONan group, O
N;
the HaradaNorton group, HN;
the Lyons group, Ly;
the Thompson group, Th;
the Baby Monster group, B; and
the FischerGriess Monster group, M.
The group M
11
is the smallest sporadic simple group, of order 7920, and the
group M, the largest of the sporadic simple groups, has order
2
46
3
20
5
9
7
6
11
2
13
3
17 19 23 29 31 41 47 59 71
= 808017424794512875886459904961710757005754368000000000.
The nite simple groups still have much to tell us, particularly with regard
to their representation theory. The study of nite simple groups is by no means
over, and certainly the study of nite groups is not. However, our study of
them is at an end. For further reading, a good starting place is [Neu]. Recent
books on the subject include [Ron] and [dS], both popular books but with the
professional reader in mind, also. Those interested in the developmental his-
tory of modern abstract algebra more generally, i.e. encompassing elds, rings,
algebraic geometry and more, should consider [Der].
Appendix A
The groups of order less
than 16
The new material in this chapter is non-examinable.
We will now identify every group (up to isomorphism) having order less than
16. We begin, logically enough, with order 1.
Order 1: there is a unique group of order 1, the trivial group, e.
We may cut down our task somewhat using the following corollary to Lagranges
theorem:
Corollary A.1. If [G[ = p is prime, then G

= C
p
, so up to isomorphism there
is a unique group of order p.
This disposes of [G[ = 2, 3, 5, 7, 11 and 13. Next we tackle squares of
prime numbers: recall that in Corollary 5.6 we proved that if [G[ = p
2
, p prime,
then G is Abelian. Then we may use the classication of nite Abelian groups
(Theorem 4.8). Hence
Order 4: G

= C
4
or C
2
C
2
Order 9: G

= C
9
or C
3
C
3
Next, we consider the case of [G[ = pq for p and q prime. We dealt with this
in Proposition 5.19.
Order 6: G

= C
6
or C
3
C
2
where C
3
C
2
= C
3
C
2
with
(b)(a) =
b
a = a
2
= a
1
(C
3
= a , C
2
= b ). But this is exactly that used to dene D
6
, i.e.
G

= C
6
or D
6
. The groups C
2
C
3
and S
3
(which we know to be of order
6) are in fact these: C
6

= C
2
C
3
, D
6

= S
3
. Since D
6
is not Abelian,
D
6
,
= C
6
.
Order 10: G

= C
10
or C
5
C
2
= C
5

C
2
with (b)(a) =
b
a = a
4
(choosing
r = 4, since r
q
= r
2
= 4
2
1 (mod 5)). Again, the latter is dihedral and
thus is D
10
, and D
10
,
= C
10
.
86
87
Order 14: G

= C
14
or D
14
as above and these are not isomorphic.
Remark A.2. In fact, q = 2 in Proposition 5.19 gives:
Corollary A.3. Let G be a group of order 2p for p prime. Then G

= C
2p
or D
2p
(and C
2p
,
= D
2p
).
Order 15: G

= C
15
(by Proposition 5.19: 3 (5 1)).
So, we have dealt with thirteen of fteen orders. The remaining orders, 8
and 12, have three prime factors (including multiplicities) and are not directly
covered by our results to date, so we have to do some work. We can deal with
the Abelian groups rst, though.
Order 8:
Abelian: the classication yields C
8
, C
2
C
4
and C
2
C
2
C
2
, pairwise
non-isomorphic.
Non-Abelian: if Gis non-Abelian of order 8, then there exists no element
of order 8 (else G is cyclic and hence Abelian) but some element must
have order not 2 (else gh = g
1
h
1
= (hg)
1
= hg for all g, h so G
is Abelian) and hence this element has order 4. Let i be an element
of order 4 and let A = i . Since [G : A[ = 2, A is normal in G. Now
if G A contains an element j of order 2, then i
j
= j
1
ij = i
1
(else
i
j
= i and G is Abelian). Hence G

= D
8
.
It is possible to show, as below, that there is at most one most group
to add to our list of isomorphism types of groups of order 8.
However, we have not proved the existence of such a group. For
all we know, a contradiction may lurk somewhere in the argument
which follows. So we give a concrete realisation of this group, which
we call Q, the quaternion group (for reasons to be explained shortly).
Consider the subgroup of S
8
generated by
I = (1 2 3 4)(5 6 7 8)
and
J = (1 5 3 7)(2 8 4 6).
Then let
K = IJ = (1 8 3 6)(2 7 4 5)
and
Z = I
2
= J
2
= K
2
= (1 3)(2 4)(5 7)(6 8).
We see that Q = I, J = , I, I
2
, I
3
, J, IJ, I
2
J, I
3
J has exactly
the properties of the group we sought. This group is called the group
of quaternions. The algebra of quaternions H is a 4-dimensional real
vector space, spanned by the basis 1, i, j, k, with multiplication sat-
isfying the relations i
2
= j
2
= k
2
= ijk = 1. The quaternions are a
generalisation of the complex numbers (C span1, i), discovered
by William Rowan Hamilton in 1843 (hence the use of H as symbol
88 APPENDIX A. THE GROUPS OF ORDER LESS THAN 16
for the quaternions). The algebra H is a non-commutative division
algebra and Q = i, j .
Coming back to the group Q, we have a matrix representation,
by matrices in M
2
(C),
1
H
=
_
1 0
0 1
_
, i =
_
i 0
0 i
_
, j =
_
0 1
1 0
_
, k =
_
0 i
i 0
_
(where inside the matrices i means
1 as usual in C)
Then Q

= 1
H
, i, j, k.
Alternatively, G A must only contain elements of order 4: let
j G A and B = j , [B[ = 4. Then [AB[ = 1 or 2 (not 4, since
A ,= B), but if [A B[ = 1 then [AB[ = [A[[B[/[A B[ = 16, which
is absurd since AB is a subset of G. So [A B[ = 2 and [AB[ = 8.
But then we must have AB = G so AB = BA. Furthermore, if z
is the non-trivial element of A B then zA = Az (A Abelian) and
zB = Bz so zAB = ABz, i.e. zG = Gz and z is central.
Now k = ij / A (else j = i
3
ij A) so k has order 4. Note
that G = AAj and we know the structure of A (A = i is cyclic),
so the structure of G will be determined once we can express ji as
an element of A Aj. G contains an element of order 2, namely
z ([A B[ = z = 2), but this is the only one: z is the only
element of order 2 in A (else A is not cyclic) and there are none
in G A by assumption (we dealt with the alternative above). So
we see that z = i
2
= j
2
= k
2
since i, j, k have order 4 and thus
their squares have order 2. In particular, k
2
= ijij = j
2
so that
ji = (i
1
)(j
2
)(j
1
) = i
3
j
2
j
3
= i
3
j (= zij). To see that we really
have determined the structure of G, observe that we have shown that
G = i
m
j
n
[ m 0, 1, 2, 3, n 0, 1 (n = 0 gives elements of A)
and to multiply two such expressions, concatenate them and use the
relation ji = i
3
j successively to simplify. For example,
(ij)(i
2
j) = i(ji)(ij) = i(i
3
j)(ij) = jij = (i
3
j)j = i
3
i
2
= i
Then G is isomorphic to Q. The element z is 1
H
, which corresponds
to Z. As stated previously, a presentation for Q is
Q = a, b [ a
4
, a
2
= b
2
, b
1
ab = a
1
.
To summarise, we have ve isomorphism types of groups of order 8:
C
8
, C
2
C
4
, C
2
C
2
C
2
, D
8
and Q.
This leaves only order 12. We will use a dierent method to that for order
8, based on the fact (which we rst establish) that a group of order 12 is a
semi-direct product. We then look at various automorphism groups to see what
possible semi-direct products can exist.
Order 12:
Abelian: C
12
, C
2
C
6
89
Non-Abelian: Let G be a group of order 12. From Sylows theorems,
we have that [Syl
2
(G)[ is odd and divides 3, so is 1 or 3 and similarly
[Syl
3
(G)[ 1 (mod 3) and divides 4, so is 1 or 4. It is not possible to
have [Syl
2
(G)[ = 3 and [Syl
3
(G)[ = 4, as this would require at least 4
elements of order 2 and 8 elements of order 3, as well as the identity,
totalling more than 12. Hence there is exactly one Sylow 2-subgroup
or exactly one Sylow 3-subgroup.
Now let p denote the prime for which there is exactly one Sylow
p-subgroup, and q the other prime dividing 12. Then the unique
Sylow p-subgroup must be normal, its intersection with any Sylow q-
subgroup must be trivial and its product with any Sylow q-subgroup
must be the whole of G. Hence G is a semi-direct product.
By our earlier classications, a Sylow 2-subgroup, being of order
4, must be isomorphic to either C
4
or C
2
C
2
. A Sylow 3-subgroup
must be isomorphic to C
3
. Therefore, to classify the groups of order
12, we must nd all possible semi-direct products involving these
groups. For this, we need to know their automorphism groups: these
are as follows.
Aut(C
3
)

= C
2
Aut(C
2
C
2
)

= S
3
Aut(C
4
)

= C
2
We also use Proposition 3.42 to tell us about isomorphisms.
C
4
C
3
: there can be no non-trivial homomorphisms from C
3
to
Aut(C
4
)

= C
2
, as any non-trivial homomorphism would have
non-trivial kernel and thus be injective, which is impossible since
[C
3
[ > [C
2
[. So the only semi-direct product of this form is in
fact direct, so we obtain C
4
C
3
(which we already had in our
Abelian classication, since this is isomorphic to C
12
).
(C
2
C
2
) C
3
: since Aut(C
2
C
2
)

= S
3
has a unique subgroup of
order 3, any two non-trivial (hence injective) homomorphisms
from C
3
to S
3
have the same image and so dene isomorphic
semi-direct products by Proposition 3.42. Now A
4
has a unique
Sylow 2-subgroup isomorphic to C
2
C
2
and hence is of the right
form, so we must have (C
2
C
2
) C
3

= A
4
.
C
3
(C
2
C
2
): since Aut(C
3
)

= C
2
there are three non-trivial in-
jective homomorphisms from C
2
C
2
to Aut(C
3
)

= C
2
but again
they dene isomorphic semi-direct products by Proposition 3.42.
This time, the candidate D
12
can be seen to have a unique Sylow
3-subgroup isomorphic to C
3
and thus C
3
(C
2
C
2
)

= D
12
.
C
3
C
4
: there is exactly one non-trivial homomorphism from C
4
to
Aut(C
3
)

= C
2
(it is the natural projection onto the quotient
of C
4
by the subgroup x
2
). This semi-direct product is not
isomorphic to A
4
or D
12
and completes our list of groups of order
12. It may be realised as the subgroup (5 6 7), (1 2 3 4)(6 7)
of S
7
.
Hence the non-Abelian groups of order 12 are A
4
, D
12
and C
3
C
4
.
90 APPENDIX A. THE GROUPS OF ORDER LESS THAN 16
Hence the full list of isomorphism types of groups of order less than 16 is as
given in Table 8.1 on page 76.
Continuing in this way is clearly not very practical, or at least not by hand.
For a very rough upper bound on the number of non-isomorphic groups of
a given order, recall that non-isomorphic groups have dierent multiplication
tables (also called Cayley tables). So the number of dierent possible n n
arrays of n elements is an upper bound on the number of isomorphism types,
though few arrays will actually describe groups. Unfortunately for order n,
this bound is n
n
2
, which is rather large, to put it mildly. It is nite, though,
which is something! For n = 8, 8
64
6.2 10
57
whereas the actual number is
5something of an over-estimate.
Hard work by many mathematicians has established the following:
there are 49,487,365,422 non-isomorphic groups of order 2
10
= 1024.
there are 423,164,062 pairwise non-isomorphic groups of order at most
2000 but not 1024.
the majority of these have order 1536.
Indeed a data library of the groups of order at most 2000 but not 1024 has been
constructed for the computer programs GAP and Magma (by Besche, Eick and
OBrien).
Index
A, ring of algebraic integers, 71
A
n
, alternating group
of degree n, 78
Aut(G), automorphism group, 25
C
G
(h), centralizer of h G, 11, 41
C
n
, the cyclic group
of order n, 14
C
, the innite cyclic group, 14

D
2n
, the dihedral group
of order 2n, 26
D
, the innite dihedral group, 27

FP
n
, projective n-space, 82
Fix
G
(), global xed points
for the action of G on , 11
H, the algebra of quaternions, 87
N
G
(H), normalizer of H in G, 11, 42
Orb(), orbit of , 10
Q, the quaternion group, 87
S
n
, symmetric group of degree n, 78
Stab(x), stabilizer of x, 10
Syl
p
(G), set of Sylow p-subgroups, 43
Z(G), the centre of G, 41
cos(G : H), coset space of H in G, 10
o(x), the order of an element, 14, 18
GL(n, F), general linear group, 80
PGL(n, F), projective general
linear group, 80
PSL(n, F), projective special
linear group, 80
SL(n, F), special linear group, 80
[ a, b ], the commutator of a and b, 50
, external direct product, 22
(G), derived length, 52
X [ R, presentation of a group
with generating set X
and relations R, 21
X, subgroup generated by X, 17
[G : H[, index of H in G, 9
, denoting normal subgroup, 9
, denoting proper normal
subgroup, 9
<, denoting proper subgroup, 9
G/N, quotient of G by N, 9
, internal semi-direct product, 24
, external semi-direct product, 25

, denoting subgroup, 7
(A), torsion subgroup, 39
, wreath product, 27
kG, group algebra, 56
[G[, the order of a group G, 7
action of a group on a set, 10
algebra of quaternions, 87
algebraic
integer, 71
number, 72
alternating group of degree n, A
n
, 78
automorphism group, Aut(G), 25
Burnsides p
theorem, 54, 73
Cauchys theorem, 40
centralizer of an element, 11, 41
centre of G, Z(G), 41
character
degree of, 62
irreducible, 61
lift of, 68
linear, 62
of group, 61
of module, 61
of representation, 61
permutation, 62
regular, 62
table, 66
theory, 6174
trivial, 61
character theory, 6174
class function, 62
classication
of nite Abelian groups, 37
of nite simple groups, 84
91
92 INDEX
of nitely generated
Abelian groups, 39
column orthogonality relations, 66
commutator, 23
commutator subgroup, 51
completely reducible module, 58
composition
factors, 30
series, 30
coset space, 10
countably generated, 18
cyclic, 18
group, 1415
group of order n, C
n
, 14
degree of character, 62
derived
group, 51
length, 52
series, 52
dihedral group
D
of innite order, 27
D
n
of order 2n, 26
generalised, 27
direct product, 2224
dual group, 67
extension, 2729
external direct product, 22
external semi-direct product, 25
factors of a subnormal series, 30
factors, composition, 30
faithful
module, 59
representation, 59
Fano plane, 82
FeitThompson theorem, 54
nite Abelian groups, 3438
nitely generated, 18
Abelian groups, 3839
rst isomorphism theorem, 9
free group, 19
Galois group, 54
general linear group, GL(n, F), 80
generalised dihedral group, 27
global xed points for an action, 11
group, 6
action, 10
algebra, 56
homomorphism, 9
isomorphism, 9
property, 42
representations, 5661
homomorphism of groups, 9
image of a homomorphism, 9
innite cyclic group, C
, 14
innite dihedral group, D
, 27
internal direct product, 23
internal semi-direct product, 24
irreducible character, 61
isomorphism theorems, 910
JordanHolder theorem, 31
kernel of a homomorphism, 9
length of a series, 30
lift of character, 68
linear
character, 62
representation, 57
Maschkes theorem, 58
matrix representation, 57
maximal subgroup, 42
module
category, 57
for an algebra, 57
homomorphism, 57
n-generator group, 18
normal
closure, 21
series, 30
subgroup, 9
normalizer of a subgroup,
N
G
(H), 11, 42
orbit, 10
order
of a group, 7
of an element, o(x), 14, 18
p-group, 35, 4142
permutation character, 62
poly-P group, 49
polyabelian, 49
INDEX 93
polycyclic, 49
presentation, 1922
Primary Decomposition Theorem
for nite Abelian groups, 35
product of subsets of a group, 22
projective
general linear group,
PGL(n, F), 80
space, 82
special linear group,
PSL(n, F), 80
property of a group, 42
quaternion group, 87
regular
character, 62
module, 59
ring of algebraic integers, A, 71
row orthogonality relations, 64, 66
second isomorphism theorem, 10
semi-direct product, 2427
external, 25
internal, 24
series, 30
composition, 30
derived, 52
normal, 30
subnormal, 30
short exact sequence, 27
simple, 75
simple module, 57
soluble, 50
soluble by radicals, 55
special linear group, SL(n, F), 80
split extension, 29
stabilizer, 10
subgroup generated by a subset, 17
subnormal series, 30
Sylow p-subgroup, 43
Sylows theorems, 43
symmetric group of degree n, S
n
, 78
third isomorphism theorem, 10
torsion subgroup, (A), 39
torsion-free, 39
transitive action, 10
trivial character, 61
wreath product, 27

B2bGroupTheoryHT11 Notes PDF

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

B2bGroupTheoryHT11 Notes PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

B2bGroupTheoryHT11 Notes PDF

Uploaded by

Copyright:

Available Formats

Group Theory

theorem, which states that all groups

innite dihedral group

group of non-zero rational numbers under multiplication

group of non-zero real numbers under multiplication

group of non-zero complex numbers under multiplication

for some n. The group operation is given by a

= Q 0 under multiplication). Observe

. The set underlying this subgroup is

. This is the precise

is also free on the set X then F

: F H. The set map j : X F extends

: F F. But i : X F extends to the

is also free on X, let i : X F and i

. Similarly, i extends to a homomorphism

F. We wish to prove that f f

(x)) = f(x) = x so (ff

in X and (formal) inverses of elements,

: a word is reduced if it contains no adjacent pair of mutually inverse

be a subset of the set of

. Dene the group X [ R to be the quotient group

such that G = X [ R, we say X [ R

B is non-Abelian, since there must exist some a A, b B such

B is the internal semi-direct

B are isomorphic and indeed in

G by ((h, g)) = (h, (g)). To distinguish the

a for (b)(a) and (b)(a), respectively.

G, ((h, g)) = (h,

B for some : B Aut(A).

does not have a composition series.

being the rst obvious exampleit is no surprise that

. No other innite Abelian

such that 1 < d

[ < [G[ also divisible by p does contain an

is in fact maximal (i.e. Sylow) then

is unique, any action of Q on P by left conjugation is an injective

. Thus by Proposition 3.42, all the resulting

H G then H G and G/H is Abelian.

, we then have a homomorphism (G

) n. However we showed that

theorem). If p and q are primes then any

as) of varying degrees. Then roots

theorem, which we will

(g)[ for all g G.

is the multiplicative group of the eld C. The trivial

given by taking the sign of a permutation is a linear character of S

are the linear characters.

(g)[). This is a general phenomenon: for any subgroup

(g) = tr(g ) = tr(Ng ) =

is the derived group of G.

has m linear characters

in their kernel. Now if

theorem and, along the way, demonstrate a pair of very

m of a natural number m (consider (

theorem, [JL, Thm. 29.4]). If p and q are

and Z(Q) ,= e, since Q is a non-trivial q-group. Let

. By induction on +, N and G/N are soluble and hence