Notes for a MSc Course in

Group Theory

Andrea Caranti
Dipartimento di Matematica, Università degli Studi di Trento,
via Sommarive 14, 38123 Trento
E-mail address: andrea.caranti@unitn.it
URL: http://www.science.unitn.it/∼caranti/
 Introduction

In the Spring semester of 2018–19 I am giving for the first time a course (6
ECTS) in “Advanced Group Theory” for the MSc in Mathematics in Trento (the
MSc is in English, in case you were wondering). This is meant as a second course
in group theory, after a 6 ECTS course in Group Theory in the BSc, which followed
a 12 ECTS basic course in Algebra.
Before and during the course, I plan to write up some notes, which are what
you are reading right now. The most recent version can be downloaded at
http://www.science.unitn.it/∼caranti/
Useful Texts
Among the many good texts in group theory, I recommend
• [Mac12] (I graduated with the author);
• [Hup67], an excellent text in German;
• [Gor80], another great classic;
• [Rob96], complete and crystal clear;
• [Rot95], a nice selection of arguments;
• [Ser78], a compact classic;
• [Ser16], a bit intense, but magnificent.
For the course, in the end I am mainly using [Ser16] and sometimes [Ser78].
To be written up next
Expand Subsection 6.3.
Determination of character tables, employing among others the following.
The irreducible characters are a basis for the space of class functions. Gener-
alised characters.
Linear characters. Characters from quotients. Kernel of a character.
Permutation characters. Induced characters and Frobenius reciprocity. M-
groups (are soluble).
Every non-linear character has a zero.
Normal subgroups and Clifford?
Frobenius groups? (Want to focus on applications to group theory.)

3
 Contents

Introduction 3
Useful Texts 3
To be written up next 3

Part 1. Preliminaries 7
Chapter 1. Preliminaries 9
1.1. Linear maps 9
1.2. The trace 9
1.3. Centre of the matrix algebra 9
1.4. Simultaneous diagonalization 11
1.5. Inner products 11
1.6. The class equation and the centre of finite p-groups 12
1.7. Endomorphisms, automorphisms and inner automorphisms 12
1.8. Elementary abelian groups 13
1.9. Integrality and algebraic integers 14

Part 2. Soluble and nilpotent groups 17

Chapter 2. Series 19
2.1. Series 19
2.2. Ω-groups and Ω-series 19
2.3. Ω-composition series 20
2.4. Uniqueness of the factors of an Ω-composition series 22
Chapter 3. Soluble groups 23
3.1. Commutators 23
3.2. The derived series and soluble groups 25
Chapter 4. Nilpotent groups 27
4.1. Central series and nilpotent groups 27
4.2. Finite nilpotent groups 28
4.3. Nilpotent groups are soluble 29

Part 3. Representations and Characters 31

Chapter 5. Representations 33
5.1. Inner products 33
5.2. Examples 34
5
6 CONTENTS

5.3. The group algebra 35

5.4. Maschke’s Theorem 36
5.5. Irreducible representations 37
5.6. Schur’s Lemma 38
5.7. Orthogonality Relations 39
Chapter 6. Characters 41
6.1. Characters 41
6.2. Decomposing the regular representation 42
6.3. Number of irreducible characters 43
6.4. Character tables 44
6.5. Characters of abelian groups 44
6.6. Character tables of small groups 44
6.7. Integrality 45
Chapter 7. Applications 47
7.1. Burnside pa q b 47
Bibliography 49
 Part 1

Preliminaries
 CHAPTER 1

Preliminaries

Recalling some basic stuff, and establishing notation.

1.1. Linear maps

If V is a vector space over some field F , then GL(V ) denotes the general linear
group of invertible linear maps on V .
If V is of finite dimension n, and we fix a basis, then the elements of GL(V )
can be represented via n × n matrices. We denote by GL(n, F ) the corresponding
group of matrices, which acts on the space F n of row vectors. When F is finite of
order q (this is not going to happen in these notes), one writes also GL(n, q).
1.1.1. Lemma. Let V be a finitely generated vector space.
Let π be a linear map on V such that π 2 = π.
Then
V = ker(π) ⊕ V π,
and π is the projection on V π along ker(π).
Proof. Let v ∈ V . Then (v − vπ)π = vπ − vπ 2 = 0, so that v − vπ ∈ ker(π).
If v ∈ ker(π) ∩ V π, then v = uπ for some u ∈ V , and thus v = uπ = uπ 2 = vπ =
0. □

1.2. The trace

It is a well-known and elementary fact that the trace (of a square matrix)
satisfies trace(AB) = trace(BA) for all matrices A, B. In particular, if C is
invertible we have
trace(C −1 AC) = trace(C −1 (AC)) = trace((AC)C −1 ) = trace(A).
Note actually that this reminds us that the trace is defined for a linear map on a
vector space, and it does not depend on the base with respect to which one writes
it down as a matrix.

1.3. Centre of the matrix algebra

1.3.1. Definition. An algebra over the field F is a ring A with unity, which
is also a vector space over F (with respect to the same “+” operation), such that
for a, b ∈ A, λ ∈ F ,
(1.3.1) λ(ab) = (λa)b = a(λb).
In other words, the ring product is bilinear.
9
10 1. PRELIMINARIES

Note that if we take a = 1 in (1.3.1), we obtain

(1.3.2) λb = (λ · 1)b.
It is easy to see that
F →A
λ 7→ λ · 1
is an injective morphism of rings with unity. Therefore F · 1 = { λ · 1 : λ ∈ F } is
a subring of A isomorphic to F , which is often identified with F , because (1.3.2)
shows that scalar multiplication by λ, or multiplication by λ · 1 in A, are the very
same thing.
1.3.2. Example. The n × n matrices over a field A are an algebra. The
identification just mentioned means that we consider λ ∈ F and the scalar matrix
λ · I with λ on the diagonal as the same thing.
Recall that if G is a group, then its centre is
Z(G) = { z ∈ G : gz = zg, for all g ∈ G } .
If A is an algebra, its centre is
Z(A) = { z ∈ A : az = za, for all a ∈ A } .
1.3.3. Exercise. Show that the centre of the algebra of n × n matrices over a
field F consists of the scalar matrices.
Proof. Let A be the algebra of n × n matrices, and z ∈ Z(A). Proceeding
by way of contradiction, assume there is a vector v ̸= 0 such that v and vz are
independent. Complete v, vz to a basis, and consider the linear map T that is zero
on all basis vectors, except that it takes vz to v. Then v(zT ) = (vz)T = v, but
v(T z) = (vT )z = 0z = 0 ̸= v, a contradiction.
Thus for all 0 ̸= v ∈ V there is a uniquely defined a(v) ∈ F such that vz =
a(v)z. (We are saying that all non-zero vectors are eigenvectors for z.) If v, w are
independent vectors then
a(v + w)v + a(v + w)w = a(v + w)(v + w) = (v + w)z =
= vz + wz = a(v)v + a(w)w,
whence
a(v) = a(v + w) = a(w),
and z is a scalar matrix. □
1.3.4. Exercise. Find other proofs for this elementary, but basic, fact. (Books
are OK, but Internet search is also allowed.)
1.3.1. Centre of the general linear group. GL(V ), the general linear
group, is the group of invertible linear maps on the finite-dimensional vector space
V . Its centre consists of the non-zero scalar matrix. A proof is a variant of the
previous one, in which one takes at T the linear map that is the identity on all
basis elements, except that (vz)T = vz + v. One checks easily that this is indeed
invertible, and then vT z = vz ̸= vz + v = vzT .
 1.5. INNER PRODUCTS 11

1.4. Simultaneous diagonalization

Let Ao , A1 , . . . , Ak be n × n matrices over a field F , such that each of them
is diagonalizable, and the matrices commute pairwise, that is Ai Aj = Aj Ai . then
there is a basis with respect to which all Ai are diagonal.
Let V = F n . For each eigenvalue λ of B = A0 , consider the eigenspace
V (λ) = { v ∈ V : vB = λv }. By hypotesis, V is a direct sum of the V (λ).
We claim that for each λ and i ≥ 1 we have V (λ)Ai ⊆ V (λ). In fact if v ∈ V (λ)
one has
(vAi )B = v(Ai B) = v(BAi ) = (vB)Ai = λ(vAi ),
so that vAi ∈ V (λ). By induction on k, the restrictions of the Ai , for i ≥ 1 to
V (λ) can be simultaneously diagonalized. As the restriction of B to V (λ) is scalar
with respect to any base (see Section 1.3) , all the Ai are then diagonalized at
once.
1.5. Inner products
Let V be a finite-dimensional vector space over C. An inner product on V is
a map
⟨ ·, · ⟩ : V × V → C
which satisfies the following properties.
(1) For λ ∈ C, x, y, z ∈ V
⟨ x, λy ⟩ = λ ⟨ x, y ⟩ , ⟨ x, y + z ⟩ = ⟨ x, y ⟩ + ⟨ x, z ⟩ .
This states that ⟨ ·, · ⟩ is linear in the second component.
(2) For x, y ∈ V
⟨ x, y ⟩ = ⟨ y, x ⟩.
Here the bar denotes complex conjugation. This implies that
(3) for x ∈ V
⟨ x, x ⟩ = ⟨ x, x ⟩
is real, and
(4) ⟨ ·, · ⟩ is semilinear in the first component, that is, for λ ∈ C, x, y, z ∈ V
⟨ λx, y ⟩ = λ ⟨ x, y ⟩ , ⟨ x + y, z ⟩ = ⟨ x, z ⟩ + ⟨ y, z ⟩ .
The first part of (4) follows from
⟨ λx, y ⟩ = ⟨ y, λx ⟩ = λ ⟨ y, x ⟩ = λ⟨ y, x ⟩ = λ ⟨ x, y ⟩ .
(5) For x ∈ V
⟨ x, x ⟩ ≥ 0,
and
(6) for x ∈ V
⟨ x, x ⟩ = 0 if and only if x = 0.
n
If V = C , the standard inner product is given by
∑
n
⟨ x, y ⟩ = xi yi .
i=1
12 1. PRELIMINARIES

1.6. The class equation and the centre of finite p-groups

Let G be a finite group acting on itself by conjugation. Since the orbits (that
is, the conjugacy classes) form a partition, we have the class equation
∑
| G | = | Z(G) | + | G : CG (a) | ,
where the sum is over a set of representatives
 of the conjugacy classes aG ̸= { a },
so that each | G : CG (a) | =  aG  > 1.
 If G ̸= { 1 } is a finite p-group, then p | | G | and p divides each | G : CG (a) | =
 G
 a . Thus p | | Z(G) |, so that | Z(G) | > 1.

1.7. Endomorphisms, automorphisms and inner automorphisms

Let G be a group. A (homo)morphism G → G is called an endomorphism of
G. The set of all endomorphisms of a group G is denoted by End(G), and is a
monoid under the composition ◦ of map.
1.7.1. Lemma. If (G, +, 0) is an abelian group, then End(G) becomes a ring
under pointwise addition
g s+t = g s + g t , for g ∈ G and s, t ∈ End(G),
and composition of maps
g s◦t = (g s )t , for g ∈ G and s, t ∈ End(G).
1.7.2. Exercise. Prove the Lemma. The point where the hypothesis that G is
abelian plays a crucial role is when proving that if s, t ∈ End(G), then s + t ∈
End(G).
1.7.3. Exercise. If (G, ·, 1) is any group, and s, t are maps on G, one can
define a map s + t on G by g s+t = g s · g t .
Let I be the identity map on G, so that I ∈ End(G). Show that I +I ∈ End(G)
if and only if G is abelian.

(Hint: Note that for g ∈ G one has g I+I = g I · g I = g 2 . So one has to prove
that the map g 7→ g 2 is a homomorphism if and only if G is abelian.)
An invertible element of End(G), that is, an endomorphism which is a bijective
map, is called an automorphism of G. The automorphisms of G form thus a group
Aut(G).
For g ∈ G, the map
ι(g) : G → G
x 7→ g −1 xg
is an automorphism of G, the inner automorphism induced by g. The map
ι : G → Aut(G)
g 7→ ι(g)
is a group morphism.
 1.8. ELEMENTARY ABELIAN GROUPS 13

The image of ι is the group Inn(G) of the inner automorphisms of G

1.7.4. Exercise. Prove these statements.
(Hint: It is possibly better to start by proving that each ι(g) is an endomor-
phism of G. Then show that ι(gh) = ι(g)ι(h) and ι(1) = 1, and then show that
ι(g −1 ) = ι(g)−1 .)
1.7.5. Exercise. Prove that Inn(G) ⊴ Aut(G).
(Hint: Let g ∈ G and φ ∈ Aut(G). For x ∈ G we have
−1 ι(g)φ −1
= (g −1 xφ g)φ = (g φ )−1 xg φ = xι(g ) ,
φ φ
xι(g) = xφ
so that ι(g)φ = ι(g φ ) ∈ Inn(G).)
Now
{ }
ker(ι) = { g ∈ G : ι(g) = 1 } = g ∈ G : g −1 xg = x for all x ∈ G = Z(G),
the centre of G. The first isomorphism theorem implies
Inn(G) ∼
= G/Z(G).
1.8. Elementary abelian groups
Let p be a prime. A finite abelian group G such that xp = 1 for all x ∈ G is
said to be elementary abelian. So if G has order pn , it is isomorphic to the direct
product of n (cyclic) groups of order p.
1.8.1. Lemma. Let (V, +, 0) be an abelian group, and F a field. The following
are equivalent
(1) a structure of F -vector space on (V, +, 0), and
(2) a homomorphism of rings with unity F → End(V ).
Proof. Suppose φ : F → End(V ) is a homomorphism of rings with unity. For
a ∈ F and v ∈ V , denote by va = v φ(a) , the image of v under the endomorphism
φ(a). Let us see what this means.
Every φ(a) is an endomorphism of V , that is
(1) for u, v ∈ V and a ∈ F , one has
(u + v)a = (u + v)φ(a) = uφ(a) + v φ(a) = ua + va.
Then φ(a + b) = φ(a) + φ(b), that is
(2) for a, b ∈ F and v ∈ V one has
v(a + b) = v φ(a+b) = v φ(a)+φ(b) = v φ(a) + v φ(b) = va + vb.
We have used the fact that addition of f, g ∈ End(V ) is defined pointwise, that
is, for v ∈ V
v f +g = v f + v g .
Now multiplication in End(V ) is defined as composition of maps, so that φ(ab) =
φ(a) ◦ φ(b), that is
(3) for a, b ∈ F and v ∈ V one has
v(ab) = v φ(ab) = v φ(a)◦φ(b) = (v φ(a) )φ(b) = (va)φ(b) = (va)b.
14 1. PRELIMINARIES

Finally,
(4) v1 = v φ(1) = v I = v, where 1 denotes the unity of F and I the identity
map on V .
So we see that (1)–(4) are just the axioms of an F -vector space.
The converse is immediate. □
1.8.2. Theorem. Let (V, +, 0) be an elementary abelian group of order pn > 1,
where p is a prime. For a ∈ Z and v ∈ V , the notation va stands for the a-th
multiple of v.
Let F = Z/pZ be the field with p elements.
Then for a ∈ Z and v ∈ V the operation
(1.8.1) v(a + pZ) = va
defines a multiplication by scalars that makes V into an F -vector space.
Proof. The map
ψ : Z → End(V )
a 7→ (v 7→ va)
is a morphism of rings with unity, by the properties of multiples.
Since V ̸= { 0 }, and vp = 0 for all v ∈ V , its kernel is pZ. The first isomor-
phism theorem for rings yields a morphism of rings with unity as in (1.8.1), so
that we may appeal to lemma 1.8.1. □
Recall that if V is a finite-dimensional vector space over a field F , then GL(V )
denotes the group of invertible linear maps on the V . If F is the field with p
elements, then, once a basis is chosen, GL(V ) is isomorphic to the group GL(n, p)
of n × n matrices with non-zero determinant in F .
1.8.3. Proposition. Let (V, +, 0) be an elementary abelian group of order pn ,
where p is a prime, so that V can be regarded as a vector space on the field with p
elements.
Then the group Aut(V ) of automorphisms of V coincides with the group GL(V ) ∼ =
GL(n, p).
Proof. If β ∈ Aut(V ), then for a ∈ Z and v ∈ V one has
(v(a + pZ))β = (va)β = v β a = v β (a + pZ),
so that β is also a linear map on the vector space V . □

1.9. Integrality and algebraic integers

Let R be a ring of characteristic zero, so that Z ⊆ R. (This theory could be
developed in more generality, see [Ser16, 8.6.1].)
1.9.1. Proposition. For x ∈ R, the following are equivalent:
(1) there exists n ≥ 1 and a1 , . . . , an ∈ Z such that
xn + a1 xn−1 + · · · + an = 0.
 1.9. INTEGRALITY AND ALGEBRAIC INTEGERS 15

(2) The subring Z[x] of R is finitely generated as an abelian group, that is,
as a Z-module.
(3) The subring Z[x] of R is contained in a finitely generated Z-submodule of
R.
Proof. If (1) holds, then Z[x] is generated as a Z-module by 1, x, . . . , xn−1 ,
so that (2) holds.
Clearly (2) implies (3). We are thus left with proving that (3) implies (1).
The fact that (3) implies (2) in Proposition 1.9.1 follows from the general fact
that a submodule of a finitely generated Z-module (more generally, a module over
a PID) is finitely generated itself, see for instance [Jac85, Chapter 3].
□
1.9.2. Remark. See [Mar18, Theorem 2, p. 11] for a slick proof in the com-
mutative case. We will need to use these results for non-commutative rings R,
though, namely the group algebra.
1.9.3. Definition. An element x which satisfies the properties of Proposi-
tion 1.9.1 is said to be integral. If R = C, one speaks of an algebraic integer
1.9.4. Proposition. The set of the integral elements in R is a subring of R.
Proof. If x, y ∈ R are integral, then x, y ∈ Z[x] + Z[y], and the latter is
finitely generated as a Z-module.
If b1 , . . . , bs generate Z[x] as a Z-module, and c1 , . . . , ct generate Z[y] as a
Z-module, then xy is in the Z-module generated by the products bi cj . □
1.9.5. Remark. The proof could be made slicker using tensor products, which
I may have to use anyway for the product of characters.
1.9.6. Proposition. A rational number is an algebraic integer if and only if
it is an integer.
This result will turn out to be very useful to prove that b | a, for given integers
a and b ̸= 0, Just prove that the rational number a/b is an algebraic integer!
Proof. If p/q, with p, q coprime integers, is a root of
xn + a1 xn−1 + · · · + an = 0,
for ai ∈ Z, then
pn + a1 pn−1 q + · · · + an q n = 0,
so that q | p, and thus q = ±1. □
Clearly roots of unity (i.e., roots of xk − 1 for some k) are algebraic integers.
1.9.7. Lemma. Let ω1 , . . . , ωn ∈ C be roots of unity.
If
ω1 + · · · + ωn
(1.9.1) α=
n
is an algebraic integer, then either α = 0, or ω1 = · · · = ωn = α.
16 1. PRELIMINARIES

Proof. Consider the normal closure E of Q[α]. Thus E/Q is a Galois exten-
sion. An element g ∈ Gal(E/Q) maps roots of unity to roots of unity, so αg has
the same form.
Note that
| ω1 + · · · + ωn | | ω1 | + · · · + | ωn |
|α| = ≤ = 1,
n n
and thus the same holds for the conjugates αg .
Now the norm ∏
αg
g∈Gal(E/Q)
is fixed by Gal(E/Q), and thus is in Q. It is an algebraic integer, as a product of
algebraic integers, and thus it is an integer. Since it is a product of elements of
absolute value at most 1, the norm has absolute value at most 1.
Now there are two possibilities. Either the norm is zero, so that α = 0, or the
norm is 1, so
| ω1 + · · · + ωn | = n.
But this can hold only if all ωi are equal, see [Ser16, Lemma 8.6]. □
 Part 2

Soluble and nilpotent groups

 CHAPTER 2

Series

For further details of the arguments of this section, see [Rob96].

In this section G will be a finite group.
2.1. Series
A series in G is a sequence of distinct subgroups
(2.1.1) 1 = H0 ⊴ H1 ⊴ . . . ⊴ Hn = G,
each normal in the next.
One says that the series (2.1.1) is normal if each Hi ⊴ G.
2.1.1. Exercise. Show that the sequence
1 < ⟨ (12)(34) ⟩ < ⟨ (12)(34), (13)(24) ⟩ < S4
is a series which is not normal.
2.2. Ω-groups and Ω-series
The concept of an Ω-series is sometimes useful.
2.2.1. Definition. Let G be a group, Ω a set, and
α:G×Ω→G
a function.
A right operator group is a triple (G, Ω, α), such that for each ω ∈ Ω the map
g 7→ (g, ω)α
is an endomorphism of G.
One says that G is an Ω-group, and writes simply g ω for (g, ω)α.
An Ω-subgroup is a subgroup H ≤ G such that hω ∈ H for each h ∈ H and
ω ∈ Ω.
(1) If Ω = ∅, we get simply a group, and the Ω-subgroups of G are just the
subgroups of G.
(2) If Ω = Inn(G), then the Ω-subgroups are the subgroups that are normal
in G.
(3) If Ω = Aut(G), then the Ω-subgroups are the so-called characteristic
subgroups.
(4) If Ω = End(G), then the Ω-subgroups are the so-called fully invariant
subgroups.
2.2.2. Exercise. Find examples of groups G and subgroups H ≤ G such that
19
20 2. SERIES

(1) H is normal, but not characteristic in G.

(2) H is characteristic, but not fully invariant in G.

(Hint: For the first question, one can consider an elementary abelian group
of order p2 , where p is a prime, that is, a group of the form Cp × Cp , where Cp ,
where Cp is cyclic of order p.
For the second question, consider the group A5 , which is known to be simple,
and a group B = ⟨ b ⟩ of order 2, say, and the product G = A5 × B. Then
B = Z(G) is characteristic in G. But the endomorphism φ, which has kernel A5
and maps b 7→ (12)(34), does not map B into B.
Alternatively, take G to be the dihedral group of order 8. Show that it has a
unique cyclic subgroup C of order 4, which is thus characteristic. Show that C is
not fully invariant.)
An Ω-series is a series where each term is an Ω-subgroup.
A series is said to be normal if each term is normal in G, that is, it is an
Inn(G)-series.

2.3. Ω-composition series

A series is a refinement of another if it can be obtained by inserting further
subgroups. For instance
1 < ⟨ (12)(34) ⟩ < ⟨ (12)(34), (13)(24) ⟩ < S4
refines
1 < ⟨ (12)(34), (13)(24) ⟩ < S4 .
An Ω-series which has no proper refinement is called a Ω-composition series.
If Ω = ∅ one speaks simply of a composition series. If Ω = Inn(G), one speaks of
a principal series.
It follows from the third isomorphism theorem that the factors Hi+1 /Hi of a
composition series are simple groups. One sees that the factors of a principal series
are characteristically simple, that is, they have no proper, non-trivial characteristic
subgroups.
2.3.1. Proposition. Let G ̸= 1 be a finite group which is characteristically
simple.
Then there is a simple group S (abelian or non-abelian) and a positive integer
such that
G∼= S n = S × · · · × S.
Proof. Let us first consider the special case when G is abelian. Let p be a
prime dividing its order. Then a Sylow p-subgroups is normal, and thus charac-
teristic in G, so that G is a p-group. The subgroup
{ g ∈ G : gp = 1 }
is characteristic in G, so that G is elementary abelian, so G = S n where S is cyclic
of order p.
 2.3. Ω-COMPOSITION SERIES 21

In the general case, let N be a minimal normal subgroup of G. Then for

each automorphism φ of G such that N ̸= N φ one has N ∩ N φ = 1, so that
φ−1
[N, N φ ] ≤ N ∩ N φ = 1. (Note that N φ ⊴ G, as N φι(g) = N ι(g )φ = N φ , see the
proof of Exercise 1.7.5.) Since G = ⟨ N φ : φ ∈ Aut(G) ⟩, one sees (argument below)
that G is a direct product of some of the N φ , including N . Thus if 1 ̸= K ⊴ N ,
then K ⊴ G, so K = N and N is simple.
To see that G is a direct product of some of the N φ , start with M = N . If
M = G, we are done. Now let M be a direct product of some of the N φ , including
N , so that M ⊴ G. If M < G, there is a N ψ ≰ M . Since M ∩ N ψ ̸= N ψ , we have
M ∩ N ψ = 1 by the minimality of N ψ , so that M N ψ = M × N ψ . □
And now for the converse.
2.3.2. Proposition. A direct product of isomorphic simple groups is charac-
teristically simple.
2.3.3. Exercise. Using Proposition 1.8.3, show that if G is an elementary
abelian p-group, for a prime p, then its automorphism group acts transitively on
the non-zero elements.
(Hint: One has to show that given a finite-dimensional vector space V (over
any field, actually), and two non-zero vectors v, w ∈ V , there is a linear map
taking v to w.)
Proof. If the simple group S is abelian, then it has order a prime p, so that
G is elementary abelian, and one can use Exercise 2.3.3.
So let S be non-abelian simple, so that G = S n = T1 × . . . Tn for some n > 1,
the case n = 1 being trivial.
Let L ̸= 1 be a normal subgroup of G, and let
1 ̸= (s1 , s2 , . . . , sn ) ∈ L,
where we may assume s1 ̸= 1. Since Z(S) = { 1 }, there is x ∈ S such that sx1 ̸= s1 .
Therefore
(s1 , s2 , . . . , sn )−1 · (s1 , s2 , . . . , sn )(x,1,...,1) = (y, 1, . . . , 1)
for y = s−1
1 s1 ̸= 1. Thus L ∩ T1 ̸= 1. Since T1 is minimal normal, we have
x

L ≥ T1 . Note that we have proved so far is that the Ti are the unique minimal
normal subgroups of G. This is a special case of the Krull-Remak-Schmidt theory,
see [Rob96, p. 80].
Now suppose L is indeed characteristic in G. It remains to note that there is a
subgroup of Aut(G) isomorphic to Sn , which permutes the Ti . (See Exercise 2.3.4
just below.) Therefore L = G. □
2.3.4. Exercise. Let S be a set, n ≥ 1 an integer, G = S n be the direct product
of n copies of S.
(1) Show that the assignment, for σ ∈ Sn ,
(s1 , . . . , sn )σ = (s1σ−1 , . . . , snσ−1 )
defines a right action of Sn on G.
22 2. SERIES

(Hint: This is slightly tricky: the inverse is needed to make this into a
right action.)
Let now S be a group, so that G is a direct product.
(2) Show that for each σ ∈ Sn the map σ ′ given by
(s1 , . . . , sn ) 7→ (s1 , . . . , sn )σ
defines an automorphism of the group G, and actually σ 7→ σ ′ is a morphism
Sn → Aut(G).
2.4. Uniqueness of the factors of an Ω-composition series
2.4.1. Theorem. Let G be a finite Ω-group. Suppose G has two Ω-composition
series.
Then the factors of the two series are pairwise isomorphic.
Proof. Let
(2.4.1) 1 = H0 ⊴ H1 ⊴ . . . ⊴ Hn−1 ⊴ Hn = G,
and
(2.4.2) 1 = K0 ⊴ K1 ⊴ . . . ⊴ Km−1 ⊴ Km = G
be the two Ω-composition series, and proceed by induction on the order of G.
If Hn−1 = Km−1 , we are done by induction. If Hn−1 ̸= Km−1 , consider the Ω-
subgroup L = Hn−1 ∩ Km−1 , and refine the Ω-series
L ⊴ Hn−1 ⊴ G, L ⊴ Km−1 ⊴ G
to two Ω-composition series H and K, by taking the same refinement of L for both.
Since Hn−1 Km−1 is an Ω-subgroup properly containing both Hn−1 and Km−1 , we
have Hn−1 Km−1 = G. Therefore
(2.4.3) G/Hn−1 = Hn−1 Km−1 /Hn−1 ∼
= Km−1 /L, and
G/Km−1 = Hn−1 Km−1 /Km−1 ∼
= Hn−1 /L.
Proceeding by induction, the factors of (2.4.1) and H are pairwise isomorphic; they
are the factors of the Ω-series for L, plus Hn−1 /L and G/Hn−1 . Also, the factors
of (2.4.2) and K are pairwise isomorphic; they are the factors of the Ω-series for
L, plus Km−1 /L and G/Km−1 . By (2.4.3), we are done. □
2.4.2. Exercise. The factors of an Ω-composition series do not determine the
group uniquely. For instance both C6 and S3 have two composition factors which
are cyclic of orders 2, 3, and both C4 and C2 × C2 have two composition factors
which are cyclic of orders 2.
 CHAPTER 3

Soluble groups

3.1. Commutators
3.1.1. Definition. Let G be a group, a, b ∈ G. The commutator of a, b is
[a, b] = (ba)−1 ab = a−1 b−1 ab.
The name is justified by the
3.1.2. Lemma. Let G be a group, a, b ∈ G. The following are equivalent
(1) ab = ba, and
(2) [a, b] = 1.
3.1.3. Exercise. Show that [a, b]−1 = [b, a].
3.1.4. Definition. The subgroup
G′ = ⟨ [a, b] : a, b ∈ G ⟩
of G generated by the commutator is referred to as the derived subgroup or the
commutator subgroup.
3.1.5. Remark. In general not all elements of G′ will be commutators, but just
products of commutators. But it has been proved that if G is a finite, nonabelian
simple group, then every element of G′ is a commutator [LOST10].
3.1.6. Exercise. Show that Sn′ = An for all n > 1.
(Hint:
(1) Show that
(132)(123 . . . k) = (145 . . . k) (12)(34) = (123)(143).
(2) Show that An is generated by the 3-cycles.
(3) Show that [(12), (23)] = (123).
)
3.1.7. Exercise. Show that A′n = An for n ≥ 5.
More generally, if A, B ≤ G, then we define the subgroup
[A, B] = ⟨ [a, b] : a ∈ A, b ∈ B ⟩ .
3.1.8. Exercise.
(1) Prove the identities, for a, b, c in a group.
[a, bc] = [a, c][a, b]c , [ab, c] = [a, c]b [b, c].
(2) Prove that if A, B are subgroups of a group G, then A, B ≤ NG ([A, B]).
23
24 3. SOLUBLE GROUPS

Note the following

3.1.9. Lemma.
• If G, H are groups, φ : G → H is a morphism, and a, b ∈ G, then
φ([a, b]) = [φ(a), φ(b)].
• G′ is a fully invariant, and thus characteristic, subgroup of G, that is,
φ(G′ ) = G′ for all φ ∈ End(G).
• If K ≤ G, then the following are equivalent
(1) K ⊴ G, and
(2) [K, G] ≤ K.
Proof. The first statement is clear, as a morphism respects products and
inverses.
The second one follows immediately.
As to the third one, just note that for a ∈ K and b ∈ G we have
ab = b−1 ab = a[a, b].
□
Note the following
3.1.10. Lemma. Let G be a group, N ⊴ G.
• If H ⊴ N , then H is not necessarily normal in G.
• If H is characteristic in N , then H ⊴ G.
Proof. For the first claim, take G = A4 , N = ⟨ (12)(34), (13)(24) ⟩ to be the
2-Sylow subgroups of G, and H = ⟨ (12)(34) ⟩.
For the second one, consider for each g ∈ G, the map
N →N
n 7→ g −1 ng.
This is well defined, as N ⊴ G, and it is an automorphism of N . Since H is
characteristic in N , we have g −1 hg ∈ H for all g ∈ G and h ∈ H, that is,
H ⊴ G. □
It is easy to see that G/G′ is abelian. More generally, we have
3.1.11. Proposition. Let G be a group, H ≤ G. The following are equivalent:
(1) H ⊴ G and G/H is abelian, and
(2) G′ ≤ H.
Thus G′ is the smallest normal subgroup of G with abelian quotient.
Proof. Assuming H ⊴ G and G/H abelian, we have for all a, b ∈ G
H = [aH, bH] = [a, b]H,
so that [a, b] ∈ H and thus G′ ≤ H.
Conversely if G′ ≤ H, then [H, G] ≤ G′ ≤ H, so H ⊴ G, and for a, b ∈ G we
have
[aH, bH] = [a, b]H = H.
□
 3.2. THE DERIVED SERIES AND SOLUBLE GROUPS 25

3.2. The derived series and soluble groups

3.2.1. Definition. Given a group G, one constructs its derived series
G0 = G, G(1) = G′ , . . . , G(n+1) = (Gn )′ , . . .
G is said to be soluble (solvable in American English) if there is n such that
G(n) = { 1 }.
3.2.2. Remark. The name comes form Galois theory, as an equation is soluble
by radicals if and only if its Galois group is soluble.
3.2.3. Proposition. Let G be a group. The following are equivalent:
(1) G is soluble;
(2) there is a normal series G = G0 ≥ G1 ≥ · · · ≥ Gm = { 1 } with Gi /Gi+1
abelian for all i;
(3) there is a series G = G0 ≥ G1 ≥ · · · ≥ Gm = { 1 } with Gi /Gi+1 abelian
for all i;
Proof. If the first condition holds, then Gi = G(i) satisfies the second one.
If the second condition holds, then the series Gi satisfies the third one.
Let now Gi be a series as in (3). Then G1 ⊴ G and G/G1 is abelian, so that
G′ = G(1) ≤ Gi . Proceeding by induction, assume G(i) ≤ Gi . Since Gi+1 ⊴ Gi , and
Gi /Gi+1 is abelian, we have G(i+1) = (G(i) )′ ≤ G′i ≤ Gi+1 , so that G(n) = { 1 }. □
3.2.4. Proposition. Let G be a finite group. Then the following are equiva-
lent.
(1) G is soluble,
(2) there is a composition series whose factors are of prime order,
(3) the factors of any composition series are of prime order.
Proof. If G is soluble, refine the derived series to a composition series.
If a composition series has all factors of prime order, then by Theorem 2.4.1
this holds for any composition series.
A composition series with factors of prime order satisfies (3) of Proposition 3.2.3.
□
In a similar manner one proves
3.2.5. Proposition. Let G be a finite group. Then the following are equiva-
lent.
(1) G is soluble,
(2) there is a principal series whose factors are elementary abelian.,
(3) the factors of any composition series are elementary abelian.
3.2.6. Theorem. Let G be a group.
(1) If G is soluble and H ≤ G, then H is soluble.
(2) If G is soluble and N ⊴ G, then G/N is soluble.
(3) If G is soluble, and φ : G → K is a morphism, then φ(K) is soluble.
(4) If N ⊴ G, and both N and G/N are soluble, then G is soluble.
26 3. SOLUBLE GROUPS

Proof. If G is soluble and H ≤ G, just note that H (i) ≤ G(i) for all i.
If G is soluble and N ⊴ G, consider a series Gi as in Proposition 3.2.3(2).
Then Gi N ⊴ G for all i, so that
Gi N G
⊴ ,
N N
and
( ) ( ) ( ) ( )
Gi N Gi+1 N ∼ Gi N Gi Gi+1 N Gi ∼ Gi Gi ∩ Gi+1 N
/ = = = = /
N N Gi+1 N Gi+1 N Gi ∩ Gi+1 N Gi+1 Gi+1
is abelian, as a quotient of the abelian group Gi /Gi+1 .
(3) follows from the first isomorphism theorem.
As to (4), let Xi /N be a series with abelian quotients for G/N , and Yi be a
series with abelian quotients for G. Since
Xi−1 /N ∼ Xi−1
= ,
Xi /N Xi
we have that the series obtained by starting with the Xi , and continuing with the
Yi , is a series with abelian quotients for G. □
 CHAPTER 4

Nilpotent groups

4.1. Central series and nilpotent groups

4.1.1. Definition. A series G = G1 ≥ G2 ≥ · · · ≥ Gn ≥ . . . is said to be
central if for each i, we have [Gi , G] ≤ Gi+1 .
Note that the condition implies that all Gi ⊴ G.
4.1.2. Definition. A group G is said to be nilpotent if it has a central series
that terminates at { 1 }.
4.1.3. Lemma. A nilpotent group is soluble.
Proof. The quotients of a central series are abelian, as [Gi , Gi ] ≤ [Gi , G] ≤
Gi+1 . □
4.1.4. Exercise. Show that S3 , A4 , S4 are soluble but not nilpotent.
4.1.5. Theorem. Let G be a group.
(1) If G is nilpotent and H ≤ G, then H is nilpotent.
(2) If G is nilpotent and N ⊴ G, then G/N is nilpotent.
(3) If G is nilpotent, and φ : G → K is a morphism, then φ(K) is nilpotent.
4.1.6. Exercise. Show that item (4) of Theorem 3.2.6 does not hold with
soluble replaced by nilpotent.
4.1.7. Lemma. For a group G and a series G = G0 ≥ G1 ≥ · · · ≥ Gn = { 1 },
the following are equivalent
(1) the series is central, and
(2) for each i we have Gi ⊴ G, and
( )
Gi−1 G
≤Z .
Gi Gi
Recall that for a group G
Z(G) = { z ∈ G : zx = xz for all x ∈ G }
is the centre (center) of G.
4.1.8. Exercise. Show that the two conditions of Lemma 4.1.7 are equivalent,
noting that z ∈ Z(G) iff [z, x] = 1 for all x ∈ G.
4.1.9. Definition. The lower central series of the group G is defined by
γ1 (G) = G, and γi+1 (G) = [γi (G), G], for i ≥ 1.
27
28 4. NILPOTENT GROUPS

The upper central series of the group G is defined by Z0 (G) = { 1 }, and

( )
Zi (G) G
=Z
Zi−1 (G) Zi−1 (G)
for i ≥ 1.
4.1.10. Exercise. Show that these are indeed two central series.
4.1.11. Theorem. Let G be a group, and
G = G1 ≥ G2 ≥ · · · ≥ Gn = { 1 }
a central series.
Then for each i one has
γi (G) ≤ Gi ≤ Zn−i (G).
Proof. We have γ1 (G) = G = G1 , and then proceeding by induction
γi+1 (G) = [γi (G), G] ≤ [Gi , G] ≤ Gi+1 .
We have Gn−1 /Gn ≤ Z(G/Gn ) = Z1 (G)/ { 1 }, so that Gn−1 ≤ Z1 (G) = Z(G).
Proceeding by backward induction, [Gi , G] ≤ Gi+1 ≤ Zn−i−1 (G), so that
Gi Zn−i−1 (G)/Zn−i−1 (G) ≤ Z(G/Zn−i−1 (G)) = Zn−i (G)/Zn−i−1 (G),
that is, Gi ≤ Zn−i (G). □
4.1.12. Lemma. Let G be a nilpotent group.
If H < G, then H < NG (H).
Proof. Let i the the smallest number such that γi (G) ≤ H, so that γi−1 (G) ≰
H. Then [γi−1 (G), H] ≤ [γi−1 (G), G] = γi (G) ≤ H, so that γi−1 (G) ≤ NG (H). □
4.2. Finite nilpotent groups
4.2.1. Exercise. Let G be a finite group. Show that if for each H < G we
have H < NG (H), then G is nilpotent.
4.2.2. Lemma (Frattini argument).
(1) Suppose the group G acts on the set Ω, and H ≤ G acts transitively on
Ω.
Then for α ∈ Ω we have G = Gα H.
(2) Let X be a finite group, H ⊴ X and let S a Sylow p-subgroup of H.
Then X = NX (S)H.
(3) Let G be a finite group, and S be a Sylow p-subgroup of G.
Then NG (NG (S)) = NG (S).
Proof. If g ∈ G, then since H is transitive there is h ∈ H such that α =
(αg )h = αgh , so that gh ∈ Gα and g ∈ Gα H.
Let Ω be the set of Sylow p-subgroup of H. Now X acts by conjugation on Ω,
as H ⊴ X; by Sylow’s theorems, H acts transitively on Ω; the stabiliser of S is
NX (S).
Let H = NG (S), and X = NG (NG (S)). Then H ⊴ X, and thus X =
NX (S)NG (S) = NG (S). □
 4.3. NILPOTENT GROUPS ARE SOLUBLE 29

4.2.3. Theorem.
(1) A direct product of finitely many nilpotent groups is nilpotent.
(2) A finite p-group is nilpotent.
(3) A finite group G is nilpotent if and only if each p-Sylow subgroup is nor-
mal, so that G is the direct product of its distinct Sylow sybgroups.
Proof. Note that if G = A × B, and a1 , a2 ∈ A, b1 , b2 ∈ B, then [a1 b1 , a2 b2 ] =
[a1 , a2 ][b1 , b2 ]. It follows that if G = H1 × · · · × Hn , and k is large enough so that
γk (Hi ) = { 1 } for all i, then γk (G) = { 1 }.
By the arguments of Section 1.6, if P ̸= { 1 } is a p-group, then Z(P ) ̸= { 1 }.
It follows by induction that the upper central series terminates at P .
If G is nilpotent, and S is a Sylow p-subgroup, then we have NG (S) =
NG (NG (S)). It follows that NG (S) = G, that is, S ⊴ G. □

4.3. Nilpotent groups are soluble

We have already seen that a nilpotent group is soluble. Let us look at the
relations between the lower central series and the derived series in a (nilpotent)
group.
Let G be a group. We have γ2 (G) = G(1) . Then γ3 (G) = [γ2 (G), G] ≥
[G(1) , G(1) ] = G(2) . Proceeding by induction, if γi (G) ≥ G(i−1) , then γi+1 (G) =
[γi (G), G] ≥ [G(i−1) , G(i−1) ] = G(i) .
4.3.1. Proposition (Hall-Witt Identity). Let G be a group, a, b, c ∈ G. Then
[a, b−1 , c]b [b, c−1 , a]c [c, a−1 , b]a = 1.
The formula can be thought of as a group-theoretic version of the Jacobi iden-
tity in Lie algebras. It has wonderful geometric interpretations, see the blog post
of the Fields medalist Terence Tao [Tao12], and also [Cal11].
Proof. Note that each factor comes from the previous one by the cyclic per-
mutation a 7→ b 7→ c 7→ a.
[a, b−1 , c]b = b−1 [a, b−1 ]−1 c−1 [a, b−1 ]cb =
= b−1 [b−1 , a]c−1 [a, b−1 ]cb = b−1 ba−1 b−1 ac−1 a−1 bab−1 cb =
= (a−1 b−1 ac−1 a−1 )(bab−1 cb) = (aca−1 ba)−1 (bab−1 cb).
Set
U = aca−1 ba, V = bab−1 cb, W = cbc−1 ac.
Note that each element is obtained from the previous one by the cyclic permutation
a 7→ b 7→ c 7→ a. Thus
[a, b−1 , c]b = U −1 V, [b, c−1 , a]c = V −1 W, [c, a−1 , b]a = W −1 U,
and the formula follows. □
4.3.2. Theorem (Hall’s three-subgroup Lemma). Let G be a group, A, B, C ≤
G, and N ⊴ G.
If [A, B, C], [B, C, A] ≤ N , then [C, A, B] ≤ N
30 4. NILPOTENT GROUPS

Proof. Let a ∈ A, b ∈ B, c ∈ C. Then

[c, a−1 , b]a = [a, b−1 , c]−b [b, c−1 , a]−c ∈ N.
□
4.3.3. Corollary. [γi (G), γj (G)] ≤ γi+j (G).
It can be shown that an analogous formula does not hold for the upper central
series.
Proof. Argue by induction on j, the case j = 1 following from the definition.
If j > 1 we have
[γi (G), γj (G)] = [γi (G), [γj−1 (G), G]] = [γj−1 (G), G, γi (G)]
Now
[γi (G), γj−1 (G), G] ≤ [γi+j−1 (G), G] = γi+j (G),
[G, γi (G), γj−1 (G)] = [γi+1 (G), γj−1 (G)] ≤ γi+j (G).
Therefore [γi (G), γj (G)] ≤ γi+j (G). □
4.3.4. Corollary. G(i) ≤ γ2i (G).
Proof. We have G(1) = γ2 (G). Proceeding by induction on i,
G(i+1) = [G(i) , G(i) ] ≤ [γ2i (G), γ2i (G)] ≤ γ2i+1 (G).
□
 Part 3

Representations and Characters

 CHAPTER 5

Representations

Let G be a finite group, V a vector space of finite dimension n over the complex
numbers C. We will usually implicitly assume that V = Cn is the space of row
vectors.
A (linear) representation of degree n of G is a morphism
ρ : G → GL(V ) = GL(n, C).
So the group ρ(G) is a group of matrices.
As usual we have ρ(g −1 ) = ρ(g)−1 . If n = | G |, we have g n = 1 for g ∈ G,
so that ρ(g)n = I, where I is the identity map on V . This ρ(g) has minimal
polynomial dividing xn − 1. It follows that the eigenvalues of ρ(g) are n-th roots
of unity, and they are distinct, so that ρ(g) is diagonalizable.

5.1. Inner products

See Section 1.5 for the proper definitions.
We now show
5.1.1. Lemma. There is an inner product on V such that the ρ(g) are unitary
matrices.
Proof. In fact, let ⟨ ·, · ⟩ be the standard inner product on V . Define, for
x, y ∈ V
∑
⟨⟨x, y⟩⟩ = ⟨xρ(g), yρ(g)⟩ .
g∈G

It is easy to show that ⟨⟨·, ·⟩⟩ is again an inner product. Moreover for h ∈ G we
immediately have
∑ ∑
⟨⟨xρ(h), yρ(h)⟩⟩ = ⟨xρ(h)ρ(g), yρ(h)ρ(g)⟩ = ⟨xρ(k), yρ(k)⟩ = ⟨⟨x, y⟩⟩,
g∈G k∈G

so that each ρ(h) is unitary with respect to ⟨⟨·, ·⟩⟩. In particular, with respect to a
basis which is orthonormal for ⟨⟨·, ·⟩⟩, we will have
ρ(g)−1 = ρ(g)∗ = ρ(g)t ,
as for x, y ∈ V and g ∈ G we have
⟨⟨xρ(g)∗ , y⟩⟩ = ⟨⟨x, yρ(g)⟩⟩ = ⟨⟨xρ(g)−1 ρ(g), yρ(g)⟩⟩ = ⟨⟨xρ(g)−1 , y⟩⟩,
and then since when x, y ∈ V are written with respect to such a basis we have
⟨⟨x, y⟩⟩ = x · y t ,
33
34 5. REPRESENTATIONS

then
⟨⟨x, yρ(g)⟩⟩ = x · ρ(g)t y t , = xρ(g)t · y t , = ⟨⟨xρ(g)t , y⟩⟩.
□

5.2. Examples
As a first set of examples, let G act on a Ω = { 1, 2, . . . , n }. Let V be a space
of dimension n, with basis v1 , v2 , . . . , vn . Then
ρ : G → GL(V )
g 7→ (vi 7→ vig )
is a representation, the permutation representation associated to the action.
In the particular case when G acts on itself by right multiplication, we obtain
the (right) regular representations.
If W is a subspace of V of dimension m, it may happen that W is invariant
under ρ(G), that is, for all w ∈ W and g ∈ G, then wρ(g) ∈ W . Then we get a
subrepresentation of G
ρW : G → GL(W ) = GL(m, C).
As an example, consider the cyclic group G = ⟨ a ⟩ ≤ S3 of order three, where
a = (123), so that G acts naturally on Ω = { 1, 2, 3 }. Then
 
0 1 0
 
ρ(a) = 0 0 1 .
1 0 0
Consider u = e1 + e2 + e3 . (Here the ei are the elements of the standard basis of
C3 .) Then
uρ(a) = e1 ρ(a) + e2 ρ(a) + e3 ρ(a) = e2 + e3 + e1 = u,
so that U = ⟨ u ⟩ is invariant under ρ(G). Let ω = exp(i 2π
3
) be a primitive third
root of unity. Consider the elements

w = e + ωe + ω 2 e
1 1 2 3
w2 = ω 2 e1 + e2 + ωe3

Then
w1 ρ(a) = e2 + ωe3 + ω 2 e1 = w2
and
w2 ρ(a) = ωe1 + ω 2 e2 + e3 = −w1 − w2 ,
as 1 + ω + ω 2 = 0. It follows that W = ⟨ w1 , w2 ⟩ is also ρ(G)-invariant.
5.2.1. Exercise. Show that V = U ⊕ W .
5.2.2. Exercise. Show that in the basis w1 , w2 of W , the (sub)representation
ρ : G → GL(W ) is given by
W
[ ]
W 0 1
ρ (a) = .
−1 −1
 5.3. THE GROUP ALGEBRA 35

5.3. The group algebra

There is another way of describing representations. Let G be a finite group,
and consider the set C[G] = CG of maps from G to C, which is a vector space of
dimension | G | over C. Define on C[G] a convolution product by
∑
a ∗ b(g) = a(x)b(y).
x,y∈G,xy=g

This can be rewritten of course as

∑
(5.3.1) a ∗ b(g) = a(x)b(x−1 g),
x∈G

but the previous symmetric form is handier for the proofs. Note that this is similar
to the product of polynomials.
With this operation, C[G] turns out to be an algebra, that is, a vector space
over C endowed with an associative, bilinear product. Let us check associativity.
∑ ∑
(a ∗ b) ∗ c(g) = a ∗ b(x)c(y) = a(s)b(t)c(y).
x,y∈G,xy=g s,t,y∈G,sty=g

One obtains the very same result with a ∗ (b ∗ c).

5.3.1. Exercise. Try and do this with the asymmetric form (5.3.1).
5.3.2. Exercise. Check the remaining properties.
The vector space C[G] has a basis given by the δg , for g ∈ G, given by

1 if x = g
δg (x) =
0 otherwise,

as for a ∈ C[G] we have uniquely

∑
a= a(g)δg .
g∈G

Note that ∑
δg ∗ δh (x) = δg (y)δh (z),
y,z∈G,yz=x

and the only non-zero summand is when y = g, z = h so that x = gh. It follows

that δg ∗ δh = δgh . It is therefore customary to write δg as g, and say that
 
∑ 
C[G] =  ag g : ag ∈ C 
g∈G

is a vector space with basis the elements of G, and product that extends by
linearity that of G.
Now if ρ : G → GL(V ) is a representation, this extends uniquely to a morphism
of algebras ρ′ : C[G] → End(V ) given by
∑ ∑
ρ′ ( ag g) = ag ρ(g).
g∈G g∈G
36 5. REPRESENTATIONS

Note that for a ∈ C one has ρ′ (a · 1) = a · I, where 1 ∈ G, and I is the identity

map on V . And conversely, given such an algebra morphism ρ′ , its restriction to
G is a representation of G.
Note that the right regular representation can be regarded as a representation
G → GL(C[G]).
5.3.3. Proposition. The centre Z(C[G]) of C[G] consists of the elements
∑
f (g)g
g∈G

where f : G → C is a class function, that is f (g h ) = f (g) for all g, h ∈ G.

Proof. Immediate, just conjugate an element in the centre by h ∈ G. □

5.4. Maschke’s Theorem

5.4.1. Theorem (Maschke). Let ρ : G → GL(V ) be a representation of G.
Suppose U is a ρ(G)-invariant subspace of V .
Then there is a ρ(G)-invariant subspace W of V such that
V = U ⊕ W.
Since this is a key result, we will give two proofs.

First proof of Maschke’s theorem. The key to this is an averaging ar-

gument.
Let us choose an arbitrary subspace X of V such that V = U ⊕ X. Of course
X need not be ρ(G)-invariant.
Let π : V → U be the projection of V onto U along X, that is, if we write
v ∈ V as v = u + x, with u ∈ U and x ∈ X, then π(v) = u.
Consider the following linear map on V
1 ∑
ψ= ρ(g)−1 πρ(g).
| G | g∈G

Thus for v ∈ V
1 ∑
vψ = ((vρ(g)−1 )π)ρ(g).
| G | g∈G
Note first that V ψ ⊆ U , as π maps V onto U , and U is ρ(G)-invariant. But in
fact V ψ = U : if u ∈ U , then
1 ∑ 1 ∑ 1 ∑
uψ = ((uρ(g)−1 )π)ρ(g) = (uρ(g)−1 )ρ(g) = u = u,
| G | g∈G | G | g∈G | G | g∈G

as uρ(g) ∈ U for all g ∈ G. Now ψ 2 = ψ, since vψ 2 = (vψ)ψ = vψ, as vψ ∈ U ,

and we have just shown that ψ restricts to the identity on U . By Lemma 1.1.1,
V = U ⊕ ker(ψ).
 5.5. IRREDUCIBLE REPRESENTATIONS 37

Now ker(ψ) is ρ(G)-invariant, as for w ∈ ker(ψ) and h ∈ G we have

∑
| G | wρ(h)ψ = wρ(h) ρ(g −1 )πρ(g)
g∈G
∑
=w ρ(hg −1 )πρ(g)
g∈G
∑
=w ρ(gh−1 )−1 πρ(g)
g∈G
∑
=w ρ(k)−1 πρ(kh)
k∈G
= | G | wψρ(h) = 0.
□
Second proof of Maschke’s theorem. We employ the inner product of
Lemma 5.1.1. Let W be the orthogonal of U with respect to ⟨⟨·, ·⟩⟩, that is,
W = U ⊥ = { v ∈ V : ⟨⟨u, x⟩⟩ = 0 for all u ∈ U } .
We claim that W is ρ(G)-invariant. In fact if w ∈ W , then for all g ∈ G and
u ∈ U we have
⟨⟨u, wρ(g)⟩⟩ = ⟨⟨uρ(g)−1 , w⟩⟩ = 0,
as U is ρ(G)-invariant. □
5.4.2. Definition. If ρi : G → GL(Vi ) are representations of the group G, for
i = 1, . . . , n, then the direct sum representation is defined as
ρ1 ⊕ ρ2 ⊕ · · · ⊕ ρn : G → GL(V1 ) × GL(V2 ) × · · · × GL(Vn ) ≤ GL(V1 ⊕ V2 ⊕ · · · ⊕ Vn )
g 7→ (v1 ⊕ v2 ⊕ · · · ⊕ vn 7→ v1 ρ1 (g) ⊕ v2 ρ2 (g) ⊕ · · · ⊕ vn ρn (g)

5.5. Irreducible representations

5.5.1. Definition. A representation ρ : G → GL(V ) is said to be irreducible
if the only ρ(G)-invariant subspaces of V are { 0 } and V itself.
Maschke’s Theorem implies immediately (by induction on the dimension of the
vector space)
5.5.2. Theorem. Every finite-dimensional representation over C is a direct
sum of irreducible ones.
One should compare this result to the fact that two groups may have the same
composition factors, without being isomorphic. For instance, as noted earlier,
both C6 and S3 have two composition factors that are cyclic groups of order 2
and 3. So this simple constituents alone do not determine a group uniquely, as
they can be put together in different ways. (In this particular case, there are two
non-isomorphic semidirect products of C3 by C2 .)
With a representation, instead, if you know its irreducible subrepresentations,
the representation is uniquely determined as the direct sum of them, there is only
one (trivial) way of gluing them together.
38 5. REPRESENTATIONS

5.5.3. Exercise. The statement is not true anymore over fields F whose char-
acteristic divides the order of G. The simplest example is given by G = ⟨ a ⟩ cyclic
of order 2, and by the representation over the field F with two elements defined by
ρ : G → F2
[ ]
1 1
a 7→ .
0 1
5.5.4. Definition. Let ρi : G → GL(Vi ) be representations of G, for i = 1, 2.
A morphism of representations is a linear map f : V1 → V2 such that for g ∈ G
ρ1 (g)f = f ρ2 (g).
These must be thought as “linear maps” with respect to the ρi (g), more about
this later when we will discuss modules. If such an f is bijective, we call it of
course an isomorphism. So ρ1 , ρ2 are isomorphic if there is f such that
ρ2 (g) = f −1 ρ1 (g)f, for all g ∈ G.
5.6. Schur’s Lemma
5.6.1. Theorem (Schur’s Lemma). Let ρi : G → GL(Vi ) be irreducible repre-
sentations of G, for i = 1, 2, and f : V1 → V2 a morphism.
Then
• either f = 0,
• or f is an isomorphism (that is, f is bijective).
Proof. One sees easily that ker(f ) ⊆ V1 is ρ1 (G)-invariant, and V1 f ⊆ V2 is
ρ2 (G)-invariant.
If V1 f = { 0 }, then f = 0. If V1 f ̸= { 0 }, then V1 f = V2 , as ρ2 is irreducible,
and ker(f ) ̸= V1 , so that ker(f ) = { 0 }, as ρ1 is irreducible. □
5.6.2. Corollary. Let ρ : G → GL(V ) be irreducible. If f : V → V is an
isomorphism, then there is λ ∈ C∗ such that vf = λv is multiplication by the
scalar λ.
Proof. Let λ be an eigenvalue of f . (We are exploiting for the first time the
fact that C is algebraically closed.)
Then from
ρ1 (g)f = f ρ2 (g), and ρ1 (g)(λI) = (λI)ρ2 (g),
where I is the identity on V , we get
ρ1 (g)(f − λI) = (f − λI)ρ2 (g).
Thus f − λI is also a morphism of representations. Since it is singular, it must be
zero, so that f = λI is scalar multiplication by λ. □
5.6.3. Lemma. Let ρi : G → GL(Vi ) be representations of G, for i = 1, 2. Let
f : V1 → V2 be any linear map. Then
∑
f′ = ρ1 (g −1 )f ρ2 (G)
g∈G
 5.7. ORTHOGONALITY RELATIONS 39

is a morphism of representations V1 → V2 .
Proof. Very much as in Maschke’s Theorem. □

5.7. Orthogonality Relations

5.7.1. Proposition. Let ρ1 , ρ2 be non-isomorphic, irreducible representations.
Take V1 , V2 to be spaces of row vectors, with standard bases u1 , . . . , un and v1 , . . . , vm .
Let ρjk
i (g) denote the (j, k)-component of ρi (g).
Then for all j, s, t, l we have
∑
−1 tl
(5.7.1) ρjs
1 (g )ρ2 (g) = 0
g∈G

5.7.2. Remark. By choosing the bases to be orthonormal as in Lemma 5.1.1,

then (5.7.1) can be rewritten as
∑
ρsj tl
1 (g)ρ2 (g) = 0
g∈G

because
ρ1 (g −1 ) = ρ1 (g)−1 = ρ1 (g)∗ = ρ1 (g)t .
Proof. Let Est : V1 → V2 be the linear map that sends all ui to 0, except
us Es,t = vt .
Then
∑
n
ej ρ1 (g −1 )Es,t ρ2 (g) = ρjk −1
1 (g )Es,t ρ2 (g)
k=1
−1
= ρjs
1 (g )vt ρ2 (g)
∑m
−1 tl
= vl ρjs
1 (g )ρ2 (g).
l=1
Summing over g ∈ G we get
∑
m ∑
−1 tl
0= vl ρjs
1 (g )ρ2 (g),
l=1 g∈G

so that
∑
−1 tl
ρjs
1 (g )ρ2 (g) = 0
g∈G

for all j, s, t, l. □
5.7.3. Proposition. Let ρ be an irreducible representation. Take V to be
spaces of row vectors, with standard basis u1 , . . . , un . Let ρjk (g) denote the (j, k)-
component of ρ(g).
Then for all j, s, t, l we have

∑ | G |

sj tl if j = l and s = t
ρ (g)ρ (g) = n

0
g∈G otherwise.
40 5. REPRESENTATIONS

Proof. In the notation of the previous proof, take first s ̸= t. Then

trace(ρ(g −1 )Es,t ρ(g)) = trace(Es,t ) = 0,
so that ∑
ρjs (g −1 )ρtl (g) = 0
g∈G
as in the previous proof.
When s = t, we have
trace(ρ(g −1 )Es,s ρ(g)) = trace(Es,s ) = 1,
so that ∑
trace( ρjs (g −1 )ρtl (g)) = | G | ,
g∈G
and thus ∑
ρjs (g −1 )ρtl (g)
g∈G
|G|
is scalar multiplication by . The result follows as in the previous proof. □
n
5.7.4. Remark. The formula of Proposition 5.7.3 also holds if instead of a
single irreducible representation ρ one considers two (possibly different) irreducible
representation isomorphic representations ρ1 , ρ2 . This is because by definition
there is a bijective linear map f such that for all g ∈ G one has ρ2 (g) = f −1 ρ1 (g)f ,
so that
trace(ρ1 (g −1 )Es,t ρ2 (g)) = trace(ρ1 (g −1 )Es,t f −1 ρ1 (g)f ) = trace(Es,t ).
 CHAPTER 6

Characters

6.1. Characters
6.1.1. Definition. The character of a representation ρ is the map

χ:G→C
g 7→ trace(ρ(g)).

One says that ρ affords χ.

We have seen in Subsection 5.5 that two representations ρ1 , ρ2 are isomorphic

if and only if there is a bijective linear map such that

ρ2 (g) = f −1 ρ1 (g)f, for all g ∈ G.

Therefore
trace(ρ2 (g)) = trace(f −1 ρ1 (g)f ) = trace(ρ1 (g)).
In other words

6.1.2. Proposition. The character of a representation only depends on the

isomorphism type of the representation.

6.1.3. Remark. If a, b : G → C are two functions, then setting

1 ∑
(a, b) = a(g)b(g)
| G | g∈G

we obtain an inner product in the space of all such functions.

6.1.4. Theorem (Orthogonality relations).

(1) If χ is the character of an irreducible representation, then

(χ, χ) = 1.

(2) If χ, ψ are the characters of two non-isomorphic, irreducible representa-

tions, then
(χ, ψ) = 0,
(3) Summing it up, the characters of the irreducible representations are or-
thonormal.
41
42 6. CHARACTERS

Proof. The first statement is clear from Proposition 5.7.1. The second one
follows from Proposition 5.7.1, as
( ) 
1 ∑ ∑ n ∑n
(χ, χ) = ρii (g)  ρjj (g)
| G | g∈G i=1 j=1

1 ∑ n ∑
= ρii (g)ρii (g)
| G | i=1 g∈G
1 |G|
= ·n· = 1.
|G| n
□
We obtain an important fact
6.1.5. Theorem. The character of a representation determines the represen-
tation up to isomorphism.
In case you are interested in how to get back from a character to the (unique)
representation it affords it, check [Spe10].
Proof. Let ρ be a representation, and decompose it as
ρ = n1 ρ1 ⊕ n2 ρ2 ⊕ · · · ⊕ nt ρt ,
where the ρi are pairwise non-isomorphic irreducible representations, and ni de-
notes the number of times ρi occurs. It is easy to see that the ni determine the
isomorphism type of ρ.
Taking the trace, we get
χ = n1 χ 1 ⊕ n2 χ 2 ⊕ · · · ⊕ nt χ t ,
where χ is the character of ρ, and χi is the character of ρi .
Now
(χi , χ) = ni .
Thus the character χ determines the ni , and thus the isomorphism type of ρ. □
This allows for the following
6.1.6. Definition. A character is said to be irreducible if the corresponding
representation is irreducible.
The set of the irreducible characters of G is denoted by Irr(G).

6.2. Decomposing the regular representation

Let ρ : G → GL(C[G]) be the right regular representation, and ψ its character.
Since xρ(g) = xg = x only if g = 1, we get

| G | if g = 1
ψ(g) =
0 otherwise.
 6.3. NUMBER OF IRREDUCIBLE CHARACTERS 43

Let χ be an irreducible character. Then

1 ∑ 1 1
(ψ, χ) = ψ(g)χ(g) = ψ(1)χ(1) = | G | χ(1) = χ(1).
| G | g∈G |G| |G|
6.2.1. Definition. The degree of a character χ is χ(1) = trace(I), which
equals the dimension of the vector space associated to representation that affords
χ.
It follows
6.2.2. Theorem.
(1) Every irreducible representation with character χ occurs χ(1) times in
the decomposition of the regular representation as a sum of irreducible
representation.
(2) If ψ is the character of the regular representation, then
∑
ψ= χ(1)χ,
χ

where χ ranges over the irreducible characters of G.

(3)
∑
(6.2.1) |G| = χ(1)2 ,
χ

where χ ranges over the irreducible characters of G.

Proof. It remains only to prove the important last statement, which follows
by evaluating (2) at 1. □
6.2.3. Corollary. Let ρi be the pairwise non-isomorphic, irreducible repre-
sentations, and ρjk jk
i be their components. Then the ρi are an orthonormal basis
for the space of functions G → C.
Proof. We have see in Subsection 5.7 that these functions are orthonormal.
Now (6.2.1) shows that their number equals the dimension of the space of functions
G → C. □

6.3. Number of irreducible characters

Here I am following [Ser16].
If ρi : G → GL(Vi ) are the pairwise non-isomorphic irreducible representations
of G, then we have an algebra morphism
∑
r : C[G] → End(Vi ).
i

The two spaces have the same dimension, by (6.2.1). We will show that r is injec-
tive, and this will imply that r is an algebra isomorphism. But an element of ker(r)
acts as 0 in every irreducible representation, thus in the regular representation,
which is faithful (action on 1 ∈ C[G]).
∑
Now the centre of the algebra i End(Vi ) has dimension the number of distinct
irreducible characters. The centre of C[G] has a basis given by the sums over the
44 6. CHARACTERS

conjugacy classes, and thus the dimension of the centre equals the number of
conjugacy classes. We obtain
6.3.1. Theorem. The number of distinct irreducible characters (that is, of
pairwise non-isomorphic representations) of G equals the number of conjugacy
classes of G.
The isomorphism r induces an isomorphism from Z(C[G]) to the centre of
∑
i End(V i ), which is a sum of copies of C (scalar matrices for each End(Vi )).
Choose one of two Vi , associated to the irreducible representation ρ with character
χ. Consider the map rχ , which is the composition of r with the projection on
End(Vi ), and which is simply given by
∑ ∑
(6.3.1) rχ ( f (g)g) = f (g)ρ(g)
g∈G g∈G
∑
Let now α = g∈G f (g)g ∈ Z(C[G]); this goes under rχ to a scalar matrix S in
Z(End(Vi )), which is multiplication by some a. Since a = trace(S)/χ(1), (6.3.1)
yields that the value of a is
1 ∑ 1 ∑ |G|
· trace( f (g)ρ(g)) = f (g)χ(g) = (f, χ).
χ(1) g∈G χ(1) g∈G χ(1)
∑
6.3.2. Theorem. rχ ( g∈G f (g)g) is the scalar matrix that is multiplication by
|G|
(f, χ).
χ(1)
6.4. Character tables
6.4.1. Definition. The character table of G is a square matrix, which the
rows labelled by the distinct irreducible characters, and the columns labelled by
the conjugacy classes.
The χ, aG entry is χ(a).
6.5. Characters of abelian groups
(To be revised.)
An irreducible character χ of a finite abelian group is linear, that is, χ(1) = 1,
so that χ : G → C∗ is a morphism of groups.
This follows from Section 1.4: a finite number of commuting, diagonalizable
matrices can be diagonalized simultaneously, that is, there is a basis with respect
to which they are all diagonal.
Alternatively, an abelian group∑of order n has n conjugacy classes, and thus n
irreducible characters. Since n = χ∈Irr(G) χ(1)2 ≥ n, each χ(1) must be 1.
(To be expanded.) Now note that representations and characters of quotient
groups G/N lift to representation and characters of G to get the linear characters
of a group. Define the kernel of a character.
6.6. Character tables of small groups
Small abelian groups (dual group), S3 , A4 , S4 . (Explanations to be added)
 6.7. INTEGRALITY 45

6.6.0.1. S3 .
1 3 3
1 (123) (12)
1 1 1
1 1 −1
2 −1 0
6.6.0.2. A4 .
1 3 4 4
1 (12)(34) (123) (132)
1 1 1 1
1 1 ω ω −1
1 1 ω −1 ω
3 −1 0 0
6.6.0.3. S4 .
1 3 6 8 6
1 (12)(34) (12) (123) (1234)
1 1 1 1 1
1 1 −1 1 −1
2 2 0 −1 0
3 −1 1 0 −1
3 −1 −1 0 1
(The last one is the previous one times sign, that is, the second one.)

6.7. Integrality
If ρ is a linear representation of G, and g ∈ G has order n, then
1 = ρ(1) = ρ(g n ) = ρ(g)n .
It follows that ρ(g) is a root of xn − 1, so that the minimal polynomial of ρ(g) is
a divisor of xn − 1, and thus the eigenvalues of ρ(g) are n-th roots of unity. We
obtain
6.7.1. Proposition. Character values, as sum of roots of unity, are algebraic
integers.
6.7.2. Exercise.
(1) Show that is ρ is the right regular representation, and g ∈ G has order n,
then the minimal polynomial of ρ(g) is xn − 1.
(2) Show that the above need not hold for an arbitrary representation. Avoid
the trivial case when ρ(g) = 1, and try and find an example in which
| g | = | ρ(g) |.
6.7.3. Theorem. Let f : G → C be a class function on the group G, whose
∑
values are algebraic integers, so that α = g∈G f (g)g ∈ Z(C[G]).
(1) α is integral.
46 6. CHARACTERS

(2) If χ is an irreducible character of G, then

1 ∑
f (g)χ(g)
χ(1) g∈G
is an algebraic integer.
∑
Proof. Let S be the subring of Z(C[G]) consisting of the g∈G f (g)g, where
∑
f is a class function with integer values. The elements g∈C g, for C a conjugacy
class, are a basis of S as a Z-module. Since S is finitely generated, all of its
elements are integral. Since the value of f (g) are algebraic integers, it follows that
α is integral.
Applying rχ to α, and appealing to Theorem 6.3.2, we obtain the second part,
since clearly the image of an integral element is integral, and thus that number is
an algebraic integer. □
6.7.4. Corollary. If χ is an irreducible character, then χ(1) divides the order
of G.
Proof. In part (2) of Theorem 6.7.3, take f = χ to get that | G | /χ(1) is a
rational number which is an algebraic integer, and thus it is an integer. □
6.7.5. Proposition. Let ρ be an irreducible representation, and χ be its char-
acter. Let g ∈ G. We have
(1)  
 G
 g  · χ(g)
χ(1)
is an algebraic
 
integer.
 G
Suppose now gcd( g  , χ(1)) = 1. Then
(2) χ(g)/χ(1) is an algebraic integer.
(3) If χ(g) ̸= 0, then ρ(g) is a scalar matrix.
Proof. Let f : G → C be the class function that is zero everywhere, except
on the conjugacy class of g, where its value is 1, so that
∑  
 
f (x)χ(x) =  g G  χ(g).
x∈G
 
 
By Theorem 6.7.3(2),  g G  · χ(g)/χ(1) is an algebraic integer.
   
   
If gcd( g G  , χ(1)) = 1, by Bézout, there are a, b ∈ Z such that a  g G  +bχ(1) =
1, so that
χ(g)   χ(g)
 
= a  gG  · + bχ(g).
χ(1) χ(1)
 
 
Since both  g G  · χ(g)/χ(1) and χ(g) are algebraic integers, so is χ(g)/χ(1).
χ(g)/χ(1) is of the form of (1.9.1) of Lemma 1.9.7, therefore ρ(g) is a scalar. □
 CHAPTER 7

Applications

7.1. Burnside pa q b
In this section we report the celebrated theorem of Burnside [Bur04], that
shows that finite group of order divisible of at most two primes are soluble. (A5
is non-abelian simple, of order 22 · 3 · 5.)
Burnside’s proof makes use of characters, whose theory he had contributed to
developing. Burnside published his proof in 1904. One had to wait until the 1970’s
for proofs not using characters [Gol70, Ben72, Mat73].
7.1.1. Lemma. Let G be a finite group, 1 ̸= g ∈ G such that its conjugacy class
has order a power of a prime p.
Then there is a proper normal subgroup N of G such that gN ∈ Z(G/N ).
Proof. By the orthogonality property of the character table we have
∑
χ(1)χ(g) = 0,
χ∈Irr(G)

and thus
∑ χ(1)χ(g) 1
=− .
1̸=χ∈Irr(G)
p p
Since the right-hand side is not an algebraic integer, there is 1 ̸= χ ∈ Irr(G) such
that
  χ(1)χ(g)/p is not an algebraic integer,
  so that χ(g) ̸= 0, and p ∤ χ(1). Since
 G  G
 g  is a power of p, it follows that gcd( g  , χ(1)) = 1. Now Proposition 6.7.5(3)
yields that ρ(g) is scalar, where ρ is a representation that affords χ. Since χ ̸= 1,
we have that ker(ρ) is a proper normal subgroup of G, and the first isomorphism
theorem yields G/ ker(ρ) ∼ = ρ(G). Since ρ(g) is scalar, it is in the centre of ρ(G),
and thus gN is in the centre of G/N . □
7.1.2. Theorem (Burnside pa q b ). A group of order pa q b , where p and q are
primes, is soluble.
Proof. Note first of all that groups of order the power of a prime are nilpotent
(and thus soluble), as follows from the class equation. Therefore we may assume
that both p and q divide the order of G.
Considering a composition series of G, we see that we may assume G to be
non-abelian simple. Were { 1 } the only conjugacy class of order not divisible by q,
then the order of G would congruent to 1 modulo q. So there is an element g ∈ G
whose conjugacy class g G has order not divisibile by q and thus, by orbit-stabiliser,
g G has order a power of p. By Lemma 7.1.1, g ∈ Z(G), a contradiction. □

47
 Bibliography

[Ben72] Helmut Bender, A group theoretic proof of Burnside’s pa q b -theorem, Math. Z. 126
(1972), 327–338. MR 0322048
[Bur04] W. Burnside, On Groups of Order pα q β , Proc. London Math. Soc. (2) 1 (1904),
388–392. MR 1576790
[Cal11] Danny Calegari, The Hall-Witt identity, November 2011,
https://lamington.wordpress.com/2011/11/20/the-hall-witt-identity/.
[Gol70] David M. Goldschmidt, A group theoretic proof of the pa q b theorem for odd primes,
Math. Z. 113 (1970), 373–375. MR 0276338
[Gor80] Daniel Gorenstein, Finite groups, second ed., Chelsea Publishing Co., New York,
1980. MR 569209
[Hup67] B. Huppert, Endliche Gruppen. I, Die Grundlehren der Mathematischen Wis-
senschaften, Band 134, Springer-Verlag, Berlin-New York, 1967. MR 0224703 (37
#302)
[Isa06] I. Martin Isaacs, Character theory of finite groups, AMS Chelsea Publishing, Provi-
dence, RI, 2006, Corrected reprint of the 1976 original [Academic Press, New York;
MR0460423]. MR 2270898
[Jac85] Nathan Jacobson, Basic algebra. I, second ed., W. H. Freeman and Company, New
York, 1985. MR 780184
[LOST10] Martin W. Liebeck, E. A. O’Brien, Aner Shalev, and Pham Huu Tiep, The Ore
conjecture, J. Eur. Math. Soc. (JEMS) 12 (2010), no. 4, 939–1008. MR 2654085
[Mac12] Antonio Machì, Groups, Unitext, vol. 58, Springer, Milan, 2012, An introduction to
ideas and methods of the theory of groups. MR 2987234
[Mar18] Daniel A. Marcus, Number fields, Universitext, Springer, Cham, 2018, Second edition
of [MR0457396], with a foreword by Barry Mazur. MR 3822326
[Mat73] Hiroshi Matsuyama, Solvability of groups of order 2a pb , Osaka J. Math. 10 (1973),
375–378. MR 0323890
[Rob96] Derek J. S. Robinson, A course in the theory of groups, second ed., Graduate Texts
in Mathematics, vol. 80, Springer-Verlag, New York, 1996. MR 1357169 (96f:20001)
[Rot95] Joseph J. Rotman, An introduction to the theory of groups, fourth ed., Graduate Texts
in Mathematics, vol. 148, Springer-Verlag, New York, 1995. MR 1307623
[Ser78] Jean-Pierre Serre, Représentations linéaires des groupes finis, revised ed., Hermann,
Paris, 1978. MR 543841
[Ser16] , Finite groups: an introduction, Surveys of Modern Mathematics, vol. 10,
International Press, Somerville, MA; Higher Education Press, Beijing, 2016, With
assistance in translation provided by Garving K. Luli and Pin Yu. MR 3469786
[Spe10] David E. Speyer, Recovering representation from its character, MathOverflow, 2010,
URL:https://mathoverflow.net/q/32846 (version: 2010-07-21).
[Tao12] Terence Tao, Cayley graphs and the algebra of groups, May 2012,
https://terrytao.wordpress.com/tag/hall-witt-identity/.

49