Algebra I: Group Theory

Dr. Arjun Paul

Assistant Professor
Department of Mathematics and Statistics
Indian Institute of Science Education and Research Kolkata,
Mohanpur - 741 246, Nadia,
West Bengal, India.
Email: [email protected].

Version: September 15, 2023 at 11:18pm (IST).

Available at: https://arjunpaul29.github.io/home/notes/Algebra-I.pdf

Note: This note will be updated from time to time.

If you find any potential mistakes/typos, please bring it to my notice.
 iii

To my students . . .
 v

Contents

List of Symbols vii

1 Foundation of Arithmetic 1
1.1 What is a Natural Number? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Integers: Construction & Basic Operations . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Division Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2 Group Theory 13
2.1 Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Subgroup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3 Cyclic group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4 Product of subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5 Permutation Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.6 Group homomorphism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.7 Notion of Quotient & Cosets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.8 Normal Subgroup & Quotient Group . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.9 Isomorphism Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.10 Direct Product & Direct Sum of Groups . . . . . . . . . . . . . . . . . . . . . . . . 56
2.11 Group Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.12 Conjugacy Action & Class Equations . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.12.1 p-group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.13 Simple Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.14 Sylow’s Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.14.1 Simplicity of An , n ≥ 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.14.2 Simplicity of PSLn (C) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.14.3 Groups of small orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.15 Semi-direct product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.15.1 Finitely Generated Abelian Groups . . . . . . . . . . . . . . . . . . . . . . 75
2.16 Free Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.17 Solvable & Nilpotent Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.18 Linear Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
 vii

List of Symbols

∅ Empty set
Z The set of all integers
Z≥0 The set of all non-negative integers
N The set of all natural numbers (i.e., positive integers)
Q The set of all rational numbers
R The set of all real numbers
C The set of all complex numbers
< Less than
≤ Less than or equal to
> Greater than
≥ Greater than or equal to
⊂ Proper subset
⊆ Subset or equal to
⊊ Subset but not equal to (c.f. proper subset)
∃ There exists
∄ Does not exists
∀ For all
∈ Belongs to
∈
/
P Does not belong to
Q Sum
Product
± Plus and minus
∞
√ Infinity
a Square root of a
∪
F Union
Disjoint union
∩ Intersection
A→B A mapping into B
a 7→ b a maps to b
,→ Inclusion map
A\B A setminus B
∼
= Isomorphic to
A := ... A is defined to be ...
□ End of a proof
viii

Symbol Name Symbol Name

α alpha β beta
γ gamma δ delta
π pi ϕ phi
φ var-phi ψ psi
ϵ epsilon ε var-epsilon
ζ zeta η eta
θ theta ι iota
κ kappa λ lambda
µ mu ν nu
υ upsilon ρ rho
ϱ var-rho ξ xi
σ sigma τ tau
χ chi ω omega
Ω Capital omega Γ Capital gamma
Θ Capital theta ∆ Capital delta
Λ Capital lambda Ξ Capital xi
Σ Capital sigma Π Capital pi
Φ Capital phi Ψ Capital psi

Some of the useful Greek alphabets

 1

Chapter 1

Foundation of Arithmetic

1.1 What is a Natural Number?

We begin with axiomatic definition of the set of all natural numbers, known as Peano’s axioms,
also known as Dedekind–Peano axioms. This was originally proposed by Richard Dedekind in
1988, and was published in a simplified version as a collection of axioms in 1989 by Giuseppe
Peano in his book Arithmetices principia, nova methodo exposita (in English: The principles of arith-
metic presented by a new method). We define addition and multiplication of natural numbers,
and briefly discuss their useful arithmetic properties (with outline of proofs) that we are famil-
iar with from elementary mathematics courses, without possibly thinking why and how these
work? The purpose of this section is to provide a logical foundation of natural numbers and their
arithmetic.

Axiom 1.1.1 (Peano’s axioms). There is a set N satisfying the following axioms.

(P1) 1 ∈ N (so N ̸= ∅); the element 1 is called one.

(P2) Axiom of equality: There is a relation “ = ” on N, called the equality, satisfying the following
properties.
(i) a = a, ∀ a ∈ N,
(ii) given a, b ∈ N, we have a = b ⇒ b = a, and
(iii) given a, b, c ∈ N, if a = b and b = c, then a = c.
In other words, the relation “ = ” on N is an equivalence relation on N. If “a = b”, we say that “a
is equal to b”. If “a = b” is not true, we say that “a is not equal to b”, expressed symbolically as
“a ̸= b”.
(Remark: The axiom (P2) was included in the original list of axioms published by Peano in 1889.
However, since the axiom (P2) is logically valid in first-order logic with equality, this is always
accepted, and is not considered to be a part of Axiom 1.1.1 in modern treatments.)
(P3) For each n ∈ N, there is a unique s(n) ∈ N, called the successor of n.
(P4) 1 is not a successor of any element of N.

(P5) Given m, n ∈ N with m ̸= n, we have s(m) ̸= s(n).

(P6) Principle of Mathematical Induction: If a subset S ⊆ N has properties that
(i) 1 ∈ S, and
(ii) n ∈ S ⇒ s(n) ∈ S,

then S = N.
2 Chapter 1. Foundation of Arithmetic

The elements of N are called natural numbers, and hence N is called the set of all natural numbers.

Exercise 1.1.2. Verify that s : N → N, n 7→ s(n) is injective but not surjective.

Remark 1.1.3. In contrast to our naive intuition, the properties (P1)–(P5) in Peano’s Axioms
1.1.1 do not guarantee that the successor function generates all natural numbers (we are famil-
iar with) except for 1. To make our naive intuition works, we need the assumption (P6), known
as the Principle of Mathematical Induction.

Lemma 1.1.4. If n ∈ N with n ̸= 1, then there is a unique element p(n) ∈ N, called the predecessor of
n, such that s(p(n)) = n.

Proof. Since s : N → N, n 7→ s(n) is injective by (P5), uniqueness of p(n) follows. To show

existence of p(n), for each n ∈ N \ {1}, it is enough to show that

s(N) := {s(n) : n ∈ N} = N \ {1}.

Since 1 ∈
/ s(N) := {s(n) : n ∈ N} by (P4), to show that s(N) = N \ {1}, it is enough to show that

T := s(N) ∪ {1} = N.

Clearly T ⊆ N and 1 ∈ T . If m ∈ T , then m = 1 or m = s(n), for some n ∈ N, and so in

both cases, s(m) ∈ T by construction of T . Then (P6) tells us that T = N. This completes the
proof.

Definition 1.1.5. A binary operation on a set S is a map S × S → S.

Definition 1.1.6. On the set N, we define two binary operations

(1.1.7) Addition + : N × N → N, (m, n) 7→ m + n,

(1.1.8) and Multiplication · : N × N → N, (m, n) 7→ m · n .

using the following rules given by the recurrence relations1 :

Rule for addition of natural numbers:

(1.1.9) n + 1 := s(n), ∀ n ∈ N, and

(1.1.10) n + s(m) := s(n + m), ∀ n, m ∈ N.

Rule for multiplication of natural numbers:

(1.1.11) n · 1 := n, ∀ n ∈ N, and
(1.1.12) n · s(m) := (n · m) + n, ∀ n, m ∈ N.

Lemma 1.1.13. The above rules (1.1.9)–(1.1.10) defines a unique binary operation on N, called addition
of natural numbers satisfying those properties.

Proof. To check uniqueness of the binary operation + satisfying the properties (1.1.9)–(1.1.10),
let ⊕ : N × N → N be any binary operation on N satisfying the following properties:

(A’) n ⊕ 1 = s(n), ∀ n ∈ N, and

(B’) n ⊕ s(m) = s(n ⊕ m), ∀ n, m ∈ N.

Let m ∈ N be arbitrary but fixed after choice. Let A := {n ∈ N : m + n = m ⊕ n} ⊆ N. Since

m + 1 = s(m) = m ⊕ 1, 1 ∈ A. If n ∈ A, then m + n = m ⊕ n, and so m + s(n) = s(m + n) =
1A relation that recalls itself repeatedly to generate its complete meaning.
1.1. What is a Natural Number? 3

s(m ⊕ n) = m ⊕ s(n). Therefore, s(n) ∈ A, and hence by principle of mathematical induction

(see (P6) in Axiom 1.1.1) we have A = N. This proves uniqueness of the binary operation + on
N.
Let n ∈ N be arbitrary but fixed after choice. Let

Tn := {m ∈ N : n + m is defined}.

Clearly Tn ⊆ N. We want to show that Tn = N. Now 1 ∈ Tn by axiom (1.1.9). If m ∈ Tn , then

n + m is defined, and so by axiom 1.1.10 n + s(m) is defined. So s(m) ∈ Tn . Then by principle
of mathematical induction (see (P6) in Peano’s Axiom 1.1.1) we have Tn = N.

Lemma 1.1.14. The above rules (1.1.11)–(1.1.12) defines a unique binary operation on N, called the
multiplication of natural numbers satisfying those properties.

Proof. Left as an exercise.

Now you know why and how you could add and multiply any two natural numbers!

Definition 1.1.15. Let ∗ : S × S → S be a binary operation on a set S. We say that ∗

(i) is associative if (a ∗ b) ∗ c = a ∗ (b ∗ c), ∀ a, b, c ∈ S;

(ii) is commutative if a ∗ b = b ∗ a, ∀ a, b ∈ S;
(iii) distributes over a binary operation ⊞ : S × S → S if for all a, b, c ∈ S we have

a ∗ (b ⊞ c) = (a ∗ b) ⊞ (a ∗ c),
(a ⊞ b) ∗ c = (a ∗ c) ⊞ (b ∗ c).

The following result is well-known, however, it is strongly recommended to verify these

in details purely using Peano’s Axioms 1.1.1, and the axioms (or, definition) for addition and
multiplication (1.1.9)–(1.1.12).
Theorem 1.1.16. For all a, b, c ∈ N, the following statements hold.

(i) Associativity for addition: (a + b) + c = a + (b + c).

(ii) Commutativity for addition: a + b = b + a.

(iii) Left distribution of multiplication over addition: a · (b + c) = (a · b) + (a · c).
(iv) Right distribution of multiplication over addition: (a + b) · c = (a · c) + (b · c).
(v) Commutativity for multiplication: a · b = b · a.

(vi) Associativity for multiplication: (a · b) · c = a · (b · c).

Proof. (i) Proof of associativity of addition: Let a, b ∈ N be arbitrary but fixed after choices. Let

Ta,b := {c ∈ N : a + (b + c) = (a + b) + c}.

Clearly Ta,b ⊆ N. To prove associativity for addition, we need to show that Ta,b = N.
Since

a + (b + 1) = a + s(b), by axiom (1.1.9).

= s(a + b), by axiom 1.1.10.
= (a + b) + 1, by axiom 1.1.9,
4 Chapter 1. Foundation of Arithmetic

we conclude that 1 ∈ Ta,b . Suppose that c ∈ Ta,b be arbitrary. Then

a + (b + s(c)) = a + s(b + c), by axiom (1.1.10).

= s(a + (b + c)), by axiom (1.1.10).
= s((a + b) + c), by axiom (1.1.10).
= (a + b) + s(c), by axiom (1.1.10).

Therefore, s(c) ∈ Ta,b . Then by principle of mathematical induction (see (P6) in Peano’s
Axiom 1.1.1) we have Ta,b = N. (Now you know why 1 + (2 + 3) = (1 + 2) + 3.)
(ii) Proof of commutativity of addition: For each a ∈ N, let Sa := {b ∈ N : a + b = b + a} ⊆ N. We
first show that S1 = N. Clearly 1 ∈ S1 . If b ∈ S1 , then

s(b) + 1 = s(s(b)), by axiom 1.1.9.

= s(b + 1), by axiom 1.1.9.
= s(1 + b), since b ∈ S1 by assumption.
= 1 + s(b), by axiom 1.1.10.

Therefore, s(b) ∈ S1 , and hence S1 = N by (P6) in Axiom 1.1.1. Now let a ∈ N be arbitrary
but fixed after choice. Since S1 = N, we have 1 ∈ Sa . If b ∈ Sa , then a + b = b + a, and so
we have

a + s(b) = s(a + b), by axiom (1.1.10).

= s(b + a), since b ∈ Sa by assumption.
= (b + a) + 1, by axiom (1.1.9).
= 1 + (b + a), since b + a ∈ N = S1 .
= (1 + b) + a, using associativity of addition.
= (b + 1) + a, since b ∈ N = S1 .
= s(b) + a, by axiom (1.1.9).

Then s(b) ∈ Sa , and hence by (P6) in Peano’s Axiom 1.1.1 we have Sa = N.

(iii) Proof of left distribution of multiplication over addition: Let a, b ∈ N be arbitrary but fixed after
choices. Let
Da,b := {c ∈ N : a · (b + c) = (a · b) + (a · c)} ⊆ N.
We need to show that Da,b = N. Note that,

a · (b + 1) = a · s(b), by axiom (1.1.9);

= (a · b) + a, by axiom (1.1.12);
= (a · b) + (a · 1), by axiom (1.1.9).

Therefore, 1 ∈ Da,b . Suppose that c ∈ Da,b . Then

a · (b + s(c)) = a · s(b + c), by axiom (1.1.10);

= a · (b + c) + a, by axiom (1.1.12);
= ((a · b) + (a · c)) + a, since c ∈ Da,b by assumption;
= (a · b) + ((a · c) + a), by associativity for addition;
= (a · b) + (a · s(c)), by axiom (1.1.12).

So s(c) ∈ Da,b . Therefore, by (P6) of Axiom 1.1.1 we have Da,b = N.

(iv) Proof of right distribution of multiplication over addition: Left as an exercise.
1.1. What is a Natural Number? 5

(v) Proof of commutativity of multiplication: Given a ∈ N, let

Sa := {b ∈ N : a · b = b · a}.

We first consider the case a = 1. Clearly 1 ∈ S1 . If b ∈ S1 , then

1 · s(b) = 1 · (b + 1), by axiom (1.1.9).

= (1 · b) + (1 · 1), by left distribution of multiplication over addition.
= (b · 1) + (1 · 1), since b ∈ S1 .
= (b + 1) · 1, by right distribution of multiplication over addition.
= s(b) · 1, by axiom (1.1.9).

Thus, s(b) ∈ S1 . Therefore, by principle of mathematical induction we have S1 = N. Now

assume that a ̸= 1. Since S1 = N, we have 1 ∈ Sa . Suppose that b ∈ Sa . Then

a · s(b) = (a · b) + a, by axiom (1.1.12).

= (b · a) + (1 · a), since 1 ∈ Sa ⇒ 1 · a = a · 1 = a.
= (b + 1) · a, by Theorem 1.1.16 (iv).
= s(b) · a, by axiom 1.1.9.

So s(b) ∈ Sa , and hence Sa = N by principle of mathematical induction.

(vi) Proof of associativity of multiplication: Left as an exercise! Let a, b ∈ N be arbitrary but fixed
after choice. Let
Ma,b := {c ∈ N : a · (b · c) = (a · b) · c}.
Clearly Ma,b ⊆ N. To prove associativity for multiplication, we need to show that Ma,b =
N. Since n · 1 = n, ∀ n ∈ N by axiom (1.1.11), we have a · (b · 1) = a · b = (a · b) · 1. So
1 ∈ Ma,b . Suppose that c ∈ Ma,b . Then

a · (b · s(c)) = a · ((b · c) + b), by axiom (1.1.12).

= a · (b · c) + (a · b), by Theorem 1.1.16 (iii).
= (a · b) · c + (a · b), by Theorem 1.1.16 (v).
= (a · b) · s(c), by axiom 1.1.12.

Therefore, s(c) ∈ Ma,b , and hence by principle of mathematical induction we have Ma,b =
N.

Proposition 1.1.17. For each n, a ∈ N, we have sn (a) = a + n, where sn : N → N is the n-times

composition of s with itself (e.g., s2 = s ◦ s, s3 = s ◦ s ◦ s etc.).

Proof. Let T := {n ∈ N : sn (a) = a + n, ∀ a ∈ N}. Clearly T ⊆ N, and 1 ∈ T by axiom 1.1.9.

Assume that n ∈ T . Then ss(n) (a) = sn+1 (a) = s(sn (a)) = s(a + n) = (a + n) + 1 = a + (n + 1) =
a + s(n). So s(n) ∈ T . Then by principle of mathematical induction we have T = N.
Lemma 1.1.18. Let a, b, n ∈ N. If a + n = b + n, then a = b.

Proof. Note that the successor map s : N → N is injective by (P5) in Axiom 1.1.1. Since sn (a) =
a + n = b + n = sn (b) by Proposition 1.1.17, and composition of injective maps is injective, we
have a = b.

Exercise 1.1.19 (Cancellation for multiplication). Let a, b, r, ℓ ∈ N.

(i) If ℓa = ℓb, show that a = b.

6 Chapter 1. Foundation of Arithmetic

(ii) If ar = br, show that a = b.

Exercise 1.1.20. Let a ∈ N. Show that the equation x + a = 1 has no solution for x in N.
Theorem 1.1.21 (Law of trichotomy for natural numbers). Given a, b ∈ N, exactly one of the
following three conditions holds:

(i) a = b,
(ii) a = b + c, for some c ∈ N, or
(iii) b = a + d, for some d ∈ N.

Proof. We first show that no two conditions among (i)–(iii) can hold simultaneously. If (i) and
(ii) holds simultaneously, then b = b + c implies s(b) = s(b + c) ⇒ b + 1 = (b + c) + 1 =
b + (c + 1) ⇒ 1 = c + 1 = s(c), which contradicts axiom (P4) in Peano’s Axioms 1.1.1.
The same argument shows that (i) and (iii) cannot hold simultaneously. If (ii) and (iii) hold
simultaneously, then we have a = b + c = (a + c) + d = a + (c + d), for some c, d ∈ N. Then
applying successor map we see that a + 1 = (a + (c + d)) + 1 = a + ((c + d) + 1). Then by
Lemma 1.1.18 we have 1 = (c + d) + 1 = s(c + d), which contracts (P4) in Peano’s Axioms 1.1.1.
Therefore, no two conditions among (i)–(iii) can hold simultaneously.
We now show that at least one of (i)–(iii) holds. For each a ∈ N, let

Sa := {b ∈ N : at least one of (i) or (ii) or (iii) holds}.

Consider the case a = 1. Clearly 1 ∈ S1 . Suppose that b ∈ S1 . Then s(b) = b + 1 = a + b satisfies

condition (iii), and so s(b) ∈ S1 . Then by (P6) in Peano’s Axioms we have S1 = N. Suppose
that a ∈ N \ {1} be arbitrary but fixed after choice. Since b = 1 satisfies a = s(p(a)) = p(a) + 1 =
p(a) + b, with p(a) ∈ N, the condition (ii) holds for b = 1, and so 1 ∈ Sa . Suppose that b ∈ Sa .
Then we have the following cases:

(I) If a = b, then s(b) = b + 1 = a + 1, and so s(b) satisfies condition (iii). So s(b) ∈ Sa .

(II) If a = b + c, for some c ∈ N, then a = s(b) or a = s(b) + p(c) depending on whether c = 1

or c ∈ N \ {1}, respectively. So in both cases, s(b) ∈ Sa .
(III) If b = a + d, for some d ∈ N, then s(b) = b + 1 = a + (d + 1) satisfies condition (iii), and
hence s(b) ∈ Sa .

Therefore, Sa = N by principle of mathematical induction.

The law of trichotomy in Theorem 1.1.21 allow us to define usual order relation “ < ” on N
as follow.
Definition 1.1.22. Given a, b ∈ N, we define a < b if ∃ c ∈ N such that a + c = b. If a < b, we
say that “a is strictly less than b”.

Note that “ < ” is a relation on N which is neither reflexive nor symmetric or anti-symmetric.
We show that it is a transitive relation on N. If a < b and b < c, then a + r = b and b + s = c, for
some r, s ∈ N, and then a + (r + s) = (a + r) + s = b + s = c shows that a < c. If a < b we say
that “a is less than b”. The relation “ < ” is called the usual ordering relation on N. Define another
relation “ ≤ ” on N by setting

a ≤ b if either a = b or a < b.

If a ≤ b, we say that “a is less than or equal to b′′ . It is easy to see that “ ≤ ” is reflexive and
transitive. We show that “ ≤ ” is anti-symmetric, and hence is a partial order relation on N.
1.1. What is a Natural Number? 7

Suppose that a, b ∈ N with a ≤ b and b ≤ a. We want to show that a = b. Suppose on the

contrary that a ̸= b. Then we must have a < b and b < a. Then there exist c, d ∈ N such that
a + c = b and b + d = a. Then

(b + d) + c = a + c = b
⇒ b + (d + c) = b, using associativity of addition.
⇒ s(b + (d + c)) = s(b), applying successor map.
⇒ (b + (d + c)) + 1 = b + 1, using axiom 1.1.10.
⇒ b + ((d + c) + 1) = b + 1, using associativity of addition.
⇒ (d + c) + 1 = 1, using Lemma 1.1.18.
⇒ s(d + c) = 1.

This contradicts axiom (P4) in Peano’s Axioms 1.1.1. Therefore, we must have a = b as required.
Definition 1.1.23. A partial order relation ρ on a set S is called a total order if for any two
elements a, b ∈ S, at least one of a ρ b and b ρ a holds. A non-empty set S together with a total
order relation is called a well-ordered set.

As an immediate consequence of the law of trichotomy of natural numbers (Theorem 1.1.21)

we see that “ ≤ ” is a total order relation on N, and hence (N, ≤) is a well-ordered set.
Theorem 1.1.24. The following are equivalent.

(i) Principle of mathematical induction (regular version): Let S ⊆ N be such that

(a) 1 ∈ S, and
(b) for each n ∈ N, n ∈ S implies s(n) ∈ S.
Then S = N.
(ii) Principle of mathematical induction (strong version): Let T ⊆ N be such that
(a’) 1 ∈ T , and
(b’) for each n ∈ N, Jn := {k ∈ N : k ≤ n} ⊆ T implies s(n) ∈ T .
Then T = N.

Proof. (i) ⇒ (ii): Suppose that the conditions (a′ ) and (b′ ) holds for T ⊆ N. Since 1 ∈ T by (a′ ),
to show T = N using the regular version of principle of mathematical induction (i), it is enough
to show that for each n ∈ N, the statement

Pn : “n ∈ T implies s(n) ∈ T.”

holds. Consider the set

S := {n ∈ N : Pk holds, ∀ k ≤ n} ⊆ N.
Since 1 ∈ T by (a ), we have J1 = {1} ⊆ T , and hence by (b′ ) we have s(1) ∈ T . Therefore, P1
′

holds, and so 1 ∈ S. Let n ∈ S be arbitrary but fixed after choice. Then P1 , . . . , Pn hold, and
hence we have Js(n) = {k ∈ N : k ≤ s(n)} ⊆ T . Then by the condition (b′ ) we have s(s(n)) ∈ T ,
and hence Ps(n) holds. Therefore, Pk holds, ∀ k ≤ s(n), and hence s(n) ∈ S. Then by (i) we
have S = N. Thus, T = N.
(ii) ⇒ (i): Let S ⊆ N be such that 1 ∈ S, and n ∈ S implies s(n) ∈ S. To show S = N using
the strong version of principle of mathematical induction (ii), we just need to ensure that for
each n ∈ N, if Jn ⊆ S then s(n) ∈ S. But this follows because n ∈ Jn implies that n ∈ S, and so
s(n) ∈ S by (a). Then by (ii) we have S = N. This proves (i).
Theorem 1.1.25. The following are equivalent.
8 Chapter 1. Foundation of Arithmetic

(i) Principle of Mathematical Induction (strong version): Let S ⊆ N be such that

(a) 1 ∈ S, and
(b) for each n ∈ N with n > 1, if {k ∈ N : k < n} ⊆ S then n ∈ S.
Then S = N.
(ii) Well-ordering principle of (N, ≤): Any non-empty subset of N has a least element.

Proof. (i) ⇒ (ii): Suppose on the contrary that there is a non-empty subset S ⊆ N which has no
least element. Let
T := N \ S = {n ∈ N : n ∈ / S}.
Since 1 is the least element of N, we have 1 ∈
/ S; for otherwise 1 would be the least element of S.
Therefore, 1 ∈ T and hence T is a non-empty subset of N. Let n ∈ N with n > 1, and suppose
that for any k ∈ N with k < n, we have k ∈ T . Then n ∈ / S, for otherwise n would be the least
element of S. So n ∈ T . Then by principle of mathematical induction (strong version), we have
T = N. This contradicts our assumption that S is non-empty. So S must have a least element.
(ii) ⇒ (i): Let S ⊆ N be such that

(a) 1 ∈ S, and
(b) for each n ∈ N with n > 1, if {k ∈ N : k < n} ⊆ S then n ∈ S.

Assuming well-ordering principle of (N, ≤), we want to show that S = N. Suppose on the
contrary that S ̸= N. Then T := N \ S is a non-empty subset of N, and so by (i) it has a least
element, say n ∈ T . Since 1 ∈ S by assumption, n > 1. Since n is the least element of T , for any
k ∈ N with k < n, we have k ∈ N \ T = S. Then by property (b) of S we have n ∈ S, which is a
contradiction. This completes the proof.

1.2 Integers: Construction & Basic Operations

Let a, b ∈ N. Suppose that we want to solve the equation

(1.2.1) x+a=b

to find x. If a < b in N, then there is r ∈ N such that b = r + a. If there is another number s ∈ N

such that b = s + a, then r + a = s + a implies r = s. So the solution of the equation (1.2.1) exists
and is unique; we denote this solution by a − b ∈ N. Now the problem is if a ≤ b, we don’t
have any solution of this equation in N. This forces us to enlarge our natural number system to
a bigger number system where we can find solutions to such linear equations.
Define a relation ∼ on the Cartesian product N × N by setting

(1.2.2) (a, b) ∼ (c, d), if a + d = b + c.

It is an easy exercise to show that ∼ is an equivalence relation on N × N. The ∼-equivalence

class of (a, b) ∈ N × N is the subset

(1.2.3) [(a, b)] := {(c, d) ∈ N × N (a, b) ∼ (c, d)}.

Let Z := {[(a, b)] : a, b ∈ N} be the associated set of all ∼-equivalence classes. The idea is to
think of the equivalence class [(a, b)] to be the solution of the equation x + b = a. The elements
of Z are called integers, and Z is called the set of all integers.
Define a map
ι:N→Z
1.2. Integers: Construction & Basic Operations 9

by
ι(n) = [(s(n), 1)], ∀ n ∈ N.
Then ι(n) = ι(m) ⇒ [(s(n), 1)] = [(s(m), 1)] ⇒ s(n) + 1 = s(m) + 1 ⇒ s(s(n)) = s(s(m)) ⇒
s(n) = s(m) ⇒ n = m, since s : N → N is an injective map. Therefore, ι : N → Z is an injective
map, and hence we can use it to identify N as a subset of Z. For notational simplicity, we may
denote by n the element [(s(n), 1)] ∈ Z, for all n ∈ N.
Define a binary operation on Z, called addition of integers, by

(1.2.4) [(a, b)] + [(c, d)] := [(a + c, b + d)], ∀ [(a, b)], [(c, d)] ∈ Z.

Note that, if (a, b) ∼ (a′ , b′ ) and (c, d) ∼ (c′ , d′ ), then (a + c, b + d) ∼ (a′ + c′ , b′ + d′ ). Therefore,
we have a well-defined binary operation + on Z.
Exercise 1.2.5. Show that the addition of integers is associative and commutative.

Exercise 1.2.6. Verify that,

ι(m + n) = ι(m) + ι(n), ∀ m, n ∈ N.

Therefore, the addition operation on integers preserves the addition operation on natural num-
bers defined earlier.

Note that, the element [(1, 1)] ∈ Z satisfies

[(a, b)] + [(1, 1)] = [(a, b)] = [(1, 1)] + [(a, b)].

We denote by 0 (pronounced as zero) the element [(1, 1)] ∈ Z. Since

[(s(n), 1)] + [(1, s(n))] = [(1, 1)] = 0, ∀ n ∈ N,

for notational simplicity (for peaceful working notations), we denote by −n the element [(1, s(n))] ∈
Z, for all n ∈ N. The element of Z of the form n and −n are called positive integers and negative
integers, respectively.
Exercise 1.2.7. The subsets Z− := {[(1, s(n))] : n ∈ N}, {0} := {[(1, 1)]} and Z+ := {[(s(n), 1)] :
n ∈ N} are mutually disjoint, and their union is Z. As a result, we may write the set Z as

Z = {−n : n ∈ N} ∪ {0} ∪ N.

The elements of Z− and Z+ are called the negative integers and the positive integers, respectively.

We define another binary operation on Z, called the product operation, by

(1.2.8) [(a, b)] · [(c, d)] := [(ac + bd, ad + bc)], ∀ [(a, b)], [(c, d)] ∈ Z.

It is easy to check that, if (a, b) ∼ (a′ , b′ ) and (c, d) ∼ (c′ , d′ ), then (ac + bd, ad + bc) ∼ (a′ c′ +
b′ d′ , a′ d′ + b′ c′ ), and hence the product operation is well-defined. One can easily check that,

(i) [(a, b)] · [(c, d)] = [(c, d)] · [(a, b)],

(ii) [(s(m), 1)] · [(s(n), 1)] = [(s(mn), 1)].
Remark 1.2.9. With the above definitions and notations, one can check that the binary opera-
tions addition and multiplication of integers are associative, commutative, and multiplication
distributes over addition. In other words, the following properties hold.

(i) (a + b) + c = a + (b + c), ∀ a, b, c ∈ Z;
10 Chapter 1. Foundation of Arithmetic

(ii) (ab)c = a(bc), ∀ a, b, c ∈ Z;

(iii) a + b = b + a, ∀ a, b ∈ Z;
(iv) ab = ba, ∀ a, b ∈ Z;
(v) a(b + c) = (ab) + (ac), ∀ a, b, c ∈ Z;
(vi) (a + b)c = (ac) + (bc), ∀ a, b, c ∈ Z.

Exercise 1.2.10. Let n ∈ Z. If a + n = b + n, for some a, b ∈ Z, show that a = b.

We define the usual ordering relation “ ≤ ” on Z as follow: given m, n ∈ Z, we define

m ≤ n if ∃ r ∈ N ∪ {0} such that m + r = n.

Exercise 1.2.11. Verify that (Z, ≤) is a well-ordered set.

1.3 Division Algorithm

Recall that the well-ordering principle of natural numbers says that any non-empty subset S of
N has a least element. This means, there exists n ∈ S such that n ≤ m, for all m ∈ S. This
statement is equivalent to the principle of mathematical induction, which says that if S ⊆ N is
such that 1 ∈ S, and for each n ∈ N, n ∈ S ⇒ n + 1 ∈ S, then S = N.
Theorem 1.3.1 (Division algorithm). Given a, d ∈ Z with d > 0, there exists unique q, r ∈ Z with
0 ≤ r < d such that a = qd + r.

Proof. We first show uniqueness of q and r. Suppose that we have another pair of integers
q ′ , r′ ∈ Z such that 0 ≤ r′ < d and a = q ′ d + r′ . Without loss of generality we may assume that
r ≤ r′ . Then qd + r = a = q ′ d + r′ implies r′ − r = (q − q ′ )d. Since 0 ≤ r ≤ r′ < d, we have
0 ≤ (q − q ′ )d = r′ − r < d. Therefore, (q − q ′ )d is a non-negative integer which is strictly less
than d and is a multiple of d. This is possible only if (q − q ′ )d = 0. Since d ̸= 0, we must have
q = q ′ , and hence r = r′ . This proves uniqueness part.
To show existence, consider the set

S := {a − dq : q ∈ N} ∩ N.

Since d > 0, choosing q sufficiently small we can ensure that a − dq ∈ N, and hence S ̸= ∅. Then
by well-ordering principle of (N, ≤), S has a least element, say r0 . Then 0 ≤ r0 = q − dq0 , for
some q0 ∈ Z. We claim that r0 < d. If not, then r0 ≥ d and hence 0 ≤ r0 − d = a − d(q + 1)
implies that r0 − d ∈ S. Since d > 0, it contradicts the fact that r0 is the least element of S.
Therefore, we must have r0 < d. This completes the proof.
Definition 1.3.2. The absolute value of n ∈ Z is the integer |n| defined by

n, if n ≥ 0,
|n| :=
−n, if n < 0.

Corollary 1.3.3. Given a, d ∈ Z with d ̸= 0, there exists unique q, r ∈ Z with 0 ≤ r < |d| such that
a = dq + r.

Proof. If d > 0, this is precisely Theorem 1.3.1. If d < 0, then d′ := −d > 0, and so by division
algorithm (Theorem 1.3.1) we find unique integers q, r ∈ Z with 0 ≤ r < d′ such that a = d′ q+r.
Then the integers q ′ := −q and r satisfies 0 ≤ r < |d| with a = q ′ d + r.
1.3. Division Algorithm 11

Definition 1.3.4. Given n, d ∈ Z, with d ̸= 0, we say that d divides n, written as d | n, if there is

an element q ∈ Z such that n = qd. Given finitely many integers a1 , . . . , an ∈ Z, which are not
all zero, we define their greatest common divisor to be a positive integer d ∈ Z+ such that

(i) d divides each of the numbers a1 , . . . , an , and

(ii) if an integer r divides ai , for all i = 1, . . . , n, then r divides d.

Remark 1.3.5. Given a finite number of integers a1 , . . . , an ∈ Z, if d and d′ are two greatest
common divisors of a1 , . . . , an , then d | d′ and d′ | d implies d ∈ {d′ , −d′ }. Since both d and d′
are positive integers, we must have d = d′ . Therefore, the greatest common divisor of a1 , . . . , an
is unique, and we denote it by gcd(a1 , . . . , an ). However, it is not yet clear if gcd(a1 , . . . , an )
exists in N. This requires a proof.
Lemma 1.3.6. Given m, n ∈ Z, not all zero, the greatest common divisor gcd(m, n) exists in N.
Moreover, there exist a, b ∈ Z such that gcd(m, n) = am + bn.

Proof. Let S := {am + bn : a, b ∈ Z}. Since at least one of m and n is non-zero, there is a
non-zero element, say x, in S. Then x = am + bn, for some a, b ∈ Z. If x < 0, then −x =
(−a)m + (−b)n ∈ S ∩ N. Therefore, S ∩ N is a non-empty subset of N. Then by well-ordering
principle of N, the non-empty subset S ∩ N has a least element, say d. Then d = a0 m + b0 n, for
some a0 , b0 ∈ Z. We claim that d = gcd(m, n).
If r | m and r | n, then r | (a0 m+b0 n) and so r | d. Now we need to show that d | m and d | n.
Let x ∈ S be arbitrary. Then x = am + bn ∈ S, for some a, b ∈ Z. By division algorithm we can
find q, r ∈ Z with 0 ≤ r < d such that x = qd + r. Then am + bn = x = qd + r = q(a0 m + b0 n) + r
implies r = (a − qa0 )m + (b − qb0 )n ∈ S. Since 0 ≤ r < d and d is the smallest positive integer
in S ∩ N, we must have r = 0. Therefore, x = qd and hence d | x, for all x ∈ S. In particular,
choosing (a, b) ∈ {(1, 0), (0, 1)}, we see that d | m and d | n. This completes the proof.

Definition 1.3.7. Given m, n ∈ Z, we say that m and n are relatively prime (or, coprime) if
gcd(m, n) = 1.
Corollary 1.3.8. Two integers m and n are coprime if and only if there exists a, b ∈ Z such that
am + bn = 1.

Proof. If gcd(m, n) = 1, then by above Lemma 1.3.6, there exists a, b ∈ Z such that am + bn = 1.
Conversely, suppose that am + bn = 1, for some a, b ∈ Z. If d = gcd(m, n), then d | m and d | n
implies d | 1. Then d ∈ {1, −1}. Since d > 0, we have d = 1.

Exercise 1.3.9. Given a finite number of integers a1 , . . . , an , not all zero, show that gcd(a1 , . . . , an )
exists in N.
Definition 1.3.10. An integer p ∈ Z is said to be a prime number if p > 1 and its only divisors in
Z are ±1, ±p.

Exercise 1.3.11 (Principle of mathematical induction). Fix n0 ∈ N. Prove that the following are
equivalent.

(i) Regular version: Let S ⊆ N be such that

(a) n0 ∈ S, and
(b) for any n ∈ N with n ≥ n0 , if n ∈ S then n + 1 ∈ S.
Then S = {n ∈ N : n ≥ n0 }.

(ii) Strong version: Let T ⊆ N be such that

(a’) n0 ∈ T , and
12 Chapter 1. Foundation of Arithmetic

(b’) for any n ∈ N with n ≥ n0 , if {k ∈ N : n0 ≤ k ≤ n} ⊆ T then n + 1 ∈ T .

Then T = {n ∈ N : n ≥ n0 }.

Assuming well-ordering principle of (N, ≤) show that the above two versions of induction
holds true.
Theorem 1.3.12 (Fundamental theorem of Arithmetic). Given a positive integer n > 1, there exists
a unique factorization of n as a product of positive integer powers of prime numbers. More precisely,
there exist finite number of unique prime numbers p1 , . . . , pk ∈ N with p1 > · · · > pk and positive
αk
integers α1 , . . . , αk ∈ N such that n = pα
1 · · · pk .
1
 13

Chapter 2

Group Theory

2.1 Group
A binary operation on a set A is a map ∗ : A × A → A; given (a, b) ∈ A × A its image under
the map ∗ is denoted by a ∗ b. We consider some examples of non-empty set together with a
natural binary operation and study list down their common properties.
Example 2.1.1. The set of all integers

Z = {0, ±1, ±2, ±3, ±4, ±5, . . .}

admits a binary operation, namely addition of integers:

+ : Z × Z −→ Z, (a, b) 7−→ a + b.

This binary operation has the following interesting properties:

(i) a + (b + c) = (a + b) + c, ∀ a, b, c ∈ Z,
(ii) there is an element 0 ∈ Z such that a + 0 = 0 + a = a, ∀ a ∈ Z,
(iii) for each a ∈ Z, there exists an element b ∈ Z (depending on a) such that a + b = b + a = 0;
the element b is denoted by −a.
Example 2.1.2. A symmetry on a non-empty set X is a bijective map from X onto itself. The set
of all symmetries of X is denoted by S(X). Note that S(X) admits a binary operation given by
composition of maps:

◦ : S(X) × S(X) −→ S(X), (f, g) 7−→ g ◦ f.

Note that

(i) given any f, g, h ∈ S(X), we have (f ◦ g) ◦ h = f ◦ (g ◦ h).

(ii) there is a distinguished element, the identity map IdX ∈ S(X) such that f ◦ IdX = f =
IdX ◦f , for all f ∈ S(X).
(iii) given any f ∈ S(X), there is a element g := f −1 ∈ S(X) such that f ◦ g = IdX = g ◦ f .
Example 2.1.3. Fix a natural number n ≥ 1, and consider the set GLn (R) of all invertible n × n
matrices with entries from R. Note that GLn (R) admits a natural binary operation given by
matrix multiplication:

· : GLn (R) × GLn (R) → GLn (R), (A, B) 7−→ AB.

Note that
14 Chapter 2. Group Theory

(i) given any A, B, C ∈ GLn (R), we have (AB)C = A(BC).

(ii) there is a distinguished element, the identity matrix In ∈ GLn (R) such that AIn = In A =
A, for all A ∈ GLn (R).
(iii) given any A ∈ GLn (R), there is a element B := A−1 ∈ GLn (R) such that AB = BA = In .

A non-empty set together with a binary operation satisfying the three properties listed in
the above examples is a mathematical model for many important mathematical and physical
systems; such a mathematical model is called a group. Here is a formal definition.
Definition 2.1.4. A group is a pair (G, ∗) consisting of a non-empty set G together with a binary
operation
∗ : G × G −→ G, (a, b) 7−→ a ∗ b,
satisfying the following conditions:

(G1) Associativity: a ∗ (b ∗ c) = (a ∗ b) ∗ c, for all a, b, c ∈ G.

(G2) Existence of neutral element: ∃ an element e ∈ G such that a ∗ e = e ∗ a = a, ∀ a ∈ G.
(G3) Existence of inverse: for each a ∈ G, there exists an element b ∈ G, depending on a, such
that a ∗ b = e = b ∗ a.

A semigroup is a pair (G, ∗) consisting of a non-empty set G together with an associative

binary operation ∗ : G × G → G (i.e., the condition (G1) holds). A monoid is a semigroup (G, ∗)
satisfying the condition (G2) as above. For example, (N, +) is a semigroup but not a monoid,
and (Z≥0 , +) is a monoid but not a group. However, we shall not deal with these two notations
in this text.
Example 2.1.5. (i) Trivial group: A singleton set {e} with the binary operation e ∗ e := e is a
group; such a group is called a trivial group.
(ii) The set G := {e, a}, with the binary operation ∗ given by a ∗ e = e ∗ a = a and a ∗ a = e, is
a group with two elements.
(iii) Verify that G := {e, a, b} together with the binary operation ∗ given by the following
multiplication table, is a group (with three elements).

∗ e a b
e e a b
a a b e
b b e a

TABLE 2.1: A group with 3 elements

Remark 2.1.6. For a group consisting of small number of elements, it is convenient to

write down the associated binary operation explicitly using a table as above, known as
the Cayley table.

(iv) The sets Z, Q, R and C form groups with respect to usual addition.
(v) The set Q∗ = Q \ {0} forms a group with respect to usual multiplication.
Exercise 2.1.7. Let (G, ∗) be a group.

(i) Uniqueness of neutral element: Show that the neutral element (also known as the identity
element) e ∈ G is unique.
2.1. Group 15

(ii) Uniqueness of inverse: Show that, for each a ∈ G, there is a unique element b ∈ G such that
a ∗ b = b ∗ a = e. The element b is called the inverse of a, and denoted by the symbol a−1 .
(iii) Cancellation Law: If a ∗ c = b ∗ c, for some a, b, c ∈ G, show that a = b.
(iv) Let a, b ∈ G. Show that ∃ unique x, y ∈ G such that a ∗ x = b and y ∗ a = b.

Let (G, ∗) be a group. We say that G is finite or infinite according as its underlying set G is
finite or infinite; the cardinality of G is called the order of the group (G, ∗), and we denote it by
the symbol |G|. For notational simplicity, we write ab to mean a ∗ b, for all a, b ∈ G; and for any
integer n ≥ 1, we denote by an the n-fold product of a with itself, i.e.,

an := | ∗ ·{z
a · · ∗ a} .
n-fold product of a

For a negative integer n, we define an := (a−1 )−n . When there is no confusion likely to arise,
we simply denote a group (G, ∗) by G without specifying the binary operation.
Exercise 2.1.8. Let G be a group.

(i) Show that (a−1 )−1 = a, for all a ∈ G.

(ii) Show that (ab)−1 = b−1 a−1 , for all a, b ∈ G.
(iii) Show that am an = am+n , for all m, n ∈ Z and a ∈ G.
(iv) Show that (am )n = amn , for all m, n ∈ Z and a ∈ G.
(v) Let a, b ∈ G be such that ab = ba. Show that (ab)n = an bn , for all n ∈ Z.
Example 2.1.9. (i) The set C∗ := C \ {0} of non-zero complex numbers forms a group with
respect to multiplication of complex numbers.
(ii) Circle group: The set
S 1 := {z ∈ C : |z| = 1}
forms a group with respect to multiplication of complex numbers.
(iii) Klein four-group: Consider the set K4 = {e, a, b, c} together with the binary operation

∗ : K4 × K4 −→ K4

defined by the Cayley table 2.2 below. Verify that K4 is a group.

∗ e a b c
e e a b c
a a e c b
b b c e a
c c b a e

TABLE 2.2: Klein four group

Exercise 2.1.10. Define a binary operation on R2 = R × R by

(x1 , y1 ) + (x2 , y2 ) := (x1 + x2 , y1 + y2 ), ∀ (x1 , y1 ), (x2 , y2 ) ∈ R2 .

Verify that (R2 , +) is a commutative group. Similarly, for each n ∈ N, show that the component-
wise addition of real numbers:

(2.1.11) (a1 , . . . , an ) + (b1 , . . . , bn ) := (a1 + b1 , . . . , an + bn ), ∀ aj , bj ∈ R,

defines a binary operation + on Rn which makes the pair (Rn , +) a commutative group.
16 Chapter 2. Group Theory

Definition 2.1.12. A map f : A → B is said to be

(i) injective if given any a1 , a2 ∈ A with f (a1 ) = f (a2 ), we have a1 = a2 ,

(ii) surjective if given any b ∈ B, there is an element a ∈ A such that f (a) = b,
(iii) bijective if f is both injective and surjective.
Exercise 2.1.13. Let A, B and C be three sets. Given maps f : A → B and g : B → C, we define
the composition of g with f , also called “g composed f ", to be the map g ◦ f : A → C defined by

(g ◦ f )(a) = g(f (a)), ∀ a ∈ A.

Prove the following.

(i) If both f and g are injective, so is g ◦ f : A → C.

(ii) If both f and g are surjective, so is g ◦ f : A → C.
(iii) If g ◦ f is injective, show that f is injective.
(iv) Give an example to show that g ◦ f could be injective without g being injective.
(v) If g ◦ f is surjective, show that g is surjective.
(vi) Give an example to show that g ◦ f could be surjective without f being surjective.
(vii) Given any set A, there is a map IdA : A → A defined by IdA (a) = a, ∀ a ∈ A, known as
the identity map of A. Verify that IdA is bijective.

(viii) If f : A → B is bijective, show that there is a bijective map fe : B → A such that fe◦f = IdA
and f ◦ fe = IdB . The bijective map fe : B → A, defined above, is called the inverse of f ,
and is usually denoted by f −1 .
Definition 2.1.14. A permutation on a set A is a bijective map from A onto itself.

For a non-empty set A, we denote by SA the set of all permutations on A. Let A be a non-
empty set. Define a binary operation on SA by

◦ : SA × SA −→ SA , (f, g) 7−→ g ◦ f.

Verify that (SA , ◦) is a group. (Hint: Use Exercise 2.1.13).

Example 2.1.15 (Symmetric group S3 ). Consider an equilateral triangle △ in a plane with its
vertices labelled as 1, 2 and 3. Consider the symmetries of △ obtained by its rotations by angles
2nπ/3, for n ∈ Z, around its centre, and reflections along a straight line passing through its top
vertex and centre. Note that, we have only six possible symmetries of △ as follow:
  
1 7→ 1 1 7→ 2 1 7→ 3
σ0 = 2 7→ 2 , σ1 = 2 7→ 3 , σ2 = 2 7→ 1 ,
3 7→ 3 3 7→ 1 3 7→ 2
  
  
1 7→ 1 1 7→ 3 1 7→ 2
σ3 = 2 7→ 3 , σ4 = 2 7→ 2 , σ5 = 2 7→ 1 .
3 7→ 2 3 7→ 1 3 7→ 3
  

Let S3 := {σ0 , σ1 , σ2 , σ3 , σ4 , σ5 }. Note that, each of symmetries are bijective maps from the set
J3 := {1, 2, 3} onto itself, and any bijective map from J3 onto itself is one of the symmetries
in S3 . Since composition of bijective maps is bijective (see Exercise 2.1.13), we get a binary
operation
S3 × S3 −→ S3 , (σi , σj ) 7−→ σi ◦ σj .
2.1. Group 17

Exercise 2.1.16. Write down the Cayley table for this binary operation on S3 defined by compo-
sition of maps, and show that S3 together with this binary operation is a group. Find σ1 , σ2 ∈ S3
such that σ1 ◦ σ2 ̸= σ2 ◦ σ1 .
Definition 2.1.17. The order of a group G is the cardinality of its underlying set G. We denote
this by |G|. In particular, if G is a finite set, then |G| is the number of elements of the set G.
Example 2.1.18. Let S4 be the set of all bijective maps from J4 := {1, 2, 3, 4} onto itself. Given
any two elements σ, τ ∈ S4 , note that their composition σ ◦ τ ∈ S4 . Thus we have a binary
operation on S4 given by sending (σ, τ ) ∈ S4 × S4 to σ ◦ τ ∈ S4 . Show that the set S4 together
with this binary operation (composition of bijective maps) is a non-commutative group of order
4! = 24.
Definition 2.1.19. Let A ⊆ R. A map f : A → R is said to be continuous at a ∈ A if given any
real number ϵ > 0, there is a real number δ > 0 (depending on both ϵ and a) such that for each
x ∈ A satisfying |a − x| < δ, we have |f (a) − f (x)| < ϵ. If f is continuous at each point of A, we
say that f is continuous on A.
Exercise 2.1.20. Let A ⊆ R, and let C(A) := {f : A → R f is continuous}. Verify that C(A) is
a group with respect to the binary operation defined for all f, g ∈ C(A) by the formula

(f + g)(x) := f (x) + g(x), ∀ x ∈ A.

Solution. Let f1 , f2 ∈ C(A). Let a ∈ A be arbitrary but fixed after choice. Since both f1 and f2
are continuous at a, given a real number ϵ > 0, there exist real numbers δ1 , δ2 > 0 such that
for each x ∈ A satisfying |a − x| < δj we have |fj (a) − fj (x)| < ϵ/2, for all j = 1, 2. Let δ :=
min{δ1 , δ2 }. Then δ > 0, and for any x ∈ A satisfying |a − x| < δ, we have |fj (a) − fj (x)| < ϵ/2,
for all j = 1, 2. Then we have,

|(f1 + f2 )(a) − (f1 + f2 )(x)| = |f1 (a) + f2 (a) − f1 (x) − f2 (x)|

= | f1 (a) − f1 (x) + f2 (a) − f2 (x) |
≤ |f1 (a) − f1 (x)| + |f2 (a) − f2 (x)|
ϵ ϵ
< + = ϵ.
2 2
Therefore, f1 + f2 is continuous at a ∈ A. Since a ∈ A is arbitrary, f1 + f2 is continuous at every
points of A, and hence f1 + f2 ∈ C(A). Since for given f1 , f2 , f3 ∈ C(A) and any x ∈ A, we
have



(f1 + f2 ) + f3 (x) = f1 + f2 (x) + f3 (x)


= f1 (x) + f2 (x) + f3 (x)


= f1 (x) + f2 (x) + f3 (x)


= f1 (x) + f2 + f3 (x)


= f1 + (f2 + f3 ) (x),

we have (f1 + f2 ) + f3 = f1 + (f2 + f3 ). Note that, the constant function

0:A→R

defined by sending all points of A to 0 ∈ R, given by 0(a) = 0, ∀ a ∈ A, is continuous (Hint:

given ϵ > 0, take any δ > 0), and satisfies f + 0 = f = 0 + f , for all f ∈ A. Given f ∈ C(A),
note that the function −f defined by (−f )(a) = −f (a), for all a ∈ A, is continuous on A (Hint:
given ϵ > 0, take the
same δ > 0 which works for f ), and satisfies f + (−f ) = (−f ) + f = 0.
Therefore, C(A), + satisfies all axioms of a group, and hence is a group.
Example 2.1.21 (Matrix groups). (i) Fix two integers m, n ≥ 1, and let Mm×n (R) be the set of
all m × n matrices with entries from R. Given A, B ∈ Mm×n (R), we define their addition
18 Chapter 2. Group Theory

to be the matrix A + B ∈ Mm×n (R) whose (i, j)-th entry is given by aij + bij , where aij
and bij are the (i, j)-th entries of A and B, respectively. Then we have a binary operation

+ : Mm×n (R) × Mm×n (R) −→ Mm×n (R), (A, B) 7−→ A + B.

Clearly, the set Mm×n (R) is non-empty, and the pair (Mm×n (R), +) satisfies the properties
(G1)–(G3) in Definition 2.1.4.
(ii) Matrix multiplication: Fix positive integers m, n, p, and let A ∈ Mm×n (R) and B ∈ Mn×p (R).
Define the product of A and B to be the m × p matrix AB ∈ Mm×p (R), whose (i, j)-th entry
is
X
(2.1.22) cij = aik bkj ,
k=1

where aik is the (i, k)-th entry of A, and bkj is the (k, j)-th entry of B.
Let A ∈ Mn×n (R). A matrix B ∈ Mn×n (R) is said to be the left inverse (resp., right inverse)
of A if BA = In (resp., AB = In ), where In ∈ Mn×n (R) whose (i, j)-th entry is

1, if i = j,
δij =
0, if i ̸= j.

Exercise 2.1.23. Show that the left inverse and the right inverse of A ∈ Mn×n (R), when
they exists, are the same. In other words, if AB = In and CA = In , for some B, C ∈
Mn×n (R), show that B = C.

A matrix A ∈ Mn×n (R) is said to be invertible if there is a matrix B ∈ Mn×n (R) such that
AB = BA = In .
General linear group: Let

GLn (R) = {A ∈ Mn×n (R) : A is invertible}

be the set of all invertible n × n matrices with real entries.

(a) Show that GLn (R) is a group with respect to matrix multiplication.
(b) Give examples of A, B ∈ GLn (R) such that A + B ∈
/ GLn (R).
(c) Give an example of A ∈ Mn×n (R) such that AB ̸= In , ∀ B ∈ Mn×n (R).
(d) Assuming n ≥ 2 give examples of A, B ∈ GLn (R) such that AB ̸= BA.

The group GLn (R) is called the general linear group (of degree n).

As we see in Example 2.1.21 that the relation ab = ba need not hold for all a, b ∈ G, in
general. We shall see later that the symmetric group S3 in Example 2.1.9 (2.1.15) is the smallest
such group; in this case, we have σ3 ◦ σ1 = σ4 while σ1 ◦ σ3 = σ5 .
Definition 2.1.24. A group G is said to be commutative (or, abelian) if ab = ba, for all a, b ∈ G. A
group which is not commutative (or, abelian) is called a non-commutative (or, non-abelian) group.

Exercise 2.1.25. (i) Verify that {e}, Z, C∗ , S 1 , K4 are abelian groups.

(ii) Show that S3 and GL2 (R) are non-abelian groups.
Exercise 2.1.26. Show that GLn (R) is not abelian, for all n ≥ 2.
Definition 2.1.27. A relation on a non-empty set A is a non-empty subset ρ ⊆ A×A. If (a, b) ∈ ρ,
sometimes we may express it as a ρ b, and call a is ρ-related to b in A. A relation ρ on A is said to
be
2.1. Group 19

(i) reflexive if (a, a) ∈ ρ, ∀ a ∈ A;

(ii) symmetric if (a, b) ∈ ρ implies (b, a) ∈ ρ;
(iii) anti-symmetric if (a, b) ∈ ρ and (b, a) ∈ ρ implies a = b;
(iv) transitive if (a, b) ∈ ρ and (b, c) ∈ ρ implies (a, c) ∈ ρ;
(v) equivalence if ρ is reflexive, symmetric and transitive; and
(vi) partial order if ρ is reflexive, anti-symmetric and transitive.

Let A be a non-empty set, and let ρ be an equivalence relation on A. The ρ-equivalence class
of an element a ∈ A is the subset

[a]ρ := {b ∈ A : (b, a) ∈ ρ} ⊆ A.

Proposition 2.1.28. With the above notations, given any a, b ∈ A, [a]ρ = [b]ρ if and only if (a, b) ∈ ρ.

Proof. Suppose that (a, b) ∈ ρ. Then for any c ∈ [a]ρ , we have (c, a) ∈ ρ. Since ρ is transitive,
from (c, a), (a, b) ∈ ρ we have (c, b) ∈ ρ, and so c ∈ [b]ρ . Therefore, [a]ρ ⊆ [b]ρ . Since ρ is
symmetric, (a, b) ∈ ρ implies (b, a) ∈ ρ. Then following above arguments, we conclude that
[b] ⊆ [a]. Therefore, [a]ρ = [b]ρ .
Conversely, suppose that [a]ρ = [b]ρ . Since ρ is reflexive, a ∈ [a]ρ . Then [a]ρ = [b]ρ implies
that a ∈ [b]ρ , and so (a, b) ∈ ρ. This completes the proof.
Proposition 2.1.29. With the above notations, given a, b ∈ A, either [a]ρ ∩ [b]ρ = ∅ or [a]ρ = [b]ρ .

Proof. It is enough to show that if [a]ρ ∩ [b]ρ ̸= ∅, then [a]ρ = [b]ρ . Let c ∈ [a]ρ ∩ [b]ρ . Then
(c, a), (c, b) ∈ ρ. Since ρ is symmetric, (c, a) ∈ ρ implies (a, c) ∈ ρ. Then (a, c) ∈ ρ and (c, b) ∈ ρ
together implies (a, b) ∈ ρ, since ρ is transitive. Then by Proposition 2.1.28 we have [a]ρ =
[b]ρ .
Definition 2.1.30. Let A be a non-empty set. A partition on A is a non-empty collection P :=
{Aα : α ∈ Λ}, where

(i) Aα ⊆ A, for all α ∈ Λ,

(ii) Aα ∩ Aβ = ∅, for α ̸= β in Λ, and
S
(iii) A = Aα .
α∈Λ

Proposition 2.1.31. To give an equivalence relation on a non-empty set is equivalent to give a partition
on it.

S an equivalence relation ρ on A. Since ρ is reflexive, a ∈ [a]ρ ,

Proof. Suppose that we have given
for all a ∈ A, and hence A = [a]ρ . Since ρ-equivalence classes of elements of A are either
a∈A
disjoint or equal (see Proposition 2.1.29), the collection P consisting of all distinct ρ-equivalence
classes of elements of A is a partition of A.
Conversely, suppose that P = {Aα : α ∈ Λ} be a partition of A. Define

ρ = {(a, b) ∈ A × A : a, b ∈ Aα , for some α ∈ Λ}.

Note that (a, a) ∈ ρ, for all a ∈ A. If (a, b) ∈ ρ, then both a and b are in the same Aα , for some
α ∈ Λ, and so (b, a) ∈ ρ. So ρ is symmetric. If (a, b), (b, c) ∈ ρ, then a, b ∈ Aα and b, c ∈ Aβ , for
some α, β ∈ Λ. Since b ∈ Aα ∩ Aβ , so we must have Aα = Aβ . Therefore, (a, c) ∈ ρ. Thus ρ is
transitive. Therefore, ρ is an equivalence relation on A. One should note that the elements of P
are precisely the ρ-equivalence classes in A (verify!).
20 Chapter 2. Group Theory

Example 2.1.32 (The groups Zn and Un ). Fix an integer n ≥ 2. Define a relation ≡n on Z by

setting
a ≡n b, if a − b = nk, for some k ∈ Z.
If a ≡n b sometimes we also express it as a ≡ b (mod n), and say that a is congruent to b modulo
n. Verify that ≡n is an equivalence relation on Z. Given any a ∈ Z, let

[a] := {b ∈ Z : b ≡n a} ⊆ Z

be the ≡n -equivalence class of a in Z. Let

Zn := {[a] : a ∈ Z}

be the set of all ≡n -equivalence classes of elements of Z. Let a, b ∈ Z. If c ∈ [a] ∩ [b], then
c = a + nk1 and c = b + nk2 , for some k1 , k2 ∈ Z. Then a − b = n(k1 − k2 ), and hence
a ≡n b. Then [a] = [b] in Zn . Therefore, the ≡n -equivalence classes are either disjoint or
identical (c.f. Proposition 2.1.29). Use division algorithm (Theorem 1.3.1) to show that ≡n -
equivalence classes [0], [1], . . . , [n − 1] are all distinct, and

Zn = {[k] : 0 ≤ k ≤ n − 1}.

In particular, Zn is a finite set containing n elements.

We now define two binary operations on Zn . Suppose that [a] = [a′ ] and [b] = [b′ ] in Zn , for
some a, a′ , b, b′ ∈ Z. Then we have

a − a′ = nk1 ,
and b − b′ = nk2 ,

for some k1 , k2 ∈ Z. Therefore,

(a + b) − (a′ + b′ ) = n(k1 − k2 ),

and hence [a + b] = [a′ + b′ ] in Zn . Therefore, we have a well-defined binary operation on Zn

(called addition of integers modulo n) given by

[a] + [b] := [a + b], ∀ [a], [b] ∈ Zn .

Now it is easy to see that,

(i) ([a] + [b]) + [c] = [a] + ([b] + [c]), for all [a], [b], [c] ∈ Zn .
(ii) [a] + [0] = [a] = [0] + [a], for all [a] ∈ Zn .
(iii) [a] + [−a] = [0], for all [a] ∈ Z.

Therefore, (Zn , +) is a group. Note that, for all [a], [b] ∈ Zn we have

[a] + [b] = [a + b] = [b + a], since addition in Z is commutative,

= [b] + [a].

Therefore, (Zn , +) is an abelian group.

Now we define multiplication operation on Zn . Suppose that [a] = [a′ ] and [b] = [b′ ]. Then
a − a′ = nk1 and b − b′ = nk2 , for some k1 , k2 ∈ Z. Then

ab − a′ b′ = (a − a′ )b + a′ (b − b′ )
= nk1 b + a′ nk2
= n(k1 b + a′ k2 ),
2.1. Group 21

implies that [ab] = [a′ b′ ]. Thus we have a well-defined binary operations on Zn (called the
multiplication of integers modulo n) defined by

[a] · [b] := [ab], ∀ [a], [b] ∈ Zn .

Clearly the multiplication modulo n operation on Zn is both associative and commutative.

Note that,
[1] · [a] = [a] = [a] · [1], ∀ [a] ∈ Zn .
Therefore, [1] ∈ Zn is the multiplicative identity in Zn . Moreover, the multiplication distributes
over addition from left and right on Zn . Indeed, we have

[a] · ([b] + [c]) = [a] · [b] + [a] · [c],

and ([a] + [b]) · [c] = [a] · [c] + [b] · [c] .

Such a triple (Zn , +, ·) is called a ring. Since n ≥ 2 by assumption, n does not divide 1 in Zn . So
[0] ̸= [1] in Zn by Proposition 2.1.28. Since for any [a] ∈ Zn , we have [0] · [a] = [0 · a] = [0] ̸= [1],
we see that [0] ∈ Zn has no multiplicative inverse in Zn . Therefore, (Zn , ·) is just a commutative
monoid, but not a group.
We now find out elements of Zn that have multiplicative inverse in Zn , and use them to
construct a subset of Zn which forms a group with respect to the multiplication modulo n
operation. Recall that given n, k ∈ Z, we have gcd(n, k) = 1 if and only if there exists a, b ∈ Z
such that an + bk = 1 (see Corollary 1.3.8). Use this to verify that if [k] = [k ′ ] in Zn , then
gcd(n, k) = 1 if and only if gcd(n, k ′ ) = 1. Thus we get a well-defined subset

Un := {[k] ∈ Zn : gcd(k, n) = 1} ⊂ Zn .

Note that, [0] ∈ / Un . If [k1 ], [k2 ] ∈ Un , then gcd(k1 , n) = 1 = gcd(k2 , n). Then there exists
a1 , b1 , a2 , b2 ∈ Z such that

a1 k1 + b1 n = 1
and a2 k2 + b2 n = 1.

Multiplying these two equations, we have

(a1 a2 )(k1 k2 ) + (a1 k1 b2 + a2 k2 b1 + b1 b2 )n = 1.

Then we have gcd(k1 k2 , n) = 1. Therefore,

[k1 ] · [k2 ] = [k1 k2 ] ∈ Un , ∀ [k1 ], [k2 ] ∈ Un .

Verify that (Un , ·) is an abelian group. If p > 1 is a prime number (see Definition 1.3.10), show
that Up = Zp \ {[0]}, as sets.
Exercise 2.1.33. Let X be a non-empty set. Let P(X) be the set of all subsets of X; called the
power set of X. Given any two elements A, B ∈ P(X), define

A △ B := (A \ B) ∪ (B \ A).


The set A △ B is known as the symmetric difference of A and B. Show that P(X), △ is a
commutative group. (Hint: The empty subset ∅ ⊂ X acts as the neutral element in P(X), and
every element of P(X) is inverse of itself).
Exercise 2.1.34 (Direct product of two groups). Let (A, ∗) and (B, ⋆) be two groups. Show that
the Cartesian product G1 × G2 is a group with respect to the binary operation on it defined by

(a1 , b1 )(a2 , b2 ) := (a1 ∗ a2 , b1 ⋆ b2 ), ∀ (a1 , b1 ), (a2 , b2 ) ∈ A × B.

The group A × B defined above is called the direct product of A with B.

22 Chapter 2. Group Theory

2.2 Subgroup
Definition 2.2.1 (Subgroup). Let G be a group. A subgroup of G is a subset H ⊆ G such that H
is a group with respect to the binary operation induced from G. A subgroup H of G is said to
be proper if H ̸= G. A subgroup whose underlying set is singleton is called a trivial subgroup.

For example, Z is a subgroup of Q; S 1 is a subgroup of C∗ etc.

Exercise 2.2.2. For each integer n, let nZ := {nk : k ∈ Z}.

(i) Show that nZ is a proper subgroup of Z, for all n ∈ Z \ {1, −1}.

(ii) Show that any subgroup of Z is of the form nZ, for some n ∈ Z.
Exercise 2.2.3 (Group of nth roots of unity). Fix an integer n ≥ 1, and let

µn := {ζ ∈ C | ζ n = 1}.

Show that µn is a subgroup of the circle group S 1 .

Exercise 2.2.4. Show that a finite subgroup of C∗ of order n is µn .
√
Exercise 2.2.5. Show that {1, −1, i, −i} is a subgroup of C∗ , where i = −1.
Exercise 2.2.6. For each integer n ≥ 1, show that there is a commutative group of order n.
Remark 2.2.7. It is easy to see that any subgroup of an abelian group is abelian. However,
the converse is not true, in general. For example, one can easily check that S3 is a non-abelian
group whose all proper subgroups are abelian.
Lemma 2.2.8. Let G be a group. A non-empty subset H ⊆ G forms a subgroup of G if and only if
ab−1 ∈ H, for all a, b ∈ H.

Proof. Since H ̸= ∅, there is an element a ∈ H. Then e = aa−1 ∈ H. In particular, for any b ∈ H,

its inverse b−1 = eb−1 ∈ H. Then for any a, b ∈ H, their product ab = a(b−1 )−1 ∈ H. Thus
H is closed under the binary operation induced from G. Associativity is obvious. Thus, H is a
subgroup of G.
Exercise 2.2.9 (Special linear group). Fix an integer n ≥ 1, and let

SLn (R) = {A ∈ GLn (R) : det(A) = 1},

where det(A) denotes the determinant of the matrix A. Show that SLn (R) is a non-trivial proper
subgroup of GLn (R). Also show that SLn (R) is non-commutative for n ≥ 2.
Proposition 2.2.10 (Center of a group). Let G be a group. Then

Z(G) := {a ∈ G : ab = ba, ∀ b ∈ G}

is a commutative subgroup of G, called the center of G.

Proof. Clearly e ∈ Z(G). Let a ∈ Z(G). Then for any c ∈ G we have

ac = ca ⇒ c = a−1 ca ⇒ ca−1 = a−1 caa−1 = a−1 c,

and hence a−1 ∈ Z(G). Then for any a, b ∈ Z(G), we have c(ab−1 )c−1 = cac−1 cb−1 c−1 = ab−1 ,
for all c ∈ G, and hence ab−1 ∈ Z(G). Therefore, Z(G) is a subgroup of G. Clearly Z(G) is
commutative.
Exercise 2.2.11. Show that a group G is commutative if and only if Z(G) = G.
2.2. Subgroup 23

Exercise 2.2.12. Find the centers of S3 , GLn (R) and SLn (R), where n ∈ N.

Exercise 2.2.13 (Centralizer). Let G be a group. Given an element a ∈ G show that the subset

CG (a) := {b ∈ G : ab = ba}
T
is a subgroup of G, called the centralizer of a in G. Show that Z(G) = CG (a).
a∈G

Lemma
T 2.2.14. Let G be a group, and let {Hα }α∈Λ be a non-empty collection of subgroups of G. Then
Hα is a subgroup of G.
α∈Λ

T T
Proof. Since e ∈ Hα , for all α ∈ Λ, we have e ∈ Hα . Let a, b ∈Hα be arbitrary. Since
α∈Λ α∈Λ
a, b ∈ Hα , for all α ∈ Λ, we have ab−1 ∈ Hα , for all α ∈ Λ, and hence ab−1 ∈
T
Hα . Thus
T α∈Λ
Hα is a subgroup of G.
α∈Λ

T and S a subset of G. Let CS be the collection of all subgroups of G

Corollary 2.2.15. Let G be a group
that contains S. Then ⟨ S ⟩ := H is the smallest subgroup of G containing S.
H∈CS

H is a subgroup of G containing S. If H ′ is any subgroup

T
Proof. By Lemma 2.2.14, ⟨ S ⟩ :=
H∈CS
of G containing S, then H ′ ∈ CS , and hence ⟨ S ⟩ := H ⊆ H ′.
T
H∈CS

Exercise 2.2.16. Recall Exercise 2.2.2, and find the subgroup 2Z ∩ 3Z of Z.

Exercise 2.2.17. Is 2Z ∪ 3Z a subgroup of Z? Justify your answer.

Exercise 2.2.18. Show that a group cannot be written as a union of its two proper subgroups.
T
Definition 2.2.19. Let G be a group and S ⊆ G. The group ⟨ S ⟩ := H is called the subgroup
H∈CS
of G generated by S. If S is a singleton subset S = {a} of G, we denote by ⟨ a ⟩.
Exercise 2.2.20. Let G be a group. Find the subgroup of G generated by the empty subset of G.
Proposition 2.2.21. Let G be a group, and let S be a non-empty subset of G. Then

⟨ S ⟩ = {ae11 · · · aenn | n ∈ N, and ai ∈ S, ei ∈ {1, −1}, ∀ i ∈ {1, 2, . . . , n}} .

Proof. Let

K := {ae11 · · · aenn | n ∈ N, and ai ∈ S, ei ∈ {1, −1}, ∀ i ∈ {1, 2, . . . , n}} .

Clearly S ⊂ K ⊆ G. Taking n = 2, a1 = a2 = a ∈ S, e1 = 1 and e2 = −1, we have e = a a−1 ∈

K. Let a, b ∈ K. Then a = ae11 · · · aenn and b = bf11 · · · bfmm , for some ai , bj ∈ S, ei , fj ∈ {1, −1},
1 ≤ i ≤ n, 1 ≤ j ≤ m, and m, n ∈ N. Then ab−1 = ae11 · · · aenn · (bf11 · · · bfmm )−1 = ae11 · · · aenn ·
b−f
m
m
· · · b−f
1
1
∈ K. Therefore, K is a subgroup of G containing S. Then by Proposition 2.2.15,
we have ⟨ S ⟩ ⊆ K. To see the reverse inclusion, note that ifTS ⊆ H, for some subgroup H of G,
then all the elements of K lies inside H. Therefore, K ⊆ H = ⟨ S ⟩.
H∈CS

Definition 2.2.22. A group G is said to be finitely generated if there exists a finite subset S ⊆ G
such that the subgroup generated by S is equal to G, i.e., ⟨ G⟩ = G.
Example 2.2.23. (i) Any finite group is finitely generated.

(ii) The additive group (Z, +) is finitely generated.

24 Chapter 2. Group Theory

Exercise 2.2.24. Let G and H be finitely generated groups. Verify if the direct product G × H
of G and H, as defined in Exercise 2.1.34, is finitely generated.
Example 2.2.25. Let G be a group. Given an element a ∈ G, the subgroup of G generated by a
can be written as
⟨ a ⟩ = {an : n ∈ Z};
and is called the cyclic subgroup of G generated by a.
Definition 2.2.26. Let G be a group. The order of an element a ∈ G is the smallest positive
integer n, if exists, such that an = e. If no such positive integer n exists, we say that the
order of a is infinite. We denote by ord(a) the order of a ∈ G. In other words, if we set
Sa := {n ∈ Z : n ≥ 1 and an = e}, then

inf Sa , if Sa ̸= ∅, and
ord(a) :=
∞, if Sa = ∅.

Exercise 2.2.27. Let G be a group and a, b ∈ G be such that ab = ba. Show that (ab)n = an bn ,
for all n ∈ N.
Exercise 2.2.28. Let G be a group. Let a, b ∈ G be elements of finite orders.

(i) If am = e, for some m ∈ N, then show that ord(a) | m.

ord(a)
(ii) Show that ord(an ) =
, for all n ∈ N.
gcd n, ord(a)

(iii) Show that both a and a−1 have the same order in G.
(iv) Show that both ab and ba have the same finite order in G.
Exercise 2.2.29. Let G be a group, and let a and b two elements of G of finite orders with ab = ba.

(i) Show that ord(ab) divides lcm(ord(a), ord(b)).

(ii) If gcd(ord(a), ord(b)) = 1, show that ord(ab) = ord(a) ord(b).
Remark 2.2.30. If we remove the assumption that ab = ba from the above Exercise 2.2.29 we can
say absolutely nothing about the order of the product ab. In fact, given any integers m, n, r > 1,
there exists a finite group G with elements a, b ∈ G such that ord(a) = m, ord(b) = n and
ord(ab) = r. The proof of this surprising fact requires some advanced techniques, and may
appear at the end of this course.
Exercise 2.2.31. Let G be an abelian group. Let H := {a ∈ G : ord(a) is finite}. Show that H is
a subgroup of G.
Exercise 2.2.32. Show that (Q, +) is not a finitely generated group.
Exercise 2.2.33. Find two elements σ and τ of S3 that generates it.
Exercise 2.2.34 (Derived subgroup). Let G be a group. The commutator of two elements a, b ∈ G
is the element [a, b] := aba−1 b−1 ∈ G. Given a, b ∈ G, show that

(i) [a, b] = e if and only if ab = ba;

(ii) [a, b]−1 = [b, a]; and
(iii) g[a, b]g −1 = [gag −1 , gbg −1 ], for all g ∈ G.

The subgroup [G, G] := ⟨ [a, b] : a, b ∈ G⟩ of G generated by all commutators of elements of

G is called the derived subgroup or the commutator subgroup of G. Show that [G, G] is a trivial
subgroup of G if and only if G is abelian.
2.3. Cyclic group 25

2.3 Cyclic group

Let G be a group. For any element a ∈ G, we consider the subset

⟨a⟩ := {an : n ∈ Z} ⊆ G.

Clearly e ∈ ⟨a⟩, and for any two elements an , am ∈ ⟨a⟩, we have an · (bm )−1 = an−m ∈ ⟨a⟩.
Therefore, ⟨a⟩ is a subgroup of G, called the cyclic subgroup of G generated by a. If H is any
subgroup of G with a ∈ H, then a−1 ∈ H, and hence an ∈ H, for all n ∈ Z. Therefore, ⟨ a ⟩ ⊆ H.
Therefore, ⟨ a ⟩ is the smallest subgroup of G containing a.
Definition 2.3.1. A group G is said to be cyclic if there is an element a ∈ G such that G = ⟨a⟩.
The element a is called the generator of ⟨a⟩.
Remark 2.3.2. If G is a cyclic group generated by a ∈ G, then ⟨ a−1 ⟩ = G. Therefore, if a2 ̸= e,
the cyclic group ⟨ a ⟩ has at least two distinct generators, namely a and a−1 . We shall see later
that if a cyclic group ⟨ a ⟩ has at least two distinct generators, then we must have a2 ̸= e.

For example, the additive group Z is a cyclic group generated by 1 or −1. It is clear that a
cyclic group may have more than one generators. For example, Z3 is a cyclic group that can be
generated by [1] or [2].

Example 2.3.3. Zn is a finite cyclic group generated by [1] ∈ Zn . To see this, note that for any
[m] ∈ Zn , we have [m] = [m · 1] = m[1] ∈ ⟨ [1] ⟩ ⊆ Zn . Therefore, Zn ⊆ ⟨ [1] ⟩, and hence
Zn = ⟨ [1] ⟩.
Proposition 2.3.4. Fix an integer n ≥ 2. Then [a] ∈ Zn is a generator of the group Zn if and only if
gcd(a, n) = 1.

Proof. Suppose that ⟨ [a] ⟩ = Zn . Then there exists m ∈ Z such that [1] = m[a] = [ma]. Then
n | (ma − 1) and so ma − 1 = nd, for some d ∈ Z. Therefore, ma + n(−d) = 1, and hence by
Corollary 1.3.8 we have gcd(a, n) = 1. Conversely, if gcd(a, n) = 1, then there exists m, q ∈ Z
such that am+nq = 1. Then n | (1−am) and hence [a] = [1] in Zn . Hence the result follows.
Corollary 2.3.5. For a prime number p > 0, Zp has p − 1 distinct generators.

Clearly any cyclic group is abelian. However, the converse is not true in general. For exam-
ple, the Klein four-group K4 in Example 2.1.9 (iii) is abelian but not cyclic (verify).
Exercise 2.3.6. Give an example of an infinite abelian group which is not cyclic.

Proposition 2.3.7. Subgroup of a cyclic group is cyclic.

Proof. Let G = ⟨ a ⟩ be a cyclic group generated by a ∈ G. Let H ⊆ G be a subgroup of G. If

H = {e} is the trivial subgroup of G, then H = ⟨ e ⟩. Suppose that H ̸= {e}. Then there exists
b ∈ G such that b ̸= e and b ∈ H. Since G = ⟨ a ⟩, we have b = an , for some n ∈ Z. Since H is a
group and an = b ∈ H, we have a−n = b−1 ∈ H. Therefore,

S := {k ∈ N : ak ∈ H} ⊆ N

is a non-empty subset of N. Then by well-ordering principle of (N, ≤) (see Theorem 1.1.25) S

has a least element, say m ∈ S. We claim that H = ⟨ am ⟩. Clearly ⟨ am ⟩ ⊆ H. Let h ∈ H be
arbitrary. Since H ⊆ G = ⟨ a ⟩, we have h = an , for some n ∈ Z. Then by division algorithm
(see Theorem 1.3.1) there exists q, r ∈ Z with 0 ≤ r < m such that n = mq + r. Then ar =
an−mq = an (am )−q = h(am )−q ∈ H. Since m is the least element of S, we must have r = 0.
Then n = mq, and so we have h = an = am q ∈ ⟨ am ⟩. Therefore, H ⊆ ⟨ am ⟩, and hence
H = ⟨ am ⟩.
26 Chapter 2. Group Theory

Exercise 2.3.8. Show that any subgroup of Z is of the form nZ := {nk : k ∈ Z}, for some n ∈ Z.

Lemma 2.3.9. Let G = ⟨ a ⟩ be an infinite cyclic group. Then for all m, n ∈ Z with m ̸= n, we have
an ̸= am .

Proof. Suppose not, then there exists m, n ∈ Z with m > n such that am = an . Then am−n =
am (an )−1 = e. Since m − n is a positive integer, the subset

S := {k ∈ N : ak = e} ⊆ N

is non-empty. Then by well-ordering principle S has a least element, say d. We claim that
G = {ak : k ∈ Z with 0 ≤ k ≤ d − 1}. Clearly {ak : k ∈ Z with 0 ≤ k ≤ d − 1} ⊆ G.
Let b ∈ G be arbitrary. Then b = an , for some n ∈ Z. Then by division algorithm (Theorem
1.3.1), there exists q, r ∈ Z with 0 ≤ r < d such that n = dq + r. Since d ∈ S, we have
ad = e. Then b = an = adq+r = (ad )q ar = ar ∈ {ak : k ∈ Z with 0 ≤ k ≤ d − 1} implies
G ⊆ {ak : k ∈ Z with 0 ≤ k ≤ d − 1}, and hence G = {ak : k ∈ Z with 0 ≤ k ≤ d − 1}. This is
not possible since G is infinite by our assumption. Hence the result follows.

Corollary 2.3.10. Let G = ⟨ a ⟩ be a cyclic group generated by a ∈ G. Then G is infinite if and only if
ord(a) is infinite.

Proof. If G = ⟨ a ⟩ is infinite, then for any non-zero integer n, we have an ̸= a0 = e by Lemma

2.3.9. Therefore, ord(a) is infinite. Conversely, if ord(a) is infinite, then an ̸= e, for all n ∈ Z\{0}.
Since an = am implies am−n = e, the map f : Z → G given by f (n) = an , ∀ n ∈ Z, is injective.
Therefore, since Z is infinite, G must be infinite.
Corollary 2.3.11. Let G be a finite cyclic group generated by a. Then |G| = ord(a).

Proof. Since G is finite, ord(a) must be finite by Corollary 2.3.10. Suppose that ord(a) = n ∈ N.
Then for any two integers r, s ∈ {k ∈ Z : 0 ≤ k ≤ n − 1}, ar = as implies ar−s = e, and hence
r = s, because |r − s| < n = ord(a). Then all the elements in the collection C := {ak : k ∈
Z with 0 ≤ k ≤ n − 1} are distinct, and that C has n elements. Clearly C ⊆ G. Given any
b ∈ G = ⟨ a ⟩, b = am , for some m ∈ Z. Then by division algorithm (Theorem 1.3.1) there exists
q, r ∈ Z with 0 ≤ r < n such that m = nq + r. Then b = am = anq+r = (an )q ar = ar ∈ C , since
an = e. Therefore, G ⊆ C , and hence G = C . Thus, |G| = ord(a).

Corollary 2.3.12. Let G be a finite group of order n. Then G is cyclic if and only if it contains an
element of order n.

Proof. If G is cyclic, then the result follows from Corollary 2.3.11. Conversely, if G contains an
element a of order n, then it follows from the proof of Corollary 2.3.11 that the cyclic subgroup
⟨ a ⟩ of G has n elements, and hence ⟨ a ⟩ = G.
Corollary 2.3.13. Any non-trivial subgroup of an infinite cyclic group is infinite and cyclic.

Proof. Let G be an infinite cyclic group generated by a ∈ G. Let H be a non-trivial subgroup of

G. Since H is cyclic by Proposition 2.3.7, we have H = ⟨ b ⟩, where b = ar for some r ∈ Z \ {0}.
Since G is an infinite cyclic group, by above Lemma 2.3.9, we have bm = amr ̸= anr = bn for
m ̸= n in Z. Therefore, H = ⟨ b ⟩ = {bk : k ∈ Z} is infinite.
Proposition 2.3.14. Let G be a finite cyclic group of order n. Then for each positive integer d such that
d | n, there is a unique subgroup H of G of order d.

Proof. Let G = ⟨ a ⟩ be a finite cyclic group of order n. Then ord(a) = n by Corollary 2.3.11.
Since d | n, there exists q ∈ Z such that
n = dq.
2.4. Product of subgroups 27

Let H := ⟨ aq ⟩ be the cyclic subgroup of G generated by aq . Since G is finite, so is H. Since

ord(a) = n, we see that d is the least positive integer such that (aq )d = aqd = an = e. Therefore,
ord(aq ) = d, and hence |H| = d by Corollary 2.3.11.
We now show uniqueness of H in G. If d = 1, then the trivial subgroup {e} ⊆ G is the only
subgroup of G of order d = 1. Suppose that d > 1. Let H and K be two subgroups of G of order
d, where d | n. Then by Proposition 2.3.7 we have H = ⟨ an ⟩ and K = ⟨ am ⟩, for some m, n ∈ N.
Since subgroup of a finite group is finite, by Corollary 2.3.10 we have ord(an ) = d = ord(am ).
By division algorithm (Theorem 1.3.1) there exists unique integers k, r with 0 ≤ r < q such that
m = kq + r. Then dm = kdq + dr = kn + dr gives

e = (am )d = adm = (an )k adr = adr .

Since 0 ≤ r < q, we have 0 ≤ dr < dq = n. If r ̸= 0, this contradicts the fact that ord(a) = n.
Therefore, we must have r = 0, and hence am = akq+r = (ak )q ∈ ⟨ ak ⟩ = H. Therefore, K ⊆ H.
Since |H| = |K| = d, we have H = K.
Proposition 2.3.15. An infinite cyclic group has exactly two generators.

Proof. Let G = ⟨ a ⟩ = {an : n ∈ Z} be an infinite cyclic group. Let b ∈ G be any generator of G.

Then b = an , for some n ∈ Z. Similarly, since a ∈ G = ⟨ b ⟩, we have a = bm , for some m ∈ Z.
Then we have a = bm = (an )m = amn . Then by Lemma 2.3.9 we have mn = 1. Since both m
and n are integers, we must have m, n ∈ {1, −1}. Therefore, b ∈ {a, a−1 }.
Exercise 2.3.16. Let G = ⟨ a ⟩ be a finite cyclic group of order n. Given any k ∈ N with 1 ≤
k ≤ n − 1, show that ⟨ ak ⟩ = G if and only if gcd(n, k) = 1. Conclude that G has exactly ϕ(n)
number of generators, where ϕ(n) is the number of elements in the set {k ∈ N : gcd(n, k) = 1}.
(Hint: Use the idea of the proof of Proposition 2.3.4.)
Remark 2.3.17. The map ϕ : N → N given by sending n ∈ N to the cardinality of the set

{k ∈ N : 1 ≤ k ≤ n and gcd(n, k) = 1},

is called the Euler phi function.

Exercise 2.3.18. Give an example of a non-abelian group G such that all of its proper subgroups
are cyclic.
Exercise 2.3.19. Show that a non-commutative group always has a non-trivial proper sub-
group.
Exercise 2.3.20. Show that a group having at most two non-trivial subgroups is cyclic.
Exercise 2.3.21. Let G be a finite group having exactly one non-trivial subgroup. Show that
|G| = p2 , for some prime number p.
Exercise 2.3.22. Give examples of infinite abelian groups having

(i) exactly one element of finite order;

(ii) all of its non-trivial elements have order 2.

2.4 Product of subgroups

Definition 2.4.1. Let G be a group. For any two non-empty subsets H and K of G, we define
their product HK := {hk : h ∈ H, k ∈ K}.
Exercise 2.4.2. Show by example that HK need not be a group in general even if both H and
K are subgroups of a group.
28 Chapter 2. Group Theory

Theorem 2.4.3. Let H and K be two subgroups of G. Then HK is a group if and only if HK = KH.

Proof. Note that, for any h ∈ H and k ∈ K we have h = h · e ∈ HK and k = e · k ∈ HK.

Therefore, H ⊆ HK and K ⊆ HK.
Suppose that HK is a group. Then kh ∈ HK, for all h ∈ H ⊆ HK and k ∈ K ⊆ HK, and
hence KH ⊆ HK. Let h ∈ H and k ∈ K. Since HK is a group, hk ∈ HK implies (hk)−1 ∈ HK,

−1
and so (hk)−1 = h1 k1 , for some h1 ∈ H and k1 ∈ K. Then hk = (hk)−1 = k1−1 h−1
1 ∈ KH.
Therefore, HK ⊆ KH, and hence HK = KH.
Conversely suppose that HK = KH. Let h1 k1 , h2 k2 ∈ HK with h1 , h2 ∈ H and k1 , k2 ∈ K.
Since k2−1 h−1 −1 −1
2 ∈ KH = HK, there exists h3 ∈ H and k3 ∈ K such that k2 h2 = h3 k3 . Again
k1 h3 ∈ KH = HK implies there exists h4 ∈ H and k4 ∈ K such that k1 h3 = h4 k4 . Now

(h1 k1 )(h2 k2 )−1 = h1 k1 k2−1 h−1

2
= h1 k1 h3 k3
= h1 h4 k4 k3 ∈ HK.

Therefore, HK is a subgroup of G.

Corollary 2.4.4. If H and K are subgroups of a commutative group, then HK is a group.

Notation: For a finite set S, we denote by |S| the number of elements of S.

Remark 2.4.5. The phrase “number of elements of S" is ambiguous when S is not a finite
set. For example, both Z and R are infinite sets, but there are some considerable differences
between “the number of elements” of them; Z is a countable set, while R is an uncountable
set. So the “number of elements" (whatever that means) for Z and R should not be the same.
For this reason, we need an appropriate concept of “number of elements" for an infinite set
S, known as the cardinality of S, also denoted by |S|. When S is a finite set, the cardinality
of S is determined by the number of elements of S. The cardinality of Z is denoted by ℵ0
(aleph-naught) and the cardinality of R is 2ℵ0 , which is also denoted by ℵ1 or c.
Definition 2.4.6. The order of a group G is the cardinality |G| of its underlying set G. For a
finite group, its order is precisely the number of elements in it.

For example, the order of S3 is 6, while the order of Z is ℵ0 .

Lemma 2.4.7. If H and K are finite subgroups of a group G, then

|H| · |K|
|HK| = .
|H ∩ K|

Proof. For each positive integer n, let Jn := {k ∈ N : k ≤ n}. Let H = {hi : i ∈ Jn } and
K = {kj : j ∈ Jm }. Then HK = {hi kj : i ∈ Jn , j ∈ Jm }. To find the number of elements
of HK, for each pair (i, j) ∈ Jn × Jm , we need to count the number of times hi kj repeats in
the collection C := {hi kj : (i, j) ∈ Jn × Jm }. Fix (i, j) ∈ Jn × Jm . If hi kj = hp kq , for some
−1
(p, q) ∈ Jn × Jm , then t := h−1p hi = kq kj ∈ H ∩ K. So any element hp kq ∈ C , which coincides
−1
with hi kj is of the form (hi t )(tkj ), for some t ∈ H ∩ K. Conversely, for any t ∈ H ∩ K, we
have (hi t−1 )(tkj ) = hi (t−1 t)kj = hi ekj = hi kj . Therefore, the element hi kj appears exactly
|H ∩ K|-times in the collection C , and hence we have

|H| · |K|
|HK| = .
|H ∩ K|

This completes the proof.

2.5. Permutation Groups 29

Proposition 2.4.8. Let H and K be subgroups of G. Then HK is a subgroup of G if and only if

HK = ⟨ H ∪ K ⟩.

Proof. Suppose that HK is a subgroup of G. Since H ⊆ HK and K ⊆ HK, we have H ∪ K ⊆

HK, and hence ⟨ H ∪ K ⟩ ⊆ HK. Since ⟨ H ∪ K ⟩ is a group containing H ∪ K, for any h ∈ H
and k ∈ K we have hk ∈ ⟨ H ∪ K ⟩. Therefore, HK ⊆ ⟨ H ∪ K ⟩, and hence HK = ⟨ H ∪ K ⟩.
Converse is obvious since ⟨ H ∪ K ⟩ is a group and HK = ⟨ H ∪ K ⟩ by assumption.

2.5 Permutation Groups

Let X be a non-empty set. A permutation on X is a bijective map σ : X → X. We denote by
SX the set of all permutations on X. For notational simplicity, when |X| = n, fixing a bijection
of X with the subset Jn := {1, 2, 3, . . . , n} ⊂ N we may identify SX with Sn . An element σ ∈ Sn
can be described by a two-column notation as follow.

 1 7→ σ(1)

   2 7→ σ(2)

1 2 3 ··· n
(2.5.1) σ= or, σ = .. .
σ(1) σ(2) σ(3) · · · σ(n) 
 .

n 7→ σ(n)


Since elements of Sn are bijective maps of Jn onto itself, composition of two elements of Sn is
again an element of Sn . Thus we have a binary operation

◦ : Sn × Sn −→ Sn , (σ, τ ) 7−→ τ ◦ σ.

For example, consider the elements σ, τ ∈ S4 defined by

   
1 2 3 4 1 2 3 4
σ= , τ= .
2 4 1 3 1 3 2 4

Then their composition τ ◦ σ is the permutation

 
1 2 3 4
τ ◦σ =
3 4 1 2

Clearly composition of functions Jn → Jn is associative, and for any σ ∈ Sn its pre-composition

and post-composition with the identity map of In is σ itself. Also inverse of a bijective map is
again bijective. Thus for all integer n ≥ 1, (Sn , ◦) is a group, called the Symmetric group (or, the
permutation group) on Jn .
Remark 2.5.2. For each integer n ≥ 0, the symmetric group Sn+1 can be understood as the
group of symmetries of a regular n-simplex inside Rn+1 . The standard n-simplex
n
X
∆n := {(t0 , . . . , tn ) ∈ Rn+1 : tj = 1, tj ≥ 1, ∀ j = 0, 1, . . . , n.} ⊂ Rn+1
j=1

is an example of a regular n-simplex. This has vertices the unit vectors {e0 , e1 , . . . , en } in Rn+1 ,
where
e0 = (1, 0, 0, . . . , 0, 0),
e1 = (0, 1, 0, . . . , 0, 0),
.. ..
. .
en = (0, 0, 0, . . . , 0, 1).
For example,
30 Chapter 2. Group Theory

• ∆0 is a point,
• ∆1 is the straight line segment [−1, 1] ⊂ R ⊂ R2 ,
• ∆2 is an equilateral triangle in the plane R2 ,
• ∆3 is a regular tetrahedron in R3 , and so on.
Exercise 2.5.3. Show that S1 is a trivial group, and S2 is an abelian group with two elements.
Lemma 2.5.4. For all integer n ≥ 3, the group Sn is non-commutative.

Proof. Let σ, τ ∈ Sn be defined by

 
 2, if k = 1  3, if k = 1
σ(k) = 1, if k = 2 , and τ (k) = 1, if k = 3 .
k, if k ∈ In \ {1, 2} k, if k ∈ In \ {1, 3}
 

Since τ ◦ σ(1) = 2 and σ ◦ τ (1) = 3, we have σ ◦ τ ̸= τ ◦ σ. Therefore, Sn is non-commutative.

Let σ ∈ Sn be given. Consider its two-column notation as in (2.5.1).

(R1) If σ(k) = k, for some k ∈ Jn , we may drop the corresponding column from its two-column
notation, and rearrange its columns, if required, to get an expression of the form
 
k1 k2 ··· kr−1 kr
σ= ,
σ(k1 ) σ(k2 ) · · · σ(kr−1 ) σ(kr )

where k1 , . . . , kr are all distinct.

By re-indexing, if required, we can find a partition of {k1 , . . . , kr } into disjoint subsets, say
m
[
{k1 , . . . , kr } = {ki,1 , . . . , ki,ri }
i=1

with m ≥ 1, 2 ≤ ri ≤ r, for all i ∈ {1, . . . , m}, and r1 + · · · + rm = r, such that for all
i ∈ {1, . . . , m} we have

 ki,j+1 , if j ∈ {1, . . . , ri − 1},
(2.5.5) σ(ki,j ) = ki,1 , if j = ri , and
kij , if kij ∈ Jn \ {k1 , . . . , kr }.


Then σ can be expressed as

 
k1,1 · · · k1,r1 −1 k1,r1 ······ km,1 ··· km,rm km,rm −1
(2.5.6) σ= .
k1,2 · · · k1,r1 k1,1 ······ km,2 ··· km,rm km,1

When m = 1 in the above notation, σ can be expressed as

 
k1 k2 · · · kr−1 kr
(2.5.7) σ= .
k2 k3 · · · kr k1

Such a permutation is called a cycle.

Definition 2.5.8 (Cycle). An element σ ∈ Sn is called a r-cycle or a cycle of length r if there
exists distinct r elements, say k1 , . . . , kr ∈ Jn := {1, . . . , n} such that σ(k) = k, for all k ∈
Jn \ {k1 , . . . , kr } and 
ki+1 if i ∈ {1, . . . , r − 1},
σ(ki ) =
k1 if i = r.
2.5. Permutation Groups 31

In this case, σ is expressed as σ = (k1 k2 · · · kr ). A 2-cycle is called a transposition.

Remark 2.5.9. Note that according to our definition 2.5.8, a cycle in Sn always have length at
least 2. So we don’t talk about 1-cycle as used in some of the standard text books.

With the notation above, the permutation σ in (2.5.6) can be written as a product of cycles
   
k1,1 · · · k1,r1 −1 k1,r1 k · · · km,rm −1 km,rm
σ= ◦ · · · ◦ m,1
k1,2 · · · k1,r1 k1,1 km,2 · · · km,rm km,1
= (k1,1 · · · k1,r1 −1 k1,r1 ) ◦ · · · ◦ (km,1 · · · km,rm −1 km,rm )

Remark 2.5.10. Transpositions are of particular interests. We shall see later that any σ ∈ Sn can
be written as product of either even number of transpositions or odd number of transpositions,
and accordingly we call σ ∈ Sn an even permutation or an odd permutation.
Example 2.5.11. Using cycle notation, the group S3 can be written as

S3 = {e, (1 2), (1 3), (2 3), (1 2 3), (1 3 2)},

where (1 2), (1 3) and (2 3) are transpositions. However, we can write 3-cycles as product of
2-cycles as (1 2 3) = (2 3) ◦ (1 3) and (1 3 2) = (2 3) ◦ (1 2). Also, the identity element e
can be written as e = (1 2) ◦ (1 2) or e = (1 3) ◦ (1 3) etc. So the decomposition of σ ∈ Sn as a
product of transpositions is not unique.
Proposition 2.5.12. Let σ = (k1 k2 · · · kr ) ∈ Sn be a r-cycle. Then for any τ ∈ Sn we have

τ στ −1 = (τ (k1 ) τ (k2 ) · · · τ (kr )).

Proof. Note that we have

(τ στ −1 )(τ (ki )) = τ (σ(ki )) = τ (ki+1 ), ∀ i ∈ {1, . . . , r − 1},

and (τ στ −1 )(τ (kr )) = τ (σ(kr )) = τ (k1 ) .

It remains to show that (τ στ −1 )(k) = k, ∀ k ∈ Jn \ {τ (k1 ), . . . , τ (kr )}. For this, note that
τ −1 (k) ∈ Jn \ {k1 , . . . , kr }, and so σ(τ −1 (k)) = τ −1 (k). Therefore, we have (τ στ −1 )(k) =
τ (σ(τ −1 (k))) = τ (τ −1 (k)) = k. This completes the proof.
Corollary 2.5.13. Let σ ∈ Sn is a product of pairwise disjoint cycles σ1 , . . . , σr in Sn . Suppose
that σi = (ki1 · · · kiℓi ) ∈ Sn , for all i ∈ {1, . . . , r}. Then for any τ ∈ Sn we have τ στ −1 =
(τ (k11 ) · · · τ (k1ℓ1 )) ◦ · · · ◦ (τ (kr1 ) · · · τ (krℓr )). In particular, both σ and τ στ −1 have the same
cycle type.

Proof. Since τ στ −1 = (τ σ1 τ −1 ) ◦ · · · ◦ (τ σr τ −1 ), the result follows from Proposition 2.5.12.

Proposition 2.5.14. Let σ ∈ Sn be a cycle. Then σ is a r cycle if and only if ord(σ) = r.

Proof. Let σ = (k1 k2 · · · kr ), for some distinct elements k1 , . . . , kr ∈ Jn . Then for any
k ∈ Jn \ {k1 , . . . , kr } we have σ(k) = k. It follows from the definition of the cyclic expression
of σ given in (2.5.5) that σ i (k1 ) = ki+1 , for all i ∈ {1, . . . , k − 1} and σ r (k1 ) = k1 . In general, for
any ki with 1 ≤ i ≤ r we have σ r−i (ki ) = kr and so σ r−i+1 (ki ) = k1 . Therefore, σ r−i+ℓ (ki ) = kℓ
for all ℓ ∈ {1, . . . , r − 1}, and hence σ r (ki ) = ki , for all i ∈ {1, . . . , r}. Combining all these, we
have σ r (k) = k, for all k ∈ Jn . In other words, σ r = e, where e is the identity element in Sn .
Since σ s (k1 ) = ks+1 , for all s ∈ {1, . . . , r − 1} (see (2.5.5)), we conclude that r is the smallest
positive integer such that σ r = e in Sn . Therefore, ord(σ) = r. Conversely, suppose that σ is a t
cycle with ord(σ) = r. But then as shown above ord(σ) = t, and hence t = r.
n!
Exercise 2.5.15. Show that the number of distinct r cycles in Sn is .
r(n − r)!
32 Chapter 2. Group Theory

Solution: Note that, we can choose a r cycle from Sn in

 
n n n!
Cr = =
r r!(n − r)!

ways. Fix a r-cycle σ = (k1 k2 · · · kr ) ∈ Sn . Note that, the cycles

(k1 k2 · · · kr ) and (k2 k3 · · · kr k1 )

represents the same element σ ∈ Sn . Note that, given any two permutations (bijective maps)

ϕ, ψ : {2, 3, . . . , r} → {2, 3, . . . , r},

two r cycles (note that k1 is fixed!)

(k1 kϕ(2) · · · kϕ(r) ) and (k1 kψ(2) · · · kψ(r) )

represents the same element of Sn if and only if ϕ = ψ. Since there are (r − 1)! number of
distinct bijective maps {2, 3, . . . , r} → {2, 3, . . . , r} (verify!), fixing k1 in one choice of r cycle
(k1 k2 · · · kr ) in Sn , considering all permutations of the remaining (r − 1) entries k2 , . . . , kr ,
we get (r − 1)! number of distinct r cycles in Sn . Therefore, the total number of distinct r cycles
in Sn is precisely
n! n!
(r − 1)! · = .
r!(n − r)! r(n − r)!
This completes the proof.
Definition 2.5.16. Two cycles σ = (i1 i2 · · · ir ) and τ = (j1 j2 · · · js ) in Sn are said to be
disjoint if {i1 , i2 , . . . , ir } ∩ {j1 , j2 , · · · , js } = ∅.
Proposition 2.5.17. If σ and τ are disjoint cycles in Sn , show that σ ◦ τ = τ ◦ σ.

Proof. Let σ = (i1 i2 · · · ir ) and τ = (j1 j2 · · · js ) be two disjoint cycles in Sn . Let k ∈ Jn be

arbitrary. If k ∈
/ {i1 , . . . , ir } ∪ {j1 , . . . , js }, then σ(k) = k = τ (k) and hence (στ )(k) = (τ σ)(k) in
this case. Suppose that k ∈ {i1 , . . . , ir }. Then σ(k) ∈ {i1 , . . . , ir } and k ∈ / {j1 , . . . , js } together
gives τ σ(k) = σ(k) = στ (k). Interchanging the roles of σ and τ we see that τ σ(k) = σ(k) =
στ (k) holds for the case k ∈ {j1 , . . . , js }. Therefore, στ = τ σ.
Lemma 2.5.18. For n ≥ 2, any non-identity element of Sn can be uniquely written as a product of
disjoint cycles of length at least 2. This expression is unique up to ordering of factors.

Proof. For n = 2, S2 has only one non-identity element, which is a 2-cycle (1 2). Assume that
n ≥ 3 and the result is true for any non-identity element of Sr for 2 ≤ r < n. Let σ ∈ Sn be
a non-identity element. Since {σ i (1) : i ∈ N} ⊆ Jn and Jn is a finite set, there exists distinct
integers i, j ∈ N such that σ i (1) = σ j (1). Without loss of generality we may assume that
i − j ≥ 1. Then σ i−j (1) = 1. Then

{i ∈ N : σ i (1) = 1}

is a non-empty subset of N, and hence it has a least element, say r. Then all the elements in

A := {1, σ(1), σ 2 (1), . . . , σ r−1 (1)}

are all distinct, and defines an r-cycle

τ := (1 σ(1) σ 2 (1) · · · σ r−1 (1))

in Sn . Let B := Jn \ A. In cases σ B is the identity map of B onto itself or B = ∅, we have

τ = σ and so σ is a cycle in Sn . Assume that B ̸= ∅ and π := σ B is not the identity map.
2.5. Permutation Groups 33

Then π is a non-identity element of Sk , where 2 ≤ k := |B| < n. Then by induction hypothesis

π = π1 · · · πℓ is a finite product of disjoint cycles π1 , . . . , πℓ of lengths at least 2 in Sk . Then for
each i ∈ {1, . . . , ℓ} we define σi ∈ Sn by setting

πi (a), if a ∈ B,
σi (a) =
a, if a ∈ Jn \ B.

Then σ1 , . . . , σℓ , τ are pairwise disjoint cycles in Sn and that σ = σ1 · · · σℓ τ .

For the uniqueness part, let σ = σ1 · · · σr = τ1 · · · τs be two decomposition of σ into product
of disjoint cycles of lengths ≥ 2 in Sn . We need to show that r = s, and there is a permutation
δ ∈ Sr such that σi = τδ(i) , for all i ∈ {1, . . . , r}. Suppose that σi = (k1 k2 · · · kt ) with t ≥ 2.
Then σ(k1 ) ̸= k1 . Since τ1 , . . . , τr are pairwise disjoint cycles of lengths ≥ 2 in Sn , there is a
unique element, say δ(i) ∈ {1, . . . , r} such that τδ(i) (k1 ) ̸= k1 . By reordering, if required, we
may write τδ(i) = (k1 v2 · · · vu ). Then we have

k2 = σi (k1 ) = σ(k1 ) = τδ(i) (k1 ) = v2 ,

k3 = σi (k2 ) = σ(k2 ) = σ(v2 ) = τδ(i) (v2 ) = v3 ,
.. .. .. .. .. ..
. . . . . .
kt = σi (kr−1 ) = σ(kr−1 ) = σ(vr−1 ) = τδ(i) (vr−1 ) = vt .

If t < u, then k1 = σi (kt ) = σ(kt ) = σ(vt ) = vt+1 , which is a contradiction. Therefore, t = u and
hence σi = τδ(i) . Hence the result follows by induction on r.
Definition 2.5.19 (Cycle type). Given σ ∈ Sn , by Lemma 2.5.18 there exists a unique finite set
of pairwise disjoint cycles {σ1 , . . . , σr } in Sn such that σ = σ1 ◦ · · · ◦ σr . Since disjoint cycles
commutes by Proposition 2.5.17, by reindexing σj ’s, if required, we may assume that n1 ≥ . . . ≥
nr , where nj = length(σj ), for all j ∈ {1, . . . , r}. Since σ1 , . . . , σr are pairwise disjoint cycles
Pr
in Sn , we have ℓ + nj = n, for some non-negative integer ℓ. If ℓ = 0, then the sequence
j=1
(n1 , . . . , nr ) is called the cycle type of σ, and if ℓ > 0, then the sequence (n1 , . . . , nr , f1 , . . . , fℓ ),
where f1 = . . . = fℓ = 1, is called the cycle type of σ.
Example 2.5.20. (i) The cycle type of σ := (1 2) ◦ (3 6) ◦ (4 5 7) ∈ S7 is (3, 2, 2).
(ii) The cycle type of τ := (1 4 3) ◦ (2 5) ∈ S7 is (3, 2, 1, 1).

(iii) The cycle type of δ := (1 3 5) ◦ (2 4 7) ∈ S6 is (3, 3, 1).

Definition 2.5.21. Two permutations σ and τ in Sn are said to be conjugate in Sn if there exists
δ ∈ Sn such that τ = δ ◦ σ ◦ δ −1 .
Theorem 2.5.22. Two elements σ, τ ∈ Sn are conjugate if and only if they have the same cycle type.

Proof. Conjugate permutations in Sn have the same cycle type by Corollary 2.5.13. Conversely
suppose that σ, τ ∈ Sn have the same cycle type, say (n1 , . . . , nr , f1 , . . . , fℓ ), where n1 ≥ · · · ≥
Pr
nr ≥ 2 and f1 = · · · = fℓ = 1, ℓ ≥ 0 and that nj + ℓ = n. Let σ = σ1 ◦ · · · ◦ σr
j=1
and τ = τ1 ◦ · · · ◦ τr , where σi , τj are cycles in Sn of lengths ni and nj , respectively. Sup-
pose that σi = (ai1 · · · aini ) and τj = (bj1 · · · bjnj ). If ℓ > 0, then we write the sub-
set In \ {aij : 1 ≤ i ≤ r, 1 ≤ j ≤ ni } as {a1 , . . . , aℓ }. Then In is a disjoint union of the
subsets {a11 , . . . , a1n1 }, . . . , {ar1 , . . . , arnr }, {a1 , . . . , aℓ }. Similarly if we write the subset In \
{bij : 1 ≤ i ≤ r, 1 ≤ j ≤ ni } as {b1 , . . . , bℓ }, then In is a disjoint union of the subsets
{b11 , . . . , b1n1 }, . . . , {br1 , . . . , brnr }, {b1 , . . . , bℓ }. Then we define a map δ : In → In by sending
aij to bij , for all (i, j) ∈ {1, . . . , r} × {1, . . . , ni }, and by sending ak to bk , for all k ∈ {1, . . . , ℓ},
if ℓ > 0. Clearly δ is a bijective map, and hence is an element of Sn . Then Proposition 2.5.12
34 Chapter 2. Group Theory

ensures that δσi δ −1 = τi , for all i ∈ {1, . . . , r}. Then we have

δσδ −1 = δ(σ1 · · · σr )δ −1
= (δσ1 δ −1 ) · · · (δσr δ −1 )
= τ1 · · · τr
= τ.

This completes the proof.

Corollary 2.5.23. For n ≥ 2, every element of Sn can be written as a finite product of transpositions.

Proof. In view of above Lemma 2.5.18 it suffices to show that every cycle of Sn is a product
of transpositions. Clearly the identity element e ∈ Sn can be written as e = (1 2)(1 2). If
σ = (k1 k2 · · · kr ) is an r-cycle, r ≥ 2, in Sn , then we can rewrite it as

σ = (k1 k2 · · · kr ) = (k1 kr )(k1 kr−1 ) · · · (k1 k2 ).

Hence the result follows.

Note that decompositions of σ ∈ Sn into a finite product of transpositions is not unique.

For example, when n ≥ 3 we have e = (1 2)(1 2) = (1 3)(1 3). However, we shall see shortly
that the number of transpositions appearing in such a product expression for σ ∈ Sn is either
odd or even, but cannot be both in two such decompositions.
Fix an integer n ≥ 2, and consider the action of a permutation σ ∈ Sn on the formal
Lemma 2.5.24. Q
product χ := (xi − xj ) given by
1≤i<j≤n

Y
σ(χ) := (xσ(i) − xσ(j) ).
1≤i<j≤n

If σ ∈ Sn is a 2-cycle (transposition), then σ(χ) = −χ.

Proof. Since σ ∈ Sn is a 2-cycle, there exists a unique subset {p, q} ⊆ Jn with p < q such that
σ = (p q). Then σ(k) = k, ∀ k ∈ Jn \{p, q}. Consider the factor (xi −xj ) of χ with 1 ≤ i < j ≤ n.
We have the following situations:

(a) If {i, j} = {p, q}, then σ(xi − xj ) = xσ(i) − xσ(j) = −(xi − xj ).

(b) If {i, j} ∩ {p, q} = ∅, then σ(xi − xj ) = xσ(i) − xσ(j) = (xi − xj ).

(c) If {i, j} ∩ {p, q} is singleton set, then we have the following subcases.
I. If t < p < q, then σ((xt −xp )(xt −xq )) = (xσ(t) −xσ(p) )(xσ(t) −xσ(q) ) = (xt −xq )(xt −xp ).
II. If p < t < q, then σ((xp −xt )(xt −xq )) = (xσ(p) −xσ(t) )(xσ(t) −xσ(q) ) = (xq −xt )(xp −xt ).
III. If p < q < t, then σ((xp −xt )(xq −xt )) = (xσ(p) −xσ(t) )(xσ(q) −xσ(t) ) = (xq −xt )(xp −xt ).
Therefore, in the above three subcases the product (xt − xp )(xt − xq ) remains fixed under
the action of σ.

From these it immediately follows that σ(χ) = −χ, for all 2-cycle σ ∈ Sn .

Corollary 2.5.25. Fix an integer n ≥ 2, and let σ ∈ Sn . If σ = σ1 · · · σr = τ1 · · · τs , where σi , τj are

all transpositions in Sn , then both r and s are either even or odd.
2.5. Permutation Groups 35

Q
Proof. Consider the formal product χ := (xi − xj ). Then σ(χ) = (σ1 ◦ · · · ◦ σr )(χ) =
1≤i<j≤n
(−1)r χ and σ(χ) = (τ1 ◦ · · · ◦ τs )(χ) = (−1) χ together implies that (−1)r = (−1)s , and hence
s

both r and s are either even or odd.

Definition 2.5.26. A permutation σ ∈ Sn is called even (respectively, odd) if σ can be written as
a product of even (respectively, odd) number of transpositions in Sn .

Note that given a permutation σ ∈ Sn , if σ = σ1 ◦ · · · ◦ σr , where σ1 , . . . , σr are 2-cycles in

Sn , then by Corollary 2.5.25 we see that σ is even if and only if (−1)r = 1. Thus we have a
well-defined map sgn : Sn → {1, −1} given by sending σ ∈ Sn to (−1)r , where r is a number of
2-cycles appearing in the decomposition of σ into a product of 2-cycles in Sn . In other words,

−1, if σ is odd,
(2.5.27) sgn(σ) =
1, if σ is even,

The number sgn(σ) is called the signature of the permutation σ ∈ Sn .

Proposition 2.5.28. An r-cycle σ ∈ Sn is even if and only if r is odd.

Proof. Let σ = (k1 k2 · · · kr ) be an r-cycle in Sn . Then we can write it as a product σ =

(k1 k2 · · · kr ) = (k1 kr )(k1 kr−1 ) · · · (k1 k2 ) of r − 1 number of transpositions in Sn . Hence
the result follows.
Exercise 2.5.29. Express the following permutations as product of disjoint cycles, and then
express them as a product of transpositions. Determine if they are even or odd permutations.
 
1 2 3 4 5 6 7 8
(i) σ = ∈ S8 .
2 3 8 5 6 4 7 1
Answer: Note that,

σ = (1 2 3 8) ◦ (4 5 6)
= (1 8) ◦ (1 3) ◦ (1 2) ◦ (4 6) ◦ (4 5).

Since σ is a product of 5 transpositions in S8 , we conclude that σ is odd.

 
1 2 3 4 5 6
(ii) τ = ∈ S6 .
5 1 4 2 3 6
 
1 2 3 4 5 6 7
(iii) π = ∈ S7 .
2 4 6 1 3 7 5
Proposition 2.5.30. Let An = {σ ∈ Sn : σ is even} be the set of all even permutations in Sn . Then
An is a subgroup of Sn , known as the alternating group on Jn .

Proof. Since e = (1 2) ◦ (1 2), we see that e ∈ An . Thus An is a non-empty subset of Sn . Let

σ, τ ∈ An be arbitrary. Suppose that τ = τ1 ◦ · · · ◦ τ2r , where τ1 , . . . , τ2r are transpositions in Sn .
Since transpositions are elements of order 2 (see Proposition 2.5.14), they are self inverse in Sn .
Now it follows from Exercise 2.1.8 (ii) that

τ −1 = τ2r ◦ · · · ◦ τ1 .

Therefore, τ −1 is also an even permutation. Since σ and τ −1 are even, their product σ ◦ τ −1 ∈
An . Therefore, An is a subgroup of Sn by Lemma 2.2.8.
Remark 2.5.31. Assume that n ≥ 3. Note that, any transposition (i j) ∈ Sn , with i ̸= 1 and
j ̸= 1, can be written as
(i j) = (1 i) ◦ (1 j) ◦ (1 i).
36 Chapter 2. Group Theory

Again (1 i) ◦ (1 j) = (1 j i). Since each element of An are product of even number of

transpositions, using above two observations, one can write each element of An as product of
3 cycles in Sn .
Exercise 2.5.32. For all n ≥ 3, show that An is generated by 3-cycles.

Solution: Note that any 3-cycle is an even permutation by Proposition 2.5.28, and hence is in
An . Therefore, the subgroup of Sn generated by all 3-cycles is a subgroup of An . For the
converse part, we show that any even permutation can be written as product of 3-cycles. Note
that any element of An is a product of even number of 2-cycles in Sn . Let σ = (i j) and
τ = (k ℓ) be two 2-cycles in Sn . If σ and τ are not disjoint, then we may assume that j = k.
Then σ ◦ τ = (i j)(j ℓ) = (i j ℓ) is a 3-cycle. If σ and τ are disjoint, then

σ ◦ τ = (i j)(k ℓ)
= (i j)(j k)(j k)(k ℓ)
= (i j k)(j k ℓ),

where the last equality is due to the first case. Hence the result follows.
Exercise 2.5.33. Show that |An | = n!/2.

Solution: Let {σ1 , . . . , σr } and {τ1 , . . . , τs } be the set of all even permutations and the set of all
odd permutations in Sn , respectively. Since r + s = n!, it suffices to show that r = s. Fix a
transposition π ∈ Sn . Then πσ1 , . . . , πσr are all distinct (verify) odd permutations in Sn , and
hence r ≤ s. Similarly s ≤ r, and hence r = s, as required.
Exercise 2.5.34. Determine the groups A3 and A4 .
Exercise 2.5.35. Given σ, τ ∈ Sn , show that [σ, τ ] := σ ◦ τ ◦ σ −1 ◦ τ −1 ∈ An . The element [σ, τ ]
is called the commutator of σ and τ in Sn .
Example 2.5.36 (Dihedral group Dn ). Consider a regular n-gon in the plane R2 whose vertices
are labelled as 1, 2, 3, . . . , n in clockwise order. Let Dn be the set of all symmetries of this regular
n-gon given by the following operations and their finite compositions:

a := The rotations about its centre through the angles 2π/n, and
b := The reflections along the vertical straight line passing through the centre of the regular
n-gon.

Note that ord(a) = n, ord(b) = 2 and that an−1 b = ba. Therefore, the group generated by all
such symmetries of the regular n-gon can be expressed in terms of generators and relations as

Dn := ⟨ a, b | ord(a) = n, ord(b) = 2, and an−1 b = ba ⟩.

This group is called the dihedral group of degree n. Note that Dn is a non-commutative finite
group of order 2n and its elements can be expressed as

Dn = {e, a, a2 , a3 , . . . , an−1 , b, ba, ba2 , ba3 , . . . , ban−1 }.

Note that each element of Dn is given by a bijection of the set Jn := {1, 2, . . . , n} onto itself,
and hence is a permutation on Jn . However, not all permutations of the set Jn corresponds to
a symmetry of a regular n-gon as described above (see Exercise 2.5.37 below). We can define a
binary operation on Dn by composition of bijective maps. Then it is easy to check using Lemma
2.2.8 that Dn is a subgroup of Sn . The group Dn is called the Dihedral group of degree n. It is a
finite group of order 2n which is non-commutative for n ≥ 3.
Exercise 2.5.37. Show that D3 = S3 , and Dn is a proper subgroup of Sn , for all n ≥ 4.
2.6. Group homomorphism 37

Exercise 2.5.38. Let G be the subgroup of S4 generated by the cycles

a := (1 2 3 4) and b := (2 4)

in S4 . Show that G is a dihedral group of degree 4.

2.6 Group homomorphism

A group homomorphism is a map from a group G into another group H that respects the
binary operations on them. Here is a formal definition.
Definition 2.6.1. Let G and H be two groups. A group homomorphism from (G, ∗) into (H, ⋆) is
a map f : G → H satisfying f (a ∗ b) = f (a) ⋆ f (b), for all a, b ∈ G.
Example 2.6.2. (i) For any group G, the constant map ce : G → G, which sends all points
of G to the neutral element e ∈ G, is a group homomorphism, called the trivial group
homomorphism of G.

(ii) Let H be a subgroup of a group G. Then the set theoretic inclusion map H ,→ G is a group
homomorphism. In particular, for any group G, the identity map

IdG : G → G, a 7→ a

is a group homomorphism.

(iii) Fix an integer m, and define a function

φm : Z −→ Z, n 7−→ mn, ∀ n ∈ Z.

Then φm (n1 + n2 ) = m(n1 + n2 ) = mn1 + mn2 = φm (n1 ) + φm (n2 ), for all n1 , n2 ∈ Z.

Therefore, φm is a group homomorphism. Note that, φm is always injective, and it is
surjective only for m ∈ {1, −1}.
(iv) Let R∗ := R \ {0}, and consider the exponential map

f : R −→ R∗ , x 7−→ ex , ∀ x ∈ R.

Since f (a + b) = ea+b = ea · eb = f (a) · f (b), for all a, b ∈ R, the map f is a group

homomorphism from (R, +) into (R∗ , ·). Verify that f is injective.
(v) The map f : R → S 1 := {z ∈ C∗ : |z| = 1} defined by f (t) = e2πit , ∀ t ∈ R is a surjective
group homomorphism. Is it injective?
(vi) Let  
1 a
ϕ : R −→ SL2 (R), a 7−→ , ∀ a ∈ R.
0 1
Verify that ϕ is an injective group homomorphism from the additive group R into the
multiplicative group SL2 (R).

(vii) Fix an integer n ≥ 2, and consider the map

ψ : Z −→ Zn , a 7−→ [a], ∀ a ∈ Z.

Verify that ψ is a surjective group homomorphism.

38 Chapter 2. Group Theory

(viii) Fix a prime number p > 0, and let F : Zp → Zp be the map defined by F(a) = ap , for all
a ∈ Zp . Since any multiple of p is 0 in Zp , using binomial expansion we have
p  
X p
F(a + b) = (a + b)p = ap−j bj = ap + bp .
j=0
j

Therefore, F is a group homomorphism.

(ix) Fix an integer n ≥ 1, and let f : GLn (R) → R∗ be the map defined by

f (A) = det(A), ∀ A ∈ GLn (R).

Verify that f is a group homomorphism.

(x) Let m, n > 1 be integers such that n | m in Z. Verify that the map φ : Zm → Zn defined by
sending [a] ∈ Zm to [a] ∈ Zn is a well-defined map that is a group homomorphism.
(xi) Let G be a group. For each a ∈ G, the map φa : G → G defined by φa (b) = aba−1 , ∀ b ∈ G,
is a group homomorphism.
Exercise 2.6.3. For each integer n ≥ 1, let Jn := {k ∈ Z : 1 ≤ k ≤ n}. For each σ ∈ Sn , consider
the map σe : Jn+1 → Jn+1 defined by

σ(k), if 1 ≤ k ≤ n,
σ
e(k) =
n + 1, if k = n + 1.

Note that, σ
e is a bijective map, and hence is an element of Sn+1 . Show that the map

f : Sn → Sn+1 , σ 7→ σ
e,

is an injective group homomorphism. Thus, we can identify Sn as a subgroup of Sn+1 .

Lemma 2.6.4. Let n ≥ 2 be an integer. Then the map sgn : Sn → {1, −1} defined by sending σ ∈ Sn
to 
−1, if σ is odd,
sgn(σ) =
1, if σ is even,
is a group homomorphism, called the signature homomorphism for Sn .

Proof. Let σ, τ ∈ Sn be arbitrary. Let σ = σ1 ◦ · · · ◦ σr and τ = τ1 ◦ · · · ◦ τs , where σi , τj are

all 2-cycles in Sn . Then σ ◦ τ = σ1 ◦ · · · ◦ σr ◦ τ1 ◦ · · · ◦ τs , and hence sgn(σ ◦ τ ) = (−1)r+s =
(−1)r (−1)s = sgn(σ)sgn(τ ).
Proposition 2.6.5. Let f : G → H be a group homomorphism. Let eG ∈ G and eH ∈ H be the neutral
elements of G and H, respectively. Then we have the following.

(i) f (eG ) = eH .
−1
(ii) f (a−1 ) = (f (a)) , for all a ∈ G.
(iii) If ord(a) < ∞, then ord(f (a)) | ord(a).

Proof. (i) Since f (eG )f (eG ) = f (eG · eG ) = f (eG ) = f (eG ) · eH , applying cancellation law we
have f (eG ) = eH . The second statement follows immediately.
(ii) Since f is a group homomorphism, for any a ∈ G, we have

f (a)f (a−1 ) = f (a · a−1 ) = f (eG ) = eH

and f (a−1 )f (a) = f (a−1 · a) = f (eG ) = eH ,
−1
and hence f (a−1 ) = (f (a)) .
2.6. Group homomorphism 39

(iii) Let n := ord(a) < ∞. Since f (a)n = f (an ) = f (eG ) = eH , it follows from Exercise 2.2.28
(i) that ord(f (a)) | n.

Exercise 2.6.6. Let G and H be two groups. Show that there is a unique constant group homo-
morphism from G to H.
Proposition 2.6.7. Let f : G → H be a group homomorphism.

(i) For any subgroup G′ of G, its image f (G′ ) := {f (a) : a ∈ G′ } is a subgroup of H. Moreover, if
G′ is commutative, so is f (G′ ).
(ii) For any subgroup H ′ of H, its inverse image f −1 (H ′ ) := {a ∈ G : f (a) ∈ H ′ } is a subgroup of
G.

Proof. (i) Clearly, f (G′ ) ̸= ∅ as e ∈ G′ . For h1 , h2 ∈ f (G′ ), we have h1 = f (a1 ) and h2 = f (a2 ),
for some a1 , a2 ∈ G′ . Since a1 a−1
2 ∈ G′ , we have h1 h−1 2 = f (a1 )f (a2 )−1 = f (a1 a−1
2 ) ∈
′ ′
f (G ). If G is commutative, we have f (a)f (b) = f (ab) = f (ba) = f (b)f (a), for all a, b ∈
G′ . Hence the result follows.
(ii) Let eG ∈ G and eH ∈ H be the neutral elements of G and H, respectively. Since f (eG ) =
eH by Proposition 2.6.5 (i), we have eG ∈ f −1 (H ′ ). Since H ′ is a subgroup of H, for any
a, b ∈ f −1 (H ′ ) we have f (ab−1 ) = f (a)f (b)−1 ∈ H ′ , and hence ab−1 ∈ f −1 (H ′ ). Thus
f −1 (H ′ ) is a subgroup of G.

Proposition 2.6.8. Composition of group homomorphisms is a group homomorphism.

Proof. Let f : G1 → G2 and g : G2 → G3 be two group homomorphisms. Since (g ◦ f )(ab) =

g(f (ab)) = g(f (a)f (b)) = g(f (a))g(f (b)) = (g ◦ f )(a)(g ◦ f )(b), for all a, b ∈ G1 , the result
follows.
Definition 2.6.9. Let f : G → H be a group homomorphism. We say that

(i) f is trivial if f (a) = eH , for all a ∈ G.

(ii) f is a monomorphism if f is injective (c.f. Proposition 2.6.22),
(iii) f is an epimorphism if f is surjective (c.f. Proposition 2.6.23), and

(iv) f is an isomorphism if f is bijective. In that case, we say that G is isomorphic to H, and

express it as G ∼
= H.
Corollary 2.6.10. Being isomorphic groups is an equivalence relation.

Proof. Given any group G, the identity map IdG : G → G given by IdG (a) = a, for all a ∈ G, is
an isomorphism of groups. Therefore, being isomorphic is a reflexive relation. If f : G → H is
an isomorphism of groups, then its inverse map f −1 : H → G is also a group homomorphism,
and hence is an isomorphism because it is bijective. Therefore, being isomorphic groups is
a symmetric relation. If f : G → H and g : H → K be isomorphism of groups. Then the
composite map g ◦ f : G → K is a group homomorphism, which is an isomorphism of groups.
Therefore, being isomorphic groups is a transitive relation. Hence the result follows.
Proposition 2.6.11. Given a group G, the set Aut(G) consisting of all group isomorphisms from G onto
itself is a group with respect to the binary operation given by composition of maps; the group Aut(G) is
known as the automorphism group of G.
40 Chapter 2. Group Theory

Proof. Since composition of two bijective group homomorphisms is bijective and a group ho-
momorphism, we see that the map

G × G → G, (f, g) 7−→ f ◦ g,

is a binary operation on Aut(G). Clearly composition of maps is associative. The identity

map IdG : G → G plays the role of a neutral element in a group. Given f ∈ Aut(G), its
inverse f −1 : G → G is again a group homomorphism. Indeed, given a, b ∈ G there exists
unique x, y ∈ G such that f (x) = a and f (y) = b. Then we have f −1 (ab) = f −1 (f (x)f (y)) =
f −1 (f (xy)) = xy = f −1 (a)f −1 (y), and hence f −1 ∈ Aut(G). This proves that Aut(G) is a
group.
Example 2.6.12. The complex conjugation map z 7→ z from the additive group C into itself is
an automorphism of C.
Exercise 2.6.13. Show that Aut(K4 ) is isomorphic to S3 . (Hint: Note that K4 = {e, a, b, c},
where a2 = b2 = c2 = e and ab = ba = c, bc = cb = a, ac = ca = b. If f ∈ Aut(K4 ), then f (e) = e
and hence f {a,b,c} is a bijection of the subset {a, b, c} ⊂ K4 onto itself, producing an element
of S3 . Thus we get a map φ : Aut(K4 ) → S3 . Verify that φ is a group isomorphism.)
Definition 2.6.14. The kernel of a group homomorphism f : G → H is the subset

Ker(f ) := {a ∈ G : f (a) = eH } ⊆ G.

Since f (eG ) = eH by Proposition 2.6.5 (i), we have eG ∈ Ker(f ). Therefore, Ker(f ) is a non-
empty subset of G. Given any two elements a, b ∈ Ker(f ) we have f (ab−1 ) = f (a)f (b−1 ) =
f (a)f (b)−1 = eH · e−1
H = eH . Therefore, Ker(f ) is a subgroup of G.

Example 2.6.15. (i) Fix an integer n and consider the homomorphism

f : Z → Zn , a 7→ [a].

Then Ker(f ) = {a ∈ Z : n divides a} = nZ.

(ii) Let S 1 := {z ∈ C : |z| = 1}. Consider the homomorphism
√
−1 t
f : R −→ S 1 , t 7→ e2π .
√
−1 t
Then Ker(f ) = {t ∈ R : e2π = 1} = Z.

The following lemma shows that the kernel of a group homomorphism can be uniquely
determined purely using its universal property. Interesting fact to note is that this description
of kernel of a group homomorphism use only arrows and not any points.
Proposition 2.6.16 (Universal Property of Kernel). Let f : G → H be a group homomorphism.
Then there is a unique subgroup K of G satisfying the following properties.

(K1) f ◦ ιK is trivial, where ιK : K ,→ G is the inclusion map, and

(K2) given any group homomorphism ϕ : G′ → G with f ◦ ϕ trivial, there is a unique group homomor-
phism ψ : G′ → K such that ιK ◦ ψ = ϕ.

G′
∃!ψ f ◦ϕ=eH
(2.6.17) ϕ
x  &/
K
ιK
/G f
H

Proof. We first show the uniqueness of K. Let ιK ′ : K ′ ,→ G be any subgroup of G satisfying

(K1) and (K2). Since the homomorphism f ◦ ιK ′ is trivial, applying (K2) for K we have a
2.6. Group homomorphism 41

unique group homomorphism η : K ′ → K such that ιK ′ = ιK ◦ η. Similarly replacing (K, ιK )

with (K ′ , ιK ′ ), and (G′ , ϕ) with (K, ιK ) in the above diagram (2.6.17), we get a unique group
homomorphism η ′ : K → K ′ such that ιK = ιK ′ ◦ η ′ . Now replace (G′ , ϕ) with (K, ιK ) in the
above diagram 2.6.17. Since both the group homomorphisms IdK : K → K and η ◦ η ′ : K → K
satisfies ιK ◦ (η ◦ η ′ ) = ιK and ιK ◦ IdK = ιK , by uniqueness assumption in (K2), we have
η ◦ η ′ = IdK . Similarly, we have η ′ ◦ η = IdK ′ . Therefore, both η ′ : K → K ′ and η : K ′ → K are
isomorphisms. Since both ιK : K ,→ G and ιK ′ : K ′ ,→ G are inclusion maps, and ιK ◦ η ′ = ιK ′ ,
we must have η ′ is an inclusion map, and hence K ⊆ K ′ . Similarly, we have K ′ ⊆ K, and
hence K = K ′ .
To prove existence, take K = Ker(f ) and ιK : K ,→ G the inclusion map. Clearly, f ◦ ιK is
trivial. For any group homomorphism ϕ : G′ → G with f ◦ ϕ trivial, we have ϕ(a) ∈ K, for all
a ∈ G′ . Thus the image of ϕ lands inside K and hence we have a group homomorphism

ψ : G′ → K, a 7→ ϕ(a)

such that ιK ◦ ψ = ϕ as required.

Proposition 2.6.18. A group homomorphism f : G → H is injective if and only if Ker(f ) is trivial.

Proof. If Ker(f ) ̸= {e}, clearly f is not injective. Conversely, suppose that Ker(f ) = {e}. If
f (a) = f (b), for some a, b ∈ G with a ̸= b, then ab−1 ̸= e and f (ab−1 ) = f (a)f (b−1 ) =
f (a)f (b)−1 = eH , which contradicts our assumption that Ker(f ) = {e}. This completes the
proof.
Proposition 2.6.19. Any infinite cyclic group is isomorphic to Z.

Proof. Let G = ⟨a⟩ be an infinite cyclic group. Define a map f : Z → G by f (n) = an , for all
n ∈ Z. Since
f (n + m) = an+m = an am = f (n)f (m), ∀ m, n ∈ Z,
the map f is a group homomorphism. Since G is infinite, we have an ̸= e, ∀ n ∈ Z \ {0}.
Therefore, Ker(f ) = {e}, and so f is injective. Clearly f is surjective, and hence is an isomor-
phism.

Proposition 2.6.20. Let G be a cyclic group generated by a ∈ G. A homomorphism f : G → G is an

automorphism of G if and only if f (a) is a generator of G.

Proof. Let f : G → G be an automorphism of G. Let b = f (a). Let x ∈ G be arbitrary. Since f

is surjective, there exists y ∈ G such that f (y) = x. Since G = ⟨ a⟩, we have y = an , for some
n ∈ Z. Then x = f (y) = f (an ) = [f (a)]n = bn ∈ ⟨ b ⟩. This shows that G = ⟨ b ⟩, and hence b is a
generator of G. Conversely if f : G → G is a homomorphism such that f (a) generates G, then
f is surjective. If |G| is finite, we must have f is bijective. If G is not finite, then G has only two
generators, namely a and a−1 by Proposition 2.3.15, and hence f must be either IdG or the map
given by sending b ∈ G to b−1 . In both cases, f is injective, and hence is in Aut(G).
Theorem 2.6.21 (Cayley). Every group is a subgroup of a symmetric group.

Proof. Let G be a group. Let S(G) be the symmetric group on G; its elements are all bijective
maps from G onto itself and the group operation is given by composition of bijective maps.
Define a map
φ : G −→ S(G)
by sending an element a ∈ G to the map

φa : G → G, g 7→ ag,
42 Chapter 2. Group Theory

which is bijective (verify!), and hence is an element of S(G). Then given any g ∈ G we have

φ(ab)(g) = φab (g)

= (ab)g = a(bg)
= (φa ◦ φb )(g)
= (φ(a) ◦ φ(b))(g),

and hence φ is a group homomorphism. Note that φa = IdG if and only if a = e in G (ver-
ify!). Therefore, φ is an injective group homomorphism, and hence we can identify G with the
subgroup φ(G) of the symmetric group S(G).

We end this section with the following two results which justify the terminologies intro-
duced in Definition 2.6.9 in the light of category theory.
Proposition 2.6.22. Let f : G → H be a group homomorphism. Then the following are equivalent.

(i) f is injective.
(ii) Given a group T and group homomorphisms ϕ, ψ : T → G with f ◦ ϕ = f ◦ ψ, we have ϕ = ψ.
In other words, f is a monomorphism in the category of groups.
(iii) Given a group T and a group homomorphism ϕ : T → G with f ◦ ϕ trivial, we have ϕ is trivial.

Proof. (i) ⇒ (ii) is Clear. To show (ii) ⇒ (iii), take ψ : T → G to be the trivial group homomor-
phism. Then both f ◦ ϕ and f ◦ ψ are trivial, and hence ϕ is trivial by (ii). To show (iii) ⇒ (i),
take T = Ker(f ) and ϕ : T → G the inclusion map of Ker(f ) into G. Then f ◦ ϕ is trivial, and
hence the inclusion map ϕ : Ker(f ) ,→ G is a trivial group homomorphism by (iii). This forces
Ker(f ) = {e}, and hence f is injective.
Proposition 2.6.23. Let f : G → H be a group homomorphism. Then the following are equivalent.

(i) f is surjective.
(ii) Given a group T and group homomorphisms ϕ, ψ : H → T with ϕ ◦ f = ψ ◦ f , we have ϕ = ψ.
In other words, f is an epimorphism in the category of groups.
(iii) Given a group T and a group homomorphism ϕ : H → T with ϕ ◦ f trivial, we have ϕ is trivial.

Proof. (i) ⇒ (ii): Let ϕ, ψ : H → T be group homomorphisms with ϕ ◦ f = ψ ◦ f . Since f is

surjective, given h ∈ H there exists g ∈ G such that f (g) = h. Then (ϕ ◦ f )(g) = (ψ ◦ f )(g) gives
ϕ(h) = ψ(h). Since h ∈ H is arbitrary, we have ϕ = ψ.
(ii) ⇒ (iii): Take ψ : H → T to be the trivial group homomorphism.
(iii) ⇒ (i): We use the notion of coset of a subgroup. See Proposition 2.7.18 for a proof.

2.7 Notion of Quotient & Cosets

Let G be a group, and H a subgroup of G. In this section we introduce the notion of a
quotient group of G by H and prove its uniqueness. In the process of construction of quotient,
we identify a class of subsets of G, known as cosets of H in G, and discuss their basic properties
with some applications. An explicit construction of quotient group will appear in the next
section.
Definition 2.7.1 (Quotient Group). Let H be a subgroup of a group G. The quotient of G by H
is a pair (Q, π), where Q is a group and π : G → Q is a surjective group homomorphism such
that
2.7. Notion of Quotient & Cosets 43

(QG1) π(h) = eQ , the neutral element of Q, for all h ∈ H, and

(QG2) Universal property of quotient: given a group T and a group homomorphism t : G → T

satisfying H ⊆ Ker(t), there exists a unique group homomorphism et : Q → T such that
t ◦ π = t; i.e., the following diagram commutes.
e

G
t /8 T
(2.7.2) π
 ∃!e
t
Q

Interesting point is that, without knowing existence of such a pair (Q, q), it follows imme-
diately from the properties (QG1) and (QG2) that such a pair (Q, q), if it exists, must be unique
up to a unique isomorphism of groups in the following sense.

Proposition 2.7.3 (Uniqueness of Quotient). With the above notations, if (Q, π) and (Q′ , π ′ ) are two
quotients of G by H, then there exists a unique group isomorphism φ : Q → Q′ such that φ ◦ π = π ′ .

Proof. Taking (T, t) = (Q′ , π ′ ) by universal property of quotient (Q, π) we have a unique group
homomorphism πe′ : Q → Q′ such that πe′ ◦ π = π ′ . Similarly, taking (T, t) = (Q, π) by universal
property of quotient (Q′ , π ′ ) we have a unique group homomorphism π e : Q′ → Q such that
′
e ◦ π = π. Since both π
π e ◦ πe′ and IdQ are group homomorphisms from Q into itself making the
following diagram commutative,

G
π π
π′
x 
Q
πe′ / Q′ π
e /& 9 Q

IdQ

e = IdQ′ . Therefore, πe′ : Q → Q′ is the unique group

e ◦ πe′ = IdQ . Similarly πe′ ◦ π
it follows that π
′ ′
isomorphisms such that π ◦ π = π . This completes the proof.
e

Now question is about existence of quotient. We shall see shortly that we need to impose
an additional hypothesis on H (namely H should be a normal subgroup of G) for existence of
quotient. The condition (QG1) says that π(H) = {eQ }. Since π : G → Q is a group homomor-
phism by assumption, given any two elements a, b ∈ G with a−1 b ∈ H we have π(a−1 b) = eQ ,
and hence π(a) = π(b). In other words, two elements a, b ∈ G are in the same fiber of the map
π : G → Q if a−1 b ∈ H. Since the set of all fibers of any set map f : G → Q gives a partition of
G, and hence an equivalence relation on G, the condition (QG1) suggests us to define a relation
ρL on G by setting
(a, b) ∈ ρL if a−1 b ∈ H.
It is easy to check that ρL is an equivalence relation on G (verify!). The ρL -equivalence class of
an element a ∈ G is the subset

[a]ρL := {b ∈ G : a−1 b ∈ H} = {ah : h ∈ H},

which we denote by aH; the subset aH is called the left coset of H in G represented by a. Note
that (verify!), given a, b ∈ G,

(i) either aH ∩ bH = ∅ or aH = bH,

(ii) aH = bH if and only if a−1 b ∈ H, and
44 Chapter 2. Group Theory

S
(iii) G = aH.
a∈G

Proposition 2.7.4. For each a ∈ G, the map φa : H → aH defined by φa (h) = ah, for all h ∈ H, is
bijective. Consequently, |aH| = |bH|, for all a, b ∈ H.

Proof. Since every element of aH is of the form ah, for some h ∈ H, we see that φa (h) = ah,
and hence φa is surjective. Since ah = ah′ implies that h = (a−1 a)h = a−1 (ah) = a−1 (ah′ ) =
(a−1 a)h′ = h′ , we see that φa is injective. Therefore, φa is bijective. Thus, both H and aH have
the same cardinality.

Let G/H = {aH : a ∈ G} be the set of all distinct left cosets of H in G.

Theorem 2.7.5 (Lagrange’s Theorem). Let G be a finite group, and H a subgroup of G. Then |H|
divides |G|.

Proof. Since ρL is an equivalence relation on G, it follows from Proposition 2.1.31 that G is a

disjoint union of distinct left cosets of H in G. Since G is finite, there can be at most finitely
many distinct left cosets of H in G. Since |aH| = |bH|, for all a, b ∈ G (see Proposition 2.7.4), it
follows that
|G| = |G/H| · |H|,
where |G/H| is the cardinality of the set G/H, i.e., the number of distinct left cosets of H in G.
This completes the proof.
Exercise 2.7.6. Let G be a finite group of order mn having subgroups H and K of orders m and
n, respectively. If gcd(m, n) = 1 show that HK := {hk ∈ G : h ∈ H, k ∈ K} is a group.
Corollary 2.7.7. Let G be a finite group of order n. Then for any a ∈ G, ord(a) divides n. In particular,
an = e, ∀ a ∈ G.

Proof. Let H be the cyclic subgroup of G generated by a. Since G is a finite group, so is H. Then
by Lagrange’s theorem 2.7.5, |H| divides |G| = n. Since |H| = ord(a), the result follows. To see
the second part, note that if ord(a) = k, then n = km, for some m ∈ N, and so an = (ak )m =
em = e.
Corollary 2.7.8. Any group of prime order is cyclic.

Proof. Let G be a finite group of order p, where p is a prime number. If p = 2, then clearly G
is cyclic. Suppose that p > 2. Then there is an element a ∈ G such that a ̸= e. Since the cyclic
subgroup ∈ Ha := ⟨a⟩ = {an : n ∈ Z} contains both a and e, we have |Ha | ≥ 2. Since |Ha |
divides |G| = p by Lagrange’s theorem, we must have |Ha | = p, because p is prime. Then we
must have G = Ha , and hence G is cyclic.
Corollary 2.7.9 (Euler’s Theorem). Let n ≥ 2 be an integer. Then for any positive integer a with
gcd(a, n) = 1, we have aϕ(n) ≡ 1 (mod n), where ϕ(n) is the number of elements in the set {k ∈ N :
1 ≤ k < n and gcd(k, n) = 1}.

Proof. Note that, Un := {[a] ∈ Zn : gcd(a, n) = 1} is a finite subset of Zn containing ϕ(n)

elements. Since Un is a group with respect to the multiplication operation modulo n, for any
[a] ∈ Un we have [a]ϕ(n) = [1]. In other words, aϕ(n) ≡ 1 (mod n).
Corollary 2.7.10 (Fermat’s little theorem). If p > 0 is a prime number, then for any positive integer
a with gcd(a, p) = 1, we have ap−1 ≡ 1 (mod p).

Proof. Since ϕ(p) = |Up | = p − 1, the result follows from the Corollary 2.7.9.
Exercise 2.7.11. Show that 26000 − 1 is divisible by 7.
2.7. Notion of Quotient & Cosets 45

Solution. Since gcd(2, 7) = 1, by Fermat’s little theorem we have 27−1 ≡ 1 (mod 7). So [26 ] = [1]
in Z7 . Then [26 ]1000 = [1]1000 = [11000 ] = [1] in Z7 . Therefore, 26000 ≡ 1 (mod 7), and hence
26000 − 1 is divisible by 7.
Exercise 2.7.12. Show that 151000 − 1 and 1051200 − 1 are divisible by 8.
Exercise 2.7.13. Define a relation ρR on G by setting

(a, b) ∈ ρR if ab−1 ∈ H.

(i) Show that ρR is an equivalence relation on G.

(ii) Show that the ρR -equivalence class of a ∈ G in G is the subset of G defined by

[a]ρR := {b ∈ G : a−1 b ∈ H} = {ha : h ∈ H} =: Ha.

The subset Ha ⊆ G is called the right coset of H in G represented by a.

(iii) Show that if G is abelian then aH = Ha, for all a ∈ G.
(iv) Give an example of a group G, two subgroups H and K of G, and an element b ∈ G such
that that bK ̸= Kb, while aH = Ha holds, for all a ∈ G. (Hint: Take G = S3 , and

H := {e, (1 2 3), (1 3 2)} ⊂ S3 and K := {e, (2 3)} ⊂ S3 .

Note that both H and K are subgroups of S3 . Verify that aH = Ha, ∀ a ∈ S3 , while for
b = (1 3 2) ∈ S3 we have bK ̸= Kb.)
(v) Show that H and Ha have the same cardinality, for all a ∈ G.

The set of all distinct right cosets of H in G is denoted by

H\ G = {Ha : a ∈ G}.

Lemma 2.7.14. Let H be a subgroup of a group G. Then there is a one-to-one correspondence between
the set of all left cosets of H in G and the set of all right cosets of H in G. In other words, there is a
bijective map φ : G/H −→ H\G. Therefore, both the sets G/H and H\G have the same cardinality.

Proof. Define a map φ : {aH : a ∈ G} −→ {Hb : b ∈ G} by sending φ(aH) = Ha−1 , for all
a ∈ G. Note that, aH = bH if and only if a−1 b ∈ H if and only if a−1 (b−1 )−1 ∈ H if and only if
Ha−1 = Hb−1 . Therefore, φ is well-defined and injective. To show φ bijective, note that given
any Hb ∈ {Hb : b ∈ G} we have φ(b−1 H) = Hb. Thus, φ is surjective, and hence is a bijective
map.
Definition 2.7.15. Let H be a subgroup of a group G. We define the index of H in G, denoted as
[G : H], to be the cardinality |G/H| = |H\G|. In case, this is a finite number, the index [G : H]
is the number of distinct left (and right) cosets of H in G.
Exercise 2.7.16. Let H and K be two subgroups of G of finite indices. Show that H ∩ K is a
subgroup of G of finite index.
Example 2.7.17. The index of nZ in Z is n. Indeed, given any two elements a, b ∈ Z, we have
a − b ∈ nZ if and only if a ≡ b (mod n). Therefore, the left coset of nZ represented by a ∈ Z is
precisely the equivalence class

[a] := {b ∈ Z : a ≡ b (mod n)} = a + nZ.

Since there are exactly n such distinct equivalence classes by division algorithm, namely

a + nZ, where 0 ≤ a ≤ n − 1;
46 Chapter 2. Group Theory

(c.f. Example 2.1.32), we conclude that the index of nZ in Z is [Z : nZ] = n. We shall explain it
later using group homomorphism and quotient group.
Proposition 2.7.18 (Epimorphism of groups is surjective). Let f : G → H be a group homomor-
phism satisfying the following property:

• Given a group T and a group homomorphism ϕ : H → T with ϕ ◦ f trivial, we have ϕ is trivial.

Then f is surjective.

Proof. Note that A := f (G) is a subgroup of H, and so we can consider the set

A\H = {Ah : h ∈ H}

consisting of all distinct right cosets of A in H. Let A′ be a subset of H which is not a right coset
of A in H, and let S = {A′ } ∪ H/A. Let T = Aut(S) be the symmetric group on S; its elements
are bijective maps from S onto itself and the group operation is given by composition of maps.
Note that, given h ∈ H, consider the map

φh : A\H → A\H
−1 −1
that sends Ah′ ∈ A\H to A(h′ h) ∈ A\H. Since (h′ h)(h′′ h)−1 = h′ hh−1 h′′ = h′ h′′ , it
follows that φh is well-defined and injective. Since φh−1 ◦ φh = IdA\H = φh ◦ φh−1 , the map φh
is bijective.
Let φ : H → T := Aut(S) be the map given by sending h ∈ H to the permutation φ(h) ∈
Aut(S) which is defined by

φ(h)(A′ ) = A′ and φ(h) A\H

= φh .

It is easy to verify that φ is a group homomorphism. Let σ ∈ T = Aut(S) be the permutation

that interchanges A and A′ , and keeps everything else fixed; i.e., σ is the 2-cycle σ = (A A′ ).
Then the map

(2.7.19) ψ : H → T, h 7→ σ −1 φ(h)σ,

is a group homomorphism (verify!).

If a ∈ A, then φ(a)(A) = Aa = A and φ(a)(A′ ) = A′ . Then φ(a) ∈ T is disjoint from the
2-cycle σ = (A A′ ), and hence they commute to give ψ(a) = σ −1 φ(a)σ = φ(a). Therefore,
φ A = ψ A and hence φ ◦ f = ψ ◦ f . Since f is an epimorphism, we have φ = ψ. Then
φ(h) = σ −1 φ(h)σ, for all h ∈ H. Since σ = (A A′ ) and φ(h)(A′ ) = A′ , we have φ(h)(A) =
(σ −1 φ(h)σ)(A) = σ −1 φ(h)(A′ ) = σ −1 (A′ ) = A. Since φ(h)(A) = Ah by definition, we have
Ah = A, and hence h ∈ A. Since h ∈ H is arbitrary, we have A = H, as required.
Exercise 2.7.20. (i) Does there exists a group isomorphism from (Q, +) onto (Q∗ , ·)?
(ii) Does there exists a surjective group homomorphism from (Q, +) onto (Q+ , ·)?
(iii) Does there exists a non-trivial group homomorphism from Q into Z?

2.8 Normal Subgroup & Quotient Group

In this section we introduce the notion of normal subgroup and give a construction of quo-
tient of a group by its normal subgroup. Recall that the condition (QG1) in Definition 2.7.1 of
quotient group suggests us to consider the set

G/H := {gH : g ∈ G}
2.8. Normal Subgroup & Quotient Group 47

consisting of all left cosets of H in G as a possible candidate for the set Q. Now question is what
should be the appropriate group structure on it? Take any group homomorphism f : G → T
such that H ⊆ Ker(f ). Then we have f (a) = f (b) if a−1 b ∈ H. The commutativity of the
diagram (2.7.2) tells us to send aH ∈ Q to f (a) ∈ T to define the map fe : Q → T which needs
to be a group homomorphism. Then we should have

(2.8.1) fe((aH)(bH)) = f (ab) = fe((ab)H), ∀ a, b ∈ G.

This suggests us to define a binary operation on the set G/H = {gH : g ∈ G} by

(2.8.2) (aH)(bH) := (ab)H, ∀ a, b ∈ G.

Proposition 2.8.3. The map G/H × G/H → G/H defined by sending (aH, bH) to (ab)H is well-
defined if and only if

(2.8.4) g −1 hg ∈ H, ∀ g ∈ G and h ∈ H.

Proof. Suppose the the above map is well-defined. Let h ∈ H and g ∈ G be arbitrary. Then
hH = H, and hence (hH) · (gH) = H · (gH). Since the above defined binary operation on G/H
is well-defined, we have (hg)H = gH and hence g −1 hg ∈ H.
Conversely, suppose that g −1 hg ∈ H, for all g ∈ G and h ∈ H. Let a1 H = a2 H and
b1 H = b2 H, for some a1 , a2 , b1 , b2 ∈ G. Then h := a−1 −1
1 a2 ∈ H and b1 b2 ∈ H. Then

(a1 b1 )−1 (a2 b2 ) = b−1 −1

1 a1 a2 b2
= b−1 −1
1 hb2 , since h := a1 a2 .
= (b−1 −1
1 hb1 )(b1 b2 ) ∈ H,

since H is a group and both b−1 −1

1 hb1 and b1 b2 are in H. Therefore, (a1 b1 )H = (a2 b2 )H, as
required.

Proposition 2.8.3 suggests us to reserve a terminology for those subgroups H of G that

satisfies the property (2.8.4).
Definition 2.8.5 (Normal Subgroup). A subgroup H of a group G is said to be normal in G if
g −1 hg ∈ H, ∀ g ∈ G, h ∈ H.
Exercise 2.8.6. Let G be a group and H a subgroup of G. Given a ∈ G, let

Ha := {ha : h ∈ H} ⊆ G.

Show that the following are equivalent.

(i) aH = Ha, for all a ∈ G.

(ii) a−1 Ha = H, for all a ∈ G.

(iii) a−1 Ha ⊆ H, for all a ∈ G.

(iv) a−1 ha ∈ H, for all a ∈ G and h ∈ H.
Proposition 2.8.7. Any subgroup of index 2 is normal.

Proof. Let H be a subgroup of G such that [G : H] = 2. Then H has only two left (resp., right)
cosets, namely H and aH (resp., H and Ha), where a ∈ G \ H. Since G = H ⊔ aH = H ⊔ Ha,
for any a ∈ G \ H, we see that aH = Ha, for all a ∈ G, and hence aHa−1 = H, for all a ∈ G.
This completes the proof.
48 Chapter 2. Group Theory

Corollary 2.8.8. For all n ≥ 3, An is a normal subgroup of Sn .

Exercise 2.8.9. (i) Show that any subgroups of an abelian group G is normal in G.
(ii) Let H = ⟨ (1 2 3)⟩ be the cyclic subgroup of S3 generated by the 3-cycle (1 2 3) ∈ S3 .
Show that H is a normal subgroup of S3 .
(iii) Verify if the subgroup K := ⟨ (1 2)⟩ of S3 is normal or not.

(iv) Determine all normal subgroups of S3 .

(v) Show that SLn (R) is a normal subgroup of GLn (R), for all n ∈ N.
Exercise 2.8.10. Let H be a subgroup of G. Let ρ = {(a, b) ∈ G × G : a−1 b ∈ H} ⊆ G × G. Note
that ρ is an equivalence relation on G. Show that H is a normal subgroup of G if and only if ρ
is a subgroup of the direct product group G × G (see Exercise 2.1.34).

Lemma 2.8.11. The kernel of a group homomorphism f : G → H is a normal subgroup of G.

Proof. For any a ∈ G and b ∈ Ker(f ), we have f (aba−1 ) = f (a)f (b)f (a−1 ) = f (a)eH f (a)−1 =
eH , and hence aba−1 ∈ Ker(f ). Therefore, Ker(f ) is a normal subgroup of G.
Exercise 2.8.12. For n ≥ 2, show that An is a normal subgroup of Sn by constructing a group
homomorphism φ : Sn → µ2 = {1, −1} such that Ker(φ) = An .
Exercise 2.8.13. For n ≥ 1, show that SLn (R) is a normal subgroup of GLn (R) by constructing
a group homomorphism φ : GLn (R) → R∗ such that Ker(φ) = SLn (R).
Lemma 2.8.14. Let f : G → H be a group homomorphism. If K is a normal subgroup of H, then
f −1 (K) is a normal subgroup of G.

Proof. Suppose that K is a normal subgroup of H. Then for any a ∈ G and b ∈ f −1 (K), we
have f (aba−1 ) = f (a)f (b)f (a)−1 ∈ K, and hence aba−1 ∈ f −1 (K).
Exercise 2.8.15. Show that N := {A ∈ GLn (C) : | det(A)| = 1} is a normal subgroup of GLn (C).

Remark 2.8.16. Normal subgroup of a normal subgroup need not be normal. To elaborate it,
there exists a group G together with a normal subgroup H of G such that H has a normal
subgroup K which is not a normal subgroup of G. Can you give such an example?
Theorem 2.8.17 (Existence of Quotient Group). Let H be a normal subgroup of a group G. Then
the quotient group (Q, π) of G by H exists and is unique in the sense that if (Q, π) and (Q′ , π ′ ) are two
quotients of G by H, then there exists a unique isomorphism of groups φ : Q → Q′ such that φ◦π ′ = π.
We denote Q by G/H.

Proof. Since H is a normal subgroup of G,

(aH)(bH) := (ab)H, ∀ a, b ∈ G,

is a well-defined binary operation on the set G/H := {aH : a ∈ G}; see Proposition 2.8.3.
Given any a, b, c ∈ G, we have

(aH · bH) · cH = (ab)H · cH = ((ab)c)H = (a(bc))H = aH · (bc)H = aH · (bH · cH).

Therefore, the binary operation on G/H is associative. Given any aH ∈ G/H, we have

aH · eH = (ae)H = aH
and eH · aH = (ea)H = aH.
2.8. Normal Subgroup & Quotient Group 49

Therefore, eH = H ∈ G/H is neutral element for the binary operation on G/H. Given any
aH ∈ G/H, note that

aH · a−1 H = (aa−1 )H = eH
and a−1 H · aH = (a−1 a)H = eH.

Therefore, G/H is a group. Set Q := G/H and consider the map

(2.8.18) π : G −→ Q defined by π(a) = aH, ∀ a ∈ G.

Clearly π is surjective and given a, b ∈ G we have π(ab) = (ab)H = (aH)(bH) = π(a)π(b).

Therefore, π is a group homomorphism. Since for any h ∈ H, we have π(h) = hH = eH = H,
the neutral element of the group G/H, we see that H ⊆ Ker(π). Let T be any group and
t : G → T be a group homomorphism satisfying t(h) = eT , the neutral element of T , for all
t ∈ T . Since aH = bH if and only if a−1 b ∈ H, applying π on a−1 b we see that π(a) = π(b).
Therefore, the map

(2.8.19) t : G/H → T, aH 7−→ t(a),

e

is well-defined. Since

t((aH)(bH)) = e
e t((ab)H) = f (ab) = f (a)f (b) = e
t(aH)e
t(bH),

we conclude that e t is a group homomorphism. Since (e t ◦ π)(a) = e

t(aH) = f (a), ∀ a ∈ G, we
t ◦ π = f . If ξ : G/H → T is any group homomorphism satisfying ξ ◦ π = t, then for any
have e
a ∈ G we have e t(aH) = (e t ◦ π)(a) = t(a) = (ξ ◦ π)(a) = ξ(aH), and hence et = ξ. Therefore,
the pair (G/H, π) satisfy the properties (QG1) and (QG2), and hence is a quotient of G by H.
Uniqueness is already shown in Proposition 2.7.3.
Corollary 2.8.20. Let H be a normal subgroup of a group G, and let (G/H, π) be the associated quotient
of G by H. Then Ker(π) = H.

Proof. Since the group operation on the quotient group G/H := {aH : a ∈ G} is given by
(aH)(bH) := (ab)H, ∀ aH, bH ∈ G/H, we have

Ker(π) = {a ∈ G : π(a) = H}
= {a ∈ G : aH = H}
= {a ∈ G : a ∈ H} = H.

This completes the proof.

Exercise 2.8.21. Let G be a group such that G/Z(G) is cyclic. Show that G is abelian.

Solution: Let Z := Z(G). Suppose that G/Z is cyclic. Then G/Z = ⟨ aZ ⟩, for some a ∈ G. Let
x ∈ G be arbitrary. Then xZ = (aZ)n = an Z, for some n ∈ Z. Then a−n x = (an )−1 x ∈ Z.
Therefore, a−n x = z, for some z ∈ Z, and so x = an z, for some z ∈ Z = Z(G). Let y ∈ G be
given. Then as before, y = am w, for some m ∈ Z and w ∈ Z(G). Since z, w ∈ Z(G), we have
xy = an zam w = am wan z = yx, as required.
Corollary 2.8.22. There is no group G such that |G/Z(G)| is a prime number.
50 Chapter 2. Group Theory

2.9 Isomorphism Theorems

Let G be a group. Given a normal subgroup K of G, let (G/K, π) be the associated quotient
group of G by K, where
π : G → G/K = {aK : a ∈ G}
is the natural quotient homomorphism given by

π(a) = aK, ∀ a ∈ G.

Theorem 2.9.1. Let f : G → H be a group homomorphism. Let K be a normal subgroup of G such

that K ⊆ Ker(f ). Then there is a unique group homomorphism fe : G/K −→ H such that fe ◦ π = f ,
where π : G → G/K is the quotient homomorphism.

7/ H
f
G
π
 fe
G/K

Furthermore, fe is injective if and only if K = Ker(f ).

Proof. Since K is a normal subgroup of G, the quotient group G/K exists with the natural
surjective group homomorphism π : G → G/K defined by π(a) = aK, ∀ a ∈ G. Since
K ⊆ Ker(f ), by universal property of quotient (see Definition 2.7.1) we have a unique group
homomorphism fe : G/K → H such that fe ◦ π = f . The fact that fe is a well-defined group
homomorphism can also be directly checked by observing that

fe(aK) = (fe ◦ π)(a) = f (a), ∀ a ∈ G.

Since Ker(fe) = {gK : f (g) = eH } = {gK : g ∈ Ker(f )}, we see that Ker(fe) is trivial
(meaning that, it is a trivial subgroup) if and only if gK = K, ∀ g ∈ Ker(f ). This is equivalent
to say that, g ∈ K, ∀ g ∈ Ker(f ), i.e., Ker(f ) ⊆ K. Since K ⊆ Ker(f ) by assumption, it follows
from Proposition 2.6.18 that fe is injective if and only if K = Ker(f ).

As an immediate corollary, we have the following.

Corollary 2.9.2 (First Isomorphism Theorem). Let f : G → H be a surjective homomorphism of
groups. Then f induces a natural isomorphism of groups fe : G/Ker(f ) → H.

Proof. Note that Ker(f ) is a normal subgroup of G. It follows from Theorem 2.9.1 that the
group homomorphism fe : G/Ker(f ) → H induced by f is injective. Since f is surjective and
fe ◦ π = f , where π : G → G/Ker(f ) is the natural surjective homomorphism, it follows that fe
is surjective. Therefore, fe is a bijective group homomorphism, and hence is an isomorphism of
groups.

Let G be a group. Note that given a normal subgroup N of G, the quotient group G/N
of G by N comes with a natural surjective group homomorphism πN : G → G/N such that
Ker(πN ) = N (see Definition 2.7.1 and Corollary 2.8.20). On the other hand, given a group Q
and a surjective group homomorphism π : G → Q, its kernel Ker(π) is a normal subgroup of
G such that G/Ker(π) ∼= Q by the First isomorphism theorem (Corollary 2.9.2) for groups. This
motivates us to define the following (c.f. Definition 2.7.1).
Definition 2.9.3. A quotient group of G is a pair (Q, π), where Q is a group and π : G → Q is a
surjective group homomorphism.
2.9. Isomorphism Theorems 51

As an immediate consequence, we have the following.

Corollary 2.9.4. Given a group G, there is a one-to-one correspondence between the following two sets:

(i) NG := the set of all normal subgroups of G, and

(ii) QG := the set of all quotient groups of G.

Proof. Define a map Φ : NG → QG by sending a normal subgroup N of G to the associ-

ated quotient group (G/N, πN ) ∈ QG . Since πN is a surjective group homomorphism with
Ker(πN ) = N , the map Φ admits an inverse, namely Ψ : QG → NG given by sending a quotient
group (Q, π) of G to the kernel N := Ker(π) ∈ NG . Since the pairs (G/N, πN ) and (Q, π) are
uniquely isomorphic, we conclude that Φ and Ψ are inverse to each other. This completes the
proof.

Proposition 2.9.5. The group Zn is isomorphic to Z/nZ.

Proof. Let f : Z → Zn be the map defined by

f (k) = [k], ∀ k ∈ Z.

Since
f (k1 + k2 ) = [k1 + k2 ] = [k1 ] + [k2 ] = f (k1 ) + f (k2 ), ∀ k1 , k2 ∈ Z,
we see that f is a group homomorphism. Clearly f is surjective (verify!). Note that Ker(f ) =
{k ∈ Z : [k] = [0]} = nZ. Then by first isomorphism theorem we have Z/nZ ∼ = Zn .
Proposition 2.9.6. Any finite cyclic group of order n is isomorphic to Zn .

Proof. Let G be a finite cyclic group of order n. Then there exists a ∈ G such that ⟨a⟩ = {ak :
k ∈ Z} = G. Define a map f : Z → G by

f (k) = ak , ∀ k ∈ Z.

Since
f (k1 + k2 ) = ak1 +k2 = ak1 ak2 = f (k1 )f (k2 ), ∀ k1 , k2 ∈ Z,
f is a group homomorphism. Clearly f is surjective because every element of G is of the form
ak , for some k ∈ Z. Then by first isomorphism theorem G is isomorphic to Z/Ker(f ). Note
that, Ker(f ) = {k ∈ Z : ak = e}. Since G is a cyclic group of order n generated by a, we have
ord(a) = n (see Corollary 2.3.11). Then we have Ker(f ) = {k ∈ Z : ak = e} = nZ. Therefore,
G∼ = Z/nZ. Since Z/nZ ∼ = Zn by Theorem 2.9.5, we have G ∼ = Zn .
Exercise 2.9.7. Fix an integer n ≥ 2. Show that Z(GLn (C)) = {λIn : λ ∈ C∗ } ∼ = C∗ and
∗ n ∼
Z(SLn (C)) = {λIn : λ ∈ C , λ = 1} = µn (C), where µn (C) is the group of all complex
n-th roots of unity. The quotient groups PGLn (C) := GLn (C)/Z(GLn (C)) and PSLn (C) :=
SLn (C)/Z(SLn (C)) are called the projective general linear group over C and the projective special
linear group over C, respectively.
Exercise 2.9.8 (Projective linear groups). Fix an integer n ≥ 2, and consider the map

f : SLn (C) → PGLn (C) := GLn (C)/Z(GLn (C))

defined by the composition of the inclusion of SLn (C) into GLn (C) with the quotient homo-
morphism GLn (C) → PGLn (C). Show that f is a surjective group homomorphism with kernel
Z(SLn (C)), and hence we have an isomorphism of complex projective linear groups PSLn (C) :=
SLn (C)/Z(SLn (C)) ∼
= PGLn (C). Show that surjectivity of f fails if we work over R instead of
C.
52 Chapter 2. Group Theory

Let G be a group. Given a ∈ G, the map φa : G → G defined by

φa (b) = aba−1 , ∀ b ∈ G,

is a group homomorphism. Indeed,

φa (bc) = a(bc)a−1 = (aba−1 )(aca−1 ) = φa (b)φa (c), ∀ b, c ∈ G.

Since Ker(φa ) = {b ∈ G : aba−1 = e} = {e}, φa is injective. Given c ∈ G, note that φa (a−1 ca) =
a(a−1 ca)a−1 = c, and so φa is surjective. Therefore, φa is an isomorphism.
Definition 2.9.9. An automorphism φ ∈ Aut(G) is said to be an inner automorphism of G if there
exists a ∈ G such that φ(b) = aba−1 , for all b ∈ G.
Proposition 2.9.10. Let G be a group. Let Inn(G) be the set of all inner autormorphisms of G. Then
Inn(G) is a subgroup of Aut(G).

Proof. Note that the identity map IdG : G → G is in Inn(G). Given f, g ∈ Inn(G), there exists
a, b ∈ G such that f and g(x) = bxb−1 , for all x ∈ G. Then f −1 = φa−1 , and that (φ−1
a ◦ φb )(x) =
a−1 bxb−1 a = (a−1 b)x(a−1 b)−1 = φa−1 b (x), for all x ∈ G. Therefore, Inn(G) is a subgroup of
Aut(G).
Proposition 2.9.11. The map φ : G → Inn(G) that sends a ∈ G to the map φa : G → G defined by

φ(a)(b) = aba−1 , ∀ b ∈ G,

is a surjective group homomorphism with kernel Z(G). Consequently, G/Z(G) ∼

= Inn(G).

Proof. Let a, b ∈ G be given. Then for any x ∈ G we have φ(ab)(x) = (ab)x(ab)−1 = a(bxb−1 )a−1 =
a(φb (x))a−1 = (φa ◦ φb )(x), and hence φ(ab) = φ(a) ◦ φ(b). Therefore, φ is a group homomor-
phism. Since every element of Inn(G) is of the form φa , for some a ∈ G, the map φ is surjective.
Since Ker(φ) = {a ∈ G : φ(a) = IdG } = {a ∈ G : aba−1 = b, ∀ b ∈ G} = Z(G), by the first
isomorphism theorem for groups we have G/Z(G) ∼ = Inn(G).
Exercise 2.9.12. Let G be a group such that G/Z(G) is cyclic. Show that Inn(G) is a trivial
subgroup of Aut(G).
Theorem 2.9.13 (Second Isomorphism Theorem). Let G be a group. Let H and K be subgroups of
G with K normal in G. Then

(i) HK is a subgroup of G,
(ii) K is a normal subgroup of HK, and
(iii) H/(H ∩ K) ∼
= HK/K.

Proof. (i) Let h ∈ H and k ∈ K be arbitrary. Since K is a normal subgroup of G, we have

hk = (hkh−1 )h ∈ KH and so HK ⊆ KH. Similarly, kh = h(h−1 kh) ∈ HK shows that
KH ⊆ HK. Thus HK = KH and hence HK is a subgroup of G by Theorem 2.4.3.
(ii) Clearly K is a subgroup of HK. Since K is normal in G, given any a ∈ HK ⊆ G and
k ∈ K we have aka−1 ∈ K, and hence K is a normal subgroup of HK.
(iii) Define a map φ : H → HK/K by φ(a) = aK, for all a ∈ H. Since φ(ab) = (ab)K =
(aK)(bK) = φ(a)φ(b), for all a, b ∈ H, φ is a group homomorphism. Since K ∈ HK/K is the
neutral element, given any h ∈ H and k ∈ K we have (hk)K = (hK)(kK) = hK = φ(h), and
so φ is surjective. Since

Ker(φ) = {h ∈ H : hK = K} = {h ∈ H : h ∈ K} = H ∩ K,

by first isomorphism theorem (see Corollary 2.9.2) we have H/(H ∩ K) ∼

= HK/K.
2.9. Isomorphism Theorems 53

Example 2.9.14. Let m, n ∈ N with gcd(m, n) = 1. Consider the subgroups H = mZ and

K = nZ of (Z, +). Since Z is abelian, K is a normal subgroup of Z. Since gcd(m, n) = 1, there
exists a, b ∈ Z such that am + bn = 1, and so 1 ∈ H + K. Since gcd(m, n) = 1, we have
lcm(m, n) = mn, and so H ∩ K = mnZ. Then by the second isomorphism theorem we have
mZ/mnZ = H/(H ∩ K) ∼ = (H + K)/K = Z/nZ. Generalize this to the case when m and n are
not necessarily coprime.
Exercise 2.9.15. Use the second isomorphism theorem for groups to prove the following.

(i) 3Z/15Z ∼
= Z/5Z, and
(ii) 6Z/30Z ∼
= 2Z/10Z. (Hint: Take H = 6Z and K = 10Z).
Theorem 2.9.16 (Abelianization). Let G be a group. Then upto isomorphism there exists a unique
pair (Gab , Φ) consisting of an abelian group Gab and a surjective group homomorphism Φ : G → Gab
satisfying the following universal property: given any abelain group H and a group homomorphism
f : G → H, there exists a unique group homomorphism fe : Gab → H such that fe ◦ Φ = f .

7/ H
f
G
Φ
 fe
Gab

The group Gab is known as the maximal abelian quotient or the abelianization of G.

Proof. Uniqueness: First we prove uniqueness of the pair (Gab , Φ) upto unique isomorphism
of groups. Suppose that (K, g) be another such pair consisting of an abelian group K and
a surjective group homomorphism g : G → K such that the pair (K, g) satisfies the above
e :K→
universal property. Taking (H, f ) = (Gab , Φ) we find a unique group homomorphism Φ
Gab such that Φe ◦ g = Φ.
G
g g
Φ
w  '/
K
Φ
e
/ Gab g
e
K
Applying universal property of (Gab , Φ) with (H, f ) = (K, g), we have a unique group homo-
morphism ge : Gab → K such that ge ◦ Φ = g. Since the composite map ge ◦ Φ e : K → K is
a group homomorphism, by the universal property of the pair (K, g) we have ge ◦ Φ e = IdK ,
where IdK : K → K is the identity map of K. Similarly, we have Φ e ◦ ge = IdG . Therefore, both
ab

ge : K → Gab and Φ : Gab → K are isomorphism of groups. Since both Φ

e e and ge are unique and
Φe ◦ g = Φ and ge ◦ Φ = g, we conclude that the pair (K, g) is uniquely isomorphic to (Gab , Φ).

Existence: To prove existence of the pair (Gab , Φ), consider the elements of G of the form

[a, b] := aba−1 b−1 ,

where a, b ∈ G, called commutators in G. Clearly [a, b] = e if G is abelian. Let

[G, G] := ⟨aba−1 b−1 : a, b ∈ G⟩

be the subgroup of G generated by all commutators of elements of G. The subgroup [G, G] is

known as the commutator subgroup or the derived subgroup of G. Since

ghg −1 = ghg −1 h−1 h = [g, h]h, ∀ g, h ∈ G,

taking h ∈ [G, G] we see that [G, G] is a normal subgroup of G. Let Gab := G/[G, G] be the
associated quotient group, and let Φ : G → Gab be the natural quotient map which sends a ∈ G
54 Chapter 2. Group Theory

to the coset a[G, G] ∈ G/[G, G] = Gab . Let us denote by a the image of a ∈ G in G/[G, G] under
the quotient map Φ : G → G/[G, G]. Since

(ab)(ba)−1 = aba−1 b−1 ∈ [G, G], ∀ a, b ∈ G,

we have ab = ba in G/[G, G]. Therefore, G/[G, G] is commutative. If f : G → H is a group

homomorphism, then

f ([a, b]) = f (aba−1 b−1 ) = [f (a), f (b)], ∀ a, b ∈ G.

Now suppose that H is abelian. Then for any a, b ∈ G, we have [f (a), f (b)] = e, and so
[a, b] ∈ Ker(f ). Therefore, [G, G] ⊆ Ker(f ). Consequently, by universal property of quotient
(see Definition 2.7.1) there is a unique homomorphism fe : G/[G, G] → H such that fe ◦ Φ = f .
This completes the proof of existence part.
Proposition 2.9.17. The commutator subgroup of Sn is An , for all n ≥ 3.

Proof. Since the signature map sgn : Sn → µ2 = {1, −1} defined by


1, if σ is even,
sgn(σ) =
−1, if σ is odd,

is a group homomorphism (see Lemma 2.6.4), we have sgn(σ)−1 = sgn(σ), for all σ ∈ Sn .
Therefore, given σ, τ ∈ Sn we have

sgn([σ, τ ]) = sgn(σ ◦ τ ◦ σ −1 τ −1 ) = sgn(σ) sgn(τ ) sgn(σ)−1 sgn(τ )−1 = 1.

Therefore, [σ, τ ] ∈ An , for all σ, τ ∈ Sn , and hence [Sn , Sn ] ⊆ An . To show the reverse inclusion,
note that An is generated by 3-cycles, for all n ≥ 3 (see Exercise 2.5.32), and any 3-cycle (i j k)
in Sn can be written as

(i j k) = (i j) ◦ (i k) ◦ (i j)−1 ◦ (i k)−1 ,

which is an element of [Sn , Sn ]. Thus An ⊆ [Sn , Sn ]. This completes the proof.

Exercise 2.9.18. Show that the abelianization of Sn is µ2 , for all n ≥ 3.

Theorem 2.9.19 (Third Isomorphism Theorem). Let H and K be normal subgroups of G with
K ⊆ H. Then we have an isomorphism of groups (G/K)/(H/K) ∼
= G/H.

Proof. Since H and K are normal subgroups of G and K ⊆ H, that K is a normal subgroup of
H, and the associated quotient groups

(i) ϕ : G → G/H,
(ii) ψ : G → G/K, and

(iii) η : H → H/K

exist. Let ιH : H ,→ G be the inclusion of H into G. Then the composite map

ιH ψ
H ,→ G −→ G/K

is a group homomorphism with kernel K, and hence we get an injective group homomorphism

H/K ,→ G/K.
2.9. Isomorphism Theorems 55

Given h ∈ H and a ∈ G, we have aha−1 ∈ H, and so (aK)(hK)(aK)−1 = (ah)K · a−1 K =

(aha−1 )K ∈ H/K. Therefore, H/K is a normal subgroup of G/K, and hence the associated
quotient group π : G/K → (G/K)/(H/K) exists. Consider the diagram

G
ψ
/ G/K

ϕ π
 
/ (G/K)/(H/K)
]
π◦ψ
G/H

Note that H/K ∈ (G/K)/(H/K) is the neutral element of the group (G/K)/(H/K). Moreover,
the composite map π ◦ ψ is a surjective group homomorphism with kernel

Ker(π ◦ ψ) = {a ∈ G : π(ψ(a)) = e}
= {a ∈ G : π(aK) = e}
= {a ∈ G : aK(H/K) = H/K}
= {a ∈ G : aK ∈ H/K}
= {a ∈ G : a ∈ H}, since the map H/K ,→ G/K is injective.
=H

Then by first isomorphism theorem (Corollary 2.9.2) applied to the group homomorphism π ◦ψ
we have the required isomorphism G/H ∼ = (G/K)/(H/K) of groups.
Corollary 2.9.20 (Correspondence Theorem). Let f : G → H be a surjective group homomorphism.
Consider the following two sets:

(i) A:= the set of a subgroups of G containing Ker(f ), and

(ii) B:= the set of all subgroups of H.

Then there is an inclusion preserving bijective map

Φ:A→B

such that a subgroup N ∈ A of G is normal in G if and only if Φ(N ) is normal in H.

Proof. Define a map Φ : A → B by sending a subgroup N of G containing Ker(f ) to its image

f (N ). Note that f (N ) is a subgroup of H by Proposition 2.6.7 (i), and hence is an element of B.
Conversely, given a subgroup K of H, its preimage f −1 (K) is a subgroup of G by Proposition
2.6.7 (ii). Since eH ∈ K we have Ker(f ) = f −1 (e) ⊆ f −1 (K). Thus, f −1 (K) ∈ A. This gives a
map
Ψ : B → A, K 7→ f −1 (K).
It remains to show that Φ and Ψ are inverse to each other. Given N ∈ A, we have (Ψ ◦ Φ)(N ) =
f −1 (f (N )) ⊇ N . If a ∈ f −1 (f (N )), then f (a) = f (b), for some b ∈ N . Then f (ab−1 ) =
f (a)f (b)−1 = eH implies ab−1 ∈ Ker(f ) ⊆ N , and so a = (ab−1 )b ∈ N . Therefore, (Ψ ◦ Φ)(N ) =
f −1 (f (N )) = N , for all N ∈ A, and hence Ψ ◦ Φ = IdA . Conversely, given K ∈ B, we have
(Φ ◦ Ψ)(K) = f (f −1 (K)) = K, since f is surjective. Thus Φ ◦ Ψ = IdB . This completes the
proof.
Exercise 2.9.21. Let H be a normal subgroup of a group G. Show that every subgroup of G/H
is of the form K/H, for some subgroup K of G containing H.
56 Chapter 2. Group Theory

2.10 Direct Product & Direct Sum of Groups

Definition 2.10.1. The direct product of a family of groups {Gα : α ∈ Λ} is a pair (G, {πα }α∈Λ ),
where G is a group and {πα : G → Gα }α∈Λ is a family of group homomorphisms such that
given any group H and a family of group homomorphisms {fα : H → Gα }α∈Λ there exists a
unique group homomorphism f : H → G such that πα ◦ f = fα , for all α ∈ Λ.
Theorem 2.10.2 (Existence & Uniqueness of Product of Groups). The direct product of a family of
groups exists and is unique upto a unique isomorphism in the sense that if (G, {gα : G → Gα }α∈Λ )
and (H, {hα : H → Gα }α∈Λ ) are direct products of the family of groups {Gα : α ∈ Λ}, then there
Q a unique isomorphism of groups ϕ : G → H such that hα ◦ ϕ = gα , for all α ∈ Λ. We denote by
exists
Gα the underlying group of the direct product of the family of groups {Gα : α ∈ Λ}.
α∈Λ

Proof. Since (G, {gα }α∈Λ ) is a direct product by assumption, for the test object (H, {hβ : H →
Gβ }β∈Λ ) we have a group homomorphism φ : G → H such that πα ◦ φ = hα , ∀ α ∈ Λ.
Interchanging the roles of (G, {gα }α∈Λ ) and (H, {hα }α∈Λ ) we have a group homomorphism
ψ : H → G such that πα ◦ ψ = gα , ∀ α ∈ Λ. Since both ψ ◦ φ : G → G and IdG : G → G are
group homomorphisms satisfying

fα ◦ (ψ ◦ φ) = fα and fα ◦ IdG = fα , ∀ α ∈ Λ,

it follows that ψ ◦ φ = IdG . Similarly, φ ◦ ψ = IdH , and hence φ : G → H is the unique

isomorphism such that hα ◦ φ = gα , ∀ α ∈ Λ.
For a construction, let
Y a
Gα := {f : Λ → Gα f (α) ∈ Gα , ∀ α ∈ Λ}.
α∈Λ α∈Λ
Q
Given f, g ∈ Gα we define
α∈Λ
a
fg : Λ → Gα
α∈Λ

by
(f g)(α) := f (α)g(α), ∀ α ∈ Λ.
Q Q
Clearly f g ∈ Gα , and (f g)h = f (gh), ∀ f, g, h ∈ Gα . Let eα ∈ Gα be the neutral
α∈Λ Q α∈Λ
element, for all α ∈ Λ. Then the map e : Λ → Gα given by e(α) = eα , ∀ α ∈ Λ satisfies
α∈Λ
Gα we define f −1 ∈ Gα by f −1 (α) = (fα )−1 ∈
Q Q Q
ef = f e = f, ∀ f ∈ Gα . Given f ∈
α∈Λ α∈Λ α∈Λ
Gα , ∀ α ∈ Λ. Then f f −1 = e = f −1 f . Therefore,
Q
Gα is a group. For each β ∈ Λ, we define a
Q α∈Λ
map πβ : Gα → Gβ by πβ (f ) = f (β). Then πβ is a group homomorphism. Given a group H
α∈Λ Q
and a family {hα : H → Gα }α∈Λ of group homomorphisms, we define a map ψ : H → Gα
` α∈Λ
that sends a ∈ H to the function ψa : Λ → Gα defined by ψa (α) = hα (a), ∀ α ∈ Λ. Then it is
α∈Λ
straight forward to verify that ψ is a group homomorphism satisfying πα ◦ψ = hα , ∀ α ∈ Λ.
Example 2.10.3 (External Direct Product of G1 , . . . , Gn ). Let G1 , . . . , Gn be a finite family of
groups, not necessarily distinct. Define a binary operation on the Cartesian product G :=
G1 × · · · × Gn by

(2.10.4) (a1 , . . . , an ) · (b1 , . . . , bn ) := (a1 b1 , . . . , an bn ),

2.10. Direct Product & Direct Sum of Groups 57

where ai , bi ∈ Gi , for all i = 1, . . . , n. Given ai , bi , ci ∈ Gi , for each i ∈ {1, . . . , n}, we have

(a1 , . . . , an ) · (b1 , . . . , bn ) · (c1 , . . . , cn ) = (a1 b1 , . . . , an bn ) · (c1 , . . . , cn )


= (a1 b1 )c1 , . . . , (an bn )cn


= a1 (b1 c1 ), . . . , an (bn cn )


= (a1 , . . . , an ) · (b1 , . . . , bn ) · (c1 , . . . , cn )

Therefore, the above defined binary operation on the set G is associative. Let ei ∈ Gi be the
neutral element of Gi , for all i ∈ {1, . . . , n}. Then given any ai ∈ Gi , for each i, we have

(a1 , . . . , an ) · (e1 , . . . , en ) = (a1 , . . . , an ) = (e1 , . . . , en ) · (a1 , . . . , an ).

Since
(a1 , . . . , an ) · (a−1 −1 −1 −1
1 , . . . , an ) = (e1 , . . . , en ) = (a1 , . . . , an ) · (a1 , . . . , an ),

we conclude that (a1 , . . . , an )−1 = (a−1 −1

1 , . . . , an ) ∈ G. Therefore, G = G1 × · · · × Gn is a group
with respect to the binary operation defined in (2.10.4).
For each i ∈ {1, . . . , n}, let

(2.10.5) pi : G1 × · · · × Gn → Gi

be the map defined by

(2.10.6) pi (a1 , . . . , an ) = ai , ∀ (a1 , . . . , an ) ∈ G1 × · · · × Gn .

Clearly pi is a surjective group homomorphism (verify!). Let H be a group and let {fi : H →
Gi }1≤i≤n be a family of group homomorphisms. Define a map f : H → G1 × · · · × Gn by

(2.10.7) f (h) = (f1 (h), . . . , fn (h)), ∀ h ∈ H.

Then given any a, b ∈ H we have

f (ab) = (f1 (ab), . . . , fn (ab))

= (f1 (a)f1 (b), . . . , fn (a)fn (b))
= (f1 (a), . . . , fn (a))(f1 (b), . . . , fn (b))
= f (a)f (b).

Therefore, f is a group homomorphism. Clearly pi ◦ f = fi , for all i ∈ {1, . . . , n}. Suppose

that f ′ : H → G1 × · · · × Gn is any group homomorphism such that pi ◦ f ′ = fi , for all
i ∈ {1, . . . , n}. Let h ∈ H be arbitrary. Let f ′ (h) = (a1 , . . . , an ) ∈ G1 × · · · × Gn . Then
fi (h) = (pi ◦ f ′ )(h) = pi (a1 , . . . , an ) = ai , for all i ∈ {1, . . . , n}, and hence f ′ (h) = (a1 , . . . , an ) =
(f1 (h), . . . , fn (h)) = f (h). Therefore, f ′ = f , and hence by universal property of product of
groups (see Definition 2.10.1) we conclude that G1 × · · · × Gn is a direct product of G1 , . . . , Gn .
The group G1 × · · · × Gn is also known as the external direct product of G1 , . . . , Gn .
Corollary 2.10.8. The direct product of a finite family of finite groups G1 , . . . , Gn is a group of order
|G1 | · · · |Gn |. Moreover, G1 × · · · × Gn is abelian if and only if Gi is abelian, for all i ∈ In .
Exercise 2.10.9. Given any two groups G and H, show that Z(G × H) = Z(G) × Z(H).
Proposition 2.10.10. Let G := G1 × · · · × Gn be the external direct product of the family of groups
G1 , . . . , Gn . For each i ∈ In := {1, . . . , n}, let Hi = {(a1 , . . . , an ) ∈ G : aj = ej , ∀ j ̸= i} ⊆ G.
Then we have the following.

(i) Hi is a normal subgroup of G, for all i ∈ In .

(ii) Every element a ∈ G can be uniquely expressed as a = h1 · · · hn , with hi ∈ Hi , for all i ∈ In .
58 Chapter 2. Group Theory

(iii) Hi ∩ (H1 · · · Hi−1 Hi+1 · · · Hn ) = {e}, for all i ∈ In .

(iv) G = H1 · · · Hn .

Proof. (i) Since (e1 , . . . , en ) ∈ Hi , so Hi ̸= ∅. Let a := (a1 , . . . , an ), b := (b1 , . . . , bn ) ∈ Hi .

Then aj = ej = bj , ∀ j ̸= i, and hence a−1 j bj = ej , for all j ̸= i. Therefore, a
−1
b =
−1 −1
(a1 b1 , . . . , an bn ) ∈ Hi , and hence Hi is a subgroup of G. Let a = (a1 , . . . , an ) ∈ G and
b := (b1 , . . . , bn ) ∈ Hi be arbitrary. Then bj = ej , for all j ̸= i, and so aj bj a−1 j = aj ej a−1
j = ej ,
−1 −1 −1
for all j ̸= i. This shows that aba = (a1 , . . . , an )(b1 , . . . , bn )(a1 , . . . , an ) ∈ Hi . Therefore, Hi
is a normal subgroup of G, for all i ∈ In .
(ii) Let a ∈ G be given. Then a = (a1 , . . . , an ), where ai ∈ Gi , ∀ i ∈ In . Let hi ∈ G be
the element whose i-th entry is ai and for j ̸= i, its j-th entry is ej ∈ Gj . In other words,
hi := (hi1 , . . . , hin ) ∈ G, where

ej , if j ̸= i,
hij :=
ai , if j = i.

Then hi ∈ Hi , for all i ∈ In , and h1 · · · hn = (a1 , . . . , an ) = a. To show uniqueness of this

expression, let a = k1 · · · kn , where ki ∈ Hi , for all i ∈ In . If kij ∈ Gj denote the j-th entry of
ki ∈ Hi , then kij = ej , for j ̸= i. Therefore,

(a1 , . . . , an ) = a = h1 · · · hn = k1 · · · kn = (k11 , . . . , knn ).

Then ai = hii , for all i ∈ In . This shows that ki = hi , for all i ∈ In . This proves uniqueness.
(iii) Let a = (a1 , . . . , an ) ∈ Hi ∩ (H1 · · · Hi−1 Hi+1 · · · Hn ). Since a ∈ Hi , we have aj =
ej , ∀ j ̸= i. Since a ∈ H1 · · · Hi−1 Hi+1 · · · Hn , we have

(2.10.11) a = h1 · · · hi−1 hi+1 · · · hn

for some hj ∈ Hj , ∀ j ̸= i. Since hj = (h1j , . . . , hnj ) ∈ Hj , we have

hkj = ek ∈ Gk , ∀ k ̸= j.

If bk denote the k-th component of the product h1 · · · hi−1 hi+1 · · · hn in G1 × · · · × Gn , then


ei , if k = i,
(2.10.12) bk =
hkk , if k ̸= i.

Comparing the j-th component of both sides of the equation (2.10.11), we have

aj = ej ∈ Gj , ∀ j ∈ In .

(iv) It follows from (ii) that G ⊆ H1 · · · Hn . Since Hi is a subgroup of G, for all i ∈ In , we

have H1 · · · Hn ⊆ G. Hence the result follows.
Lemma 2.10.13. Let G be a group. Let H, K be two normal subgroups of G such that H ∩ K = {e}
Then given any h ∈ H and k ∈ K we have hk = kh. Consequently, [H, K] = {e}.

Proof. Since H is normal in G, we have (hk)(kh)−1 = h(kh−1 k −1 ) ∈ H. Similarly, since K is

normal in G, we have (hk)(kh)−1 = (hkh−1 )k −1 ∈ K. Therefore, (hk)(kh)−1 ∈ H ∩ K = {e},
and hence hk = kh in G.
Exercise 2.10.14. Is the conclusion of the Lemma 2.10.13 still holds if we assume exactly one of
H and K is normal in G?
Lemma 2.10.15. Let G be a group. Let H and K be normal subgroups of G. Then HK is a normal
subgroup of G.
2.10. Direct Product & Direct Sum of Groups 59

Proof. Since H and K are normal in G, it follows that HK is a subgroup of G. Let a ∈ G and
h ∈ H, k ∈ K be arbitrary. Then a(hk)a−1 = (aha−1 )(aka−1 ) ∈ HK. Therefore, HK is a normal
subgroup of G.
Definition 2.10.16. Let G be a group and let H1 , . . . , Hn be normal subgroups of G. Then G
is said to be an internal direct product of H1 , . . . , Hn if every element a ∈ G can be uniquely
expressed as a = h1 · · · hn with hi ∈ Hi , for all i ∈ {1, . . . , n}.
Proposition 2.10.17. Let G = G1 × · · · × Gn be the external direct product of a finite collection of (not
necessarily distinct) groups G1 , . . . , Gn , and Hi := {(a1 , . . . , an ) ∈ G : aj = ej , ∀ j ̸= i}, for each
i ∈ In . Then G is an internal direct product of H1 , . . . , Hn , respectively.

Proof. It follows from Proposition 2.10.10 (ii) that given a ∈ G there exists ai ∈ Hi , for each
i ∈ In , such that a = a1 · · · an . To show that this expression for a is unique, let

a = a1 · · · an = b1 · · · bn ,

for some ai , bi ∈ Hi , ∀ i ∈ In . Note that each Hi is a normal subgroup of G by Proposition

2.10.10 (i), and Ki := H1 · · · Hi−1 Hi+1 · · · Hn is a normal subgroups of G by Lemma 2.10.15.
Moreover, Hi ∩ Ki = {e} by Proposition 2.10.10 (iii). Then using Lemma 2.10.13 we have

e = a−1 a = (a1 · · · an )−1 b1 · · · bn

−1
= a−1
n · · · a1 b1 · · · bn
= (a−1 −1
1 b1 ) · · · (an bn ).

Then for each i ∈ In , we have

b−1 −1 −1 −1 −1
i ai = (a1 b1 ) · · · (ai−1 bi−1 )(ai+1 bi+1 ) · · · (an bn ) ∈ Hi ∩ Ki = {e},

and hence ai = bi , for all i ∈ In . This completes the proof.

Theorem 2.10.18. Let {H1 , . . . , Hn } be a finite collection of normal subgroups of G. Let Ki :=
H1 · · · Hi−1 Hi+1 · · · Hn , ∀ i ∈ In . Then G is an internal direct product of H1 , . . . , Hn if and only
if

(i) G = H1 · · · Hn , and
(ii) Hi ∩ Ki = {e}, for all i ∈ In .

Moreover, in this case we have an isomorphism of groups G ∼

= H1 × · · · × Hn .

Proof. Suppose that G is an internal direct product of H1 , . . . , Hn , respectively. Let a ∈ G be

given. Then for each i ∈ In , there exists unique ai ∈ Hi such that a = a1 · · · an . There-
fore, G ⊆ H1 · · · Hn , and hence G = H1 · · · Hn . Let a ∈ Hi ∩ Ki . Then a ∈ Hi gives
a = e1 · · · ei−1 aei+1 · · · en , where ej ∈ Hj is the neutral element of Hj , for all j. Again,
a ∈ Ki = H1 · · · Hi−1 Hi+1 · · · Hn gives a = a1 · · · ai−1 eai+1 · · · an , where aj ∈ Hj , ∀ j ̸= i.
Then form the uniqueness of representation of a as product of elements from Hj ’s, we see that
a = e. Therefore, Hi ∩ Ki = {e}.
Conversely, suppose that (i) and (ii) holds. By (i) given a ∈ G, there exists ai ∈ Hi , for each
i ∈ In , such that a = a1 · · · an . Suppose that for each i ∈ In , there exists bi ∈ Hi such that
a = b1 · · · bn . Then as shown in the proof of the above Proposition, we have

e = a−1 a = (a1 · · · an )−1 b1 · · · bn

−1
= a−1
n · · · a1 b1 · · · bn
= (a−1 −1
1 b1 ) · · · (an bn ).
60 Chapter 2. Group Theory

Then for each i ∈ In , we have

b−1 −1 −1 −1 −1
i ai = (a1 b1 ) · · · (ai−1 bi−1 )(ai+1 bi+1 ) · · · (an bn ) ∈ Hi ∩ Ki = {e},

and hence ai = bi , for all i ∈ In . This completes the proof.

Exercise 2.10.19. Let G be a finite group of order mn, where gcd(m, n) = 1. If H and K are
normal subgroups of G of orders m and n, respectively, show that G is isomorphic to the direct
product group H × K.

Corollary 2.10.20. If m, n ∈ Z with gcd(m, n) = 1, then Zmn ∼

= Zm × Zn .
Proposition
L Q2.10.21 (Direct Sum of Groups). Let {Gα : α ∈ Λ} be a family of groups. Let ι :
Gα ,→ Gα be a subset such that πα ◦ ι = ceα , for all but finitely many α ∈ Λ, where ceα is
α∈Λ α∈Λ L
the constant map sending all elements to the neutral element eα ∈ Gα , for all α ∈ Λ. Then Gα is a
Q α∈Λ
subgroup of Gα , called the direct sum of the family of groups {Gα : α ∈ Λ}.
α∈Λ

Exercise 2.10.22. Let G and H be cyclic groups of prime order p generated by x ∈ G and y ∈ H,
respectively. Show that G × H is an abelian group of order p2 that is not cyclic. Show that

⟨ x ⟩, ⟨ xy ⟩, ⟨ xy 2 ⟩, . . . , ⟨ xy p−1 ⟩ and ⟨ y ⟩

are all possible distinct subgroups of G × H of order p.

Exercise 2.10.23. Find the number of distinct subgroups of order p of the cyclic group Zpn ,
where p > 0 is a prime number and n ∈ N.

2.11 Group Action

Let G be a group and let X be a non-empty set.

Definition 2.11.1. A left G-action on X is a map

σ :G×X →X

satisfying the following conditions:

(i) σ(e, x) = x, ∀ x ∈ X, and

(ii) σ(b, σ(a, x)) = σ(ba, x), ∀ a, b ∈ G, x ∈ X.

For notational simplicity, we write ax for σ(a, x).

Remark 2.11.2. We can define a right G-action on X to be a map

τ :X ×G→X

satisfying the following conditions:

(i) τ (x, e) = x, ∀ x ∈ X, and

(ii) τ (τ (x, a), b) = τ (x, ab), ∀ a, b ∈ G, x ∈ X.

For notational simplicity, we write xa for τ (a, x).

2.11. Group Action 61

Example 2.11.3. (i) Given a group G and a non-empty set X, the map

σ :G×X →X

defined by
σ(a, x) = x, ∀ a ∈ G and x ∈ X,
is a left G-action on X, known as the trivial left G-action on X. Similarly, we have a trivial
right G-action τ : X × G → X on X that sends (x, a) ∈ X × G to x ∈ X.
(ii) For each integer n ≥ 2, the group Sn acts on the set In := {k ∈ N : 1 ≤ k ≤ n} by sending
(σ, i) ∈ Sn × In to σ(i) ∈ In . Clearly for σ = e ∈ Sn we have σ(i) = i, ∀ i ∈ In , and
(στ )(i) = σ(τ (i)), ∀ i ∈ In , σ, τ ∈ Sn .
(iii) Given a non-empty set X, let S(X) be the group of all symmetries on X; its elements are
bijective maps from X onto itself, and the group operation is given by composition of
maps. Then the group S(X) acts on X from the left.

(iv) Let H be a normal subgroup of a group G. For example, H = Z(G). Then the map
φ : G × H → H defined by

φ(a, h) = aha−1 , ∀ a ∈ G, h ∈ H,

is a G-action on H. Indeed, φ(e, h) = ehe−1 = h, ∀ h ∈ H, and

φ(a, φ(b, h)) = φ(a, bhb−1 ) = a(bhb−1 )a−1 = (ab)h(ab)−1 = φ(ab, h), ∀ a, b ∈ G, h ∈ H.

Lemma 2.11.4. Given a group G and a non-empty set X, there is a one-to-one correspondence between
the set of all left G-actions on X and the set of all group homomorphisms from G into the symmetric
group S(X) on X.

Proof. Let A be the set of all left G-actions on X, and let B := Hom(G, S(X)) be the set of all
group homomorphisms from G into S(X). Define a map Φ : A → B by sending a left G-action
σ : G × X → X to the map

(2.11.5) fσ : G → S(X)

that sends a ∈ G to the map

(2.11.6) fσ (a) : X → X, x 7→ σ(a, x).

We first show that fσ (a) is bijective and hence is an element of ∈ S(X). Let x, y ∈ X be such
that σ(a, x) = σ(a, y). Then we have

x = σ(e, x) = σ(a−1 , σ(a, x))

= σ(a−1 , σ(a, y))
= σ(e, y) = y.

Therefore, fσ (a) is injective. Given y ∈ X, note that x := σ(a−1 , y) ∈ X, and that

fσ (a)(x) = σ(a, x) = σ(a, σ(a−1 , y)) = σ(e, y) = y.

This shows that σa is surjective. Therefore, fσ (a) ∈ S(X), for all a ∈ G. To show fσ : G → S(X)
is a group homomorphism, note that given a, b ∈ G we have

fσ (ab)(x) = σ(ab, x) = σ(a, σ(b, x))

= fσ (a)(fσ (b)(x))
= (fσ (a) ◦ fσ (b))(x), ∀ x ∈ X,
62 Chapter 2. Group Theory

and hence fσ (ab) = fσ (a) ◦ fσ (b), ∀ a, b ∈ G. Therefore, fσ is a group homomorphism, known

as the permutation representation of G associated to the left G-action σ on X. Thus, fσ ∈ B.
Given a group homomorphism f : G → S(X), consider the map σf : G × X → X defined
by
σf (a, x) = f (a)(x), ∀ a ∈ G, x ∈ X.
We show that σf is a left G-action on X. Since f : G → S(X) is a group homomorphism,
f (e) = IdX in S(X). Therefore, σf (e, x) = f (e)(x) = x, ∀ x ∈ X. Since f : G → S(X) is a group
homomorphism, given a, b ∈ G we have f (ab) = f (a) ◦ f (b), and hence given any x ∈ X we
have

f (ab)(x) = (f (a) ◦ f (b)) (x)

⇒ σf (ab, x) = f (a)(σf (b, x))
⇒ σf (ab, x) = σf (a, σf (b, x)).

Therefore, σf is a left G-action on X. Thus we get a map Ψ : B → A defined by

Ψ(f ) = σf , ∀ f ∈ B.

It remains to check that Ψ ◦ Φ = IdA and Φ ◦ Ψ = IdB . Given a left G-action τ : G × X → X

on X, we have (Ψ ◦ Φ)(τ ) = Ψ(fτ ) = σfτ . Since

σfτ (a, x) = fτ (a)(x) = τ (a, x), ∀ (a, x) ∈ G × X,

we have (Ψ ◦ Φ)(τ ) = τ, ∀ τ ∈ A . Therefore, Ψ ◦ Φ = IdA . Conversely, given a group

homomorphism g : G → S(X), we have (Φ ◦ Ψ)(g) = Φ(σg ) = fσg . Since fσg (a) = σg (a, −) =
g(a), ∀ a ∈ G, we conclude that (Φ ◦ Ψ)(g) = g, ∀ g ∈ B. Therefore, Φ ◦ Ψ = IdB . This
completes the proof.
Definition 2.11.7 (Faithful action). A left G-action σ : G × X → X on a non-empty set X is
said to be faithful if Ker(fσ ) = {e}, where fσ : G → S(X) is the permutation representation of
G associated to σ (see (2.11.5) and (2.11.6) in Lemma 2.11.4).
Example 2.11.8. The multiplicative group R∗ := R\{0} acts on V := Rn by scalar multiplication

σ : R∗ × V → V

defined by
σ(t, (a1 , . . . , an )) := (ta1 , . . . , tan ), ∀ t ∈ R∗ , (a1 , . . . , an ) ∈ Rn .
Note that σ is a left R∗ -action on V = Rn . The permutation representation

fσ : R∗ → S(V )

associated to σ is given by sending t ∈ R∗ to the map

fσ (t) : V → V, (a1 , . . . , an ) 7→ (ta1 , . . . , tan ).

Since

Ker(fσ ) = {t ∈ R∗ : fσ (t) = IdV }

= {t ∈ R∗ : tv = v, ∀ v ∈ V }
= {1}

is trivial, we conclude that σ is a faithful left R∗ -action on V = Rn .

Example 2.11.9. Recall that Cayley’s theorem (Theorem 2.6.21) says that any group G is isomor-
phic to a subgroup of the permutation group S(G) on G. This can be explained using group
2.11. Group Action 63

action as follow. Consider the left translation map

σ :G×G→G

defined by
σ(a, x) = ax, ∀ a, x ∈ G.
Note that σ is a left G-action on itself, called the left regular action of G on itself, and the associated
permutation representation fσ : G → S(G) that sends a ∈ G to the bijective map

fσ (a) : G → G, x 7→ ax,

Then fσ is a group homomorphism with

Ker(fσ ) = {a ∈ G : fσ (a) = IdG }

= {a ∈ G : ax = x, ∀ x ∈ G}
= {eG }

is trivial, and hence σ is a faithful action.

Given a left G-action σ : G × X → X on X, we define a relation ∼σ on X by setting

(2.11.10) x ∼σ y if y = σ(a, x), for some a ∈ G.

Note that ∼σ is an equivalence relation on X (verify!). The ∼σ -equivalence class of x ∈ X is

the subset

(2.11.11) OrbG (x) := {σ(a, x) : a ∈ G} ⊆ X,

called the orbit of x under the left G-action σ on X. Note that

(i) x ∈ OrbG (x), ∀ x ∈ X, and

(ii) given x, y ∈ X, either OrbG (x) = OrbG (y) or OrbG (x) ∩ OrbG (y) = ∅.

Therefore, X is a disjoint union of distinct G-orbits of elements of X. A G-action σ : G×X → X

is said to be transitive if OrbG (x) = OrbG (y), for all x, y ∈ X. Therefore, σ is transitive if and
only if given any two elements x, y ∈ X, there exists a ∈ G such that σ(a, x) = y.
Proposition 2.11.12. Let σ : G × X → X be a left G-action on X. For each x ∈ X the subset

Gx := {a ∈ G : σ(a, x) = x}

is a subgroup of G, called the stabilizer or the isotropy subgroup of x.

Proof. Since σ(e, x) = x, e ∈ Gx . Let a, b ∈ Gx be arbitrary. Then x = σ(a, x) gives

σ(a−1 , x) = σ(a−1 , σ(a, x)) = σ(a−1 a, x) = σ(e, x) = x.

Since σ(b, x) = x, we have σ(a−1 b, x) = σ(a−1 , σ(b, x)) = σ(a−1 , x) = x. Therefore, a−1 b ∈ Gx .
Thus Gx is a subgroup of G.
Exercise 2.11.13. Let σ : G × X → X be a left G-action on X.
T If fσ : G → S(X) is the group
homomorphism induced by σ, then show that Ker(fσ ) = Gx , where Gx is the isotropy
x∈X
subgroup of x ∈ X.
Exercise 2.11.14. Let σ : G × X → X be a left G-action on X. Given x ∈ X and a ∈ G, show
that Gy = aGx a−1
T , where−1y = σ(a, x) ∈ X. Deduce that if σ is a transitive G-action on X, show
that Ker(fσ ) = aGx a .
a∈G
64 Chapter 2. Group Theory

Exercise 2.11.15. Let X be a non-empty set. Let G be a subgroup of the symmetric group S(X)
T σ∈G
on X. Given and x ∈ X we have σGx σ −1 = Gσ(x) . Deduce that if G acts transitively on
−1
X, then σGx σ = {e}.
σ∈G

Corollary 2.11.16 (Generalized Cayley’s Theorem). Let H be a subgroup of G, and let X = {aH :
a ∈ G} be the set of all distinct left cosets of H in G. Let S(X) be the symmetric group on the set X.
Then there exists a group homomorphism φ : G → S(X) such that Ker(φ) ⊆ H.

Proof. Consider the map σ : G × X → X defined by

σ(a, bH) = (ab)H, ∀ a ∈ G, bH ∈ X.

If bH = cH, for some b, c ∈ G, then given any a ∈ G, we have (ab)−1 (ac) = b−1 a−1 ac = b−1 c ∈
H. Therefore, σ is well-defined. Note that σ(e, bH) = bH, ∀ bH ∈ X, and σ(a1 , σ(a2 , bH)) =
σ(a1 , a2 bH) = (a1 a2 b)H = σ(a1 a2 , bH), for all a1 , a2 ∈ G and bH ∈ X. Therefore, σ is a left
G-action on X. Then σ give rise to the group homomorphism

fσ : G → S(X)

that sends a ∈ G to the map

σ(a, −) : X → X, x 7→ σ(a, x).

Since Ker(fσ ) ⊆ Gx , for all x ∈ X by Exercise 2.11.13, taking x = H ∈ X we see that

GH = {a ∈ G : σ(a, H) = H} = {a ∈ G : a ∈ H} = H,

and hence Ker(fσ ) ⊆ H.

Exercise 2.11.17. Let H be a subgroup of G, and let X be the set of all left cosets of H in X. Let
σ : G × X → X be the left G-action on X defined by σ(a, bH) = (ab)H, ∀ a, b ∈ G. Show that σ
is a transitive action.

Exercise 2.11.18. Let G be a group and H a subgroup of G with [G : H] = n < ∞. Show that
there is a normal subgroup K of G with K ⊆ H and [G : K] ≤ n!.
Corollary 2.11.19 (Cayley’s Theorem). Any group G is isomorphic to a subgroup of the symmetric
group S(G) on G.

Proof. Take H = {e} in Corollary 2.11.16.

Corollary 2.11.20. Let G be a finite group of order n. Let p > 0 be a smallest prime number that divides
n. If H is subgroup of G with [G : H] = p, then H is normal in G.

Proof. Let H be a subgroup of index p in G. Let X := {aH : a ∈ G} be the set of all distinct left
cosets of H in G. Then |X| = p. Let f : G → S(X) be the map that sends a ∈ G to

f (a) : X → X, bH 7→ (ab)H.

Then f is a group homomorphism. Then K := Ker(f ) ⊆ H by Corollary 2.11.16, and [G : K] =

[G : H] · [H : K] = pk, where k := [H : K]. Since |X| = [G : H] = p, the quotient group
G/K is isomorphic to a subgroup of the symmetric group Sp by first isomorphism theorem
(see Theorem 2.9.2). Then by Lagrange’s theorem pk = |G/K| divides |Sp | = p!. Then k divides
(p−1)!. Since k is a divisor of n and p is the smallest prime divisor of n, unless k = 1, any prime
divisor of k must be greater than or equal to p. But since k divides (p − 1)!, any prime divisor
of k is less than p. Thus we get a contradiction unless k = 1. Therefore, [H : K] = k = 1, and so
H = K = Ker(f ). Thus H is a normal subgroup of G.
2.11. Group Action 65

Question 2.11.21. Let G be a group. Let H be a subgroup of G of smallest prime index. Is it

necessary that H is normal in G?

Exercise 2.11.22. Let G be a finite group of order pn , for some prime number p and integer
n > 0. Show that every subgroup of G of index p is normal in G. Deduce that every group of
order p2 has a normal subgroup of order p.
Exercise 2.11.23. Let G be a non-abelian group of order 6. Show that G has a non-normal
subgroup of order 2. Use this to classify groups of order 6. (Hint: Produce a monomorphism
into S3 ).
Proposition 2.11.24. Let σ : G×X → X be a left G-action on X. Fix x ∈ X, and let G/Gx = {aGx :
a ∈ G} be the set of all distinct left cosets of Gx in G. Then the map φ : G/Gx → OrbG (x) defined by
φ(aGx ) = σ(a, x), ∀ a ∈ G, is a well-defined bijective map. Consequently, [G : Gx ] = |OrbG (x)|.

Proof. Let a, b ∈ G be such that aGx = bGx . Then a−1 b ∈ Gx , and so σ(a−1 b, x) = x. Applying
σ(a, −) both sides, we have σ(b, x) = σ(a, σ(a−1 b, x)) = σ(a, x). Therefore, the map φ is well-
defined. To show that φ is injective, suppose that σ(a, x) = σ(b, x), for some a, b ∈ G. Then
σ(a−1 b, x) = σ(a−1 , σ(b, x)) = σ(a−1 , σ(a, x)) = σ(e, x) = x. Therefore, a−1 b ∈ Gx , and hence
aGx = bGx . Thus φ is injective. To show φ is surjective, note that σ(a, x) = φ(aGx ), for all
a ∈ G. Therefore, φ is bijective.
Corollary 2.11.25 (Class Equation). Let σ : G × X → X be a left G-action on a non-empty finite set
X, and let O be a subset of X containing exactly one element from each G-orbits in X. Then we have
X
|X| = [G : Gx ].
x∈O

F
Proof. Since X = OrbG (x), the result follows from Proposition 2.11.24.
x∈O

Exercise 2.11.26. Let G be a group of order 2n, where n ≥ 1 is an odd integer. Show that G has
a normal subgroup of order n.
Exercise 2.11.27. Let G be a group. Let H be a subgroup of G such that |H| = 11 and [G : H] =
4. Show that H is a normal subgroup of G.
Exercise 2.11.28. Fix n ∈ N. Show that the map σ : GLn (R) × Rn → Rn defined by

σ(A, v) = Av, ∀ A ∈ GLn (R), v = (v1 , . . . , vn )t ∈ Rn ,

is a left GLn (R)-action on Rn . Is σ transitive? Find the set of all GLn (R)-orbits in Rn .
Exercise 2.11.29. Let σ : G × G → G be the left G-action on itself given by

σ(a, b) = aba−1 , ∀ a, b ∈ G.

If fσ : G → S(G) is the permutation representation of G associated to σ, show that Ker(fσ ) =

Z(G).
Theorem 2.11.30 (Burnside’s Theorem). Let G be a finite group acting from the left on a non-empty
finite set X. Then the number of distinct G-orbits in X is equal to

1 X
F (a),
|G|
a∈G

where F (a) = #{x ∈ X : ax = x}, the number of elements of X fixed by a.

66 Chapter 2. Group Theory

P P
Proof. Let T := {(a, x) ∈ G × X : ax = x}. Note that |T | = F (a). Also |T | = |Gx |,
a∈G x∈X
where Gx is the stabilizer of x ∈ X. Let {x1 , . . . , xn } be the subset of X consisting of exactly
one element from each of the G-orbits in X. Note that two elements x and y of X are in the
same G-orbit if and only if OrbG (x) = OrbG (y). Since |G|/|Gx | = [G : Gx ] = |OrbG (x)|, we
conclude that |Gx | = |Gy | whenever x and y are in the same G-orbit. Then we have
X X
F (a) = |T | = |Gx |
a∈G x∈X
n
X
= |OrbG (xi )||Gxi |
i=1
Xn
= |G| = n|G|,
i=1

1 P
and hence n = F (a). This completes the proof.
|G| a∈G

2.12 Conjugacy Action & Class Equations

Let G be a group. Consider the map

(2.12.1) σ : G × G → G, (a, b) 7→ aba−1 .

Note that σ is a left action of G on itself, known as the conjugation action. Given a ∈ G, its
σ-stabilizer

Ga = {g ∈ G : gag −1 = a} = {g ∈ G : ga = ag}.

is a subgroup of G, called the centralizer or the normalizer of a in G. The equivalence relation

∼σ on G induced by the conjugation action of G on itself is known as the conjugate relation on
G. An element b ∈ G is said to be a conjugate of a ∈ G if there exists g ∈ G such that b = gag −1 .
Given a ∈ G, its G-orbit

(2.12.2) OrbG (a) = {gag −1 : g ∈ G}

consists of all conjugates of a in G, and is called the conjugacy class of a in G.

Definition 2.12.3. A partition of an integer n ≥ 1 is a finite sequence of positive integers
r
P
(n1 , . . . , nr ) such that n1 ≥ · · · ≥ nr and nj = n.
j=1

Exercise 2.12.4. Fix an integer n ≥ 2. Show that the number of conjugacy classes in Sn is the
number of partitions of n.

Solution: Let C = {C1 , . . . , Ck } be the set of all distinct conjugacy classes in Sn . Let Pn be the
set of all partitions of n. Define a map t : C → Pn by sending Ci ∈ C to the cycle type of an
element of Ci , for all i. Since two elements of Sn are conjugate in Sn if and only if they have the
same cycle type by Theorem 2.5.22, the map t is well-defined and injective. Given a partition
(n1 , . . . , nr ) of n, we have a permutation σ = (1 · · · n1 ) ◦ · · · ◦ (n1 + · · · + nr−1 + 1 · · · n1 +
· · · + nr ) ∈ Sn whose cycle type is precisely (n1 , . . . , nr ). Therefore, t is surjective, and hence is
bijective, as required.

More generally, G acts on its power set X := P(G) by conjugation:

(2.12.5) σ : G × P(G) → P(G), (a, S) 7→ aSa−1 ,

2.12. Conjugacy Action & Class Equations 67

where
{aga−1 ∈ G : g ∈ S},

if S≠ ∅, and
aSa−1 :=
∅, if S = ∅.
Two non-empty subset S and T of G are said to be conjugates if there exists a ∈ G such that
T = aSa−1 . Given a subset S ⊆ G, its stabilizer

(2.12.6) NG (S) := {a ∈ G : aSa−1 = S}

for the conjugation action in (2.12.5), is a subgroup of G, known as the normalizer of S in G.

Then we have the following.
Corollary 2.12.7. Let S be a non-empty subset of G. Then the number of distinct conjugates of S in
G is the index [G : NG (S)]. In particular, the number of distinct conjugates of an element a ∈ G is
[G : CG (a)], where CG (a) is the centralizer of a in G.

Proof. Follows from Proposition 2.11.24.

Exercise 2.12.8. Let G be a group and S a non-empty subset of G. If H is the subgroup of G

generated by S, show that NG (S) = NG (H).

Note that given a ∈ G we have CG (a) = G if and only if a ∈ Z(G). Therefore, we have the
following.
Theorem 2.12.9 (Class Equation). Let G be a finite group, and let {a1 , . . . , an } be the subset of G
consisting of exactly one element from each conjugacy class that are not contained in Z(G). Then we
have
Xn
|G| = |Z(G)| + [G : CG (ai )].
i=1

Proof. Follows from Corollary 2.11.25 by taking X = G and σ to be the conjugation action of G
on itself.

Corollary 2.12.10. Let G be a group of order pn , where p > 0 is a prime number and n ∈ N. Then G
has non-trivial center.

Proof. The class equation (see Theorem 2.12.9) for the conjugacy action of G on itself gives
r
X
pn = |G| = |Z(G)| + [G : CG (ai )],
i=1

where {a1 , . . . , an } is a subset consisting of exactly one element from each conjugacy class that
are not in the center Z(G). Since CG (ai ) is a subgroup of G, by Lagrange’s theorem |CG (ai )|
divides |G| = pn , and hence its index [G : CG (ai )] = |G|/|CG (ai )| is of the form pni , for
some ni ∈ N ∪ {0}. Since ai ∈ / Z(G), we have CG (ai ) ̸= G, and so ni ≥ 1, for all i. Since
Z(G) is a subgroup of G, we have |Z(G)| ≥ 1. Then by above class equation we see that
r
|Z(G)| = pn − pni is divisible by p. Therefore, Z(G) ̸= {e}.
P
i=1

Corollary 2.12.11. Let G be a group of order p2 , where p > 0 is a prime number. Then G is isomorphic
to either Zp2 or Zp × Zp .

Proof. Since Z(G) ̸= {e} by Corollary 2.12.10, we see that G/Z(G) has order p or 1, and hence
is cyclic. Then G is abelian by Exercise 2.8.21. If G has an element of order p2 , then G is cyclic.
Suppose that G has no element of order p2 . Then every non-neutral element of G has order p.
Fix an a ∈ G \ {e}, and take b ∈ G \ ⟨ a ⟩. Then we have |⟨a, b⟩| > |⟨a⟩| = p, and hence ⟨a, b⟩ = G.
Since both a and b has order p, it follows that ⟨a⟩ × ⟨b⟩ ∼= Zp × Zp . Note that both H := ⟨ a ⟩
68 Chapter 2. Group Theory

and K := ⟨ b ⟩ are normal subgroups of G of order p. Since H ∩ K is a subgroup of both H

and K, |H ∩ K| is either p or 1 by Lagrange’s theorem (Theorem 2.7.5). If |H ∩ K| = p, then
K = H ∩ K = H, which contradicts the choice of b ∈ G \ H. Therefore, H ∩ K = {e}. Since
HK is a subgroup of G by Theorem 2.4.3 with

|H| · |K|
|HK| = = p2 = |G|
|H ∩ K|

by Lemma 2.4.7, we have G = HK. Then G ∼

= H × K by Theorem 2.10.18.
Proposition 2.12.12. Let G be a finite abelian group of order n ≥ 2. If p > 0 is a prime number
dividing n, then G has an element of order p.

Proof. We prove this by induction on n = |G|. The case n = 2 is trivial. Assume that n > 2,
and the result holds for any abelian group of order r with 2 ≤ r < n. Let a ∈ G \ {e} be
given. If ⟨ a ⟩ = G, then we are done by Proposition 2.3.14. Assume that H := ⟨ a ⟩ is a proper
non-trivial subgroup of G. Let m := ord(a). Then 1 < m < n. If p | m, then by induction
hypothesis H has an element, say b, of order p, and we are done. Assume that p ∤ m. Since G
is abelian, H is a normal subgroup of G. Then p divides the order of the quotient group G/H.
Since |G/H| = n/m < n, by induction hypothesis G/H has an element, say bH ∈ G/H, of
order p. Then bp H = (bH)p = H in G/H, and so bp ∈ H. Since H = ⟨ a ⟩ is a cyclic group
of order m, we have (bm )p = (bp )m = e. Then ord(bm ) | p. Since p is a prime number, either
bm = e or ord(bm ) = p. If bm = e, then (bH)m = bm H = eH = H, and so p = ord(bH) | m. This
contradicts our assumption that p ∤ m. Therefore, bm ̸= e, and hence ord(bm ) = p.
Theorem 2.12.13 (Cauchy). Let G be a finite group of order n. Then for each prime number p > 0
dividing n, G has an element of order p.

Proof. Fix a prime number p > 0 that divides n. The case n = 2 is trivial. Suppose that n > 2,
and the statement holds for any finite group of order r with 2 ≤ r < n. The class equation for
G associated to the conjugacy action of G on itself is given by
r
X
(2.12.14) |G| = |Z(G)| + [G : CG (ai )],
i=1

where {a1 , . . . , ar } is the subset of G consisting of exactly one element from each G-orbits of that
does not intersect Z(G). Since a ∈ Z(G) if and only if CG (a) = G, we see that |CG (ai )| < n, for
all i ∈ {1, . . . , r}. If p | |CG (ai )|, for some i ∈ {1, . . . , r}, then by induction hypothesis CG (ai ) ⊆
G has an element of order p, and we are done. Suppose that p ∤ |CG (ai )|, ∀ i ∈ {1, . . . , r}. Since
p | n = |G| and |G| = |CG (ai )|[G : CG (ai )], we see that p | [G : CG (ai )], ∀ i ∈ {1, . . . , r}. Since
Z(G) is a subgroup of G, |Z(G)| ≥ 1. Then from class equation above, we see that p divides
|Z(G)|. Since Z(G) is abelian, it contains an element of order p by Proposition 2.12.12. This
completes the proof.

As an immediate corollary, we have the following result, known as the converse of Lagrange’s
theorem for finite abelian groups.

Corollary 2.12.15. Let G be a finite abelian group of order n. Let m > 0 be an integer that divides n.
Then G has a subgroup of order m.

Proof. The cases n = 2 and m = 1 are trivial. So we assume that m > 1 and n > 2, and we
prove it by induction on n. Suppose that the statement holds for any finite abelian group of
order r with 2 ≤ r < n. Let G be an abelian group of order n. Since m > 1, there is a prime
number, say p ∈ N, such that p | m. Then m = pk, for some k ∈ N. Then by Cauchy’s theorem
(Theorem 2.12.13) G has a subgroup, say H, of order p. Since G is abelian, that H is normal in
2.12. Conjugacy Action & Class Equations 69

G. Then the quotient group G/H exists and we have 1 ≤ |G/H| = n/p < n. Since m | n, we
have n = mℓ, for some ℓ ∈ N. Then
n mℓ pkℓ
|G/H| = = = = kℓ.
p p p

Since G/H is abelian group with |G/H| < n and k | |G/H, by induction hypothesis G/H has
a subgroup, say S, of order k. Now S = K/H, for some subgroup K of G containing H by
Exercise 2.9.21. Since |K| = |S| · |H| = kp = m, that K is a required subgroup of G of order m.
This completes the proof.

2.12.1 p-group

Definition 2.12.16 (p-group). Let p ∈ N be a prime number. A group G is said to be a p-group if

every element of G has order equal to a power of p. A subgroup H of G is called a p-subgroup
of G if H is a p-group.

Example 2.12.17. D4 and K4 are 2-groups.

Proposition 2.12.18. A finite group G is a p-group if and only if |G| = pn , for some n ∈ N.

Proof. If |G| = pn , for some n ∈ N, then given a ∈ G, ord(a) | pn by Lagrange’s theorem

(Theorem 2.7.5), and hence ord(a) = pr , for some r ∈ {1, . . . , n}, since p is a prime number.
Conversely suppose that G is a finite p-group. If |G| = ̸ pn , for all n ∈ N ∪ {0}, then there
exists a prime number q ̸= p such that q | |G|. Then by Cauchy’s theorem G has an element of
order q, which is not of the form pn , for any n ∈ N. This contradicts our assumption that G is a
p-group. This completes the proof.
Proposition 2.12.19. Any finite non-trivial p-group have non-trivial center.

Proof. Let G be a p-group of order pn , for some prime number p > 0 and positive integer n > 0.
Then the class equation for the conjugacy action of G on itself gives
X
|G| = |Z(G)| + [G : CG (a)],
a∈O\Z(G)

where O is a subset of G consisting of exactly one element from each G-orbits. Since CG (a) = G
if and only if a ∈ Z(G), we see that [G : CG (a)] > 1 for all a ∈ O \ Z(G). Since |G| = pn , it
follows from Lagrange’s theorem that p divides [G : CG (a)], ∀ a ∈ O\Z(G). Then from the class
equation above we see that p divides |Z(G)|. Since |Z(G)| ≥ 1, it follows that Z(G) ̸= {e}.
Corollary 2.12.20. Let p > 0 be a prime number. Then every group of order p2 is abelian.

Proof. Let G be a group of order p2 . Then by Proposition 2.12.19 above, Z(G) ̸= {e}. Then
|Z(G)| ∈ {p, p2 } by Lagrange’s theorem. If |Z(G)| = p, then the quotient group G/Z(G) has
order p, and hence is cyclic by Corollary 2.7.8. Then G is abelian by Exercise 2.8.21, which is a
contradiction. Therefore, |Z(G)| = p2 = |G|, and hence G = Z(G). Therefore, G is abelian.
Lemma 2.12.21. Let G be a group of order pn , where p > 0 is a prime number and n ∈ N. Let X be a
non-empty finite set admitting a left G-action. Let

X0 := {x ∈ X : ax = x, ∀ a ∈ G}

be the subset of X consisting of elements with singleton G-orbits. Then |X| ≡ |X0 | (mod p). In
particular, if p ∤ |X|, there exists x ∈ X with singleton G-orbit.
70 Chapter 2. Group Theory

Proof. The class equation for the left G-action on X gives

X
|X| = |X0 | + [G : Gx ],
x∈O\X0

where O is the subset of X consisting of exactly one element from each G-orbits of X. Since
[G : Gx ] = |OrbG (x)| > 1, for all x ∈ O \ X0 , and |G| = pn , we conclude that p divides [G : Gx ],
for all x ∈ O \ X0 . Then the result follows by reducing the class equation above modulo p. If
p ∤ |X|, then |X0 | =
̸ 0 (mod p), and hence the second part follows.
Corollary 2.12.22. Let G be a finite group having a subgroup H of order pn , where p > 0 is a prime
number and n ∈ N. Then [G : H] ≡p [NG (H) : H]. In particular, if p | [G : H], then NG (H) ̸= H.

Solution: Take X = {aH : a ∈ G} to be the set of all left cosets of H in G. Then H acts on X by

σ : H × X → X, (h, aH) 7→ (ha)H.

Note that σ is a well-defined map and is a left H-action on X. Moreover the subset of X
consisting of singleton H-orbits is given by

X0 = {aH ∈ X : σ(h, aH) = aH, ∀ h ∈ H}

= {aH ∈ X : a−1 ha ∈ H, ∀ h ∈ H}
= {aH ∈ X : a ∈ NG (H)},

we have |X0 | = [NG (H) : H]. Since |X| = [G : H], the result follows from Lemma 2.12.21.

2.13 Simple Groups

Definition 2.13.1. A group is said to be simple if it has no non-trivial proper normal subgroup.
Example 2.13.2. Any group of prime order is simple (c.f. Lagrange’s theorem).
Lemma 2.13.3. A finite abelian group G is simple if and only if |G| is a prime number.

Proof. If |G| = p, for some prime number, then its only subgroups are {e} and G, and hence G
is simple in this case. To see the converse, note that if |G| is composite, then |G| = pk, for some
prime number p and an integer k > 1. Then by Cauchy’s theorem (Theorem 2.12.13) G has an
element, say a ∈ G, of order p. Since G is abelian, the cyclic subgroup H := ⟨ a ⟩ of G is normal
in G. Since 1 < |H| = p < |G|, it follows that H is a non-trivial proper normal subgroup of G.
Thus G is not simple.
Exercise 2.13.4. Let G be a finite group of order pq, where p and q are primes (not necessarily
distinct). Show that G is not simple.

Solution: If p = q, then |G| = p2 , and so G is abelian by Corollary 2.12.20. Then G is not simple
by Lemma 2.13.3. If p ̸= q, without loss of generality we assume that p > q. Then by Cauchy’s
theorem G has a subgroup, say H, of order p. To show G is not simple, it suffices to show that
H is normal. If possible suppose that there exists a ∈ G such that aHa−1 ̸= H. Since both
H and Ka := aHa−1 are subgroups of G of order p, their intersection H ∩ Ka is a subgroup
(see Lemma 2.2.14) of order 1 or p by Lagrange’s theorem (Theorem 2.7.5). Since H ̸= Ka by
assumption, |H ∩ Ka | = 1. Then the subset HKa ⊆ G has cardinality

|H| · |Ka |
|HKa | = = p2 > pq = |G|,
|H ∩ Ka |

which is a contradiction. Therefore, aHa−1 = H, ∀ a ∈ G, and hence H is normal in G.

2.14. Sylow’s Theorems 71

Exercise 2.13.5. Let G be an abelian group having finite subgroups H and K of orders m and
n, respectively. Show that G has a subgroup of order d := lcm(m, n).

Solution. Since G is abelian, both H and K are normal in G, and hence HK is a subgroup of G
of order at most |H| · |K| = mn. Since H and K are subgroups of HK, by Lagrange’s theorem
both m and n divides |HK|, and hence d := lcm(m, n) divides |HK|. Since G is abelian, so is
its subgroup HK. Then by Corollary 2.12.15 HK has a subgroup, say V of order d. Since V is
also a subgroup of G, we are done.
Exercise 2.13.6. Let G be a non-abelian group of order p3 , where p is a prime number. Show
that |Z(G)| = p.

Proof. Since G has order p3 , it has non-trivial center. Since G is non-abelian, so Z(G) ̸= G.
Then by Lagrange’s theorem Z(G) has order p or p2 . If |Z(G)| = p2 , then G/Z(G) has order
p, and hence is a cyclic group. Then G is abelian by Exercise 2.8.21, which is a contradiction.
Therefore, |Z(G)| = p.
Exercise 2.13.7. Let G be a finite abelian group. Let n ∈ N be such that n | |G|. Show that the
number of solutions of the equation xn = e in G is a multiple of n.

Solution: The set of all solutions of xn = e in G is given by

H := {a ∈ G : an = e}.

Since en = e, we see that H ̸= ∅. Let a, b ∈ H be given. Since G is abelian, we have (a−1 b)n =
(an )−1 bn = e−1 e = e, and so a−1 b ∈ H. Therefore, H is a subgroup of G. Since G is a finite
abelian group and n | |G|, by Corollary 2.12.15 G has a subgroup, say K of order n. Then by
Corollary 2.7.7 we have an = e, ∀ a ∈ K, and hence K ⊆ H. Since |K| = n, by Lagrange’s
theorem we have n | |H|.

Exercise 2.13.8. Let G be a group of order pn , where p > 0 is a prime number and n ∈ N. Let H
be a subgroup of G of order pn−1 . Show that H is normal in G.

Solution: Follows from Corollary 2.11.20.

2.14 Sylow’s Theorems

Theorem 2.14.1 (Sylow’s First Theorem). Let G be a finite group of order pr m, where p > 0 is a
prime number, r and m are positive integers, and gcd(m, p) = 1. Then for each k ∈ {1, . . . , r}, G has
a subgroup of order pk .

Proof. Since p | |G|, by Cauchy’s theorem G has a subgroup of order p, which proves the state-
ment for the case k = 1. Suppose that 1 < k ≤ r, and G has a subgroup, say H, of order pk−1 .
Since H is a finite p-subgroup of the finite group G, considering the class equation of the set
LH := {aH : a ∈ G} associated to the left translation action of H on LH we see that

[G : H] ≡p [NG (H) : H]

(see Corollary 2.12.22). Since p | [G : H] and [NG (H) : H] ≥ 1, we conclude that p | [NG (H) :
H], and hence H is a proper normal subgroup of NG (H). Then p divides the order of the
quotient group NG (H)/H. Then by Cauchy’s theorem NG (H)/H has a subgroup, say S, of
order p. By Exercise 2.9.21 we have S = K/H, for some subgroup K of NG (H) containing H.
Therefore, K is a subgroup of G of order |K| = |K/H| · |H| = p · pk−1 = pk , as required. This
completes the proof by induction on k.
72 Chapter 2. Group Theory

Exercise 2.14.2. Let G be a finite group of order pr m, where p > 0 is a prime number, r, m ∈ N
and gcd(p, m) = 1. Let H be a subgroup of G of order pk , for some k ∈ {1, . . . , r − 1}. Show
that there exists a subgroup K of G containing H such that |K| = pk+1 and H is normal in K.

Solution: Follows from the proof of Sylow’s first theorem (Theorem 2.14.1) by noting that if K
is a subgroup of NG (H) containing H, then H is a normal subgroup of K.

As an immediate corollary, we have the following generalization of Cauchy’s theorem.

Corollary 2.14.3. Let G be a finite group. If pn | |G|, for some prime number p > 0 and an integer
n ≥ 0, then G has a subgroup of order pn .
Definition 2.14.4 (Sylow p-subgroup). Let G be a finite group and p > 0 a prime number. A
subgroup P of G is said to be a Sylow p-subgroup of G if P is a maximal p-subgroup of G; i.e., P
is a p-subgroup of G that is not properly contained in any other p-subgroup of G.
Example 2.14.5. The symmetric group S3 has three Sylow 2-subgroups, namely

H1 := {e, (1 2)}, H2 := {e, (1 3)}, and H3 := {e, (2 3)},

and one Sylow 3-subgroup, namely K := {e, (1 2 3), (1 3 2)}. Therefore, a Sylow p-subgroup,
for certain prime p, need not be unique. In this case, the unique Sylow 3-subgroup K of S3 is
normal; this is not a coincidence. We shall see later that a unique Sylow p-subgroup of G must
be normal.
Proposition 2.14.6. Let G be a finite group. Then for each prime number p > 0 G has a Sylow
p-subgroup.

Proof. If p ∤ |G|, then {e} is the Sylow p-subgroup of G. If p | |G|, then there exists r ∈ N
such that |G| = pr m, for some integer m ≥ 1 with gcd(p, m) = 1 by the fundamental theorem
of arithmetic. Then G has a subgroup, say P , of order pr by Sylow’s first theorem (Theorem
2.14.1). If Q is any p-subgroup of G containing P , then |Q| = pk , for some k ∈ N and that
pr = |P | divides pk = |Q| by Lagrange’s theorem. Then r ≤ k. Since pk = |Q| divides
|G| = pr m by Lagrange’s theorem and since gcd(p, m) = 1, we must have k ≤ r, and hence
k = r. Then |P | = pr = |Q| gives P = Q. Therefore, P is a required Sylow p-subgroup of G.

Exercise 2.14.7. Let G be a finite group of order pr m, where p > 0 is a prime number, r, m ∈ N
and gcd(p, m) = 1. Let P be a subgroup of G. Show that P is a Sylow p-subgroup of G if and
only if |P | = pr .
Exercise 2.14.8. Let G be a finite group of order pr m, where p > 0 is a prime number, r, m ∈ N
and gcd(p, m) = 1. Let P be a subgroup of G. Prove the following statements.

(i) If P is a p-subgroup of G, so is aP a−1 , for all a ∈ G.

(ii) If P is a Sylow p-subgroup of G, so is aP a−1 , for all a ∈ G.

(iii) If P is the only Sylow p-subgroup of G, then P is normal in G.

Solution: (i) Follows from the fact that |aP a−1 | = |P |, for all a ∈ G.
(ii) Follows from (i) and Exercise 2.14.7.
(iii) Since P is the only Sylow p-subgroup of G, using part (ii) we have P = aP a−1 , for all
a ∈ G. Therefore, P is normal in G.

Lemma 2.14.9. Let H be a normal subgroup of a group G. If both H and G/H are p-groups, then G is
a p-group.
2.14. Sylow’s Theorems 73

Proof. Let a ∈ G be arbitrary. Since G/H is a p-group, ord(aH) = pr , for some integer r ≥ 0. If
r = 0, then a ∈ H, and then the result follows since H is a p-group. Assume that r > 0. Then
r r r r
ap H = (aH)p = H gives ap ∈ H. Since H is a p-group, ord(ap ) = pn , for some integer
r+n
n ≥ 0. Then ap = e, and hence ord(a) | pr+m . Hence the result follows. Let G be a finite
r
group of order p m, where p > 0 is a prime number, r, m ∈ N and gcd(p, m) = 1.
Exercise 2.14.10. Let G be a finite group. Let P be a Sylow p-subgroup of G. Let a ∈ G be such
that ord(a) = pr , for some integer r ≥ 0. If aP a−1 = P , show that a ∈ P .

Solution: Since aP a−1 = P , we have a ∈ NG (P ). Note that P ⊆ NG (P ). Let b ∈ NG (P ) \ P be

arbitrary. If ord(b) = pk , for some integer k ≥ 0, then considering the quotient homomorphism
π : NG (P ) → NG (P )/P we see that ord(bP ) divides ord(b) = pk . Then the cyclic subgroup ⟨ bP ⟩
of NG (P )/P has order pm , for some integer m ≥ 0. Now ⟨ bP ⟩ = K/P , for some subgroup K
of NG (P ) containing P by Exercise 2.9.21. Since b ∈ / P by assumption, P ⫋ K. Since both P
and K/P are p-groups, so is K by Lemma 2.14.9. But this contradicts the maximality of P as it
is a Sylow p-subgroup of G. Therefore, ord(b) cannot be a power of p, and hence a ∈ P .
Theorem 2.14.11 (Sylow’s Second Theorem). Let G be a finite group of order pr m, where p > 0 is a
prime number, r, m ∈ N and gcd(p, m) = 1. Then any two Sylow p-subgroups of G are conjugate, and
hence are isomorphic.

Proof. Let H and K be two Sylow p-subgroups of G. Let LH := {aH : a ∈ G} be the set of all
left cosets of H in G. Since G is finite, so is the set LH . Define a map

σ : K × LH → LH

by
σ(b, aH) = (ba)H, ∀ (b, aH) ∈ K × LH .
If aH = a′ H, for some a, a′ ∈ G, then (ba)−1 (ba′ ) = a−1 b−1 ba′ = a−1 a′ ∈ H, and hence
(ba)H = (ba′ )H, for all b ∈ K. Therefore, the map σ is well-defined. It is easy to check that σ is
a left action of K on LH . The subset of all elements of LH with singleton K-orbits is given by

LH,0 := {aH ∈ LH : baH = aH, ∀ b ∈ K}

= {aH ∈ LH : aba−1 ∈ H, ∀ b ∈ K}
= {aH ∈ LH : aKa−1 ⊆ H}
= {aH ∈ LH : aKa−1 = H},

since both H and K are finite Sylow p-subgroups of G, they have the same cardinality. Since K
is a finite p-group, considering the class equation for LH associated to the K-action σ on it, we
have
[G : H] ≡p |LH,0 |.
Since H is a Sylow p-subgroup of G, we see that p ∤ [G : H]. Therefore, |LH,0 | ≥ 1, and hence
there exists a ∈ G such that aKa−1 = H. This completes the proof.
Theorem 2.14.12 (Sylow’s Third Theorem). Let G be a finite group of order pr m, where p > 0 is a
prime number, r, m ∈ N and gcd(p, m) = 1. Let np be the number of Sylow p-subgroups of G. Then
np = 1 + kp, for some k ∈ N ∪ {0}, and np | pr m.

Proof. Let X be the set of all Sylow p-subgroups of G. Note that X ̸= ∅ by Sylow’s first theorem.
Fix a Sylow p-subgroup P ∈ X. Note that P acts on X by conjugation:

P × X → X, (a, Q) 7→ aQa−1 .

Let X0 := {Q ∈ X : aQa−1 = Q, ∀ a ∈ P }. Since aP a−1 = P, ∀ a ∈ P , we have P ∈ X0 . So

X0 ̸= ∅. Let Q ∈ X0 be arbitrary. Then aQa−1 = Q, for all a ∈ P , and so P ⊆ NG (Q). Since
74 Chapter 2. Group Theory

both P and Q are Sylow p-subgroups of G contained in NG (Q), we conclude that P and Q are
also Sylow p-subgroups of NG (Q). Then by Sylow’s second theorem (Theorem 2.14.11) P and
Q are conjugate in NG (H). So there exists a ∈ NG (Q) such that aQa−1 = P . But aQa−1 = Q,
since a ∈ NG (Q). Therefore, P = Q. Thus we see that X0 = {P } is singleton. Then by Lemma
2.12.21 we have np = |X| ≡p 1, and hence np = 1 + kp, for some k ∈ N ∪ {0}.
For the second part, we consider the conjugation action of G on X. Since any two Sylow
p-subgroups of G are conjugate by Sylow’s second theorem (Theorem 2.14.11), we see that X
has only one G-orbit. Then given P ∈ X, its stabilizer GP = {a ∈ G : aP a−1 = P } = NG (P ).
Since np = |X| = |OrbG (P )| = [G : GP ] and [G : GP ] divides |G| = pr m, we conclude that
np | pr m.

2.14.1 Simplicity of An , n ≥ 5

2.14.2 Simplicity of PSLn (C)

2.14.3 Groups of small orders

2.15 Semi-direct product

Let H and K be groups. We say that K acts on H by automorphisms if there is a group
homomorphism f : K → Aut(H), where Aut(H) is the group of all automorphisms of H. To
simplify the notation, we denote by fk the automorphism f (k) ∈ Aut(H).
On the Cartesian product H × K of H with K, we define a binary operation by setting

(2.15.1) (h1 , k1 ) · (h2 , k2 ) := (h1 fk1 (h2 ), k1 k2 ), ∀ (h1 , k1 ), (h2 , k2 ) ∈ H × K.

Note that if f is the trivial homomorphism, then fk (h) = h, ∀ h ∈ H, k ∈ K, and in that case
the above binary operation become the component-wise binary operation on the direct product
group H × K. For notational simplicity, we denote by k1 · h2 the element fk1 (h2 ) ∈ H whenever
there is no ambiguity with the homomorphism f . Given (h1 , k1 ), (h2 , k2 ), (h3 , k3 ) ∈ H × K, we
have

((h1 , k1 )(h2 , k2 )) (h3 , k3 ) = (h1 fk1 (h2 ), k1 k2 )(h3 , k3 )

= (h1 fk1 (h2 )fk1 k2 (h3 ), (k1 k2 )k3 )
= (h1 fk1 (h2 )fk1 (fk2 (h3 )), k1 (k2 k3 ))
= (h1 fk1 (h2 fk2 (h3 )), k1 (k2 k3 ))
= (h1 , k1 )(h2 fk2 (h3 ), k2 k3 )
= (h1 , k1 ) ((h2 , k2 )(h3 , k3 )) .

Therefore, the binary operation on H × K defined in (2.15.1) is associative. Note that given
(h, k) ∈ H × K, we have

(h, k)(eH , eK ) = (hfk (eH ), keK ) = (heH , k) = (h, k),

and (eH , eK )(h, k) = (eH feK (h), eK k) = (h, k),
2.16. Free Group 75

where eH ∈ H and eK ∈ K are the neutral elements of H and K, respectively. Finally, given
(h, k) ∈ H × K, we have

(h, k)(fk−1 (h−1 ), k −1 ) = (hfk (fk−1 (h−1 )), kk −1 )

= (h(fk ◦ fk−1 )(h−1 ), eK )
= (hh−1 , eK )
= (eH , eK ),
−1 −1
and (fk−1 (h ), k )(h, k) = (fk−1 (h−1 )fk−1 (h), k −1 k)
= (fk−1 (h−1 h), eK )
= (fk−1 (eH ), eK )
= (eH , eK ).

Therefore, (h, k)−1 = (fk−1 (h−1 ), k −1 ), ∀ (h, k) ∈ H × K. Therefore, the binary operation
(2.15.1) on H × K makes it a group, called the semidirect product of H with K along f , and is
denoted by H ⋊f K or simply by H ⋊ K, if there is no confusion about f .

2.15.1 Finitely Generated Abelian Groups

2.16 Free Group

2.17 Solvable & Nilpotent Groups

2.18 Linear Groups