MTH 581-582

Introduction to Abstract Algebra

D. S. Malik
Creighton University

John N. Mordeson
Creighton University

M.K. Sen
Calcutta University

c 2007
COPYRIGHT

Department of Mathematics
c
2007
by D.S. Malik. All rights reserved. No part of this book may be reproduced, stored in a retrieval system,
or transcribed, in any form or by any meanselectronic, mechanical, photocopying, recording, or otherwise
without the prior written permission of the authors. The software described in this document is furnished under
a license agreement and may be used or copied only in accordance with the terms of the agreement. It is against
the law to copy the software on any medium except as specifically allowed in the agreement.

Printed in the United States of America

This document was produced with Scientific Word.

iii

iv

Preface
This book is intended for a one-year introductory course in abstract algebra with some topics of an
advanced level. Its design is such that the book can also be used for a one-semester course. The book
contains more material than normally would be taught in a one-year course. This should give the
teacher flexibility with respect to the selection of the content and the level at which the book is to be
used. We give a rigorous treatment of the fundamentals of abstract algebra with numerous examples
to illustrate the concepts. It usually takes students some time to become comfortable with the seeming
abstractness of modern algebra. Hence we begin at a leisurely pace paying great attention to the clarity
of our proofs. The only real prerequisite for the course is the appropriate mathematical maturity of the
students. Although the material found in calculus is independent of that of abstract algebra, a year of
calculus is typically given as a prerequisite. Since many of the examples in algebra comes from matrices,
we assume that the reader has some basic knowledge of matrix theory. The book should prepare the
student for higher level mathematics courses and computer science courses. We have many problems of
varying diculty appearing after each section. We occasionally leave as an exercise the verification of
a certain point in a proof. However, we do not rely on exercises to introduce concepts which will be
needed later on in the text.
A distinguishing feature of the book is the Worked-Out Exercises which appear after every section.
These Worked-Out Exercises provide not only techniques of problem solving, but also supply additional
information to enhance the level of knowledge of the reader. The reader should study the WorkedOut Exercises that are marked with along with the chapter. Those not marked with may be
skipped during the first reading. Sprinkled throughout the book are comments dealing with the historical
development of abstract algebra.
We welcome any comments concerning the text. The comments may be forwarded to the following
e-mail addresses: [email protected] or [email protected]

D.S. Malik
J. N. Mordeson
M.K. Sen

vi

PREFACE

Contents
Preface

List of Symbols
1 Sets, Relations, and Integers
1.1 Sets . . . . . . . . . . . . .
1.2 Integers . . . . . . . . . . .
1.3 Relations . . . . . . . . . .
1.4 Functions . . . . . . . . . .
1.5 Binary Operations . . . . .

ix

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

3
. 3
. 7
. 17
. 24
. 32

2 Introduction to Groups
35
2.1 Elementary Properties of Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3 Permutation Groups
59
3.1 Permutation Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4 Subgroups and Normal Subgroups
4.1 Subgroups . . . . . . . . . . . . . . . . .
4.2 Cyclic Groups . . . . . . . . . . . . . . .
4.3 Lagranges Theorem . . . . . . . . . . .
4.4 Normal Subgroups and Quotient Groups

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

71
71
78
81
88

5 Homomorphisms and Isomorphisms of Groups

97
5.1 Homomorphisms of Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.2 Isomorphism and Correspondence Theorems . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.3 The Groups D4 and Q8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
6 Direct Product of Groups
123
6.1 External and Internal Direct Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7 Introduction to Rings
129
7.1 Basic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
7.2 Some Important Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8 Subrings, Ideals, and Homomorphisms
145
8.1 Subrings and Subfields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
8.2 Ideals and Quotient Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
8.3 Homomorphisms and Isomorphisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
9 Ring Embeddings
165
9.1 Embedding of Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
vii

viii

CONTENTS

10 Direct Sum of Rings

171
10.1 Complete Direct Sum and Direct Sum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
11 Polynomial Rings
177
11.1 Polynomial Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
12 Euclidean Domains
185
12.1 Euclidean Domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
12.2 Greatest Common Divisors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
12.3 Prime and Irreducible Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
13 Unique Factorization Domains
199
13.1 Unique Factorization Domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
13.2 Factorization of Polynomials over a UFD . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
13.3 Irreducibility of Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
14 Maximal, Prime, and Primary Ideals
213
14.1 Maximal, Prime, and Primary Ideals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
15 Modules and Vector Spaces
221
15.1 Modules and Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
16 Field Extensions
229
16.1 Algebraic Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
16.2 Splitting Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
16.3 Algebraically Closed Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
17 Multiplicity of Roots
247
17.1 Multiplicity of Roots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
18 Finite Fields
257
18.1 Finite Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
References

261

List of Symbols

A\B
(a, b)
A0
N
Z
Z#
Q
Q+
Q
R
R+
R
C
C
P(S)

 
n
i

n!
a|b
a /| b
gcd(a, b)
lcm(a,
Sn b)
Si=1 ai
aS a
n
f :AB
f (x)
D(f )
I(f )
gf
f 1
In
f (A)
f 1 (B)

belongs to
does not belong to
subset
proper subset
contains
properly contains
symmetric dierence
set dierence
ordered pair
complement of a set A
set of positive integers
set of integers
set of nonnegative integers
set of rational numbers
set of positive rational numbers
set of nonzero rational numbers
set of real numbers
set of positive real numbers
set of nonzero real numbers
set of complex numbers
set of nonzero complex numbers
power set of the set S
union of sets
intersection of sets
number of combinations of n objects
taken i at a time
n factorial
a divides b
a does not divide b
greatest common divisor of a and b
least common multiple of a and b
a1 + a2 + + an
sum of all elements of S
congruence modulo n
f is a function from a set A into a set B
image of x under f
domain of f
image of f
composition of mappings g and f
inverse of a mapping f
In = {1, 2, . . . , n}
f (A) = {f (a) | a A}, A is a set
contained in the domain of the function f
f 1 (B) = {x X | f (x) B},
where f : X Y and B Y
composition
product

ix

LIST OF SYMBOLS
Mn (R)
|X|
|G|
(a)
Zn
Z(G)
G/H
aH, Ha
aHa1
[G : H]
K4
Sn
An
Dn
hSi
hai

N(H)
C(a)
Ker f
'
Aut(G)
Inn(G)
Ga
Cl(a)
G0
G[n]
nG
C(R)
QR
Z[ n]
Z[i]
Z[i n]

Q[ n]
Q[i]
Q[i n]
hail
hair
hai
R/I
Q(R)
R[x]
deg f (x)
R[x1 , x2 , . . . , xn ]

I
radR
F/K
K(C)
[F : K]
GF(n)
G(F/K)
FG
n (x)
PF

set of all n n matrices over R

number of elements in a set X
order of the group G
order of an element a
set of integers modulo n
center of the group G
quotient group
left, right coset of a in H
aHa1 = {aha1 | h H}
index of the subgroup H in G
Klein 4-group
symmetric group on n symbols
alternating group on n symbols
dihedral group of degree n
the subgroup generated by S
the subgroup generated by a
direct sum
Normalizer of H
centralizer of a
kernel of f
isomorphism
set of all automorphisms of the group G
set of all inner automorphisms of the group G
stabilizer of a or isotropy group of a
conjugacy class of a
commutator subgroup of the group G
set of all x G with nx = 0, G is a group
nG = {nx | x G}
center of the ring R
real
quaternions
Z[ n] = {a + b n | a, b Z}, n is a
fixed positive integer
Z[i]= {a + bi | a,b Z}
Z[i n] = {a + bi n | a, b Z}, n is a
fixed
positive integer

Q[ n] = {a + b n | a, b Q}, n is a
fixed positive integer
Q[i]= {a + bi | a,b Q}
Q[i n] = {a + bi n | a, b Q}, n is a
fixed positive integer
the left ideal generated by a
the right ideal generated by a
the ideal generated by a
quotient ring
quotient field of the ring R
polynomial ring in x
degree of the polynomial f (x)
polynomial ring in n indeterminates
radical of an ideal I
Jacobson radical of a ring R
field extension
smallest subfield containing the subfield K
and the subset C of a field
degree of the field F over the field K
Galois field of n elements
Galois group of the field F over the field K
fixed field of the group G
nth cyclotomic polynomial
plane of the field F

1
Bn
x
Kn
I(V )

l
grl
grel
multidegf
LC(f )
LM(f )
LT(f)

set of all binary n-tuples

n
1
x
1 xn
ane space over the field K
ideal of the variety V
total ordering
lexicographic order
graded lexicographic order
graded reverse lexicographic order
multidegree of the polynomial f
leading coecient of the polynomial f
leading monomial of the polynomial f
leading term of the polynomial f
end of proof

LIST OF SYMBOLS

Chapter 1

Sets, Relations, and Integers

The purpose of this introductory chapter is mainly to review briefly some familiar properties of sets,
functions, and number theory. Although most of these properties are familiar to the reader, there are
certain concepts and results which are basic to the understanding of the body of the text.
This chapter is also used to set down the conventions and notations to be used throughout the book.
Sets will always be denoted by capital letters. For example, we use the notation N for the set of positive
integers, Z for the set of integers, Z# for the set of nonnegative integers, E for the set of even integers,
Q for the set of rational numbers, Q+ for the set of positive rational numbers, Q for the set of nonzero
rational numbers, R for the set of real numbers, R+ for the set of positive real numbers, R for the
set of nonzero real numbers, C for the set of complex numbers, and C for the set of nonzero complex
numbers.

1.1

Sets

We will not attempt to give an axiomatic treatment of set theory. Rather we use an intuitive approach
to the subject. Consequently, we think of a set as some given collection of objects. A set S with only a
finite number of elements is called a finite set; otherwise S is called an infinite set. We let |S| denote
the number of elements of S. We quite often denote a finite set by a listing of its elements within braces.
For example, {1, 2, 3} is the set consisting of the objects 1, 2, 3. This technique is sometimes used for
infinite sets. For instance, the set of positive integers N may be denoted by {1, 2, 3, . . .}.
Given a set S, we use the notation x S and x
/ S to mean x is a member of S and x is not a
member of S, respectively. For the set S = {1, 2, 3}, we have 1 S and 4
/ S.
A set A is said to be a subset of a set S if every element of A is an element of S. In this case,
we write A S and say that A is contained in S. If A S, but A 6= S, then we write A S and
say that A is properly contained in S or that A is a proper subset of S. As an example, we have
{1, 2, 3} {1, 2, 3} and {1, 2} {1, 2, 3}.
Let A and B be sets. If every member of A is a member of B and every member of B is a member
of A, then we say that A and B are the same or equal. In this case, we write A = B. It is immediate
that A = B if and only if A B and B A. Thus, we have the following theorem.
Theorem 1.1.1 Let A and B be sets. Then A = B if and only if A B and B A.
The null set or empty set is the set with no elements. We usually denote the empty set by .
For any set A, we have A. The later inclusion follows vacuously. That is, every element of is an
element of A since has no elements.
We also describe sets in the following manner. Given a set S, the notation
A = {x | x S, P (x)}
or
A = {x S | P (x)}
3

1. Sets, Relations, and Integers

means that A is the set of all elements x of S such that x satisfies the property P. For example,
N = {x | x Z, x > 0}.
We can combine sets in several ways.
Definition 1.1.2 The union of two sets A and B, written A B, is defined to be the set
A B = {x | x A or x B}.
In the above definition, we mean x is a member of A or x is a member of B or x is a member of both
A and B.
Definition 1.1.3 The intersection of two sets A and B, written A B, is defined to be the set
A B = {x | x A and x B}.
Here x is an element of A B if and only if x is a member of A and at the same time x is a member
of B.
Let A and B be sets. By the definition of the union of sets, every element of A is an element of
A B. That is, A A B. Similarly, every element of B is also an element of A B and so B A B.
Also, by the definition of the intersection of sets, every element of A B is an element of A and also an
element of B. Hence, A B A and A B B. We record these results in the following theorem.
Theorem 1.1.4 Let A and B be sets. Then the following statements hold:
(i) A A B and B A B.
(ii) A B A and A B B.
The union and intersection of two sets A and B is described pictorially in the following diagrams.
The shaded area represents the set in question.

U
A

AB

B
AB

Two sets A and B are said to be disjoint if A B = .

Example 1.1.5 Let A be the set {1, 2, 3, 4} and B be the set {3, 4, 5, 6}. Then
A B = {1, 2, 3, 4, 5, 6}
and A B = {3, 4}. If C is the set {5, 6}, then
A C = {1, 2, 3, 4, 5, 6}
while A C = .
Now that the union and intersection have been defined for two sets, these operations can be similarly
defined for any finite number of sets. That is, suppose that A1 , A2 , . . . , An are n sets. The union of
A1 , A2 , . . . , An , denoted by ni=1 Ai or A1 A2 An , is the set of all elements x such that x is
an element of some Ai , where 1 i n. The intersection of A1 , A2 , . . . , An , denoted by ni=1 Ai or
A1 A2 An , is the set of all elements x such that x Ai for all i, 1 i n.
We say that a set I is an index set for a collection of sets A if for any I, there exists a set
A A and A = {A | I}. I can be any nonempty set, finite or infinite.
The union of the sets A , I, is defined to be the set {x | x A for at least one I} and is
denoted by I A . The intersection of the sets A , I, is defined to be the set {x | x A for all
I} and is denoted by I A .

1.1. Sets

Definition 1.1.6 Given two sets A and B, the relative complement of B in A, denoted by the set
dierence A\B, is the set
A\B = {x | x A, but x
/ B}.
The following diagram describes the set dierence of two sets.

U
A

B
A\B

Example 1.1.7 Let A = {1, 2, 3, 4} and B = {3, 4, 5, 6}. Then A\B = {1, 2}.
We now define a concept which is a building block for all of mathematics, namely, the concept of an
ordered pair.
Definition 1.1.8 Let A and B be nonempty sets and x A, y B.
(i) The ordered pair (x, y) is defined to be the set {{x}, {x, y}}.
(ii) The Cartesian cross product (Cartesian product) of A and B, written A B, is defined
to be the set
A B = {(x, y) | x A, y B}.
Let (x, y), (z, w) A B. We claim that (x, y) = (z, w) if and only if x = z and y = w. First suppose
that x = z and y = w. Then {{x}, {x, y}} = {{z}, {z, w}} and so (x, y) = (z, w). Now suppose that
(x, y) = (z, w). Then
{{x}, {x, y}} = {{z}, {z, w}}.
Since {x} {{x}, {x, y}}, it follows that {x} {{z}, {z, w}}. This implies that {x} = {z} or {x} =
{z, w}. If {x} = {z}, then we must have {x, y} = {z, w}. From this, it follows that x = z and y = w. If
{x} = {z, w}, then we must have {x, y} = {z}. This implies that x = z = w and x = y = z. Thus, in
this case, x = y = z = w. This establishes our claim.
It now follows that if A has m elements and B has n elements, then A B has mn elements.
Example 1.1.9 Let A = {1, 2, 3} and B = {3, 4}. Then
A B = {(1, 3), (1, 4), (2, 3), (2, 4), (3, 3), (3, 4)}.
For the set R of real numbers, the Cartesian product R R is merely the Euclidean plane.
Definition 1.1.10 For any set X, the power set of X, written P(X), is defined to be the set {A | A is
a subset of X}.
Example 1.1.11 Let X = {1, 2, 3}. Then
P(X) = {, {1}, {2}, {3}, {1, 2}, {1, 3}, {2, 3}, {1, 2, 3}}.

Here P(X) has 23 elements.

Remark 1.1.12 Let P and Q be statements. Throughout the text we will encounter questions in which
we will be asked to show that P if and only if Q; that is, show that statement P is true if and only
if statement Q is true. In situations like this, we first assume that statement P is true and show that
statement Q is true. Then we assume that statement Q is true and show that statement P is true. The
statement P if and only if Q is also equivalent to the statement: if P, then Q, and if Q, then P. For
example, see Worked-Out Exercise 1, below.

1. Sets, Relations, and Integers

Worked-Out Exercises
Exercise 1 Prove for sets A and B that A B if and only if A B = B.
Solution First suppose A B. We now show that A B = B. Let x be any element of A B. Then either
x A or x B. This implies that x B since A B. Thus, we find that every element of A B
is an element of B and so A B B. Also, B A B by Theorem 1.1.4(i). Hence, A B = B.

Conversely, suppose A B = B. Now by Theorem 1.1.4(i), A A B. Since A B = B, it now

follows that A B.

Exercise 2 For a subset A of a set S, let A0 denote the subset S\A. A0 is called the complement of A in S.
Let A and B be subsets of S. Prove that (A B)0 = A0 B 0 , DeMorgans law.
Solution First we show that (A B)0 A0 B 0 . Then we show that A0 B 0 (A B)0 . The result then
follows by Theorem 1.1.1.
Let x be any element of (A B)0 . Now (A B)0 = S\(A B) and so x S and x
/ A B. Also,
x
/ A B implies that either x
/ A or x
/ B. If x S and x
/ A, then x A0 , and if x S and
x
/ B, then x B 0 . Thus, either x A0 or x B 0 , i.e., x A0 B 0 . Hence, (A B)0 A0 B 0 .

Let us now show that A0 B 0 (A B)0 . Suppose x is any element of A0 B 0 . Then either x A0
or x B 0 . Suppose x A0 , then x S and x
/ A. Since A B A and x
/ A, we must have
x
/ A B. This implies that x (A B)0 . Similarly, we can show that if x B 0 , then x
/ A B,
i.e., x (A B)0 . Hence, A0 B 0 (A B)0 . Consequently, (A B)0 = A0 B 0 .
Exercise 3 Let A, B, and C be sets. Prove that
A (B C) = (A B) (A C).
Solution

As in the previous exercise, we first show that A (B C) (A B) (A C) and then

(A B) (A C) A (B C). The result then follows by Theorem 1.1.1.

Let x be any element of A (B C). Then x A and x B C. Thus, x A and x B or x C.

If x A and x B, then x A B, and if x A and x C, then x A C. Therefore, x A B
or x A C. Hence, x (A B) (A C). This shows that A (B C) (A B) (A C).

Let us now show that (AB)(AC) A(B C). Suppose x is any element of (AB)(AC).
Then x A B or x A C. Suppose x A B, then x A and x B. Since B B C,
we have x B C. Thus, x A and x B C and so x A (B C). Similarly, if x A
and x C, then x A (B C). Hence, (A B) (A C) A (B C). Consequently,
A (B C) = (A B) (A C).

Exercises
1. Let A = {x, y, z} and B = {y, w}. Determine each of the following sets: A B, A B, A\B, B\A,
A B, and P(A).
2. Prove for sets A and B that A B if and only if A B = A.
3. Prove for sets A, B, and C that
(i) A B = B A and A B = B A,

(ii) (A B) C = A (B C) and (A B) C = A (B C),

(iii) A (B C) = (A B) (A C),

(iv) A (A B) = A,
(v) A (A B) = A.

4. If a set S has 12 elements, how many elements does P(S) have? How many of these are properly
contained in S?

1.2. Integers

5. For subsets A and B of a set S, prove DeMorgans law:

(A B)0 = A0 B 0 .
6. The symmetric dierence of two sets A and B is the set
A 4 B = (A B)\(A B).
(i) If A = {a, b, c} and B = {b, c, d, e}, find A 4 B.
(ii) Show that A 4 B = (A\B) (B\A).

7. Let A and B be finite subsets of a set S. Show that

(i) if A B = , then |A B| = |A| + |B| ,

(ii) |A\B| = |A| |A B| ,

(iii) |A B| = |A| + |B| |A B| .

8. In each of the following exercises, write the proof if the statement is true; otherwise give a counterexample. The sets A, B, and C are subsets of a set U.
(i) A (B\C) = (A B)\(A C).

(ii) A\(B C) = (A\B) C.

(iii) (A\B)0 = (B\A)0 .

(iv) A (B C) = (A B) (A C).

(v) A 4 C = B 4 C implies A = B.

1.2

Integers

Throughout abstract algebra, the set of integers provides a source of examples. In fact, many algebraic
abstractions come from the integers. An axiomatic development of the integers is not given in this text.
Instead, certain basic properties of integers are taken for granted. For example, if n and m are integers
with n < m, then there exists a positive integer t Z such that m = n + t. In this section, we review
and prove some important properties of the integers.
The proofs of many results of algebra depend on the following basic principle of the integers.
Principle of Well-Ordering: Every nonempty subset of Z# has a smallest (least) element, i.e., if
6= S Z# , then there exists x S such that x y for all y S.
Let S be a subset of Z# . Suppose that S has the following properties:
(i) n0 S, i.e., there exists an element n0 S.
(ii) For all n n0 , n Z# , if n S, then n + 1 S.
We show that the set of all integers greater than or equal to n0 is a subset of S, i.e.,
{n Z# | n n0 } S.
Let T denote the set {n Z# | n n0 }. We wish to show that T S. On the contrary, suppose
T 6 S. Then there exists a T such that a
/ S. Let T1 be the set of all elements of T that are not in S,
i.e., T1 = T \S. Since a T and a
/ S, we have a T1 . Thus, T1 is a nonempty subset of Z# . Hence, by
the principle of well-ordering, T1 has a smallest element m, say. Then m T and m
/ S. Since m T,
m n0 . If m = n0 , then m S, a contradiction. Thus, m > n0 . This implies that m 1 n0 and so
m 1 T. Now m 1
/ T1 since m is the smallest element of T1 . Since m 1 T and m 1
/ T1 , we
must have m 1 S. But then by (ii), m = (m 1) + 1 S, which is a contradiction. Hence, T S.
Thus, from the principle of well-ordering, we deduce another important property of integers. This
property is known as the principle of mathematical induction. We thus have the following theorem.

1. Sets, Relations, and Integers

Theorem 1.2.1 (Principle of Mathematical Induction) Let S Z# . Let n0 S. Suppose S satisfies either of the following conditions.
(i) For all n n0 , n Z# , if n S, then n + 1 S.
(ii) For all n0 m < n, n Z# , if m S, then n S.
Then
{n Z# | n n0 } S.
We proved, above, Theorem 1.2.1, when S satisfies (i). We leave it for the reader to prove Theorem
1.2.1 if S satisfies (ii).
We have seen the following mathematical statement in a college algebra or in a calculus course.
1 + 2 + + n =

n(n + 1)
,
2

n 1.

We now show how this statement can be proved using the principle of mathematical induction. Let
S(n) denote the above mathematical statement, i.e.,
S(n) :

1 + 2 + + n =

n(n + 1)
,
2

n 1.

This statement will be true if the left-hand side of the statement is equal to the right-hand side. Let
S = {n Z# | S(n) is true}.
That is, S is the set of all nonnegative integers n for which the statement S(n) is true. We will show
that S is the set of all positive integers. Now
1=

1 (1 + 1)
,
2

i.e., S(1) is true. Hence, 1 S. Let n be an integer such that n 1 and suppose S(n) is true, i.e., n S.
We now show that S(n + 1) is true. Now
S(n + 1) :

1 + 2 + + n + (n + 1) =

(n + 1)(n + 2)
.
2

Consider the left-hand side.

1 + 2 + + n + (n + 1) =
=

n(n+1)
+ (n
2
(n+1)(n+2)
.
2

+ 1) (since S(n) is true)

Hence, the left-hand side is equal to the right-hand side and so S(n + 1) is true. Thus, n + 1 S. Hence,
by the principle of mathematical induction, S = {n Z# | n 1}. This proves our claim, which in turn
shows that
n(n + 1)
1 + 2 + + n =
2
is true for all positive integers n.
Sometimes we use the word induction for the principle of mathematical induction.
A proof by the principle of mathematical induction consists of three steps.
Step 1: Show that n0 S, i.e., the statement S(n0 ) is true for some n0 Z# .
Step 2: Write the induction hypothesis: n is an integer such that n n0 and n S, i.e., S(n) is
true for some integer n such that n n0 (or k is an integer such that n0 k n and S(k) is true).
Step 3: Show that n + 1 S, i.e., S(n + 1) is true.

1.2. Integers

Example 1.2.2 In this example, we show that 2n + 1 2n for all n 3.

Let S(n) be the statement:
S(n) : 2n + 1 2n , n 3.
Since we want to show that S(n) is true for all n 3, as the first step of our induction, we must
verify that S(3) is true. Let n = 3. Now 2n + 1 = 2 3 + 1 = 7 and 2n = 23 = 8. Thus, for n = 3,
2n + 1 2n . This shows that S(3) is true. Suppose that 2n + 1 2n for some n 3, i.e., S(n) is true
for some n 3. Consider S(n + 1),
S(n + 1) :

2(n + 1) + 1 2n+1 .

Let us evaluate the left-hand side of S(n + 1). We have

2(n + 1) + 1 =
=

2n + 2 + 1
(2n + 1) + 2
2n + 2
since S(n) is true
2n + 2n
(since n 3, 2 2n )
n+1
2
.

Thus, S(n + 1) is true. Hence, by the principle of mathematical induction, 2n + 1 2n for all n 3.
The principle of mathematical induction is a very useful tool in mathematics. We will make use of
this result throughout the text.
We now prove the following important properties of integers with the help of the principle of wellordering.
Theorem 1.2.3 (Division Algorithm) Let x, y Z with y 6= 0. Then there exist unique integers q
and r such that x = qy + r, 0 r < |y| .
Proof. Let us first assume y > 0. Then y 1. Consider the set
S = {x uy | u Z, x uy 0}.
Since y 1, we have x ( |x|)y = x + |x| y 0 so that x ( |x|)y S. Thus, S is a nonempty set of
nonnegative integers. Hence, by the principle of well-ordering, S must have a smallest element, say, r.
Since r S, we have r 0 and r = x qy for some q Z. Then x = qy + r. We must show that r < |y| .
Suppose on the contrary that r |y| = y. Then
x (q + 1)y = (x qy) y = r y 0
so that r y S, a contradiction since r is the smallest nonnegative integer in S and r y < r. Hence,
it must be the case that r < |y| . This proves the theorem in case y > 0.
Suppose now that y < 0. Then |y| > 0. Thus, there exist integers q 0 , r such that x = q 0 |y| + r,
0 r < |y| by the above argument. Since y < 0, |y| = y. Hence, x = q 0 y + r. Let q = q 0 . Then
x = qy + r, 0 r < |y| , the desired conclusion.
The uniqueness of q and r remains to be shown. Suppose there are integers q 0 , r0 such that
x = qy + r = q 0 y + r0 ,
0 r0 < |y|, 0 r < |y| . Then

r0 r = (q q 0 )y.

Thus,
|r0 r| = |q q 0 | |y| .

Now |y| < r 0 and 0 r0 < |y| . Therefore, if we add these inequalities, we obtain
|y| < r0 r < |y| ,

10

1. Sets, Relations, and Integers

or |r0 r| < |y| . Hence, we have

0 |q q 0 | < 1.

Since q q 0 is an integer, we must have 0 = |q q 0 | . It now also follows that |r r0 | = 0. Thus, q q 0 = 0

and r r0 = 0 or q = q 0 and r = r0 . Consequently, q and r are unique.
In Theorem 1.2.3, the integer q is called the quotient of x and y on dividing x by y and the integer
r is called the remainder of x and y on dividing x by y.
The following corollary is a special case of Theorem 1.2.3.
Corollary 1.2.4 For any two integers x and y with y > 0, there exist unique integers q and r such that
x = qy + r, where 0 r < y.
Proof. By Theorem 1.2.3, there exist unique integers q and r such that x = qy+r, where 0 r < |y| .
Since y > 0, |y| = y. Hence, x = qy + r, where 0 r < y.
Definition 1.2.5 Let x, y Z with x 6= 0. Then x is said to divide y or x is a divisor (or factor)
of y, written x|y, provided there exists q Z such that y = qx. When x does not divide y, we sometimes
write x - y.
Let x, y, z be integers with x 6= 0. Suppose x|y and x|z. Then for all integers s and t, x|(sy + tz). We
ask the reader to prove this fact in Exercise 5(iii) (page 16).
Definition 1.2.6 Let x, y Z. A nonzero integer c is called a common divisor of x and y if c|x and
c|y.
Definition 1.2.7 A nonzero integer d is called a greatest common divisor (gcd) of the integers x
and y if
(i) d|x and d|y,
(ii) for all c Z if c|x and c|y, then c|d.
Let d and d0 be two greatest common divisors of integers x and y. Then d|d0 and d0 |d. Hence, there
exist integers u and v such that d0 = du and d = d0 v. Therefore, d = duv, which implies that uv = 1
since d 6= 0. Thus, either u = v = 1 or u = v = 1. Hence, d0 = d. It now follows that two dierent
gcds of x and y dier in their sign. Of the two gcds of x and y, the positive one is denoted by gcd(x, y).
For example, 2 and 2 are the greatest common divisors of 4 and 6. Hence, 2 = gcd(4, 6).
In the next theorem, we show that the gcd always exists for any two nonzero integers.
Theorem 1.2.8 Let x, y Z with either x 6= 0 or y 6= 0. Then x and y have a positive greatest common
divisor d. Moreover, there exist elements s, t Z such that d = sx + ty.
Proof. Let
S = {mx + ny | m, n Z, mx + ny > 0}.
Suppose x 6= 0. Then

x
if x > 0
x
if x < 0

1x + 0y
if x > 0
=
(1)x + 0y
if x < 0.

|x| =

Hence, |x| S and so S 6= . By the well-ordering principle, S contains a smallest positive integer, say,
d. We now show that d is the greatest common divisor of x and y.
Since d S, there exist s, t Z such that d = sx + ty. First we show that d|x and d|y. Since d 6= 0,
by the division algorithm (Theorem 1.2.3), there exist integers q and r such that
x = dq + r,

1.2. Integers

11

where 0 r < |d| = d. Thus,

r

= x dq
= x (sx + ty)q
(substituting for d)
= (1 qs)x + (qt)y.

Suppose r > 0. Then r S, which is a contradiction since d is the smallest element of S and r < d.
Thus, r = 0. This implies that x = dq and so d|x. Similarly, d|y. Hence, d satisfies (i) of Definition 1.2.7.
Suppose c|x and c|y for some integer c. Then c|(sx + ty) by Exercise 5(iii) (page 16), i.e., c|d. Thus, d
satisfies (ii) of Definition 1.2.7. Consequently, d = gcd(x, y).
Let x and y be nonzero integers. By Theorem 1.2.8, gcd(x, y) exists and if d = gcd(x, y), then there
exist integers s and t such that d = sx + ty. The integers s and t in the representation d = sx + ty are
not unique. For example, let x = 45 and y = 126. Then gcd(x, y) = 9, and 9 = 3 45 + (1) 126 =
129 45 + (46) 126.
The proof of Theorem 1.2.8 does not indicate how to find gcd(x, y) or the integers s, t. In the
following, we indicate how these integers can be found.
Let x, y Z with y 6= 0. By the division algorithm, there exist q1 , r1 Z such that
x = q1 y + r1 ,

0 r1 < |y| .

If r1 6= 0, then by the division algorithm, there exist q2 , r2 Z such that

y = q2 r1 + r2 ,

0 r2 < r1 .

If r2 6= 0, then again by the division algorithm, there exist q3 , r3 Z such that

r1 = q3 r2 + r3 ,

0 r3 < r2 .

Since r1 > r2 > r3 0, we must in a finite number of steps find integers qn , qn+1 , and rn > 0 such
that
rn2 = qn rn1 + rn , 0 < rn < rn1
rn1 = qn+1 rn + 0.
We assert that rn (the last nonzero remainder) is the greatest common divisor of x and y. Now
rn |rn1 . Since rn |rn , rn |rn1 , and rn2 = qn rn1 + rn , we have rn |rn2 by Exercise 5(iii) (page 16).
Working our way back in this fashion, we have rn |r1 and rn |r2 . Thus, rn |y since y = q2 r1 + r2 . Since
rn |y, rn |r1 , and x = q1 y + r1 , we have rn |x. Hence, rn is a common divisor of x and y. Now if c is any
common divisor of x and y, then we see that c|r1 . Since c|y and c|r1 , c|r2 . Continuing, we finally obtain
c|rn . Thus, rn = gcd(x, y).
We now find s, t Z such that gcd(x, y) = sx + ty as follows:
rn

= rn2 + rn1 (qn )

= rn2 + [rn3 + rn2 (qn1 )](qn )
= rn3 (qn ) + rn2 (1 + qn1 qn )
(simplifying).

We now substitute rn4 + rn3 (qn2 ) for rn2 . We repeat this back substitution process until we
reach rn = sx + ty for some integers s and t.
We illustrate the above procedure for finding the gcd and integers s and t with the help of the
following example.
Example 1.2.9 Consider the integers 45 and 126. Now
126 = 2 45 + 36
45 = 1 36 + 9
36 = 4 9 + 0

12

1. Sets, Relations, and Integers

Thus, 9 = gcd(45, 126). Also,
9 = 45 1 36
= 45 1 [126 2 45]
= 3 45 + (1) 126.
Here s = 3 and t = 1.
We now define prime integers and study their basic properties.

Definition 1.2.10 (i) An integer p > 1 is called prime if the only divisors of p are 1 and p.
(ii) Two integers x and y are called relatively prime if gcd(x, y) = 1.
The following theorem gives a necessary and sucient condition for two nonzero integers to be
relatively prime.
Theorem 1.2.11 Let x and y be nonzero integers. Then x and y are relatively prime if and only if
there exist s, t Z such that 1 = sx + ty.
Proof. Let x and y be relatively prime. Then gcd(x, y) = 1. By Theorem 1.2.8, there exist integers
s and t such that 1 = sx + ty.
Conversely, suppose 1 = sx + ty for some pair of integers s, t. Let d = gcd(x, y). Then d|x and d|y
and so d|(sx + ty) (by Exercise 5(iii) (page 16)) or d|1. Since d is a positive integer and d|1, d = 1. Thus,
gcd(x, y) = 1 and so x and y are relatively prime.
Theorem 1.2.12 Let x, y, z Z with x 6= 0. If x|yz and x, y are relatively prime, then x|z.
Proof. Since x and y are relatively prime, there exist s, t Z such that 1 = sx + ty by Theorem
1.2.11. Thus, z = sxz + tyz. Now x|x and by hypothesis x|yz. Thus, x|(sxz + tyz) by Exercise 5(iii)
(page 16) and so x|z.
Corollary 1.2.13 Let x, y, p Z with p a prime. If p|xy, then either p|x or p|y.
Proof. If p|x, then we have the desired result. Suppose that p does not divide x. Since the only
positive divisors of p are 1 and p, we must have that p and x are relatively prime. Thus, p|y by Theorem
1.2.12.
The following corollary is a generalization of Corollary 1.2.13.
Corollary 1.2.14 Let x1 , x2 , . . . , xn , p Z with p a prime. If
p|x1 x2 xn ,
then p|xi for some i, 1 i n.
Proof. The proof follows by Corollary 1.2.13 and induction.
Consider the integer 24. We can write 24 = 23 3. That is, 24 can be written as product of prime
powers. Similarly, 49500 = 22 32 53 11. In the next theorem, called the fundamental theorem of
arithmetic, we prove that any positive integer can be written as product of prime powers.
Theorem 1.2.15 (Fundamental Theorem of Arithmetic) Any integer n > 1 has a unique factorization (up to order)
n = pe11 pe22 pess ,
(1.1)
where p1 , p2 , . . . , ps are distinct primes and e1 , e2 , . . . , es are positive integers.

1.2. Integers

13

Proof. First we show that any integer n > 1 has a factorization like Eq. (1.1) and then we show the
uniqueness of the factorization.
We show the existence of the factorization by induction. If n = 2, then clearly n has the above
factorization as a product of prime powers. Make the induction hypothesis that any integer k such that
2 k < n has a factorization like Eq. (1.1). If n is prime, then n already has the above factorization
as a product of prime powers, namely n itself. If n is not prime, then n = xy for integers x, y, with
1 < x < n and 1 < y < n. By the induction hypothesis, there exist primes q1 , q2 , . . . , qk , q10 , q20 , . . . , qt0
and positive integers e1 , e2 , . . . , ek , e01 , e02 , . . . , e0t such that q1 , q2 , . . . , qk are distinct primes, q10 , q20 , . . . ,
qt0 are distinct primes and
x = q1e1 q2e2 qkek
0e0 0e0
0e0
y = q1 1 q2 2 qt t .
Thus,
0e0 0e0

0e0

n = q1e1 q2e2 qkek q1 1 q2 2 qt t ,

0

0e0j

i.e., n can be factored as a product of prime powers. If qi = qj for some i and j, then we replace qiei qj
ei +e0

by qi j . It now follows that n = pe11 pe22 pess , where p1 , p2 , . . . , ps are distinct primes and e1 , e2 , . . . ,
es are positive integers. Hence, by induction, any integer n > 1 has a factorization like (1.1).
We now prove the uniqueness property by induction also. If n = 2, then clearly n has a unique
factorization as a product of prime powers. Suppose the uniqueness property holds for all integers k
such that 2 k < n. Let
(1.2)
n = pe11 pe22 pess = q1c1 q2c2 qtct
be two factorizations of n into a product of prime powers. Suppose n is prime. Then in Eq. (1.2), we
must have s = t = 1 and e1 = 1 = c1 since the only positive divisors of n are 1 and n itself. This implies
that n = p1 = q1 and so the factorization is unique.
Suppose n is not a prime. Now p1 |n and
n
= p1e1 1 pe22 pess
p1
is an integer. If s = 1, then n = pe11 and since n is not a prime, we have e1 > 1. Hence, pn1 = pe11 1 2.
If s > 1, then pn1 = p1e1 1 pe22 pess 2. Thus, in either case, pn1 is an integer 2. Now p1 |n implies that
p1 |q1c1 q2c2 qtct and so by Corollary 1.2.14, p1 |qici for some i. By reordering the qi if necessary, we can
assume that i = 1. Thus, p1 |q1c1 and so by Corollary 1.2.14, p1 |q1 . Since p1 and q1 are primes, p1 = q1 .
Thus,
n
= p1e1 1 pe22 pess = pc11 1 q2c2 qtct .
(1.3)
p1
Now e1 1 = 0 if and only if c1 1 = 0. For suppose e1 1 = 0 and c1 1 > 0. Then pn1 = pe22 pess
implies that p1 |/ pn1 and pn1 = pc11 1 q2c2 qtct implies that p1 | pn1 , which is of course impossible. We can
get a similar contradiction if we assume e1 1 > 0 and c1 1 = 0.
Now pn1 is an integer and 2 pn1 < n. Hence, by the induction hypothesis, we obtain from Eq. (1.3)
that s = t, and p1 = q1 , . . . , ps = qs (without worrying about the order), and e1 1 = c1 1, e2 = c2 , . . . ,
es = cs . Hence, by induction, we have the desired uniqueness property.
Corollary 1.2.16 Any integer n < 1 has a unique factorization (up to order)
n = (1)pe11 pe22 pess ,
where p1 , p2 , . . . , ps are distinct primes and e1 , e2 , . . . , es are positive integers.
Proof. Since n < 1, n > 1. Hence, by Theorem 1.2.15, n has a unique factorization (up to
order)
n = pe11 pe22 pess ,

14

1. Sets, Relations, and Integers

where p1 , p2 , . . . , ps are distinct primes and e1 , e2 , . . . , es are positive integers. Thus,

n = (1)pe11 pe22 pess ,
where p1 , . . . , ps are distinct primes and e1 , . . . , es are positive integers.
Theorem 1.2.15 says that any positive integer greater than 1 can be written as a product of prime
powers. Now we pose the obvious question: How many prime numbers are there? This is answered by
the following theorem due to Euclid.
Theorem 1.2.17 (Euclid) There are an infinite number of primes.
Proof. Let p1 , p2 , . . . , pn be a finite number of distinct primes. Set x = p1 p2 pn + 1. Since pi does
not divide 1, pi does not divide x, i = 1, 2, . . . , n. By the fundamental theorem of arithmetic, it follows
that there is some prime p such that p|x. Thus, p is distinct from p1 , p2 , . . . , pn so that we have n + 1
distinct primes. That is, for any finite set of primes we can always find one more. Thus, there must be
an infinite number of primes.
We close this section with the following definition. There are a few places in the text where we will
be making use of it.
Definition 1.2.18 Let n be a positive integer. Let (n) denote the number of positive integers m such
that m n and gcd(m, n) = 1, i.e.,
(n) = |{m N | m n and gcd(m, n) = 1}| .
(n) is called the Euler -function.
Clearly (2) = 1, (3) = 2, (4) = 2. Since 1, 5, 7, 11 are the only positive integers less than 12 and
relatively prime to 12, (12) = 4.
Pn
Let {a1 , . . . , an } Z. We use the notation i=1 ai to denote the sum of a1 , . . . , an , i.e.,
n
X
i=1

ai = a1 + + an .

P
If S is any finite subset
of Z, then
aS a denotes the sum of all elements of S. For example, if
P
S = {2, 4, 7}, then aS a = 2 + 4 + 7 = 13.

Worked-Out Exercises

Exercise 1 By the principle of mathematical induction, prove that

32n+1 + (1)n 2 0(mod 5)
for all positive integers n. (For integers a and b, a b(mod 5) means 5 divides a b.)
Solution Let S(n) be the statement
S(n) :

32n+1 + (1)n 2 0(mod 5),

n 1.

We wish to show that S(n) is true for all positive integers. We first must verify that S(1) is true
as the first step of our induction. Let n = 1. Then
32n+1 + (1)n 2 = 32+1 + (1)2 = 27 2 = 25 0(mod 5).
Thus, S(1) is true. Now suppose that S(n) is true for some positive integer n, i.e., 32n+1 +(1)n 2
0(mod 5) for some integer n 1. We now show that
S(n + 1) :

32(n+1)+1 + (1)n+1 2 0(mod 5)

1.2. Integers

15

is true. Now
32(n+1)+1 + (1)n+1 2 = 32n+1 32 (1)n 2
= 9(32n+1 + (1)n 2) (1)n 18 (1)n 2
= 9(32n+1 + (1)n 2) (1)n 20.
Since 32n+1 + (1)n 2 0(mod 5) and 20 0(mod 5), it follows that 32(n+1)+1 + (1)n+1 2
0(mod 5). This shows that S(n + 1) is true. Hence, by the principle of mathematical induction,
32n+1 + (1)n 2 0(mod 5) for all positive integers n.
Exercise 2 Let a and b be integers such that gcd(a, 4) = 2 and gcd(b, 4) = 2. Prove that gcd(a + b, 4) = 4.
Solution Since gcd(a, 4) = 2, 2|a, but 4 does not divide a. Therefore, a = 2x for some integer x such that
gcd(2, x) = 1. Similarly, b = 2y for some integer y such that gcd(2, y) = 1. Thus, x and y are both
odd integers. This implies that x + y is an even integer and so x + y = 2n for some integer n. Now
a + b = 2(x + y) = 4n. Hence, gcd(a + b, 4) = gcd(4n, 4) = 4.
Exercise 3 Let a, b, and c be integers such that gcd(a, c) = gcd(b, c) = 1. Prove that gcd(ab, c) = 1.
Solution If c = 0, then gcd(a, 0) = gcd(b, 0) = 1 implies that a = 1 and b = 1. Thus, gcd(ab, c) =
gcd(1, 0) = 1. Suppose now c 6= 0. By Theorem 1.2.8, gcd(ab, c) exists. Let d = gcd(ab, c). Also,
by Theorem 1.2.8, there exist integers x1 , y1 , x2 , y2 such that 1 = ax1 + cy1 , 1 = bx2 + cy2 . Thus,
(ax1 )(bx2 ) = (1 cy1 )(1 cy2 ) = 1 cy1 cy2 + cy1 cy2 . Hence, 1 = (ab)x1 x2 + c(y1 + y2 cy1 y2 ).
Thus, any common divisor of ab and c is also a divisor of 1. Hence, d|1. Since d > 0, d = 1.
Exercise 4 Let a, b Z with either a 6= 0 or b 6= 0. Prove that for any integer c,
gcd(a, b) = gcd(a, b) = gcd(a, b + ac).
Solution Suppose a 6= 0. Then gcd(a, b), gcd(a, b) and gcd(a, b + ac) exist. Let d = gcd(a, b). Then there
exist integers x and y such that d = ax + by = ax + (b)(y). Thus, any common divisor of
a and b is also a divisor of d. Hence, gcd(a, b)|d. Similarly, d| gcd(a, b). Since gcd(a, b) and
gcd(a, b) are positive, gcd(a, b) = gcd(a, b).
Let e = gcd(a, b + ac). Then there exist integers p and q such that e = ap + (b + ac)q = ap + bq + acq =
a(p + cq) + bq. Since d|a and d|b, d|e. Also, d = ax + by = ax + (b + ac)y acy = a(x cy) + (b + ac)y.
Since e|a and e|b + ac, e|d. Hence, e = d.
Exercise 5 Find integers x and y such that 512x + 320y = 64.
Solution
512
320
192
128

=
=
=
=

320 1 + 192
192 1 + 128
128 1 + 64
64 2 + 0.

Thus, 64 = 192 128 = 192 (320 192) = 192 2 + 320 (1) = (512 320) 2 + 320 (1) =
512 2 + 320 (3). Hence, x = 2 and y = 3.

Exercises
1. Determine gcd(90, 252). Find integers s and t such that
gcd(90, 252) = s 90 + t 252.
2. Find integers s and t such that gcd(963, 652) = s 963 + t 652.
3. Find integers s and t such that 657s + 963t = 9.

16

1. Sets, Relations, and Integers

4. Use the principle of mathematical induction to prove the following.
(i) 12 + 22 + 32 + + n2 =

n(n+1)(2n+1)
,
6
#

n = 1, 2, . . . .

(ii) 7n 1 is divisible by 6 for all n Z .

(iii) 6 7n 2 3n is divisible by 4 for all n Z# .

(iv) 52n + 3 is divisible by 4 for all n Z# .

(v) n < 2n for all n Z# .

(vi) 2n n2 , n = 4, 5, . . . .

(vii) n! 3n , n = 7, 8, . . . .
5. Let a, b, and c be three integers such that a 6= 0. Prove the following:
(i) If a|b, then a|bc for all c Z.

(ii) If b 6= 0, a|b and b|c, then a|c.

(iii) If a|b and a|c, then a|(bx + cy) for all x, y Z.

(iv) If a, b are positive integers such that a|b, then a b.

(v) If b 6= 0, a|b, and b|a, then a = b.

6. Let a, b, and c be integers. Prove that if ac 6= 0 and ac|bc, then a|b.

7. Let a, b, c, and d be integers such that a 6= 0 and b 6= 0. Prove that if a|c and b|d, then ab|cd.
8. Let p be a prime integer, m, n integers and r a positive integer. Suppose pr |mn and p 6 |m. Show
that pr |n.
9. Let a and b be integers and gcd(a, b) = d. If a = dm and b = dn, prove that gcd(m, n) = 1.
10. Let a, b, and c be positive integers. Prove that gcd(ab, ac) = a gcd(b, c).
11. Prove that if gcd(x, y) = gcd(x, z) = 1, then gcd(x, yz) = 1 for all x, y, z N.
12. Prove that if gcd(x, y) = 1, x|z, and y|z, then xy|z for all x, y, z N.
13. Let a, b N. Show that gcd(a, b) = gcd(a, a + b).
14. Prove that gcd(a, b) = 1 for any two positive consecutive integers a and b.
15. Let x and y be nonzero integers. The least common multiple of x and y, written lcm(x, y), is
defined to be a positive integer m such that
(i) x|m and y|m and
(ii) if x|c and y|c, then m|c.
Prove that lcm(x, y) exists and is unique.
16. Let x and y be nonzero integers. Prove that lcm(x, y) gcd(x, y) = |xy|.
17. Let x and y be nonzero integers. Show that lcm(x, y) = |xy| if and only if gcd(x, y) = 1.
18. Show that there are infinitely many prime integers of the form 6n 1, n 1.
19. Let S be a set with n elements, n 1. Show by mathematical induction that |P(S)| = 2n .
20. Determine whether the following assertions are true or false. If true, then prove it, and if false give
a counterexample.
(i) If p is a prime such that p|a5 , then p|a, where a is an integer.
(ii) If p is a prime such that p|(a2 + b2 ) and p|a, then p|b, where a and b are integers.
(iii) For any integer a, gcd(a, a + 3) = 1 or 3.
(iv) If gcd(a, 6) = 3 and gcd(b, 6) = 3, then gcd(a + b, 6) = 6, where a and b are integers.
(v) If gcd(b, c) = 1 and a|b, then gcd(a, c) = 1.

1.3. Relations

1.3

17

Relations

Some describe or define mathematics as the study of relations. Since a relation is a set of ordered pairs,
we get our first glimpse of the fundamental importance of the concept of an ordered pair.
Definition 1.3.1 A binary relation or simply a relation R from a set A into a set B is a subset of
A B.
Let R be a relation from a set A into a set B. If (x, y) R, we write xRy or R(x) = y. If xRy, then
sometimes we say that x is related to y (or y is in relation with x) with respect to R or simply x is
related to y. If A = B, then we speak of a binary relation on A.
Example 1.3.2 Let A denote the names of all states in the USA and B = Z. With each state a
in A associate an integer n which denotes the number of people in that state in the year 1996. Then
R = {(a, n) | a A and n is the number of people in state a in 1996} is a subset of A Z. Thus, R
defines a relation from A into Z.
Example 1.3.3 Consider the set of integers Z. Let R be the set of all ordered pairs (m, n) of integers
such that m < n, i.e.,
R = {(m, n) ZZ | m < n}.
Then R is a binary relation on Z.
Let R be a relation from a set A into a set B. By looking at the elements of R, we can find out
which elements of A are related to elements of B with respect to R. The elements of A that are related
to elements of B form a subset of A, called the domain of R, and the elements of B that are in relation
with elements of A form a subset of B, called the range of R. More formally, we have the following
definition.
Definition 1.3.4 Let R be a relation from a set A into a set B. Then the domain of R, denoted by
D(R), is defined to be the set
{x | x A and there exists y B such that (x, y) R}.
The range or image of R, denoted by I(R), is defined to be the set
{y | y B and there exists x A such that (x, y) R}.
Example 1.3.5 Let A = {4, 5, 7, 8, 9} and B = {16, 18, 20, 22}. Define R A B by
R = {(4, 16), (4, 20), (5, 20), (8, 16), (9, 18)}.
Then R is a relation from A into B. Here (a, b) R if and only if a divides b, where a A and
b B. Note that for the domain of R, we have D(R) = {4, 5, 8, 9} and for the range of R, we have
I(R) = {16, 18, 20}.
Example 1.3.6 Let S = {(x, y) | x, y R, x2 + y 2 = 1, y > 0}. Then S is a binary relation on R. S
is the set of points in the Euclidean plane constituting the semicircle lying above the x-axis with center
(0, 0) and radius 1.
Definition 1.3.7 Let R be a binary relation on a set A. Then R is called
(i) reflexive if for all x A, xRx,
(ii) symmetric if for all x, y A, xRy implies yRx,
(iii) transitive if for all x, y, z A, xRy and yRz imply xRz.
Definition 1.3.8 A binary relation E on a set A is called an equivalence relation on A if E is
reflexive, symmetric, and transitive.

18

1. Sets, Relations, and Integers

The important concept of an equivalence relation is due to Gauss. We will use this concept repeatedly
throughout the text.
Example 1.3.9 Let A = {1, 2, 3, 4, 5, 6} and E = {(1, 1), (2, 2), (3, 3), (4, 4), (5, 5), (6, 6), (2, 3), (3, 2)}.
Then E is an equivalence relation on A.
Example 1.3.10 (i) Let L denote the set of all straight lines in the Euclidean plane and E be the
relation on L defined by for all l1 , l2 L, (l1 , l2 ) E if and only if l1 and l2 are parallel. Then E is an
equivalence relation on L.
(ii) Let L be defined as in (i) and P be the relation defined on L by for all l1 , l2 L, (l1 , l2 ) P if and
only if l1 and l2 are perpendicular. Let l be a line in L. Since l cannot be perpendicular to itself, (l, l)
/ P.
Hence, P is not reflexive and so P is not an equivalence relation on L. Also, P is not transitive.
Example 1.3.11 Let n be a fixed positive integer in Z. Define the relation n on Z by for all x, y Z,
x n y if and only if n|(x y), i.e., x y = nk for some k Z. We now show that n is an equivalence
relation on Z.
(i) For all x Z, x x = 0 = 0n. Hence, for all x Z, x n x. Thus, n is reflexive.
(ii) Let x, y Z. Suppose x n y. Then there exists q Z such that qn = x y. Thus, (q)n = y x
and so n|(y x), i.e., y n x. Hence, n is symmetric.
(iii) Let x, y, z Z. Suppose x n y and y n z. Then there exist q, r Z such that qn = x y and
rn = y z. Thus, (q + r)n = x z and q + r Z. This implies that x n z. Hence, n is transitive.
Consequently, n is an equivalence relation on Z.
The equivalence relation, n , as defined in Example 1.3.11 is called congruence modulo n. (Another commonly used notation for x n y is x y(mod n).)
Definition 1.3.12 Let E be an equivalence relation on a set A. For all x A, let [x] denote the set
[x] = {y A | yEx}.
The set [x] is called the equivalence class (with respect to E) determined by x.
In the following theorem, we prove some basic properties of equivalence classes.
Theorem 1.3.13 Let E be an equivalence relation on the set A. Then
(i) for all x A, [x] 6= ,
(ii) if y [x], then [x] = [y], where x, y A,
(iii) for all x, y A, either [x] = [y] or [x] [y] = ,
(iv) A = xA [x], i.e., A is the union of all equivalence classes with respect to E.
Proof. (i) Let x A. Since E is reflexive, xEx. Hence, x [x] and so [x] 6= .
(ii) Let y [x]. Then yEx and by the symmetric property of E, xEy. In order to show that [x] = [y],
we will show that [x] [y] and [y] [x]. The result then will follow by Theorem 1.1.1. Let u [y].
Then uEy. Since uEy and yEx, the transitivity of E implies that uEx. Hence, u [x]. Thus, [y] [x].
Now let u [x]. Then uEx. Since uEx and xEy, uEy by transitivity and so u [y]. Hence, [x] [y].
Consequently, [x] = [y].
(iii) Let x, y A. Suppose [x] [y] 6= . Then there exists u [x] [y]. Thus, u [x] and u [y],
i.e., uEx and uEy. Since E is symmetric and uEy, we have yEu. Now yEu and uEx and so by the
transitivity of E, yEx. This implies that y [x]. Hence, by (ii), [y] = [x].
(iv) Let x A. Then x [x] xA [x]. Thus, A xA [x]. Also, xA [x] A. Hence, A = xA [x].
One of the main objectives of this section is to study the relationship between an equivalence relation
and a partition of a set. We now focus our attention to partitions. We begin with the following definition.

1.3. Relations

19

Definition 1.3.14 Let A be a set and P be a collection of nonempty subsets of A. Then P is called a
partition of A if the following properties are satisfied:
(i) for all B, C P, either B = C or B C = .
(ii) A = BP B.
In other words, if P is a partition of A, then (i) B A for all B P, i.e., every element of P is a
subset of A, (ii) distinct elements of P are either equal or disjoint, and (iii) the union of the members of
P is A.
Example 1.3.15 (i) Let A = {1, 2, 3, 4, 5, 6}. Let A1 = {1}, A2 = {2, 4, 6}, and A3 = {3, 5}. Now
A = A1 A2 A3 , A1 A2 = , A1 A3 = , and A2 A3 = . Hence, P = {A1 , A2 , A3 } is a partition
of A.
(ii) Consider Z. Let A be the set of all even integers and B be the set of all odd integers. Then
A B = and A B = Z. Thus, {A, B} is a partition of Z.
The following theorem is immediate from Theorem 1.3.13.
Theorem 1.3.16 Let E be an equivalence relation on the set A. Then
P = {[x] | x A}
is a partition of A.
Example 1.3.17 Consider the equivalence relation n on Z as defined in Example 1.3.11. Let Zn =
{[x] | x Z}. By Theorem 1.3.16, Zn is a partition of Z. Suppose n = 6. We claim that
Z6 = {[0], [1], [2], [3], [4], [5]}
and
[i] = {0 + i, 6 + i, 12 + i, . . .} = {6q + i | q Z} for all i Z.
Let 0 n < m < 6. Suppose [n] = [m]. Then m [n] and so 6|(m n). This is a contradiction since
0 < m n < 6. Hence, the equivalence classes [0], [1], [2], [3], [4], [5] are distinct. We now show that these
are the only distinct equivalence classes.
Let k be any integer. By the division algorithm, k = 6q + r for some integers q and r such that
0 r < 6. Thus, k r = 6q and so 6|(k r). This implies that k 6 r and so [k] = [r]. Since 0 r < 6
we have [r] {[0], [1], [2], [3], [4], [5]} and so [k] {[0], [1], [2], [3], [4], [5]}. This proves our first claim.
Let i Z. Then x [i] if and only if 6|(x i) if and only if 6q = x i for some q Z if and only
if x = 6q + i for some q Z. This proves our second claim. It now follows that for all i = 0, 1, . . . , 5,
[i] = [6q + i] for all q Z. Hence,
for
for
for
for
for
for

i = 0,
i = 1,
i = 2,
i = 3,
i = 4,
i = 5,

[0] = [6] = [12] = = [6] = [12] = ;

[1] = [7] = [13] = = [5] = [11] = ;
[2] = [8] = [14] = = [4] = [10] = ;
[3] = [9] = [15] = = [3] = [9] = ;
[4] = [10] = [16] = = [2] = [8] = ;
[5] = [11] = [17] = = [1] = [7] = .

By Theorem 1.3.16, given an equivalence relation E on a set A, the set of all equivalence classes forms
a partition of A. We now prove that corresponding to any partition, we can associate an equivalence
relation.
Theorem 1.3.18 Let P be a partition of the set A. Define a relation E on A by for all x, y A, xEy
if there exists B P such that x, y B. Then E is an equivalence relation on A and the equivalence
classes are precisely the elements of P.

20

1. Sets, Relations, and Integers

Proof. Note that if two elements x and y of A are related, i.e., xEy, then x and y must belong to
the same member of P. Also, if B P, then any two elements of B are related, i.e., xEy for all x, y B.
We now prove the result.
Since P is a partition of A, A = BP B. First we show that E is reflexive. Let x be any element
of A. Then there exists B P such that x B. Since x, x B, we have xEx. Hence, E is reflexive.
We now show that E is symmetric. Let xEy. Then x, y B for some B P. Thus, y, x B and so
yEx. Hence, E is symmetric. We now establish the transitivity of E. Let x, y, z A. Suppose xEy and
yEz. Then x, y B and y, z C for some B, C P. Since y B C, B C 6= . Also, since P is a
partition and B C 6= , we have B = C so that x, z B. Hence, xEz. This shows that E is transitive.
Consequently, E is an equivalence relation.
We now show that the equivalence classes determined by E are precisely the elements of P. Let
x A. Consider the equivalence class [x]. Since A = BP B, there exists B P such that x B. We
claim that [x] = B. Let u [x]. Then uEx and so u B since x B. Thus, [x] B. Also, since x B,
we have yEx for all y B and so y [x] for all y B. This implies that B [x]. Hence, [x] = B.
Finally, note that if C P, then C = [u] for all u C. Thus, the equivalence classes are precisely the
elements of P.
The relation E in Theorem 1.3.18 is called the equivalence relation on A induced by the partition P.
New relations can be constructed from existing relations. For example, given relations R and S from
a set A into a set B, we can form relations R S, R S, R\S, (A B)\R in a natural way. In all these
relations, the domain and range of the relations under consideration are subsets of A and B, respectively.
Now given a relation R from a set A into a set B and a relation S from B into a set C, there is a relation
from A into C that arises in a natural way as follows: Let us denote the new relation by T. Suppose
(a, b) R and (b, c) S. Then we make (a, c) T. Every element of T is constructed in this way. That
is, (a, c) T for some a A and c C if and only if there exists b B such that (a, b) R and
(b, c) S. This relation T is called the composition of R and S and is denoted by S R. Note that to
form the composition of R and S, we must have the domain of S and the range of R to be subsets of
the same set. More formally we have the following definition.
Definition 1.3.19 Let R be a relation from a set A into a set B and S be a relation from B into a set
C. The composition of R and S, denoted by S R, is the relation from A into C defined by
x(S R)y if there exists z B such that xRz and zSy
for all x A, y C.
Let R be a relation on a set A. Recursively, we define a relation Rn , n N, as follows:
R1
Rn

= R
= R Rn1 if n > 1.

Definition 1.3.20 Let R be a relation from a set A into a set B. The inverse of R, denoted by R1 ,
is the relation from B into A defined by
xR1 y if yRx
for all x B, y A.
The following theorem gives a necessary and sucient condition for a binary relation to be an
equivalence relation.
Theorem 1.3.21 Let R be a relation on a set A. Then R is an equivalence relation on A if and only if
(i) 4 R, where 4 = {(x, x) | x A},
(ii) R = R1 , and
(iii) R R R.

1.3. Relations

21

Proof. Suppose R is an equivalence relation. Let (x, x) 4, where x A. Since R is reflexive,

(x, x) R. Hence, 4 R, i.e., (i) holds. Let (x, y) R. Since R is symmetric, (y, x) R. Thus, by the
definition of R1 , (x, y) R1 . Hence, R R1 . On the other hand, let (x, y) R1 . Then (y, x) R.
Therefore, by the symmetric property, (x, y) R. Hence, R1 R. Thus, R = R1 , i.e., (ii) holds. We
now prove (iii). Let (x, y) R R. Then there exists z A such that (x, z) R and (z, y) R. Since R
is transitive, (x, y) R. Thus, R R R, i.e., (iii) holds.
Conversely, suppose that (i), (ii), and (iii) hold for R. For all x A, (x, x) R. Thus, R is
reflexive. Next, we show that R is symmetric. Let (x, y) R. Then by (ii), (x, y) R1 . This implies
that (y, x) R. Hence, R is symmetric. For the transitivity of R, let (x, z) R and (z, y) R. Then
(x, y) R R by the definition of composition of relations. Since R R R, (x, y) R. Hence, R is
transitive. Consequently, R is an equivalence relation.

Worked-Out Exercises
Exercise 1 In Z10 , which of the following equivalence classes are equal: [2], [5], [5], [8], [12], [15], [3], [7],
[22]?
Solution We note that [2] = [2 + 10] = [12], [8] = [8 + 10] = [2], [12] = [12 + 10] = [22], [5] =
[5 + 10] = [5] = [5 + 10] = [15] and [3] = [3 + 10] = [7]. Also, [2] 6= [5], [2] 6= [7] and [5] 6= [7].
Hence, it now follows that [2] = [12] = [8] = [22], [5] = [5] = [15] and [3] = [7].
Exercise 2 Let R be a reflexive and transitive relation on a set S. Prove that RR1 is an equivalence relation.
Solution Since (x, x) R for all x S, (x, x) R1 for all x S. Thus, (x, x) R R1 for all x S.
Hence, R R1 is reflexive. Let (x, y) R R1 . Then (x, y) R and (x, y) R1 . Thus,
(y, x) R1 and (y, x) R. Therefore, (y, x) R R1 . Hence, R R1 is symmetric. Now
suppose that (x, y), (y, z) R R1 . Then (x, y), (y, z) R and (x, y), (y, z) R1 . Since R
is transitive, (x, z) R. Now since (x, y), (y, z) R1 , (y, x), (z, y) R. Since R is transitive,
(z, x) R and so (x, z) R1 . Thus, (x, z) R R1 . Hence, R R1 is transitive. We have thus
proved that R R1 is reflexive, symmetric, and transitive and hence R R1 is an equivalence
relation.
Exercise 3 Give an example of an equivalence relation on the set S = {1, 2, 3, 4, 5, 6, 7, 8} such that R has
exactly four equivalence classes.
Solution R = {(1, 1), (2, 2), (3, 3), (4, 4), (5, 5), (6, 6), (7, 7), (8, 8), (1, 2), (2, 1), (3, 4), (4, 3), (5, 6), (6, 5),
(7, 8), (8, 7)}. The equivalence classes are [1] = [2], [3] = [4], [5] = [6], and [7] = [8].
Exercise 4 Let R1 and R2 be two symmetric relations on a set S. Prove that R1 R2 is symmetric if and only
if R1 R2 = R2 R1 .
Solution Suppose R1 R2 is symmetric. Let (x, y) be any element of R1 R2 . Then (y, x) R1 R2
since R1 R2 is symmetric. Thus, there exists z S such that (y, z) R2 and (z, x) R1
by the definition of composition of relations. Since R1 and R2 are symmetric, (z, y) R2 and
(x, z) R1 . Hence, (x, y) R2 R1 . Thus, R1 R2 R2 R1 . Similarly, R2 R1 R1 R2 . Hence,
R1 R2 = R2 R1 .
Conversely, suppose that R1 R2 = R2 R1 . Let (x, y) R1 R2 . Then (x, y) R2 R1 . Thus, there
exists z S such that (x, z) R1 and (z, y) R2 . Since R1 and R2 are symmetric, (z, x) R1
and (y, z) R2 . Hence, (y, x) R2 R1 = R1 R2 . Thus, R1 R2 is symmetric.

Exercise 5 Let A = {1, 2, 3, 4, 5} and R = {(1, 1), (2, 2), (3, 3), (4, 4), (5, 5), (1, 2), (2, 1), (4, 5), (5, 4)}. Show
that R is an equivalence relation.
Solution Let B = {1, 2}, C = {3}, and D = {4, 5}. Let P = {B, C, D}. Then P is a partition of A. Also,
note that if x, y A, then (x, y) R if and only if x, y X for some X P, i.e., the relation R
is induced by the partition P. Hence, R is an equivalence relation on A by Theorem 1.3.18.

22

1. Sets, Relations, and Integers

Exercise 6 Let X = {1, 2, 3, 4, 5, 6, 7}. Then

P = {{1, 3, 5}, {2, 6}, {4, 7}}
is a partition of X. List the elements of the corresponding equivalence relation R on X induced by
P.
Solution R = {(a, b) X X | a and b both belong to the same element of P}. Then R = {(1, 1), (2, 2),
(3, 3), (4, 4), (5, 5), (6, 6), (7, 7), (1, 3), (3, 1), (1, 5), (5, 1), (3, 5), (5, 3), (2, 6), (6, 2), (4, 7), (7, 4)}.
Exercise 7 Let R be a relation on a set S. Prove that the following conditions are equivalent.
(i) R is an equivalence relation on S.
(ii) R is reflexive and for all a, b, c S, if aRb and bRc, then cRa.
Solution (i)(ii): Suppose R is an equivalence relation on S. Then R is reflexive. Let a, b, c S. Suppose
aRb and bRc. The transitive property of R implies that aRc. Hence, cRa since R is symmetric.
(ii)(i): Since R is given to be reflexive, to show that R is an equivalence relation, we only need
to check that R is symmetric and transitive. For symmetry, suppose aRb. Since R is reflexive, we
have aRa. Now since we have aRa and aRb, bRa by hypothesis. This shows that R is symmetric.
To show that R is transitive, suppose aRb and bRc. Then by the hypothesis, cRa. Since we have
shown that R is symmetric, cRa implies that aRc. Hence, R is transitive. Consequently, R is an
equivalence relation on S.

Exercises
1. Let R be a relation on the set A = {1, 2, 3, 4, 5, 6, 7} defined by R = {(a, b) A A | 4 divides
a b}.
(i) List the elements of R.

(ii) Find the domain of R.

(iii) Find the range of R.
(iv) Find the elements of R1 .
(v) Find the domain of R1 .
(vi) Find the range of R1 .
2. Let R be a relation on the set A = {1, 2, 3, 4, 5, 6} defined by R = {(a, b) A A | a + b 9}.
(i) List the elements of R.

(ii) Is 4 R, where = {(x, x) | x A}?

(iii) Is R = R1 ?

(iv) Is R R R?
3. Which of the following relations E are equivalence relations on the set of integers Z?
(i) xEy if and only if x y is an even integer.

(ii) xEy if and only if x y is an odd integer.

(iii) xEy if and only if x y.

(iv) xEy if and only if x divides y.

(v) xEy if and only if x2 = y 2 .
(vi) xEy if and only if |x| = | y|.

(vii) xEy if and only if |x y| 2.

4. Let R = {(a, b) | a, b Q and a b Z}. Prove that R is an equivalence relation on Q.

1.3. Relations

23

5. Let A = {1, 2, 3, 4, 5, 6, 7, 8}. Define a relation R on A by

aRb if and only if 3 divides a b
for all a, b A. Show that R is an equivalence relation on A. Find the equivalence classes [1], [2],
[3], and [4].
6. Let R be an equivalence relation on a set A. Find the domain and range of R.
7. Find all equivalence relations on the set S = {a, b, c}.
8. In Z6 , which of the following equivalence classes are equal: [1], [2], [8], [5], [2], [11], [23]?
9. Let x, y Z be such that x n y, where n N. Show that for all z Z,
(i) x + z n y + z,

(ii) xz n yz

10. Let x, y, z, w Z and n be a positive integer. Suppose that x n y and z n w. Show that
x + z n y + w and xz n yw.
11. Let n be a positive integer and [x], [y] Zn . Show that the following conditions are equivalent.
(i) [x] = [y].

(ii) x y = nr for some integer r.

(iii) n|(x y).

12. (Chinese Remainder Theorem) Let m and n be positive integers such that gcd(m, n) = 1.
Prove that for any integers a and b, the congruences x m a and x n b have a common solution
in Z. Furthermore, if u and v are two solutions of these congruences, prove that u mn v.
13. Define relations R1 , R2 , R3 such that R1 is reflexive and symmetric but not transitive, R2 is
reflexive and transitive but not symmetric, and R3 is symmetric and transitive but not reflexive.
14. Prove that the intersection of two equivalence relations on a set S is an equivalence relation on S.
15. Let R be a relation on a set A. Define T (R) = R R1 {(x, x) | x A}. Show that T (R) is
reflexive and symmetric.
16. Let R be a relation on a set S. Set R = R R2 R3 . Prove the following:
(i) R is a transitive relation on S.

(ii) If T is a transitive relation on A such that R T, then R T.

(R is called the transitive closure of R.)

17. Let R1 and R2 be symmetric relations on a set S such that R1 R2 R2 R1 . Prove that R2 R1
is symmetric and R1 R2 = R2 R1 .
18. Let R1 and R2 be equivalence relations on a set S such that R1 R2 = R2 R1 . Prove that R1 R2
is an equivalence relation.
19. Let R1 and R2 be relations on a set S. Determine whether each statement is true or false. If the
statement is false, give a counterexample.
(i) If R1 and R2 are reflexive, then R1 R2 is reflexive.

(ii) If R1 and R2 are transitive, then R1 R2 is transitive.

(iii) If R1 and R2 are symmetric, then R1 R2 is symmetric.

(iv) If R1 is transitive, then R11 is transitive.

(v) If R1 is reflexive and transitive, then R1 R1 is transitive.

24

1.4

1. Sets, Relations, and Integers

Functions

Like sets, functions play a central role in mathematics. Readers may already be familiar with the notion
of a function either through a college algebra or a calculus course. In these courses, functions were
usually real valued. Throughout the text we will encounter functions which do not have to be real
valued. Functions help us study the relationship between various algebraic structures. In this section,
we review some of their basic properties. Roughly speaking, a function is a special type of correspondence
between elements of one set and those of another set. More precisely, a function is a particular set of
ordered pairs.
Definition 1.4.1 Let A and B be nonempty sets. A relation f from A into B is called a function (or
mapping) from A into B if
(i) D(f ) = A and
(ii) for all (x, y), (x0 , y 0 ) f, x = x0 implies y = y 0 .
When (ii) is satisfied by a relation f, we say that f is well defined or single-valued.
We use the notation f : A B to denote a function f from a set A into a set B. For (x, y) f, we
usually write f (x) = y and say that y is the image of x under f and x is a preimage of y under f.
Leibniz seems to be the first to have used the word function to stand for any quantity related to
a curve. Clairant (1734) originated the notation f (x) and Euler made extensive use of it. Dirichlet is
responsible for the current definition of a function.
Let us now explain the above definition. Suppose f : A B. Then f is a subset of A B such that
for all x A, there exists a unique y B such that (x, y) f. Hence, we like to think of a function
as a rule which associates to each element x of A exactly one element y of B. In order to show that a
relation f from A into B is a function, we first show that the domain of f is A and next we show that
f well defined or single-valued, i.e., if x = y in A, then f (x) = f (y) in B for all x, y A.
We now consider some examples of relations, some of which are functions and some of which are not.
Example 1.4.2 Let f be the subset of ZZ defined by
f = {(n, 2n + 3) | n Z}.
Then D(f ) = {n | n Z} =Z. We now show that f is well defined. Let n, m Z. Suppose n = m.
Then 2n + 3 = 2m + 3, i.e., f (n) = f (m). Therefore, f is well defined. Hence, f satisfies (i) and (ii) of
Definition 1.4.1 and so f is a function.
Example 1.4.3 Let A = {1, 2, 3, 4} and B = {a, b, c}. Let f be the subset of A B defined by
f = {(1, a), (2, b), (3, c), (4, b)}.
First note that D(f ) = {1, 2, 3, 4} = A and so f satisfies (i) of Definition 1.4.1. From the definition of
f, it is immediate that for all x A, there exists a unique y B such that (x, y) f. Therefore, f is
well defined and so f satisfies (ii) of Definition 1.4.1. Hence, f is a function.
Example 1.4.4 Let f be the subset of QZ defined by
p
f = {( , p) | p, q Z, q 6= 0}.
q
First we note that D(f ) = { pq | p, q Z, q 6= 0} = Q. Thus, f satisfies (i) of Definition 1.4.1. Now
( 23 , 2) f, ( 46 , 4) f and 23 = 46 . But f ( 23 ) = 2 6= 4 = f ( 46 ). Thus, f is not well defined. Hence, f is not
a function from Q into Z.
Example 1.4.5 Let f be the subset of ZZ defined by
f = {(mn, m + n) | m, n Z}.

1.4. Functions

25

First we show that f satisfies (i) of Definition 1.4.1. Let x be any element of Z. Then we can write
x = x 1. Hence, (x, x + 1) = (x 1, x + 1) f. This implies that x D(f ). Thus, Z D(f ). However,
D(f ) Z and so D(f ) = Z. Thus, f satisfies (i) of Definition 1.4.1. Now 4 Z and 4 = 4 1 = 2 2.
Thus, (4 1, 4 + 1) f and (2 2, 2 + 2) f. Hence, we find that 4 1 = 2 2 and f (4 1) = 5 6= 4 = f (2 2).
This implies that f is not well defined, i.e., f does not satisfy (ii) of Definition 1.4.1. Hence, f is not a
function from Z into Z.
We now explore the meaning of equality of two functions.
Let f : A B and g : A B be two functions. Then f and g are subsets of A B. Suppose
f = g. Let x be any element of A. Then (x, f (x)) f = g. Also, (x, g(x)) g. Since g is a function and
(x, f (x)), (x, g(x)) g, we must have g(x) = f (x). Conversely, assume that g(x) = f (x) for all x A.
Let (x, y) f. Then y = f (x) = g(x). Thus, (x, y) g. This implies that f g. Similarly, we can show
that g f. It now follows that f = g. Thus, two functions f : A B and g : A B are equal if and
only if f (x) = g(x) for all x A.
Example 1.4.6 Let f : Z Z# and g : Z Z# be defined by f = {(n, n2 ) | n Z} and g = {(n, |n|2 ) |
n Z}. Now for all n Z,
f (n) = n2 = |n|2 = g(n).
Hence, f = g.

Definition 1.4.7 Let f be a function from a set A into a set B. Then

(i) f is called one-one if for all x, x0 A, f (x) = f (x0 ) implies x = x0 .
(ii) f is called onto B (or f maps A onto B) if I(f ) = B.
We note that if f : A B, then I(f ) = B if and only if for all y B, there exists x A such that
f (x) = y. In other words, I(f ) = B if and only if every element of B has a preimage. We also note that
f is one-one if and only if every element of B has at most one preimage.
Let A be a nonempty set. The function iA : A A defined by iA (x) = x for all x A is a one-one
function of A onto A. iA is called the identity map on A.
Example 1.4.8 Consider the relation f from Z into Z defined by
f (n) = n2
for all n Z. Now D(f ) = Z. Also, if n = n0 , then n2 = (n0 )2 , i.e., f (n) = f (n0 ). Hence, f is well
defined. Thus, f is a function. Now f (1) = 1 = f (1) and 1 6= 1. This implies that f is not one-one.
Now for all n Z, f (n) is a nonnegative integer. This shows that a negative integer has no preimage.
Hence, f is not onto Z. Note that f is onto {0, 1, 4, 9, . . .}.
Example 1.4.9 Consider the relation f from Z into Z defined by for all n Z, f (n) = 2n. As in the
previous examples, we can show that f is a function. Let n, n0 Z and suppose that f (n) = f (n0 ). Then
2n = 2n0 , i.e., n = n0 . Hence, f is a one-one function. Since for all n Z, f (n) is an even integer, we
see that an odd integer has no preimage. Thus, f is not onto Z. However, we note that f is onto E.
Definition 1.4.10 Let A, B, and C be nonempty sets and f : A B and g : B C. The composition
of f and g, written g f, is the relation from A into C defined as follows:
gf

= {(x, z) | x A, z C, there exists y B

such that f (x) = y and g(y) = z}.

Let f : A B and g : B C and (x, z) g f, i.e., (g f )(x) = z. Then by the definition of

composition of functions, there exists y B such that f (x) = y and g(y) = z. Now
z = g(y) = g(f (x)).
Hence, (g f )(x) = g(f (x)).
In the following, we describe some properties of composition of functions.

26

1. Sets, Relations, and Integers

Theorem 1.4.11 Suppose that f : A B and g : B C. Then

(i) g f : A C, i.e., g f is a function from A into C.
(ii) If f and g are one-one, then g f is one-one.
(iii) If f is onto B and g is onto C, then g f is onto C.
Proof. (i) Let x A. Since f is a function and x A, there exists y B such that f (x) = y. Now
since g is a function and y B, there exists z C such that g(y) = z. Thus, (g f )(x) = g(f (x)) =
g(y) = z, i.e., (x, z) g f. Hence, x D(g f ). This shows that A D(g f ). But D(g f ) A and
so D(g f ) = A. Next, we show that g f is well defined.
Suppose that (x, z) g f, (x1 , z1 ) g f and x = x1 , where x, x1 A and z, z1 C. By the
definition of composition of functions, there exist y, y1 B such that f (x) = y, g(y) = z, f (x1 ) = y1
and g(y1 ) = z1 . Since f is a function and x = x1 , we have y = y1 . Similarly, since g is a function and
y = y1 , we have z = z1 . Thus, g f is well defined. Hence, g f is a function from A into C.
(ii) Let x, x0 A. Suppose (g f )(x) = (g f )(x0 ). Then g(f (x)) = g(f (x0 )). Since g is one-one,
f (x) = f (x0 ). Since f is one-one, x = x0 . Thus, g f is one-one.
(iii) Let z C. Then there exists y B such that g(y) = z since g is onto C. Since f is onto B,
there exists x A such that f (x) = y. Thus, (g f )(x) = g(f (x)) = g(y) = z. Hence, g f is onto C.
Example 1.4.12 Consider the function f : Z Z and g : Z E, where f (n) = n2 and g(n) = 2n for
all n Z. Then g f : Z E and (g f )(n) = g(f (n)) = g(n2 ) = 2n2 .
Theorem 1.4.13 Let f : A B, g : B C, and h : C D. Then
h (g f ) = (h g) f.
That is, composition of functions is associative.
Proof. First note that h (g f ) : A D and (h g) f : A D. Let x A. Then
[h (g f )](x) = h((g f )(x)) = h(g(f (x))) = (h g)(f (x)) = [(h g) f ](x).
Thus, by the equality of two functions, h (g f ) = (h g) f.
Let A be a set and f : A A. Recursively, we define
f 1 (x) = f (x)
f
(x) = (f f n )(x)
n+1

for all x A, n N.
Let A and B be sets. A and B are said to be equipollent, written A B, if there exists a one-one
function from A onto B, i.e., the elements of A and B are in one-one correspondence.
From Theorem 1.4.11, it follows that is an equivalence relation. If A B, then sometimes we
write |A| = |B| . It is immediate that if A and B are finite sets, then |A| = |B| if and only if A and B
have the same number of elements.
The following lemma, which follows from Theorem 1.4.11(ii), is of independent interest. We give a
direct proof of this result.
Lemma 1.4.14 Let A be a set and f : A A be a one-one function. Then f n : A A is a one-one
function for all integers n 1.
Proof. Suppose there exists n > 1 such that f n is not one-one. Let m > 1 be the smallest positive
integer such that f m is not one-one. Then there exist x, y A such that x 6= y and f m (x) = f m (y).
But then f (f m1 (x)) = f (f m1 (y)) and hence f m1 (x) = f m1 (y) since f is one-one. Now since m is
the smallest positive integer such that f m is not one-one, f m1 is one-one. Hence, x = y, which is a
contradiction. Thus, f n is one-one for all n 1.
That one-one functions on a finite set are onto is proved next.
Theorem 1.4.15 Let A be a finite set. If f : A A is one-one, then f is onto A.

1.4. Functions

27

Proof. Let y A. Now f n (y) A for all n 1. Hence,

{y, f (y), f 2 (y), . . .} A.
Since A is finite, all elements of the set {y, f (y), f 2 (y), . . .} cannot be distinct. Thus, there exist positive
integers s and t such that s > t and f s (y) = f t (y). Then f t (f st (y)) = f t (y). Hence, f st (y) = y since
by Lemma 1.4.14, f t is one-one. Let x = f st1 (y) A. Then f (x) = y. Hence, f is onto A.
Definition 1.4.16 Let A and B be sets and f : A B.
(i) f is called left invertible if there exists g : B A such that
g f = iA .
(ii) f is called right invertible if there exists h : B A such that
f h = iB .
A function f : A B is called invertible if f is both left and right invertible.
Example 1.4.17 Let f : Z Z and g : Z Z be as defined below.
f (n) = 3n
g(n) =
for all n Z. Now

n
3

if n is a multiple of 3
0 if n is not a multiple of 3

(f g)(n) = f(g(n))
n if n is a multiple of 3
=
0 if n is not a multiple of 3.

Hence, f g 6= iZ . But (g f )(n) = g(f (n)) = g(3n) = n for all n Z. Thus, g f = iZ . Hence, g is a
left inverse of f.
Often we are required to find a left (right) inverse of a function. However, not every function has
a left (right) inverse. Thus, before we attempt to find a left (right) inverse of a function, it would be
helpful to know if a given function has a left (right) inverse or not. The following theorem is very useful
in determining whether a function is left (right) invertible or invertible.
Theorem 1.4.18 Let A and B be sets and f : A B. Then the following assertions hold.
(i) f is one-one if and only if f is left invertible.
(ii) f is onto B if and only if f is right invertible.
(iii) f is one-one and onto B if and only if f is invertible.
Proof. (i) Suppose f is left invertible. Then there exists g : B A such that g f = iA . Let x,
y A be such that f (x) = f (y). Then g(f (x)) = g(f (y)) or (g f )(x) = (g f )(y). Hence, iA (x) = iA (y),
i.e., x = y. Thus, f is one-one.
Conversely, suppose f is one-one. Then for y B, either y has no preimage or there exists a unique
xy A such that f (xy ) = y. Fix x A. Define g : B A by

x if y has no preimage under f

g(y) =
xy if y has a preimage under f and f (xy ) = y
for all y B. By the definition of g, D(g) = B. To show g is well defined, suppose y, y 0 B and y = y 0 .
Then either both y and y 0 have no preimages or there exist unique xy , xy0 A such that f (xy ) = y and
f (xy0 ) = y 0 . Suppose both y and y 0 have no preimages. Then g(y) = x = g(y 0 ). Now suppose there exist
unique xy , xy0 A such that f (xy ) = y and f (xy0 ) = y 0 . Thus, g(y) = xy and g(y 0 ) = x0y . Since y = y 0 ,
we have f (xy ) = f (xy0 ). Since f is one-one, xy = xy0 and so g(y) = g(y 0 ). We have thus shown that g is

28

1. Sets, Relations, and Integers

well defined and so g is a function. We now show that g f = iA . Let u A and suppose f (u) = v for
some v B. Then by the definition of g, g(v) = u. Thus,
(g f )(u) = g(f (u)) = g(v) = u = iA (u).
Hence, g f = iA .
(ii) Suppose f is right invertible. Then there exists g : B A such that f g = iB . Let y B. Let
x = g(y) A. Now y = iB (y) = (f g)(y) = f (g(y)) = f (x). Hence, f is onto B.
Conversely, suppose f is onto B. Let y B. Since f is onto, there exists x A such that f (x) = y.
Let Ay = {x A | f (x) = y}. Then Ay 6= . Choose xy Ay for all y B. Define h : B A such that
h(y) = xy for all y B. Then h is a function. Let y B. Then (f h)(y) = f (h(y)) = f (xy ) = y = iB (y).
Hence, f h = iB and so f is right invertible.
(iii) The result here follows from (i) and (ii).
Let f : A B be invertible. Let g be a left inverse of f and h be a right inverse of f. Then g f = iA
and f h = iB . Now g = g iB = g (f h) = (g f ) h = iA h = h. Thus, if f is invertible, then
left and right inverses of f are the same. This also proves that the inverse of a function, if it exists, is
unique.
If f is an invertible function, then the inverse of f is denoted by f 1 .
Let f : A B and A0 A. Then f induces a function from A0 into B in a natural way as defined
next.
Definition 1.4.19 Let f : A B and A0 be a nonempty subset of A. The restriction of f to A0 ,
written f |A0 , is defined to be
f |A0 = {(x0 , f (x0 )) | x0 A0 }.
We see that f |A0 is really the function f except that we are considering f on a smaller domain.
Definition 1.4.20 Let f : A0 B and A be a set containing A0 . A function g : A B is called an
extension of f to A if g|A0 = f.
Example 1.4.21 Consider the function f : E Z and g : Z Z, where f (2n) = 2n+1 and g(n) = n+1
for all n Z. Then g is an extension of f to Z and f is the restriction of g to E. Let the function
h : Z Z be defined by for all m Z, h(m) = m + 1 if m E and h(m) = m if m
/ E. Then h is an
extension of f to Z. However, h 6= g. Thus, a function may have more than one extension.
In Section 1.1, we defined the Cartesian cross product, A B, of two sets A and B. We now extend
this notion to a family of sets {A | I}, where I is an index set. First let us make the following
observation: Suppose I = {1, 2}. Let S be the set of all functions f : I A B such that f (1) A
and f (2) B. Then every function f S defines an ordered pair (f (1), f (2)) A B. Conversely,
given x A and y B, define f S by f (1) = x and f (2) = y. Then the ordered pair (x, y) defines
a function f S. Hence, there is a one-one correspondence between the elements of S and A B. We
now define the Cartesian product of {A | I}.
Let {A | I} be a family of sets. The Cartesian (cross) product of {A | I}, denoted
Q
by I A , is defined to be the set
{f | f : I I A and f () A for all I}.

Q
Let f I A . Then f () A for all
Q I. Let us write f () = x for all I. We usually write
(x )I for f, i.e., a typical member of I A is denoted by (x )I , where
x A for all I.
Q
Suppose I = {1, 2, . . . , n} is a finite set. Then the Cartesian product iI A , is denoted by A1
A2 An . A typical member of A1 A2 An is denoted by (x1 , x2 , . . . , xn ), xi Ai for all
i = 1, 2, . . . , n. The elements of A1 A2 An are called ordered n-tuples. For two elements
(x1 , x2 , . . . , xn ), (y1 , y2 , . . . , yn ) A1 A2 An , (x1 , x2 , . . . , xn ) = (y1 , y2 , . . . , yn ) if and only if
xi = yi for all i.

1.4. Functions

29

Worked-Out Exercises
Exercise 1 Determine which of the following mappings f : R R are one-one and which are onto R :
(i) f (x) = x + 4,
(ii) f (x) = x2
for all x R.
Solution (i) Let x, y R. Suppose f (x) = f (y). Then x + 4 = y + 4 or x = y. Hence, f is one-one. Now
f is onto R if and only if for all y R there exists x R such that f (x) = y. Let y R. If
f (x) = y, then x + 4 = y or x = y 4. Also, y 4 R. Thus, we can take x to be y 4. Now
f (y 4) = y 4 + 4 = y. Hence, f is onto R.
(ii) We note that f (x) is a nonnegative real number for all x R. This means that negative real
numbers have no preimages. In particular, for all x R, f (x) = x2 6= 1. Hence, f is not onto R.
Also, f (1) = 1 = f (1) and 1 6= 1. Thus, f is not one-one. Thus, f is neither one-one nor onto
R.

Exercise 2 (i) Let f : Z Z be a mapping defined by

x if x is even
f (x) =
2x + 1 if x is odd
for all x Z. Find a left inverse of f if one exists.

(ii) Let f : Z Z be the mapping defined by f (x) = |x| + x for all x Z. Find a right inverse of
f if one exists.

Solution

(i) By Theorem 1.4.18, f has a left inverse if and only if f is one-one. Before we attempt
to find a left inverse of f, let us first check whether f is one-one or not. Let x, y R and
f (x) = f (y). By the definition of f, f (x) is even if x is even and f (x) is odd if x is odd. Thus,
since f (x) = f (y), we have both x and y are either even or odd. If x and y are both even then
f (x) = x and f (y) = y and so x = y. Suppose x and y are odd. Then f (x) = 2x + 1 and
f (y) = 2y + 1. Then 2x + 1 = 2y + 1 or x = y. Hence, f is one-one and so f has a left inverse.
Thus, there exists a function g : Z Z such that g f = iZ . Let x Z. Suppose x is even. Now
x = iZ (x) = (g f )(x) = g(f (x)) = g(x). This means g(x) = x when x is even. Now suppose x
is odd. Then x = iZ (x) = (g f )(x) = g(f (x)) = g(2x + 1). Put t = 2x + 1. Then x = t1
2 . This
shows that g(x) = x1
if
x
is
odd.
Thus,
our
choice
of
g
is
2

x
if x is even
g(x) =
x1
if x is odd.
2

(ii) Note that f (x) = |x| + x 0 for all x Z. This shows that negative integers do not belong to
I(f ). In particular, f (x) 6= 1 for all x Z. Thus, f is not onto Z and so f does not have a right
inverse.
Exercise 3 Let X and Y be nonempty sets and f : X Y. If T X, then f (T ) denotes the set {f (x) | x T }.
f (T ) is called the image of T under f. Prove that f is one-one if and only if
f (A B) = f (A) f (B)
for all nonempty subsets A and B of X.
Solution Suppose that f is one-one. Let A and B be nonempty subsets of X. Let y f (A B). Then
y = f (x) for some x A B. Hence, y f (A) f (B). Thus, f (A B) f (A) f (B). Now let
y f (A) f (B). Then y f (A) and y f (B). Thus, y = f (a) for some a A and y = f (b) for
some b B. Since f is one-one and f (a) = f (b), we find that a = b. Thus, y f (A B). Hence,
f (A) f (B) f (A B). Consequently, f (A B) = f (A) f (B).

30

1. Sets, Relations, and Integers

Conversely, suppose that f (A B) = f (A) f (B) for all subsets A and B of X. Suppose f is not
one-one. Then there exist x, y X such that f (x) = f (y) and x 6= y. Let A = {x} and B = {y}.
Since AB = , f (AB) = . However, f (A)f (B) = {f (x)} 6= . Thus, f (AB) 6= f (A)f (B),
a contradiction. Hence, f is one-one.

Exercise 4 Let A be a nonempty set and E be an equivalence relation on A. Let B = {[x] | x A}, i.e., B is
the set of all equivalence classes with respect to E. Prove that there exists a function f from A
onto B. The set B is usually denoted by A/E and is called the quotient set of A determined by
E.
Solution Define f : A B by f (x) = [x] for all x A. By the definition of f, D(f ) = A. Let x, y A.
Suppose x = y. Then [x] = [y] and so f (x) = f (y). Thus, f is well defined. Let [a] B. Then
a A and f (a) = [a]. Hence, f is onto B.
Exercise 5 Let S = {x R | 1 < x < 1}. Show that R S.
Solution Define f : R S by

f (x) =

x
1 + |x|

for all x R. Let x R. Then |x| x |x| , 1 |x| < |x| , and |x| 1 + |x| . Hence,
x
1 |x| < x < 1 + |x| . Thus, 1 < 1+|x|
< 1 and so 1 < f (x) < 1. This shows that f (x) S.

|x|
|y|
y
x
= 1+|y|
. Thus, 1+|x|
= 1+|y|
. This implies that
Let x, y R and f (x) = f (y). Then 1+|x|
y
x
|x| + |x| |y| = |y| + |x| |y| and so |x| = |y| . Now 1+|x| = 1+|y| implies that x 0 if and only if
y 0. Therefore, since |x| = |y| , x = y. Thus, f is one-one.

Now let z R and 1 < z < 1. If 0 z < 1, then

f(

If 1 < z < 0, then

f(

z
1z
1z

)=
z = 1 + z = z.
1z
1z
1 + 1z
z
1+z
1+z

)=
z = 1 + z = z.
1+z
1 + 1+z
1+z

Hence, f is onto R. Consequently, R S.

Exercises
1. Determine which of the following mappings f : R R are one-one and which are onto R :
(i) f (x) = x + 1,
(ii) f (x) = x3 ,
(iii) f (x) = |x| + x
for all x R.
2. Consider the function f = {(x, x2 ) | x S} of S = {3, 2, 1, 0, 1, 2, 3} into Z. Is f one-one?
Is f onto Z?

3. Let f : R+ R+ and g : R+ R+ be functions defined by f (x) = x and g(x) = 3x + 1 for all

x R+ , where R+ is the set of all positive real numbers. Find f g and g f. Is f g = g f ?
4. Let f : Q+ R and g : R R be defined by f (x) = 1 + x1 for all x Q+ and g(x) = x + 1 for all
x R, where Q+ is the set of all positive rational numbers. Find g f.

1.4. Functions

31

5. For each of the mappings f : Z Z given below, find a left inverse of f whenever one exists.
(i) f (x) = x + 2,

(ii) f (x) = 2x,

x
2 if x is even
(iii) f (x) =
5 if x is odd
for all x Z.

6. For each of the mappings f : Z Z given below, find a right inverse of f whenever one exists.
(i) f (x) = x 3,

(ii) f (x) = 2x,

x if x is even
(iii) f (x) =
x + 1 if x is odd

for all x Z.

7. Let A = {1, 2, 3}. List all one-one functions from A onto A.

8. Let A = {1, 2, . . . , n}. Show that the number of one-one functions of A onto A is n!
9. Let f : A B be a function. Define a relation R on A by for all a, b A, aRb if and only if
f (a) = f (b). Show that R is an equivalence relation.
10. Given f : X Y and A, B X, prove that
(i) f (A B) = f (A) f (B),

(ii) f (A B) f (A) f (B),

(iii) f (A\B) f (A)\f (B) if f is one-one.

11. Given f : X Y. Let S Y. Define f 1 (S) = {x X | f (x) S}. Let A, B Y. Prove that
(i) f 1 (A B) = f 1 (A) f 1 (B),

(ii) f 1 (A B) = f 1 (A) f 1 (B),

(iii) f 1 (A\B) = f 1 (A)\f 1 (B).

12. Let f : A B. Let f be the inverse relation, i.e.,

f = {(y, x) B A | f (x) = y}.
(i) Show by an example that f need not be a function.
(ii) Show that f is a function from I(f ) into A if and only if f is one-one.

(iii) Show that f is a function from B into A if and only if f is one-one and onto B.
(iv) Show that if f is a function from B into A, then f 1 = f .
13. Show that ZE, where E is the set of all even integers.
14. Let A = {x R | 0 x 1} and B = {x R | 5 x 8}. Show that f : A B defined by
f (x) = 5 + (8 5)x is a one-one function from A onto B.
15. (i) Show that Z and 3Z are equipollent.
(ii) Show that 5Z and 7Z are equipollent.
16. Let S = {x R | 0 < x < 1}. Show that R+ S.
17. (Schrder-Bernstein) Let A and B be sets. If A Y for some subset Y of B and B X for
some subset X of A, prove that A B.

32

1. Sets, Relations, and Integers

18. Find a one-one mapping from R onto R+ .

19. Is Z Q?
20. Let A = {x R | 0 x 1} and B = {x R | 0 < x < 1}. Is it true that A B?
21. For each of the following statements, write the proof if the statement is true, otherwise give a
counterexample.
(i) A function f : A B is one-one if and only if g f = h f for all functions g, h : B A.
(ii) A function f : A B is one-one if and only if for all subsets C of A, f (A\C) B\f (C).

1.5

Binary Operations

The concept of a binary operation is very important in abstract algebra. Throughout the text we will
be concerned with sets together with one or more binary operations. In this section, we define binary
operations and examine their basic properties.
Definition 1.5.1 Let S be a nonempty set. A binary operation on S is a function from S S into
S.
For any ordered pair (x, y) of elements x, y S, a binary operation assigns a third member of S. For
example, + is a binary operation on Z which assigns 3 to the pair (2, 1).
If is a binary operation on S, we write x y for (x, y), where x, y S. Since the image of is a
subset of S, we say S is closed under .
Z is closed under + since if we add two integers we obtain an integer. Since 2, 5 N and 2 5 =
3
/ N, we see that (subtraction) is not a binary operation of N and we say that N is not closed under
.
Definition 1.5.2 A mathematical system is an ordered (n + 1)-tuple (S, 1 , . . . , n ), where S is a
nonempty set and i is a binary operation on S, i = 1, 2, . . . , n. S is called the underlying set of the
system.
Definition 1.5.3 Let (S, ) be a mathematical system. Then
(i) is called associative if for all x, y, z S, x (y z) = (x y) z.
(ii) is called commutative if for all x, y S, x y = y x.
Example 1.5.4 Consider the mathematical system (Z, +). Since addition of integers is both associative
and commutative, + is both associative and commutative.
Example 1.5.5 Let A be a nonempty set. Let S be the set of all functions on A, i.e.,
S = {f | f : A A}.
Since composition of functions is a function (Theorem 1.4.11), (S, ) is a mathematical system. By
Theorem 1.4.13, is associative.
Example 1.5.6 Let M2 (R) be the set of all 2 2 matrices over R, i.e.,

a b
M2 (R) =
| a, b, c, d R .
c d
Let + denote the usual addition of matrices and denote the usual multiplication of matrices. Since
addition (multiplication) of 2 2 matrices over R is a 2 2 matrix over R, it follows that + ( )
is a binary operation on M2 (R). Hence, (M2 (R), +, ) is a mathematical system. Note that + is both
associative and commutative and is associative, but not commutative.

1.5. Binary Operations

33

The following is an example of a mathematical system for which the binary operation is neither
associative nor commutative.
Example 1.5.7 Consider the mathematical system (Z, ), where denotes the binary operation of
subtraction on Z. Then 3 (2 1) = 2 6= 0 = (3 2) 1 and so is not associative. Also, since
3 2 6= 2 3, is not commutative.
A convenient way to define a binary operation on a finite set S is by means of an operation or
multiplication table. For example, let S = {a, b, c}. Define on S by the following operation table.

a
b
c

a
c
a
b

b
b
a
b

c
a
a
b

To determine the element of S assigned to a b, we look at the intersection of the row labeled by a
and the column headed by b. We see that a b = b. Note that b a = a.
Definition 1.5.8 Let (S, ) be a mathematical system. An element e S is called an identity of (S, )
if for all x S,
e x = x = x e.
Example 1.5.9 Let S = {e, a, b}. Define on S by the following multiplication table

e
a
b

e
e
a
b

a
a
a
a

b
b
a
a

We note that e a = a = a e, e b = b = b e and e e = e = e e. Thus, e is an identity of (S, ).

Example 1.5.10 (i) In Example

1.5.5, iA is an identity element of (S, ).

0 0
(ii) In Example 1.5.6,
is an identity element for the mathematical system (M2 (R), +) and
0 0

1 0
is an identity element for the mathematical system (M2 (R), ).
0 1
Theorem 1.5.11 An identity element (if it exists) of a mathematical system (S, ) is unique.
Proof. Let e, f be identities of (S, ). Since e is identity, e a = a for all a S. Substituting f for
a, we get
e f = f.
(1.4)
Now f is identity and so a f = a for all a S. Substituting e for a we get
e f = e.
From Eqs. (1.4) and (1.5), we get e = f. Hence, an identity element (if it exists) is unique.

Worked-Out Exercises
Exercise 1 Which of the following are associative binary operations?
(i) (Z, ), where x y = (x + y) (x y) for all x, y Z.

(ii) (R, ), where x y = max(x, y) for all x, y R.

(iii) (R, ), where x y = |x + y| for all x, y R.

(1.5)

34

1. Sets, Relations, and Integers

Solution (i) (x y) z = ((x + y) (x y)) z = (x + y) (x y) + z ((x + y) (x y)) z = x + y+

z x y x z y z + x y z. Similarly, x (y z) = x + y + z x y x z y z+ x y z.
Thus, (x y) z = x (y z). Hence, is associative.
(ii) (xy)z = max(x, y)z = max(max(x, y), z) = max(x, y, z) = max(x, max(y, z)) = xmax(y, z)
= x (y z). Thus, is associative.

(iii) (2 (3)) 6 = |2 + (3)| 6 = 1 6 = |1 + 6| = 7 and 2 ((3) 6) = 2 (|(3) + 6|) =

2 3 = |2 + 3| = 5. Hence, (2 (3)) 6 6= 2 ((3) 6) and so is not associative.

Exercises
1. Which of the following are associative binary operations?
(i) (N, ), where x y = xy for all x, y N.

(ii) (Z, ), where x y = x + y + 1 for all x, y Z.

(iii) (N, ), where x y = gcd(x, y) for all x, y N.

(iv) (N, ), where x y = lcm(x, y) for all x, y N.

(v) (R, ), where x y = min(x, y) for all x, y R.

(vi) (R, ), where x y = |x| + |y| for all x, y R.

2. In Exercise 1, which of the operations are commutative?

3. In Exercise 1, which mathematical systems have an identity?

Chapter 2

Introduction to Groups
There are four major sources from which group theory evolved, namely, classical algebra, number theory, geometry, and analysis. Classical algebra originated in 1770 with J. L. Lagranges work on polynomial equations. His work appeared in a memoir entitled, Rflexions sur la rsolution algbrique des
quations. C. F. Gauss is considered the originator of number theory with his work, Disquistiones
Arithmeticae, which was published in 1801. F. Kleins lecture in 1872, A Comparative Review of Recent Researches in Geometry, dealt with the classification of geometry as the study of invariants under
groups of transformations. The impact of his lecture was so strong as to allow Klein to be considered
as the originator of this source of group theory. The originators of the analysis source are S. Lie (1874)
and H. Poincar and F. Klein (1876).

2.1

Elementary Properties of Groups

In this chapter, and in fact in the remainder of the text, we will be concerned with mathematical systems.
These systems are composed of a nonempty set together with binary operations defined on this set so
that certain properties hold. From these properties, results concerning these systems are derived. This
axiomatic approach to abstract algebra unifies diverse examples and also strips away nonessential ideas.
Although noted for his geometry, Euclid inspired the use of the axiomatic method, which has proved
so indispensable in mathematics. His axiomatic approach also aected philosophy, where in the 17th
century Baruch Spinoza laid down (in The Ethics) an axiomatic system from which he was able to prove
the existence of God. His proof, of course, depended on his axioms. His proof lost its conviction with
the emergence of noneuclidean geometries whose axioms were as logical and practical as Euclids.
We will be primarily concerned with mathematical systems called groups in this chapter. The theory
of groups is one of the oldest branches of abstract algebra. The first eective use of groups was in
the early nineteenth century by A. Cauchy and E. Galois. They used groups to describe the eect of
permutations of roots of a polynomial equation. Their use of groups was not based on an axiomatic
approach. In 1854, A. Cayley gave the first postulates for a group. However, his definition was lost sight
of. Kronecker again set down the axioms for an Abelian group in 1870. H. Weber gave the definition for
finite groups (in 1882) and the definition for infinite groups in 1883.
As previously mentioned, the notion of a group arose from the study of one-one functions on the set
of roots of a polynomial equation. We have seen that the set S of all one-one functions from a set X
onto itself satisfies the following properties:
(i) Composition of functions, , is a binary operation on S.
(ii) For all f, g, h S, f (g h) = (f g) h.
(iii) There exists i S such that f i = f = i f for all f S.
(iv) For all f S there exists an element f 1 S such that f f 1 = i = f 1 f.
These properties lead us to the definition of an abstract group.

35

36

2. Introduction to Groups

Definition 2.1.1 A group is an ordered pair (G, ), where G is a nonempty set and is a binary
operation on G such that the following properties hold:
(G1) For all a, b, c G, a (b c) = (a b) c (associative law).
(G2) There exists e G such that for all a G, a e = a = e a (existence of an identity).
(G3) For all a G, there exists b G such that a b = e = b a (existence of an inverse).
Thus, a group is a mathematical system (G, ) satisfying axioms G1 to G3.
Example 2.1.2 Consider Z, the set of integers, together with the binary operation +, where + is the
usual addition. We know that + is associative. Now 0 Z and for all a Z,
a + 0 = a = 0 + a.
So 0 is an identity. Also, for all a Z, a Z and
a + (a) = 0 = (a) + a.
That is, a is an inverse of a. It now follows that (Z, +) satisfies axioms G1 to G3, so (Z, +) is a group.
As in Example 2.1.2, we can show that (Q, +), (R, +), (C, +) are also groups, where + is the usual
addition.
Example 2.1.3 Consider Q\{0}, the set of nonzero rational number, together with the binary operation
, where is the usual multiplication. We know that is associative. Now 1 Q and for all a Q,
a 1 = a = 1 a.
So 1 is an identity. Also, for all a Q\{0},

1
a

Q\{0} and
1
1
= 1 = a.
a
a

This implies that a1 is an inverse of a. It now follows that (Q\{0}, ) satisfying axioms G1 to G3, so
(Q\{0}, ) is a group.
As in Example 2.1.3, we can show that (R\{0}, ), (C\{0}, ) are also groups, where is the usual
multiplication. Note that for each of the groups (R\{0}, ), (C\{0}, ) the identity element is 1.
Example 2.1.2 shows that 0 is an identity of (Z, +) and for each element a Z, a is an inverse of
a. Similarly, Example 2.1.3 shows that 1 is an identity of (Q\{0}, ) and for each element a Q\{0},
1
a is an inverse of a. The next theorem shows that in a group there is only one identity element, i.e.,
identity element is unique. Similarly, in a group, every element has only one inverse, i.e., the inverse of
an element is unique.
Theorem 2.1.4 Let (G, ) be a group.
(i) There exists a unique element e G such that e a = a = a e for all a G.
(ii) For all a G, there exists a unique b G such that a b = e = b a.
Proof. (i) Now (G, ) is group. Therefore, by G2, there exists e G such that e a = a = a e for
all a G. Because (G, ) is a mathematical system, e is unique by Theorem 1.5.11.
(ii) Let a G. By G3, there exists b G such that a b = e = b a. Suppose there exists c G such
that a c = e = c a. We show that b = c. Now
b =
=
=
=
=

be
b (a c)
(b a) c
ec
c.

(substituting e = a c)
(using the associativity of )
(because b a = e)

Thus, b is unique.
The unique element e G that satisfies G2 is called the identity element of the group (G, ). Let
a G. Then the unique element b G that satisfies G3 is called the inverse of a and is denoted by
a1 .

2.1. Elementary Properties of Groups

37

Remark 2.1.5 By Theorem 2.1.4, it follows that for each of the groups (Z, +), (Q, +), (R, +), and
(C, +), the identity element is 0. Similarly, for each of the groups (Q\{0}, ), (R\{0}, ), and (C\{0}, ),
the identity element is 1.
Before giving additional examples of groups, let us make the following definition.
Definition 2.1.6 Let (G, ) be a group. If for all a, b G
a b = b a,
then (G, ) is called a commutative or Abelian group. A group (G, ) is called noncommutative if
it is not commutative.
Example 2.1.7 Consider the group (Z, +) of Example 2.1.2. Because a + b = b + a for all a, b Z, it
follows that + is commutative. Hence, (Z, +) is a commutative group.
Similarly, (Q, +), (R, +), (C, +), (Q\{0}, ), (R\{0}, ), (C\{0}, ) are also commutative groups, where
+ is the usual addition and is the usual multiplication.
Next we consider additional examples of (commutative) groups.
Example 2.1.8 Consider Z, the set of integers. Let a be a fixed integer. Let
G = {na | n Z}.
That is, G consists of all multiples of a. Note that G Z.
Now 0 = 0a G. So it follows that G is nonempty. Because + is commutative and associative on Z
and G is a subset of Z, it follows that + is commutative and associative on G. Moreover, note that 0 is
the identity element of G. Also for each na G, (na) = (n)a G and
na + ((na)) = 0 = ((na)) + na.
We can now conclude that (G, +) is a commutative group.
Let n be a fixed positive integer, Chapter 1 extensively describes the set Zn and the binary relation
n on Zn . The next example shows that Zn together with the binary relation +n , as defined in that
example, is a commutative group. The next two examples are, in fact, due to Gausss, whose work
yielded many new directions of research in Abelian groups.
Example 2.1.9 Let n be a fixed positive integer. Consider Zn (as defined in Examples 1.3.11). Let +n
be defined on Zn by
[a] +n [b] = [a + b].
Recall that [a] = {x Z | x n a}. We show that (Zn , +n ) is a commutative group.
First, we show that +n is a binary operation on Zn . Let [a], [b] Zn . Then [a] +n [b] = [a + b] Zn .
Next let, [a], [b], [c], [d] Zn . Suppose [a] = [c] and [b] = [d]. Then a c(mod n) and b d(mod n).
Thus, there exist s, t Z such that
a c = ns and b d = nt.
This implies that
a + b (c + d) = (a c) + (b d) = ns + nt = n(s + t).

Thus, a+b (c+d)(mod n), so [a+b] = [c+d]. This implies that [a]+n [b] = [c]+n [d]. Consequently,
+n is well-defined. It now follows that +n is a binary operation on Zn .
For all [a], [b], [c] Zn ,
([a] +n [b]) +n [c] =
=
=
=
=

[a + b] +n [c]
[(a + b) + c]
[a + (b + c)]
[a] +n [b + c]
[a] +n ([b] +n [c])

by the definition of +n
by the definition of +n
becasue + is associative on Z
by the definition of +n
by the definition of +n .

38

2. Introduction to Groups

This implies that, +n is associative.

Now [0] Zn and for all [a] Zn ,
[a] +n [0] = [a + 0] = [a] = [0 + a] = [0] +n [a].
This shows that [0] is the identity element. Also, for all [a] Zn , [a] Zn we have
[a] +n [a] = [a a] = [0] = [a + a] = [a] +n [a].
Thus, [a] is the inverse of [a]. Finally, for all [a], [b] Zn
[a] +n [b] = [a + b] = [b + a] = [b] +n [a],
so +n is commutative. Hence, (Zn , +n ) is a commutative group.
Example 2.1.10 Let n be a fixed positive integer. Consider Zn (as defined in Examples 1.3.11). Let n
be define on Zn by for all [a], [b] Zn
[a] n [b] = [ab].
First we show that n is a binary operation on Zn . Let [a], [b] Zn . Then [a] n [b] = [ab] Zn . Next
let, [a], [b], [c], [d] Zn . Suppose [a] = [c] and [b] = [d]. Then a c(mod n) and b d(mod n). Thus,
there exist s, t Z such that
a c = ns and b d = nt.
This implies that
ab cd = ab bc + bc cd = b(a c) + c(b d)
= bns + cnt = n(bs + ct).
Thus, ab cd(mod n), so [ab] = [cd]. This implies that [a] n [b] = [c] n [d]. Consequently, n is welldefined. It now follows that n is a binary operation on Zn . Thus, (Zn , n ) is a mathematical system.
Moreover, as in Example 2.1.9, we can show that n is associative.
Now [1] Zn and for all [a] Zn ,
[a] n [1] = [a 1] = [a] = [1 a] = [1] n [a].
This implies that [1] is the identity element.
Let [a] Zn and [a] 6= [0]. We leave it as an exercise to show that [a] has an inverse if and only if
gcd(a, n) = 1. Thus, we see that in general, not every element of Zn \{[0]} has an inverse.
For example if n = 6, then the only elements of Z6 that have inverses are [1], [3], and [5]. Hence, in
general (Zn \{[0]}, n ) is not a group.
Let Un be the set of all elements of Zn \{[0]} that have an inverse in (Zn \{[0]}, n ), i.e.,
Un = {[a] Zn \{[0]} | gcd(a, n) = 1}.
We ask the reader to verify in Exercise 10 (page 55) that (Un , n ) is a group.
Note that for n = 8, U8 = {[1], [3], [5], [7]} and for n = 7,
U7 = {[1], [2], [3], [4], [5], [6]} = Z7 \{[0]}.
Example 2.1.11 Let

Q[ 2] = {a + b 2 | a, b Q}.

Note that Q[ 2] R.
Now 0= 0 + 0 2 Q[ 2]. This shows that Q[ 2] 6= . Define + on Q[ 2] as
follows: for all a + b 2, c + d 2 Q[ 2],

(a + b 2) + (c + d 2) = (a + b) + (c + d) 2).

2.1. Elementary Properties of Groups

39

It
is easy to see that + is a binary operation on Q[ 2]. Note that
+ is the usual addition.
Because

Q[ 2] R, it follows that + is associative and commutative on Q[ 2]. Next, for all a + b 2 Q[ 2],

(a + b 2) + (0 + 0 2) = a + b 2 = (0 + 0 2) + (a + b 2).

Thus, 0 = 0+0
2 is the identity element of (Q[ 2], +). Note that the inverse of a+b 2 is a+(b) 2.

Hence, (Q[ 2], +) is a commutative groups.

In a similar manner, we can show that (Q[
group, where is the
2]\{0}, ) is a commutative

usual
multiplication. Note
that
the
identity
of
(Q[
2]\{0},
)
is
1
=
1
+
0
2
and
the
inverse
of
a
+
b
2 6= 0

a
b
is a2 2b

2.
2
a2 2b2
Example 2.1.12 Let P(X) be the power set of a set X. Consider the operation on P(X). Then for
all A, B P(X),
AB = (A\B) (B\A).
It can be verified that is a binary operation on P(X) and is associative.
Now P(X). Let A P(X). Then
A = (A\) (\A)
= A
because A\ = A and \A =
= A.
Similarly, A = A. Thus, A = A = A. It now follows that is the identity element.
Next,
AA = (A\A) (A\A)
=
because A\A = and A\A =
= .
This implies that A is the inverse of A, i.e., A is its own inverse. We can now conclude that (P(X),
) is a group.
We also note that for all A, B P(X)
AB

= (A\B) (B\A)
= (B\A) (A\B)
= BA.

This shows that is commutative on P(X). Consequently, (P(X), ) is a commutative group.

Example 2.1.13 Let X be a set and SX the set of all one-one functions of X onto X, i.e.,
SX = {f : X X | f is a one-one function of X onto X}.
Because iX , the identity function on X, is one-one and onto X, iX SX . Thus, SX 6= .
Let f, g SX . Then f g is a one-one function of X onto X by Theorem 1.4.11. Hence, f g SX .
By Theorem 1.4.13, is associative. Also, for all f SX , f 1 SX and f f 1 = iX = f 1 f.
Consequently, (SX , ) is a group.
Note that, (SX , ) is not necessarily commutative. For example, let X = {a, b, c}. Let f, g SX be
defined by
f (a) = b, f (b) = a, f (c) = c,
g(a) = b, g(b) = c, g(c) = a.
Then
(f g)(b) = f (g(b)) = f (c) = c
and
(g f )(b) = g(f (b)) = g(a) = b.
This implies that (f g)(b) 6= (g f )(b). Hence, f g 6= g f. Thus, (SX , ) is not commutative.

40

2. Introduction to Groups

Example 2.1.14 Let GL(2, R) =

on GL(2, R) by

for all

0
that
0

a b
c d

a b
c d

| a, b, c, d R, ad bc 6= 0 . Define a binary operation

v
s

au + bw
cu + dw

av + bs
cv + ds

a b
u v
,
GL(2, R). This binary operation is the usual matrix multiplication. Note
c d
w s
0
GL(2, R). Thus, GL(2, R) 6= .
0

1 0
Because matrix multiplication is associative, is associative. Next consider the matrix
.
0 1

1 0
a b
Because 1100 = 1 6= 0, it follows that
GL(2, R). Moreover, for any
GL(2, R)
0 1
c d

a b
1 0
a b
1 0
a b

=
=

.
c d
0 1
c d
0 1
c d

1 0
This implies that
is the identity element of GL(2, R).
0 1

d
b
a b
adbc
. Because
Let
GL(2, R). Then ad bc 6= 0. Consider the matrix adbc
c
a
c d
adbc
adbc
a
b
c
1
d

=
6= 0,
ad bc ad bc ad bc ad bc
ad bc
we have

Now

and

Thus,

d
adbc
c
adbc

b
adbc
a
adbc

Next note that

and

a b
c d

b
adbc
a
adbc

d
adbc
c
adbc

d
adbc
c
adbc

GL(2, R).

b
adbc
a
adbc

1 0
0 1

a b
1 0

=
.
c d
0 1

a b
is the inverse of
. Hence, (GL(2, R), ) is a group.
c d

1 1
1 0
,
GL(2, R)
0 1
1 1

1 1
0 1

d
adbc
c
adbc

b
adbc
a
adbc

1 0
2 1
1 1
1 0
1 1

=
6=
=

.
1 1
1 1
1 2
1 1
0 1

Hence, (GL(2, R), ) is a noncommutative group.

Remark 2.1.15 The group in Example 2.1.14 is known as the general linear group of degree 2.
We now prove some elementary properties of a group in the following theorem.
Theorem 2.1.16 Let (G, ) be a group.
(i) (a1 )1 = a for all a G.
(ii) (a b)1 = b1 a1 for all a, b G.
(iii) (Cancellation Law) For all a, b, c G, if either a c = b c or c a = c b, then a = b.
(iv) For all a, b G, the equations a x = b and y a = b have unique solutions in G for x and y.

2.1. Elementary Properties of Groups

41

Proof. (i) Let a G. Then a1 a = e = a a1 . Thus, by the definition of an inverse of an

element, a is an inverse of a1 . However, by Theorem 2.1.4, the inverse of an element is unique in a
group. Therefore, because (a1 )1 denotes the inverse of a1 , it follows that (a1 )1 = a.
(ii) Let a, b G. Then
(a b) (b1 a1 ) =
=
=
=
=

((a b) b1 ) a1
(a (b b1 )) a1
(a e) a1
a a1
e.

Similarly, (b1 a1 ) (a b) = e. Hence, b1 a1 is an inverse of a b. Because the inverse of an element

is unique in a group and (a b)1 denotes the inverse of a b, it follows that (a b)1 = b1 a1 .
(iii) Let a, b, c G. Suppose a c = b c. Now
(a c) c1 = (b c) c1
a (c c1 ) = b (c c1 ) (by the associativity of )
ae=be
(becasue c c1 = e)
a = b.
Similarly, if c a = c b, then we can show that a = b.
(iv) Let a, b G. First we consider the equation
a x = b.
Now a1 b G. Substituting a1 b for x in the equation a x = b, we obtain
a (a1 b) = (a a1 ) b = e b = b.
This implies that a1 b is a solution of the equation a x = b.
We now establish the uniqueness of the solution. Suppose c is any solution of ax = b. Then ac = b.
Hence,
c = ec
= (a1 a) c (because a1 a = e)
= a1 (a c) (because is associative)
= a1 b
(because a c = b).
This yields the uniqueness of the solution.
Similar arguments hold for the equation y a = b.
Corollary 2.1.17 Let (G, ) be a group and a G. If a a = a, then a = e.
Proof. Suppose a = a a. This implies that a a = a e. By the cancellation law, a = e.
Corollary 2.1.18 In a multiplication table for a group (G, ), each element appears exactly once in each
row and exactly once in each column.
Proof. Let b G be such that b occurs twice in the row labeled by a G. Then there exists u, v G
with u 6= v such that a u = b and a v = b. This implies that the equation a x = b has (at least) two
distinct solutions, which are u and v. This is a contradiction to Theorem 2.1.16(iv) because the equation
a x = b has a unique solution for x.
A similar argument for columns can be used.
Let (G, ) be a group and a, b, c G. Then by the associative law,
a (b c) = (a b) c.
Hence, we can define
a b c = a (b c) = (a b) c.

42

2. Introduction to Groups

Let a, b, c, d G. Then

(a b c) d =
=
=
=
=

(a (b c)) d
a ((b c) d))
a (b (c d))
(a b) (c d)
((a b) c) d.

Thus, there is more than one way of inserting parentheses in the expression a b c d to produce a
meaningful product of a, b, c, d (in this order). We now extend this notion to any finite number of
elements.
Definition 2.1.19 Let (G, ) be a group and a1 , a2 , . . . , an G be n elements of G (not necessarily
distinct). The meaningful product of a1 , a2 , . . . , an (in this order) is defined as follows:
If n = 1, then the meaningful product is a1 .
If n > 1, then the meaningful product of a1 , a2 , . . . , an is any product of the form
(a1 am ) (am+1 an ),
where 1 m < n and (a1 am ) and (am+1 an ) are meaningful products of m and n m
elements, respectively.
Definition 2.1.20 Let (G, ) be a group and a1 , a2 , . . . , an G, n 1. The standard product of
a1 , a2 , . . . , an denoted by a1 a2 an is defined recursively as
a1
a1 a2 an

= a1
= (a1 a2 an1 ) an if n > 1.

In the next theorem, we establish the equality between any meaningful product and standard product.
Theorem 2.1.21 Let (G, ) be a group and a1 , a2 , . . . , an G, n 1. Then all possible meaningful
products of a1 , a2 , . . . , an (in this order) are equal to the standard product of a1 , a2 , . . . , an (in this
order).
Proof. We prove the result by induction.
Basis step: If n = 1, then a1 is the only meaningful product of a1 , which is equal to the standard
product a1 of a1 . Thus, the result is true if n = 1.
Inductive hypothesis: Suppose that the theorem is true for all integers m such that 1 m < n.
Inductive step: Let a1 , a2 , . . . , an G. Let (a1 at ) (at+1 an ) be a meaningful product
of a1 , a2 , . . . , an (in this order). Now t < n and n t < n. If t = n 1, then
(a1 a2 at ) at+1 = a1 a2 at at+1 .
Suppose t < n 1. Then
(a1 at ) (at+1 an )
= (a1 at ) ((at+1 an1 ) an )
= ((a1 at ) (at+1 an1 )) an
= (a1 a2 an1 ) an
by the inductive hypothesis,
(a1 at ) (at+1 an1 )
= a1 a2 an1 .
= a1 an
This shows that the result is true for n. The result now follows by induction.
We have seen several examples of groups. In order to show that a given set with a given binary
operation is a group, we need to verify G1 to G3 of Definition 2.1.1. However, it would be helpful if we
had some criteria that could be used to show whether a given set with a binary operation is a group
or not instead of verifying all the properties G1G3 explicitly. Partly for this reason we define what a
semigroup is. Following the examples, we develop some results that can be used to test whether a given
set with a binary operation is a group or not.

2.1. Elementary Properties of Groups

43

Definition 2.1.22 A semigroup is an ordered pair (S, ), where S is a nonempty set and is an
associative binary operation on S.
Thus, a semigroup is a mathematical system with one binary operation such that the binary operation
is associative.
Remark 2.1.23 For any group (G, ), the binary operation is associative. Therefore, every group
(G, ) is a semigroup.
As in the case of a group, next we define a commutative semigroup.
Definition 2.1.24 A semigroup (S, ) is commutative if is commutative, i.e., a b = b a for all
a, b S. A semigroup (S, ) which is not commutative is called noncommutative.
Definition 2.1.25 Let (S, ) be a semigroup.
(i) We say that (S, ) is with identity if the mathematical system (S, ) has an identity.
(ii) An element a S is called idempotent if a a = a.
Example 2.1.26 (i) Consider N, the set of positive integers. We know that the sum of positive integers
is again a positive integer. Thus, + is a binary operation on N. We also know that + is associative and
commutative. Thus, (N, +) is a commutative semigroup. In a similar manner, (N, ) is a commutative
semigroup, where denotes the usual multiplication of integers.
(ii) Because (Z, +) is a commutative group, it is a commutative semigroup. Also note that (Z, ) is a
commutative semigroup.
Example 2.1.27 Let X be a nonempty set and S the set of all functions f : X X. If denotes the
composition of functions, then (S, ) is a semigroup with identity. The associativity of follows from
Theorem 1.4.13.
When X has two or more elements, the semigroup (S, ) is noncommutative. For example, let
X = {a, b}. Let g, h S be defined by
g(a) = b, g(b) = b, h(a) = b, h(b) = a.
Then
(g h)(a) = b 6= a = (h g)(a).
Therefore, g h 6= h g.
Let f S be defined by f (a) = a and f (b) = a. Now
(f g)(x) = f (g(x)) = a = f (h(x)) = (f h)(x)
for all x G. Hence, f g = f h. But g 6= h. This shows that the cancellation laws do not hold in S.
Thus, (S, ) is not a group.
Example 2.1.28 Let X be a set with two or more elements and S 0 the set of all functions f : X X
which are not one-one. Then (S 0 , ) is a noncommutative semigroup without identity.
Example 2.1.29 Let X be a set and P(X) the power set of X. We leave it as an exercise that (P(X), )
and (P(X), ) are commutative semigroups with identity. The identity of (P(X), ) is and the identity
of (P(X), ) is X.
The following three theorems give necessary and sucient conditions for a semigroup to be a group.
Theorem 2.1.30 A semigroup (S, ) is a group if and only if
(i) there exists e S such that e a = a for all a S, (i.e., e is a left identity), and
(ii) for all a S there exists b S such that b a = e, (i.e., every element has a left inverse).

44

2. Introduction to Groups

Proof. Suppose (S, ) is a semigroup that satisfies (i) and (ii). Let a be any element of S. Then
there exists b S such that b a = e by (ii). For b S, there exists c S such that c b = e by (ii). Now
a = e a = (c b) a = c (b a) = c e
and
a b = (c e) b = c (e b) = c b = e.
Hence, a b = e = b a. Also,
a e = a (b a) = (a b) a = e a = a.
Thus, a e = a = e a. This shows that e is the identity element of S. Now because a b = e = b a, we
have b = a1 . Therefore, (S, ) is a group.
The converse follows from the definition of a group.
Remark 2.1.31 To verify that a specific nonempty set is a group, we can use Theorem 2.1.30 as follows:
Show that (1) the operation, say , defined on the set is well-defined; (2) is associative; (3) the set has
a left identity; and (4) every element has a left inverse. For example, see Worked-Out Exercises 1, 2,
and 3 at the end of this section.
Remark 2.1.32 The analog of Theorem 2.1.30 is given in Exercise 39 at the end of this section.
Theorem 2.1.33 A semigroup (S, ) is a group if and only if for all a, b S the equations a x = b
and y a = b have solutions in S for x and y.
Proof. Suppose the given equations have solutions in S. Let a S. Consider the equation y a = a.
By our assumption, y a = a has a solution u S, say. Then u a = a. Let b be any element of S.
Consider the equation a x = b. Again by our assumption, a x = b has a solution in S. Let c S be a
solution of a x = b. Then a c = b. Now
ub =
=
=
=

u (a c)
(u a) c
ac
b.

(because b = a c)
(because is asociative)
(because u a = a)

Because b was an arbitrary element of S, we find that u b = b for all b S. Thus, (S, ) satisfies (i) of
Theorem 2.1.30. Consider the equation y a = u. Let d S be a solution of y a = u. Then d a = u.
This shows that (S, ) satisfies (ii) of Theorem 2.1.30. Hence, (S, ) is a group by Theorem 2.1.30.
The converse follows by Theorem 2.1.16(iv).
The next theorem gives a necessary and sucient condition for a finite semigroup to be a group.
Theorem 2.1.34 A finite semigroup (S, ) is a group if and only if (S, ) satisfies the cancellation laws
(i.e., a c = b c implies a = b and c a = c b implies a = b for all a, b, c S).
Proof. Let (S, ) be a finite semigroup satisfying the cancellation laws. Let a, b S. Consider the
equation a x = b. We show that this equation has a solution in S.
Let us write S = {a1 , a2 , . . . , an }, where the ai s are all distinct elements of S. Because S is a
semigroup, a ai S for all i = 1, 2, . . . , n. Thus,
{a a1 , a a2 , . . . , a an } S.
Suppose a ai = a aj for some i 6= j. Then by the cancellation law we have ai = aj , which is a
contradiction because ai 6= aj . Hence, all elements in {a a1 , a a2 , . . . , a an } are distinct. Thus,
S = {a a1 , a a2 , . . . , a an }.

2.1. Elementary Properties of Groups

45

Let b S. Then b = a ak for some ak S. Therefore, the equation a x = b has a solution in S.

Similarly, we can show that the equation y a = b has a solution in S. Hence, by Theorem 2.1.33, (S, )
is a group.
The converse follows by Theorem 2.1.16(iii).
Let (G, ) be a group, a G, and n Z. We now define the integral power an of a as follows:
a0
an
an

= e
= a an1 if n > 0
= (a1 )n if n < 0.

Note that an = (an )1 if n < 0.

Example 2.1.35 Let X = {a, b, c} be a nonempty set and SX be the set of all one-one functions from
X onto X. Then as in Example , (SX , ) is a group, where is the composition of functions. Consider
the function f : X X such that
f (a) = b, f (b) = c, f (c) = a.
Note that f is a one-one function of X onto X, so f SX . Now f 2 = f f. Let us determine f 2 . We
have
f 2 (a) = (f f )(a) = f (f (a)) = f (b) = c.

Similarly, f 2 (b) = a and f 2 (c) = b. Thus, f 2 SX is defined by f 2 (a) = c, f 2 (b) = a and f 2 (c) = b.

It should be pointed out that when we use additive notation for the binary operation , we speak of
multiples of an element a of the group (G, +), which are defined as follows:
0a = 0, where the 0 on the right-hand side denotes the identity of the
group (G, +) and the 0 on the left-hand side denotes the integer 0.
na = a + (n 1)a if n > 0.
na = (n)(a) if n < 0.
Note that, by the notation na, we do not mean n and a multiplied together because no multiplicative
operation between elements of Z and G has been defined.
Example 2.1.36 Consider the group (Z6 , +6 ) and [3] Z6 . We have
2[3] = [3] +6 [3] = [6] = [0].
Similarly,
3[5] = [5] +6 2[5] = [5] +6 ([5] +6 [5]) = [5] +6 [10] = [15] = [3].
Note that 15 = 2 6 + 3, so [15] = [3].
Remark 2.1.37 Let n be a fixed positive integer. Consider the group (Zn , +n ). Let [a] Zn . For any
integer k, by induction, we can show that
k[a] = [ka].
In the exercises at the end of this section, we ask the reader to verify certain basic properties of
integral powers.
Definition 2.1.38 A group (G, ) is called a finite group if G has only a finite number of elements.
The order, written |G|, of a group (G, ) is the number of elements of G.
Remark 2.1.39 Example 2.1.9 shows that for every positive integer n, there is a commutative group of
order n.
The groups in Examples 2.1.9 and 2.1.10 are finite groups.

46

2. Introduction to Groups

Definition 2.1.40 A group with an infinite number of elements is called an infinite group.
Klein and Lies use of groups in geometry influenced the turn from finite groups to infinite groups.
Example 2.1.41 The groups in Examples 2.1.7, 2.1.8, and 2.1.11 are infinite groups.
Let G be a finite group and a G. Now a2 = aa G and by induction, we can show that am G for
all m 1. Thus, {a, a2 , . . . , am , . . .} G. Because G is finite, all elements of the set {a, a2 , . . . , am , . . .}
cannot be distinct. Hence, ak = al for some positive integers k, l, k > l. This implies that akl = e.
Let us write n = k l. Therefore, an = e for some positive integer n. Also, if G is an infinite group
and a G, then it may still be possible that an = e for some positive integer n. This leads us to the
following definition.
Definition 2.1.42 Let (G, ) be a group and a G. If there exists a positive integer n such that an = e,
then the smallest such positive integer is called the order of a. If no such positive integer n exists, then
we say that a is of infinite order.
Notation 2.1.43 We denote the order of an element a of a group (G, ) by (a).
The concept of the order of an element is very important in group theory. We shall see in later
chapters how eectively information about the order of an element of a group reveals the nature of the
group and in several instances leads us to determine the structure of the group itself.
Example 2.1.44 Consider the group (Z6 , +6 ). Z6 has order 6. Consider the element [1]. Now
1[1] = [1] 6= [0],
3[1] = [3] 6= [0]
4[1] = [4] 6= [0]
6[1] = [6] = [0].

2[1] = [2] 6= [0],

4[1] = [4] 6= [0]
5[1] = [5] 6= [0]

This implies that 6 is the smallest positive integer such that 6[1] = [0]. Hence, ([1]) = 6. For [2] we have
1[2] = [2] 6= [0],

2[2] = [4] 6= [0],

3[2] = [6] = [0].

That is, 3 is the smallest positive integer such that 3[2] = [0]. Hence, ([2]) = 3.
In a similar manner, we can show that ([0]) = 1, ([3]) = 2, ([4]) = 3, and ([5]) = 6.
Example 2.1.45 Consider the group G(2, R) of Example 2.1.14. Also consider the elements

1 1
and
. Note that both these elements are in G(2, R). Now
0 1

0 1
0 1
1 0

=
.
1 0
1 0
0 1

0 1
1 1
This implies that the order of
is 2. Next, we consider
. Here
1 0
0 1

1 1
1 1
1 1
1 2
=

=
0 1
0 1
0 1
0 1
and

1 1
0 1

1 1
0 1

1 1
1 1
1 2
1 3

=
.
0 1
0 1
0 1
0 1

By induction, we can show that

1 1
1 n
=
for all positive inetegers n.
0 1
0 1

1 1
This implies that the order of
is infinite.
0 1

0 1
1 0

2.1. Elementary Properties of Groups

47

Let G be a group and a G. If (a) is infinite, then by the definition of the order of an element it
follows that (ak ) is also infinite for all k 1, i.e., the order of every positive power of a is also infinite.
If (a) is finite, then the next theorem tells us how to compute the order of various powers of a.
Theorem 2.1.46 Let (G, ) be a group and a be an element of G such that (a) = n.
(i) If am = e for some positive integer m, then n divides m.
(ii) For every positive integer t,
n
.
(at ) =
gcd(t, n)
Proof. (i) By the division algorithm, there exist q, r Z such that m = nq + r, where 0 r < n.
Now
ar = amnq = am anq = am (an )q = e (e)q = e.
Now (a) = n. Therefore, n is the smallest positive integer such that an = e. However, ar = e and
0 r < n. Thus, we must have that r = 0. This implies that m = nq. Hence, n divides m.
(ii) Let (at ) = k. Then akt = e. By (i), n divides kt. Thus, there exists r Z such that kt = nr.
Let gcd(t, n) = d. Then there exist integers u and v such that
t = du, n = dv, and gcd(u, v) = 1
by Exercise 9 (page 16).
Now kt = nr implies that kdu = dvr. Thus, ku = rv, i.e., v divides ku. Now gcd(u, v) = 1 and v
divides ku. So v divides k. Because nd = v, we have nd divides k.
Now
n

nt

(at ) d = a d = a

ndu
d

= anu = (an )u = eu = e.

We therefore have (at ) = k and (at ) d = e . Therefore, as in (i), k divides nd . Because k and
positive integers such that k divides nd and nd divides k, we must have k = nd . Hence,
(at ) = k =

n
d

are

n
n
=
.
d
gcd(t, n)

Definition 2.1.47 A group (G, ) is called a torsion group if every element of G is of finite order. If
every nonidentity element of G is of infinite order, then G is called a torsion-free group.
Exercise 2.1.48 (i) The group of Example 2.1.44 is a torsion group.
(ii) The groups (R, +), (R+ , ), (Q+ , ) are torsion-free groups.
(iii) Consider the group (R\{0}, ). Now (1)2 = 1, so (1) = 2. However, if x R\{0} such that
x 6= 1 and x 6= 1, then x is of infinite order. It now follows that the groups (R\{0}, ) has elements of
finite as well as infinite orders. Hence, (R\{0}, ) is neither a torsion group nor a torsion-free group.
We close this chapter with the following example. The ideas set forth in this example are due to
Klein.
Example 2.1.49 Imagine a square having its sides parallel to the axes of a coordinate system and its
center at the origin.

48

2. Introduction to Groups

v
d2
1

2
h
0

d1

Figure 2-1 Rigid motions of a square

We label the vertices as in the figure and we allow the following rigid motions of the square: clockwise
rotations of the square about the center and through angles of 90 , 180 , 270 , 360 , say, r90 , r180 , r270 ,
r360 , respectively; reflections h and v about the horizontal and vertical axes; reflections d1 , d2 about the
diagonals. The diagrams in Figure 2-2 should prove helpful.
1

r 360
4

r 90
4

h
2

r 270
4

d2

r 180
4

1
d1

Figure 2-2 Rigid motions of a square

A multiplication on two rigid motions can be defined by performing two such motions in succession.
For example, r90 h is determined by first performing motion h and then the motion r90 , see Figure 2-3.
1

h
4

r90
1

Figure 2-3 Motion h followed by the motion r90 ; r90 h

2.1. Elementary Properties of Groups

49

In Figure 2-3, we see that r90 h = d1 . The complete multiplication table for the operation follows.

r360
r90
r180
r270
h
v
d1
d2

r360
r360
r90
r180
r270
h
v
d1
d2

r90
r90
r180
r270
r360
d2
d1
h
v

r180
r180
r270
r360
r90
v
h
d2
d1

r270
r270
r360
r90
r180
d1
d2
v
h

h
h
d1
v
d2
r360
r180
r90
r270

v
v
d2
h
d1
r180
r360
r270
r90

d1
d1
v
d2
h
r270
r90
r360
r180

d2
d2
h
d1
v
r90
r270
r180
r360

(2.1)

We leave it for the reader to verify that the set of rigid motions of a square is a group under the operation
. We also note that r360 is the identity element.
This group of rigid motions of a square is known as the group of symmetries of the square. Let
us denote this group by Sym. Then
Sym = {r360 , r90 , r180 , r270 , h, v, d1 , d2 }.
Now h r270 = d1 6= d2 = r270 h. Therefore, the group Sym is noncommutative.
Let us now determine the order of the elements. Consider r90 . Now
2
3
2
r90
= r90 r90 = r180 , r90
= r90
r90 = r270 ,

and
Thus, (r90 ) = 4. Next, we have

4
3
r90
= r90
r90 = r360 .

2
(r180 ) = (r90
)
4
= gcd(4,2)
= 42
= 2.

(by Theorem 2.1.46)

Similarly,
(r270 ) = 4, (h) = 2, (v) = 2, (d1 ) = 2, and (d2 ) = 2.

Let us write = r90 and = d2 . Then

2 = r180 , 3 = r270 , 4 = r360 , = v, 2 = d1 , and 3 = h.

Also, note that
Thus, we see that

= 1 = 3 .
Sym = {e, , 2 , 3 , , , 2 , 3 }.

Finally, we make the following observations. Consider r90 . We can think of r90 as a one-one function
of {1, 2, 3, 4} onto {1, 2, 3, 4} by defining
r90 (1) = 2, r90 (2) = 3, r90 (3) = 4, r90 (4) = 1.
In a similar manner, we can consider other rigid motions of the square as one-one functions of {1, 2, 3, 4}
onto {1, 2, 3, 4}.
Remark 2.1.50 A fundamental phenomenon of nature is that of symmetry. A figure or an object is
said to have a symmetry if a rotation, a translation, an inversion, a minor reflection, or a combination
of these operations leaves the figure or object indistinguishable from its original position. The 1890s saw
the first application of group theory to the natural and physical sciences. An important application of
group theory was to crystallography. Groups were used to give a theoretical classification of the dierent
kinds of symmetry arrangements possible within crystalline matter 20 years before experimental means
were available for analyzing the crystals themselves.

50

2. Introduction to Groups

Remark 2.1.51 Group theory is used in quantum mechanics. It is used to study the atoms internal
structure. In the 1950s, a new generation of particle accelerators produced a variety of subatomic particles. Group theory was used to predict the existence of a tenth nucleon in a tenfold symmetry scheme
of nucleons of which nine particles had already been detected. In 1964, the tracks of Omega-Minus, the
tenth nucleon, were identified.

Worked-Out Exercises
Exercise 1 Let G = {a R | 1 < a < 1}. Define a binary operation on G by
ab=

a+b
1 + ab

for all a, b G. Show that (G, ) is a group.

Solution: Note that 1 < x < 1 if and only if x2 < 1 for all x R.

Let a, b G. First we show that a b G. Now a2 < 1 and b2 < 1. Thus,

(1 a2 )(1 b2 ) > 0.
This implies that
1 a2 b2 + a2 b2 > 0.

Now (1 + ab)2 (a + b)2 = 1 + a2 b2 + 2ab a2 b2 2ab = 1 a2 b2 + a2 b2 > 0, so

(

a+b 2
) < 1.
1 + ab

Therefore, a b G. Hence, G is closed under . We now show that is well defined. Let
a, b, c, d G and (a, b) = (c, d). Then a = c and b = d. Thus,
ab=

a+b
c+d
=
= c d.
1 + ab
1 + cd

So is well defined. To show that is associative, let a, b, c G. Now

(a b) c =

a+b
+c
a + b + c + abc
a+b
=
c = 1+aba+b
.
1 + ab
1
+ ab + ac + bc
1 + ( 1+ab )c

Similarly,
a (b c) =

a + b + c + abc
.
1 + ab + ac + bc

Therefore, (ab)c = a(bc), so is associative. So far, we have shown that (G, ) is a semigroup.
Now 0 G and

0a=

0+a
=a
1 + 0a

for all a G.

This shows that (G, ) satisfies (i) of Theorem 2.1.30.

Let a G. Then a G and

(a) a =

a + a
= 0.
1 + (a)a

Thus, (G, ) satisfies (ii) of Theorem 2.1.30. Consequently, by Theorem 2.1.30, (G, ) is a group.
Exercise 2 Let G = {(a, b) | a, b R, a 6= 0} = R\{0}R. Define a binary operation on G by
(a, b) (c, d) = (ac, b + d)
for all (a, b), (c, d) G. Show that

2.1. Elementary Properties of Groups

51

(a) (G, ) is a group,

(b) G has exactly one element of order 2,

(c) G has no elements of order 3.
Solution: (a) As in Worked-Out Exercise 1, we show that (G, ) satisfies the conditions of Theorem 2.1.30.
Let (a, b), (c, d) G. Then a 6= 0 and c 6= 0, so ac 6= 0. Thus, (a, b) (c, d) = (ac, b + d) G.
Hence, G is closed under .
It is a direct computation to verify that is well defined and associative, so we ask the reader
to do the verification. Now (1, 0) G and
(1, 0) (a, b) = (1a, 0 + b) = (a, b)

for all (a, b) G.

This implies that (G, ) satisfies (i) of Theorem 2.1.30.

Let (a, b) G. Then a 6= 0, so a1 R and a1 6= 0. Thus, ( a1 , b) G. Now
1
1
( , b) (a, b) = ( a, b + b) = (1, 0).
a
a
This implies that (G, ) satisfies (ii) of Theorem 2.1.30.
By Theorem 2.1.30, we can conclude that (G, ) is a group.

(b) First note that (1, 0) G and

(1, 0) (1, 0) = (1, 0).

Thus, (1, 0) is of order 2. We now show that this is the only element of order 2 by showing
that if (a, b) is any other element of G of order 2, then (a, b) = (1, 0).
Let (a, b) G be an element of order 2. Then
(a, b) (a, b) = (1, 0)
implies that
(a2 , b + b) = (1, 0).
Therefore, a2 = 1 and b = 0. Now a2 = 1 implies that a = 1. If a = 1, then (a, b) = (1, 0),
which is a contradiction because (1, 0) is of order 1. Hence, a = 1. This implies that (a, b) =
(1, 0). It now follows that (1, 0) is the only element of order 2.
(c) Suppose that (a, b) is an element of order 3. Then
(a, b) (a, b) (a, b) = (1, 0).
This implies that
(a3 , 3b) = (1, 0).
Thus, a3 = 1 and b = 0. Now a3 = 1 implies that a = 1. Hence, (a, b) = (1, 0). This means that
(1, 0) is of order 2. However, (1, 0) is of order 1, so we have a contradiction. Consequently, G
has no element of order 3.
Exercise 3 Let G be the set of all rational numbers except 1. Show that (G, ) is a group where
a b = a + b + ab
for all a, b G.
Solution: As in Worked-Out Exercise 1, we show that (G, ) satisfies the conditions of Theorem 2.1.30. Our
first step is to show that is well defined. Let a, b, c, d G and (a, b) = (c, d). Then a = c and
b = d. Thus,
a b = a + b + ab = c + d + cd = c d.

52

2. Introduction to Groups
Hence, is well defined. Let a, b G. Then a 6= 1 and b 6= 1. We now show that a b G by
showing that a b 6= 1 and a b is a rational number. Suppose a b = 1, i.e., a + b + ab = 1.
Then
(a + 1)(b + 1) = 0.
This implies that either (a + 1) = 0 or (b + 1) = 0, i.e., either a = 1 or b = 1, which is a
contradiction. Therefore, a b 6= 1.
Now the addition and multiplication of rational numbers is a rational number, so it follows that
a b is a rational number. Hence, a b G. Thus, is a binary operation on G.

Let a, b, c G. Then

(a b) c =
=
=
=
=

(a + b + ab) c
a + b + ab + c + ac + bc + abc
a + (b + c + bc) + a(b + c + bc)
a + b c + a(b c)
a (b c).

This shows that is associative. Thus, (G, ) is a semigroup. Now 0 G and

0a=0+a+0a=a
for all a G. Hence, (G, ) satisfies (i) of Theorem 2.1.30. Now for all a G, a + 1 6= 0. Note that
a
a
a+1
6= 1. Therefore, a+1
G and

a
a
a
a + a + a2 a2
a=
+ a + (
)a =
= 0.
a+1
a+1
a+1
a+1

This implies that (G, ) satisfies (ii) of Theorem 2.1.30. Hence, by Theorem 2.1.30, (G, ) is a
group.
Exercise 4 Let G be a group and x G. Suppose (x) = mn, where m and n are relatively prime. Show that
there exist y, z G such that x = y z = z y and (y) = m and (z) = n.
Solution: Because gcd(m, n) = 1 there exist s, t Z such that 1 = ms + nt. Now x = xms+nt = xms xnt .
Let y = xnt and z = xms . Then x = y z = z y. Now y m = (xnt )m = xmnt = e. Hence, (y)
divides m. Similarly, (z) divides n. Suppose (y) = m1 and (z) = n1 . It is an easy exercise to
verify that (y z)l = y l z l for all positive integers l. Thus, xm1 n1 = (y z)m1 n1 = y m1 n1 z m1 n1 =
e e = e. Hence, mn | m1 n1 . But because m1 | m and n1 | n, we must have m = m1 and n = n1 .
Exercise 5 Let (G, ) be a group of even order. Show that there exists a G such that a 6= e, a2 = e.
/ A. If g A, then g 1 A, i.e., elements of A occurs in
Solution: Let A = {g G | g 6= g 1 } G. Then e
pairs. Therefore, the number of elements in A is even. This implies that the number of elements
in {e} A is odd. Because the number of elements in G is even and {e} A G, there exists
a G such that a
/ {e} A. But then a 6= e and a
/ A. Hence, there exists a G such that a 6= e
and a = a1 or a2 = e.
Exercise 6 Let (G, ) be a group and a, b G. Suppose that a b = b a1 and b a = a b1 . Show that
a4 = b4 = e.
Solution: Because a b = b a1 , a = b a1 b1 . Similarly, b = a b1 a1 . Thus, b a = a b1 =
(b a1 b1 ) b1 = b a1 b2 . Multiply both sides of the equation b a = b a1 b2 by b1 to
get a = a1 b2 . This implies that a2 = b2 . Hence, a4 = a2 a2 = a2 b2 = a (a b1 ) b1 =
a (b a) b1 = (a b) a b1 = (b a1 ) a b1 = b (a1 a) b1 = b e b1 = e. Also,
b4 = a4 = e.
Exercise 7 Let (G, ) be a group and a, b G. Suppose that a bn = bn+1 a and b an = an+1 b for some
n N. Show that a = b = e.

2.1. Elementary Properties of Groups

53

Solution: Multiply both sides of the equation a bn = bn+1 a by bn to get a = bn+1 a bn . Thus,
a2 = aa = a bn+1 abn = (a bn )babn = (bn+1 a)babn = bn+1 (ab)abn . Now
a3 = aa2 = a(bn+1 (ab)abn ) = (abn )b(ab)abn = (bn+1 a)b(ab)abn =
bn+1 (a b)2 a bn . Hence, we see that we could use induction to obtain
an = bn+1 (a b)n1 a bn
for all n N. Also,

b an

=
=
=
=
=
=
=

which implies that

(2.2)

an+1 b
a an b
a (bn+1 (a b)n1 a bn ) b
a bn+1 (a b)n1 a b1n
(a bn ) b (a b)n1 a b1n
(bn+1 a) b (a b)n1 a b1n
bn+1 (a b)n a b1n ,

an = bn (a b)n a b1n .

(2.3)

From Eqs. (2.2) and (2.3),

bn+1 (a b)n1 a bn = bn (a b)n a b1n ,
which implies that
b (a b)n1 a = (a b)n a b = (a b)n+1 .
Thus,
(a b)n+1

= b (a b)n1 a
= b ((a b) (a b)) a
{z
}
|
n1 tim es

(2.4)

= (b a) (b a)
{z
}
|
n times

Interchange the role of a and b to get

= (b a)n .

(b a)n+1 = (a b)n .

(2.5)

Hence, (a b)n = (b a)n+1 = (b a)n (b a) = (a b)n+1 (b a), so e = (a b) (b a), which

implies that
a2 = b2 .
(2.6)
Now
b an = b a2 an2 = b b2 an2 = b1 an2

(2.7)

an+1 b = an1 a2 b = an1 b2 b = an1 b1 .

(2.8)

and
1

Thus, from Eqs. (2.7) and (2.8) it follows that b

n2

n1

=a

an1 = b1 an2 b = (b1 a b)n2 .

, so
(2.9)

Now b an = an+1 b implies that

an = (b1 a b)n+1 .

(2.10)

Hence, an = (b1 a b)n+1 = (b1 a b)n2 (b1 a b)3 = an1 (b1 a b)3 , which implies
that a = (b1 a b)3 = b1 a3 b. Thus, a3 b = b a. Therefore, b a = a3 b = a a2 b =
a b2 b = a b1 by Eq. (2.6). That is, we have
b a = a b1 .

(2.11)

54

2. Introduction to Groups
Similarly,
a b = b a1 .

(2.12)

Now a b = b a1 implies that a b a = b. Thus, b = a b a = a a b1 [by Eq. (2.11)]. Hence,

a2 = b2 .

Suppose n is even. Then a2 = b2 implies that an = bn . Hence, a bn = bn+1 a implies that

an+1 = an b a, so b = e. Similarly, a = e. Suppose n is odd. Let n = 2k + 1. Then a2k = b2k .
Now a bn = bn+1 a a b2k+1 = b2k+2 a a a2k b = a2k+2 a. Thus, b = a2 = b2 . Hence,
b = e. Similarly, a = e.
Exercise 8 (Hays) Let (S, ) be a semigroup. Show that S is a group if and only if for all a S there exists
a unique b S such that a b a = a.
Solution: Suppose for all a S, there exists a unique b S such that a b a = a. Let a S. Then there
exists b S such that a b a = a. Thus, a b a b = a b, so (a b)2 = a b. Hence, S has an
idempotent element. If (S, ) is to be a group, then it can have only one idempotent (Corollary
2.1.17), namely, the identity element. Therefore, first we show that S has only one idempotent.
Suppose e and f are two idempotents in S. Because e f S, there exists a unique g such
that (e f ) g (e f ) = e f. Now (e f ) (g e) (e f ) = (e f ) g (e e) f =
(ef )g e2 f = (ef )g (ef ) = ef. Because g is unique such that (ef )g(ef ) = (ef ), it
follows that ge = g. Similarly, because (ef )(f g)(ef ) = (ef )g(ef ) = ef, the uniqueness
of g implies that f g = g. Also, (e f ) (g (e f ) g) (e f ) = ((e f ) g (e f )) g (e f ) =
(e f ) g (e f ). Again, the uniqueness of g implies that g (e f ) g = g. Hence, g 2 = g g =
(g e) (f g) = g (e f ) g = g. Thus, g is an idempotent. Now g = g g g and g (e f ) g = g.
Hence, by the uniqueness of the middle element g = e f. Therefore, e f is an idempotent. Now
(ef )f (ef ) = (e(f f ))(ef ) = (ef )(ef ) = ef and similarly (ef )e(ef ) = ef.
By the uniqueness of the middle element, it follows that e = f. Hence, S has a unique idempotent
element.
Let e be the idempotent element of S. Let a S. Then there exists b S such that a b a = a,
which implies that (a b)2 = a b. Hence, a b = e. Also, a b a = a implies that b a b a = b a.
Thus, b a is an idempotent. Hence, b a = e. Also, a b a = a together with a b = e = b a
implies that e a = a = a e. Therefore, e is the identity element. Because a b = e = b a, b is
an inverse of a. Consequently, (S, ) is a group.

Conversely, suppose (S, ) is a group. Let a S. Note that a a1 a = a. This shows the
existence of an element b S such that a b a = a, namely, b = a1 . To show the uniqueness,
suppose there exist b, c S such that a b a = a and a c a = a. Then a b a = a c a and
by the cancellation laws, b = c. Thus, b is unique such that a b a = a.

Exercises
1. Which of the following mathematical systems are semigroups? Which are groups?
(a)
(b)
(c)
(d)
(e)
(f)
(g)

(N, ), where a b = a for all a, b N.

(Z, ), where a b = a b for all a, b Z.
(R, ), where a b = |a|b for all a, b R.
(R, ), where a b = a + b + 1 for all a, b R.
(R, ), where a b = a + b ab for all a, b R.
(Q,), where a b = ab
2 for all a, b Q.
(G, ), where

a b
a b
0 0
G=
|
6=
and a, b R
b a
b a
0 0
and is the usual matrix multiplication.

2.1. Elementary Properties of Groups

55

(h) (G, ), where G is the set of all matrices of the following form over Z

1 a b
0 1 c
0 0 1
and is the usual matrix multiplication.

2. Let G = {(a, b) | a, b R, b 6= 0}. Define a binary operation on G by (a, b) (c, d) = (a + bc, bd)
for all (a, b), (c, d) G. Show that (G, ) is a noncommutative group.

a b
3. Let G =
| a, b, c, d R, ad bc = 1 . Show that G is a group under usual matrix
c d
multiplication. (This group is usually denoted by SL(2, R) and is called the special linear group
of degree 2.)

1 n
4. Let G =
| n Z . Show that (G, ) is a commutative group, where denotes the usual
0 1
matrix multiplication. Also, show that (G, ) is torsion-free.
5. In Z14 , find the smallest positive integer n such that n[6] = [0].
6. Find an element [b] Z9 such that [8] 9 [b] = [1]. Does [b] U9 ?
7. In U24 , find the smallest positive integer n such that [7]n = [1].
8. Describe U6 , U9 , U12 , U24 of Example 2.1.10.
9. Let p be a prime. Show that Up = Zp \{[0]}.
10. Let Un = {[a] Zn \{[0]}| gcd(a, n) = 1}. Show that (Un , n ) is a group, where n is the multiplication modulo n.
11. Show that Un = {[a] Zn \{[0]} | additive order of [a] = n }.
12. Let (G, ) be a group and a, b G. Suppose that a2 = e and a b4 a = b7 . Show that b33 = e.
13. Let (G, ) be a group and a, b G. Suppose that a1 b2 a = b3 and b1 a2 b = a3 . Show that
a = b = e.
14. Let (G, ) be a group. If a, b G are such that a4 = e and a2 b = b a, show that a = e.
15. Let (G, ) be a group and x, a, b G. Let c = x a x1 and d = x b x1 . Show that a b = b a
if and only if c d = d c.
16. Let (G, ) be a group such that a2 = e for all a G. Show that G is commutative.
17. Prove that a group (G, ) is commutative if and only if (a b)1 = a1 b1 for all a, b G.
18. Let (G, ) be a group. Prove that if (a b)2 = a2 b2 for all a, b G, then (G, ) is commutative.
19. Prove that a group (G, ) is commutative if and only if for all a, b G, (a b)n = an bn for any
three consecutive integers n.
20. Let (G, ) be a group. If G has only two elements, prove that G is commutative.
21. Let (G, ) be a group and a, b, c G. Find an element x G such that a x b = c. Is x unique?
22. Let (G, ) be a group and a, b G. Show that (a b a1 )n = a bn a1 for all integers n.
23. Let (G, ) be a finite group and a G. Show that there exists n N such that an = e.
1
24. If (G, ) is a group and a1 , . . . , an G, prove that (a1 an )1 = a1
n a1 .

56

2. Introduction to Groups

25. Let (G, ) and (H, ) be groups. Define the operation on G H = {(a, b) | a G, b H} by
(a, b) (c, d) = (a c, b d). Prove that (G H, ) is a group. If (G, ) and (H, ) are commutative,
prove that (G H, ) is commutative. The group (G H, ) is called the direct product of G
and H.
26. Let (G, ) be a finite group and a G. Show that (a) |G| .
27. Let (G, ) be a group and a, b G.
(a) Prove that a and a1 have the same order.
(b) Prove that a and b a b1 have the same order.
(c) Prove that a b and b a have the same order.

28. Let (G, ) be a group and a, b G.

(a) Suppose that a b = b5 a3 . Show that (b a1 ) = (b5 a) = (b3 a3 ).

(b) Generalize (a) to arbitrary powers of a and b.

29. Let (G, ) be a group, a G and (a) = n. Let 1 p n be such that p and n are relatively
prime. Show that (ap ) = n.
30. Let (G, ) be a group, a G, and (a) = p, where p is a prime.
(a) Prove that (ak ) = p for all 1 k < p.

(b) Prove that for all m N, either am = e or (am ) = p.

31. Let (G, ) be a group and a G. Suppose that (a) = n and n = mk for some m, k N. What is
(ak )?
32. Let (G, ) be a group.
(a) Let a, b G, (a) = n, (b) = m, gcd(m, n) = 1, and a b = b a. Show that (a b) = mn.

(b) Let ai G, (ai ) = ni , 1 i m. Suppose gcd(ni , nj ) = 1 and ai aj = aj ai for all i and j.

Let x = a1 a2 am . Show that (x) = n1 n2 nm .
6 j, ni and nj
33. Let (G, ) be a group and x G. Suppose (x) = n = n1 n2 nk , where for all i =
are relatively prime. Show that there exists xi G such that (xi ) = ni for all i = 1, 2, . . . , k,
x = x1 x2 xk and xi xj = xj xi for all i and j.
34. Let G = {(a, b) | a, b R, a 6= 0}. Then G is a group under the binary operation (a, b) (c, d) =
(ac, bc + d) for all (a, b), (c, d) G. Prove that G has infinitely many elements of order 2, but G
has no element of order 3.
35. Let a, b Sym. As remarked in Example 2.1.49, every rigid motion of the square can be considered
a one-one function of {1, 2, 3, 4} onto itself. Consider a b as a function. Show that a b = a b,
where represents the binary operation of rigid motions of the square and is the composition of
functions.
36. Let (S, ) be a finite semigroup. Prove that there exists a S such that a2 = a.
37. Let (G, ) be a finite semigroup with identity. Prove that (G, ) is a group if and only if G has
only one element a such that a2 = a.
38. Prove that a semigroup (S, ) is a group if and only if a S = S and S a = S for all a S, where
a S = {a s | s S} and S a = {s a | s S}.
39. Prove that a semigroup (S, ) is a group if and only if

2.1. Elementary Properties of Groups

57

(a) there exists e S such that a e = a for all a S, and

(b) for all a S there exists b S such that a b = e.

40. Rewrite the statements and proofs of the theorems in this chapter using additive notation.
41. Let (G, ) be a group, a, b G and m, n Z. Prove that
(a) an am = an+m = am an ,

(b) (an )m = anm ,

(c) an = (an )1 ,
(d) en = e,
(e) (a b)n = an bn , if (G, ) is commutative.
42. Write the proof if the following statements are true; otherwise, give a counterexample.
(a) Let T (S) be the set of all functions on S = {1, 2, 3}. T (S) is a group under composition of
functions.

a b
(b) M2 (R) =
| a, b, c, d R is a group under usual matrix multiplication.
c d
(c) Every group of four elements is commutative.

(d) A group has only one idempotent element.

(e) A semigroup with only one idempotent is a group.
(f) If a semigroup S satisfies the cancellation laws, then S is a group.

58

2. Introduction to Groups

Niels Henrik Abel (18021829) was born on August 5, 1802, in Finny, Norway. He was the second
of six children. Abel and his brothers received their first education from their father.
At the age of 13, Abel along with his older brother, was sent to the Cathedral sch-ool in Christiania
(Oslo). In 1817, his mathematics teacher was Bernt Michael Holmb, who was seven years older than
Abel. Holmb recognized Abels talent and started giving him special problems and recommended special
books outside the curriculum. Abel and Holmb read the calculus text of Euler and the work of Lagrange
and Laplace. Soon Abel became familiar with most of the important mathematical literature.
Abels father died when he was 18 years old and the responsibility of supporting the family fell
on his shoulders. He gave private lessons and did odd jobs. However, he continued to carry out his
mathematical research.
Abel, in his last year of school, attacked the problem of the solvability of the quintic equation, a
problem that had been unsettled since the sixteenth century. Abel thought that he had solved the
problem and submitted his work for publication. Unable to find an error and understand his arguments,
he was asked by the editor to illustrate his method. In 1824, during the process of illustration he
discovered an error. This discovery led Abel to a proof that no such solution exists. He also worked on
elliptic functions and in essence revolutionized the theory of elliptic functions.
He traveled to Paris and Berlin in order to find a teaching position. Then poverty took its toll, and
Abel died from tuberculosis on April 6, 1829. Two days later a letter from Crelle reached his address,
conveying the news of his appointment to the professorship of mathematics at the University of Berlin.
Abel is honored by such terms as Abelian group and Abelian function.

Chapter 3

Permutation Groups
Permutation groups is one of the specialized theories of groups which arose from the source, classical
algebra, in the evolution of group theory.

3.1

Permutation Groups

As stated earlier, there are four major sources from which abstract group theory evolved. Mathematicians interest in finding formulas to solve polynomial equations by means of radicals led some
mathematicians to the study of permutations of the roots of rational functions. Lagrange, Rufini, and
Cauchy were among the earlier mathematicians to work with permutation groups. However, it was
Cauchy whose systematic study of permutation groups (between 1815 and 1845) is believed, by some,
to be the origin of abstract group theory. Many of the concepts and major results in this chapter are
due to Cauchy.
We begin our study of permutation groups by defining what a permutation is.
Definition 3.1.1 Let X be a nonempty set. A permutation of X is a one-one function from X
onto X.
Example 3.1.2 (i) Let X be a nonempty set. Define : X X by (x) = x for all x X. Then
is one-one function of X onto X. Thus, is a permutation of X. Nothe that is called the identity
permutations and is, usually, denoted by iX or e.
(ii) Let X = {a, b, c}. Define : X X such that (a) = b, (b) = a, and (c) = c. By the definition
of it follows that is is one-one function of X onto X. Thus, is a permutation of X.
(iii) Consider R, the set of real numbers. Define : R R by (x) = 3x + 5 for all x R. It can be
shown that is a one-one function of R onto R. Thus, is a permutation of R. Similarly, if : R R
by (x) = x3 for all x R. It can be shown that is a one-one function of R onto R. Thus, is a
permutation of R.
Definition 3.1.3 A group (G, ) is called a permutation group on a nonempty set X if the elements
of G are permutations of X and the operation is the composition of two functions.
Example 3.1.4 Let X be any nonempty set and SX be the set of all one-one functions from X onto X,
as defined in Example 2.1.13. Then (SX , ) is a group as we have shown in Example 2.1.13, where is
the composition of functions. Hence, (SX , ) is a permutation group.
Example 3.1.5 Let X = {1, 2}. Define : X X such that (1) = 1, (2) = 2. Then is a one-one
function of X onto X, so is a permutation of X. Next define : X X such that (1) = 2 and
(2) = 1. Then is a one-one function of X onto X, so is a permutation of X. Let SX = {, }.
Then (SX , ) is a group, where is the composition of functions. Note that on this set X, and b are
the only permutations on X. Moreover, is the identity permutation and 1 = .
59

60

3. Permutation Groups

In this chapter, and in fact in this text, our study of permutation groups will focus on permutation
groups on finite sets, i.e., X is a finite set.
Before we consider more examples of permutation groups, let us fix some notation which will be
useful when working with permutations.
Let In = {1, 2, . . . , n}, n 1. Let be a permutation on In . Then
= {(1, (1)), (2, (2)), . . . , (n, (n))}.
(Recall that a function f : A A is a subset of A A.) It is sometimes convenient to describe a
permutation by means of the following notational device:

1
2
3

n
=
.
(1) (2) (3) (n)
This notation is due to Cauchy and is called the two-row notation. In the upper row, we list all the
elements of In and in the lower row under each element i In , we write the image of the element, i.e.,
(i).
Example 3.1.6 Let n = 4 and be the permutation on I4 defined by (1) = 2, (2) = 4, (3) = 3, and
(4) = 1. Then using the two-row notation we can write

1 2 3 4
=
.
2 4 3 1
As we shall see, the two-row notation of permutations is quite convenient while doing computations
such as determining the composition of permutations.
Let n = 7 and and be two permutations on I7 defined by

1 2 3 4 5 6 7
=
1 3 4 6 7 2 5
and
=

1 2 3 4 5 6 7
2 5 3 1 7 6 4

Let us compute . Now by the definition of the composition of functions

( )(i) = ((i))
for all i I7 . Thus,

( )(1) = ((1)) = (2) = 3,

( )(2) = ((2)) = (5) = 7,

and so on. From this, it is clear that when determining, say, ( )(1), we start with and finish with
and read as follows: 1 goes to 2 (under ) and 2 goes to 3 (under ), so 1 goes to 3 (under ). We
can exhibit this in the following form:

123

257

334

411

575

662

746
Thus,
=

1 3

2 7

3 4

4 1

5 5

6 2

7 6.

1 2 3 4 5 6 7
3 7 4 1 5 2 6

3.1. Permutation Groups

61

Example 3.1.7 Let n = 6 and and be permutations on I6 defined by

1 2 3 4 5 6
=
3 1 4 6 5 2
and
=
Let us first determine . Now

1 2 3 4 5 6
1 3 5 4 2 6

1 1 3, i.e., 1 3.
Similarly,

2 4, 3 5, 4 6, 5 1, 6 2.
Thus,
=

1 2 3 4 5 6
3 4 5 6 1 2

Similarly, for ; 1 3 5, i.e., 1 5, and so on. In this case, we start with and finish with .
Note that

1 2 3 4 5 6
=
.
5 1 4 6 2 3
We note that 6= .
Let Sn denote the set of all permutations on In , n 1.
Example 3.1.8 In this example, we describe S3 , i.e., the set of all permutations on I3 = {1, 2, 3}. From
Exercise 8 (page 31), we know that the number of one-one functions
of I3 onto I3 is 3! = 6. Thus,

1 2 3
|S3 | = 6. Let e denote the identity permutation on I3 , i.e., e =
. Let 1 be a nonidentity
1 2 3
permutation on I3 . Let us see some of the choices for 1 . Suppose 1 (1) = 1. If 1 (2) = 2, then we must
have 1 (3) = 3 because 1 is a permutation. In this
case, wesee that 1 = e, a contradiction. Thus, we
1 2 3
must have 1 (2) = 3 and 1 (3) = 2, i.e., 1 =
. In a similar manner, we can show that
1 3 2

1 2 3
1 2 3
1 2 3
the other four permutations on I3 are 2 =
, 3 =
, 4 =
,
2 1 3
3 2 1
2 3 1

1 2 3
and 5 =
. Thus,
3 1 2
S3 = {e, 1 , 2 , 3 , 4 , 5 }.
Let us denote 2 by and 4 by . We ask the reader to check that
2 = 5 , = 1 , and 2 = 3 .
Hence, we can write
S3 = {e, , 2 , , , 2 }.
Because (S3 , ) is also a group, we ask the reader to show that () = 2 and () = 3 by showing that
2 = e and 2 6= e, but 3 = e.
In the previous example, the permutation group (S3 , ) consisted of all the permutations on the set
I3 . Next, we give an example of a permutation group that does not contain all the permutations on a
given set.

62

3. Permutation Groups

Example 3.1.9 Let n = 4 and consider I4 = {1, 2, 3, 4}. Recall that in Example 2.1.49, we remarked
that rigid motions of the square can be viewed as permutations on I4 . Let S be the set of all permutations
that corresponds to the rigidmotions of the
square. We will use the same notation for the permutations,
1 2 3 4
i.e., r90 is the permutation
, r360 is the identity permutation, and so on. By Exercise
2 3 4 1
35 (page 56), it follows that the multiplication table of (S, ) is the same as the multiplication table of the
group (Sym, ). Now composition of functions is associative and from the multiplication table, it follows
that S is closed under , r360 is the identity of (S, ), and every element of S has an inverse. Thus,
(S, ) is a group. Hence, the group of symmetries of a square can be thought of as a permutation group
on I4 .
The following theorem describes some basic properties of Sn .
Theorem 3.1.10 (i) (Sn , ) is a group for any positive integer n 1.
(ii) If n 3, then (Sn , ) is noncommutative.
(iii) |Sn | = n!
Proof. (i) We have already noted that the set of all one-one functions of any nonempty set onto
itself forms a group under composition of functions in Example 2.1.13. Thus, (Sn , ) is a group for any
positive integer n 1.
(ii) Let n 3. Let , Sn be defined by

1 2 3 4 n
1 2 3 4 n
=
and =
.
1 3 2 4 n
3 2 1 4 n
Now
=

1 2 3 4
2 3 1 4

n
n

1 2 3 4
3 1 2 4

n
n

and

Thus, ( )(1) = 2 6= 3 = ( )(1). Hence, 6= , so Sn is noncommutative.

(iii) This follows from Exercise 8 (page 31).
Definition 3.1.11 The group (Sn , ) is called the symmetric group on In .

1
2

n
Consider the permutation =
. If (i) = i, then we drop the column
(1) (2) (n)

i
1 2 3 4
2 4
. For example, =
is denoted by
.
(i)
1 4 3 2
4 2
Definition 3.1.12 Let be an element of Sn . Then is called a k-cycle, written (i1 i2 ik ), if

i1 i2 ik1 ik
=
,
i2 i3
ik
i1
i.e., (ij ) = ij+1 , j = 1, 2, . . . , k 1, (ik ) = i1 , and (a) = a for any other element of In .
Note that if = (i1 i2 ik ), then

= (i1 i2 ik )
= (i2 i3 ik i1 )
..
.
= (ij ij+1 ik i1 ij1 ).

A k-cycle is called a transposition when k = 2.

3.1. Permutation Groups

63

We know that in Example 3.1.9, the permutation r90 is a 4-cycle and d2 is a 2-cycle. We write
r90 = (1 2 3 4)
and
d2 = (1 3).
The identity of Sn is sometimes denoted by (1) or e.
Example 3.1.13 Using the cycle notation, we can write
S3 = {e, (1 2), (1 3), (2 3), (1 2 3), (1 3 2)}.
We now note some of the properties of the group (S3 , ).
(i) (S3 , ) is a noncommutative group of order 6 by Theorem 3.1.10.
(ii) S3 contains two elements of order 3; for (1 2 3) (1 2 3) = (1 3 2) 6= e and (1 2 3) (1 2 3) (1
2 3) = e. Hence, the order of (1 2 3) is 3. Similarly, the order of (1 3 2) is 3. The order of (1 2), (1 3),
and (2 3) is 2 because (1 2) (1 2) = e, (1 3) (1 3) = e, and (2 3) (2 3) = e.
(iii) In S3 , the product of distinct elements of order 2 is an element of order 3. (1 2) (2 3) = (1 2
3), (1 3) (1 2) = (1 2 3), (1 2) (1 3) = (1 3 2), (2 3) (1 2) = (1 3 2), (1 3) (2 3) = (1 3 2), and (2
3) (1 3) = (1 2 3).
Definition 3.1.14 Let , Sn . Then and are called conjugate if there exists Sn such that
1 = .
The following theorem shows how to compute the conjugate of a cycle.
Theorem 3.1.15 Let = (i1 i2 il ) Sn be a cycle. Then for all Sn ,
1 = ((i1 ) (i2 ) (il )).
Proof. Because Sn , is a one-one mapping of In onto In . Thus, the elements (1), . . . , (n) In
are all distinct, so In = {(1), (2), . . . , (n)}. Let r be any integer such that 1 r < l. Then
( 1 )((ir )) = ((1 ((ir ))))
= ((ir ))
= (ir+1 ).
Also, ( 1 )((il )) = ((1 ((il )))) = ((il )) = (i1 ). Now let a In be such that a 6= (ir )
for all r, 1 r l. Then 1 (a) In and 1 (a) 6= ir for all r, 1 r l, so (1 (a)) = 1 (a). Thus,
( 1 )(a) = ((1 (a)))
= (1 (a))
= a.
It now follows that 1 = ((i1 ) (i2 ) (il )).
Definition 3.1.16 Let 1 , 2 , . . . , k Sn . Then 1 , 2 , . . . , k are called disjoint if for all i, 1 i k
and for all a In , i (a) 6= a implies j (a) = a for all j 6= i, 1 j k.
In other words, 1 , 2 , . . . , k Sn are disjoint if for all 1 i k and for all a In , if i moves a,
then all other permutations j must fix a, i.e., j (a) = a for all j 6= i, 1 j k.
Let and be disjoint permutations on In . Let a S be such that (a) 6= a. Then (a) = a. Let
(a) = b. Then
( )(a) = ((a)) = (a) = b.
Also,
( )(a) = ((a)) = (b).

64

3. Permutation Groups

If (b) = b, then (b) = b = (a), so a = b. Thus, (a) = b = a, a contradiction. Hence, (b) 6= b, so

(b) = b. Thus,
( )(a) = ((a)) = (b) = b.

Hence, ( )(a) = ( )(a). Suppose (a) = a. If (a) = a, then ( )(a) = a = ( )(a). Suppose
(a) 6= a. By a similar argument as before, ( )(a) = ( )(a). Therefore, = . Consequently,
if and are disjoint permutations, then they commute. We record this result in the following theorem.
Theorem 3.1.17 Let , Sn such that and are disjoint. Then = , i.e., and
commute.

1 2 3 4 5 6 7 8
Consider =
Sn . Then = (1 2 5 3) (4 8) (6 7) can be written
2 5 1 8 3 7 6 4
as a product of disjoint cycles. This leads us to the following theorem.
Theorem 3.1.18 Any nonidentity permutation of Sn (n 2) can be uniquely expressed (up to the
order of the factors) as a product of disjoint cycles, where each cycle is of length at least 2.
Proof. We prove the result by induction on n.
Basis step: Suppose n = 2. Now |S2 | = 2 and the nonidentity element of S2 is =

1 2
2 1

. Now

= (1 2), i.e., is a cycle. Thus, the theorem is true for n = 2.

Inductive hypothesis: Suppose that the theorem is true for all Sk such that 2 k < n.
Inductive step: Suppose n > 2. We show that the result is true for n.
Let be a nonidentity element of Sn . Now i (1) In for all integers i, i 1. Therefore, {(1), 2 (1),
. . . , i (1), . . .} In . Because In is a finite set, we must have l (1) = m (1) for some integers l and m
such that l > m 1. This implies that lm (1) = 1. Let us write j = l m. Then j > 0 and j (1) = 1.
Let i be the smallest positive integer such that i (1) = 1. Let
A = {1, (1), 2 (1), . . . , i1 (1)}.
Then all elements of the set A are distinct. Let Sn be the permutation defined by
= (1 (1) 2 (1) i1 (1)),
i.e., is a cycle. Let B = In \A. If B = , then is a cycle. Suppose B 6= . Let = |B . If is the
identity, then is a cycle. Suppose that is not the identity. Now by the induction hypothesis, is a
product of disjoint cycles on B, say, = 1 2 r . Now for 1 i r, define i by

i (a) if a B
i (a) =
a if a
/ B.
Then 1 , 2 , . . . , r and are disjoint cycles in Sn . It is easy to see that = 1 2 r . Thus,
is a product of disjoint cycles.
To prove the uniqueness, let = 1 2 r = 1 2 s , a product of r disjoint cycles and
also a product of s disjoint cycles, respectively. We show that every i is equal to some j and every k
is equal to some t. Consider i , 1 i r. Suppose i = (i1 i2 . . . il ). Then (i1 ) 6= i1 . This implies that
i1 is moved by some l . By the disjointness of the cycles, there exists unique j , 1 j s, such that i1
appears as an element in j . By reordering, if necessary, we may write j = (i1 c2 . . . cm ). Now
i2
i3

il

= i (i1 ) = (i1 ) = j (i1 ) = c2

= i (i2 ) = (i2 ) = (c2 ) = j (c2 )
= c3

= i (il1 ) = (il1 ) = (cl1 ) = j (cl1 ) = cl.

If l < m, then i1 = i (il ) = (il ) = (cl ) = j (cl ) = cl+1 , a contradiction. Thus, l = m. Hence, i = j
for some j, 1 j s. Similarly, every k = t for some t, 1 t r.

3.1. Permutation Groups

65

Corollary 3.1.19 Let n 2. Any permutation of Sn can be expressed as a product of transpositions.

Proof. In view of the preceding theorem, it suces to show that every k-cycle can be expressed as
a product of transpositions. This fact is immediate from the following equations:
e = (1) = (1 2) (1 2)
and for k 2

(i1 i2 ik ) = (i1 ik ) (i1 ik1 ) (i1 i2 ),

where {i1 , i2 . . . , ik } In .
Let Sn . Because Sn is a finite group, we know that () is finite. Thus, in order to find the
order of , we need to compute , 2 , 3 , . . . , until we find the first positive integer k such that k = e.
Finding such a positive integer could be a tedious task. However, we can eectively make use of the
decomposition of as a product of disjoint cycles, compute the order of each cycle, which is nothing
but the length of the cycle (Exercise 17, page 69) and from the order of the cycles deduce the order of
. We ask the reader to consider this problem in Exercise 18 (page 69).
Theorem 3.1.18 tells us that any permutation Sn , n 2, can be written as a product of disjoint
cycles. However, the theorem does not tell us how to find the disjoint cycles in the decomposition of .
Next, we illustrate how to find these cycles.
Let be a permutation on In , n 2. In order to express as a product of disjoint cycles, first
consider 1, (1), 2 (1), 3 (1), . . . and find the smallest positive integer r such that r (1) = 1. Let
1 = (1 (1) 2 (1) r1 (1)).
Then 1 is a cycle of length r. Let i be the first element of In not appearing in 1 . Now consider i, (i),
2 (i), 3 (i), . . . and find the smallest positive integer s such that s1 (i) = i. Let
2 = (i (i) 2 (i) s1 (i)).
Then 2 is a cycle of length s. Now
{1, (1), 2 (1), . . . , r1 (1)} {i, (i), 2 (i), . . . , s1 (i)} = ,
for if j {1, (1), 2 (1), . . . , r1 (1)} {i, (i), 2 (i), . . . , s1 (i)}, then j = p (i) for some p,
1 p < r, and j = k (1) for some k, 1 k < s. Thus, {1, (1), 2 (1), . . . , r1 (1)} = {i, (i), 2 (i),
. . . , s1 (i)}, which is a contradiction. Hence, 1 and 2 are disjoint cycles. If {1, (1), 2 (1), . . . ,
r1 (1)} {i, (i), 2 (i), . . . , s1 (i)} 6= In , then consider the first element of In not appearing in {1,
(1), 2 (1), . . . , r1 (1)} {i, (i), 2 (i), . . . , s1 (i)} and continue the above process to construct the
cycle 3 . Because In is finite, the above process must stop with some cycle m . Then = 1 2 m .
We illustrate the above procedure with the help of the following example.
Example 3.1.20 Consider the permutation

1 2 3 4 5 6 7
=
6 3 5 2 4 7 1

on I7 . Here (1) = 6, 2 (1) = (6) = 7, and 3 (1) = (7) = 1. That is, 1 6 7 1. Hence, 1 = (1 6
7) is a 3-cycle. Now 2 is the first element of I7 not appearing in (1 6 7). Also, (2) = 3, 2 (2) = (3) = 5,

3 (2) = (5) = 4, and 4 (2) = (4) = 2. That is, 2 3 5 4 2. Hence, 2 = (2 3 5 4) is a cycle

of length 4. Now 1 and 2 are disjoint and = 1 2 .
While writing a permutation as a product of disjoint cycles, it is customary not to write cycles of
length one in the product. Thus, if some element of In does not appear in any of the cycles, then it is
assumed to be fixed. For example, if = (1 2 5) (4 6) S7 , then because 3 and 7 neither appear in (1
2 5) nor in (4 6), they are fixed, i.e., (3) = 3 and (7) = 7.
Given a permutation Sn , n 2, we can write as a product of disjoint cycles. We can also write
as a product of transpositions. However, the representation of as a product of transposition need

66

3. Permutation Groups

not be unique. For example, (1 2 3) = (1 3) (1 2) = (2 1) (2 3). Also, (1 3) = (1 2) (1 3) (2 3).

That is, (1 3) can be written as a product of one transposition or as a product of three transpositions.
However, we will show that the number of transpositions in any representation of a permutation is either
even or odd, but not both. We now proceed to prove this result.
Consider the formal product
Q
X =
1i<jn (ai aj ) = (a1 a2 )(a1 a3 ) (a1 an )
(a2 a3 ) (a2 an )
..
.
(an1 an ).
If n = 4, then X = (a1 a2 )(a1 a3 )(a1 a4 )(a2 a3 )(a2 a4 )(a3 a4 ).
For any permutation Sn , let
Y
(a(i) a(j) ).
(X ) =
1i<jn

Let us first examine (X ) for any transposition Sn .

Lemma 3.1.21 Let n 2. Let = (i j) Sn , i < j, be a transposition. Then (X ) = X .
Proof. First consider the factor (ai aj ) in the product X . The corresponding factor in (X ) is
a(i) a(j) . Now
a(i) a(j) = aj ai = (ai aj ).
Next, consider the factor ak al , where both k and l are neither equal to i nor equal to j. The corresponding factor in (X ) is a(k) a(l) and
a(k) a(l) = ak al .
Thus, the factor ak al remains unaltered. Now consider the factor ak al , where either k or l (but
not both) is equal to i or j. Let 1 t n. Suppose t < i < j. We have the pair of factors (at ai ) and
(at aj ) in the product X . The corresponding factors in (X ) are a(t) a(i) and a(t) a(j) and
(a(t) a(i) )(a(t) a(j) ) = (at aj )(at ai ) = (at ai )(at aj ).
Therefore, the product (at ai )(at aj ) remains unchanged. Now suppose i < t < j. Then we have the
pair of factors (ai at ) and (at aj ) in the product X . The corresponding factors in (X ) are a(i) a(t)
and a(t) a(j) and
(a(i) a(t) )(a(t) a(j) ) = (aj at )(at ai ) = (ai at )(at aj ).
Hence, the product (ai at )(at aj ) remains unaltered. Finally, let i < j < t. Then we have the pair
of factors (ai at ) and (aj at ) in the product X . The corresponding factors in (X ) are a(i) a(t)
and a(j) a(t) and
(a(i) a(t) )(a(j) a(t) ) = (aj at )(ai at ) = (ai at )(aj at ).
Therefore, the product (ai at )(aj at ) remains unaltered. Thus, all factors other than ai aj and
ak al , where both k and l are neither equal to i nor equal to j, can be paired so that the product of
factors under remains unaltered. Hence, it now follows that (X ) = X .
Theorem 3.1.22 Let n 2. Let Sn . Suppose
= 1 2 r = 1 2 s ,
where i , j Sn are transpositions, i = 1, 2, . . . , r, and j = 1, 2, . . . , s. Then both r and s are either
even or odd.

3.1. Permutation Groups

67

Proof. By Lemma 3.1.21, i (X ) = X and j (X ) = X for all i = 1, 2, . . . , r, and j = 1, 2, . . . , s.

First we compute ( 1 2 r )(X ). Now
( 1 2 r )(X ) = 1 ( 2 ( (r (X ))))
= (1)r X .
Similarly, ( 1 2 s )(X ) = (1)s X . Hence, (1)r = (1)s . Thus, both r and s are either even of
odd.
By the above theorem, if Sn , then can be written as a product of either an even or an odd
number of transpositions, but not both. This leads us to the following definition.
Definition 3.1.23 Let Sn . If is a product of an even number of transpositions, then is called
an even permutation; otherwise is called an odd permutation.
Corollary 3.1.24 Let Sn be a k-cycle. Then is an even permutation if and only if k is odd.
Proof. Let = (1 2 k). Then = (1 k) (1 k 1) (1 2), i.e., is a product of k 1
transposition. If is an even permutation then k 1 is even, so k is odd. On the other hand, if k is
odd, then k 1 is even, so is an even permutation. This completes the proof.
Let An denote the subset of Sn consisting of all even permutations, n 2.
Theorem 3.1.25 For n 2, the pair (An , ) is a group, called the alternating group on In .
Proof. Because e = (1 2) (1 2), e An . Thus, An 6= . A product 1 2 is even if and only if 1
and 2 are both even or both odd by Theorem 3.1.22. Therefore, An is closed under . If An , then
1 = e is even and hence 1 An . Hence, (An , ) is a group.
Cauchy recognized many important properties of An . Among others, he proved the following theorem.
Theorem 3.1.26 Every element in An is a product of 3-cycles, n 3.
Proof. Let An . Then = 1 2 r , where i is a transposition, 1 i r, and r is even.
Now for any transposition (a b),
(a b) = (1 a) (1 b) (1 a).
Thus,
= (1 i1 ) (1 i2 ) (1 im )
where m is even. Because (1 i1 ) (1 i2 ) = (1 i2 i1 ), it follows that is a product of 3-cycles.

Worked-Out Exercises
Exercise 1 Prove that two cycles in Sn are conjugate if and only if they have the same length.
Solution: Let = (i1 i2 ir ) and = (j1 j2 js ) be two cycles in Sn . First suppose that and are
conjugate. Then = 1 for some Sn . Because is onto and il In , there exists kl
such that (kl ) = il for all l = 1, 2, . . . , r . Now
(j1 j2 js ) = (1 (i1 ) 1 (i2 ) 1 (ir ))
= (k1 k2 kr ).

( by Theorem 3.1.15)

Hence, s = r, so and are of the same length.

Conversely,
ir ) and = (j1 j2 jr ) be two cycles in Sn of the same length.
let = (i1 i2
i1 i2 ir
Let =
, i.e., (il ) = jl for all l = 1, 2, . . . , r, and (a) = a for all a
j1 j2 jr
In \{i1 , i2 , . . . , ir }. Then Sn . Now
1 = (1 (j1 ) 1 (j2 ) 1 (jr )) = (i1 i2 ir ) = .

68

3. Permutation Groups

Exercise 2 Express the permutation

=

1 2 3 4 5 6 7 8
2 3 8 5 6 4 7 1

on I8 as a product of disjoint cycles and then as a product of transposition. Is an even permutation?

Solution: We have (1) = 2, 2 (1) = (2) = 3, 3 (1) = (3) = 8, and 4 (1) = (8) = 1. Thus, (1 2 3
8) is a cycle. Now 4 is the first element of I8 not appearing in (1 2 3 8). We have (4) = 5,
2 (4) = (5) = 6, and 3 (4) = (6) = 4. Hence, (4 5 6) is also a cycle in . Next, 7 is the first
element of I8 not appearing in (1 2 3 8) and (4 5 6). Now (7) = 7. Because all the elements of I8
appear in one of the cycles (1 2 3 8), (4 5 6), and (7), we have = (1 2 3 8) (4 5 6). Now (1 2 3
8) = (1 8) (1 3) (1 2) and (4 5 6) = (4 6) (4 5). Thus, = (1 8) (1 3) (1 2) (4 6) (4 5).
Because is a product of five transpositions, is not an even permutation.
Exercise 3 Write all elements of S4 . Show that S4 has no elements of order 5.
Solution: Let S4 and = 1 2 k , a product of disjoint cycles. Because S4 is a permutation
group on I4 , k 2. If k = 1, then is a 2-cycle, 3-cycle, or 4-cycle. If k = 2, then is a product of
two disjoint transpositions. The number of distinct cycles of length 2 is 6, the number of distinct
cycles of length 3 is 8, and the number of distinct cycles of length 4 is 6. Hence, S4 = {e, (1 2), (1
3), (1 4), (2 3), (2 4), (3 4), (1 2 3), (1 3 2), (2 3 4), (2 4 3), (1 3 4), (1 4 3), (1 2 4), (1 4 2), (1 2
3 4), (1 3 2 4), (1 4 2 3), (1 2 4 3), (1 3 4 2), (1 4 3 2), (1 2) (3 4), (1 4) (3 2), (1 3) (2 4)}.
Because each 2-cycle is of order 2, each 3-cycle is of order 3, each 4-cycle is of order 4, and the
order of the product of two disjoint 2-cycles is 2, S4 has no element of order 5.

Exercise 4 Find the order of (1 2 3 4) (5 6 7) in S7 .

Solution: (1 2 3 4) = 4, (5 6 7) = 3. Now (1 2 3 4) and (5 6 7) are disjoint. Hence, (1 2 3 4) (5 6 7) = (5
6 7) (1 2 3 4). If a and b are two elements of a group G such that (a) = m, (b) = n, and
gcd(m, n) = 1, then (ab) = mn. Using this result, we find that the order of (1 2 3 4) (5 6 7) is
12.
Exercise 5 Find the order of (1 2 3 4) (5 6) in S6 .
Solution: (1 2 3 4) = 4, (5 6) = 2. Now (1 2 3 4) and (5 6) are disjoint, so they commute. Thus, ((1 2 3
4) (5 6))4 = e. Now ((1 2 3 4) (5 6))1 6= e, ((1 2 3 4) (5 6))2 = (1 2 3 4)2 (5 6)2 = (1 2 3
4)2 6= e. If ((1 2 3 4) (5 6))3 = e, then the order of (1 2 3 4) (5 6) will be 3 and 3 divides 4, a
contradiction. Hence, the order of (1 2 3 4) (5 6) is 4.

Exercises
1. Express the following permutations as (i) a product of disjoint cycles and (ii) a product of transpositions:

1 2 3 4 5 6
1 2 3 4 5 6
,
.
3 5 4 1 6 2
3 2 1 5 4 6
2. Let = (1 2 5 7) and = (2 4 6) S7 . Find 1 .
3. Let = (1 3 5 7) and = (2 4 8) (1 3 6) S8 . Find 1 .
4. Let = (1 3) (5 8) and = (2 3 6 7) S8 . Find 1 .
5. Let = (2 5 9) (1 3 6) and = (1 5 7) (2 4 6 9) S9 . Find 1 .
6. Let (1 3 5 7) and (2 3 6 8) S8 . Find S8 such that (1 3 5 7) 1 = (2 3 6 8).
7. If = (1 2 3 4 5 6), show that = (1 6) (1 5) (1 4) (1 3) (1 2).

3.1. Permutation Groups

69

8. Find the order of (1 2 3) (4 5) in S5 .

9. Prove that (1 2 n 1 n)1 = (n n 1 2 1).
10. Prove that every transposition is its own inverse.
11. Prove that the symmetric group on two symbols (S2 , ) is commutative.
12. Let = (a1 a2 ak ) Sn be a k-cycle. Show that

(a1 a3 a2m1 ) (a2 a4 a6 a2m ) if k = 2m, i.e., k is even

2 =
(a1 a3 a2m+1 a2 a4 a2m ) if k = 2m + 1, i.e., k is odd.
13. Determine A4 .
14. Let , Sn . Show that 1 1 An .
15. Prove that |An | =

n!
2.

16. Show that the number of distinct cycles of length r in Sn is

1 n!
r (nr)! .

17. Let n 2 and Sn be a cycle. Show that is a k-cycle if and only if () = k.

18. Let Sn and = 1 2 k be a product of disjoint cycles. Suppose ( i ) = ni ,
i = 1, 2, . . . , n. Show that () = lcm(n1 , n2 , . . . , nk ).
19. Let Sn and p be a prime.
(a) Show that () = p if and only if either is a p-cycle or is a product of disjoint cycles,
where each cycle is either of length 1 or length p and at least one cycle is of length p.
(b) If is a p-cycle, prove that either m = e or m is a p-cycle for all m N.
20. Let and Sn . Let = 1 2 k and = 1 2 s be a product of disjoint
cycles. Let length(i ) = di and length( j ) = mj for all i = 1, 2, . . . , k and j = 1, 2, . . . , s and
d1 d2 dk and m1 m2 ms . We say that and have the same cyclic
structure if k = s and di = mi for all i = 1, 2, . . . , k. Prove that and have the same cyclic
structure if and only if and are conjugate.
21. Prove that for Sn , is an even permutation if and only if (X ) = X .
22 (a) Let = (k l), Sn be two distinct transpositions, n 3. Show that there exist transpositions , Sn such that = , (k) = k and moves k.

(b) Prove that if the identity permutation e Sn can be written as a product of r ( 3) transpositions, then e can be written as a product of r 2 transpositions.
(c) Prove that if e = 1 2 r Sn as a product of transpositions, then r is even.

(d) Use (a), (b), and (c) to prove that if Sn , then can be written as a product of either an
even or an odd number of transpositions, but not both.

70

3. Permutation Groups

Augustin-Louis Cauchy (1789 1857) was born on August 21, 1789, in Paris, France. He received
his first education from his father. He was a neighbor of Laplace and Berthollet. Cauchy became acquainted
with famous scientists at a young age. Lagrange is said to have warned his father not to show Cauchy any
mathematics book before the age of seventeen.
At the age of fifteen, he completed his classic studies with distinction. He became an engineer in 1810, in the
Napoleon army. In 1813, he returned to Paris.
In 1811, Cauchy started his mathematical career by solving a problem sent to him by Lagrange on convex
polygons. In 1812, he solved Fermats famous classical problem on polygon numbers. His treatise on the definite
integral, which he submitted in 1814 to the French Academy, later became a basis of the theory of complex
functions.
In 1816, he was appointed full professor at the cole Polytechnique. More theorems and concepts have
been named for Cauchy than for any other mathematician. There are sixteen concepts and theorems named for
Cauchy in elasticity alone.
He worked on mathematics, mathematical physics, and celestial mechanics. In mathematics, he worked on
several areas, such as calculus, complex functions, algebra, dierential equations, geometry, and analysis. The
notion of continuity used today was invented by Cauchy. He also proved that a continuous function has a zero
between two points where the function changes its signs, a result also proved by Bolzano. The first adequate
definitions of indefinite integral and definite improper integral are due to Cauchy
In algebra, the notion of the order of an element, a subgroup, and conjugates are found in his papers. He
proved the famous Cauchys theorem for finite groups, that is, if the order of a finite group is divisible by a
prime p, then the group has a subgroup of order p. Cauchys role in shaping the theory of permutation groups is
central. He is regarded by some to be the founder of finite group theory. The two-row notation for permutations
was introduced by Cauchy. He also defined the product of permutations, inverse permutations, transpositions,
and the cyclic notation. He wrote his first paper on this subject in 1815, but did not return to it for nearly thirty
years. In 1844, he proved that every permutation is a product of disjoint cycles.
He also did work of fundamental importance in the theory of determinants. His treatise on determinants,
published in 1812, contains important results concerning product theorems and the inverse of a matrix.
Cauchy enjoyed teaching. He published more than 800 papers and eight books. He died on May 22, 1857.

Chapter 4

Subgroups and Normal Subgroups

In Chapter 2, we began a discussion of the evolution of group theory. This chapter seems a good place to renew
the discussion. It took more than 100 years for the abstract concept of a group to evolve. The evolution followed
lines similar to the evolution of other theories. First came the discovery of isolated phenomena, followed by the
recognition of features common to all. Then came the search and classification of other instances. Next, general
principles emerged. Last, the abstract postulates which define the system were uncovered. A deeper account can
be found in Bell.

4.1

Subgroups

Let us consider the groups (Z, +) and (Q, +), where + is the usual addition of numbers, and note the following:
1. Both these groups have the same binary operation.
2. Z is a subset of Q.
The same is true for the groups (Z, +) and (R, +); (Q, +) and (R, +); (R, +) and (C, +). Similarly, as seen in
the previous chapter, both the groups (An , ) and (Sn , ) have the same binary operations and An is a subset of
Sn .
One can think of many examples, where the underlying set of one group is a subset of the underlying set of
another group. This leads us to the concept of a subgroup. Before formally defining subgroups, let us also note
the following:
Let (G, ) be a group and H be a nonempty subset of G. Then H is said to be closed under the binary
operation if a b H for all a, b H.
Suppose H is closed under the binary operation . Then the restriction of to H H is a mapping from H H
into H. Thus, the binary operation defined on G induces a binary operation on H. We denote this induced
binary operation on H by also. Thus, (H, ) is a mathematical system. It also follows that is associative as
a binary operation on H, i.e., a (b c) = (a b) c for all a, b, c H. If (H, ) is a group, then we call H a
subgroup of G. More formally, we have the following definition.
Definition 4.1.1 Let (G, ) be a group and H be a nonempty subset of G. Then (H, ) is called a subgroup of
(G, ) if (H, ) is a group.
Consider the group (Q, +) and its subgroups (Z, +). Now the identity elements of both these groups is 0.
Next, let a Z. Then a Q. Also, the inverse of a in Z as well as in Q is a. In other words, the inverse of a in
Z and the inverse of a in Q is the same. In general, we have the following result.
Theorem 4.1.2 Let (G, ) be a group and (H, ) be a subgroup of (G, ).
(i) The identity elements of (H, ) and (G, ) are the same.
(ii) If h H, then the inverse of h in H and the inverse of h in G is the same.
Proof. (i) Let eH denote the identity of H and e denote the identity of G. Note that
eH eH = eH = eH e.
Hence, by the cancellation property, eH = e. This implies that the identity elements of G and H are the same.

71

72

4. Subgroups and Normal Subgroups

(ii) Let h H. Let h0 denote the inverse of h in H and h1 denote the inverse of h in G. Then
h h0 = e = h0 h

and
Now

h h1 = e = h1 h.
h0 = h0 e = h0 (h h1 ) = (h0 h) h1 = e h1 = h1 .

This implies that the inverse of h in H and the inverse of h in G are the same.

Remark 4.1.3 If (G, ) is a group, then ({e}, ) and (G, ) are subgroups of (G, ). These subgroups are called
trivial.
Example 4.1.4 Consider the following list of groups.
(i) ({0}, +), (Z, +), (Q, +), (R, +), (C, +),
(ii) ({1}, ), (Q\{0}, ), (R\{0}, ), (C\{0}, ),
where + is the usual addition operation and is the usual multiplication operation. Each group is a subgroup
of the group listed to its right. For example, (Z, +) is a subgroup of (Q, +), (R, +), and (C, +), and (R\{0}, ) is
a subgroup of (C\{0}, ).
Notation 4.1.5 In the remainder of the text, we shall generally use the notation G instead of (G, ) for a
group and we write ab for a b. We shall refer to ab as the product of a and b. This notation is usually called
multiplicative notation.
Readers with some knowledge of linear algebra should notice the similarity with respect to the type of results
and order of presentation of those which immediately follow. First comes a result which gives an easy method
of determining if a nonempty subset is a substructure. This is followed by the result that the intersection
of any collection of substructures is a substructure. Next, comes the definition of a substructure generated
by a subset. Finally, a theorem describing the substructure generated by a given subset. These ideas appear
throughout algebra. We will encounter them again, for example, when we examine the ideals of a ring.
Let G be a group and H be a nonempty subset of G. To show that H is a subgroup of G, we need to show
that H is a group under the binary operation of G. This requires us to show that H is closed under the binary
operation, the binary operation is associative, H contains the idenntiy element, and every element of H has an
inverse in H. Next theorem gives a criteria that can be eectively used to show that a nonempty subset of a
group is a subgroup.
Theorem 4.1.6 Let G be a group and H be a nonempty subset of G. Then H is a subgroup of G if and only if
for all a, b H, ab1 H.
Proof. Suppose H is a subgroup of G. Let a, b H. Because H is a subgroup, it is a group. Therefore,
b H implies that b1 H. Thus, ab1 H because H is closed under the binary operation.
Conversely, suppose H is a nonempty subset of G such that a, b H implies ab1 H. Because H 6= ,
there exists a H. Now a, a H. Therefore, e = aa1 H, i.e., H contains the identity. Next, let b H. Then
e, b H, implies that b1 = eb1 H. Thus, every element of H has an inverse in H.
To show that H is closed under the binary operation, let a, b H. Then a, b1 H. Thus, ab = a(b1 )1 H.
Hence, H is closed under the binary operation. From the statements preceding Definition 4.1.1, associativity
holds for H. Hence, H is a group, so H is subgroup of G.

Corollary 4.1.7 Let G be a group and H be a finite nonempty subset of G. Then H is a subgroup of G if and
only if for all a, b H, ab H.
Proof. If H is a subgroup, then for all a, b H, ab H.
Conversely, suppose that for all a, b H, ab H. Let h H. Then h, h2 , . . . , hn , . . . H, so
{h, h2 , . . . , hn , . . .} H.
Because H is finite and the set {h, h2 , . . . , hn , . . .} is a subset of H, it follows that all the elements of {h, h2 , . . . , hn , . . .}
cannot be distinct. Thus, there exist integers r and s such that 0 r < s and hr = hs . This implies that
e = hsr H. Now s r 1. Thus, e = hhsr1 implies that h1 = hsr1 H.
Let a, b H. Then a, b1 H, so ab1 H by the hypothesis. Hence, by Theorem 4.1.6, H is a subgroup.

4.1. Subgroups

73

Theorem 4.1.8 Let G be a group and Z(G) = {b G | ab = ba for all a G}. Then Z(G) is a commutative
subgroup of G. Z(G) is called the center of G.
Proof. Since ae = a = ea for all a G, e Z(G) and so Z(G) 6= . Let a, b Z(G). Then bc = cb for all
c G. From this, it follows that cb1 = b1 c for all c G and so b1 Z(G). Now
(ab1 )c = a(b1 c) = a(cb1 ) = (ac)b1 = (ca)b1 = c(ab1 )

for all c G and so ab1 Z(G). Hence by Theorem 4.1.6, Z(G) is a subgroup of G. That Z(G) is commutative
follows by the definition of Z(G).
In the remainder of this section, we will see how new subgroups arise from existing subgroups of a group.
Theorem 4.1.9 Let G be a group and {H | I} be any nonempty collection of subgroups of G. Then
I H is a subgroup of G.
Proof. Each H is a subgroup. Therefore, e H for all I. This implies that e I H , so I H
6= . To show that I H is a subgroup, we will use Theorem 4.1.6.
Let a, b I H . Now

a, b I H
a, b H for all I
ab1 H for all I
ab1 I H .

because H is a subgroup for all I

Consequently, I H is a subgroup of G by Theorem 4.1.6.

Definition 4.1.10 Let G be a group and S be a subset of G. Let
S = {H | H is a subgroup of G and S H}.
Define
hSi = HS H,

i.e., hSi is the intersection of all subgroups H of G such that S H.

(i) Thesubgroup hSi of G is called the subgroup generated by S.
(ii) If G = hSi , then S is called a set of generators for G.
Remark 4.1.11 Let G be a group and S be a subset of G. Note that if either S = or S = {e}, then hSi = {e}.
Moreover, hGi = G.
We now proceed to obtain a characterization of a subgroup generated by a nonempty subset in terms of the
elements of the group.
Let S = {H | H is a subgroup of G and S H}, where S 6= . It can be shown that (S, ) is a partially
ordered set, where denotes the set inclusion relation. In this poset, hSi is the least element. Hence, hSi is the
smallest subgroup of G which contains S.
Because hSi is a subgroup of G, we must have for any s1 , . . . , sn S, the product sk1 1 sknn hSi , where
ki = 1 for i = 1, 2, . . . , n. Thus, if A denotes the set {sk1 1 sknn | si S, ki = 1, i = 1, 2, . . . , n; n = 1, 2,
. . .}, then A hSi . Note that if s S, then e = ss1 A. In the following theorem, we show that A = hSi .
Therefore, S does generate hSi in the sense of multiplying elements of S or their inverses together to build up
the smallest subgroup containing S.
Theorem 4.1.12 Let S be a nonempty subset of a group G. Then
hSi = {sk1 1 sknn | si S, ki = 1, i = 1, 2, . . . , n; n = 1, 2, . . .}.
Proof. Let
A = {sk1 1 sknn | si S, ki = 1, i = 1, 2, . . . , n; n = 1, 2, . . .}.

We have already noted that A hSi . We show that hSi A by showing that A is a subgroup of G containing
S. (Recall that hSi is the smallest subgroup of G containing S.)
j
Because S 6= , there exists s S. Then s = s1 A, so S A. Let a = si11 simm , b = tj11 tqq A. Then
j

jq

ab1 = (si11 simm )(tj11 tqq )1 = si11 simm tq

Thus, A is a subgroup of G by Theorem 4.1.6. Hence, hSi A.

1
tj
A.
1

74

4. Subgroups and Normal Subgroups

Notation 4.1.13 For a G, we use the notation hai rather than h{a}i to denote the subgroup of G generated
by {a}.
Corollary 4.1.14 Let G be a group and a G. Then hai = {an | n Z}.
Proof. By Theorem 4.1.12, we have hai = {ak1 akm | ki = 1, i = 1, 2, . . . , m; m = 1, 2, . . .} =
{a
| ki = 1, i = 1, 2, . . . , m; m = 1, 2, . . .} = {an | n Z}.
k1 ++km

In the additive notation, we would have hai = {na | n Z}.

Let n 3. From Chapter 3, recall that An is the set of all even permutations on the set In . Moreoevr,
in Chapter 3, we proved that every element of An is a product of 3-cycles (Theorem 3.1.26). In the following
theorem, we conclude that An is generated by the set of all 3-cycles.
Theorem 4.1.15 Let n 3. Then An is generated by the set of all 3-cycles.
Proof. Let (a b c) be a 3-cycle. Then (a b c) = (a c) (a b), i.e., (a b c) is a product of even number of
transposition. Hence, every 3-cycle is an even permutation. This implies that every 3-cycle is in An . By Theorem
3.1.26, every element of An is a product of 3-cycles. Hence, An is generated by the set of all 3-cycles.
We now turn our attention to the product of subgroups.
Definition 4.1.16 Let H and K be nonempty subsets of a group G. The product of H and K is defined to be
the set
HK = {hk | h H, k K}.
Let H1 , H2 , . . . , Hn be nonempty subsets of a group G. We define the product, H1 H2 Hn , of H1 , H2 , . . . ,
Hn to be the set
H1 H2 Hn = {h1 h2 hn | hi Hi , i = 1, 2, . . . , n}.
Example 4.1.17 Consider the group of symmetries of the square. Let H = {r360 , d1 } and K = {r360 , h}. Then
H and K are subgroups of G. Now
HK = {r360 r360 , r360 h, d1 r360 , d1 h} = {r360 , h, d1 , r90 }.
/ HK, so HK is not closed under the binary operation. Hence, HK is not a subgroup of
Now hd1 = r270
the symmetries of the square. Also, note that
KH = {r360 r360 , r360 d1 , hr360 , hd1 } = {r360 , d1 , h, r270 },
and
hH Ki = {r360 , r90 , r180 , r270 , h, v, d1 , d2 }.
Example 4.1.17 shows that in general the product of subgroups need not be a subgroup. In the following
theorem, we give a necessary and sucient condition for the product of subgroups to be a subgroup.
Theorem 4.1.18 Let H and K be subgroups of a group G. Then HK is a subgroup of G if and only if
HK = KH.
Proof. Suppose HK is a subgroup of G. Let kh KH, where h H and k K. Now h = he HK and
k = ek HK. Because HK is a subgroup, it follows that kh HK. Hence, KH HK. On the other hand, let
hk HK. Then (hk)1 HK, so (hk)1 = h1 k1 for some h1 H and k1 K. Thus,
hk = (h1 k1 )1 = k11 h1
1 KH.

This implies that HK KH. Hence, HK = KH.

Conversely, suppose HK = KH. Let h1 k1 , h2 k2 HK, where h1 , h2 H and k1 , k2 K. We show that
(h1 k1 )(h2 k2 )1 HK.
Now k2 K and h2 H. Therefore, k21 h1
KH = HK. This implies that k21 h1
= h3 k3 for some
2
2
h3 H and k3 K. Similarly, k1 h3 KH = HK, so k1 h3 = h4 k4 for some h4 H and k4 K. Thus,
(h1 k1 )(h2 k2 )1

=
=
=

h1 k1 k21 h1
2
h1 k1 h3 k3
h1 h4 k4 k3 HK.

Hence, HK is a subgroup of G by Theorem 4.1.6.

because (h2 k2 )1 = k21 h1

2
substitute k21 h1
2 = h3 k3
substitute k1 h3 = h4 k4

4.1. Subgroups

75

Corollary 4.1.19 If H and K are subgroups of a commutative group G, then HK is a subgroup of G.

Proof. Since G is commutative, HK = KH. The result now follows by Theorem 4.1.18
The following theorem gives another necessary and sucient condition for a product of subgroups to be a
subgroup.
Theorem 4.1.20 Let H and K be subgroups of a group G. Then HK is a subgroup of G if and only if
HK = hH Ki .
Proof. First suppose that HK is a subgroup of G. We show that HK = hH Ki .
Let h H. Then h = he HK. Thus, H HK. Similarly, K HK. Hence, H K HK. Now hH Ki
is the smallest subgroup of G containing H K, so it follows that hH Ki HK.
Let hk HK, where h H and k K. Because H hH Ki and K hH Ki , we have h, k hH Ki .
Thus, hk hH Ki . This implies that HK hH Ki . Hence, HK = hH Ki .
The converse is immediate because hH Ki is a subgroup and HK = hH Ki .
Example 4.1.21 Let G = ha, bi , where a3 = e, b2 = e, and (ab)2 = e. Then
(i) ab = ba1 , ba = a1 b, and a2 b = ba.
(ii) G is not commutative because ab 6= ba.
(iii) bas = as b for all positive integers s.
(iv) By (i) and (iii)
 r+s j
if i = 0
a b
ar bi as bj =
ars bi+j
if i = 1.

(v) Because a3 = e = b2 , every element of G is of the form ar bi , 0 r < 3, i = 0, 1 by (iv).

(vi) G = {e, a, b, ab, a2 , a2 b}. Thus, |G| = 6.
(vii) (a) = 3 = (a2 ), (b) = (ab) = (a2 b) = 2.

(viii) The only subgroups of G are {e}, hai = a2 , hbi , habi , a2 b , and G.
G is called a dihedral group of degree 3 and is denoted by D3 . In general, a dihedral group1 of degree n is
Dn = ha, bi , where (ab)2 = e, (a) = n, and (b) = 2. In Chapter 5, we consider, D4 , a dihedral group of degree
4, and study this group in detail.

Worked-Out Exercises
Exercise 1 Let H be a subgroup of a group G. Let g G. Prove that
(a) gHg 1 = {ghg 1 | h H} is a subgroup of G,


(b) gHg 1  = |H| .

(a) We first show that gHg1 6= and then use Theorem 4.1.6. Since e = geg 1 gHg 1 , gHg1 6= .
Let gh1 g1 , gh2 g 1 gHg 1 . Then

Solution

1
1
(gh1 g1 )(gh2 g 1 )1 = gh1 g 1 gh1
= gh1 h1
gHg 1 .
2 g
2 g

Hence, gHg 1 is a subgroup of G.



(b) Let g G. To prove that gHg 1  = |H| , we show that there exists a one-one onto function of H
1
onto gHg . Define f : H gHg1 by f(h) = ghg 1 for all h H. Let h, h0 H. If h = h0 , then
ghg1 = gh0 g1 , i.e., f is well defined. Also, ghg1 gHg1 . Thus, f is a function of H into gHg1 .
Suppose f (h) = f (h0 ). Then ghg 1 = gh0 g 1 . From this it follows that h = h0 . This shows that f is
one-one. To show f is onto gHg1 , let a gHg 1 . Then a = gbg1 = f (b) for some b H, namely,
b = g 1 ag. Thus, f is onto gHg1 .
Exercise 2 Prove that Sn is generated by {(1 2), (1 3), (1 4), . . . , (1 n)}.

Solution Let be any permutation in Sn . Then is a product of transpositions. Thus, it is sucient to show that
if (i j) is any transposition in Sn , i < j, then
(i j) h(1 2), (1 3), (1 4), . . . , (1 n)i .
This follows from the fact that (i j) = (1 i) (1 j) (1 i). Hence, Sn is generated by {(1 2), (1 3), (1
4), . . . , (1 n)}.
1 We

show the existence of such groups in Chapter 7.

76

4. Subgroups and Normal Subgroups

Exercise 3 Find all subgroups of (Z, +).

Solution Let H be a subgroup of Z. Suppose H 6= {0}. Let a be a nonzero element of H. Then a H. Since either
a or a is a positive integer, H contains a positive integer. With the help of the principle of well-ordering,
we can show that H contains a smallest positive integer. Let a be the smallest positive integer in H. We
claim that H = {na | n Z}.
Now na H for all n Z and so {na | n Z} H. On the other hand, let b H. By the division algorithm,
there exist c and r in Z such that b = ca + r, where 0 r < a. Suppose r 6= 0. Then r = b ca H. Thus,
H contains a positive integer smaller than a, a contradiction. Hence, r = 0 and so b = ca {na | n Z}.
This implies that H {na | n Z}. Thus, H = {na | n Z} for some a Z. Also, for all n Z, the set
T = {nm | m Z} = nZ is a subgroup of Z. Hence, nZ, n = 0, 1, 2, . . . are the subgroups of Z.

Exercises
1. Prove that H is a subgroup of the group G, where
(a) H = {[0], [2], [4], [6], [8], [10]}, G = Z12 ,

(b) H = {[0], [3], [6], [9]}, G = Z12

and where the group operation under consideration is +12 .
2. Let GL(2, R) denote the group of all nonsingular 2 2 matrices over R. Show that each of the following
sets is a subgroup of GL(2, R).



a b
(a) S =
| a, b, c, d R and ad bc = 1 .
c d



a 0
(b) S =
| a, b, c, d R and a 6= 0 .
0 a



a b
(c) S =
| a, b, c, d R and either a or b is nonzero .
b a



a b
(d) S =
| a, b, c, d R and ad 6= 0 .
0 d



a b
(e) S =
| a, b R and a2 + b2 6= 0 .
b a



a b
(f) S =
| a, b, c, d Z and ad bc = 1
c d
3. Show that the set H = {a + bi C | a2 + b2 = 1} is a subgroup of (C , ), where is the multiplication
operation of complex numbers.
4. Let G = {(a, b) | a, b R, b 6= 0}. Prove that (G, ) is a noncommutative group under the binary operation
(a, b) (c, d) = (a + bc, bd) for all (a, b), (c, d) G.
(a) Let H = {(a, b) G | a = 0}. Show that H is a subgroup of G.

(b) Let K = {(a, b) G | b > 0}. Show that K is a subgroup of G.

(c) Let T = {(a, b) G | b = 1}. Show that T is a subgroup of G.

(d) Find all elements of order 2 in G.

5. In S3 , determine the set T = {x S3 | x2 = e}. Is T a subgroup of S3 ?

6. Determine the subgroup h4, 6i in (Z,+).

7. In (Z, +), determine the subgroup generated by {4, 5}.

8. List the elements of the following subgroups.


 

1 2 3 4
1 2 3 4
(a)
,
in S4 .
4 3 2 1
2 1 4 3
(b) hh, vi in the symmetries of the square.

9. Let a = (1 2 3 4) and b = (2 4) S4 .

4.1. Subgroups

77

(a) Find (a) and (b).

(b) Show that ba = a3 b = a1 b.

(c) Find H = ha, bi in S4 .

(d) Find |H| .

10. Let G be a group generated by a, b such that (b) = 2, (a) = 6, and (ab)2 = e. Show that
(a) aba = b,
(b) (a2 b)2 = e,
(c) ba2 b = a4 ,
(d) ba3 b = a3 .
11. Let G be a group. Prove that a nonempty subset H of G is a subgroup if and only if for all a, b H,
ab H and a1 H.
12. Let G be a commutative group. Show that the set H of all elements of finite order is a subgroup of G.

13. Let G be a group and a G. Show that if a is the only element of order n in G, then a Z(G).
14. Show that Z(Sn ) = {e} for all n 3.

15. Let G be a group and a G. Let C(a) = {b G | ba = ab}. Prove that C(a) is a subgroup of G and that
Z(G) = aG C(a). C(a) is called the centralizer of a in G.
16. Prove that a group G cannot be written as the union of two proper subgroups.
17. Let G be a group and H be a nonempty subset of G.
(a) Show that if H is a subgroup of G, then HH = H.
(b) If H is finite and HH H, prove that H is a subgroup of G.

(c) Give an example of a group G and a nonempty subset H of G such that HH H, but H is not a
subgroup of G.

18. Let H be a subgroup of a group G. Prove that hHi = H.

19. If A and B are subgroups of a group G, prove that A B is a subgroup of G if and only if A B or
B A. If C is also a subgroup of G, does a similar necessary and sucient condition hold for A B C
to be a subgroup of G?
20. Let G be a commutative group. If a and b are two distinct elements of G such that (a) = 2 = (b), show
that |ha, bi| = 4.
21

(a) Prove that Sn is generated by {(1 2), (1 2 3 n)}.

(b) Prove that Sn is generated by {(1 2), (2 3), (3 4), . . . , (n 1 n)}.

22 Show that (Q, +) is not finitely generated.
23 Let G be a group. Prove that if G is finite, then G has finitely many subgroups.
24 Does there exist an infinite group with only a finite number of subgroups?
25 For the following statements, write the proof if the statement is true; otherwise, give a counterexample.
(a) All nontrivial subgroups of (Z, +) are infinite groups.
(b) If A, B, and C are subgroups of a group G such that A B C, then ABC C.
(c) If G is a noncommutative group, then Z(G) = {e}.

(d) Let G be a group. If H is a nonempty subset of G such that a1 H for all a H, then H is a
subgroup of G.
(e) There exists a proper subgroup A of (Z, +) such that A contains both 2Z and 3Z.
(f) If H is a subgroup of (Q, +) such that Z H, then H = Q.

(g) If H is a subgroup of (Q , ) such that Z\{0} H, then H = Q .

78

4. Subgroups and Normal Subgroups

4.2

Cyclic Groups

In the previous section, we introduced the notion of a subgroup generated by a set. Groups that are generated
by a single element, called cyclic groups, are of special importance. As we shall see throughout the text, these
groups play an important role in studying the structure of a group. In fact, all of Chapter 9 revolves around
these groups. Cyclic groups are easier to study than any other group. They have special properties, some of
which we will discover in this section.
Definition 4.2.1 A group G is called a cyclic group if there exists a G such that
G = hai .
We recall that hai in Definition 4.2.1 is the set {an | n Z} (Corollary 4.1.14).
Let G = hai be a cyclic group and b, c G. Then b = an and c = am for some n, m Z. Now
bc = an am = an+m = am+n = am an = cb.
This shows that G is commutative. Hence, every cyclic group is commutative. We record this result in the
following theorem.
Theorem 4.2.2 Every cyclic group is commutative.
Example 4.2.3 (i) (Z, +) is a cyclic group because Z = h1i .
(ii) ({na | n Z}, +) (Example 2.1.8) is a cyclic group, where a is any fixed element of Z.
(iii) (Zn , +n ) is a cyclic group because Zn = h[1]i .
Example 4.2.4 Let a be a symbol and n a positive integer. Define by means of the following operation table.

a0
a1
a2
..
.
an2
an1

a0
a0
a1
a2
..
.
an2
an1

a1
a1
a2
a3
..
.
an1
a0

a2
a2
a3
a4
..
.
a0
a1

an2
an2
an1
a0
..
.
an4
an3

..
.

an1
an1
a0
a1
..
.
an3
an2

Then ({a0 , a1 , . . . , an1 }, ) is a cyclic group generated by a1 .

Example 4.2.5 Consider the set G = {e, a, b, c}. Define on G by means of the following operation table.

e
a
b
c

e
e
a
b
c

a
a
e
c
b

b
b
c
e
a

c
c
b
a
e

From the multiplication table, it follows that (G, ) is a commutative group. However, G is not a cyclic group
because
hei = {e}, hai = {e, a}, hbi = {e, b}, and hci = {e, c}
and each of these subgroups is properly contained in G. G is known as the Klein 4-group.
The next theorem gives the exact description of a finite cyclic group.
Theorem 4.2.6 Let hai be a finite cyclic group of order n. Then hai = {e, a, a2 , . . . , an1 }.

4.2. Cyclic Groups

79

Proof. By Corollary 4.1.14, hai = {ai | i Z}. Because hai is finite, there exist i, j Z (j > i) such that
ai = aj . Thus, aji = e and j i is positive. Let m be the smallest positive integer such that am = e. Then for
all integers i, j such that 0 i < j < m, ai 6= aj otherwise aji = e for some 0 i < j < m, which contradicts
the minimality of m. Hence, the elements of the set S = {e, a, a2 , . . . , am1 } are distinct. Clearly S hai .
Let ak hai . By the division algorithm, there exist integers q, r such that k = qm + r, 0 r < m. Thus,
ak = aqm+r = (am )q ar = ear = ar S. Therefore, hai S. Thus, S = hai . Since the elements of S are distinct
and hai has order n, it must be the case that m = n.
The following corollaries are immediate from the proof of Theorem 4.2.6. We omit the proofs.
Corollary 4.2.7 Let hai be a finite cyclic group. Then (a) = |hai| .
Corollary 4.2.8 A finite group G is a cyclic group if and only if there exists an element a G such that
(a) = |G| .
As stated in the beginning of this section, cyclic groups have special properties. We now proceed to discover
some of these properties. Subgroups of a cyclic group are themselves cyclic; this is proved in the next theorem.
Theorem 4.2.9 Every subgroup of a cyclic group is cyclic.
Proof. Let H be a subgroup of a cyclic group G = hai . If H = {e}, then H = hei , so H is cyclic. Suppose
{e} H. Then there exists b H such that b 6= e. Since b G, we have b = am for some integer m. Thus,
m 6= 0 since b 6= e. Since H is a group, am = b1 H. Now either m or m is positive. Therefore, H contains
at least one element which is a positive power of a. Let n be the smallest positive integer such that an H. We
now show that H = han i .
Now an H, so we must have han i H. Let h H. Then h = ak for some integer k. By the division algorithm,
there exist integers q, r such that k = nq+r, 0 r < n. Since an and ak H, we have ar = aknq = ak (an )q H.
However, if r > 0, we contradict the minimality of n. Therefore, r = 0 so that ak = (an )q han i . Hence, H
han i , so H = han i . Thus, H is cyclic.
Corollary
 4.2.10 Let G = hai be a cyclic group of order m, m > 1, and H be a proper subgroup of G. Then
H = ak for some integer k such that k divides m and k > 1. Furthermore, |H| divides m.
m
Proof. If H = {e},
k then H = ha i . Suppose that H 6= {e}. Let k be the smallest positive integer such that
a H. Then H = a . Now there exist integers q and r such that m = qk + r, where 0 r < k, and
k

ar = amqk = am aqk = aqk = ((ak )1 )q H.

The minimality of k implies that r = 0. Hence, m = qk and so k divides m. Since H 6= G, k > 1. Next, we show
m
that |H| divides m. By Theorem 2.1.46(ii), (ak ) = gcd(m,k)
= m
= q. As a result Corollary 4.2.7 implies that
k
|H| = (ak ) = q.
Since m = qk, we have q | m, i.e., |H| divides m.
By Corollary 4.2.10, if G is a finite cyclic group and H is a subgroup of G, then |H| divides |G| . This is a
special case of a more general result, called Lagranges theorem, which we will prove in the next section.
Let G = hai be an infinite cyclic group. Then (a) is infinite and this implies that (ak ) is infinite for any
nonzero integer k. Thus, the order of any nonidentity element of G is infinite. Let H be a nontrivial subgroup of
G. Then H is cyclic. Let H = hbi . Then b 6= e and b G and so (b) is infinite. This in turn shows that |H| is
infinite. Thus, every nontrivial subgroup of an infinite cyclic group is infinite.
Now let G = hai be a finite cyclic group of order n and H be a proper subgroup of G. Then by Corollary
4.2.10, |H| divides |G| . If H = {e}, then |H| = 1 and if H = G, then |H| = |G| and so |H| divides |G| . Thus,
the order of every subgroup of G divides the order of G. The following theorem shows that the converse of this
result is also true for finite cyclic groups.
Theorem 4.2.11 Let G be a finite cyclic group of order m. Then for every positive divisor d of m, there exists
a unique subgroup of G of order d.

80

4. Subgroups and Normal Subgroups

Proof. Let G = hai and d be a positive divisor of m. Becuase d | m, there exists k Z such that m = kd.
Now ak G and by Theorem 2.1.46(ii),
(ak ) =

m
(a)
=
= d.
gcd(k, m)
k


Let H = ak . Then |H| = (ak ) = d. Thus, G has a subgroup of order d. Next, we establish that H is unique.

Let K be a subgroup of order d. Let t be the smallest positive integer such that at K. Then K = at .
t
t
m
Because K is of order d, (a ) = d by Corollary 4.2.7. But (a ) = gcd(t,m) by Theorem 2.1.46(ii). Hence,
m
d = gcd(t,m)
, which implies that gcd(t, m) = m
d = k. This shows that k | t. Let t = kl for some l Z. Now
t
kl
k l
a = a = (a ) H. Hence, K H. Since |K| = |H| and H and K are finite, we have H = K. Thus, there
exists a unique subgroup of order d.

Worked-Out Exercises
Exercise 1 (Q, +)is not cyclic.
Solution Suppose Q is cyclic. Then Q=

G H
p
q

for some
p
2q

there exists n Z, n 6= 0 such that

contradiction. Thus, Q is not cyclic.

n pq

p
q

Q, where p and q are relatively prime. Since

by Corollary 4.1.14. This implies that

1
2

p
2q

Q,

= n Z, which is a

Exercise 2 Let G be a group such that |G| = mn, m > 1, n > 1. Show that G has a nontrivial subgroup.

Solution First suppose that G is cyclic. Let G = hai . Then (a) = mn. Clearly (am ) = n. Let H = ham i . Then
H is a nontrivial subgroup of G. Now suppose that G is not cyclic. Then for all a G, (a) < mn by
Exercise 26 (page 56). Let e 6= a G and let H = hai . Then H is a nontrivial subgroup of G.

Exercise 3 Let G be an infinite cyclic group generated by a. Show that

(a) ar = at if and only if r = t, where r, t Z,

(b) G has exactly two generators.

Solution

Exercise 4

(a)

Suppose ar = at and r 6= t. Let r > t. Then art = e. Thus, (a) is finite, say, (a) = n.
Then G = {e, a, . . . , an1 }, which is a contradiction since G is an infinite group. The converse is
straightforward.

(b) Let G = hbi for some b G. Since a G = hbi and b G = hai , a = br and b = at for some r, t Z.
Thus, a = br = (at )r = art . Hence, by (a), rt = 1. This implies that either r = 1 = t or r = 1 = t.
Thus, either b = a or b = a1 . Now from (a), a 6= a1 . Therefore, G has exactly two generators.
(a) Let G = hai be a finite cyclic group of order n. Show that ak is a generator of G if and only if
gcd(k, n) = 1, where k is a positive integer.

(b) Find all generators of Z10 .

Solution

n
n
. Hence, gcd(k,n)
= n.
(a) Suppose ak is a generator of G. Since |G| = n, (ak ) = n. But (ak ) = gcd(k,n)
k
n
Thus, gcd(k, n) = 1. Conversely, suppose that gcd(k, n) = 1. Then (a ) = gcd(k,n) = n. Hence,
 k 


 a  = n. Since ak G and |G| = n, G = ak .

(b) Now Z10 = h[1]i and |Z10 | = 10. By (a), k[1] is a generator if and only if gcd(k, 10) = 1, where
1 k 10. Now if k = 1, 3, 7, or 9, then gcd(k, 10) = 1. Thus, the generators of Z10 are 1[1] = [1],
3[1] = [3], 7[1] = [7] and 9[1] = [9].

Exercises
1. Let G = hai be a cyclic group of order 30. Determine the following subgroups.

(a) a5 .

(b) a2 .

2. Let G be a cyclic group of order 30. Find the number of elements of order 6 in G and also find the number
of elements of order 5 in G.
3. Prove that 1 and 1 are the only generators of Z.

4.

(a) Show that (R, +) is not cyclic.

4.3. Lagranges Theorem

81

(b) Show that (Q , ) is not cyclic.

(c) Show that (R , ) is not cyclic.

5. If G is a cyclic group of order n, show that the number of generators of G is (n), where is the Euler
-function.
6. Show that every proper subgroup of S3 is cyclic.
7. Give an example of a noncyclic Abelian group all of whose proper subgroups are cyclic.
8. Let G be a group. Suppose that G has at most two nontrivial subgroups. Show that G is cyclic.
9. Let G be a finite group. Show that if G has exactly one nontrivial subgroup, then order of G is p2 for
some prime p.
10. Let G be a noncommutative group. Show that G has a nontrivial subgroup.
11. Give an example of an infinite group of order 2 which contains a nontrivial finite cyclic group.

12. Show that there are cyclic subgroups of order 1, 2, 3, and 4 in S4 , but S4 does not contain any cyclic
subgroup of order 5.

13. For the following statements, write the proof if the statement is true; otherwise, give a counterexample.
(a) For every positive integer n, there exists a cyclic group of order n.
(b) Every proper subgroup of A4 is cyclic.
(c) A3 is a cyclic group.
(d) A4 is a cyclic group.
(e) All proper subgroups of (R, +) are cyclic.

4.3

Lagranges Theorem

In the last section, we noted that the order of a subgroup of a finite cyclic group divides the order of the group
(Corollary 4.2.10). We also remarked that this is a special case of a general result, called Lagranges theorem,
i.e., the order of a subgroup of a finite group divides the order of the group. Lagrange proved this result in
1770, long before the creation of group theory, while working on the permutations of the roots of a polynomial
equation. Lagranges theorem is a basic theorem of finite group theory and is considered by some to be the
most important result in finite group theory. In this section, we prove this result. We begin with the following
definition.
Definition 4.3.1 Let H be a subgroup of a group G and a G. The sets aH = {ah | h H} and Ha = {ha |
h H} are called the left and right cosets of H in G, respectively. The element a is called a representative of
aH and Ha.
If G is commutative, then of course aH = Ha. Observe that eH = H = He and that a = ae aH and
a = ea Ha.
Example 4.3.2 Consider the symmetric group S3 (Example 3.1.8). Then
 
 

1 2 3
1 2 3
H = e,
,
2 3 1
3 1 2
and
0

H =

 
1
e,
1

2
3

3
2



are subgroups of S3 . We now compute the left and right cosets of H in S3 . The left cosets of H in S3 are






1 2 3
1 2 3
1 2 3
H=
H=
H=H
1 2 3
2 3 1
3 1 2
and






1 2 3
1 2 3
1 2 3
H=
H=
H=
1 
3 2
3 2 1
 
  2 1 3
1 2 3
1 2 3
1 2 3
,
,
1 3 2
2 1 3
3 2 1

82

4. Subgroups and Normal Subgroups

and the right cosets of H in S3 are


1 2
H
1 2
and
H

3
3

=H

1
2


=H
 
3
,
2

2
3

3
1

=H

1
3

2
1

3
2

=H




1 2 3
1 2 3
=H
=
3 2 1 
2 1 
3
1 2 3
1 2 3
,
.
2 1 3
3 2 1

1 2
1 3

1
1

3
2
2
3

1
3

2
1

3
2

1
2

2
3

3
1

=H

Thus, for all a S3 , aH = Ha.

0
0
Next, we compute the left and right cosets of H in S3 . The left cosets of H in S3 are




0
0
0
1 2 3
1 2 3
H =
H =H ,
1 2 3
1 3 2

and

1
3

2
2

3
1

1
2

2
1

3
3

H =

H =

H =



1
3

2
2

3
1

 
1
,
3

2
1

3
2





1
2

2
1

3
3

 
1
,
2

2
3

3
1



1
1

2
3

3
2

1
3

2
2

3
1

 
1
,
2

2
3

3
1



1
2

2
1

3
3

 
1
,
3

2
1

3
2



H =

and the right cosets of H in S3 are

H

and
H
We see that

1
1

1
3

2
2

3
1

=H

1
2

2
1

3
3

=H


1
3

2
2

3
3

1
2

2
3

3
1



1
3

2
1

3
2



2
1

3
2

H 6= H

1
3

2
1

3
2

Thus, the left and right cosets of H 0 in S3 are not the same.

=H ,

There are some interesting phenomena happening in the above example. We see that all left and right cosets
of H in S3 have the same number of elements, namely, 3; that there are the same number of distinct left cosets
of H in S3 as of right cosets, namely, 2; that the set of all left cosets and the set of all right cosets form partitions
of S3 ; and, finally, that 3 2 equals the order of S3 . Similar statements hold for the subgroup H 0 . We show, in
the results to follow, that these phenomena hold in general.
In the next few theorems, we prove some properties of left and right cosets of a subgroup which will eventually
lead us to the proof of Lagranges theorem. The following theorem tells us when two left (right) cosets are equal.
It is a result that is used often in the study of groups.
Theorem 4.3.3 Let H be a subgroup of a group G and a, b G. Then
(i) aH = bH if and only if b1 a H.
(ii) Ha = Hb if and only if ab1 H.
Proof. (i) Suppose aH = bH. Since a aH and aH = bH, there exists h0 H such that a = bh0 . This
implies that b1 a = h0 H.
Conversely, suppose b1 a H. Then there exists h0 H such that b1 a = h0 , i.e., a = bh0 . Let ah aH.
Then ah = bh0 h bH. This implies that aH bH. Next, we show that bH aH. Now b1 a = h0 implies that
1
1
ah0 = b. Let bh bH. Then bh = ah0 h aH. Hence, bH aH. Consequently, aH = bH.
(ii) The proof is similar to (i). We leave it as an exercise.
Theorem 4.3.4 Let H be a subgroup of a group G. Then for all a, b G, either aH = bH or aH bH =
(i.e., two left cosets are either equal or they are disjoint).

4.3. Lagranges Theorem

83

Proof. Let a, b G. Suppose that aH bH 6= . We wish to show that aH = bH. Since aH bH 6= ,

there exists c aH bH. Hence, c aH and c bH and so there exist h1 , h2 H such that c = ah1 and
1
c = bh2 . Thus, ah1 = bh2 and from this, it follows that b1 a = h2 h1
a H. By Theorem 4.3.3(i),
1 . Therefore, b
aH = bH.
Corollary 4.3.5 Let H be a subgroup of a group G. Then {aH | a G} forms a partition of G.
Proof. Let P = {aH | a G}, i.e., P is the set of all left cosets of H in G. By Theorem 4.3.4, for all
aH, bH P, either aH = bH or aH bH = . Thus, P satisfies (i) of Definition 1.3.14. Since aH G for all
a G, aHP aH G. If a G, then a aH aHP aH. Therefore, G aHP aH. Hence, G = aHP aH.
This shows that P satisfies (ii) of Definition 1.3.14. Consequently, P is a partition of G.
Theorem 4.3.6 Let H be a subgroup of a group G. Then the elements of H are in one-one correspondence with
the elements of any left (right) coset of H in G.
Proof. Let a be any element of G and aH be a left coset of H in G. To show that the elements of H are
in one-one correspondence with the elements of aH, we show that there exists a one-one function of H onto aH.
Define f : H aH by f (h) = ah for all h H. Let h, h1 H. If h = h1 , then ah = ah1 , i.e., f (h) = f (h1 ).
Hence, f is well defined. Suppose f (h) = f(h1 ). Then ah = ah1 and this implies that h = h1 . Thus, f is a
one-one function. To show f is onto aH, let ah aH, where h H. Then ah = f (h). Hence, f maps H onto
aH. Similarly, we can show that the elements of H are in one-one correspondence with the elements of Ha.
The following corollary is immediate from Theorem 4.3.6.
Corollary 4.3.7 Let H be a subgroup of a group G. Then for all a G, |H| = |aH| = |Ha| .
The next theorem says that there are the same number of left cosets as right cosets.
Theorem 4.3.8 Let H be a subgroup of a group G. Then there is a one-one correspondence of the set of all left
cosets of H in G onto the set of all right cosets of H in G.
Proof. Let L = {aH | a G} be the set of all left cosets of H in G and R = {Ha | a G} be the set of
all right cosets of H in G. To establish a one-one correspondence between the elements of L and R, we need to
show the existence of a one-one function of L onto R.
Define f : L R by
f(aH) = Ha1
for all aH L. First note that Ha1 R for all a G. Let aH, bH L. Suppose aH = bH. Then by Theorem
4.3.3(i), b1 a H. This implies that b1 (a1 )1 = b1 a H and so by Theorem 4.3.3(ii), Hb1 = Ha1 . Thus,
f (bH) = f (aH). Hence, f is well defined. To show f is one-one, suppose f (aH) = f (bH). Then Ha1 = Hb1
and so a1 (b1 )1 H by Theorem 4.3.3(ii), i.e., a1 b H. Therefore, b1 a = (a1 b)1 H and so aH = bH.
Hence, f is one-one. Since for all Ha R, Ha = H(a1 )1 = f (a1 H) and a1 H L, it follows that f is onto
R. Thus, f is a one-one function from L onto R.
Definition 4.3.9 Let H be a subgroup of a group G. Then the number of distinct left (or right) cosets, written
[G : H], of H in G is called the index of H in G.
By Theorem 4.3.8, the number of left cosets and the number of right cosets of a subgroup H of a group G
are the same. Thus, [G : H] is well defined.
If G is finite, then of course [G : H] is finite. The following example is one, where G is infinite and [G : H] is
finite.
Example 4.3.10 Let n be a fixed positive integer. Consider the cyclic subgroup (hni , +) of (Z, +). Let k + hni
be a left coset of hni in Z. By the division algorithm, there exist integers q and r such that k = qn + r, where
0 r < n. Then k r = qn hni and so k + hni = r + hni by Theorem 4.3.3. Suppose i + hni = j + hni , where
0 i, j < n. Then i j hni by Theorem 4.3.3. This implies that n | (i j) and so we must have i j = 0 or
i = j since 0 i, j < n. Thus, the distinct left cosets of hni in Z are 0 + hni , 1 + hni , . . . , n 1 + hni .
We are now ready to prove Lagranges theorem. It is interesting to note that Lagrange proved the result for
the symmetric group Sn . Some credit Galois for proving the result in general.

84

4. Subgroups and Normal Subgroups

Theorem 4.3.11 (Lagrange) Let H be a subgroup of a finite group G. Then the order of H divides the order
of G. In particular,
|G| = [G : H] |H| .
Proof. Since G is a finite group, the number of left cosets of H in G is finite. Let {a1 H, a2 H, . . . , ar H} be
the set of all distinct left cosets of H in G. Then by Corollary 4.3.5, G = ri=1 ai H and ai H aj H = for all
i 6= j, 1 i, j r. Hence, [G : H] = r and
|G| = |a1 H| + |a2 H| + + |ar H| .
By Corollary 4.3.7, |H| = |ai H| for all i, 1 i r. Therefore,
|G|

|H| + |H| + + |H|

~}


r tim es

=
=

r |H|
[G : H] |H| .

Thus, the order of H divides the order of G.

Corollary 4.3.12 Let G be a group of finite order n. Then the order of any element a of G divides n and an = e.
Proof. Let a G and (a) = k. Let H = hai . Then by Corollary 4.2.7, |H| = |hai| = (a) = k. Hence, by
Theorem 4.3.11, k divides n. Thus, there exists q Z such that n = kq. Hence, an = akq = (ak )q = eq = e.
Let G be a finite group of order n and a G. Then (a) divides n by Corollary 4.3.12. Thus, to find (a),
we only need to check ak , where k is a positive divisor of n. For example, consider Z20 and [6] Z20 . Now
|Z20 | = 20 and 1, 2, 4, 5, 10, and 20 are the only positive divisors of 20. Now 1[6] = [6] 6= [0], 2[6] = [12] 6= [0],
4[6] = [24] = [4] 6= [0], 5[6] = [30] = [10] 6= [0], and 10[6] = [60] = [0]. Thus, ([6]) = 10. Hence, the above
corollary can be used to find the order of an element in a finite group.
Corollary 4.3.13 Let G be a group of prime order. Then G is cyclic.
Proof. Since |G| 2, there exists a G such that a 6= e. Let H = hai . Then {e} H and |H| divides |G| .
But |G| is prime and so |H| = |G| . Since H G and |H| = |G| , it follows that G = H. Therefore, G is cyclic.
Let H and K be subgroups of a group G. If either H or K is infinite, then, of course, HK is infinite. Suppose
H and K are both finite. We know that HK need not be a subgroup of G. Thus, |HK| need not divide |G| .
However, with the help of Lagranges theorem, we can determine |HK| . This is a very useful result and we will
use it very eectively in this text. In the next theorem, we determine |HK| when H and K are both finite.
Theorem 4.3.14 Let H and K be finite subgroups of a group G. Then
|HK| =

|H| |K|
.
|H K|

Proof. Let us write A = H K. Since H and K are subgroups of G, A is a subgroup of G and since A H,
A is also a subgroup of H. By Lagranges theorem, |A| divides |H| . Let n = |H|
|A| . Then [H : A] = n and so A
has n distinct left cosets in H. Let {x1 A, x2 A, . . . , xn A} be the set of all distinct left cosets of A in H. Then
H = n
i=1 xi A. Since A K, it follows that
n
HK = (n
i=1 xi A)K = i=1 xi K.

We now show that xi K xj K = if i 6= j. Suppose xi K xj K 6= for some i 6= j. Then xj K = xi K. Thus,

1
1
x1
i xj K. Since xi xj H, xi xj A and so xj A = xi A. This contradicts the assumption that x1 A, . . . , xn A
are all distinct left cosets. Hence, x1 K, . . . , xn K are distinct left cosets of K. Also, |K| = |xi K| by Corollary
4.3.7 for all i = 1, 2, . . . , n. Thus,
|HK| = |x1 K| + + |xn K|
= |K| + + |K|

~}

n tim es

=
=
=

n |K|

|H||K|
|A|
|H||K|
.
|HK|

The following corollary is an immediate consequence of the above theorem.

4.3. Lagranges Theorem

85

Corollary 4.3.15 Let H and K be finite subgroups of a group G such that H K = {e}. Then
|HK| = |H| |K| .
Let H and K be subgroups of a group G. If either H or K is infinite, then, of course, HK is infinite. Suppose
H and K are both finite. We know that HK need not be a subgroup of G. Thus, |HK| need not divide |G| .
However, with the help of Lagranges theorem, we can determine |HK| . This is a very useful result and we will
use it very eectively in this text. In the next theorem, we determine |HK| when H and K are both finite.
Theorem 4.3.16 Let H and K be finite subgroups of a group G. Then
|HK| =

|H| |K|
.
|H K|

Proof. Let us write A = H K. Since H and K are subgroups of G, A is a subgroup of G and since A H,
A is also a subgroup of H. By Lagranges theorem, |A| divides |H| . Let n = |H|
|A| . Then [H : A] = n and so A
has n distinct left cosets in H. Let {x1 A, x2 A, . . . , xn A} be the set of all distinct left cosets of A in H. Then
H = n
i=1 xi A. Since A K, it follows that
n
HK = (n
i=1 xi A)K = i=1 xi K.

We now show that xi K xj K = if i 6= j. Suppose xi K xj K 6= for some i 6= j. Then xj K = xi K. Thus,

1
1
x1
i xj K. Since xi xj H, xi xj A and so xj A = xi A. This contradicts the assumption that x1 A, . . . , xn A
are all distinct left cosets. Hence, x1 K, . . . , xn K are distinct left cosets of K. Also, |K| = |xi K| by Corollary
4.3.7 for all i = 1, 2, . . . , n. Thus,
|HK| = |x1 K| + + |xn K|
= |K| + + |K|
~}


n tim es

=
=
=

n |K|

|H||K|
|A|
|H||K|
.
|HK|

The following corollary is an immediate consequence of the above theorem.

Corollary 4.3.17 Let H and K be finite subgroups of a group G such that H K = {e}. Then
|HK| = |H| |K| .
The following is known as Fermats little theorem.
Theorem 4.3.18 (Fermat) Let p be a prime integer and a be an integer such that p does not divide a. Then p
divides ap1 1, i.e.,
ap1 p 1.
Proof. Let Up = Zp \{0}. Then by Exercise 10 (page 55), Up is a group. Also, by Exercise 9 (page 55),
|Up | = p 1. Let a be an integer such that p does not divide a. Then [a] is a nonzero element of Zp and so
[a] Up . Thus, by Corollary 4.3.12, [a]p1 = [1], i.e., [ap1 ] = [1]. Hence, ap1 p 1 by Exercise 11 (page 23).
We now discuss Eulers generalization of Fermats little theorem.
If p is a prime, then (p) = p 1. Therefore, Fermats little theorem can be written as follows:
If a is a positive integer and p is a prime such that gcd(a, p) = 1, then
a(p) p 1.
Euler generalized this result from the case of a prime to an arbitrary integer n. The following theorem is the
Eulers generalization of Fermats Little Theorem. We leave the proof as an exercise.
Theorem 4.3.19 (Euler) Let a and n be integers such that n > 0 and gcd(a, n) = 1. Then
a(n) n 1.

86

4. Subgroups and Normal Subgroups

Worked Out Exercises

Exercise 1 Let H be a subgroup of a group G. Show that for all a G, aH = H if and only if a H.
Solution Let a G. Suppose aH = H. Then a = ae aH = H. Conversely, suppose that a H. Now for any
h H, ah H. Hence, aH H. Let h H. Then a1 h H. Thus, h = a(a1 h) aH. Therefore,
H aH, proving that aH = H.

Exercise 2 Let G be a noncyclic group of order p2 , p a prime integer. Show that the order of each nonidentity element
is p.
Solution Let g G and g 6= e. Now (g) divides |G| = p2 . Hence, (g) = 1, p or p2 . Since g 6= e, (g) 6= 1. If
(g) = p2 , then G contains an element g such that (g) = |G| and this implies that G is cyclic, which
contradicts the hypothesis. Hence, (g) = p.
Exercise 3 Let G = {a, b, c, d} be a group. Complete the following Cayley table for this group.
a
a
b
c
d

b
b

Solution From the table, c2 = b and db = b. Now db = b implies that d = e, the identity element of G. Since
c2 = b 6= d, (c) 6= 2. Hence, (c) = 4. Thus, G is a cyclic group generated by c. Then G = {e , c, c2 , c3 }.
Since d = e and c2 = b, it follows that c3 = a. Hence, the Cayley table is

a
b
c
d

a
b
c
d
a

b
c
d
a
b

c
d
a
b
c

d
a
b
c
d

Exercise 4 Let G be a finite nontrivial group. Suppose for all x G, there exists y G such that x = y 2 . Prove that
the order of G is odd and conversely.
Solution Suppose G is of odd order. Then |G| = 2n + 1 for some positive integer n and for all x G, x2n+1 = e.
Now x2n+1 = e implies x = x2n = (xn )2 = y 2 , where y = xn . Conversely, suppose |G| is not odd. Let
|G| = 2n and x G. Then there exists y G such that x = y 2 . Hence, xn = y 2n = e. Thus, for all x G,
xn = e. Suppose n is odd, say, n = 2m + 1. Then x2m+1 = e for all x G. By Worked-Out Exercise 5
(page 52), there exists z G such that z 6= e and z 2 = e since |G| is even. Hence, e = z 2m+1 = zz 2m =
z(z 2 )m = ze = z, which is a contradiction. So n is even, say, n = 2m. Then x2m = e for all x G. As
before, we can show that xm = e for all x G and m is even. Continuing in this way, we can conclude
that x2 = e for all x G. Let x G. Then there exists y G such that x = y 2 . Therefore, x = e. Thus,
|G| = 1, which is a contradiction. Consequently, G is of odd order.
Exercise 5 Let G be a group such that |G| > 1. Prove that G has only the trivial subgroups if and only if |G| is prime.
Solution Let |G| = p, p a prime. Let H be a subgroup of G. Then |H| divides |G| . This implies that |H| = 1 or p.
Thus, H = {e} or H = G. Conversely, suppose that G has only the trivial subgroups. Let a G be such
that a 6= e. Now hai = {an | a Z} is a cyclic subgroup of G and hai 6= {e}. Therefore, G = hai . If G
is infinite, then ar 6= as for all r, s Z, r 6= s. Hence, {a2n | n Z} is a nontrivial subgroup of G, which
is a contradiction. Thus, |G| is a finite cyclic group of order, say, m > 1. Suppose m is not prime. Then
m = rs for some r, s Z, 1 < r, s < m. Since r | |G| and G is cyclic, G has a cyclic subgroup H of order
r. This contradicts the assumption that G has only the trivial subgroups. Hence, |G| is prime.

Exercise 6 Let G be a group of order pn , p a prime. Show that G contains an element of order p.

Solution: Let a G, a 6= e. Then H = hai is a cyclic subgroup of G. Now |H| divides |G| = pn . Thus,
|H| = pm for some m Z, 0 < m n. Now H is a cyclic group of order pm . Hence, for every divisor d of pm ,
there exists a subgroup of order d. So for p, there exists a subgroup T of H such that |T | = p. By Corollary
4.3.13, there exists b T such that T = hbi and b is of order p. Hence, G contains an element of order p.
Exercise 7 Let G be a finite commutative group such that G contains two distinct elements of order 2. Show that |G|
is a multiple of 4. Also, show that this result need not be true if G is not commutative.

4.3. Lagranges Theorem

Solution

87

Let a and b be two distinct elements of order 2. Let H = {e, a} and K = {e, b}. Now H and K are
subgroups of G. Since G is commutative, HK = {e, a, b, ab} is a subgroup of G of order 4. Now |HK| = 4
divides |G| . Thus, |G| is a multiple of 4.

The symmetric group S3 is noncommutative, (1 2) and (1 3) are elements of S3 , and each is of order 2. But
4 does not divide |S3 | = 6.
Exercise 8 Find all subgroups of S3 .
Solution S3 = {e, (1 2), (1 3), (2 3), (1 2 3), (1 3 2)}. (1 2) = 2, (1 3) = 2, (2 3) = 2, (1 2 3) = 3, and (1 3
2) = 3. Now {e}, {e, (1 2)}, {e, (1 3)}, {e, (2 3)}, {e, (1 2 3), (1 3 2)}, and S3 are subgroups of S3 . Let H
be a subgroup of S3 . Now |H| divides |G| . Thus, |H| = 1, 2, 3, or 6. If |H| = 1, then H = {e}. If |H| = 6,
then H = S3 . If |H| = 2, then H is a cyclic group of order 2. Hence, H is one of {e, (1 2)}, {e, (1 3)}, {e,
(2 3)}. Suppose |H| = 3. Then by Lagranges theorem, H has no subgroup of order 2. Thus, (1 2), (1 3),
(2 3)
/ H. Therefore, e, (1 2 3), (1 3 2) H. Also, {e, (1 2 3), (1 3 2)} is a subgroup and so H = {e, (1 2
3), (1 3 2)}. Hence, H0 = {e}, H1 = {e, (1 2)}, H2 = {e, (1 3)}, H3 = {e, (2 3)}, H4 = {e, (1 2 3), (1 3
2)}, and S3 are the only subgroups of S3 .

Exercises
1. In S3 ,
(a) find all the right cosets of H = {e, (2 3)},
(b) find a subgroup B of G such that H(1 2 3) is a left coset of B.
2. Find all right cosets of the subgroup 6Z in the group (Z, +).
3. Let
H=

 
1
e,
4

2
3

3
2

4
1

1
2

2
1

3
4

4
3

 
1
,
3

2
4

3
1

4
2

 

where e is the identity permutation. Show that H is a subgroup of S4 . List all the left and right cosets of
H in S4 .
4. Let H denote the subgroup {r360 , h} of the group of symmetries of the square. List all the left and right
cosets of H in G.
5. Find all subgroups of the Klein 4-group.
6. Find all subgroups of order 4 in S4 .
7. Let G = {a, b, c, d} be a group. Complete the following Cayley table for this group.
d
a
b
c

d
d

8. Let G be a group and H and K be subgroups of G. Show that (H K)x = Hx Kx for all x G.
9. Let G be a group and H and K be subgroups of G. Let a, b G. Show that either Ha Kb = or
Ha Kb = (H K)c for some c G.
10. (Poincar) Let G be a group and H and K be subgroups of G of finite indices. Show that H K is of
finite index.
11. Give an example of a group G and a subgroup H of G such that aH = bH, but Ha 6= Hb for some a, b G.
12. Let G be a group of order pq, where p and q are prime integers. Show that every proper subgroup of G is
cyclic.
13. Let H be a subgroup of a group G. Define a relation on G by for all a, b G, a b if and only if
b1 a H (i.e., if and only if aH = bH). Show that is an equivalence relation on G and the equivalence
classes of are the cosets aH, a G.
14. Let n > 1. Show that there exists a proper subgroup H of Sn such that [Sn : H] n.

88

4. Subgroups and Normal Subgroups

s
s
15. Let H and K be subgroups of a finite group G such that |H| >
|G| and |K| >
|G|. Show that
|H K| > 1.
16. Let |G| = pq, (p > q), where p and q are distinct primes. Show that G has at most one subgroup of order
p.
17. Let G be a group. If a subset A is a left coset of some subgroup of G, then show that A is a right coset of
some subgroup of G.
18. Let G be a finite group and A and B be subgroups of G such that A B G. Prove that
[G : A] = [G : B][B : A].
19. Let G be a group such that |G| < 200. Suppose G has subgroups of order 25 and 35. Find the order of G.
20. Let G be a group of order 35 and A and B be subgroups of G of order 5 and 7, respectively. Show that
G = AB.
21. Let A and B be subgroups of a group G. If |A| = p, a prime integer, show that either A B = {e} or
A B.
22. Let H and K be subgroups of a group G. Define a relation on G by for all a, b G, a b if and only if
b = hak for some h H and k K.
(a) Show that is an equivalence relation on G.
(b) Let a G and [a] denote the equivalence class of a in G. Show that
[a] = {hak | h H, k K} = HaK.
The set HaK is called a double coset of H and K in G.
(c) If G is a finite group, prove that
|H| |K|
|HaK| =
|H aKa1 |
for all a G.
23. For the following, if the statement is true, then write the proof. Otherwise justify why the statement is
false.
(a)
(b)
(c)
(d)
(e)
(f)

4.4

Every left coset of a subgroup of a group is also a right coset.

The product of two left cosets of a subgroup of a group is also a left coset.
There may exist a subgroup of order 12 in a group of order 40.

Let G = hai be a cyclic group of order 30. Then [G : a5 ] = 5.
Every proper subgroup of a group of order p2 (p a prime) is cyclic.
Let G be a group. If H is a subgroup of order p and K is a subgroup of order q, where p and q are
distinct primes, then |HK| = pq.

Normal Subgroups and Quotient Groups

In the previous section, we saw that a subgroup H of a group G induced two decompositions of G, one by left
cosets and another by right cosets. In other words, if H is a subgroup of a group G, then G can be written
as a disjoint union of distinct left (right) cosets of H in G. These two decompositions were first recognized
by Galois in 1831 in the context of permutation groups. Galois called the decomposition proper if the two
decompositions coincide, i.e., if left cosets are the same as right cosets. We call such a subgroup normal in our
present-day terminology. Normal subgroups are the subject of this section. Galois showed how the solvability
of a polynomial equation by means of radicals is related to the concept of a normal subgroup of the group of
permutations of the roots and the group, called the quotient group, created by the normal subgroup.
Perhaps the notion of a normal subgroup is one of the most innovative ideas in group theory. I.N. Herstein
(19231988) remarked about normal subgroups that It is a tribute to the genius of Galois that he recognized that
those subgroups for which the left and right cosets coincide are distinguished ones. Very often in mathematics
the crucial problem is to recognize and to discover what are the relevant concepts; once this is accomplished the
job may be more than half done.
Later C. Jordan defined normal subgroups without using the term normal as we define it in our present-day
terminology.
We shall see in this text that normal subgroups play a crucial role in obtaining structural results of groups.
Let us now begin our study of normal subgroups.

4.4. Normal Subgroups and Quotient Groups

89

Definition 4.4.1 Let G be a group. A subgroup H of G is said to be a normal (or invariant) subgroup of G
if aH = Ha for all a G.
From the definition of a normal subgroup, it follows that for any group G, G and {e} are normal subgroups
of G.
If H is a normal subgroup of G, this does not always mean that ah = ha for all h H and for all a G as
shown by the following example.
Example 4.4.2 Recall Example 4.3.2. H is a normal subgroup of S3 . Consider h =


1
1

2
3

3
2

1
1

2
3

3
2

and

Hence,

even though

1
1

1
1

2
3
2
3

3
2

3
2

1
3

2
2

3
1

1
2

2
1

3
3

h 6= h

3
2

h=


H=H

1
1
1
1

2
3
2
3

3
2

1
2

2
3

3
1

H. Then

The following theorem gives a necessary and sucient condition for a subgroup to be a normal subgroup.
For a G, 6= H G, let aHa1 = {aha1 | h H}.
Theorem 4.4.3 Let H be a subgroup of a group G. Then H is a normal subgroup of G if and only if for all
a G, aHa1 H.
Proof. First suppose that H is a normal subgroup of G. Let a G. We now show that aHa1 H. Let
aha1 aHa1 , where h H. Since H is a normal subgroup of G, aH = Ha. Also, since ah aH, we have
ah Ha and so ah = h0 a for some h0 H. Thus, aha1 = h0 H. Hence, aHa1 H.
Conversely, suppose aHa1 H for all a G. Let a G. We show that aH = Ha. Let ah aH, where
h H. Now aha1 aHa1 and so aha1 H. Thus, aha1 = h0 for some h0 H. This implies that
ah = h0 a Ha. Therefore, aH Ha. Similarly, we can show that Ha aH. Hence, aH = Ha. Consequently,
H is a normal subgroup of G.
There are several other criteria that can be used to test the normality of a subgroup. We consider some of
these criteria in exercises at the end of this section.
The following theorem describes some important properties of normal subgroups.
Theorem 4.4.4 Let H and K be normal subgroups of a group G. Then
(i) H K is a normal subgroup of G,
(ii) HK = KH is a normal subgroup of G,
(iii) hH Ki = HK.
Proof. (i) Since the intersection of subgroups is a subgroup, H K is a subgroup of G. Let g G. Consider
g(H K)g1 . Let gag 1 be any element of g(H K)g 1 , where a H K. Since a H K, we have a H and
a K. Hence, gag 1 H and gag1 K. Thus, gag1 H K. This shows that g(H K)g1 H K. Hence,
H K is a normal subgroup by Theorem 4.4.3.
(ii) First we show that HK = KH. Let hk HK, where h H and k K. Since K is a normal subgroup
of G and h G, we have hK = Kh. Thus, hk hK = Kh. Since Kh KH, we have hk KH. Hence,
HK KH. Similarly, KH HK and so HK = KH. Since H and K are subgroups and HK = KH, HK is a
subgroup of G by Theorem 4.1.18. To show that HK is a normal subgroup, let g G. Then gHg1 H and
gKg 1 K since H and K are normal subgroups. Now
g(HK)g1

=
=

g(Hg 1 gK)g1
(gHg 1 )(gKg1 )
HK.

Therefore, HK is a normal subgroup of G by Theorem 4.4.3.

90

4. Subgroups and Normal Subgroups

(iii) By (ii), HK is a subgroup of G. Hence, by Theorem 4.1.20,
HK = hH Ki .

We know that if H and K are subgroups of a group G, then HK need not be a subgroup of G (Example
4.1.17). By the above theorem, if H and K are normal subgroups, then HK is a normal subgroup and hence
a subgroup. However, in order to show that HK is a subgroup, we only need either H or K to be a normal
subgroup. We consider one of these situations in Exercise 14 (page 93).
We now focus our attention on the study of quotient groups. First, let us consider the following example.
Example 4.4.5 Consider the subgroup H 0 of Example 4.3.2. Now H 0 is not a normal subgroup of S3 . Let S3 /H 0
be the set of all left cosets of H 0 in S3 . Now let us try to define a binary operation on S3 /H 0 . The natural way
would be to define (1 H 0 ) (2 H 0 ) to be ( 1 2 )H 0 . Now




1 2 3
1 2 3
H0
H0 =
3 1 2
3 2 1
and

However,

and

Since






1
2

2
1

3
3

1
2

2
3

3
1

1
2

2
1

H0

3
3

H0 =

 
1

2
2

3
1

2
1

3
2

H 0 6=

 
1
H
3
0

1
3

2
1

3
2

1
2



1
1

2
3

3
1

H0

2
2

=
3
3

H0.

1
3

2
1

3
2

1
1

2
2

3
3

H0

H0.

H0,

is not well defined. That is not well defined is due to the fact that H 0 is not a normal subgroup of S3 .
Theorem 4.4.6 Let H be a normal subgroup of a group G. Denote the set of all left cosets {aH | a G} by
G/H and define on G/H by for all aH, bH G/H,
(aH) (bH) = abH.
Then (G/H, ) is a group.
Proof. First we show that is well defined. Let aH, bH, a0 H, b0 H G/H and suppose (aH, bH) = (a0 H,
b0 H). Then aH = a0 H and bH = b0 H. We need to show that aH bH = a0 H b0 H or abH = a0 b0 H. Now
aH = a0 H and bH = b0 H imply that a = a0 h1 and b = b0 h2 for some h1 , h2 H. Thus,
(a0 b0 )1 (ab)

=
=
=

b01 a01 ab
b01 a01 a0 h1 b0 h2
b01 h1 b0 h2 .

Since H is a normal subgroup and h1 H, we have b01 h1 b0 h2 = (b01 h1 b0 )h2 H and so (a0 b0 )1 (ab) H.
Hence, abH = a0 b0 H by Theorem 4.3.3(i). Thus, is well defined and so (G/H, ) is a mathematical system.
Next, we show that is associative. Let aH, bH, cH G/H. Now (aH) [(bH) (cH)] = (aH) (bcH) =
a(bc)H = (ab)cH = (abH) (cH) = [(aH) (bH)] (cH). Hence, is associative. Now eH G/H and
(aH) (eH) = aeH = aH = eaH = (eH) (aH)
for all aH G/H. Therefore, eH is the identity of G/H. Also, for all aH G/H, a1 H G/H and
(aH) (a1 H) = aa1 H = eH = a1 aH = (a1 H) (aH).
Thus, for all aH G/H, a1 H is the inverse of aH. Consequently, (G/H, ) is a group.
Definition 4.4.7 Let G be a group and H be a normal subgroup of G. The group G/H is called the quotient
group of G by H.

4.4. Normal Subgroups and Quotient Groups

91

Example 4.4.8 Consider the subgroup (hni , +) of the group (Z, +), where n is a fixed positive integer. Since Z
is commutative, hni is a normal subgroup of Z (Exercise 16, page 93). Hence, (Z/ hni , +) is a group, where
(a + hni) + (b + hni) = (a + b) + hni
for all a + hni , b + hni Z/ hni . In Example 4.3.10, we determined the distinct left cosets of hni in Z. We found
that
Z/ hni = {0 + hni , 1 + hni , 2 + hni , . . . , n 1 + hni}.
Example 4.4.9 Consider the normal subgroup H of S3 of Example 4.4.2. Since |S3 | = 6 and |H| = 3, [S3 :
H] = 2 by Lagranges theorem. Now |S3 /H| = [S3 : H] = 2 and for all h H, hH = H. Thus, eH = H, (1 2
3)H = H and (1 3 2)H = H. We have shown in Example 4.3.2 that (2 3)H = (1 3)H = (1 2)H. Thus,
S3 /H = {H, (2 3)H} .
We also note that S3 /H is cyclic and (2 3)H is a generator for S3 /H.
Example 4.4.10 Consider Z8 and let H = {[0], [4]}. Then H is a normal subgroup of Z8 . Now |H| = 2 and
8|
|Z8 | = 8. Thus, |Z8 /H| = |Z
|H| = 4. Hence, Z8 /H has four elements. Now
[0] + H = H = [4] + H,
[1] + H = {[1], [5]} = [5] + H,
[2] + H = {[2], [6]} = [6] + H,
and
[3] + H = {[3], [7]} = [7] + H.
Hence, Z8 /H = {[0] + H, [1] + H, [2] + H, [3] + H}.
Example 4.4.11 Consider Z4 Z6 , the direct product of Z4 and Z6 . Let
H = h([0], [1])i = {([0], [0]), ([0], [1]), ([0], [2]), ([0], [3]), ([0], [4]), ([0], [5])}.
Then H is a subgroup of Z4 Z6 and since Z4 Z6 is commutative, H is a normal subgroup of Z4 Z6 . Now
|Z4 Z6 | = 24 and |H| = 6. Hence,
|(Z4 Z6 )/H| =

|Z4 Z6 |
= 4.
|H|

Thus, (Z4 Z6 )/H has four elements. Since for all [n] Z6 , ([0], [n]) H, we have for all [n] Z6 , ([0], [n])+H =
H. Let ([m], [n]) Z4 Z6 . Then ([m], [n]) = ([m], [0]) + ([0], [n]) and from this, it follows that ([m], [n]) + H =
([m], [0]) + H. Let us now compute ([m], [0]) + H for m = 0, 1, 2, 3. Now ([0], [0]) + H = H,
([1], [0]) + H = {([1], [0]), ([1], [1]), ([1], [2]), ([1], [3]), ([1], [4]), ([1], [5])},
([2], [0]) + H = {([2], [0]), ([2], [1]), ([2], [2]), ([2], [3]), ([2], [4]), ([2], [5])},
and
([3], [0]) + H = {([3], [0]), ([3], [1]), ([3], [2]), ([3], [3]), ([3], [4]), ([3], [5])}.
From above, we see that ([0], [0]) + H, ([1], [0]) + H, ([2], [0]) + H, and ([3], [0]) + H are all distinct. Hence,
(Z4 Z6 )/H = {([0], [0]) + H, ([1], [0]) + H, ([2], [0]) + H, ([3], [0]) + H}.

92

4. Subgroups and Normal Subgroups

Worked-Out Exercises
Exercise 1 Let H be a subgroup of a group G. Then W = gG gHg1 is a normal subgroup of G.

Solution: By Worked-Out Exercise 1 (page 75), gHg 1 is a subgroup of G for all g G. Since the intersection of
subgroups is a subgroup, W is a subgroup of G. Let x G, w W. Then w gHg1 for all g G. We
show that xwx1 gHg1 for all g G, which in turn will yield that xwx1 W. Let g G.
Let us work our way backward and suppose xwx1 gHg1 . Then xwx1 = ghg1 for some h H.
Thus, g1 xw x1 g = h H. This implies that
(g 1 x)w(g 1 x)1 H.
Set y = x1 g. Then g = xy. Hence, in order to show that xwx1 gHg1 for a given g G, first we need
to find y G such that g = xy. Since g = x(x1 g), we can choose y = x1 g.
So there exists y G such that g = xy. Since y G, we have w yHy 1 and so w = yhy 1 for some
h H. Therefore, xwx1 = x(yhy1 )x1 = xyhy1 x1 = (xy)h(xy)1 = ghg 1 gHg 1 . Since g G
was arbitrary, xwx1 gHg 1 for all g G. Consequently, W is a normal subgroup of G.

Exercise 2 Let H be a subgroup of G.

(a) If x2 H for all x G, prove that H is a normal subgroup of G and G/H is commutative.

(b) If [G : H] = 2, prove that H is a normal subgroup of G.

Solution:

(a) Let g G and h H. Consider ghg1 and note that

ghg1 = (gh)2 h1 g 2 .
Now h1 H and by our hypothesis (gh)2 , g2 H. This implies that ghg1 H, which in turn
shows that gHg 1 H. Hence, H is a normal subgroup of G. To show that G/H is commutative, let
xH, yH G/H. We wish to show that xHyH = yHxH or xyH = yxH or (yx)1 (xy) H. Consider
(yx)1 (xy). Now
(yx)1 (xy) = (x1 y 1 )(xy) = (x1 y 1 )2 (yxy 1 )2 y 2 .
Since a2 H for all a G, it follows that (x1 y 1 )2 (yxy 1 )2 y 2 H and so (yx)1 (xy) H. Thus,
G/H is commutative.
(b) We prove that H is a normal subgroup of G first by showing that x2 H for all x G and then by
/ H. Then x
/ H and so H and xH are distinct
using (i). Suppose there exists x G such that x2
left cosets of H in G. Since [G : H] = 2, it follows that G/H = {H, xH}. Hence, G = H xH. This
implies that x2 H xH. Since x2
/ H, we must have x2 xH . Hence, x2 = xh for some h H.
But then x = h H, which is a contradiction. Hence, x2 H for all x G. By (i), H is a normal
subgroup of G.

Exercise 3 Let G be a group such that every cyclic subgroup of G is a normal subgroup of G. Prove that every
subgroup of G is a normal subgroup of G.
Solution: Let H be a subgroup of G. Let g G and a H. Then g 1 ag hai H. Hence, H is normal in G.

Exercise 4 Let H be a proper subgroup of G such that for all x, y G\H, xy H. Prove that H is a normal subgroup
of G.
Solution: Let x G\H. Then x1 G\H. Let y H. Then xy G\H. Thus, xy, x1 G\H. Hence, xyx1 H.
Therefore, H is a normal subgroup of G.
Exercise 5 Let G be a group and {Ni | i } be a family of proper normal subgroups of G. Suppose G = i Ni and
Ni Nj = {e} for i 6= j. Prove that G is commutative.

Solution: Let x, y G. Then there exist i and j such that x Ni and y Nj . If i 6= j, then since Ni Nj = {e},
xy = yx (Exercise 13, page 93). Let i = j. Now there exists z G such that z
/ Ni . Then zx
/ Ni . Hence,
zx Nl for some l 6= i and so (zx)y = y(zx). Thus, z(xy) = (zx)y = y(zx) = (yz)x = (zy)x = z(yx). This
implies that xy = yx. Consequently, G is commutative.

Exercise 6 Let H be a subgroup of a group G. Suppose that the product of two left cosets of H in G is again a left
coset of H in G. Prove that H is a normal subgroup of G.
Solution: Let g G. Then gHg1 H = tH for some t G. Thus, e = geg 1 e tH. Hence, e = th for some h H.
Thus, t = h1 H so that tH = H. Now gHg 1 gHg1 H = H. Therefore, H is a normal subgroup of
G.

4.4. Normal Subgroups and Quotient Groups

93

Exercise 7 Let G be a group. Show that if G/Z(G) is cyclic, then G is commutative.

Solution: Write Z = Z(G). Let G/Z = hgZi . Let a, b G. Then aZ, bZ G/Z. Hence, aZ = gn Z and bZ = g m Z
for some n, m Z. Then a gn Z and b gm Z. Thus, a = gn d and b = g m h for some d, h Z. Now
ab = g n dg m h = g n gm dh (since d Z) = gn+m hd (since h Z) = g m g n hd = g m hgn d = ba. Hence, G is
commutative.

Exercises
1. Let
H=

 
1
e,
4

2
3

3
2

4
1

1
2

2
1

3
4

4
3

 
1
,
3

2
4

3
1

4
2

 

where e is the identity permutation. Determine whether or not H is a normal subgroup of S4 .

2. Let H denote the subgroup {r360 , h} of the group of symmetries of the square. Determine whether or not
H is a normal subgroup of G.
3. Let G be a group and H be a subgroup of G. Show that H is normal if and only if ghg 1 H for all
g G, h H.
4. Let G be a group and H be a subgroup of G. If for all a, b G, ab H implies ba H, prove that H is a
normal subgroup of G.

5. Let H be a proper subgroup of a group G and a G, a

/ H. Suppose that for all b G, either b H or
Ha = Hb. Show that H is a normal subgroup of G.
6. Let G be a group. Prove that Z(G) is a normal subgroup of G.
7. Let G be a group. Let H be a subgroup of G such that H Z(G). Show that if G/H is cyclic, then
G = Z(G), i.e., G is commutative.
8. Let H and K be subgroups of a group G such that H is a normal subgroup of G. Prove that H K is a
normal subgroup of K.
9. Determine the quotient groups of
(a) (E, +) in (Z, +),
(b) (Z, +) in (Q, +),
(c) (h[4]i , +12 ) in (Z12 , +12 )
(d) (4Z, +) in (Z, +).
10. Let H be a normal subgroup of a group G. Prove that if G is commutative, then so is the quotient group
G/H.
11. If H is a subgroup of finite index in (C , ),then prove that H = C

12. Let H be a nonempty subset of a group G. The set N(H) = {a G | aHa1 = H} is called the normalizer
of H in G.
(a) Prove that N(H) is a subgroup of G.
Suppose H is a subgroup of G.
(b) Prove that H is normal in G if and only if N(H) = G.
(c) Prove that H is normal in N(H).
(d) Prove that N(H) is the largest subgroup of G in which H is normal, i.e., if H is normal in a subgroup
K of G, then K N(H).
13. Let H and K be normal subgroups of a group G. If H K = {e}, prove that hk = kh for all h H and
k K.

14. Let G be a group. Let H be a subgroup of G and K be a normal subgroup of G. Prove that HK is a
subgroup of G.
15. Give an example of a noncommutative group in which every subgroup is normal.
16. Show that every subgroup of a commutative group is normal.
17. Let H be a normal subgroup of a group G such that |H| = 2. Show that H Z(G).

18. Show that if H is the only subgroup of order n in a group G, then H is a normal subgroup of G.

94

4. Subgroups and Normal Subgroups

19. Let K = {e, (1 2) (3 4), (1 4) (3 2), (1 3) (2 4)}.

(a) Show that K is the only subgroup of order 4 in A4 .
(b) Show that K is a normal subgroup of A4 .
20. Show that A4 has no subgroup of order 6.
21. Find all subgroups of A4 .
22. Prove that An is the only subgroup of index 2 in Sn .
23. Let G be a group. An equivalence relation on G is called a congruence relation if
for all a, b, c G, ab implies that cacb and acbc.
Let H be a normal subgroup of G. Define the relation H on G by
for all a, b G, aH b if and only if a1 b H.
Prove that
(a) H is a congruence relation on G,
(b) The H class aH = {b G | aH b} is the left coset aH,
(c) H = eH .

24. Let H be a subgroup of a group G. Define a relation H on G by H = {(a, b) G G | a1 b H}. Show

that if H is a congruence relation, then H is a normal subgroup of G.
25. Let be a congruence relation on a group G. Show that there exists a normal subgroup H of G such that
= {(a, b) G G | a1 b H}.

26. For the following statements, write the proof if the statement is true; otherwise, give a counterexample.
(i) A subgroup H of a group G is a normal subgroup if and only if every right coset of H is also a left
coset.
(ii) If A, B and C are normal subgroups of a group G, then A(B C) is a normal subgroup of G.
(iii) If A is a normal subgroup of a finite group G, then [G : A] = 2.
(iv) Every commutative subgroup of a group G is a normal subgroup of G.
(v) If G is a group of order 2p, p an odd prime, then either G is commutative or G contains a normal
subgroup of order p.
(vi) If every element of a group G is of finite order, then G is a finite group.
(vii) A5 is the only nontrivial normal subgroup S5 .

4.4. Normal Subgroups and Quotient Groups

95

Joseph Louis Lagrange (17361813) was born on January 25, 1736, in Turin, Italy. He spent the early
part of his life in Turin. While there he was involved in carrying out research work in calculus of variations and
mechanics.
In 1766, Lagrange was invited by the Prussian king, Frederick II, to fill the position vacated by Euler in
Berlin. Frederick the Great proclaimed in his appointment that the greatest king in Europe ought to have
the greatest mathematician in Europe. In 1787, after the death of Frederick II, he went to Paris, accepting an
invitation from Louis XVI. In 1797, he accepted a position at the newly formed cole Polytechnique in Paris.
He was made a count by Napoleon and remained at the cole Polytechnique till his death. He died on April 10,
1813.
Throughout his life, Lagrange did work of fundamental importance. He made numerous contributions to
many branches of mathematics, including number theory, the theory of equations, dierential equations, celestial
mechanics, and fluid mechanics. In 1770, he proved the famous Lagranges theorem in group theory.
He is responsible for the work leading to Galois theory. In his paper, Rflexion sur la thorie algbriques des
quations, Lagrange carefully analyzed the various known methods to solve a polynomial equation of degree 4
by means of radicals. He was interested in finding a general method of solution for polynomials of higher degree.
He was unable to find a general solution, but in his paper he introduced several key ideas on the permutations
of roots which finally led Abel and Galois to develop the necessary theory to answer the question. Lagranges
work on the solution of polynomial equations is one of the sources from which modern group theory evolved.

96

4. Subgroups and Normal Subgroups

Chapter 5

Homomorphisms and Isomorphisms

of Groups
One of the main uses of the concept of an isomorphism is the classification of algebraic structuresin particular,
groups. Readers with some knowledge of linear algebra may recall that the concept of an isomorphism is used
to completely characterize vector spaces with the same field of scalars in terms of a single integer, the dimension
of the vector space. Another important use of an isomorphism is the representation of one algebraic structure
by means of another. This is done in linear algebra, where it is shown that the vector space of all linear
transformations from one finite dimensional vector space into another is isomorphic to a certain vector space of
matrices.

5.1

Homomorphisms of Groups

In this section, we consider certain mappings between groups. These mappings will be defined in such a way as
to preserve the algebraic structure of the groups involved. More precisely, suppose we are given a function f from
a group G into a group G1 , where 1 denotes the operation of G1 . Let a, b G. Then under f, a corresponds to
f (a), b to f(b), and a b to f (a b). If f is to preserve the operations of G and G1 , a b must correspond to
f (a) 1 f (b). Since f is a function, this forces the requirement that f (a b) = f (a) 1 f (b).
Definition 5.1.1 Let (G, ) and (G1 , 1 ) be groups and f a function from G into G1 . Then f is called a homomorphism of G into G1 if for all a, b G,
f (a b) = f (a) 1 f (b).
Let the identity element of the group G1 be denoted by e1 .
Define f : G G1 by f (a) = e1 for all a G. Since f (a b) = e1 = e1 1 e1 = f(a) 1 f (b) for all a, b G,
we find that f is a homomorphism from G into G1 . This shows that there always exists a homomorphism from
a group G into a group G1 . This homomorphism is called the trivial homomorphism.
The identity map from G onto G is also a homomorphism.
Before we consider more examples of homomorphisms, let us prove some basic properties of homomorphisms.
Theorem 5.1.2 Let f be a homomorphism of a group G into a group G1 . Then
(i) f (e) = e1 .
(ii) f (a1 ) = f (a)1 for all a G.
(iii) If H is a subgroup of G, then f (H) = {f (h) | h H} is a subgroup of G1 .
(iv) If H1 is a subgroup of G1 , then f 1 (H1 ) = {g G | f (g) H1 } is a subgroup of G, and if H1 is a
normal subgroup, then f 1 (H1 ) is a normal subgroup of G.
(v) If G is commutative, then f (G) is commutative.
(vi) If a G is such that (a) = n, then (f (a)) divides n.
Proof. (i) Since f is a homomorphism, f (e)f(e) = f (ee) = f (e) = f (e)e1 . This implies that f (e) = e1 by
the cancellation law.

97

98

5. Homomorphisms and Isomorphisms of Groups

(ii) Let a G. Then f(a)f (a1 ) = f (aa1 ) = f(e) = e1 . Similarly, f (a1 )f(a) = e1 . Since f (a) has a unique
inverse, f(a1 ) = f (a)1 .
(iii) Let H be a subgroup of G. Then e H and by (i), f (e) = e1 . Thus, e1 = f(e) f (H) and so f(H) 6= .
Let f (a), f (b) f (H), where a, b H. Since H is a subgroup, ab1 H. Thus, f (a)f(b)1 = f (a)f (b1 ) =
f (ab1 ) f (H). Hence, by Theorem 4.1.6, f (H) is a subgroup of G1 .
(iv) By (i), e f 1 (H1 ) and so f 1 (H1 ) 6= . Let a, b f 1 (H1 ). Then f (a), f(b) H1 . Hence, f (ab1 ) =
f (a)f(b1 ) = f (a)f (b)1 H1 and so ab1 f 1 (H1 ). Thus, by Theorem 4.1.6, f 1 (H1 ) is a subgroup
of G. Suppose H1 is a normal subgroup of G1 . Let g G. We now show that gf 1 (H1 )g 1 f 1 (H1 ).
Let a gf 1 (H1 )g1 . Then a = gbg 1 for some b f 1 (H1 ). Now f (a) = f (gbg1 ) = f (g)f(b)f (g 1 ) =
f (g)f (b)f (g)1 H1 since H1 is a normal subgroup of G1 and f (b) H1 . Hence, a f 1 (H1 ) and this shows
that gf 1 (H1 )g 1 f 1 (H1 ). Thus, f 1 (H1 ) is a normal subgroup of G.
(v) Suppose G is commutative. Let f (a), f (b) f (G). Then f (a)f (b) = f (ab) = f (ba) = f (b)f(a). Hence,
f (G) is commutative.
(vi) Since (f (a))n = f (an ) = f(e) = e1 , we have (f (a)) divides n by Theorem 2.1.46.
Definition 5.1.3 Let f be a homomorphism of a group G into a group G1 . The kernel of f, written Ker f, is
defined to be the set
Ker f = {a G | f(a) = e1 }.
By Theorem 5.1.2, e Ker f.
Example 5.1.4 Define the function f from (Z, +) into (Zn , +n ) by f (a) = [a] for all a Z. From the definition
of f, it follows that f maps Z onto Zn . Let a, b Z. Then
f (a + b) = [a + b] = [a] +n [b] = f(a) +n f (b).
Thus, f is a homomorphism of Z onto Zn . Now
Ker f

=
=
=
=
=

{a Z | f(a) = [0]}
{a Z | [a] = [0]}
{a Z | a is divisible by n}
{a Z | a = qn for some q Z}
{qn | q Z}.

The above example shows that a nontrivial finite group may be an image of an infinite group under a
homomorphism. By Theorem 5.1.2(v), a noncommutative group cannot be an image under a homomorphism of
a commutative group. In the next example, we show that two finite groups G and G1 having same number of
elements need not have a homomorphism from G onto G1 .
Example 5.1.5 The groups Z4 Z4 and Z8 Z2 are commutative and each is of order 16. Suppose there exists
a homomorphism f of Z4 Z4 onto Z8 Z2 . Now a = ([7], [0]) Z8 Z2 and (a) = 8. Since f is onto Z8 Z2 ,
there exists b Z4 Z4 such that f (b) = a. By Theorem 5.1.2(vi), (f (b)) divides (b). Since (f (b)) = 8 and
Z4 Z4 has elements of order 1, 2, and 4 only, (f (b)) cannot divide (b). This is a contradiction. Hence, there
does not exist any homomorphism from Z4 Z4 onto Z8 Z2 .
Definition 5.1.6 Let G and G1 be groups. A homomorphism f : G G1 is called an epimorphism if f is
onto G1 and f is called a monomorphism if f is one-one. If there is an epimorphism f from G onto G1 , then
G1 is called a homomorphic image of G.
The homomorphism in Example 5.1.4 is an epimorphism, but not a monomorphism.
Example 5.1.7 Let R be the group of all nonzero real numbers under multiplication. Define f : R R by
f (a) = |a| . Now f (ab) = |ab| = |a| |b| = f (a)f(b), which implies that f is a homomorphism. Since f(1) = 1 =
f (1) and 1 6= 1, f is not one-one. Also, from the definition of f, it follows that f is not onto R . Hence, f is
neither an epimorphism nor a monomorphism.
The following theorem gives a necessary and sucient condition for a homomorphism to be a one-one mapping
in terms of its kernel.
Theorem 5.1.8 Let f be a homomorphism of a group G into a group G1 . Then f is one-one if and only if Ker
f = {e}.

5.1. Homomorphisms of Groups

99

Proof. Suppose f is one-one. Let a Ker f. Then f (a) = e1 = f (e) by Theorem 5.1.2(i). Since f is
one-one, we must have a = e. Hence, Ker f = {e}. Conversely, suppose that Ker f = {e}. Let a, b G. Suppose
f (a) = f (b). Then
f (ab1 ) = f (a)f (b1 ) = f (a)f(b)1 = e1 .
Thus, ab1 Ker f = {e} and so ab1 = e, i.e., a = b. This proves that f is one-one.
Theorem 5.1.9 Let f be a homomorphism of a group G into a group G1 . Then Ker f is a normal subgroup of
G.
Proof. Since e Ker f, Ker f 6= . Let a, b Ker f. Then f (ab1 ) = f (a)f (b1 ) = f (a)f (b)1 = e1 (e1 )1 =
e1 e1 = e1 . Thus, ab1 Ker f and hence Ker f is a subgroup of G by Theorem 4.1.6. Let a G and h Ker
f. Then f (aha1 ) = f (a)f (h)f (a1 ) = f(a)f (h)f (a)1 = f(a)e1 f (a)1 = e1 . Therefore, aha1 Ker f. This
proves that aKer f a1 Ker f. Hence, Ker f is a normal subgroup of G by Theorem 4.4.3.


a b
| a, b, c, d R, ad bc 6= 0 be the noncommutative group of Exc d
ample 2.1.14. Let R be the group of all nonzero real numbers under multiplication. Define f : GL(2, R) R
by


a b
f
= ad bc
c d



 

a b
a b
u v
for all
GL(2, R). Let
,
GL(2, R). Now
c d
c d
w s
Example 5.1.10 Let GL(2, R) =



a
c

b
d





u
w

v
s



=
=
=
=




au + bw av + bs
f
cu + dw cv + ds
(au + bw)(cv + ds) (av + bs)(cu + dw)
(ad
vw)
 bc)(us


a b
u v
f
f
.
c d
w s

This proves that f is a homomorphism. To show that f is onto R , let a R . Then







a 0
a 0
a 1
f
= a. Hence, f is onto R . Since f
=a=f
and
0 1
0 1
0 1
f is not one-one.

a
0


0
1
a
0

GL(2, R) and
 

0
a 1
6=
,
1
0 1

The previous example shows that there may exist a homomorphism of a noncommutative group onto a
commutative group.
Example 5.1.11 Consider S3 and the normal subgroup

 
1 2 3
1 2
H=
,
1 2 3
2 3

3
1

 
1
,
3

2
1

3
2



Define f : S3 S3 /H by for all S3 , f () = H. Then

f ( 0 ) = ( 0 )H = (H) (0 H) = f () f(0 )
for all , 0 S3 . Hence, f is a homomorphism. Also, Ker f = { S3 | H = H} = { S3 | H} = H.
In Theorem 5.1.9, we showed that if f is a homomorphism of a group into a group G1 , then Ker f is a
normal subgroup of G. In the following theorem, we show that every normal subgroup H of a group G induces
a homomorphism g of G onto the quotient group G/H such that Ker g = H. We note that in Example 5.1.11,
the conclusion did not depend on the nature of S3 . The conclusion was made by use of general arguments. This
also leads us to the following theorem.
Theorem 5.1.12 Let H be a normal subgroup of a group G. Define the function g from G onto the quotient
group G/H by g(a) = aH for all a G. Then g is a homomorphism of G onto G/H and Ker g = H. (The
homomorphism g is called the natural homomorphism of G onto G/H.)

100

5. Homomorphisms and Isomorphisms of Groups

Proof. From the definition of g, it follows that g is a function from G onto G/H. To show g is a homomorphism, let a, b G. Then g(ab) = (ab)H = (aH)(bH) = g(a)g(b). Hence, g is a homomorphism of G onto G/H.
Finally, we show that Ker g = H. Now a Ker g if and only if g(a) = eH if and only if aH = eH if and only if
e1 a H if and only if a H. Thus, Ker g = H.
We now define a particular type of homomorphism between groups in order to introduce the important idea
of groups being algebraically indistinguishable.
Definition 5.1.13 A homomorphism f of a group G into a group G1 is called an isomorphism of G onto
G1 if f is one-one and onto G1 . In this case, we write G ' G1 and say that G and G1 are isomorphic. An
isomorphism of a group G onto G is called an automorphism.
For a group G, Aut(G), denotes the set of all automorphisms of G.
In the following theorem, we collect some properties of isomorphisms, which will be useful in determining
whether given groups are isomorphic or not.
Theorem 5.1.14 Let f be an isomorphism of a group G onto a group G1 . Then
(i) f 1 : G1 G is an isomorphism.
(ii) G is commutative if and only if G1 is commutative.
(iii) For all a G, (a) = (f (a)).
(iv) G is a torsion group if and only if G1 is a torsion group.
(v) G is cyclic if and only if G1 is cyclic.
Proof. (i) Since f is one-one and onto G1 , f 1 is one-one and onto G. Now we only need to verify that f 1
is a homomorphism. Let u, v G1 . Then there exist a, b G such that f (a) = u and f (b) = v. This implies
that a = f 1 (u), b = f 1 (v), and uv = f (a)f (b) = f (ab). Thus, f 1 (uv) = ab = f 1 (u)f 1 (v) and so f 1 is a
homomorphism. Hence, f 1 is an isomorphism.
(ii) Suppose G is commutative. Let u, v G1 . Since f is onto G1 , there exist a, b G such that f (a) = u
and f(b) = v. Now
uv = f (a)f (b) = f (ab) = f (ba) = f (b)f (a) = vu.
Thus, G1 is commutative. Conversely, suppose G1 is commutative. Let a, b G. Now
f (ab) = f(a)f (b) = f (b)f (a) = f(ba).
Since f is one-one, we have ab = ba. This proves that G is commutative.
(iii) Let a G. By induction, it follows that for all positive integers n, f(an ) = (f(a))n . Since f is one-one,
for all b G, f (b) = e1 if and only if b = e. Hence, an = e if and only if (f (a))n = e1 . Thus, a is of finite order if
and only if f (a) is of finite order. Suppose (a) = m and (f (a)) = n. Since am = e, (f (a))m = e1 . By Theorem
2.1.46, n divides m. Also, (f (a))n = e1 implies that an = e. Hence, m divides n. Since m and n are both positive
integers and m divides n and n divides m, it follows that m = n.
(iv) This follows immediately by (iii).
(v) Suppose G is cyclic. Then G = hai for some a G. Since f (a) G1 , hf (a)i G1 . Let b G1 . Since f is
onto G1 , there exists c G such that f (c) = b. Now c = an for some n Z . Thus,
b = f (c) = f (an ) = (f (a))n hf (a)i .
Hence, G1 = hf (a)i and so G1 is cyclic. The converse follows since f 1 is an isomorphism.
In order to develop a feel for two groups being algebraically indistinguishable, let us consider two sets S and
S 0 such that there is a one-one function f of S onto S 0 . Then in a set-theoretic sense, S and S 0 are the same sets
under f. For instance, let A and B be subsets of S. Then f (A) and f (B) are corresponding subsets of S 0 .
Now f (A B) = f (A) f (B) and f (A B) = f (A) f (B); that is, union and intersection are preserved under f.
Other purely set-theoretic operations can be seen to be preserved under f also. Now suppose binary operations
and 0 are defined on S and S 0 , respectively, so that (S, ) and (S 0 , 0 ) are groups. Now even though S and S 0
are the same sets under f, they need not be the same as groups, i.e., f may not preserve operations. We have
seen that the requirement for f to preserve operations is that f (a b) = f (a) 0 f (b) for all a, b S.
We now consider examples of groups that are isomorphic and examples of groups that are not isomorphic.
Example 5.1.15 Let n be a positive integer. Define f from Zn into Z/ hni by for all [a] Zn , f ([a]) = a + hni .
Then [a] = [b] if and only if n | (a b) if and only if a b = nq for some q Z if and only if a b hni if and
only if a + hni = b + hni if and only if f([a]) = f ([b]). Therefore, we find that f is a one-one function. From the
definition of f, it follows that f maps Zn onto Z/ hni . Now f ([a] +n [b]) = f ([a + b]) = (a + b) + hni = (a + hni)+
(b + hni) = f ([a]) + f([b]). Thus, f is an isomorphism of Zn onto Z/ hni .

5.1. Homomorphisms of Groups

101

Example 5.1.16 Consider the sets G = {e, a, b, c} and G1 = {1, 1, i, i}. Define and on G and G1 ,
respectively, by means of the following operation tables.

e
a
b
c

1
1
i
i

e
e
a
b
c
1
1
1
i
i

a
a
e
c
b

b
b
c
e
a

1
1
1
i
i

i
i
i
1
1

c
c
b
a
e
i
i
i
1
1

Now G1 is a cyclic group generated by i. G is also a group. However, since aa = e, bb = e, and cc = e, no

element of G has order 4 and so G is not cyclic. Thus, G and G1 are not isomorphic.
Example 5.1.17 Let (R, +) be the group of real numbers under addition and (R+ , ) be the group of positive
real numbers under multiplication. Define f : RR+ by f (a) = ea for all a R. Clearly f is well defined. Let
a, b R. Then f (a + b) = ea+b = ea eb = f (a)f (b). Hence, f is a homomorphism. Suppose f (a) = f (b). Then
ea = eb and so loge ea = loge eb . This implies that a = b, whence f is one-one. Let b R+ . Then loge b R and
f (loge b) = eloge b = b. Thus, f is onto R+ . Consequently, f is an isomorphism of (R, +) onto (R+ , ).
Example 5.1.18 Consider the groups (Z, +) and (Q, +). By Worked-Out Exercise 1 (page 80), (Q, +) is not
cyclic. Since (Z, +) is cyclic and (Q, +) is not cyclic, (Z, +) is not isomorphic to (Q, +) by Theorem 5.1.14(v).
Example 5.1.19 The group (Q, +) is not isomorphic to (Q , ) since every nonidentity element of (Q, +) is of
infinite order while 1 is a nonidentity element of (Q , ) which is of finite order.
Let us now characterize finite and infinite cyclic groups.
Theorem 5.1.20 Every finite cyclic group of order n is isomorphic to (Zn , +n ) and every infinite cyclic group
is isomorphic to (Z, +).
Proof. Let (hai , ) be a cyclic group of order n. Let G = hai . Define the function f : G Zn by for all
ai G, f (ai ) = [i]. Now ai = aj if and only if aji = e if and only if n | (j i) if and only if [i] = [j] (Exercise
11, page 23) if and only if f (ai ) = f (aj ). Thus, f is a one-one function. Now
f(ai aj ) = f (ai+j ) = [i + j] = [i] +n [j] = f (ai ) +n f (aj ).
Since f is one-one and G and Zn are finite with same number of elements, f is onto Zn . Hence, G ' Zn .
Now let G = hai be an infinite cyclic group. Define the functionf : G Z by f (ai ) = i for all i Z. Since
i
a = aj if and only if aij = e if and only if i j = 0 (since a is of infinite order) if and only if i = j, we have
that f is a one-one function of G into Z. From the definition of f, f is onto Z. Now
f (ai aj ) = f (ai+j ) = i + j = f (ai ) + f (aj ).
Hence, G ' Z.
Corollary 5.1.21 Any two cyclic groups of the same order are isomorphic.
From the above corollary, it follows that there is only one (up to isomorphism) cyclic group having a prescribed
order.
In Example 5.1.16, we saw that there are at least two nonisomorphic groups of order 4. We now show that
these are exactly two nonisomorphic groups of order 4.
Let G be a group of order 4 which is not cyclic. (Example 5.1.16 shows that such a group exists.) Then no
element of G can have order 4, for if a G has order 4, then e, a, a2 , a3 would be distinct elements of G and thus
G would be cyclic, i.e., G = hai . This is contrary to the assumption that G is not cyclic. Let G = {e, a, b, c}.
Since the order of every element of G divides the order of G, a, b, and c have order 2. If ab = a, then b = e, a
contradiction. Thus, ab 6= a. Similarly, ab 6= b. Suppose ab = e, then a(ab) = ae. Therefore, b = a since a2 = e,
a contradiction. Thus, ab = c. Similarly, ba = c. Hence, ab = ba. By similar arguments, we have ac = b = ca

102

5. Homomorphisms and Isomorphisms of Groups

and bc = a = cb. Thus, we find that G is a commutative group and its operation table is given by the table in
Example 5.1.16. Consequently, there is essentially one group of order 4 which is not cyclic. This is the Klein
4-group. Since all cyclic groups of the same orders are isomorphic, we thus have exactly two nonisomorphic
groups of order 4, namely, the Klein 4-group and the cyclic group of order 4. We have thus proved the following
result.
Theorem 5.1.22 There are only two groups of order 4 (up to isomorphism), a cyclic group of order 4 and K4
(Klein 4-group).
Since every cyclic group is commutative and every group of prime order is cyclic, it follows that that if a
group is noncommutative, then it must have order at least 6. Indeed, the symmetric group S3 is noncommutative
and of order 6. Since all cyclic groups of the same order are isomorphic and since every group of prime order
is cyclic, there is exactly one group of order 1, 2, 3, 5 (up to isomorphism), respectively. We have seen that
there are two nonisomorphic groups of order 4. In the next theorem, we show that there are only two (up to
isomorphism) nonisomorphic groups of order 6.
Theorem 5.1.23 There are only two (up to isomorphism) groups of order 6.
Proof. The group Z6 is a cyclic group of order 6 and S3 is a noncommutative group of order 6. Note that
Z6 is not isomorphic to S3 . To show that there are only two (up to isomorphism) nonisomorphic groups of order
6, we will show that any group of order 6 is isomorphic to either Z6 or S3 .
Let G be a group of order 6. Since |G| is even, there exists a G, a 6= e such that a2 = e. If x2 = e for all
x G, then G is commutative and for any two distinct nonidentity elements a and b, {e, a, b, ab} is a subgroup
of G. Since |G| = 6, G has no subgroups of order 4. Hence, there exists b G such that b2 6= e, i.e., b 6= e
and (b) 6= 2. Since (b) | 6, (b) = 6 or 3. If (b) = 6, then G = hbi is a cyclic group of order 6 and G ' Z6 .
Suppose G is not cyclic. Then (b) = 3. Let H = {e, b, b2 }. Then H is a subgroup of G of index 2. Thus, H is a
normal subgroup of G. Clearly a
/ H. Now G = H aH and H aH = . Hence, G = {e, b, b2 , a, ab, ab2 }. Now
1
aba H since H is normal and b H. Therefore, aba1 = e or aba1 = b or aba1 = b2 . If aba1 = e, then
b = e, which is a contradiction. If aba1 = b, then ab = ba. Since (a) and (b) are relatively prime and ab = ba,
(ab) = (a) (b) = 6. Thus, G is cyclic, a contradiction. Hence, aba1 = b2 . Thus, G = ha, bi , where (a) = 2,
(b) = 3, and aba1 = b2 . It is now easy to see that G ' S3 .
We conclude this section by proving Cayleys theorem, which says that any group can be realized as a
permutation group.
Theorem 5.1.24 (Cayley) Any group G is isomorphic to some subgroup of the group (S(G), ) of all permutations of the set G.
Proof. Let a be an element of a group G. Define the function fa : G G by for all b G, fa (b) = ab. Then
b = c if and only if ab = ac if and only if fa (b) = fa (c). Thus, fa is a one-one function of G into G. For any
b G,
fa (a1 b) = a(a1 b) = b.
So we find that fa maps G onto G. Hence, fa is a permutation of G. This implies that fa S(G). Let F (G) =
{fa | a G}. Then F (G) is a subset of the set S(G) of all permutations on G Define g : G S(G) by for all
a G, g(a) = fa . Then a = b if and only if ac = bc for all c G if and only if fa (c) = fb (c) for all c G if and
only if fa = fb if and only if g(a) = g(b). This proves that g is a one-one function of G into F (G). Clearly g
maps G onto F (G). Now g(ab) = fab and g(a) g(b) = fa fb . Also, for all c G,
fab (c) = (ab)c = a(bc) = fa (bc) = fa (fb (c)) = (fa fb )(c).
Thus, fab = fa fb . Hence, g(ab) = g(a) g(b) and so g is a homomorphism.This implies that F (G) is a subgroup
and G is isomorphic to this subgroup.
Cayleys theorem is another example of a representation theorem. However, Cayley realized that the best
way of studying general problems in group theory was not necessarily by the use of permutations.

Worked-Out Exercises
Exercise 1 Let f : G G1 be an epimorphism of groups. If H is a normal subgroup of G, then show that f(H) is a
normal subgroup of G1 .

5.1. Homomorphisms of Groups

103

Solution: By Theorem 5.1.2, we find that f (H) is a subgroup of G1 . Let g1 G1 . Since f is onto G1 , there exists g G
such that f (g) = g1 . Let a g1 f (H)g11 = f (g)f (H)f (g)1 . Then a = f (g)f (h)f(g)1 = f (ghg 1 ) for
some h H. Since H is a normal subgroup of G, ghg 1 H and so a f(H). Thus, g1 f (H)g11 f (H).
Hence, f (H) is a normal subgroup of G1 .
Exercise 2 Let G and H be finite groups such that gcd(|G| , |H|) = 1. Show that the trivial homomorphism is the only
homomorphism from G into H.
Solution: Let f : G H be a homomorphism and let a G. We show that every element of G is mapped onto
the identity element of H, i.e., f (a) = eH for all a G, where eH denotes the identity element of H. Now
(a) | |G| and (f (a)) | |H| . Also, by Theorem 5.1.2, (f (a)) | (a). Hence, (f (a)) | |G| . Since |G| and
|H| are relatively prime, (f(a)) = 1, proving f (a) = eH . Thus, f is the trivial homomorphism.

Exercise 3 Show that the group (Q, +) is not isomorphic to (Q/Z, +).

Solution: In (Q, +), every nonzero element is of infinite order. Let pq + Z Q/Z, where p, q Z and q 6= 0. Then
q( pq +Z) = p+Z=Z. This shows that every element of Q/Z is of finite order. Hence, (Q, +) is not isomorphic
to (Q/Z, +).

Exercise 4 Show that R , the group of all nonzero real numbers under multiplication, is not isomorphic to C , the
group of all nonzero complex numbers under multiplication.
Solution: In the group C , i is an element of order 4. But R does not contain any element of order 4. Hence, by
Theorem 5.1.14, R is not isomorphic to C .
Exercise 5 Find all homomorphisms from Z6 into Z4 .
Solution:

Z6 = h[1]i . Let f : Z6 Z4 be a homomorphism. For any [a] Z6 , f ([a]) = af([1]) shows that f is
completely known if f ([1]) is known. Now (f ([1])) divides ([1]) and 4, i.e., (f ([1])) divides 6 and 4.
Hence, (f ([1])) = 1 or 2. Thus, f ([1]) = [0] or [2]. If f ([1]) = [0], then f is the trivial homomorphism
which maps every element to [0]. On the other hand, f ([1]) = [2] implies that f ([a]) = [2a] for all [a] Z6 .
Thus, f ([a] + [b]) = f ([a + b]) = [2(a + b)] = [2a + 2b] = [2a] + [2b] = f ([a]) + f ([b]), proving that the
mapping f : Z6 Z4 defined by f ([a]) = [2a] for all [a] Z6 is a homomorphism. Hence, there are two
homomorphisms from Z6 into Z4 .

Exercise 6 Let G be a finite commutative group. Let n Z be such that n and |G| are relatively prime. Show that
the function : G G defined by (a) = an for all a G is an isomorphism of G onto G.
Solution: Let a, b G. Now

(ab)

=
=
=

(ab)n
an bn
(since G is commutative)
(a)(b).

This implies that is a homomorphism. Let (a) = (b). Then an = bn and so (ab1 )n = e. Therefore,
(ab1 ) divides n. Since (ab1 ) divides |G| and n and |G| are relatively prime, (ab1 ) = 1. This implies
that ab1 = e, i.e., a = b, proving that is one-one. Since G is a finite group and is one-one, is onto
G. Hence, is an isomorphism of G onto G.
Exercise 7

(a) Let G be a group and f : G G be defined by f(a) = an for all a G, where n is a positive integer.
Suppose f is an isomorphism. Prove that an1 Z(G) for all a G.

(b) Let G be a group and f : G G defined by for all a G, f (a) = a3 be an isomorphism. Prove that
G is commutative.
Solution:

(a) Let a, b G. Then f (a1 ba) = (a1 ba)n = a1 bn a. Thus,

an bn an = f(a1 )f (b)f(a) = f (a1 ba) = a1 bn a.
Hence, a(n1) bn an1 = bn or (a(n1) ban1 )n = bn . Thus, f (a(n1) ban1 ) = f (b). Since f is
one-one, a(n1) ban1 = b. Hence, an1 b = ban1 , proving that an1 Z(G).

(b) By (a), a2 Z(G) for all a G. Let a, b G. Then f (ab) = (ab)3 = ab(ab)2 = a(ab)2 b = aababb =
a2 bab2 = ba2 b2 a = bb2 a2 a = b3 a3 = f(b)f(a) = f (ba). Hence, ab = ba since f is one-one. Thus, G is
commutative.

Exercises
1. Determine whether the indicated function f is a homomorphism from the first group into the second group.
If f is a homomorphism, determine its kernel.

104

5. Homomorphisms and Isomorphisms of Groups

(a) f (a) = a2 ; (R+ , ), (R+ , ) for all a R+ .

(b) f (a) = 2a ; (R, +), (R+ , ) for all a R.

(c) f (a) = |a| ; (R\{0}, ), (R+ , ) for all a R\{0}.

(d) f (a) = a + 1; (Z, +), (Z, +) for all a Z.

(e) f (a) = 2a; (Z, +), (Z, +) for all a Z.
(f) f ([a]) = [5a], (Z8 , +), [Z8 , +]

2. Find all homomorphisms from Z into Z. How many homomorphisms are onto?
3. Find all homomorphisms from Z onto Z6 .
4. Find all homomorphisms from Z8 into Z12 and from Z20 into Z10 .
5. Show that Q , the group of all nonzero rational numbers under multiplication, is not isomorphic to R ,
the group of all nonzero real numbers under multiplication.
6. Show that (Q, +) is not isomorphic to (R, +).
7. Show that (Z, +) is not isomorphic to (R, +).
8. Let G be a group. Define the function f : G G by for all a G, f (a) = a1 . Prove that f is a
homomorphism if and only if G is commutative.
9. Let G = {(a, b) | a, b R, b 6= 0}. Then (G, ) is a noncommutative group under the binary operation
(a, b)(c, d) = (a+bc, bd) for all (a, b), (c, d) G. Let H = {(a, b) G | a = 0} and K = {(a, b) G | b > 0}.
Show that H K ' (R+ , ), where (R+ , ) is the group of all positive real numbers under multiplication.

10. Let G = {a R | 1 < a < 1}. Show that (G, ) ' (R,+), where the binary operation on G is defined by
ab=
11.

a+b
1 + ab

for all a, b G.

(a) Let f be a homomorphism from a cyclic group of order 8 onto a cyclic group of order 4. Determine
Ker f.

(b) Let f be a homomorphism from a cyclic group of order 8 onto a cyclic group of order 2. Determine
Ker f.
12. Prove that a homomorphic image of a cyclic group is cyclic.
13. Show that S3 and Z6 are not isomorphic groups, but for every proper subgroup A of S3 there exists a
proper subgroup B of Z6 such that A ' B.

14. Let G, H, and K be groups. Suppose that the functions f : G H and g : H K are homomorphisms.
Prove that g f : G K is also a homomorphism.

15. Let G and H be groups. Define the function f : G H G by for all (a, b) G H, f ((a, b)) = a. Prove
that f is a homomorphism from G H onto G. Determine Ker f.
16. Let f : G H be an isomorphism of groups. Prove that f 1 : H G is also an isomorphism.
17. Let G, H, and K be groups. Prove that
(a) G H ' H G.

(b) If G ' H and H ' K, then G ' K.

(c) G (H K) ' (G H) K.

18. Let G and H be groups. Let f : G H be a homomorphism of G onto H. Show that if G = hSi for some
subset S of G, then H = hf (S)i .

19. Let f : G H be an isomorphism of groups. Show that for any integer k and for any g G, the sets
A = {a G | ak = g} and B = {b H | bk = f (g)} have the same number of elements.
20. Let G be a simple group and : Sn G be an epimorphism for some positive integer n. Prove that
G ' Sk for some k n.

21. Which of the following statements are true? Justify.

(a) A cyclic group with more than one element may be a homomorphic image of a noncyclic group.

5.2. Isomorphism and Correspondence Theorems

105

(b) There does not exist a nontrivial homomorphism from a group G of order 5 into a group H of order
4.
(c) The group (Z,+) is isomorphic to (Q,+).
(d) There exists a monomorphism from a group of order 20 into a group of order 70.
(e) There exists an epimorphism of (R,+) onto (Z,+).
(f) There does not exist any epimorphism of (Q,+) onto (Z,+).
(g) If f and g are two epimorphisms of a group G onto a group H such that Ker f = Ker g, then f = g.
(h) (Z Z, +) is a cyclic group.
(i) The group (Z, +) is a homomorphic image of (Q, +).

5.2

Isomorphism and Correspondence Theorems

In this section, we continue our study of isomorphisms. Our objective is to prove the fundamental theorem
of homomorphisms, the isomorphism theorems, and the correspondence theorem. These theorems show us the
relationship between homomorphisms and quotient groups.
Theorem 5.2.1 Let f be a homomorphism of a group G onto a group G1 , H be a normal subgroup of G such
that H Ker f, and g be the natural homomorphism of G onto G/H. Then there exists a unique homomorphism
h of G/H onto G1 such that f = h g. Furthermore, h is one-one if and only if H = Ker f .

G1

f(a)

g
h

G/H
Proof. Define h : G/H G1 by

f
h

aH
h(aH) = f (a)

for all aH G/H.

Now aH = bH implies b1 a H Ker f and so f (b1 a) = e1 or f (a) = f (b). Hence, h(aH) = h(bH) and
so h is well defined. Let a G. Then
(h g)(a) = h(g(a)) = h(aH) = f (a).
Therefore, h g = f. Since f maps G onto G1 , h must map G/H onto G1 . Now
h((aH)(bH)) = h((ab)H) = f (ab) = f(a)f (b) = h(aH)h(bH).
Hence, h is a homomorphism of G/H onto G1 satisfying f = h g. To prove the uniqueness part, let us assume
f = h0 g for some homomorphism h0 from G/H onto G1 . Then
h(aH) = f (a) = (h0 g)(a) = h0 (g(a)) = h0 (aH)
for all aH G/H and so h = h0 . Hence, h is the only homomorphism of G/H onto G1 such that f = h g.
Suppose h is one-one. Let a Ker f. Then f (a) = e1 and so h(aH) = e1 . Since h(eH) = e1 and h is one-one,
aH = eH. Thus, a H and so Ker f H. By hypothesis, H Ker f and so H = Ker f . Conversely, assume
H = Ker f. Suppose h(aH) = h(bH). Then f (a) = f (b) or f (b1 a) = e1 . Thus, b1 a Ker f = H and so
aH = bH, proving that h is one-one.
From Theorem 5.2.1, it follows that if H = Ker f, then h is an isomorphism and hence G/Ker f is isomorphic
to G1 , i.e., every homomorphism of a group G onto a group G1 induces an isomorphism of G/Ker f onto G1 . This
result plays a fundamental role in group theory. It is known as the fundamental theorem of homomorphisms
for groups. This result is also called the first isomorphism theorem for groups. Considering the importance of
this theorem, we state it in its general form and also give a direct proof of it.

106

5. Homomorphisms and Isomorphisms of Groups

Theorem 5.2.2 (First Isomorphism Theorem) Let f be a homomorphism of a group G into a group G1 .
Then f (G) is a subgroup of G1 and
G/Ker f ' f (G).
Proof. By Theorem 5.1.2, f (G) is a subgroup of G1 . Let H = Ker f. Define h : G/H f (G) by
h(aH) = f (a)
for all aH G/H. Now aH = bH if and only if b1 a H = Ker f if and only if f (b1 a) = e1 if and only
if f (b1 )f (a) = e1 if and only if f (a) = f (b). Thus, h is a one-one function. Let x f (G). Then x = f (b)
for some b G. Therefore, h(bH) = f (b) = x. This shows that h is onto f (G). Finally, h(aHbH) = h(abH) =
f (ab) = f (a)f (b) = h(aH)h(bH) for all aH, bH G/H, proving that h is a homomorphism. Consequently,
G/Ker f ' f(G).
In the following example we illustrate the first isomorphism theorem.
Example 5.2.3 Let f be the homomorphism of (Z, +) onto (Z3 , +3 ) defined by f(n) = [n] for all n Z. Let
g be the natural homomorphism of Z onto Z/ h6i . Now h6i is a normal subgroup of Z and h6i h3i = Ker f.
Thus, there exists a homomorphism h of Z/ h6i onto Z3 such that f = h g. The homomorphism h is defined by
h(n + h6i) = [n].

Z3

g
h
Z/<6>
0

[0]

g
h

0 + <6>

[3]

3 + <6>

2 + <6>
f

[4]

g
h

[5]

g
h

4 + <6>

[2]
h

1 + <6>
f

[1]

h
5 + <6>

Recall that a group G1 is called a homomorphic image of a group G if there exists a homomorphism of G
onto G1 .
From Theorem 5.2.1 and Corollary 5.2.2, we find that for each normal subgroup N of a group G, G/N is a
homomorphic image of G, and for each homomorphic image G1 , there exists a normal subgroup N of G such
that G/N ' G1 .
Example 5.2.4 The group S3 has (up to isomorphism) only three homomorphic images. This follows from the
fact that S3 has only three normal subgroups. The homomorphic images are S3 , Z1 , and Z2 since {e}, S3 , and
{e, (1 2 3), (1 3 2)} are the only normal subgroups of S3 and S3 ' S3 /{e}, Z1 ' S3 /S3 , and Z2 ' S3 /{e, (1 2
3), (1 3 2)}.
Theorem 5.2.5 Let G1 be a homomorphic image of a group G. Then the following assertions hold.
(i) If G is cyclic, then G1 is cyclic.
(ii) If G is commutative, then G1 is commutative.
(iii) If G1 contains an element of order n and |G| is finite, then G contains an element of order n.

5.2. Isomorphism and Correspondence Theorems

107

Proof. (i) Follows by Exercise 12 (page 104).

(ii) Follows by Theorem 5.1.2(v).
(iii) Let f : G G1 be an epimorphism and let a0 be an element of G1 of order n. If n = 1, then e is the
required element of G of order 1. Suppose n > 1. Since f is onto G1 , there exists a G such that f(a) = a0 . Now
(a) is finite and by Theorem 5.1.2(v), (a0 ) divides (a), i.e., n divides (a). Let t Z+ be such that (a) = nt.
Then t < (a). Hence, at 6= e. Now ant = e. Let b = at . Then bn = e and by Theorem 2.1.46,
(at ) =

nt
(a)
=
= n.
gcd(t, (a))
t

Note that the result in Theorem 5.2.5(iii) does not hold if |G| is not finite. For example, Z6 is a homomorphic
image of Z; Z6 contains an element of order 3, but Z has no element of order 3.
Theorem 5.2.6 (Second Isomorphism Theorem) Let H and K be subgroups of a group G with K normal
in G. Then
H/(H K) ' (HK)/K.
Proof. Define f : H (HK)/K by f (h) = hK for all h H. Now
f(h1 h2 ) = h1 h2 K = h1 Kh2 K = f (h1 )f (h2 )
for all h1 , h2 H, proving that f is a homomorphism. Let xK (HK)/K. Then x = hk for some h H and
k K. Thus,
xK = (hk)K = (hK)(kK) = hK = f (h).
This proves that f is onto (HK)/K and so f (H) = (HK)/K. Hence, by the first isomorphism theorem, it follows
that
H/Ker f ' (HK)/K.
To complete the proof, we show that Ker f = H K. Now
Ker f

=
=
=
=

{h H | f (h) = identity element of HK/K}

{h H | hK = K}
{h H | h K}
H K.

Consequently, H/H K ' (HK)/K.

We illustrate the second isomorphism theorem with the help of the following example.
Example 5.2.7 Consider the group (Z, +) and its subgroups H = h2i and K = h3i . Then H +K = h2i+h3i = Z
and H K = h6i . Theorem 5.2.6 says that
H/(H K) ' (H + K)/K,
i.e.,
h2i / h6i ' Z/ h3i .

This isomorphism is evident if we notice that h2i / h6i = {0 + h6i , 2 + h6i , 4 + h6i} while Z/ h3i = {0 + h3i ,
1 + h3i , 2 + h3i}. The mapping
h : h2i / h6i Z/ h3i
defined by h : 0 + h6i 0 + h3i , 2 + h6i 2 + h6i h3i , 4 + h6i 1 + h3i is the desired isomorphism.

Theorem 5.2.8 Let f be a homomorphism of a group G onto a group G1 , H be a normal subgroup of G such
that H Ker f, and g, g0 be the natural homomorphisms of G onto G/H and G1 onto G1 /f (H), respectively.
Then there exists a unique isomorphism h of G/H onto G1 /f (H) such that g 0 f = h g.

g
G/H

G1

G1/f(H)

108

5. Homomorphisms and Isomorphisms of Groups

Proof. If we show Ker g 0 f = H, then there exists a unique isomorphism h of G/H onto G1 /f (H) by
Theorem 5.2.1. Let a H. Then (g 0 f )(a) = g0 (f (a)) = the identity of G1 /f (H) since f (a) f (H) = Ker g 0 .
Thus, a Ker g0 f and hence H Ker g0 f. Let a Ker g0 f. Then g0 (f(a)) = the identity of G1 /f(H) and
so f (a) Ker g0 = f (H). Therefore, there exists b H such that f(b) = f (a) or f (ab1 ) = e1 . This implies that
ab1 Ker f H and so a = (ab1 )b H. Thus, Ker g0 f H. Hence, Ker g 0 f = H.
Corollary 5.2.9 (Third Isomorphism Theorem) Let H1 , H2 be normal subgroups of a group G such that
H1 H2 . Then
(G/H1 )/(H2 /H1 ) ' G/H2 .

G/H1

(G/H1)/(H2/H1)

G/H2

Proof. Make the following substitutions in Theorem 5.2.8: G/H1 for G1 , H2 for H, and (G/H1 )/(H2 /H1 )
for G1 /f (H), where in this case f is the natural homomorphism of G onto G/H1 . Note that f (H2 ) = H2 /H1 .
We illustrate the third isomorphism theorem with the help of the following example.
Example 5.2.10 Consider the group (Z, +) and the subgroups h6i and h3i of Z. Then
Z/ h3i = {0 + h3i , 1 + h3i , 2 + h3i}.
Z/ h6i

{0 + h6i , 1 + h6i , 2 + h6i , 3 + h6i , 4 + h6i , 5 + h6i}.

h3i / h6i = {0 + h6i , 3 + h6i}.

Now,

(Z/ h6i)/(h3i / h6i) = {

0,
1,
2},

where

It is now clear that

=
=
=

0 + h6i + (h3i / h6i)

1 + h6i + (h3i / h6i)
2 + h6i + (h3i / h6i).

Z/ h3i ' (Z/ h6i)/(h3i / h6i)

since both are cyclic groups of order 3 and of course, by Corollary 5.2.9.
We can at times determine the subgroups of a group G1 from a group G whose subgroups are known if there
is a homomorphism f of G onto G1 . For if such an f exists, the following result says that the subgroups of G1
can be determined from the subgroups of G which contain Ker f.
Theorem 5.2.11 (Correspondence Theorem) Let f be a homomorphism of a group G onto a group G1 .
Then f induces a one-one inclusion preserving correspondence between the subgroups of G containing Ker f and
the subgroups of G1 . In fact, if H and K are corresponding subgroups of G and G1 , respectively, then H is a
normal subgroup of G if and only if K is a normal subgroup of G1 .
Proof. Let
H = {H | H is a subgroup of G such that Ker f H}

and

K = {K | K is a subgroup of G1 }.

Define f : H K by for all H H, f (H) = {f (h) | h H}. Then f (H) K by Theorem 5.1.2. Hence, f
is a function since f is a function. Let K K. Denote the preimage, f 1 (K), of K in G by H. Let a Ker f.
Then f (a) = e1 K and so a f 1 (K) = H. Thus, Ker f H. Let a, b H. Then f (a), f (b) K and so
f (ab1 ) = f (a)f (b1 ) = f(a)f (b)1 K. Therefore, ab1 H and so H is a subgroup of G containing Ker f,
i.e., H H. Hence, f maps H onto K. Let H1 , H2 H. Suppose f (H1 ) = f (H2 ). Let h1 H1 . Then there

5.2. Isomorphism and Correspondence Theorems

109

1
exists h2 H2 such that f (h1 ) = f (h2 ). This implies that f (h1 h1
Ker f H2 . Hence,
2 ) = e1 and so h1 h2
h1 = (h1 h1
)h

H
.
Therefore,
H

H
.
Similarly,
H

H
.
Thus,
H
=
H
and
so
f is one-one. Clearly
2
2
1
2
2
1
1
2
2

H1 H2 if and only if f (H1 ) f (H2 ). In fact, since f is one-one, H1 H2 if and only if f (H1 ) f (H2 ).
Suppose H is a normal subgroup of G such that Ker f H. Let K = f (H). We show that K is a normal
subgroup of G. Let f (a) G1 and f (h) K. Now aha1 H since H is a normal subgroup of G and so
f (a)f(h)f (a)1 = f (aha1 ) K. Hence, K is a normal subgroup of G1 . Let J be a normal subgroup of G1 and
L H be such that f (L) = J. Let a G and h L. Then f (aha1 ) = f(a)f (h)f (a)1 J and so aha1 L.
This proves that L is a normal subgroup of G.

Corollary 5.2.12 Let N be a normal subgroup of a group G. Then every subgroup of G/N is of the form K/N,
where K is a subgroup of G that contains N. Also, K/N is a normal subgroup of G/N if and only if K is a
normal subgroup of G.
Proof. Let g : G G/N be the natural homomorphism. If a G, then g(a) = aN. From Theorem 5.2.11,
we find that this homomorphism induces a one-one mapping g between the subgroups of G which contain Ker
g = N and the subgroups of G/N. Let H be a subgroup of G/N. Then there exists a subgroup K of G such that
N K and H = g (K) = {g(a) | a K} = K/N. The last part follows from Theorem 5.2.11.
The following example illustrates the correspondence theorem.
Example 5.2.13 Let f be a homomorphism of (Z, +) onto (Z12 , +12 ) defined by f (n) = [n] for all n Z. Then
for H and K of Theorem 5.2.11,
H = {h12i , h6i , h4i , h3i , h2i , Z}

and

K = {h[0]i , h[6]i , h[4]i , h[3]i , h[2]i , Z12 }.

f : h12i h[0]i ,
f : h2i h[2]i ,
f : h4i h[4]i ,

f : h3i h[3]i ,
f : h6i h[6]i ,
f : Z Z12 .

The following diagram indicates the one-one inclusion preserving the correspondence property of f .

Z12

<2>

<3>

<[2]>

<[3]>

<4>

<6>

<[4]>

<[6]>

<12>

<[0]>

Now h[9]i = {n[9] | n Z} {n[3] | n Z} = h[3]i . Also, [3] = [27] = 3[9] h[9]i . Therefore, h[3]i h[9]i .
Hence, h[3]i = h[9]i . Thus, the subgroup h9i of Z gets mapped to the subgroup h[3]i of Z12 by f. However, this
does not contradict Theorem 5.2.11 since h9i 6 h12i .
In the remainder of this section, we consider all isomorphisms of a group G onto itself. Recall that Aut(G)
is the set of all automorphisms of G.
Theorem 5.2.14 Let G be a group. Then (Aut(G), ) is a group, where denotes the composition of functions.
Proof. Since iG Aut(G), Aut(G) 6= . Let f, g Aut(G). Then f g is an automorphism by Exercise 14
(page 104) and Theorem 1.4.11. Hence, f g Aut(G). Clearly iG is the identity of Aut(G) and f 1 is the
inverse of f. Also, is associative by Theorem 1.4.13. Consequently, (Aut(G), ) is a group.
Theorem 5.2.15 Let G be a group and a G. Define a : G G by a (b) = aba1 for all b G. Then
(i) a Aut(G),
(ii) a b = ab for all a, b G,
(iii) (a )1 = a1 ,
(iv) for all Aut(G), a 1 = (a) .

110

5. Homomorphisms and Isomorphisms of Groups

Proof. (i) Let c, d G. Suppose c = d. Then aca1 = ada1 or a (c) = a (d). Therefore, a is well
defined. Now a (cd) = a(cd)a1 = (aca1 )(ada1 ) = a (c)a (d). This shows that a is a homomorphism. Also,
c = a (a1 ca), proving that a is onto G. Suppose a (c) = a (d). Then aca1 = ada1 and so c = d. Thus, a
is one-one. Consequently, a Aut(G).
(ii) Let a, b G. Then (a b )(c) = a (b (c)) = a (bcb1 ) = a(bcb1 )a1 = (ab)c(ab)1 = ab (c) for all
c G. Hence, a b = ab .
(iii) Note that a a1 = aa1 = e = iG and a1 a = a1 a = e = iG . Thus, (a )1 = a1 .
(iv) Let Aut(G). Now ( a 1 )(b) = (a (1 (b))) = (a1 (b)a1 ) = (a)(1 (b))(a1 ) =
(a)b((a))1 = (a) (b) for all b G. Hence, a 1 = (a) .
The automorphism a of Theorem 5.2.15 is called an inner automorphism of G. We denote by Inn(G) the
set of all inner automorphisms of G.
Theorem 5.2.16 Let G be a group. Then Inn(G) is a normal subgroup of Aut(G).
Proof. Since iG = e Inn(G), Inn(G) 6= . By Theorem 5.2.15(i), Inn(G) Aut(G). Let a , b Inn(G).
Then a 1
= a b1 = ab1 Inn(G). Hence, Inn(G) is a subgroup of Aut(G) by Theorem 4.1.6. Let
b
Aut(G). Then by Theorem 5.2.15(iv), a 1 = (a) Inn(G). Hence, Inn(G) is a normal subgroup of
Aut(G).
Theorem 5.2.17 Let G be a group and H be a subgroup of G. Then
N(H)
' a subgroup of Aut(H),
C(H)
where N(H) = {x G | xHx1 = H} is the normalizer of H and C(H) = {x G | xhx1 = h for all h H} is
the centralizer of H.
Proof. Define f : N(H) Aut(H) by for all a N(H),
f (a) = a |H .
Then f is well defined. Let a1 , a2 N(H).
a homomorphism. Now
Ker f =
=
=
=
=
=

Then f(a1 a2 ) = a1 a2 |H = a1 |H a2 |H = f (a1 ) f (a2 ). Thus, f is

{a G | f (a) = iH }
{a G | a = iH }
{a G | a (b) = iH (b) for all b H}
{a G | aba1 = b for all b H}
{a G | ab = ba for all b H}
C(H).

Thus, by the first isomorphism theorem, we have the desired result.

Corollary 5.2.18 Let G be a group. Then
G
' Inn(G).
Z(G)
Proof. Let H = G in Theorem 5.2.17. Then we have N(G) = G and C(G) = Z(G).

Worked-Out Exercises
Exercise 1 Find all homomorphic images of the additive group Z.

Solution: Let H be a homomorphic image of (Z, +). There exists a homomorphism f of Z onto H. By the first
isomorphism theorem, Z/Ker f ' H. Since Ker f is a subgroup of Z, Ker f = nZ for some integer n 0.
Hence, H ' Z/nZ for some integer n 0. On the other hand, for any n 0, nZ is a subgroup of Z and
since Z is commutative, nZ is a normal subgroup of Z. There exists a natural homomorphism f from Z
onto Z/nZ given by f (m) = m + nZ for all m Z. This shows that Z/nZ is a homomorphic image of Z for
all n 0. Consequently, the homomorphic images of Z are the groups (up to isomorphism) Z/nZ, n 0.
Now for n = 0, Z/nZ ' Z and for n > 0, Z/nZ ' Zn (Exercise 2, page 111). Therefore, we conclude that
the homomorphic images of Z are the cyclic groups Z and Zn , n > 0.

Exercise 2 If there exists an epimorphism of a finite group G onto the group Z8 , show that G has normal subgroups
of index 4 and 2.

5.2. Isomorphism and Correspondence Theorems

111

Solution: Let f : G Z8 be an epimorphism. Then by the first isomorphism theorem, G/Ker f ' Z8 . Hence, G/Ker
f is a cyclic group of order 8. Thus, G/Ker f has a normal subgroup H1 of order 4 and a normal subgroup
H2 of order 2. By the correspondence theorem, there exist normal subgroups N1 and N2 of G such that
Ker f N1 , Ker f N2 , N1 /Ker f = H1 , and N2 /Ker f = H2 . Thus,
8 = |G/Ker f | = [G : Ker f ] = [G : N1 ][N1 : Ker f ] = [G : N1 ]4.
This implies that [G : N1 ] = 2. Similarly, [G : N2 ] = 4.
Exercise 3 Show that 4Z/12Z ' Z3 .

Solution: Define f : 4ZZ3 by f (4n) = [n] for all 4n 4Z. One can show that f is an epimorphism. Then from
the first isomorphism theorem, 4Z/Ker f ' Z3 . Now Ker f = {4n 4Z | f (4n) = [0]} = {4n 4Z | [n] =
[0]} = 12Z.

Exercise 4 Let G be a finite group and f be an automorphism of G such that for all a G, f (a) = a if and only if
a = e. Show that for all g G, there exists a G such that g = a1 f (a).

1
Solution: Let G = {a1 , a2 , . . . , an }. Let S = {a1
1 f (a1 ), . . . , an f (an )}. Then S G. Next, we show that all
1
1
elements of S are distinct. Now a1
f
(a
)
=
a
f
(a
)
= ai a1
if and only if
i
j if and only if f (ai )f(aj )
i
j
j
1
1
1
f (ai aj ) = ai aj if and only if ai aj = e if and only if ai = aj . This shows that all elements of S are
distinct and so |S| = n. Thus, S = G. Let g G. Then g S. Hence, g = a1 f (a) for some a G.

Exercise 5 Let G be a finite group and f be an automorphism of G such that for all a G, f (a) = a if and only if
a = e. Suppose that f 2 = iG , where iG denotes the identity map. Prove that G is commutative.

Solution: Let g G. By Worked-Out Exercise 4, g = a1 f (a) for some a G. Then g = iG (g) = f 2 (a1 f (a)) =
f (f(a1 f(a))) = f (f(a1 )f 2 (a)) = f (f (a)1 a) = f(g 1 ). This implies that f (g) = g1 for all g G. Let
a, b G. Then (ab)1 = f (ab) = f (a)f (b) = a1 b1 = (ba)1 and so ab = ba. Hence, G is commutative.

Exercise 6 Let H be a subgroup of index 2 in a finite group G. If the order of H is odd and every element of G\H is
of order 2, prove that H is commutative.

Solution: Since [G : H] = 2, H is a normal subgroup of G. Now G = H Hg, where g

/ H. Then (g) = 2. Define
f : G G by for all a G, f (a) = gag 1 . Then f is an automorphism of G. Now f 2 (a) = f (f (a)) =
f (gag1 ) = g(gag 1 )g1 = g 2 ag 2 = a since g 2 = e. Hence, f 2 = iG . Since H is a normal subgroup of G,
f (h) = aha1 H for all h H. Thus, f is also an automorphism of H. Let h H. Suppose f (h) = h.
Then ghg1 = h or gh = hg. Since gh
/ H, (gh) = 2. Therefore, h2 = g 2 h2 = (gh)2 = e. Since the order
2
of H is odd, h = e implies that h = e. Hence, f (h) = h if and only h = e. Thus, f is an automorphism
of H such that f 2 = iG and f (h) = h if and only if h = e. By Worked-Out Exercise 5, H is commutative.
Exercise 7 Show that Aut(Zn ) ' Un .

Solution: Define :Aut(Zn ) Un by (f ) = f([1]) for all f Aut(Zn ). Now mf ([1]) = f ([m]). Hence, f ([m]) = [0]
if and only if m is divisible by n. Thus, (f ([1])) = n. This implies that f ([1]) Un and so is well
defined. Let f, g Aut(Zn ). Then (f g) = (f g)([1]) = f (g([1])). Suppose g([1]) = [k]. Then (f g) =
f ([k]) = kf ([1]) = k[1]f ([1]) = [k]f ([1]) = f ([1])g([1]) = (f)(g). Hence, is a homomorphism. Now
Ker

=
=
=

{f Aut(Zn ) | (f ) = [1]}
{f Aut(Zn ) | f ([1]) = [1]}
{f Aut(Zn ) | f is the identity map}.

Hence, is a monomorphism. Finally, we show that is onto Un . Let [t] Un . Then t and n are relatively
prime. Define f : Zn Zn by f ([m]) = [mt] for all [m] Zn . Let [r], [s] Zn . Suppose [r] = [s]. Then
r s = nq for some q Z. Thus, rt st = nqt. Hence, [rt] = [st], proving that f is well defined. Clearly
f is a homomorphism. Suppose f ([r]) = f ([s]). Then [rt] = [st] and so n divides rt st = (r s)t. Since t
and n are relatively prime, n divides r s. Therefore, [r] = [s]. This implies that f is one-one. Since Zn
is finite therefore we find that f is onto. Hence, f Aut(Zn ). Now (f ) = f ([1]) = [t] shows that is
onto Un . Thus, is an isomorphism. Consequently, Aut(Zn ) ' Un .

Exercises
1. Let R be the multiplicative group of all nonzero real numbers and T = {1, 1}. Then T is a subgroup
of R . Prove that the quotient group R /T is isomorphic to the multiplicative group R+ of positive real
numbers.
2. For any positive integer n, prove that Z/nZ'Zn .

112

5. Homomorphisms and Isomorphisms of Groups

3. Show that 8Z/56Z'Z7 and 4Z/8Z ' Z2

4. For any two positive integers m and n prove that mZ/mnZ'Zn .

5. Let G be a group and A and B be normal subgroups of G such that A ' B. Show by an example that
G/A 6' G/B.

6. Let G be the group of symmetries of the square and K4 the Klein 4-group.
Show that the mapping f : G K4 defines a homomorphism of G onto K4 , where f (r180 ) = f (r360 ) = e,
f (r90 ) = f (r270 ) = a, f(h) = f(v) = b, f (d1 ) = f (d2 ) = c.

7. In Exercise 6, exhibit the one-one inclusion preserving correspondence between the subgroups of G containing Z(G) and the subgroups of K4 .
8. Let G and K4 be as in Exercise 6. Let g be the natural homomorphism of G onto G/Z(G), where Z(G) is
the center of G. Prove that Z(G) = Ker f and exhibit the isomorphism h of G/Z(G) onto K4 such that
f = h g.

9. Show that Z8 is not a homomorphic image of Z15 .

10. Show that Z9 is not a homomorphic image of Z3 Z3 .

11. Show that if there exists an epimorphism from a finite group G onto the group Z15 , then G has normal
subgroups of indices 5 and 3, respectively.

12. Partition the following collection of groups into subcollections of groups such that any two groups in the
same subcollection are isomorphic.
(i) (Z, +), (ii) (Z6 , +), (iii) (Z2 , +), (iv) S2 , (v) S6 , (vi) (17Z, +), (vii) (3Z, +), (vii) (Q, +), (ix) (R, +),
(x)(R , ), (xi) (R+ , ), (xii) (Q , ), (xiii) (C , ), (xiv) (hi , ), where R denotes the set of nonzero real
numbers, Q denotes the set of nonzero rational numbers, C denotes the set of nonzero real numbers, R+
denotes the set of positive real numbers, and (hi , ) is the cyclic subgroup of (R+ , ) generated by .

13. Show that

(a) Aut(Z5 ) ' Z4 .

(b) Aut(Z8 ) ' Klein 4-group.

(c) Aut(Z2 Z2 ) ' S3

(d) Aut(Z) ' Z2

(e) Aut(Q, +) contains infinite number of elements.

14. Find all automorphisms of the group Z6 .

15. Show that |Aut(Zp )| = p 1, where p is a prime.

16. Prove that Inn(S3 ) ' S3 ' Aut(S3 ).

17. Determine Aut(S4 ).

18. Let G be a cyclic group of order n and be the Euler -function. Prove that |Aut(G)| = (n).
19. Let G be a group such that Z(G) = {e}. Prove that Z(Aut(G)) = {e}.

20. Let G be a group and H be a subgroup of G. H is called a characteristic subgroup of G if f (H) H for
all f Aut(G).
(a) Show that every characteristic subgroup of G is a normal subgroup of G.
(b) Give an example of a group G and a subgroup H such that H is a normal subgroup of G, but H is
not a characteristic subgroup of G.
(c) Show that Z(G) is a characteristic subgroup of G.
(d) Let H and K be characteristic subgroups of G. Show that HK and H K are characteristic subgroups
of G.
(e) Let H and K be subgroups of G such that H K. Show that if K is a normal subgroup of G and
H is a characteristic subgroup of G, then H is a normal subgroup of G.
(f) Let H and K be subgroups of G such that H K. Show that if H is a characteristic subgroup of K
and K is a characteristic subgroup of G, then H is a characteristic subgroup of G.
(g) Suppose G is cyclic. Show that every subgroup of G is a characteristic subgroup of G.

5.3. The Groups D4 and Q8

113

21. Show that the only characteristic subgroups of (Q, +) are {0} and Q.
22. Which of the following statements are true? Justify.
(a) Any epimorphism of Z onto Z is an isomorphism.
(b) Any epimorphism of a group G onto G is an isomorphism.
(c) The quotient group 4Z/64Z has five subgroups.
(d) Z5 has five homomorphic images.
(e) 2Z/6Z is a subgroup of Z/6Z.
(f) There exist four subgroups of Z which contain 10Z as a subgroup.
(g) Let G and H be two groups, A be a normal subgroup of G, and B be a normal subgroup of H. If
G ' H and A ' B, then G/A ' H/B.

5.3

The Groups D4 and Q8

In Section 5.1, we saw that there are two types of groups of order 4 and two types of groups of order 6. In this
section, we wish to classify all noncommutative groups of order 8. We will consider finite commutative groups
in Chapter 9. First we introduce two groups D4 and Q8 and study these groups in detail. The study of these
groups will eventually lead us to the classification of noncommutative groups of order 8.
Definition 5.3.1 A group G is called a dihedral group of degree 4 if G is generated by two elements a and b
satisfying the relations
(a) = 4, (b) = 2, and ba = a3 b.
Example 5.3.2 Let G be the subgroup of GL(2,R)(Example 2.1.10) generated by the matrices




0 1
0 1
A=
and B =
.
1 0
1 0
Then (A) = 4 and (B) = 2. Now
BA =

0
1

1
0

A3 B =

0
1

1
0

and



0
1



0
1

1
0

1
0

0
1

1
0

1
0

0
1

Thus, BA = A3 B. Hence, G is a dihedral group of degree 4.

Example 5.3.3 Consider S4 . Let G be the subgroup of S4 such that G is generated by the permutations
a = (1 2 3 4) and b = (2 4).
Then a2 = (1 3) (2 4), a3 = (1 4 3 2), a4 = e, b2 = e, and b a = (1 4) (2 3) = a3 b. Hence, (a) = 4,
(b) = 2, and b a = a3 b. Thus, G is a dihedral group of degree 4.
The following theorem reveals some interesting properties of D4 . These properties are similar to the properties
listed in Example 4.1.21 for D3 .
Theorem 5.3.4 Let G be a dihedral group of degree 4 generated by the elements a and b such that
(a) = 4, (b) = 2, and ba = a3 b.
Then the following assertions hold.
(i) Every element of G is of the form ai bj , 0 i < 4, 0 j < 2.
(ii) G has exactly eight elements, i.e., |G| = 8.
(iii) G is a noncommutative group.

114

5. Homomorphisms and Isomorphisms of Groups

Proof. (i) Since G = ha, bi ,

G = {ai1 bj1 ai2 bj2 ain bjn | it , jt Z, 1 t n, n N}.

Since a1 = a3 and b1 = b,every element of G can be expressed in the form ai1 bj1 ai2 bj2 ain bjn where
it 0 and jt 0. Again since ba = a3 b it follows that every element of G is of the form an bm , where n, m are
nonnegative integers. Now a4 = e, b2 = e. These imply that every element of G is of the form ai bj , 0 i < 4,
0 j < 2.
(ii) By (i), every element of G is of the form ai bj , 0 i < 4, 0 j < 2. Thus, |G| 8. Since (a) = 4, it
follows that e, a, a2 , a3 are distinct elements of G. Then b, ab, a2 b, a3 b are also distinct elements of G. Also, since
a1 = a3 , b1 = b, and a 6= b 6= e,
{e, a, a2 , a3 } {b, ab, a2 b, a3 b} = .

Thus, G = {e, a, a2 , a3 , b, ab, a2 b, a3 b}. Hence, G has eight elements.

(iii) Suppose ab = ba. Then ab = a3 b. This implies that a2 = e, which is a contradiction. Hence, ab 6= ba,
proving that G is noncommutative.
It is easy to see that any two dihedral groups of degree 4 are isomorphic. Hence, there exists only one dihedral
group (up to isomorphism) of degree 4. We denote a dihedral group of degree 4 by D4 .
We now describe all subgroups of D4 .
In D4 ,
(a) = 4, (a2 ) = 2, (a3 ) = 4, (b) = 2,
(ab)2 = abab = aa3 bb = e,

(a2 b)2 = a2 ba2 b = a2 (a3 b)ab = abab = e,

(a3 b)2 = a3 ba3 b = a3 (a3 b)a2 b = a2 ba2 b = e.
From this, it follows that H1 = {e, a2 }, H2 = {e, b}, H3 = {e, ab}, H4 = {e, a2 b}, and H5 = {e, a3 b} are subgroups
of order 2. By Lagranges theorem, D4 has no subgroups of order 3, 5, 6, or 7. Now
T1 = {e, a, a2 , a3 }

T2 = {e, a2 , b, a2 b}

T3 = {e, ab, a2 , a3 b}

are subgroups of order 4. We ask the reader to verify that {e}, H1 , H2 , H3 , H4 , H5 , T1 , T2 , T3 , and D4 are the
only subgroups of D4 .
It is interesting to note in D4 that H5 is a normal subgroup of T3 and T3 is a normal subgroup of D4 , but H5
is not a normal subgroup of D4 . We also note that every nontrivial subgroup of D4 is of order 2 or 4. Therefore,
every nontrivial subgroup of D4 is commutative. However, since T2 is a nontrivial subgroup of D4 and T2 is not
cyclic, it follows that not every nontrivial subgroup of D4 is cyclic. Finally, we also note that D4 is isomorphic
to Sym, the group of symmetries of a square (page 47). This follows from Theorem 5.3.4 and the group table of
the group of symmetries of the square given on page 49.
Next, we consider Q8 .
Definition 5.3.5 A group G is called a quaternion group if G is generated by two elements a, b satisfying the
relation
(a) = 4, a2 = b2 , and ba = a3 b.
Example 5.3.6 Let T be the group of all 2 2 invertible matrices over C under usual matrix multiplication.
Let G be the subgroup of T generated by the matrices




0 1
0 i
A=
and B =
.
1 0
i 0
Then (A) = 4 and

A =
Now
BA =
and
A3 B =

1
0

0
1

 
0

0
i

i
0

0
1

1
0

 
0

Thus, BA = A3 B. Hence, G is a quaternion group.


1
0
i
0

= B2 .


i
0

0
i

i
0

0
i

5.3. The Groups D4 and Q8

115

We leave the proof of the following theorem, which is similar to the proof of Theorem 5.3.4, as an exercise.
Theorem 5.3.7 Let G be a quaternion group generated by the elements a and b such that
(a) = 4, a2 = b2 , and ba = a3 b.
Then the following assertions hold.
(i) Every element of G is of the form ai bj , 0 i < 4, 0 j < 2.
(ii) G has exactly eight elements, i.e., |G| = 8.
(iii) G is a noncommutative group.
It is easy to see that any two quaternion groups are isomorphic. Hence, there exists only one quaternion
group (up to isomorphism) and we denote it by Q8 .
Next, we determine all subgroups of Q8 .
Let Q8 = ha, bi , where (a) = 4, a2 = b2 , and ba = a3 b. Then
Q8 = {e, a, a2 , a3 , b, ab, a2 b, a3 b}.
In Q8 ,
Now

(a) = 4, (a2 ) = 2, (a3 ) = 4, (b) = 4.

(ab)2 = abab = aa3 bb = b2 = a2 .

Thus, (ab) = 4. Also,

(a2 b)2 = a2 ba2 b = a2 (a3 b)ab = a5 bab = abab

and
(a3 b)2 = a3 ba3 b = a3 (a3 b)a2 b = a2 ba2 b.
Hence, (a2 b) = 4 and (a3 b) = 4. It now follows that H0 = {e}, H1 = {e, a2 }, H2 = {e, a, a2 , a3 }, H3 =
{e, ab, a2 , a3 b}, and H4 = {e, b, a2 , a2 b} are subgroups of Q8 . We ask the reader to verify that H0 , H1 , H2 , H3 ,
H4 , and Q8 are the only subgroups of Q8 .
Since [Q8 : H2 ] = [Q8 : H3 ] = [Q8 : H4 ] = 2, H2 , H3 , and H4 are normal subgroups of Q8 . Now ba2 b1 =
baab1 = a3 bab1 = a3 a3 bb1 = a2 H1 . Since Q8 = ha, bi , H1 is a normal subgroup of Q8 . Thus, every
subgroup of Q8 is a normal subgroup of G. It is also interesting to observe that all proper subgroups of Q8 are
cyclic.
Theorem 5.3.8 D4 6' Q8 .
Proof. We note from the above discussion that Q8 contains six elements of order 4 while D4 contains only
two elements of order 4. Hence, D4 6' Q8 .
The next theorem classifies all noncommutative groups of order 8.
Theorem 5.3.9 There exist (up to isomorphism) only two noncommutative nonisomorphic groups of order 8.
Proof. Let G be a noncommutative group of order 8. Since |G| is even, there exists an element u G, u 6= e,
such that u2 = e. If x2 = e for all x G, then G is commutative, a contradiction. Thus, there exists a G such
that a2 6= e. Since (a) | 8, (a) = 4 or 8. If (a) = 8, then G is cyclic and hence commutative, a contradiction.
Thus, (a) = 4. Let H = {e, a, a2 , a3 }. Then H is a subgroup of G of index 2 and so H is a normal subgroup of
G. Let b G be such that b
/ H. Then G = H Hb and H Hb = . This implies that
G = {e, a, a2 , a3 , b, ab, a2 b, a3 b} = ha, bi .
Now bab1 H. If bab1 = e, then a = e, a contradiction. Thus, bab1 6= e. If bab1 = a, then ab = ba and
hence G is commutative, a contradiction. If bab1 = a2 , then ba2 b1 = (bab1 )2 = a4 = e and so a2 = e, a
/ H, (Hb) = 2. Hence, b2 H.
contradiction. Therefore, bab1 = a3 and so ba = a3 b. Since |G/H| = 2 and b
2
3
If b = a or a , then (b) = 8 and so G is commutative, a contradiction. Therefore, either b2 = e or b2 = a2 . It
now follows that if G is a noncommutative group of order 8, then either
G = ha, bi such that (a) = 4, (b) = 2 ,and ba = a3 b
or
G = ha, bi such that (a) = 4, b2 = a2 , and ba = a3 b.

In the first case, G ' D4 and in the second case, G ' Q8 .

116

5. Homomorphisms and Isomorphisms of Groups

Worked-Out Exercises
Exercise 1 Find Z(D4 ).

Solution: It is known that Z(D4 ) is a normal subgroup of D4 . Now D4 has six normal subgroups: D4 , {e}, H1 =
{e, a2 }, T1 = {e, a, a2 , a3 }, T2 = {e, a2 , b, a2 b}, T3 = {e, ab, a2 , a3 b}. Since ab 6= ba, D4 , T1 , and T2 cannot
be Z(D4 ). If (ab)b = b(ab), then a = (ba)b = a3 b2 = a3 and so a2 = e, a contradiction. Hence, T3 6= Z(D4 ).
Now a2 b = a6 b = a3 (a3 b) = a3 (ba) = (ba)a = ba2 . Hence, a2 Z(D4 ). Thus, Z(D4 ) = {e, a2 } = H1 .

Exercise 2 Prove that D4 /Z(D4 ) is isomorphic to K4 and hence find Inn(D4 ).

Solution: By Corollary 5.2.18, Inn(D4 ) ' D4 /Z(D4 ). Now D4 /Z(D4 ) is a group of order 4 and
D4 /Z(D4 ) = {eZ(D4 ), aZ(D4 ), bZ(D4 ), abZ(D4 )}.
Since a2 Z(D4 ), b2 = e, and (ab)2 = e, we find that each nonidentity element of D4 /Z(D4 ) is of order 2.
Hence, D4 /Z(D4 ) ' K4 , the Klein 4-group.

Exercises
1. In D4 , find subgroups H and K such that K is a normal subgroup of H and H is a normal subgroup of
D4 , but K is not a normal subgroup of D4 .
2. Show that Q8 is the union of three subgroups each of index 2.
3. Find all homomorphic images of D4 .
4. Find all homomorphic images of Q8 .

Group Actions
As previously mentioned, the theory of groups first dealt with permutation groups. Later the notion of an
abstract group was introduced in order to examine properties of permutation groups which did not refer to the
set on which the permutations acted. However, one is primarily interested in permutation groups in geometry.
Also, permutation groups are used in counting techniques that are important in finite group theory. An example
of this can be seen in the proof of Lagranges theorem. We extend the notion of a permutation on a set to a group
action on a set. We use the notion of a group action on a set to determine, via counting techniques, important
properties of finite groups.
Let G be a group and S a nonempty set. A (left) action of G on S is a function : G S S (usually
denoted by (g, x) g x) such that
(i) (g1 g2 ) x = g1 (g2 x), and
(ii) e x = x, where e is the identity of G
for all x S, g1 , g2 G.
Note: If no confusion arises, we write gx for g x.
If there is a left action of G on S, we say that G acts on S on the left and S is a G-set.
Example 5.3.10 Let G be a permutation group on a set S. Define a left action of G on S by
x = (x)
for all G, x S. Let x S. Now ex = e(x) = x, where e is the identity permutation on S. Let 1 , 2 G.
Then (1 2 ) x = (1 2 )(x) = 1 (2 (x)) = 1 ( 2 (x)) = 1 (2 x). Hence, S is a G-set.
Example 5.3.11 Let G be a group and H be a normal subgroup of G. Define a left action of G on H by
(g, h) ghg1
for all g G, h H. We denote this by g h = ghg1 . Let h H. Now e h = ehe1 = ehe = h. Let g1 , g2 G.
Then (g1 g2 ) h = (g1 g2 )h(g1 g2 )1 = (g1 g2 )h(g21 g11 ) = g1 (g2 hg21 )g11 = g1 (g2 h)g11 = g1 (g2 h). Hence, H
is a G-set.
Theorem 5.3.12 Let S be a G-set, where G is a group and S is a nonempty set. Define a relation on S by
for all a, b S,
a b if and only if ga = b for some g G.

Then is an equivalence relation on S.

5.3. The Groups D4 and Q8

117

Proof. Since for all a S, ea = a, a a for all a S. Thus, is reflexive. Let a, b, c S. Suppose a b.
Then ga = b for some g G, which implies that g 1 b = g 1 (ga) = (g1 g)a = ea = a. Hence, b a and so
is symmetric. Now suppose a b and b c. Then there exist g1 , g2 G such that g1 a = b and g2 b = c. Thus,
(g2 g1 )a = g2 (g1 a) = g2 b = c and so a c. Hence, is transitive. Consequently, is an equivalence relation.
Definition 5.3.13 Let S be a G-set, where G is a group and S is a nonempty set. The equivalence classes
determined by the equivalence relation of Theorem 5.3.12 are called the orbits of G on S.
For a S, the orbit containing a is denoted by [a].
Lemma 5.3.14 Let G be a group and S be a G-set. For all a S, the subset
Ga = {g G | ga = a}
is a subgroup of G.
Proof. Let a S. Since ea = a, e Ga and so Ga 6= . Let g, h Ga . Then ga = a and ha = a. This implies
that (gh)a = g(ha) = ga = a and so gh Ga . Now h1 a = h1 (ha) = (h1 h)a = ea = a. Thus, h1 Ga .
Hence, Ga is a subgroup of G.
The subgroup Ga of Lemma 5.3.14 is called the stabilizer of a or the isotropy group of a.
Lemma 5.3.15 Let G be a group and S be a G-set. For all a S,
[G : Ga ] = |[a]| .
Proof. Let a S. Let L be the set of all left cosets of Ga in G. Now
[a] = {b S | a b} = {b S | ga = b for some g G} = {ga | g G}.
We now show that there exists a one-one function from L onto [a]. Define
f : L [a]
by
f (gGa ) = ga
for all gGa L. Let g1 , g2 G. Then g1 Ga = g2 Ga if and only if g21 g1 Ga if and only if g21 (g1 a) = (g21 g1 )a =
a if and only if g1 a = g2 a. Thus, f is a one-one function from L into [a]. Let b [a]. Then there exists g G
such that ga = b. Thus, f(gGa ) = ga = b. This implies that f is onto [a]. Consequently, [G : Ga ] = |L| = |[a]| .
Theorem 5.3.16 Let G be a group and S be a G-set. If S is finite, then
[
|S| =
[G : Ga ],
aA

where A is a subset of S containing exactly one element from each orbit [a].
Proof. By Theorem 5.3.12, S can be partitioned as the union of orbits. Therefore,
S = aA [a].
Hence,
|S| =

aA

|[a]| =

[G : Ga ] by Lemma 5.3.15.

aA

Theorem 5.3.17 Let S be a finite G-set, where G is a group of order pn (p a prime). Let S0 = {a S | ga = a
for all g G}. Then
|S| p |S0 | .

118

5. Homomorphisms and Isomorphisms of Groups

Proof. By Theorem 5.3.16,

|S| =

[G : Ga ],

aA

where A is a subset of S containing exactly one element from each orbit [a] of G. Now a S0 if and only if
ga = a for all g G, i.e., if and only if [a] = {a}. Hence,
|S| = |S0 | +
Since |Ga | 6= |G| for all a A\S0 ,
that |S| p |S0 |

|G|
|Ga |

aA\S0

|G|
.
|Ga |

is some power of p for all a A\S0 . Thus,

|G|
|Ga |

is divisible by p, proving

Corollary 5.3.18 Let G be a finite group and H be a subgroup of G such that |H| = pk , where p is a prime and
k is a nonnegative integer. Then
[G : H] p [N(H) : H],

where N(H) = {g G | gHg 1 = H}.

Proof. Let S = {xH | x G}. Define a left action of H on S by h(xH) = (hx)H for all h H, xH S.
Then S is an H-set. Let S0 = {xH S | h(xH) = xH for all h H}. By the above theorem, |S| p |S0 | .
Now xH S0 if and
only if h(xH) = xH for all h H if and only if x1 hx H for all h H if and only if
 1
1

 = |H| . Hence, xH S0 if and only if x1 Hx H if and only if x1 Hx = H (since
x Hx H. Now
 x Hx

H is finite and x1 Hx = |H|) if and only if x N(H). This shows that S0 is the set of all left cosets of H in
N(H). Thus, |S0 | = [N(H) : H]. Also, |S| = [G : H]. Hence, [G : H] p [N(H) : H] .
Theorem 5.3.19 Let G be a group and S be a G-set. Then the left action of G on S induces a homomorphism
from G into A(S), where A(S) is the group of all permutations of S.
Proof. Let g G. Define g : S S by g (a) = ga for all a S. Let a, b S. Then g (a) = g (b) if
and only if ga = gb if and only if a = b. Therefore, g is a one-one function. Now b = g(g 1 b) = g (g 1 b) and
g 1 b S. This shows that g is onto S. Thus, g A(S). Let g1 , g2 G. Then g1 g2 (a) = (g1 g2 )a = g1 (g2 a) =
g1 (g2 a) = g1 ( g2 (a)) = ( g1 g2 )(a) for all a S. This implies that g1 g2 = g1 g2 . Define
: G A(S)
by
(g) = g
for all g G. Then is a function. Now (g1 g2 ) = g1 g2 = g1 g2 = (g1 ) (g2 ) for all g1 , g2 G. This
proves that is a homomorphism.
The following corollary, which is known as the Extended Cayleys theorem, follows from the above theorem.
Theorem 5.3.20 Extended Cayleys theorem: Let G be a group and H be a subgroup of G. Let S = {aH | a G}.
Then there exists a homomorphism from G into A(S) (the group of all permutations on S) such that Ker H.
Proof. First we note that S is a G-set, where the left action of G on S is defined by g(aH) = (ga)H for all
g G. This left action induces the homomorphism of Theorem 5.3.19. Now
Ker

=
=
=

{g G | (g) = g = the identity mapping on S}

{g G | g (aH) = aH for all aH S}
{g G | g(aH) = aH for all aH S}.

Let g Ker . Then g(aH) = aH for all aH S. In particular, gH = H. Thus, g H. Hence, Ker H.
Corollary 5.3.21 Let G be a group and H be a subgroup of G of index n .Then there exists a homomorphism
from G into Sn such that Ker H.
Proof. Because [G : H] = n, the group A(S) of all permutations on S is isomrphic to Sn . Hence the corollary
follows from the theorem.

5.3. The Groups D4 and Q8

119

Corollary 5.3.22 Let H be a subgroup of a group G of index a prime integer p. Then H is isomorphic to a
subgroup of Sp .
Corollary 5.3.23 Let G be a finite group and H be a proper subgroup of G of index n such that |G| does not
divide n! Then G contains a nontrivial normal subgroup.
Proof. From Corollary 5.3.20, Ker H and G/Ker is isomorphic to a subgroup of Sn , where is
as defined in Corollary 5.3.20. Therefore, |G/Ker | divides n! But |G| does not divide n! Hence, |Ker | 6= 1,
proving that Ker is a nontrivial normal subgroup of G.
Definition 5.3.24 Let G be a group and S be a G-set. Let a S, g G. Then a is called fixed by g if ga = a.
If ga = a for all g G, then a is called fixed by G.
Theorem 5.3.25 (Burnside) Let S be a finite nonempty set and G be a finite group. If S is a G-set, then the
number of orbits of G is
1 [
F (g),
|G| gG
where F (g) is the number of elements of S fixed by g.

Proof. Let T = {(g,

Sa) G S | ga = a}. Since F (g) is the number of elements a S such that (g, a) T,
it follows
that
|T
|
=
gG F (g). Also, |Ga | is the number of elements g G such that (g, a) T. Hence,
S
|T | = aS |Ga | .
Let S = [a1 ] [a2 ] [ak ], where {[a1 ], [a2 ], . . . , [ak ]} is the set of all distinct orbits of G on S. Then
[
[
[
[
F (g) =
|Ga | +
|Ga | + +
|Ga | .
gG

a[a1 ]

a[a2 ]

a[ak ]

Suppose a, b are in the same orbit. Then [a] = [b] and [G : Ga ] = |[a]| = |[b]| = [G : Gb ]. This implies
|G|
|G|
=
|Ga |
|Gb |
and so |Ga | = |Gb | . Thus,
S

gG

F (g)

=
=
=

|[a1 ]| |Ga1 | + |[a2 ]| |Ga2 | + + |[ak ]| |Gak |

|G|
|G | + G|G| |Ga2 | + + G|G| |Gak |
|Ga1 | a1
| a2 |
| ak |
k |G| ,

where k is the number of distinct orbits. Consequently,

k=

1 [
F (g).
|G| gG

Worked-Out Exercises
Exercise 1 Let S be a finite G-set, where G is a group of order pn (p a prime) such that p does not divide |S| . Show
that there exists
Solution: Let S0 = {a S | ga = a for all g G}. By Worked-Out Exercise 1, |S| p |S0 | . Since p does not divide
|S| , p does not divide |S0 | . Thus, |S0 | 6= 0. This shows that there exists a S0 . Thus, a is fixed by G.

Exercise 2 Let G be a finite group. Let H be a subgroup of G of index p, where p is the smallest prime dividing the
order of G. Show that H is a normal subgroup of G.
Solution: Let S = {aH | a G}. Since [G : H] = p, by Extended Cayleys theorem there exista a homomorphism
: G A(S) such that Ker H. Now G/Ker is isomorphic to a subgroup of A(S). Therefore,
|G/Ker | divides |A(S)| = p! Let |G/Ker | = n. Then n = [G : H][H : Ker ] p. Let n = p1 p2 pk ,
where pi are prime integers, i = 1, 2, . . . , k. Since pi divides |G| and p is the smallest prime dividing the
order of G, pi p for all i = 1, 2, . . . , k. Since n divides p!, we have each pi divides p!. Since each pi is a
prime and pi p, we must have i = 1 and pi = p. Thus, n = p. This implies that [H : Ker ] = 1. Hence,
H = Ker and so H is a normal subgroup of G.

120

5. Homomorphisms and Isomorphisms of Groups

Exercise 3 Let G be a group of order pn, p a prime, and p n. If H is a subgroup of order p in G, prove that H is a
normal subgroup of G.
|G|
Solution: Let S = {aH | a G}. Now |S| = [G : H] = |H|
= pn
= n. By Extended Cayleys theorem there exista
p
a homomorphism : G A(S) such that Ker H. Since |H| = p, either Ker = {e} or Ker = H.
If Ker = {e}, then G is isomorphic to a subgroup of A(S). This implies that |G| divides |A(S)| , i.e.,
pn | n! Therefore, p | (n 1)! Since p n ,p does not divide (n 1)! Thus, Ker = H. Hence, H is a
normal subgroup of G.

Exercise 4 Let G be a group. Show that G is isomorphic to a subgroup of A(G). (This is Cayleys theorem. Here we
want to prove this result by the group action method.)
Solution: G is a G-set, where the left action of G on G is defined by the group operation. This left action induces
a homomorphism : G A(G) defined by (g) = g , where g (a) = ga for all a, g G. Now Ker
= {g G | g = identity permutation on G} = {g G | ga = a for all a G} = {e}. Hence, is a
monomorphism.

Exercises
1. Show that I3 = {1, 2, 3} is a S3 -set, where the left action is defined by a = (a) for all S3 , a I3 .
Find all distinct orbits of S3 . Find G1 , G2 , and G3 .
2. Let H be a subgroup of order 11 and index 4 of a group G. Prove that H is a normal subgroup of G.
3. Let H be a subgroup of a group G of index n. If H does not contain any nontrivial normal subgroups of
G, prove that H is isomorphic to a subgroup of Sn .
4. Let G = GL(2, R) and S = R2 . Show that S is a G-set under the left action defined by


a b
(x, y) = (ax + by, cx + dy)
c d


a b
for all
G, (x, y) R2 .
c d

5. Let G be a group of order 77 acting on a set S of 20 elements. Show that G must have a fixed point.
6. Let G be a group. The left action of G on the set G is defined by conjugation, i.e., (g, x) gxg 1 for all
g, x G. Show that the kernel of the homomorphism : G A(G) induced by this action is Z(G).
7. Let G be a group of order 80 such that G has a subgroup of order 16. Show that G is not a simple group.

8. Show that a group of order 22 is not a simple group.

9. Show that there are no simple groups of orders 6, 10, 14, 26, 34, and 58.
10. Show that a group of order 8 cannot be a simple group.
11. Show that a simple group of order 63 cannot contain a subgroup of order 21.
12. Let G be a group of order 70 such that G has a subgroup of order 14. Show that G has a nontrivial normal
subgroup.

5.3. The Groups D4 and Q8

121

Arthur Cayley(18211895) was born on August 16, 1821, in Cambridge, England. He was the second son.
He entered Trinity College at the age of 17, as a pensioner. In 1842, he graduated as senior wrangler. Later he
went to a law school and in 1849 he became a lawyer. As a lawyer, he made a comfortable living and in fourteen
years, during which he practiced his law profession, he wrote approximately 300 mathematical papers.
In 1863, Cayley was elected to the new Sadlerian chair of pure mathematics at Cambridge, where he remained
until his death. He died on January 26, 1895.
For most of his life, Cayley worked on mathematics, theoretical dynamics, and mathematical astronomy. In
1876, he published his only book, Treatise on Elliptic Functions. Cayley wrote 966 papers; there are thirteen
volumes of his collected papers.
Cayleys mathematical style was terse. He usually wrote out his results and published them without delay.
He, along with J. J. Sylvester, his lifelong friend, is considered to be the founder of invariant theory. He is also
responsible for matrix theory. The square notation used for determinants is due to Cayley. He proved many
important theorems of matrix theory, such as the Cayley-Hamilton theorem. He is one of the first mathematicians
to consider geometry of more than three dimensions.
In 1854, Cayley published, On the theory of groups depending on the symbolic equation n = 1. In this
paper, he considered a group as a set of symbols, 1, , , . . . , all of them dierent and such that the product of
any two of them (no matter in what order), or the product of any one of them into itself, belongs to the set.
This formulation of a group as a set of symbols and multiplications is dierent from the formulation considered
by the earlier mathematicians. The paper is generally regarded as the earliest work on abstract group theory
and Cayley is regarded as the founder of abstract group theory. He is best known for the theorem that every
finite group is isomorphic to a suitable permutation group. In his article of 1854, he introduced a procedure for
defining a finite group by listing its elements in the form of a multiplication table, known as a Cayley table.
Cayley also proved a number of important theorems.

122

5. Homomorphisms and Isomorphisms of Groups

Chapter 6

Direct Product of Groups

6.1

External and Internal Direct Product

In Section 2.1, Exercise 25, we defined the direct product G H of two groups G and H. In this section, we
extend this concept to any finite family of groups and obtain their basic properties.
The notion of a direct product is used to factor a group into a product of smaller groups. This factorization
gives structural properties of a group. In some cases, it allows for the complete characterization of a certain type
of group. In Chapter 9, the concept of direct product is used to give a complete system of invariants for a finitely
generated Abelian group, i.e., a finite set of positive integers which implies the isomorphism of any two finitely
generated Abelian groups that have this set of integers.
Recall that In = {1, 2, . . . , n}.
Let {Gi | i In } be a family of groups. Let
G = G1 G2 Gn = {(a1 , a2 , . . . , an ) | ai Gi , i In }.
Define on G as follows: for all (a1 , a2 , . . . , an ), (b1 , b2 , . . . , bn ) G
(a1 , a2 , . . . , an ) (b1 , b2 , . . . , bn ) = (a1 b1 , a2 b2 , . . . , an bn ).
In the following theorem, we show that is a binary operation on G and that the set G together with the
binary operation is a group. We also obtain several important properties of G.
Theorem 6.1.1 Let {Gi | i In } be a family of groups and G = G1 G2 Gn . Let ei be the identity of
Gi for all i In . Then (G, ), where is defined above, is a group with e = (e1 , e2 , . . . , en ) the identity element,
and for all (a1 , a2 , . . . , an ) G,
1
1
(a1 , a2 , . . . , an )1 = (a1
1 , a2 , . . . , an ).

Furthermore, let
Hi = {(e1 , e2 , . . . , ei1 , ai , ei+1 , . . . , en ) | ai Gi }
for all i In . Then the following assertions hold.
(i) Hi is a normal subgroup of G for all i In .
(ii) For all a G, a can be uniquely expressed as a = h1 h2 hn , where hi Hi , i In .
(iii) Hi (H1 H2 Hi1 Hi+1 Hn ) = {e} for all i In .
(iv) G = H1 H2 Hn .
Proof. First we note that is single-valued and if (a1 , . . . , an ), (b1 , . . . , bn ) G, then (a1 , . . . , an )
(b1 , . . . , bn ) = (a1 b1 , . . . , an bn ) G since ai bi Gi for all i. Thus, is a binary operation on G. We ask
the reader to verify that is associative. Now e = (e1 , e2 , . . . , en ) G and for all a = (a1 , a2 , . . . , an ) G,
ae

=
=
=
=

(a1 , a2 , . . . , an )(e1 , e2 , . . . , en )
(a1 e1 , a2 e2 , . . . , an en )
(a1 , a2 , . . . , an )
a.

123

124

6. Direct Product of Groups

Similarly, ea = a. Hence, e is the identity of G. To show that every element of G has an inverse in G, let
1
1
1
(a1 , a2 , . . . , an ) G. Then (a1
Gi for all i and
1 , a2 , . . . , an ) G since ai
1
1
(a1 , a2 , . . . , an )(a1
1 , a2 , . . . , an )

=
=
=

1
1
(a1 a1
1 , a2 a2 , . . . , an an )
(e1 , e2 , . . . , en )
e.

1
1
Similarly, (a1
1 , a2 , . . . , an )(a1 , a2 , . . . , an ) = e. Thus, every element of G has an inverse. Consequently, (G, )
is a group. We also note that by the uniqueness of the inverse of an element
1
1
(a1 , a2 , . . . , an )1 = (a1
1 , a2 , . . . , an ).

(i) Let i In . Since (e1 , e2 , . . . , en )

Then
ab1 =
=
=

Hi , Hi 6= . Let a = (e1 , . . . , ai , . . . , en ), b = (e1 , . . . , bi , . . . , en ) Hi .

(e1 , . . . , ai , . . . , en )(e1 , . . . , bi , . . . , en )1
(e1 , . . . , ai , . . . , en )(e1 , . . . , b1
i , . . . , en )
,
.
.
.
,
e
)

H
.
(e1 , . . . , ai b1
n
i
i

Thus, Hi is a subgroup of G by Theorem 4.1.6. Let g = (g1 , g2 , . . . , gn ) G. Then

gag 1

=
=
=

(g1 , g2 , . . . , gn )(e1 , . . . , ai , . . . , en )(g1 , g2 , . . . , gn )1

(g1 , g2 , . . . , gi ai , . . . , gn )(g11 , g21 , . . . , gn1 )
(e1 , . . . , gi ai gi1 , . . . , en ) Hi since gi ai gi1 Gi .

Hence, Hi is a normal subgroup of G.

(ii) Let a = (a1 , a2 , . . . , an ) G. Let hi = (e1 , . . . , ai , . . . , en ) for all i In . Then a = h1 h2 hn . To show
that the representation of a is unique, let a = k1 k2 kn be another representation of a, where ki Hi for all
i In . Let ki = (e1 , . . . , bi , . . . , en ) Hi for all i In . Then
(a1 , a2 , . . . , an ) = h1 h2 hn = a = k1 k2 kn = (b1 , b2 , . . . , bn ).
This implies that ai = bi for all i In and so hi = ki for all i In . Hence, the representation of a is unique.
(iii) Suppose a Hi (H1 Hi1 Hi+1 Hn ). Then a Hi and
a H1 Hi1 Hi+1 Hn .
Since a Hi , a = (e1 , . . . , ai , . . . , en ) Hi for some ai Gi and since
a H1 Hi1 Hi+1 Hn ,
we have a = h1 h2 hi1 hi+1 hn , where hj = (e1 , . . . , aj , . . . , en ) Hj for some aj Gj . Thus,
(e1 , . . . , ai , . . . , en ) = a = h1 hi1 hi+1 hn = (a1 , . . . , ai1 , ei , ai+1, . . . , an ).
This implies that ai = ei for all i In . Hence,
Hi (H1 H2 Hi1 Hi+1 Hn ) = {e}.
(iv) The desired result follows from (ii).
Definition 6.1.2 The group G of Theorem 6.1.1 is called the external direct product of the groups Gi , i =
1, 2, . . . , n.
Theorem 6.1.1 motivates the following definition.
Definition 6.1.3 Let G be a group and {Ni | i In } be a family of normal subgroups of G. Then G is called
the internal direct product of N1 , N2 , . . . , Nn if if G = N1 N2 Nn and Ni (N1 Ni1 Ni+1 Nn ) = {e}
for all i In .
Let G = G1 G2 Gn be the external direct product of the groups Gi . Let Hi be defined as in Theorem
6.1.1. Then G is the internal direct product of H1 , H2 , . . . , Hn by Theorem 6.1.1.
Theorem 6.1.4 Let G be a group and {Ni | i In } be a family of normal subgroups of G. Then G is an internal
direct product of {Ni | i In } if and only if for all a G, a can be uniquely expressed as a = a1 a2 an , where
ai Ni , i In .

6.1. External and Internal Direct Product

125

Proof. Let G be an internal direct product of {Ni | i In }. Then G = N1 N2 Nn and Ni (N1 Ni1 Ni+1 Nn ) =
{e} for all i In . Then Ni Nj = {e} for all i 6= j and hence uv = vu for all u Ni and for all v Nj by
Exercise 13 (page 93). Let a = a1 a2 an = b1 b2 bn be two representations of a, where ai , bi Ni , i In .
Then
e = a1 a
= (a1 a2 an )1 (b1 b2 bn )
1
1
= a1
n an1 a1 b1 b2 bn
1
1
= a1 b1 a2 b2 a1
n bn
since for all i 6= j if u Ni and v Nj , then uv = vu. This implies that

1
1
1
1
b1
i ai = a1 b1 ai1 bi1 ai+1 bi+1 an bn Ni N1 N2 Ni1 Ni+1 Nn

for all i In . Since Ni N1 N2 Ni1 Ni+1 Nn = {e}, we must have b1

i ai = e or ai = bi for all i In . Thus,
a can be written uniquely as a1 a2 an , where ai Ni , i In .
Conversely assume that for all a G, a can be uniquely expressed as a = a1 a2 an , where ai Ni ,
i In .This implies that G = N1 N2 Nn . We now show that Ni (N1 Ni1 Ni+1 Nn ) = {e} for all i In .
Let i In and a Ni (N1 Ni1 Ni+1 Nn ). Then a Ni and a N1 Ni1 Ni+1 Nn . This implies
that we can write a = a1 a2 ai1 ai+1 an for some aj Nj , j In \{i}. Hence,
ee a e = a = a1 a2 ai1 eai+1 an
are two representations of a, where aj Nj , j In \{i}. Since the representation of a is unique, a = e. Hence,
Ni (N1 Ni1 Ni+1 Nn ) = {e}.
In the following theorem, we show that if a group G is an internal direct product of a family of normal
subgroups {Ni | i In }, then G can be viewed as an external direct product of the groups Ni s.
Theorem 6.1.5 Let G be an internal direct product of a family of normal subgroups {Ni | i In }. Then
G ' N1 N2 Nn .
Proof. Let a G. Then a can be expressed uniquely as a = a1 a2 an , where ai Ni , i In . Define
f : G N1 N2 Nn
by
f(a) = (a1 , a2 , . . . , an )
for all a G. From the definition of f, it follows that f is well defined and onto N1 N2 Nn . And from
the uniqueness of the representation of a, it follows that f is one-one. We now show that f is a homomorphism.
Let a = a1 a2 an and b = b1 b2 bn be two elements of G, where ai , bi Ni , i In . Now Ni Nj = {e} for
all i 6= j and so uv = vu for all u Ni , v Nj . This implies that
ab = a1 a2 an b1 b2 bn = a1 b1 a2 b2 an bn .
Thus,
f (ab)

=
=
=

(a1 b1 , a2 b2 , . . . , an bn )
(a1 , a2 , . . . , an )(b1 , b2 , . . . , bn )
f (a)f (b)

and so f is a homomorphism. Consequently, G ' N1 N2 Nn .

Considering Theorem 6.1.5, let us agree to write G = N1 N2 Nn when G is an internal direct product
of a family of normal subgroups {Ni | i In }.

Worked-Out Exercises
Exercise 1 Let G and G1 be groups and f : G G1 be a homomorphism. Let H be a normal subgroup of G. Suppose
that f |H : H G1 is an isomorphism of H onto G1 . Prove that G = H Ker f. Give an example to show
that this result need not be true if H is not a normal subgroup.
Solution: Let a G. Then f (a) G1 = f (H). Thus, there exists h H such that f(a) = f (h). Now f(a) = f (h)
implies that f (h1 a) = e1 and hence h1 a Ker f. Therefore, there exists b Ker f such that b = h1 a
or a = hb. Hence, G = HKer f. Suppose a HKer f. Then a H and f (a) = e1 = f (e). Since f |H
is one-one, f(a) = f (e) implies that a = e. Therefore, HKer f = {e}. Thus, H and Ker f are normal
subgroups of G such that G = HKer f and HKer f = {e}. Consequently, G = H Ker f.

126

6. Direct Product of Groups

This result need not be true if H is not a normal subgroup of G. For let G = S3 and G1 = hg 0 i be such that
(g0 ) = 2, i.e., G1 is a cyclic group of order 2. Let H = h(1 2)i . Define f : G G1 by f (e) = e, f (x) = e if x is
an element of order 3, and f (x) = g0 if x is an element of order 2. Then f |H : H G1 is an isomorphism of H
onto G1 . Now Ker f = {e, (1 2 3), (1 3 2)} = h(1 2 3)i . But G 6= H Ker f (see Exercise 14, page 127.)
Exercise 2 Let G be a group and H and K be subgroups of G such that G = H K. Let N be a normal subgroup of
G such that N H = {e} and N K = {e}. Prove that N is commutative.
Solution: Since G = H K, H and K are normal subgroups of G. Now for all n N, h H, k K, nh = hn, and
nk = kn by Exercise 13 (page 93). Let a, b N. Then there exist h H, k K such that b = hk. Now
ab = a(hk) = (ah)k = (ha)k = h(ak) = h(ka) = (hk)a = ba. Hence, N is commutative.
Exercise 3 Let G be a group and A and B be subgroups of G. If
(i) G = AB,
(ii) ab = ba for all a A, b B, and
(iii) A B = {e},
prove that G is an internal direct product of A and B.
Solution: Let us first show that A and B are normal subgroups of G. For this, let a A, g G. There exist c A
and b B such that g = cb by (i). Now gag 1 = (cb)a(cb)1 = cbab1 c1 = cabb1 c1 = cac1 A.
Hence, A is a normal subgroup of G. Similarly, B is a normal subgroup of G. Let g G. Then g = ab
for some a A, b B. Suppose g = a1 b1 , where a1 A, b1 B. Then ab = a1 b1 , which implies that
1
a1
A B = {e}. Thus, a = a1 and b = b1 . Therefore, we find that every element g of G can
1 a = b1 b
be expressed uniquely as g = ab, a A, b B. Consequently, G is an internal direct product of A, B.
Exercise 4 Let G be a cyclic group of order mn, where m, n are positive integers such that gcd(m, n) = 1. Show that
G ' Zm Zn .
Solution: Since m divides |G| and G is cyclic, there exists a unique cyclic subgroup A of G of order m by Theorem
4.2.11. Similarly, there exists a unique cyclic subgroup B of G of order n. Now |A B| divides |A| = m
and |A B| divides |B| = n. Since gcd(m, n) = 1, |A B| = 1. Thus, by Theorem 4.3.14,
|AB| =

|A| |B|
mn
=
= mn = |G| .
|A B|
1

Since AB G, |AB| = |G| , and G is finite, we must have G = AB. Hence, G = AB, A B = {e}, and A
and B are normal subgroups of G. Thus, G = A B ' Zm Zn .
Exercise 5 Let A and B be two cyclic groups of order m and n, respectively. Show that A B is a cyclic group if and
only if gcd(m, n) = 1.
Solution: Let A = hai for some a A and B = hbi for some b B. Suppose gcd(m, n) = 1. Let g = (a, b). Then
g mn = (a, b)mn = (amn , bmn ) = (eA , eB ), where eA denotes the identity of A and eB denotes the identity
of B. Suppose (g) = t. Then (a, b)t = (eA , eB ). This implies that at = eA and bt = eB . Thus, m | t and
n | t. Since gcd(m, n) = 1, mn | t. Hence, mn is the smallest positive integer such that g mn = e. Thus,
(g) = mn. Now |A B| = mn and A B contains an element g of order mn. As a result, A B is cyclic.
Conversely, assume that A B is cyclic and gcd(m, n) = d 6= 1. Let (a, b) A B. Then (a) | m and
(b) | n. Now mn
= m
n = m nd is an integer and mn
< mn. Also,
d
d
d
(a, b)

mn
d

= (am d , bn d ) = (eA , eB ).

Hence, A B does not contain any element of order mn. This implies that A B is not cyclic, a contradiction. Therefore, gcd(m, n) = 1.
Exercise 6 Show that |Aut(Z2 Z2 )| = 6.
Solution: First note that Z2 Z2 has four elements, e = ([0], [0]), a = ([1], [0]), b = ([0], [1]), c = ([1], [1]), and (a) =
(b) = (c) = 2. Let f Aut(Z2 Z2 ). Then (f (x)) = (x) for all x Z2 Z2 . Hence, f maps {a, b, c}
onto {a, b, c}. Thus, f is a permutation of {a, b, c}. Since there are only six permutations of {a, b, c}, it
follows that |Aut(Z2 Z2 )| 6. Now a + b = c, a + c = b, b + c = a, and a + a = e = b + b = c + c. Thus,
any permutation of {a, b, c} gives rise to an automorphism of Z2 Z2 . For example, let : a b, b c,
c a, and e e. Now (a + b) = (c) = a and (a) + (b) = b + c = a. Therefore, (a + b) = (a) + (b).
Similarly, (a + c) = (a) + (c), (b + c) = (b) + (c), (a + a) = (a) + (a), (b + b) = (b) + (b),
and (c + c) = (c) + (c). Hence, is an automorphism. Thus, |Aut(Z2 Z2 )| = 6.

6.1. External and Internal Direct Product

127

Exercises
1. Prove that the direct product of two groups A and B is commutative if and only if both groups A and B
are commutative.
2. Let A, B, C, and D be four groups such that A ' C and B ' D. Show that A B ' C D.

3. Let G be a group such that G = H1 H2 Hn , where Hi is a subgroup of G. Let Ki be a normal

subgroup of G such that Ki Hi , 1 i n. Let K = K1 K2 Kn . Show that
H1
H2
Hn
G
'

.
K
K1
K2
Kn
4. Let Gi be a group, 1 i n. Show that
Z(G1 G2 Gn ) = Z(G1 ) Z(G2 ) Z(Gn ).
5. Let G be a group and H and K be subgroups of G such that G = H K. Show that G/K ' H and
G/H ' K.

6. Let G be a finite cyclic group of order mn, where m and n are relatively prime. Let H and K be subgroups
of G such that |H| = m and |K| = n. Show that G = H K.

7. Prove that Aut(Z2 Z2 ) ' S3 .

8. Let G be a group and H and K be normal subgroups of G such that G = HK. Let H K = N. Show that
G/N ' H/N K/N.

9. Prove that a finite Abelian group G is the internal direct product of subgroups H and K if and only if (i)
H K = {e} and (ii) |G| = |H| |K| .

10. Show that the Klein 4-group is isomorphic to the direct product of a cyclic group of order 2 with itself.

11. Show that a cyclic group of order 4 cannot be expressed as an internal direct product of two subgroups of
order 2.
12. Show that a cyclic group of order 8 cannot be expressed as an internal direct product of two subgroups of
order 4 and 2, respectively.
13. Can the cyclic group Z12 be expressed as an internal direct product of two proper subgroups?
14. Show that S3 cannot be written as a direct product of proper subgroups.
15. Show that D4 cannot be expressed as an internal direct product of two proper subgroups.
16. Consider the groups Z2 S3 , Z2 Z6 , and Z12 . Are any two of these groups isomorphic? Is any one
noncommutative?
17. Show that the additive group (Z,+) cannot be expressed as an internal direct product of two nontrivial
subgroups.
18. Show that the additive group (Q,+) cannot be expressed as an internal direct product of two nontrivial
subgroups.

128

6. Direct Product of Groups

Heinrich Weber (18421913) was born on May 5, 1892, in Heidelberg, Germany. In 1860, he studied
mathematics and physics at the University of Heidelberg. He received his Ph.D. in 1863. He was appointed as
extraordinary professor at the University of Heidelberg in 1869 and also taught at Edgenssische Polytechnikum
in Zurich, the University of Knigsberg, the Technische Hochschule in Charlottenburg, and the universities of
Marburg, Gttingen, and Strasbourg.
Weber was a friend of Richard Dedekind and they often collaborated. Together they edited the work of
Riemann in 1876. Herman Minkowski and David Hilbert were among Webers students.
Webers main research interests were in analysis and its applications to mathematical physics and number
theory. He was encouraged by von Neumann to investigate physical problems and by Richelot to study algebraic
functions. Along the lines of Jacobi, he worked on the theory of dierential equations. He proved Abels theorem
in its most general form. He also worked on physical problems concerning heat, static and current electricity,
the motion of rigid bodies in liquids, and electrolytic displacement.
Webers most profound and penetrating work is in algebra and number theory. He, jointly with Dedekind,
did work of fundamental importance on algebraic functions.
In 1891, Weber gave the modern definition of an abstract finite group. One of his outstanding accomplishments was the proof of Kroneckers theorem, which states that absolute Abelian fields are cyclotomic.
Weber was an enthusiastic and inspiring teacher who took great interest in educational questions. He died
on May 17, 1913.

Chapter 7

Introduction to Rings
In the previous chapters, we investigated mathematical systems with one binary operation. There are many
mathematical systems, called rings, with two binary operations. The notion of a ring is an outgrowth of such
mathematical systems as the integers, rational numbers, real numbers, and complex numbers.
Although David Hilbert coined the term ring, it was E. Noether who, under the influence of Hilbert, set
down the axioms for rings. In 1914, Fraenkel gave the first definition of a ring. However, it is no longer commonly
used.
As we shall see, a ring is a particular combination of a group and a semigroup. Hence, our previous work
will prove helpful in our examination of rings. However, it is not enough to examine a set with two independent
binary operations. In order to obtain the full power of the axiomatic approach, we need a dependency between
the two operationsin particular, the distributive laws.

7.1

Basic Properties

This section parallels Chapter 2. Furthermore, we introduce several notations and definitions which will be used
throughout the text.
Example 7.1.1 Consider Z, the set integer, together with the binary operations +, usual addition of numbers,
and , usual multiplication of numbers. By Example 2.1.9, (Z, +) is a commutative group. Also by Example
2.1.10, (Z, ) is a semigroup. Moreover, the distributive laws hold in Z. That is, for any integers a, b, c Z,
a (b + c) = (a b) + (a c) and (b + c) a = (b a) + (c a).
Example 7.1.2 Consider the set Zn and the binary operations +n and n as defined in Examples 2.1.9 and
2.1.10, respectively, where n is positive integer. As shown in Example 2.1.9, (Zn , +n ) is a commutative group.
By Example 2.1.10, (Zn , n ) is a semigroup. Moreover, for any [a], [b], [c] Zn , we have
[a] n ([b] +n [c])

=
=
=
=
=

[a] n [b + c]
[a(b + c)]
[ab + ac]
[ab] +n [ac]
([a] n [b]) +n ([a] n [c]).

Similarly,
([b] +n [c]) n [a] = ([b] n [a]) +n ([c] n [a]).

That is, the distributive laws hold in (Zn , +n , n ).

In the previous two examples, we considered a mathematical system with two binary operations. In general,
the two binary operations are denoted by + (addition) and (multiplication). Under the binary operation +
the mathematical system is a commutative group, under the binary operation the mathematical system is a
semigroup, and distributes over +. There are many such mathematical systems and such mathematical systems
are called rings. More specifically, a ring is a mathematical system (R, +, ) such that (R, +) is a commutative
group, (R, ) is a semigroup, and the distributive laws hold, i.e., for all a, b, c R,
a (b + c) = (a b) + (a c),

129

130

7. Introduction to Rings
(b + c) a = (b a) + (c a).

a.

We denote the identity of (R, +) by the symbol 0. The additive inverse of an element a R is denoted by
We now give a complete definition of a ring.

Definition 7.1.3 A ring is an ordered triple (R, +, ) such that R is a nonempty set and + and are two binary
operations on R satisfying the following axioms.
(R1) (a + b) + c = a + (b + c) for all a, b, c R.
(R2) a + b = b + a for all a, b R.
(R3) There exists an element 0 in R such that a + 0 = a for all a R.
(R4) For all a R, there exists an element a R such that
a + (a) = 0.
(R5) (a b) c = a (b c) for all a, b, c R.
(R6) a (b + c) = (a b) + (a c) for all a, b, c R.
(R7) (b + c) a = (b a) + (c a) for all a, b, c R.
We call 0, the zero element of the ring (R, +, ).
During the development of the theory of rings, we will use the following conventions.
1. Multiplication is assumed to be performed before addition.
2. We write ab for a b.
3. We write a b for a + (b).
4. We refer to a ring (R, +, ) as a ring R.
Accordingly, ab + c stands for (a b) + c, ab + ac stands for (a b) + (a c), ab ac stands for (a b) + ((a c)),
where a, b, c R.
Example 7.1.4 (i) As shown in Examples 7.1.1 and 7.1.2, repsectively, (Z, +, ) and (Zn , +n , n ) are rings.
(ii) It can be shown that (Q, +, ), (R, +, ), and (C, +, ) are rings.
(iii) Consider E, the set of even integers. Because addition and multiplication of even integers is an integer,
we can show that (E, +, ) is a ring, called the ring of even integers. We leave the details as an exercise.
However, note the 0 is the additive identity of the ring E.
The ring (Z, +, ) of Example 7.1.4(i) is called the ring of integers. This ring plays an important role in
the study of ring theory. One of the basic problems in ring theory is to determine rings, which satisfy the same
type of properties as the ring of integers.
Remark 7.1.5 The ring (Zn , +n , n ) Example 7.1.4(i) is called the ring of integers mod n.
Definition 7.1.6 A ring R is called commutative if ab = ba for all a, b R. A ring R which is not commutative
is called a noncommutative ring.
From the above definition, it follows that a ring R is commutative if and only if the semigroup (R, ) is
commutative.
Example 7.1.7 Because multipliation of numbers is commutative, it follows that (E, +, ), (Z, +, ), (Q, +, ),
(R, +, ), and (C, +, ) are commutative rings. Also for any [a], [b] Zn , [a] n [b] = [b] n [a]. Hence, (Zn , +n , n )
is also a commutative ring.
Definition 7.1.8 For a ring R, the set C(R) = {a R | ab = ba for all b R} is called the center of R.
It follows that a ring R is commutative if and only if R = C(R).

7.1. Basic Properties

131

Example 7.1.9 Let M2 (Z) denote the set of all 2 2 matrices over the ring of integers. Let + and denote the
usual matrix addition and multiplication, respectively.
Then
+ and are binary operations


 on M2(Z). It is easy
0 0
a b
is the additive identity and for
M2 (Z),
to show that (M2 (Z), +, ) is a ring. Note that
0 0
c d

 

a b
a b

=
.
c d
c d

 

1 2
5 6
Now
,
M2 (Z) and
3 4
7 8


 
 
 


1 2
5 6
19 22
23 34
5 6
1 2
=
6=
=
.
3 4
7 8
43 50
31 46
7 8
3 4
Therefore, M2 (Z) is not a commutative ring.
In Z, we have 1n = n = n1 for all n Z. Such an element 1 is called an identity of the ring Z. We thus, have
the following definition.
Definition 7.1.10 Let R be a ring. An element e R is called an identity element if ea = a = ae for all
a R.
Note that an identity element of a ring R (if it exists) is an identity element of the semigroup (R, ). Therefore,
a ring cannot contain more than one identity element (Theorem 1.5.11). The identity element of a ring (if it
exists) is denoted by 1.
Definition 7.1.11 A ring R is called a ring with identity if it has an identity.
Example 7.1.12 The ring Z of integers is a ring with identity. The integer 1 is the identity element of Z.
Example 7.1.13 Let n be a positive integer. The commutative ring (Zn , +n , n ) is with identity. The identity
element is [1].
Remark 7.1.14 Let n be a positive integer. Note that the set Zn has n elements. Therefore, by Example 7.1.13,
it follows that for every positive integer n, there exists a commutative ring R with 1 such that the number of
elements in R is n.
Example 7.1.15 Consider E, the ring of even integers. In E, there does not exist any element e such that
ex = x = xe for all x E. Hence, E, is a ring without identity.
Example

 7.1.16 The ring M2 (Z) of Example 7.1.9 is a ring with identity. The identity element of M2 (Z) is
1 0
.
0 1
Example 7.1.17 Let R denote the set of all functions f : R R. Define +, on R by for all f, g R and for
all a R,
(f + g)(a) = f (a) + g(a),
(f g)(a) = f (a)g(a).

From the definition of + and , it follows that + and are binary operations on R. Let f, g, h R. Then for all
a R, we have by using the associativity of R that ((f + g) + h)(a) = (f + g)(a) + h(a) = (f (a) + g(a)) + h(a) =
f (a)+ (g(a) + h(a)) = f (a) + (g + h)(a) = (f + (g + h))(a). Thus, (f + g) + h = f + (g + h). This shows that +
is associative.
In a similar way, we can show that the other properties of a ring hold for R by using the fact that they hold
for R. Thus, (R, +, ) is a ring.
We note that the function i0 : R R, where i0 (a) = 0 for all a R, is the additive identity of R and the
element i1 R, where i1 (a) = 1 for all a R, is the identity of R. Also, for all f, g R and for all a R,
(f g)(a) = f (a)g(a) = g(a)f (a) = (g f )(a).
Thus, for all f, g R, f g = g f. Consequently, (R, +, ) is a commutative ring with identity.

132

7. Introduction to Rings

The addition and multiplication on R in Example 7.1.17 are the same as those encountered by the student
in calculus.
Example 7.1.18 Let (G, ) be a commutative group and Hom(G, G) be the set of all homomorphisms of G into
itself. By Exercise 14, (page 104), the composition of two homomorphisms of G is again a homomorphism of
G. Thus, is a binary operation on Hom(G, G). Also, is associative by Theorem 1.4.13 and iG Hom(G, G)
is the identity. Hence, (Hom(G, G), ) is a semigroup with identity.
We now define a suitable + on Hom(G, G) so that (Hom(G, G), +, ) becomes a ring with identity. Define
+ on Hom(G, G) by for all f, g Hom(G, G),
(f + g)(a) = f (a) g(a) for all a G.
(Note that is the binary operation of the group G.) Let f, g Hom(G, G). From the definition of +, it follows
that f + g is a mapping from G into G. Let a, b G. Then
(f + g)(ab)

=
=
=
=

f (ab) g(ab)
(f(a) f (b)) (g(a) g(b))
f (a) g(a) f (b) g(b)
(f + g)(a) (f + g)(b).

This shows that f + g is a homomorphism from G into G. We omit the routine verification that + is associative
and commutative. Consider the mapping
fe : G G
such that
fe (a) = e
for all a G, where e is the identity of G. Note that fe is a constant function, that maps each element of G to
e. Now for all a, b G,
fe (ab) = e = ee = fe (a)fe (b).
Thus, fe is a homomorphism of G into G. Hence, fe Hom(G, G).
Let g Hom(G, G) and a G. Now by the definition of fe , we have fe (a) = e. Thus,
(fe + g)(a)

=
=
=

fe (a) g(a)
e g(a)
g(a).

(g + fe )(a)

=
=
=

g(a) fe (a)
g(a) e
g(a).

Also,

Because a is an arbitrary element of G, we can now conclude that

fe + g = g = g + fe .
This implies that fe is the identity of (Hom(G, G), +).
We leave it as an exercise for the reader to verify that for any f Hom(G, G), the mapping f defined by
(f )(a) = f (a)1 for all a G is the additive inverse of f. Thus, (Hom(G, G), +) is a commutative group.
We now show that the left distributive law holds. For any a G and any elements f, g, h Hom(G, G),
[f (g + h)](a)

=
=
=
=
=

f ((g + h)(a))
f (g(a) h(a))
f (g(a)) f (h(a))
(f g)(a) (f h)(a)
(f g + f h)(a).

Hence, f (g + h) = (f g) + (f h). The right distributive law holds similarly. Consequently, (Hom(G, G), +,
) is a ring.
We now prove some elementary properties of rings.

7.1. Basic Properties

133

Theorem 7.1.19 Let R be a ring and a, b, c R. Then

(i) a0 = 0a = 0,
(ii) a(b) = (a)b = (ab),
(iii) (a)(b) = ab,
(iv) a(b c) = ab ac and (b c)a = ba ca.
Proof. (i) Observe that
a0 + a0 = a(0 + 0) = a0.
Thus,

Similarly, 0a = 0.
(ii) We have

a0 + a0 = a(0 + 0) = a0
(a0 + a0) + ((a0)) = a0 + ((a0))
a0 + (a0 + ((a0))) = 0
a0 + 0 = 0
a0 = 0

because a0 + ((a0)) = 0
because a0 + ((a0)) = 0
because a0 + 0 = a0.

ab + a(b) = a(b + (b)) = a0 = 0.

Also
a(b) + ab = a(b + b) = a0 = 0.
Hence,
ab + a(b) = 0 = a(b) + ab.
This implies that a(b) is an additive inverse of ab. Because the additive inverse of an element is unique,
a(b) = (ab). Similarly, (a) b = (ab).
(iii) Using (ii), we have
(a)(b) = (a(b)) = (ab) = ab.

(iv) Because b c = b + (c), a(b c) = a(b + (c)) = ab + a(c) = ab + ((ac)) (by (ii)) = ab ac. Similarly,
(b c)a = ba ca.

Corollary 7.1.20 Let R be a ring with 1. Then R 6= {0} if and only if the elements 0 and 1 are distinct.
Proof. Suppose R 6= {0}. Let a R be such that a 6= 0. Suppose 1 = 0. Then a = a1 = a0 = 0, a
contradiction. Thus, 1 6= 0. The converse follows because R has at least two distinct elements 0 and 1.
Convention From now on, we assume that the identity element 1 (if it exists) is dierent from the zero element
of the ring.
From this convention, it follows that if R is a ring with 1, then R has at least two elements, namely the
additive and multiplicatively identities.
Definition 7.1.21 Let R be a ring with 1. An element u R is called a unit (or an invertible element) if
there exists v R such that uv = 1 = vu.
We note the following properties of invertible elements.
Theorem 7.1.22 Let R be a ring with 1 and T be the set of all units of R. Then
(i) T 6= ,
(ii) 0
/ T, and
(iii) ab T for all a, b T.
Proof. (i) Because 1 1 = 1 = 1 1, it follows that 1 is a unit. Thus, 1 T. Hence, T 6= .
(ii) Suppose that 0 T. Then there exists v R such that
0v = 1 = v0.
However, by Theorem 7.1.19(i), 0v = 0. It now follows that 0 = 1, which is a contradiction. Hence, 0
/ T.
(iii) Let a, b T. There exist c, d R such that ac = 1 = ca and bd = 1 = db. Now
(ab)(dc) = a(bd)c = a1c = ac = 1
and
(dc)(ab) = d(ca)b = d1b = db = 1.
Hence, (ab)(dc) = 1 = (dc)(ab). Thus, ab is a unit, so ab T.

134

7. Introduction to Rings

Definition 7.1.23 (i) A ring R with 1 is called a division ring (skew-field) if every nonzero element of R is
a unit.
(ii) A commutative division ring R is called a field.
Note that a ring R is a division ring (or skew-field) if and only if (R\{0}, ) is a group. Therefore, if R is a
division ring, then for all a R, a 6= 0, there exists a unique element, denoted by a1 R, such that
aa1 = 1 = a1 a.
We call a1 the multiplicative inverse of a. In a similarly manner, a ring R is a field if and only if (R\{0}, ) is a
commutative group.
Example 7.1.24 Consider Z, the ring of integers. Let a Z be such that a 6= 0, a 6= 1, and a 6= 1. Now
a a1 = 1 = a1 a. That is, the multiplicative inverse of a is a1 . However, a1
/ Z. (For example, the multiplicative
inverse of 2 is 12
/ Z.). It follows that Z is not a field. Note that in Z, the only invertible elements are 1 and 1.
Example 7.1.25 (i) From Example 2.1.7, (Q, +, ) is a field, where + and are the usual addition and multiplication, respectively. Q is called the field of rational numbers.
(ii) From Example 2.1.7, (R, +, ) is a field, where + and are the usual addition and multiplication, respectively. R is called the field of real numbers.
(iii) From Example 2.1.7, (C, +, ) is a field, where + and are the usual addition and multiplication, respectively. C is called the field of complex numbers.
The following example is due to William Rowan Hamilton. Due to physical considerations, Hamilton constructed a consistent algebra in which the commutative law of multiplication fails to hold. At the time, such
a construction seemed inconceivable. His work and H.G. Grossmans work on hypercomplex number systems
began the liberation of algebra. Their work encouraged other mathematicians to create algebras, which broke
with tradition, e.g., algebras in which ab = 0 with a 6= 0, b 6= 0 and algebras with an = 0, where a 6= 0 and n is
a positive integer.
Example 7.1.26 Let QR = {(a1 , a2 , a3 , a4 ) | ai R, i = 1, 2, 3, 4}. Define + and on QR as follows:
(a1 , a2 , a3 , a4 ) + (b1 , b2 , b3 , b4 ) = (a1 + b1 , a2 + b2 , a3 + b3 , a4 + b4 )
(a1 , a2 , a3 , a4 ) (b1 , b2 , b3 , b4 )

(a1 b1 a2 b2 a3 b3 a4 b4 ,
a1 b2 + a2 b1 + a3 b4 a4 b3 ,
a1 b3 + a3 b1 + a4 b2 a2 b4 ,
a1 b4 + a2 b3 a3 b2 + a4 b1 ).

From the definition of + and , it follows that + and are binary operations on QR . Now + is associative
and commutative because addition is associative and commutative in R. We also note that (0, 0, 0, 0) QR
is the additive identity and if (a1 , a2 , a3 , a4 ) QR , then (a1 , a2 , a3 , a4 ) QR and (a1 , a2 , a3 , a4 ) =
(a1 , a2 , a3 , a4 ). Hence, (QR , +) is a commutative group. Similarly, is associative and (1, 0, 0, 0) QR is
the multiplicative identity.
Let (a1 , a2 , a3 , a4 ) QR be a nonzero element. Then N = a21 + a22 + a23 + a24 6= 0 and N R. Thus,
(a1 /N, a2 /N, a3 /N, a4 /N) QR . We ask the reader to verify that (a1 /N, a2 /N, a3 /N, a4 /N) is the
multiplicative inverse of (a1 , a2 , a3 , a4 ). Thus, QR is a division ring and is called the ring of real quaternions.
However, QR is not commutative because
(0, 1, 0, 0) (0, 0, 1, 0) = (0, 0, 0, 1) 6= (0, 0, 0, 1) = (0, 0, 1, 0) (0, 1, 0, 0).
Therefore, QR is not a field.
Consider the ring (Z8 , +8 , 8 ). Now [2], [4] Z8 and [2] 6= [0], [4] 6= [0], and
[2] 8 [4] = [2 4] = [8] = [0].
That is, [2] and [4] are nonzero, but their product is zero. There are other rings with such a property. This
motivates the following definition.
Definition 7.1.27 A nonzero element a in a ring R is called a zero divisor if there exists b R such that
b 6= 0 and either ab = 0 or ba = 0.

7.1. Basic Properties

135

Remark 7.1.28 We do not call 0 a zero divisor.

Remark 7.1.29 An element cannot be a unit and zero divisor at the same time (Worked-Out Exercise 1, page
138). Thus, a field has no zero divisors.
Consider Z, the ring of integers. In this ring, if a, b Z and a 6= 0, b 6= 0, then ab 6= 0. From this it follows
that Z has no zero divisors. However, as noted earlier Z is not a field. This motivates the following definition.
Definition 7.1.30 Let R be a commutative ring with 1. Then R is called an integral domain if R has no zero
divisors.
Example 7.1.31 The ring of integers Z is an integral domain.
domain because it is noncommutative. Also, M2 (Z) has
Example 7.1.32 The ring M2 (Z) is
 not
 an integral

1 0
0 1
zero divisors. For example,
,
M2 (Z) and
0 0
0 0


0
0

1
0



1
0

0
0

0
0

0
0

Example 7.1.33 Let F be a field. Then F is a commutative ring with 1. By Worked-Out Exercise 1, page 138,
if a F and a 6= 0, then a is not a zero divisor. It now follows that F is an integral domain. Hence, every field
is an integral domain.

Example 7.1.34 Consider Z[ 3] = {a+b 3 | a, b Z}. Then Z[ 3] is an integraldomain, where the operations
+
and are the
usual operations of addition and multiplication.
Note that 0 + 0 3 is the additive identity of

identity
of
Z[
3].
Z[ 3] and 1 +0 3 isthe multiplicative

Consider 3 Z[ 3]. Suppose 3 is a unit in Z[ 3]. Then

( 3)1 = a + b 3

for some a, b Z. If a = 0, then ( 3)1 = b 3 or

1 = 3b,

which is a contradiction because this equation has no solutions in Z. Therefore, a 6= 0. So ( 3)1 = a + b 3

implies that

1 = a 3 + 3b
or

1 3b
Q,
3=
a

a contradiction. Hence, 3 is not a unit. We can now conclude that Z[ 3] is not a field.

By arguments similar to the ones used in Example 7.1.34, we can show that the following sets are integral
domains under the usual addition and multiplication.

Z[n] = {a + b n | a, b Z}
Z[i n] = {a + bi n | a, b Z}
Z[i] = {a + bi| a, b Z}

Q[n] = {a + b n | a, b Q}
Q[i n] = {a + bi n | a, b Q}
Q[i] = {a + bi | a, b Q},

where n is a fixed positive integer and i2 = 1. In fact, it can be shown that Q[ n], Q[i n], and Q[i] are fields.
Example 7.1.35 The ring of even integers E is a commutative ring, without identity, and without zero divisors.
Thus, E is not an integral domain.
The ring appearing in the following example is sometimes useful in the construction of counterexamples.

136

7. Introduction to Rings

Example 7.1.36 Let (R, +) be a commutative group. Define multiplication on R by ab = 0 for all a, b R,
where 0 denotes the identity element of the group (R, +). Then (R, +, ) is a ring called the zero ring. If R
contains more than one element, then R is a commutative ring without 1 and every nonzero element of R is a
zero divisor.
The following theorem establishes a relation between zero divisors and the cancellation property of a ring.
Theorem 7.1.37 Let R be a ring. If R has no zero divisors, then the cancellation laws hold, i.e., for all
a, b, c R, a 6= 0, ab = ac implies b = c (left cancellation law) and ba = ca implies b = c (right cancellation
law). If either cancellation law holds, then R has no zero divisors.
Proof. Suppose R has no zero divisors. Let a, b, c R be such that ab = ac and a 6= 0. Then ab ac = 0
or a(b c) = 0. Becuase R has no zero divisors and a 6= 0, a(b c) = 0 implies that b c = 0 or b = c. Hence,
the left cancellation law holds. Similarly, the right cancellation law holds.
Conversely, suppose one of the cancellation laws hold, say, the left, i.e., if a, b, c R, a 6= 0, then ab = ac
implies b = c.
Let a be a nonzero element of R and b R. Suppose ab = 0. Then ab = a0, from which b = 0 by canceling a.
Suppose ba = 0 and b 6= 0. Then ba = b0 and by canceling b, we obtain a = 0, a contradiction. Therefore,
b = 0. Hence, R has no zero divisors.
Similarly, the right cancellation law implies that R has no zero divisors.
Definition 7.1.38 A ring R is called a finite ring if R has only a finite number of elements; otherwise R is
called an infinite ring.
The rings Z and M2 (Z) are infinite.
Example 7.1.39 Consider the ring (Zn , +n , n ). From Example 2.1.10, not every nonzero element of Zn has
an inverse. For example, suppose n is not prime, say, n = 6. Then [4] has no multiplicative inverse in Z6 . Also,
Z6 has zero divisors. We have [3] 6= [0] 6= [2]. Because [3] 6 [2] = [6] = [0], it follows that [3] and [2] are zero
divisors. Thus, Z6 is not an integral domain and thus not a field. We can also conclude that [2] and [3] do not
have multiplicative inverses because they are zero divisors.
In the following result, we assume that the ring R is commutative. This assumption can be removed and the
conclusion that R is a field remains valid. However, we have not developed the appropriate results to remove
this assumption. We will prove the theorem in its most general form in Chapter 24.
Theorem 7.1.40 A finite commutative ring R with more than one element and without zero divisors is a field.
Proof. We must show that R has an identity and that every nonzero element of R is a unit.
Suppose that R has n elements. Let a1 , a2 , . . . , an be the distinct elements of R. Let a R, a 6= 0. Now
aai R for all i, so
{aa1 , aa2 , . . . , aan } R.

If aai = aaj , then by Theorem 7.1.37, ai = aj . Therefore, the elements aa1 , aa2 , . . . , aan must be distinct.
Because R has only n elements, it follows that
R = {aa1 , aa2 , . . . , aan }.
This implies that a {aa1 , aa2 , . . . , aan }. So one of the products must be equal to a, say, aai = a. Because R is
commutative, we also have ai a = aai = a.
We show that ai is the identity of R. Let b be any element of R. Then b {aa1 , aa2 , . . . , aan }. So there exists
aj R such that b = aaj . Thus,
bai

=
=
=
=
=

ai b
ai (aaj )
(ai a)aj
aaj
b.

(because R is commutative)
(substituting for b)

This implies that ai is the identity of R. We denote the identity of R by 1. Now 1 R = {aa1 , aa2 , . . . , aan },
so one of the products, say, aak , must equal 1, i.e., aak = 1. By commutativity, ak a = aak = 1. Hence, every
nonzero element is a unit. Consequently, R is a field.
The following corollary is immediate from Theorem 7.1.40.

7.1. Basic Properties

137

Corollary 7.1.41 Every finite integral domain is a field.

In Example 2.1.10, we showed that a nonzero element [a] of Zn has an inverse if and only if gcd(a, n) = 1.
Thus, the following corollary is an immediate consequence of this fact. We leave the details as an exercise.
Corollary 7.1.42 Let n be a positive integer. Then Zn is a field if and only if n is prime.
Let R be a ring and a R. Then for any integer n, define na as follows:
0a
na
na

=
=
=

0
a + (n 1)a
(n)(a)

if n > 0
if n < 0.

We emphasize that na is not a multiplication of elements of R because R may not contain Z. We have the
following properties holding for any a, b R and any m, n Z :
(m + n)a
m(a + b)
(mn)a
m(ab)
(ma)(nb)

=
=
=
=
=

ma + na,
ma + mb,
m(na),
(ma)b = a(mb),
mn(ab).

The proofs of the above properties can be obtained by induction and the defining conditions of a ring.
We close this chapter by introducing the concept of the characteristic of a ring and proving its basic properties.
Definition 7.1.43 Let R be a ring. If there exists a positive integer n such that for all a R, na = 0, then the
smallest such positive integer is called the characteristic of R. If no such positive integer exists, then R is said to
be of characteristic zero.
Example 7.1.44 The ring Zn , n = 1, 2, 3, . . . , has characteristic n. Note that in Z6 , 3[2] = [6] = [0] and
2[3] = [6] = [0]. However, 6 is the smallest positive integer such that 6[a] = [0] for all [a] Z6 . In particular, [1]
has additive order 6.
Example 7.1.45 The rings Z, Q, R, C have characteristic 0.
Example 7.1.46 Let X be a nonempty set and P(X) the power set of X. Then (P(X), , ) is a commutative
ring with 1, where is the operation symmetric dierence. In this example, acts as + and acts as . Now
for all A P(X), 2A = AA = (A\A) (A\A) = . Thus, P(X) has characteristic 2.
Theorem 7.1.47 Let R be a ring with 1. Then R has characteristic n > 0 if and only if n is the least positive
integer such that n1 = 0.
Proof. Suppose that R has characteristic n > 0. Then na = 0 for all a R, so in particular, n1 = 0.
Suppose that m1 = 0 for some m such that 0 < m < n. Then
ma = m(1a) = (m1)a = 0a = 0
for all a R. However, this contradicts the minimality of n. Hence, n is the smallest positive integer such that
n1 = 0.
Conversely, suppose n is the smallest positive integer such that n1 = 0. Then for all a R,
na = n(1a) = (n1)a = 0a = 0.
By the minimality of n for 1, n must be the characteristic of R.
Theorem 7.1.48 The characteristic of an integral domain R is either zero or a prime.

138

7. Introduction to Rings

Proof. If there does not exist a positive integer n such that na = 0 for all a R, then R is of characteristic
zero.
Suppose there exists a positive integer n such that na = 0 for all a R. Let m be the smallest positive
integer such that ma = 0 for all a R. That is, the characteristic of R is m. Then
m1 = 0.
Suppose m is not prime. Then there exist integers m1 , m2 such that 1 < m1 < m, 1 < m2 < m, and m = m1 m2 .
Hence,
0 = (m1 m2 )1 = (m1 1)(m2 1).
Because R has no zero divisors, either m1 1 = 0 or m2 1 = 0. This contradicts the minimality of m. Hence, m is
prime.

Worked-Out Exercises
Exercise 1 Let R be a ring.
(a) Let R be with 1. Let a R be such that a has an inverse. Show that a cannot be a zero divisor.
An element a R is called idempotent if a2 = a and nilpotent if an = 0 for some positive integer
n.
(b) Let a R be a nonzero idempotent. Show that a is not nilpotent.
(c) Let R be with 1 and suppose R has no zero divisors. Show that the only idempotents in R are 0 and
1.
Solution:

(a) There exists b R such that ab = 1 = ba. Suppose that a is a zero divisor. Then there exists c R,
c 6= 0, such that ac = 0. Thus, 0 = b0 = b(ac) = (ba)c = c, which is a contradiction. Hence, a is not
a zero divisor.
(b) From the hypothesis, a2 = a. By induction, an = a for all positive integers n. Suppose a is nilpotent.
Then am = 0 for some positive integer m, so a = am = 0, which is a contradiction, so a is not
nilpotent.
(c) Clearly 0 and 1 are idempotent elements. Let e R be an idempotent. Then e2 = e, so e(e 1) = 0.
Because R has no zero divisors, either e = 0 or e 1 = 0, i.e., either e = 0 or e = 1. Therefore, the
only idempotents of R are 0 and 1.

Exercise 2 Determine positive integers n such that Zn has no nonzero nilpotent elements.
Solution: We claim that n is a square free integer, i.e., n = p1 p2 pk , where the pi s are distinct primes.
Suppose that n = p1 p2 pk , pi s are distinct primes. Let [a] Zn be nilpotent. Then [a]m = [0] for
some integer m. Hence, n divides am , so p1 p2 pk divides am . Then pi | am for all i = 1, 2, . . . , k.
Because the pi s are prime, pi | a for all i = 1, 2, . . . , k. Because p1 , p2 , . . . , pk are distinct primes, we
must have p1 p2 pk | a, i.e., n | a, so [a] = [0]. This implies that Zn has no nonzero nilpotent elements.
mk
1 m2
Conversely, suppose that Zn has no nonzero nilpotent elements. Let n = pm
1 p2 pk , where the pi s
m
m
are distinct primes and mi 1. Let m = max{m1 , m2 , . . . , mk }. Now [p1 p2 pk ] = [p1 m pm
2 pk ] = [0]
m
m
because n | (pm
p

p
).
Also,
because
Z
has
no
nonzero
nilpotent
elements,
[p
p

p
]
=
[0].
Hence,
n
1 2
k
1 2
k
mk
1 m2
n | (p1 pk ), so (pm
p

p
)
|
(p

p
).
Thus,
m

1
for
all
i
=
1,
2,
.
.
.
,
k.
Hence,
m
=
1
for all
1
i
i
k
1
2
k
i = 1, 2, . . . , k, so n is a square free integer.
Exercise 3 Show that the number of idempotent elements in Zmn , where m > 1, n > 1, and m and n are relatively
prime, is at least 4.
Solution: Clearly, [0] and [1] are idempotent elements. Because m and n are relatively prime, there exist integers a
and b such that am + bn = 1. We now show that n does not divide a and m does not divide b. Suppose
that n | a. Then a = nr for some integer r. Thus, n(rm + b) = nrm + nb = am + nb = 1. This implies
that n = 1, which is a contradiction. Therefore, n does not divide a and similarly m does not divide b.
Now m2 a = m(1 nb). This implies that [m2 a] = [m]. Hence, [ma]2 = [ma]. If [ma] = [0], then mn | ma,
so n | a, which is a contradiction. Consequently, [ma] 6= [0]. If [ma] = [1], then mn | (ma 1). Hence,
ma + mnt = 1 for some integer t. Thus, m(a + nt) = 1. This implies m = 1, which is a contradiction.
Hence, [ma] 6= [1]. Thus, [ma] is an idempotent such that [ma] 6= [0] and [ma] 6= [1]. Similarly, [nb] is an
idempotent such that [nb] 6= [0] and [nb] 6= [1]. Clearly [ma] 6= [nb]. Thus, we find that [0], [1], [ma], and
[nb] are idempotent elements of Zmn .

7.1. Basic Properties

139

Exercise 4 Determine the positive integers n such that Zn has no idempotent elements other than [0] and [1].
Solution: We show that n = pr for some prime p and some integer r > 0.

First assume that n = pr for some prime p and some positive integer r and [x] Zn be an idempotent. Then
[x]2 = [x]. Thus, pr | (x2 x) or pr | x(x 1). Because x and x 1 are relatively prime, pr | x or pr | (x 1). If
pr | x, then [x] = [0] and if pr | (x 1), then [x] = [1]. Thus, [0] and [1] are the only two idempotent elements.
mk
1 m2
Conversely, suppose that [0] and [1] are the only two idempotent elements. Let n = pm
1 p2 pk , where the
mk
m1
m2
pi s are distinct primes, mi 1, and k > 1. Let t = p1 and s = p2 pk . Then t and s are relatively prime
and n = ts. By Worked-Out Exercise 3, Zn = Zts must have at least four idempotents, which is a contradiction.
Therefore, k = 1. Thus, n = pr for some prime p and some positive integer r.
Exercise 5 Let R be a ring. Show that the following conditions are equivalent.
(i) R has no nonzero nilpotent elements.
(ii) For all a R, if a2 = 0, then a = 0.

Solution: (i)(ii) Let a R and a2 = 0. If a 6= 0, then a is a nonzero nilpotent element of R, a contradiction. Thus,
a = 0.
(ii)(i) Let a R be such that an = 0 for some positive integer n. Suppose a 6= 0. Let n be the smallest
positive integer such that an = 0. Suppose n is even, say, n = 2m for some positive integer m. Then
(am )2 = a2m = 0, so am = 0, contradicting the minimality of n. Suppose n is odd. If n = 1, then a = 0, a
contradiction. Therefore, n > 1. Suppose n = 2m + 1. Then m + 1 < n. Thus, a2m+2 = a2m+1 a = an a = 0.
This implies that am+1 = 0, which is a contradiction of the minimality of n. Hence, R has no nonzero
nilpotent elements.

Exercise 6 An element e of a ring R is called a left (right) identity, if ea = a (ae = a) for all a R. Show that if a
ring R has a unique left identity e, then e is also the right identity of R and hence the identity of R.
Solution: Let e be the unique left identity of R. Then ex = x for all x R. Let x R. Now (xe x + e)x =
xex xx + ex = xx xx + x = x. This implies that xe x + e is a left identity. Because e is the unique
left identity, xe x + e = e, so xe = x. Thus, e is a right identity.

Exercise 7 Let R be a commutative ring with 1 and a, b R. Suppose that a is invertible and b is nilpotent. Show
that a + b is invertible. Also, show that if R is not commutative, then the result may not be true.

Solution: There exists c R such that ac = 1 = ca and there exists a positive integer n such that bn = 0. Let
d = c c2 b + c3 b2 + + (1)n+1 cn bn1 . Now (a + b)d = ac ac2 b+ ac3 b2 + + (1)n+1 acn bn1 + bc
bc2 b+ bc3 b2 + + (1)n+1 bcn bn1 = 1 cb + c2 b2 + + (1)n+1 cn1 bn1 + bc c2 b2 + c3 b3 + +
(1)n+1 cn bn = 1. Similarly, d(a + b) = 1. Hence, a + b is invertible.




0 1
0 1
and b =
. Then a is invertible and b is nilpotent.
Consider the ring M2 (Z). Let a =
1 0
0 0


0 0
Now a + b =
. Clearly a + b is a nonzero nilpotent element. Hence, a + b is not invertible.
1 0

Exercises
1. In the rings Z8 and Z6 , find the following elements:
(i) the units, (ii) the nilpotent elements, and (iii) the zero divisors.


z1
z2
, where
z 2 z 1
z denotes the complex conjugate of the complex number z. Show that (R, +, ) is a division ring, where +
and are the usual matrix addition and matrix multiplication, respectively. Is R a field?

2. Let R be the set of all 2 2 matrices over the field of complex numbers of the form

3. Let R be a ring with 1. Prove that

(i) (1)a = a = a(1) and (1)(1) = 1,
(ii) if a is a unit in R, then a is a unit in R and (a)1 = (a1 ).

4. Prove that a ring R is commutative if and only if (a + b)2 = a2 + 2ab + b2 for all a, b R.
5. Prove that a ring R is commutative if and only if a2 b2 = (a + b)(a b) for all a, b R.
6. Let R be a ring. If a3 = a for all a R, prove that R is commutative.

7. Let R be a commutative ring and a, b R. Prove that for all n N,

# $
# $
#
$
n n1
n nr r
n
n
n
(a + b) = a +
b + +
b + +
a
a
abn1 + bn .
1
r
n1

140

7. Introduction to Rings

8. If a and b are elements of a ring and m and n are integers, prove that
(i) (na)(mb) = (nm)(ab),
(ii) n(ab) = (na)b = a(nb),
(iii) n(a) = (n)a.
9. If R is an integral domain of prime characteristic p, prove that (a + b)p = ap + bp for all a, b R.

10. Let R be a ring with 1 and without zero divisors. Prove that for all a, b R, ab = 1 implies ba = 1.
11. Let R be a ring with 1. If a is a nilpotent element of R, prove that 1 a and 1 + a are units.
12. Let R be a division ring and a, b R. Show that if ab = 0, then either a = 0 or b = 0.
13. Let a R be an idempotent element. Show that (1 a)ba is nilpotent for all b R.

14. Find all idempotent elements of the ring M2 (R).

15. Let R be a ring with 1. Let 0 6= a R. If there exist two distinct elements b and c in R such that ab = ac = 1,
show that there are infinitely many elements x in R such that ax = 1. (American Mathematical Monthly
70(1961) 315).
16. Let R be an integral domain and a, b R. Let m, n Z be such that m and n are relatively prime. Prove
that am = bm and an = bn imply that a = b.
17. Let R and R0 be rings. Define + and on R R0 by for all (a, b), (c, d) R R0
(a, b) + (c, d) = (a + c, b + d) and (a, b) (c, d) = (a c, b d).
(i) Prove that (R R0 , +, ) is a ring. This ring is called the direct sum of R and R0 and is denoted by
R R0 .
(ii) If R and R0 are commutative with identity, prove that R R0 is commutative with identity.

18. Extend the notion of direct sum in Exercise 17 to any finite number of rings.
19. Prove that the characteristic of a finite ring R divides |R| .

20. Let R be a ring with 1. Prove that the characteristic of the matrix ring M2 (R) is the same as that of R.
21. If p is a prime integer, prove that (p 1)! p 1.

22. In the following exercises, write the proof if the statement is true; otherwise, give a counterexample.
(i) In a ring R, if a and b are idempotent elements, then a + b is an idempotent element.
(ii) In a ring R, if a and b are nilpotent elements, then a + b is a nilpotent element.
(iii) Every finite ring with 1 is an integral domain.
(iv) There exists a field with seven elements.
(v) The characteristic of an infinite ring is always 0.
(vi) An element of a ring R which is idempotent, but not a zero divisor, is the identity element of R.
(vii) If a and b are two zero divisors, then a + b is also a zero divisor in a ring R.
(viii) In a finite field F, a2 + b2 = 0 implies a = 0 and b = 0 for all a, b F.
(ix) In a field F, (a + b)1 = a1 + b1 for all nonzero elements a, b such that a + b 6= 0.
(x) There exists a field with six elements.

7.2

Some Important Rings

In this section, we introduce two important rings and study some of their basic properties.

Boolean Rings
We recall that in Worked-Out Exercise 1 (page 138), an element x of a ring R is called an idempotent element
if x2 = x. The zero element and identity element of a ring are idempotent elements. In the ring Z, the only
idempotent elements are0 and 1. There exist rings, which contain idempotent elements dierent from 0 and 1.
1 0
is an idempotent element.
For example, in M2 (Z),
2 0
Definition 7.2.1 A ring R with 1 is called a Boolean ring if every element of R is an idempotent.

7.2. Some Important Rings

141

Example 7.2.2 (i) Z2 is a Boolean ring.

(ii) The ring P(X) of Example 7.1.46 is a Boolean ring because for all A P(X), A A = A.
Theorem 7.2.3 Let R be a Boolean ring. Then the characteristic of R is 2 and R is commutative.
Proof. First we show that R is of characteristic 2. Let x R. Now x + x = (x + x)2 = (x + x)(x + x) =
x(x + x) + x(x + x) = x2 + x2 + x2 + x2 = x + x + x + x. This implies that 2x = 4x, so 0 = 2x. Hence,
2 1 = 0 because x was arbitrary. It follows that the characteristic of R is 2 by Theorem 7.1.47. To show R is
commutative, let x, y R. Then x + y = (x + y)2 = (x + y)(x + y) = x2 + xy + yx + y 2 = x + xy + yx + y. This
implies that 0 = xy + yx. Hence, xy = xy + 0 = xy + xy + yx or xy = 2xy + yx = yx because 2xy = 0. Thus, R
is commutative.

Regular Rings
An element x of a ring R is called a regular element if there exists y R such that x = xyx.
Definition 7.2.4 A ring R is called a regular ring if every element of R is regular.
In the ring Z, the only regular elements are 0, 1, and 1. Thus, Z is not a regular ring.
Example 7.2.5 Let R be a division ring and x R. If x = 0, then x = xxx. Suppose x 6= 0. Then xx1 = 1, so
x = xx1 x. Thus, R is a regular ring.
From the definition of a Boolean ring, it follows that every Boolean ring is a regular ring. The field R is a
regular ring, but not a Boolean ring.
Example 7.2.6 Consider R, the field of real numbers and
RR={(x, y) | x, y R}.
Define + and on RR by

(x, y) + (z, w)
(x, y) (z, w)

=
=

(x + z, y + w)
(xz, yw)

for all x, y, z, w R. Then RR is a commutative ring with identity. Now (1, 0), (0, 1) RR and (1, 0)(0, 1) =
(0, 0). This shows that RR contains zero divisors, so RR is not a field. We claim that RR is regular. Let
(x, y) RR. If x = 0 = y, then (x, y)(x, y)(x, y) = (x, y). If x 6= 0 and y 6= 0, then (x, y)(x1 , y 1 )(x, y) = (x, y).
If x = 0, but y 6= 0, then (x, y)(x, y 1 )(x, y) = (x, y). Similarly, if x 6= 0 and y = 0, then (x, y)(x1 , y)(x, y) =
(x, y). Thus, in any case, (x, y) is a regular element. Hence, RR is a regular ring.
Example 7.2.7 Let M2 (R) be the set of all 2 2 matrices over R. Now M2 (R) is a noncommutative ring with
1, where + and are theusual matrix addition and multiplication, respectively. We show that M2 (R) is a regular
x y
ring. Let A =
M2 (R).
z w


y
w
xwzy
xwzy
Case 1: xw zy 6= 0. Then B =
M2 (R) and A = ABA.
z
x
xwzy

xwzy

Case 2: xw zy = 0.

Subcase 2a: x, y, z, w are all zero. In this case, A =

0
0

0
0

, so for any B M2 (R), ABA = A.


 1
0
. Then
Subcase 2b: x, y, z, w are not all zero. Suppose x 6= 0 and let B = x
0 0
ABA

=
=
=

x
 z
1

z
x

x
z

 1
y
x
w  0
0
x
0  z
y
=
zy
x


0
x
0  z
y
w

x y
z w

y
w

142

7. Introduction to Rings

because xw zy = 0 and x 6= 0 implies w =

ABA

=
=
=

zy
.
x

If y 6= 0, then let B =


0
x y
1
z w
 y

1 0
x
w
0
z
 
 y
x
y
=
wx
w
y

0
1
y

0
0



0
x y
0
z w

y
w

x y
.
z w

. Then

Similarly, if z 6= 0 or w 6= 0, then we can find B such that ABA = A. Thus, M2 (R) is a regular ring.
Because M2 (R) is not a division ring, it follows that a regular ring need not be a division ring. However, a
division ring is a regular ring as shown in Example 7.2.5. In the next theorem, we show that a regular ring under
a suitable condition becomes a division ring.
Theorem 7.2.8 Let R be a regular ring with more than one element. Suppose for all x R, there exists a
unique y R such that x = xyx. Then
(i) R has no zero divisors,
(ii) if x 6= 0 and x = xyx, then y = yxy for all x, y R,
(iii) R has an identity,
(iv) R is a division ring.
Proof. (i) Let x be a nonzero element of R and xz = 0 for some z R. Now by the hypothesis, there exists
a unique y R such that xyx = x. Thus,
x(y z)x = xyx xzx = xyx.
Hence, by the uniqueness of y, y z = y, so z = 0. This proves that R has no zero divisors.
(ii) Let x 6= 0 and xyx = x. Then
x(y yxy) = xy xyxy = xy xy = 0.
Because R has no zero divisors and x 6= 0, y yxy = 0, so yxy = y.
(iii) Let 0 6= x R. Then there exists a unique y R such that xyx = x. Let e = yx. If e = 0, then
x = xyx = 0, which is a contradiction. Therefore, e 6= 0. Also,
e2 = yxyx = y(xyx) = yx = e.
Let z R. Then

(ze z)e = ze2 ze = ze ze = 0.

Thus, by (i), either ze z = 0 or ze = z. Similarly, e(ez z) = 0 implies that ez = z. Hence, e is the identity of
R.
(iv) By (iii), R contains an identity element e. To show R is a division ring, it remains to be shown that every
nonzero element of R has an inverse in R. Let x be a nonzero element in R. Then there exists a unique y R
such that xyx = x. Thus, xyx = xe, i.e., x(yx e) = 0. Because R has no zero divisors and x 6= 0, yx e = 0,
so yx = e. Similarly, xyx = ex implies xy = e. Therefore, xy = e = yx. Hence, R is a division ring.

Exercises
1. Prove that a Boolean ring R is a field if and only if R contains only 0 and 1.
2. Prove that a ring R with 1 is a Boolean ring if and only if for all a, b R, (a + b)ab = 0.
3. Let R be a Boolean ring with more than two elements. Find all zero divisors of R.

4. Let T = {f | f : RZ2 }. Define + and on T by for all f, g T, (f + g)(x) = f (x) + g(x) and
(f g)(x) = f (x)g(x) for all x R. Show that (T, +, ) is a Boolean ring.
5. Prove that a nonzero element of a regular ring with 1 is either a unit or a zero divisor.
6. Prove that the center of a regular ring is regular.
7. Let R be a ring in which each element is idempotent. Let R = R Z2 . Define + and on R by (a, [n]) +
(b, [m]) = (a + b, [n + m]) and (a, [n]) (b, [m]) = (na + mb + ab, [nm]) for all (a, [n]), (b, [m]) R. Show
that + and are well defined on R and R is a Boolean ring.

7.2. Some Important Rings

143

8. Let R be a regular ring with 1.

(i) Prove that for any a R, there exists an idempotent e R such that Ra = Re.
(ii) Prove that for any two idempotents e, f R, there exists an idempotent g R such that Re+Rf = Rg.

144

7. Introduction to Rings

Chapter 8

Subrings, Ideals, and

Homomorphisms
The most important substructure of a ring is a particular subset called an ideal. The term ideal was coined
by Dedekind in honor of Kummers work on ideal numbers. This notion of Kummer and Dedekind was used to
obtain unique factorization properties. Kummer introduced the idea of an ideal number in his work on Fermats
last theorem. Noether followed with some important results on the theory of ideals. Some of her ideas were
inspired by the work not only of Dedekind, but also of Kronecker and Lasker.

8.1

Subrings and Subfields

In this section, we introduce the idea of a subring of a ring. This concept is analogous to the concept of a
subgroup of a group.
Definition 8.1.1 Let (R, +, ) be a ring. Let R0 be a subset of R. Then (R0 , +, ) is called a subring of (R, +,
) if
(i) (R0 , +) is a subgroup of (R, +) and
(ii) for all x, y R0 , x y R0 .
Let (R0 , +, ) be a subring of the ring (R, +, ). Because R0 R and because the associativity for and the
distributive laws are inherited, (R0 , +, ) is itself a ring. We will usually suppress the operations + and and
call R0 a subring of R. When R0 and R are fields, R0 is called a subfield of R.
The following theorem gives a necessary and sucient condition for a subset to be a subring. With these
conditions it is easy to verify whether a nonempty subset of a ring is a subring or not.
Theorem 8.1.2 Let R be a ring. A nonempty subset R0 of R is a subring of R if and only if x y R0 and
xy R0 for all x, y R0 .
Proof. First suppose that R0 is a subring of R. Then R0 is a ring. Hence, for all x, y R, x y, xy R0 .
Conversely, suppose x y R0 and xy R0 for all x, y R0 . Because x y R0 for all x, y R0 , (R0 , +) is
a subgroup of (R, +) by Theorem 4.1.6. By the hypothesis, xy R0 for all x, y R0 . Hence, R0 is a subring of
R.
Example 8.1.3 (i) The ring E of even integers is a subring of Z. E is without 1.
(ii) Consider the subset E8 = {[0], [2], [4], [6]} of Z8 . Then E8 is a subring of Z8 . Hence, E8 is commutative.
However, E8 has no identity and E8 does have zero divisors, namely, [2], [4], and [6].
Example 8.1.4 Let QZ = {(a1 , a2 , a3 , a4 ) | ai Z, i = 1, 2, 3, 4}. Define + and on QZ as in Example 7.1.26.
Because the dierence and product of integers is an integer, we have
(a1 , a2 , a3 , a4 ) (b1 , b2 , b3 , b4 ) QZ
and
(a1 , a2 , a3 , a4 ) (b1 , b2 , b3 , b4 ) QZ

145

146

8. Subrings, Ideals, and Homomorphisms

for all (a1 , a2 , a3 , a4 ), (b1 , b2 , b3 , b4 ) QZ . Hence, QZ is a subring of QR .

We note that QZ is noncommutative, has an identity, and is without zero divisors. Now (0, 2, 0, 0) QZ and
as in Example (0, 2, 0, 0)1 = (0, 12 , 0, 0). However, (0, 12 , 0, 0)
/ QZ . Thus, QZ is not a division ring.
Example 8.1.5 Set QE = {(a1 , a2 , a3 , a4 ) | ai E, i = 1, 2, 3, 4}. Define + and on QE as in Example 7.1.26.
Because the dierence and product of even integers is an even integer,
(a1 , a2 , a3 , a4 ) (b1 , b2 , b3 , b4 ) QE
and
(a1 , a2 , a3 , a4 ) (b1 , b2 , b3 , b4 ) QE

for all (a1 , a2 , a3 , a4 ), (b1 , b2 , b3 , b4 ) QE . It follows that QE is a subring of QZ . In fact, QE is a noncommutative

ring without identity and without zero divisors.
Example 8.1.6 Consider the ring M2 (Z) of Example 7.1.9. Let M2 (E) denote the set of all 2 2 matrices with
entries from E. Because the sum, dierence, and product of even integers is an even integer, it follows that M2 (E)
is a subring of M2 (Z). Also, M2 (E) is a noncommutative ring without identity and with zero divisors.
Following along the lines of Theorem 8.1.2, we can prove the next theorem. We leave its proof as an exercise.
Theorem 8.1.7 Let F be a field. A nonempty subset S of F is a subfield of F if and only if
(i) S contains more than one element,
(ii) x y, xy S for all x, y S, and
(iii) x1 S for all x S, x 6= 0.

Example 8.1.8 Q and Q[ 2] = {a + b 2 | a, b Q} are subfields of R (see Worked-Out Exercise 4 below).

Theorem 8.1.9 Let R be a ring (field) and {Ri | i } be a nonempty family of subrings (subfields) of R. Then
i Ri is a subring (subfield) of R.
Proof. Suppose that {Ri | i } be a nonempty family of subrings of R. Then 0 Ri for all i . This
implies that 0 i Ri , so i Ri 6= . Let x, y i Ri . Then x, y Ri for all i . Because each Ri is a
subring, x y, xy Ri for all i . Hence, x y, xy i Ri . Hence, by Theorem 8.1.2, i Ri is a subring
of R.
Similarly, we can show that if each Ri is a subfield of the field R, then i Ri is a subfield of R.
Remark 8.1.10 It is interesting to note that the intersection of all subfields of R is Q.

Worked-Out Exercises
Exercise 1 Let X be an infinite set. Then (P(X), , ) is a ring with 1. Let
R = {A P(X) | A is finite}.
Prove the following assertions.
(a) R is a subring of P(X).

(b) R is without identity.

(c) For all A R, A 6= , A is a zero divisor in R.

(d) For all A P(X), A 6= X, A 6= , A is a zero divisor in P(X).

Solution:

(a) Because is finite, R, so R is nonempty. Let A, B R. Then A and B are finite, so A B is

finite. Now AB = (A B)\(A B), so AB is finite. Therefore, AB, A B R. Thus, R is
closed under the operations and . Now it is easy to verify that (R, , ) is a subring.

(b) Suppose R has an identity, say, E. Then E is finite. Because X is infinite, there exists a X such
that a
/ E. Now {a} R. Thus, {a} = E {a} = , which is a contradiction. Hence, R has no
identity.
(c) Let A R and A 6= . Because A is finite and X is infinite, there exists x X such that x
/ A. Now
{x} R. Because A {x} = , A is a zero divisor.

8.1. Subrings and Subfields

147

(d) Let A P(X) be such that A 6= X and A 6= . Then there exists x X such that x
/ A. Hence,
A {x} = , so A is a zero divisor.

Exercise 2 Let R be a ring such that a2 + a is in the center of R for all a R. Show that R is commutative.

Solution: Let x, y R. Then (x+y)2 +x+y C(R), i.e., x2 + xy+yx+ y 2 +x+ y C(R). Because x2 +x, y 2 +y C(R)
and C(R) is a subring (Exercise 14, page 148), xy + yx C(R). Therefore, x(xy + yx) = (xy + yx)x, so
x2 y + xyx = xyx + yx2 . Thus, x2 y = yx2 . Now x2 + x C(R), so y(x2 + x) = (x2 + x)y. Hence,
yx2 + yx = x2 y + xy, so xy = yx, proving that R is commutative.

Exercise 3 Find all subrings of the ring Z of integers. Find those subrings which do not contain the identity element.

Solution: Let n be a nonnegative integer and Tn = nZ = {nt | t Z}. Because 0 Tn , Tn 6= . Let a = nt, b = ns be
two elements in Tn . Then a b = nt ns = n(t s) Tn and ab = (nt)(ns) = n(t(ns)) Tn . Hence, Tn
is a subring of Z. We now show that if A is any subring of Z, then A = Tn for some nonnegative integer n.
Let A be a subring of Z. If A = {0}, then A = 0Z. Suppose A 6= {0}. Then there exists m A such
that m 6= 0. Now m A, so A contains a positive integer. By the well-ordering principle, A contains a
smallest positive integer. Let n be the smallest positive integer in A. Then nZA. Let m A. By the
division algorithm, there exist integers q and r such that m = nq + r, 0 r < n. Because n A, nq A.
Hence, r = m nq A. The minimality of n implies that r = 0, so m = nq nZ. Thus, A = nZ. If n 6= 1,
then nZ does not contain identity.

Exercise 4 Show that Q[ 2] = {a + b 2 R | a, b Q} is a subfield of the field R.

Solution: Because 0 = 0 + 0 2 Q[ 2], Q[ 2] 6= . Let a + b 2, c + d 2 Q[ 2]. Then

(a + b 2) (c + d 2) = (a c) + (b d) 2 Q[ 2]
and

(a + b 2) (c + d 2) = (ac + 2bd) + (ad + bc) 2 Q[ 2].

Now 0 + 0 2 and1 + 0 2 are distinct elements

of Q[ 2]. Therefore, Q[ 2] contains more than one
element. Let a + b 2 bea nonzero element of Q[
cannot both be zero simultaneously.
2]. Then a and b
0. Then a = b 2. If b = 0, then a = 0. Therefore,
We now show that a b 2 6= 0. Suppose a b 2 =

both a and b are zero, a contradiction. If b 6= 0, then 2 = ab Q, a contradiction. Hence, a b 2 6= 0.

Similarly, a + b 2 6= 0. Thus, a2 2b2 = (a + b 2)(a b 2) 6= 0. Now

1
a
b
(a b 2)
= 2
= 2
2
2 Q[ 2].
a 2b2
a 2b2
a 2b2
a+b 2

Because (a + b 2)( a+b12 ) = 1, (a + b 2)1 exists in Q[ 2]. Thus, we find that Q[ 2] is a subfield of R
by Theorem 8.1.7.

Exercises
1. Prove the following the statements.



a b
| a, b, c Z is a subring of M2 (Z).
(i) T1 =
0 c



a b
| a, b Z is a subring of M2 (Z).
(ii) T2 =
b a



a 0
| a Z is a subring of M2 (Z).
(iii) T3 =
0 a



a b
| a, b Z is a subring of T1 .
(iv) T4 =
0 a

2. In the ring Z of integers, find which of the following subsets of Z are subrings.
(i) The set of integers of the form 4k + 2, k Z.
(ii) The set of integers of the form 4k + 1, k Z.
(iii) The set of integers of the form 4k, k Z.
3. Show that T = {[0], [5]} is a subring of the ring Z10 .

4. Let R be a ring with 1. Show that the subset T = {n1 | n Z} is a subring of R.

5. Let R be a ring and n be a positive integer. Show that the subset T = {a R | na = 0} is a subring of R.

148

8. Subrings, Ideals, and Homomorphisms





a
b 3
| a, b R is a subring of M2 (R).
b 3
a

7. Show that Q[ 3] and Q[ 5] are subfields of the field R, but Z[ 2] = {a + b 2 | a, b Z} is not a subfield
of R.

6. Show that T =

8. Show that Q(i) = {a + bi | a, b Q} is a subfield of C, where i2 = 1.




a b
9. Show that F =
| a, b Z5 is a subring of M2 (Z5 ). Is F a field?
b a

10. Let be a root of x2 + x + 1 = 0. Prove that T = {a + b | a, b Q} is a subfield of the field of complex

numbers.
11. Let F be a field of characteristic p > 0. Show that T = {a F | ap = a} is a subfield of F.



x+y y
12. Prove that T =
| x, y Z is a subring of M2 (Z). Also, show that every nonzero element
y
x
of T is a unit in M2 (R).
13. Let R be a commutative ring. Show that the set
T = {r R | rn = 0 for some integer n}
is a subring of R.
14. Prove that C(R) is a subring of R and that C(R) is commutative.
15. Let e be an idempotent of a ring R. Prove that the set
eRe = {ere | r R}
is a subring of R with e as the identity element.
16. Find the center of the ring M2 (R).
17. Prove that the characteristic of a subfield is the same as the characteristic of the field.
18. Find all subrings with identity of the ring Z16 .
19. Find all subfields of the field Zp , p a prime integer.
20. Let R be a ring without any nonzero nilpotent elements. Show that (ara ra)2 = 0 for all r R and for
all idempotent elements a R. Hence, show that C(R) contains all idempotent elements.
21. Let C = { f : RR | f is continuous on R}. Define + and on C by
(f + g)(x)
(f g)(x)

=
=

f (x) + g(x),
f (x)g(x)

for all f, g C and for all x R.

(i) Show that C is a ring.
(ii) Let D = {f C | f is dierentiable on R}. Show that D is a subring of C.
22. Let R be a ring and f : R [0, 1] be such that
f (a b)
f (ab)

min{f(a), f (b)},
min{f(a), f (b)}

for all a, b R. Prove that for all t I(f ), Rt = {x R | f (x) t} is a subring of R.

23. In the following exercises, write the proof if the statement is true; otherwise, give a counterexample.
(i) The union of two subrings of a ring is a subring.
(ii) The identity element of a subring is always the identity element of the ring.
(iii) Q is the only subfield of the field R.

(iv) Q[ 3] = {a + b 3 | a, b Q} is the intersection of all subfields of R which contain 3.

(v) The set Z of integers is a subring of the field of real numbers.
(vi) Every additive subgroup of Z is a subring of Z.

8.2. Ideals and Quotient Rings

8.2

149

Ideals and Quotient Rings

In this section, we introduce the notions of ideals and quotient rings. These concepts are analogous to normal
subgroups and quotient groups.
The very famous problem called Fermats last theorem led to the invention of ideals. Fermat (16011665)
jotted many of his results in the margin of Diophantus Arithmetica. For this particular theorem, Fermat wrote
that he discovered a remarkable theorem whose proof was too long to put in the margin. The theorem is stated
as follows: If n is an integer greater than 2, then there exist no positive integers x, y, z such that xn + y n = z n .
However, no one was able to prove this result until recently; in 1994, Andrew Wiles found a proof after many
years of work.
In 1843, Kummer (18101893) thought that he had found a proof of Fermats last theorem. However, Kummer
had incorrectly assumed uniqueness of the factorization of complex numbers of the form x+y, where p = 1 for p
an odd prime. Dirichlet (18051859) had made an incorrect assumption about factorization of numbers. Kummer
continued his eorts to solve Fermats last theorem. He was partially successful by introducing the concept of
ideal number. Dedekind (18311916) used Kummers ideas to invent the notion of an ideal. Kronecker (1823
1891) also played an important part in the development of ring theory.
Definition 8.2.1 Let R be a ring. Let I be a nonempty subset of R.
(i) I is called a left ideal of R if for all a, b I and for all r R, a b I, ra I.
(ii) I is called a right ideal of R if for all a, b I and for all r R, a b I, ar I.
(iii) I is called a (two-sided) ideal of R if I is both a left and a right ideal of R.
From the definition of a left (right) ideal, it follows that if I is a left (right) ideal of R, then I is a subring of
R. Also, if R is a commutative ring, then every left ideal is also a right ideal and every right ideal is a left ideal.
Thus, for commutative rings every left or right ideal is an ideal.
By Theorem 8.1.2, it follows that a nonempty subset I of a ring R is an ideal if and only if (I, +) is a
subgroup of (R, +) and for all a I and for all r R, ar and ra I.
Example 8.2.2 Let R be a ring. The subsets {0} and R of R are (left, right) ideals. These ideals are called
trivial ideals. All other (left, right) ideals are called nontrivial.
An ideal I of a ring R is called a proper ideal if I 6= R.
Example 8.2.3 Let n Z and I = {nk | k Z}. As in Worked-Out Exercise 3 (page 147), I is a subring. Also,
for all r Z, (nk)r = n(kr) I and r(nk) = n(rk) I. Hence, I is an ideal of Z.
Next, we give an example of a ring in which there exists a left ideal which is not a right ideal, a right ideal
which is not a left ideal, and a subring which is not a left (right) ideal.
Example 8.2.4 Consider the ring M2 (Z). Let



a 0
I1 =
| a, b Z ,
b 0



a b
| a, b Z ,
I2 =
0 0



a c
| a, b, c and d are even integers ,
I3 =
b d
and



0
|aZ .
0

 




a 0
c 0
x y
0
6 . Let
,
I1 and
M2 (Z). Then
I1 , I1 =
b 0
d 0
z w
0

 
 

a 0
c 0
ac 0

=
I1
b 0
d 0
bd 0
I4 =

Because

and

0
0

x
z

y
w



a
b



a
0

0
0

xa + yb
za + wb

0
0

I1 ,

150

8. Subrings, Ideals, and Homomorphisms

proving that I1 is a left ideal of M2 (Z). Now



1
1

0
0

1
1



0
0
0
0


1
0

I1 and


0
0

0
0

1
1

1
0

M2 (Z), but

/ I1 .

Hence, I1 is not a right ideal of M2 (Z). Similarly, I2 is a right ideal of M2 (Z), but not a left ideal, I3 is an ideal
of M2 (Z), and I4 is a subring, but not an ideal of M2 (Z).
We remind the reader to notice the similarity of the next few results with corresponding results in linear
algebra and group theory.
Theorem 8.2.5 Let R be a ring and {I | } be a nonempty collection of left (right) ideals of R. Then
I is a left (right) ideal of R.
Proof. Suppose {I | } is nonempty a collection of left ideals of R. Because 0 I for all , 0 I .
Thus, I 6= . Let a, b I . Then a, b I for all . Because each I is a left ideal, a b I for all
. Hence, a b I . Let r R. Because each I is a left ideal of R, ra I for all . This implies that ra
I . Thus, I is a left ideal of R.
Similarly, if {I | } is a nonempty collection of right ideals of R, then I is a right ideal of R.
Notation 8.2.6 Let a1 , a2 , . . . , an R. Then by the notation

Sn

i=1

ai , we mean the sum a1 + a2 + + an .

Definition 8.2.7 Let S be a nonempty subset of a ring R.

(i) Define hSil to be the intersection of all left ideals of R which contain S.
(ii) Define hSir to be the intersection of all right ideals of R which contain S.
(iii) Define hSi to be the intersection of all right ideals of R which contain S.
Using Theorem 8.2.5, the following theorem is immediate.
Theorem 8.2.8 Let S be a nonempty subset of a ring R.
(i) hSil is a left ideal. The left ideal hSil is called the left ideal generated by S.
(ii) hSir is a right ideal of R. The right ideal hSir is called the right ideal generated by S.
(iii) hSi is an ideal of R. The ideal hSi is called the ideal generated by S.
Note that hSil is the smallest left ideal of R which contains S.
Theorem 8.2.9 Let R be a ring and S be a nonempty subset of R. Then
(i)
S
S
hSil = { ki=1 ri si + lj=1 nj tj | ri R, nj Z, si , tj S,
1 i k, 1 j l, k, l N}.
(ii)
hSir

Proof. (i) Let

A

S
S
{ ki=1 si ri + lj=1 nj tj | ri R, nj Z, si , tj S,
1 i k, 1 j l, k, l N}.
S
S
{ ki=1 ri si + lj=1 nj tj | ri R, nj Z, si , tj S,
1 i k, 1 j l, k, l N}.

S
Because hSil is the intersection of all left ideals of R which contain S, it follows that hSil S. Let ki=1 ri si +
Sl
j=1 nj tj A, where ri R, nj Z, si , tj S, 1 i k, 1 j l, k, l N. Now si , tj S hSil . Because
hSil is a left ideal, ri si , nj tj hSil , 1 i k, 1 j l. Once again using the fact that hSil is a left ideal, we
S
S
can conclude that ki=1 ri si + lj=1 nj tj hSil . Hence, A hSil .
We now show that A is a left ideal of R such that S A. We can then conclude that hSil A because hSil
is the smallest left ideal of R containing S.
S
S
S
S
Let s S. Then s = 0 s + 1s A. Hence, S A. Let ki=1 ri si + lj=1 nj tj and pi=1 ri0 s0i + qj=1 n0j t0j A.
Then
S
 S

Sl
Sq
k
p
0 0
0 0
i=1 ri si +
j=1 nj tj
i=1 ri si +
j=1 nj tj
 S

S
Sp
Sq
k
l
0 0
0 0
=
i=1 ri si +
i=1 (ri )si +
j=1 nj tj +
j=1 (nj )tj A

8.2. Ideals and Quotient Rings

151

 S
S
Sl
S
k
= ki=1 (rri )si + lj=1 (nj r)tj A. Thus, A is a left ideal of R. It
Let r R. Then r
i=1 ri si +
j=1 nj tj
now follows that hSil A.
Consequently, hSil = A
(ii) The proof is similar to (i).
Corollary 8.2.10 Let R be a ring and S be a nonempty subset of R. If R is with 1, then
(i)
+ k
,
[
ri si | ri R, si S, 1 i k, n N .
hSil =
i=1

(ii)

hSir =

+ k
[
i=1

si ri | ri R, si S, 1 i k, n N .

r
Proof. (i) Let A =
i=1 ri si | ri R, si S, 1 i k, n N . Observe that A hSil .
S
S
S
Let ti=1 ri si + lj=1 nj tj hSil . Because R has an identity 1, nj tj = (nj 1)tj and nj 1 R. Thus, ti=1 ri si +
St
Sl
Sl
j=1 nj tj =
i=1 ri si +
j=1 (nj 1)tj A. Hence, hSil A. Consequently,
qS
k

hSil =

+ k
[
i=1

ri si | ri R, si S, 1 i k, n N .

(ii) The proof is similar to (i).

Let us note the following:
1. If S = {a1 , a2 , . . . , an }, then the left ideal hSil generated by S is denoted by ha1 , a2 , . . . , an il . In this case,
we call hSil a finitely generated left ideal. Similar terminology is used for hSir and hSi.

2. If S = {a}, then

(a) hail is called the principal left ideal generated by a,

(b) hair is called the principal right ideal generated by a, and

(c) hai is called the principal ideal generated by a.

Corollary 8.2.11 Let R be a ring and a R.

(i) Then
hail = {ra + na | r R, n Z}.
(ii) If R is with 1, then

hail = {ra | r R}.

Proof. (i) This assertion follows from the equality
k
[
i=1

ri a +

m
[
j=1

k
m
[
[
nj a = (
ri )a + (
nj )a.
i=1

j=1

(ii) This follows from (i) and Corollary 8.2.10.

S
Similarly, we can prove that hair = {ar +na | r R, n Z} and hai = {ra+as+na+ ki=1 ri asi | r, s, ri , si
R, n Z,1 i k, k N}.
Consider the subsets Ra = {ra | r R} and aR = {ar | r R} of R. If R is without identity, then Ra (aR)
is still a left (right) ideal of R (Exercise 4, page 157). It is not necessarily the case that a Ra (a aR) as
illustrated by the next example.
Example 8.2.12 Consider the ring E of even integers. E does not have an identity. Now
h2i = {r2 + n2 | r E, n Z} = {0, 2, 4, . . .}
and 2 h2i . However,
which does not contain 2.

R2 = {r2 | r E} = {0, 4, 8, . . .},

152

8. Subrings, Ideals, and Homomorphisms

In the next theorem, we obtain a necessary and sucient condition for a ring with 1 to be a division ring.
Theorem 8.2.13 Let R be a ring with 1. Then R is a division ring if and only if R has no nontrivial left ideals.
Proof. Suppose R is a division ring. Let I be a left ideal of R such that I {0}. Then there exists a I
such that a 6= 0. Now a 6= 0, so a1 R. Thus, because I is a left ideal, 1 = a1 a I. This implies that for all
r R, r = r1 I. This shows that R I. Because, I R, we can now conlude that R = I. Consequently, R
has no nontrivial left ideals.
Conversely, suppose R has no nontrivial left ideals. Let a R and a 6= 0. We show that a is a unit.
Because R has no nontrivial left ideals and a 6= 0, we must have hail = R. This implies that 1 hail . By
Corollary 8.2.11(ii), hail = {ra | r R}. Thus, there exists r R such that 1 = ra. This implies that r 6= 0.
Now r 6= 0, so proceeding as in the case of the nonzero element a, we can show that tr = 1 for some t R.
Thus, we have
t = t1 = t(ra) = (tr)a = 1a = a.
This, implies that ar = 1. Hence,
ra = 1 = ar,
i.e., a is a unit. Consequently, every nonzero element of R is a unit. Hence, R is a division ring.
Following along the lines of the above theorem, we can prove that a ring R with 1 is a division ring if and
only if R has no nontrivial right ideals.
The following corollary is immediate from Theorem 8.2.13.
Corollary 8.2.14 Let R be a commutative ring with 1. Then R is a field if and only if R has no nontrivial
ideals.
Definition 8.2.15 A ring R is called a simple ring if R2 6= {0} and {0} and R are the only ideals of R.
Example 8.2.16 Every division ring is a simple ring.
Example 8.2.17 In this example, we show that
 M2 (R) is a simple ring. Let A be a nonzero ideal of M2 (R).
a b
Then there exists a nonzero element
A. Now at least one of a, b, c, d is nonzero. Because A is an
c d

 

0 0
0 1
ideal and
,
M2 (R), we have
1 0
0 0


 

a b
0 0
b 0
=
A,
c d
1 0
d 0


 

0 1
a b
c d
=
A,
0 0
c d
0 0
and

0
0

1
0



Therefore, we find that A contains a matrix


Thus,

Finally,

Hence,

1
0

0
0



a
c

b
d



a
c


a1
0

b
d
a
c
0
0



0
1

b
d

such that a 6= 0. Now a1 R and

1
0

0
0



0
0

d
0

0
0

1
ca1

0
0

1
0

0
0



0
0

1
0

0
0

1
0

A.

0
1

0
0



0
0

1
0

0
0

0
1

A.

1
0

0
1

1
0

0
0

0
0

0
1

A.

A.

This implies that A = M2 (R). Also note that M2 (R) is not a division ring.

1
0

0
0

A.

8.2. Ideals and Quotient Rings

153

Example 8.2.17 shows that there are simple rings, which are not division rings.
Notation 8.2.18 For a R, aRa denotes the set {ara | r R}.
We now consider the sum and product of left (right) ideals.
Let A and B be two nonempty subsets of a ring R. Define the sum and product of A and B as follows:
A + B = {a + b | a A, b B}
AB = {a1 b1 + a2 b2 + + an bn | ai A, bi B, i = 1, 2, . . . , n, n N}.
S
ai bi , ai A, bi B.

Thus, AB denotes the set of all finite sums of the form

Let n N. Inductively, we define
A1
An

=
=

A,
AAn1

if n > 1.

Example 8.2.19 Consider Z, the ring of integers. Let A = h2i , the ideal generated by 2, and B = h3i , the ideal
generated by 3. Note that A = {2n | n Z} and B = {3n | n Z}. Let m Z. Now m = 2(m) + 3m A + B.
This implies that A + B = Z.
Next we dsetermine AB. Let m Z. Now 6n = 2 (3n) AB. This implies that h6i = {6n | n Z} AB.
Let x AB. Then
x = a1 b1 + a2 b2 + + an bn ,
for some n N, where ai A, bi B, i = 1, 2, . . . , n. Now ai = 2ti and bi = 3si , for some ti , si Z,
i = 1, 2, . . . , n. Hence
x = a1 b1 + a2 b2 + + an bn
= (2t1 )(3s1 ) + (2t2 )(3s2 ) + + (2tn )(3sn )
= 6(t1 s1 ) + 6(t2 s2 ) + + 6(tn sn )
= 6(t1 s1 + t2 s2 + + tn sn )
= 6k h6i , where k = t1 s1 + t2 s2 + + tn sn Z.
Thus, AB h6i . Hence, AB = h6i .

We now list some interesting properties of the sum and product of left (right) ideals.
Theorem 8.2.20 Let A, B, and C be left (right) ideals of a ring R. Then the following assertions hold.
(i) A A + B.
(ii) A + B = B + A is a left (right) ideal of R.
(iii) A + A = A.
(iv) (A + B) + C = A + (B + C).
(v) AB is a left (right) ideal of R.
(vi) (AB)C = A(BC).
(vii) If A, B and C are ideals, then A(B + C) = AB + AC, (B + C)A = BA + CA.
(viii) If A is a right ideal and B is a left ideal, then AB A B.
(ix) R is a regular ring if and only if for any right ideal A and for any left ideal B, AB = A B.
Proof. We only prove (ix) and (x) and leave the other properties as exercises.
(ix) Suppose R is a regular ring. Let a A B. There exists b R such that a = aba. Because B is a left
ideal and a B, ba B. Thus, a = a(ba) AB, whence A B AB. By (vii), AB A B. Consequently,
AB = A B. Conversely, assume that AB = A B for any right ideal A and left ideal B of R. Let a R
and consider hair , the right ideal generated by a. Because hair is a right ideal, hair R hair . Also, by our
assumption hair R = hair R. Hence,
a hair R = hair R.
Sn
Therefore, a = i=1 ai bi for some ai hair , bi R, i = 1, 2, . . . , n. From the statements following Corollary
8.2.11, ai = ati + ni a for some ti R, ni Z, i = 1, 2, . . . , n. Thus,
a=

n
[
i=1

ai bi =

n
n
[
[
(ati + ni a)bi = a( (ti bi + ni bi )) aR.
i=1

i=1

This implies that hair = aR. Because aR hair , hair = aR. Similarly, hail = Ra. It now follows that a
aR Ra = (aR)(Ra) aRa. Hence, there exists b R such that a = aba, i.e., a is regular. Consequently, R is
regular.

154

8. Subrings, Ideals, and Homomorphisms

Quotient Rings
We now give the analogue of quotient groups for rings. Let R be a ring and I an ideal of R. Let x R. Let x + I
denote the set
x + I = {x + a | a I}.

The set x + I is called a coset of I. For x, y R, we leave it as exercise for the reader to verify that
x + I = y + I if and only if x y I.

This property of cosets is, in fact, analogous to property of cosets for a group, (see Theorem 4.3.3). Moreover
notice that
0 + I = I.
Let R/I denote the set
R/I = {x + I | x R}.

Because I = 0 + I R/I, R/I is a nonempty set. Define the operations + and on R/I as follows: for all x + I,
y + I R/I
(x + I) + (y + I) = (x + y) + I,
and
(x + I) (y + I) = xy + I.

We leave it as an exercise for the reader to verify that + and are binary operations on R/I.
Under these binary operations (R/I, +, ) satiesfies the properties of a ring. Let us verify some of these
properties.
Let x + I, y + I, z + I R/I. Now
(x + I) + ((y + I) + (z + I))

=
=
=
=
=

(x + I) + ((y + z) + I)
(x + (y + z)) + I
((x + y) + z) + I, because + is assocaitive in R
((x + y) + I) + (z + I)
((x + I) + (y + I)) + (z + I).

This shows that + is associative in R/I. Similarly, + is commutative. Next, note that 0 + I = I is the additive
identity and for x + I R/I, (x) + I is the additive inverse of x + I. As in the case of the associativity for +,
we can show that is associative.
Next, let us verify one of the distributive law. Now
(x + I) ((y + I) + (z + I))

=
=
=
=
=

(x + I) ((y + z) + I)
(x(y + z)) + I
(xy + xz) + I, because distributivity holds in R
(xy) + I) + (xz + I)
((x + I) (y + I)) + ((x + I) (z + I)).

In a similar manner, we can verify the right distributive property.

Remark 8.2.21 The quotient ring R/I can also be realised by oberving the (I, +) is a normal subgroup of (R, +)
because the latter group is commutative. Hence, if R/I denotes the set of all cosets x + I = {x + a | a I} for
all x R, then (R/I, +) is a commutative group, where
(x + I) + (y + I) = (x + y) + I
for all x + I, y + I R/I. Now define multiplication on R/I by (x + I) (y + I) = xy + I for all x + I, y + I R/I.
Then (R/I, +, ) forms a ring.
Definition 8.2.22 If R is a ring and I is an ideal of R, then the ring (R/I, +, ) is called the quotient ring of
R by I.
Theorem 8.2.23 Let n Z be a fixed positive integer. Then the following conditions are equivalent.
(i) n is prime.
(ii) Z/ hni is an integral domain.
(iii) Z/ hni is a field.

8.2. Ideals and Quotient Rings

155

Proof. Note that hni is the ideal of Z generated by n and hni = {nt | t Z}.
(i) (ii): Suppose n is prime. Let a + hni , b + hni Z/ hni . Suppose
(a + hni)(b + hni) = 0 + hni .
Now

(a + hni)(b + hni) = 0 + hni

ab + hni = 0 + hni
ab hni
ab = rn
n | ab.
either n | a or n | b
either a hni or b hni
either a + hni = 0 + hni or b + hni = 0 + hni

for some r Z
because n is prime

We can now conclude that Z/ hni has no zero divisors, proving that Z/ hni is an integral domain.
(ii)(iii): Because Z/ hni is a finite integral domain, the result follows from Theorem 7.1.40.
(iii)(i): Suppose n is not prime. Then n = n1 n2 for some 1 < n1 < n and 1 < n2 < n. Because 1 < n1 < n,
we have n1 <
/ n > . Similarly, n2 <
/ n > . Hence n1 + hni and n2 + hni are nonzero elements of Z/ hni and
(n1 + hni)(n2 + hni) = n1 n2 + hni = n + hni = 0 + hni .
Because Z/ hni is a field, Z/ hni has no zero divisors. Thus, either n1 + hni = 0 + hni or n2 + hni = 0 + hni , i.e.,
either n1 hni or n2 hni ,a contradiction. Therefore, n is prime.
We close this section by introducing the notions of nil and nilpotent ideals.
Definition 8.2.24 Let I be an ideal of a ring R.
(i) I is called a nil ideal if each element of I is a nilpotent element.
(ii) I is called a nilpotent ideal if I n = {0} for some positive integer n.
Example 8.2.25 In the ring Z8 , the ideal I = {[0], [4]} is a nil ideal and also a nilpotent ideal.
+ k
,
[
[ai ][bi ] | [ai ], [bi ] I,k N = {0}
I =
2

i=1

because 16 | ai bi .
From the definition, it follows that every nilpotent ideal is a nil ideal. The following example shows that the
converse is not true. In this example, we construct a ring R from the rings Zpn , n = 1, 2, . . . , i.e., from the rings
Zp , Zp2 , Zp3 , . . . , where p is a fixed prime.
Example 8.2.26 Let p be a fixed prime. Let R be the collection of all sequences {an } such that an Zpn (n 1)
and there exists a positive integer m (dependent on {an }) such that an = [0] for all n m. Define addition and
multiplication on R by
{an } + {bn } = {an + bn },
{an }{bn } = {an bn }
for all {an }, {bn } R. We ask the reader to verify that R is a commutative ring under these two operations, where
the zero element is the sequence {an } such that an = [0] for all n and the additive inverse of the sequence {an }
is the sequence {an }. Now in Zpn , [p] is a nilpotent element because [p]n = [pn ] = [0]. Thus, for any [r] Zpn ,
[p][r] = [pr] is a nilpotent element. Therefore, we find that each element of [p]Zpn is a nilpotent element.
Let
I = {{[p]a1 , [p]a2 , . . . , [p]an , [0], [0], . . .} R | n N, ai Zpi , i = 1, . . . , n}.
Then I is an ideal of R. Also, every element of I is nilpotent. Let us now show that I is not nilpotent. Suppose
I is nilpotent. Then there exists a positive integer m such that I m = {0}. Now the sequence {an } such that
an = [p] for n = 1, 2, . . . , m + 1 and an = 0 for all n m + 2 is an element of I. Then {an }m = {[0], [0], . . . , [0],
[pm ], [0], [0], . . .}, where the (m + 1)th term of this sequence is [pm ] and all other terms are 0. Because [pm ] is
not zero in Zpm+1 , we find that {an }m 6= 0 and {an }m I m = {0}, a contradiction. This implies that I is not
nilpotent.

156

8. Subrings, Ideals, and Homomorphisms

Theorem 8.2.27 Let R be a commutative ring with 1 and I denote the set of all nilpotent elements of R. Then
(i) I is a nil ideal of R,
(ii) the quotient ring R/I has no nonzero nilpotent elements.
m

Proof. (i) Because 0 I, I 6= . Let a, b I. There exist positive integers m and n such that an = 0 and
= 0. Because R is commutative, we can write
#
$
n+m
n+m
r n+m
=a
+ + (1)
(a b)
an+mr br + + (1)n+m bn+m .
r


 n+mr r
a
b , where 0 r m + n. If r m, then
The general term of the above expression is (1)r n+m
r
n + m r n and hence an+mr = an amr = 0. Again, if r > m, then
br = bm+(rm) = bm brm = 0.
Therefore, we find that

#
$
n + m n+mr r
b = 0, r = 0, 1, 2, . . . , n + m.
(1)
a
r
r

This implies that (a b)n+m = 0, i.e., a b is nilpotent, so a b I. Let r R. Then (ra)n = rn an = rn 0 = 0.

Because R is commutative, (ar)n = (ra)n = 0. Thus, ar, ra I. Consequently, I is an ideal of R. Because every
element of I is nilpotent, I is nil.
(ii) Let a+I be a nilpotent element of R/I. Then (a+I)n = I for some positive integer n. But an +I = (a+I)n .
Thus, an + I = I, which implies that an I. Because every element of I is nilpotent, there exists a positive
integer m such that (an )m = 0, i.e., anm = 0, which shows that a is nilpotent, so a I. This implies a + I = I.
Hence, R/I has no nonzero nilpotent elements.
Theorem 8.2.28 Let A and B be two nil ideals of a commutative ring R with 1. Then A + B is a nil ideal.
Proof. By Theorem 8.2.20, we know that A + B is an ideal of R. Let I be the set of all nilpotent elements of
R. Then A I, B I and by Theorem 8.2.27, I is an ideal. Hence, A + B I. Because I is nil, A + B is nil.

Worked-Out Exercises
Exercise 1 Find all ideals of Z.
Solution: From Worked-Out Exercise 3 (page 147), we know that the subrings of Z are the subsets nZ, n = 0, 1, 2, . . . .
Let us now show that these subrings are precisely the ideals of Z. If I is an ideal of Z, then I is a subring
of Z, so I = nZ for some nonnegative integer n. Now, let I = nZ (n is a nonnegative integer). Then I is a
subring. If r Z, then rI = r(nZ) =n(rZ) nZ = I. Similarly, Ir I. Hence, I is an ideal of Z.
Exercise 2 Let R be a ring such that R has no zero divisors. Show that if every subring of R is an ideal of R, then R
is commutative.
Solution: Let 0 6= a R. Then C(a) = {x R | xa = ax} is a subring of R and hence an ideal of R. Thus, ra C(a)
for all r R. Let r R. Now ara = ra2 implies that (ar ra)a = 0. Because R has no zero divisors and
a 6= 0, ar ra = 0, so ar = ra. Hence, a is in the center of R. Because a is arbitrary, R is commutative.
S
Exercise 3 Give an example of a ring R and ideals Ai , i I, such that Ai Aj = {0} if i 6= j, but Ai ( j6=i Aj ) 6= {0}.
Solution: Let R = {0, a, b, c}. Define + and on R by

2a = 2b = 2c = 0, xy = 0, for all x, y R and

a + b = b + a = c, a + c = c + a = b, and b + c = c + b = a.
Then (R, +, ) is a ring. Let A1 = {0, a}, A2 = {0, b}, and A3 = {0, c}. Then A1 + A2 = A1 + A3 =
A2 + A3 = R and A1 A2 = A1 A3 = A2 A3 = {0}.
Exercise 4 Give an example of a ring R and ideals A and B such that AB A B.
Solution: Let R be the ring of Worked-Out Exercise 3. Let A = B = {0, a}. Then AB = {0} {0, a} = A B.
Exercise 5 Characterize all commutative rings R such that R has only two ideals R and {0}.

8.2. Ideals and Quotient Rings

157

Solution: Let R be a commutative ring such that the only ideals of R are R and {0}. Now R2 is an ideal of R. Thus,
R2 = {0} or R2 = R.
Case 1. R2 = {0}. Then ab = 0 for all a, b R. In this case, every subgroup of (R, +) is an ideal. Hence,
(R, +) has no nontrivial subgroups, so (R, +) is a cyclic group of prime order by Exercise 22 (page 94).
Case 2. R2 = R. Let 0 6= a R. Then aR is an ideal of R. Hence, either aR = {0} or aR = R. Suppose
aR = {0}. Let T = hai . Then T is an ideal of R and a T. Thus, T = R. Now aR = {0} implies that
T R = {0} and hence R2 = {0}, which is a contradiction. Therefore, aR = R. Thus, for all 0 6= a R,
aR = R. We now show that R has no zero divisors. Let a, b be two nonzero elements of R such that ab = 0.
Let T = {c R | ac = 0}. It is easy to see that T is a nonzero ideal of R. Hence, by the hypothesis, T = R.
This implies that R = aR = aT = {0}, a contradiction to the fact that R = R2 6= {0}. Consequently, R
has no zero divisors. Next, for 0 6= a R, aR = R, so we find that ae = a for some e R. Because a 6= 0,
we must have e 6= 0. Also, because R has no zero divisors, a(e2 e) = 0 implies that e2 = e. Now for any
b R, eb = e2 b implies that e(b eb) = 0 and hence b = eb = be. This shows that e is the identity element
of R. Also, aR = R implies that e = ab for some b R. Hence, a1 exists in R. Consequently, R is a field.
So from the above two cases we conclude that either R is the zero ring with a prime number of elements
or R is a field.

Exercises



b
| a, b, c Z be the ring of all upper triangular matrices over Z.
c



0 b
(a) Prove that I =
| b, c Z is an ideal of T2 (Z). Find the quotient ring T2 (Z)/I.
0 c



0 a
(b) Prove that I =
| a Z is an ideal of T2 (Z). Find the quotient ring T2 (Z)/I.
0 0

1. Let T2 (Z) =



a
0

2. In the ring Z24 , show that I = {[0], [8], [16]} is an ideal. Find all elements of the quotient ring Z24 /I.

3. Show that the set I = {a + bi 5 | a, b Z and a b is even} is an ideal of the ring Z[i 5].
4. Let R be a ring and a R. Show that aR is a right ideal of R and Ra is a left ideal of R.
5. Let R be a ring. Let A be a left ideal of R and B be a right ideal of R. Show that AB is an ideal of R and
BA A B.

6. Let R be a ring such that R2 6= {0}. Prove that R is a division ring if and only if R has no nontrivial left
ideals.
7. Let R be a ring with 1. Prove that R has no nontrivial left ideals if and only if R has no nontrivial right
ideals.

8. Let I1 , I2 be ideals of a ring R. Prove that I1 I2 is an ideal of R if and only if either I1 I2 or I2 I1 .

9. Let I and J be ideals of a ring R. Prove that I + J is an ideal of R and that I + J = hI Ji , the ideal of
R generated by I J.
10. Let I be an ideal of a commutative ring R and a R. Prove that hI {a}i = {i + ra + na | i I, r R,
n Z}.
11. Let m and n be positive integers in Z. Prove that
(a) hm, ni = hmi + hni = hdi , where d is the greatest common divisor of m and n;
(b) hmi hni = hqi , where q is the least common multiple of m and n.
12. Find all ideals of the Cartesian product F1 F2 of two fields F1 and F2 .
13. Consider the Cartesian product ring R1 R2 of the rings R1 and R2 .
(a) If I1 is an ideal of R1 and I2 is an ideal of R2 , prove that I1 I2 is an ideal of R1 R2 .
(b) Suppose R1 and R2 are with 1 and I is an ideal of R1 R2 . Does there exist ideals I1 of R1 and I2
of R2 such that I = I1 I2 ?
14. Let R be an ideal of a ring R. Prove that the quotient ring R/I is a commutative ring if and only if
ab ba I for all a, b R.

158

8. Subrings, Ideals, and Homomorphisms

15. Let T = { ab | ab Q, a and b are relatively prime and 5 does not divide b}. Show that T is a ring under
the usual addition and multiplication. Also, prove that I = { ab T | 5 divides a} is an ideal of T and the
quotient ring T /I is a field.
16. Let I be an ideal of a ring R. Prove that if R is a commutative ring with identity, then R/I is a commutative
ring with identity. If R has no zero divisors, is the same necessarily true for R/I?
17. Let I be an ideal of a commutative ring R. Define the annihilator of I to be the set

Prove that annI is an ideal of R.

annI = {r R | ra = 0 for all a I}.

18. In the ring Z20 , prove that I = {[n] | n is even} is an ideal. Find annI.

19. In the ring Z[i], show that I = {a + bi | a, b Z and a, b are even} is an ideal. Find annI.

20. In a commutative regular ring R with 1, prove that every principal ideal I is generated by an idempotent
and for every principal ideal I, there exists a principal ideal J such that R = I + J and I J = {0}.

21. Prove that every ideal of a regular ring is regular.

22. Prove that a ring R is regular if and only if every principal left ideal of R is generated by an idempotent.
23. Prove that in a commutative regular ring with 1 every finitely generated ideal is a principal ideal.
24. In a ring R, prove that {0} is the only nilpotent ideal if and only if for all ideals A and B of R, AB = {0}
implies A B = {0}.

25. Let R be a ring and f : R [0, 1] be such that

f (a b)
f(rb)

min{f(a), f (b)},
f (b)

for all a, b, r R. Prove the following:

(i) f (0) f (a) for all a R;
(ii) f (a) = f (a) for all a R;
(iii) for all t I(f), Rt = {x R | f(x) t} is a left ideal of R;
(iv) R0 = {a R | f (a) = f (0)} is a left ideal of R.

26. Let R be a ring. A relation on R is called a congruence relation on the ring R if is an equivalence
relation on R and for all a, b, c R, ab implies that acbc, cacb, and (a + c)(b + c). Let I be an ideal of
R and be the relation on R defined by ab if and only if a b I. Show that is a congruence relation
on R.
27. In each of the following exercises, write the proof if the statement is true; otherwise, give a counterexample.
(i) If {Ii | i N} is a collection of ideals of R, then iN Ii is an ideal of R.
(ii) Z is a subring of R, but not an ideal of R.
(iii) If I is a nontrivial ideal of an integral domain R, then the quotient ring R/I is an integral domain.

8.3

Homomorphisms and Isomorphisms

In this section, we introduce the ideas of homomorphisms and isomorphisms of rings. These concepts are the
analogs of homomorphisms and isomorphisms for groups.
Definition 8.3.1 Let (R, +, ) and (R0 , +0 , 0 ) be rings and f a function from R into R0 . Then f is called a
homomorphism of R into R0 if
f(a + b) = f(a) +0 f(b),
for all a, b R.

f (a b) = f (a) 0 f (b)

A homomorphism f of a ring R into a ring R0 is called

(i) a monomorphism if f is one-one,
(ii) an epimorphism if f is onto R0 , and
(iii) an isomorphism if f is one-one and maps R onto R0 .
If f is an isomorphism of a ring R onto a ring R0 , then f 1 is an isomorphism of R0 onto R.
An isomorphism of a ring R onto R is called an automorphism.

8.3. Homomorphisms and Isomorphisms

159

Definition 8.3.2 Two rings R and R0 are said to be isomorphic if there exists an isomorphism of R onto R0 .
We write R ' R0 when R and R0 are isomorphic.
When speaking of two rings R and R0 , from now on we usually use the operations + and for both rings.
Let f : R R0 be a homomorphism of rings. Because f preserves +, f is a also a homomorphism of the
groups (R, +) and (R0 , +). Hence, we can immediately apply Theorem 5.1.2 to conclude that f maps 0 to 00 ,
i.e., f (0) = 00 , and for all a R, f (a) = f(a). We list some properties of homomorphisms in the following
theorem. The proofs are similar to the proof of Theorem 5.1.2, so we leave them as an exercise for the reader.
Theorem 8.3.3 Let f be a homomorphism of a ring R into a ring R0 . Then the following assertions hold.
(i) f (0) = 00 , where 00 is the zero of R0 .
(ii) f (a) = f (a) for all a R.
(iii) f (R) = {f(a) | a R} is a subring of R0 .
(iv) If R is commutative, then f (R) is commutative.
Suppose R has an identity and f (R) = R0 , i.e., f is onto R0 . Then
(v) R0 has an identity, namely, f (1).
(vi) If a R is a unit, then f(a) is a unit in R0 and
f (a)1 = f (a1 ).
We point out that in (v) of Theorem 8.3.3, if f is not onto, then R0 may or may not have an identity. Even
if R0 has an identity, the identity of R need not map onto the identity of R0 . We illustrate this point later in
Example 8.3.7.
Definition 8.3.4 Let f be a homomorphism of a ring R into a ring R0 . Then the kernel of f, written Ker f, is
defined to be the set
Ker f = {a R | f(a) = 00 },

where 00 denote the additive identity of R0 .

From Theorem 8.3.3, we know that 0 Ker f. Hence, Ker f 6= .

Example 8.3.5 (i) The identity map of a ring R is a homomorphism (in fact, an isomorphism). Its kernel is
{0}.
(ii) Let R and R0 be rings and f : R R0 be defined by f (a) = 00 for all a R. Then f is a homomorphism
of R into R0 and Ker f = R.
Example 8.3.6 Let f be the mapping from Z onto Zn defined by f (a) = [a] for all a Z. From Example 5.1.4,
f (a + b) = f (a) +n f (b) for all a, b Z. Also, f(a b) = [ab] = [a] n [b] = f (a) n f(b) for all a, b Z. Thus, f is
a homomorphism of Z onto Zn . As in Example 5.1.4, Ker f = {qn | q Z}.
In the following example, we show that if f is a homomorphism from a ring R with 1 into a ring R0 with 1
and f is not onto, then the identity of R need not map onto the identity of R0 .
Example 8.3.7 Consider the direct sum Z Z of Z with itself (see Exercise 17, page 140). Define f : Z Z Z
by f (a) = (a, 0) for all a Z. From the definition of f, f is well defined. Now for all a, b Z,
f (a + b) = (a + b, 0) = (a, 0) + (b, 0) = f (a) + f (b)
and
f (ab) = (ab, 0) = (a, 0)(b, 0) = f (a)(b).
Thus, f is a homomorphism. Also, Ker f = {0}. Now f (1) = (1, 0), but (1, 1) is the identity of ZZ. Therefore,
the identity of Z does not map onto the identity of Z Z.
Consider the rings Z and Q. Suppose Z ' Q. Then the groups (Z, +) and (Q, +) are isomorphic. However,
this is not possible because (Z,+) is a cyclic group and (Q,+) is not a cyclic group. In the following example,
we give another argument to show that Z is not isomorphic to Q.

160

8. Subrings, Ideals, and Homomorphisms

Example 8.3.8 Suppose Z ' Q. Let f : ZQ be an isomorphism. Then f (1) = 1 and f(0) = 0. Let n be a
positive integer. Then
f (n) = f (1 + + 1) = f (1) + f (1) + + f (1) = nf (1) = n1 = n.
~}


n tim es

Now suppose that n is a negative integer. Let n = m, where m is positive. Then f (n) = f(m) = f(1 1
1) = f (1) f (1) f(1) = m(f (1)) = mf (1) = m1 = m = n. Hence, f(n) = n for all n Z.
Let 0 6= ab Q\Z. Because f is onto Q, there exists n Z such that ab = f(n) = n, which is a contradiction.
Hence, Q is not isomorphic to Z.
In the following example, we consider two rings which look similar, but which are not isomorphic.

Example
8.3.9 In this example, we show that the ring Z[ 3] = {a + b 3 | a, b Z} and the ring Z[ 5] =
{a + b 5 | a, b Z} are not isomorphic.

Suppose there exists an isomorphism f : Z[ 3] Z[ 5]. Now 3 = (0 + 3)2 . Thus,

f(3) = f(( 3)2 ) = (f ( 3))2 .

Because f is an isomorphism, we have f(1) = 1. This implies that f (3) = 3. Hence,

3 = f(3) = (f ( 3))2 .

Because f ( 3) Z[ 5], f ( 3) = a + b 5 for some a + b 5 Z[ 5]. Therefore,

3 = (a + b 5)2 .
This implies that

3 = a2 + 5b2 + 2ab 5.

If ab = 0, then 3 = a2 + 5b2 . But there do not exist integers a and b such that ab = 0 and 3 = a2 + 5b2 .

2 5b2
If ab 6= 0, then 5 = 3a2ab
Q, which is a contradiction. Hence, Z[ 3] and Z[ 5] are not isomorphic.
The next example shows that the ring Zn and the ring Z/ hni are isomorphic.
Example 8.3.10 Consider the ideal hni generated by a fixed positive integer n Z. By Corollary 8.2.11, hni =
{qn | q Z}. The cosets of hni in Z are a + hni = {a + qn | q Z}, where a Z. Now
Z/ hni = {a + hni | a Z}.
Define f : Zn Z/ hni by f ([a]) = a + hni for all [a] Zn . We recall that f is an isomorphism of ( Zn , +n )
onto (Z/ hni , +) (Example 5.1.15). Now
f ([a] n [b]) = f ([ab]) = ab + hni = (a + hni)(b + hni) = f ([a])f ([b]).
Thus, f is a ring isomorphism of Zn onto Z/ hni .
Theorem 8.3.11 Let f be a homomorphism of a ring R into a ring R0 . Then Ker f is an ideal of R.
Proof. Because 0 Ker f, Ker f 6= . Let a, b Ker f. Then f (a b) = f (a) f (b) = 00 00 = 00 , so
a b Ker f. Let r R. Then f (ra) = f (r) f (a) = f (r) 00 = 00 , so ra R. Similarly, ar Ker f. Hence, Ker
f is an ideal of R.
In the remainder of the section, we consider isomorphism theorems which are parallel to those for groups
(Section 5.2).
Theorem 8.3.12 Let R be a ring and I be an ideal of R. Define the mapping g : R R/I by g(a) = a + I for
all a R. Then g is a homomorphism, called the natural homomorphism, of R onto R/I. Furthermore, Ker
g = I.

8.3. Homomorphisms and Isomorphisms

161

Proof. Now for all a, b R,

g(a + b) = (a + b) + I = (a + I) + (b + I) = g(a) + g(b)
and
g(ab) = ab + I = (a + I)(b + I) = g(a)g(b).
That Ker g = I follows from Theorem 5.1.12 in group theory.
Theorem 8.3.13 Let f be a homomorphism of a ring R onto a ring R0 and I be an ideal of R contained in Ker
f. Let g be the natural homomorphism of R onto R/I. Then there exists a unique homomorphism h of R/I onto
R0 such that f = h g. Furthermore, h is one-one if and only if I = Ker f.
Proof. Once again, we use the work already done for groups. Define h : R/I R0 by h(a + I) = f (a) for
all a R. We have the desired results by Theorem 5.2.1, once we verify that h preserves multiplication. Now
h((a + I)(b + I)) = h(ab + I) = f (ab) = f(a)f (b) = h(a + I)h(b + I).

The proof of the following theorem is similar to that of the first isomorphism theorem for groups. We omit
the proof. This theorem is also known as the fundamental theorem of homomorphisms for rings.
Theorem 8.3.14 (First Isomorphism Theorem) Let f be a homomorphism of a ring R into a ring R0 . Then
f (R) is an ideal of R0 and
R/Ker f ' f (R).
We state the following theorem without proof. Its proof is a direct translation of the proof of the corresponding
theorem for groups.
Theorem 8.3.15 (Correspondence Theorem) Let f be a homomorphism of a ring R onto a ring R0 . Then
f induces a one-one inclusion preserving correspondence between the ideals of R containing Ker f and the ideals
of R0 in such a way that if I is an ideal of R containing Ker f, then f (I) is the corresponding ideal of R0 , and
if I 0 is an ideal of R0 , then f 1 (I 0 ) is the corresponding ideal of R.
An example similar to Example 5.2.13 can be developed to illustrate Theorem 8.3.15
The next two isomorphism theorems for rings correspond to Theorems 5.2.8 and 5.2.6, respectively.
Theorem 8.3.16 Let f be a homomorphism of a ring R onto a ring R0 , I be an ideal of R such that I Ker
f, g, and g 0 be the natural homomorphisms of R onto R/I and R0 onto R0 /f (I), respectively. Then there exists
a unique isomorphism h of R/I onto R0 /f(I) such that g 0 f = h g.
Corollary 8.3.17 Let I1 , I2 be ideals of a ring R such that I1 I2 . Then
(R/I1 )/(I2 /I1 ) ' R/I2 .
Theorem 8.3.18 If I and J are ideals of the ring R, then I/(I J) ' (I + J)/J.

Worked-Out Exercises
Exercise 1 Show that the function f : Z6 Z10 defined by f ([a]) = 5[a] for all [a] Z6 is a ring homomorphism of
Z6 into Z10 .
Solution: We first show that f is well defined. Let [a] = [b] in Z6 . Then a b is divisible by 6. Thus, a = 6k + b for
some k Z. Now 5a = 30k + 5b shows that 5[a] = [5a] = [30k + 5b] = [30k] +10 [5b] = [0] +10 5[b] = 5[b] in
Z10 . Therefore, f ([a]) = f ([b]). Thus, we find that f is well defined. Let [a], [b] Z6 . Then f ([a] +6 [b]) =
f ([a + b]) = 5[a + b] = 5([a] +10 [b]) = 5[a] +10 5[b] = f (a) +10 f(b) and f ([a] 6 [b]) = f ([ab]) = 5[ab] =
25[ab] (because Z10 is of characteristic 10) = (5[a]) 10 (5[b]) = f(a) 10 f(b). Hence, f is a homomorphism.
Exercise 2 Let R be the field of real numbers. Let be an automorphism of R. Show that (x) = x for all x R.

162

8. Subrings, Ideals, and Homomorphisms

Solution: Because is an automorphism of R, (0) = 0, and (1) = 1. Let n N. Then (n) = (1 + 1 + + 1) =

(1) + (1) + + (1) = 1 + 1 + + 1 = n. Now let m Z and m < 0. Let n = m > 0. Then
(m) = (n) = (n) = n = m. This shows that (x) = x for all x Z. Let pq Q. Then
( pq ) = (pq 1 ) = (p)(q 1 ) = p(q)1 = pq 1 = pq . This shows that (x) = x for all x Q. Let x R
be such that x 0. Then x = y 2 for some y R. Thus, (x) = (y 2 ) = (yy) = (y)(y) = (y)2 0.
Now let a, b R be such that a b. Then a b 0. Hence, (a b) 0, so (a) (b) 0, i.e.,
(a) (b). Therefore, is order preserving. We now show that is continuous. Let R and > 0.
Because is onto R, there exists > 0 such that () = . Now let x, y R be such that |x y| < .
Thus,
< x y < .
Because is order preserving,

Therefore,
so
This implies that

() < (x y) < ().

< (x y) < ,
< (x) (y) < .
|(x) (y)| < .

Hence, is continuous. Now let x R. Because Q is dense in R, there exists a sequence {an } of rational
numbers such that
lim an = x.
n

Because is continuous,
(x) = ( lim an ) = lim (an ) = lim an = x,
n

proving the result.

Exercise 3 Let R be a ring with 1. If the characteristic of R is 0, then show that R contains a subring isomorphic to
Z.
Solution: Let T = {n1 | n Z}. Because 0 = 01 T, T 6= . Let a = n1 and b = m1 be two elements of T. Then
a b = n1 m1 = (n m)1 and ab = (n1)(m1) = (nm)1. Hence, a b, ab T. Thus, T is a subring of
R. Suppose n, m are two integers such that n1 = m1. If n > m, then (n m)1 = 0. This contradicts the
assumption that R is of characteristic 0. Similarly, m > n also leads to a contradiction. Hence, n = m.
Thus, we find that for each a T, there exists a unique integer n such that a = n1. Hence, the mapping
f : Z T defined by f (n) = n1 is an isomorphism.

Exercise 4 Let p be a prime integer. Show that there are only two nonisomorphic rings of p elements.

Solution: It is known that (Zp , +p ) is the only group of order p (up to isomorphism). Define 1 and 2 on Zp by
[a] 1 [b] = [0] and [a] 2 [b] = [ab] for all [a], [b] Zp . Now 1 and 2 are well defined and (Zp , +p , 1 ) and
(Zp , +p , 2 ) are rings. Let R be a ring with p elements. Then (R, +) ' (Zp , +p ). If R 6' (Zp , +p , 1 ), then
the multiplication of R is not 1 . Let [a] be a generator of (Zp , +p ). Now [a]2 = n[a] for some nonzero integer
n. There exists an integer m such that mn p 1. Let [b] = m[a]. Then [b]2 = m2 [a]2 = m2 n[a] = m[a] =
[b]. Let g be an isomorphism from (Zp , +p ) onto (R, +). Define f : Zp R by f ([u]) = ug([b]) for
all [u] Zp .Then f ([u] +p [v]) = f ([u + v]) = (u + v)g([b]) = ug([b]) + vg([b]) = f([u]) + f ([v]) and
f ([u] 2 [v]) = f ([uv]) = (uv)g([b]) = uvg([b]2 ) = uvg([b])g([b]) = ug([b])vg([b]) = f ([u])f ([v]). Hence, f is
a ring homomorphism. Let c R. Then there exists [u] Zp such that g([u]) = c. Now [u] = t[a] for some
t Z. Thus, f ([tn]) = tng([b]) = tn g(m[a]) = tg(mn[a]) = tg([a]) = g(t[a]) = g([u]) = c. Hence, f is onto
R. Because |Zp | = |R| , it follows that f is one-one. Thus, f is an isomorphism.

Exercises

a
b

b
a

1. Let R denote the set of all 2 2 matrices of the form

, where a and b are real numbers. Prove


a b
that R is a ring and the function a + bi
is an isomorphism of C onto R.
b a

2. Define the binary operations and on Z by a b = a + b 1 and a b = a + b ab for all a, b Z.

Show that (Z,, ) is a ring isomorphic to the ring (Z,+, ).

8.3. Homomorphisms and Isomorphisms

163

3. (i) Show that the rings R and Q are not isomorphic.

(ii) Show that the rings R and C are not isomorphic.
(iii) Are the rings Z6 and Z3 Z2 isomorphic?



a b
| a, b, c Z be the ring of all upper triangular matrices over Z. Define f :
4. Let T2 (Z) =
0 c


a b
T2 (Z) Z by for all
T2 (Z),
0 c


a b
f(
) = a.
0 c
(i) Show that f is a homomorphism.
(ii) Is f an epimorphism?
(iii) Is f an isomorphism?
(iv) Find Ker f.
5. Does there exist an epimorphism from the ring Z24 onto the ring Z7 ?
6. Show that there does not exist a monomorphism from the ring Z6 into the ring Z11 .
7. Show that the ring 2Z is not isomorphic to the ring 3Z.
8. Let R be a Boolean ring. If {0} and R are the only ideals of R, prove that R ' Z2 .

9. Show that the ring Z is not isomorphic to any proper subring of Z.

10. Is the ring Q[ 2] isomorphic to the ring Q[ 3]?

11. Let f : R S be a nontrivial homomorphism from a field R onto a ring S. Prove that S is a field.

12. Let R be a ring with 1. If R is of characteristic n > 0, show that R contains a subring isomorphic to the
ring Zn .
13. Show that there exist only two homomorphisms from R into R.
14. Prove that every ring R is isomorphic to a subring of Mn (R), the ring of n n matrices over R.

15. Let f be a homomorphism of a ring R onto a ring R0 . Prove that

(i) if I is an ideal of R, then f(I) is an ideal of R0 ;
(ii) if I 0 is an ideal of R0 , then f 1 (I 0 ) is an ideal of R and f 1 (I 0 ) Ker f ;
(iii) if R is commutative and I and J are two ideals of R, then f(I +J) = f (I)+f (J) and f (IJ) = f (I)f (J).

16. In each of the following exercises, write the proof if the statement is true; otherwise, give a counterexample.
(i) There exist only two homomorphisms from the ring of integers into itself.
(ii) The mapping f : Z Z defined by f (n) = 3n is a group homomorphism, but not a ring homomorphism.
(iii) The only isomorphism of a ring R onto itself is the identity mapping of R.
(iv) Let R be a ring with 1. Let f : R S be a ring homomorphism. Then f (1) is the identity element of
S.
(v) A nonzero homomorphism from a field into a ring with more than one element is a monomorphism.
(vi) Every nontrivial homomorphic image of an integral domain is an integral domain.

164

8. Subrings, Ideals, and Homomorphisms

Chapter 9

Ring Embeddings
9.1

Embedding of Rings

Sometimes it is worthwhile to study the properties of a ring by considering it as a subring of some ring with
more ring properties than itself. A ring without identity lacks important arithmetic properties, in particular, a
fundamental theorem of arithmetic. As another example, in the ring E of even integers, we cannot say that 2
divides 2 because 1
/ E. Now E is a subring of Z and 1 Z. In Z, it is true that 2 divides 2. The main aim of
this section is to embed a ring into a suitable ring with additional properties. The main feature of this section is
that any integral domain can be embedded in a field. The proof of this result yields a rigorous construction of
the rational numbers from the integers.
Definition 9.1.1 A ring R is said to be embedded in a ring S if there exists a monomorphism of R into S.
From the above definition, it follows that a ring R can be embedded in a ring S if there exists a subring T
of S such that R ' T.
In the next theorem, we show that any ring R can be embedded in a ring with identity.
Theorem 9.1.2 Any ring R can be embedded in a ring S with 1 such that R is an ideal of S. If R is commutative,
then S is commutative.
Proof. Set S = R Z. Define addition and multiplication as follows:
(a, m) + (b, n)
(a, m) (b, n)

=
=

(a + b, m + n),
(ab + na + mb, mn)

for all a, b R and m, n Z. (Here na means a adds to itself n times if n is positive, a adds to itself |n| times
if n is negative, and 0a = 0.) Then S forms a ring under these definitions of addition and multiplication, a fact
we ask the reader to prove in the exercises. We do note that (0, 0) is the additive identity and that (0, 1) is the
multiplicative identity of S.
Consider the subset R {0} of S. Because (0, 0) R {0}, R {0} 6= . Also, for all (a, 0), (b, 0) R {0},
(a, 0) (b, 0) = (a b, 0) R {0},
and
(a, 0) (b, 0) = (ab, 0) R {0}.
Thus, R {0} is a subring of S. Now for all (a, 0) R {0} and (c, n) S,
(a, 0) (c, n) = (ac + na, 0) R {0}
and
(c, n) (a, 0) = (ca + na, 0) R {0}.
This proves that R {0} is an ideal of S.
Now define f : R R {0} by f (a) = (a, 0) for all a R. Then f is an isomorphism of R onto R {0}, so
R ' R {0}. Therefore, R can be embedded in S. By identifying a R with (a, 0) R {0}, we can regard

165

166

9. Ring Embeddings

R to be an ideal of S. To show that the commutativity of R implies that of S, let (a, m), (b, n) S and R be
commutative. Then
(a, m) (b, n)

=
=
=

(ab + na + mb, mn)

(ba + mb + na, nm)
(b, n) (a, m).

(because R iscommutative, ab = ba)

Hence, S is commutative.
Our main objective in this section is to embed a ring in a field. By Theorem 9.1.2, every ring can be embedded
in a ring with identity. If S were a field, then S is commutative and has no zero divisors. This in turn implies
that R is commutative and has no zero divisors. Thus, if we were to embed a ring R in a field S, then R must
have at least these two properties, i.e., R must be commutative and have no zero divisors. In the next theorem,
we embed a commutative ring with no zero divisors into an integral domain and then we will embed an integral
domain in a field.
Theorem 9.1.3 Let R be a commutative ring with no zero divisors. Then R can be embedded in an integral
domain.
Proof. Let S be the ring as defined in Theorem 9.1.2. Let A be the annihilator of R in S. Then A is an
ideal of S by Exercise 17 (page 158). If R A = {0}, then the natural homomorphism of R onto the quotient
ring S/A must map R one-one into S/A, i.e., R can be embedded in S/A. We now show that R A = {0} and
that S/A is an integral domain. Let a R A. Then ar = 0 for all r R. Because R has no zero divisors, a = 0.
Therefore, R A = {0}. Let b + A, c + A S/A. If (b + A)( c + A) = 0 + A, then bc A. Thus, (bc)r = 0 for all
r R. Suppose c + A 6= 0 + A, i.e., c
/ A. Then there exists r R such that cr 6= 0. Because R is an ideal of S,
cr R, and for all s R, bs R. Now
(cr)(bs) = (bcr)s = 0s = 0.
Also, R has no zero divisors and cr 6= 0. Therefore, we must have bs = 0. This implies that b A, so b+A = 0+A.
Hence, S/A is an integral domain.
Suppose we are given the ring of integers Z and we are asked to construct the rational numbers from Z. We
can think of any integer as n/1, i.e., n divided by 1. However, we must somehow pick up the fractions which
cannot be reduced to having a 1 for a denominator. One idea that suggests itself is to consider the Cartesian
product Z Z and consider the first component of the elements of Z Z as the numerator and the second
component as the denominator. However, the ordered pairs (3, 2) and (6, 4) are distinct. A common technique
used in mathematics suggests putting these elements in the same equivalence class so that they become equal.
This is precisely what we shall do. Lets also remember not to have 0 in the denominator.
Theorem 9.1.4 Any integral domain R can be embedded in a field.
Proof. Let S = R (R\{0}). Define the relation on S by for all (a, b), (c, d) S, (a, b) (c, d) if and only
if ad = bc. Then is an equivalence relation. The reflexive and symmetric properties are immediate. Suppose
that (a, b) (c, d) and (c, d) (e, f ). Then ad = bc and cf = de. This implies that adf = bcf and bcf = bde, so
adf = bde. Canceling d, we obtain af = be, i.e., (a, b) (e, f ). Hence, is transitive. Now partitions S into
equivalence classes. Denote the equivalence class {(c, d) S | (c, d) (a, b)} by a/b. Set
F = {a/b | (a, b) S}.
Define + and on F as follows:

a/b + c/d
a/b c/d

=
=

(ad + bc)/bd,
ac/bd

for all a/b, c/d F. We show that + is well defined. Let a/b, c/d, a0 /b0 , c0 /d0 F. Suppose a/b = a0 /b0 and
c/d = c0 /d0 . Then ab0 = ba0 and cd0 = dc0 . Therefore, ab0 dd0 = ba0 dd0 and cd0 bb0 = dc0 bb0 . Hence,
ab0 dd0 + cd0 bb0 = ba0 dd0 + dc0 bb0 .
This implies that
(ad + bc)b0 d0 = bd(a0 d0 + b0 c0 ).
Thus,
(ad + bc, bd) (a0 d0 + b0 c0 , b0 d0 ).

9.1. Embedding of Rings

167

This implies that

(ad + bc)/bd = (a0 d0 + b0 c0 )/b0 d0 .
A similar proof shows that is well defined.
The reader is asked to verify the associative, commutative, and distributive laws for F. The additive identity
of F is 0/b and the multiplicative identity of F is b/b, where b 6= 0. For a/b F, the additive inverse is
(a)/b = a/(b)
and the multiplicative inverse is b/a (when a 6= 0). Thus, F is a field.
We now show that R can be embedded in F. Let
R0 = {a/1 | a R} F.
Then R0 is a subring of F. Define f : R R0 by f (a) = a/1 for all a R. Then a = b if and only if a 1 = 1 b
if and only if a/1 = b/1 if and only if f (a) = f (b). Hence, f is a one-one function. Now
f(a + b) = (a + b)/1 = (a 1 + 1 b)/1 1 = a/1 + b/1 = f (a) + f (b)
and
f (ab) = ab/1 = ab/1 1 = a/1 b/1 = f (a) f (b).

From the definition of f, f is onto R0 . Thus, f is an isomorphism of R onto R0 F.

The above theorem gives another instance of the power of the concept of an equivalence relation. We have
once again used the notion of an ordered pair in a fundamental manner.
Definition 9.1.5 Let R be an integral domain. A field F is called a quotient field of R or a field of quotients
of R if there exists a subring R1 of F such that
(i) R ' R1 and
(ii) for all x F, there exists a, b R1 with b =
6 0 such that x = ab1 .
Let us now show that for the given integral domain R, the field constructed in Theorem 9.1.4 is a quotient
field of R. Let x F. Then x = a/b, where (a, b) S. Now (a, 1) S and (b, 1) S. Thus, a/1, b/1 R0 and
a/b = a/1 1/b = (a/1) (b/1)1 .
Hence, F is a quotient field of R. We call F the quotient field or the field of quotients or R.
Theorem 9.1.6 Let R be an integral domain and F its field of quotients. Let R0 be an integral domain contained
in a field K 0 and set
F 0 = {a0 (b0 )1 | a0 , b0 R0 , b0 6= 0}.
Then F 0 is the smallest subfield of K 0 which contains R0 and any isomorphism of R onto R0 has a unique
extension to an isomorphism of F onto F 0 .
Proof. By Exercise 2 (page 168), F 0 is the smallest subfield of K 0 which contains R0 . Let f be an isomorphism
of R onto R0 . Let a/b F. If f (a) = a0 and f(b) = b0 , define g : F F 0 by
g(a/b) = a0 (b0 )1 = f(a)f (b)1 .
Identifying the ring R with the set {a/1 | a R}, it is clear that f = g |R . Now a/b = c/d if and only if ad = bc
if and only if f(ad) = f(bc) if and only if f (a)f (d) = f (b)f (c) if and only if f (a)f (b)1 = f (c)f (d)1 if and only
if g(a/b) = g(c/d). Therefore, g is a one-one function. From the definition of g, it follows that g is onto F 0 . Now
g(a/b + c/d)

=
=
=
=
=

g((ad + bc)/bd)
f(ad + bc)(f(bd))1
[f (a)f (d) + f(b)f(c)][f(b)1 f (d)1 ]
f(a)f (b)1 + f (c)f (d)1
g(a/b) + g(c/d)

and
g(a/b c/d)

=
=
=
=
=

g(ac/bd)
f (ac)(f (bd))1
[f (a)f (c)][f (b)1 f(d)1 ]
f (a)f (b)1 f (c)f(d)1
g(a/b)g(c/d)

168

9. Ring Embeddings

for all a/b, c/d F. Thus, g is an isomorphism of F onto F 0 .

Let g 0 be any other isomorphism of F onto F 0 such that f = g 0 |R . Then
g0 (a/b)

=
=
=
=
=

g0 (a/1 (b/1)1 )
g0 (a/1)g0 ((b/1)1 )
g0 (a/1)g0 (b/1)1
f (a)f(b)1
g(a/b)

for all a/b F, so g0 = g. Thus, there is a unique extension of f.

We can conclude from this result that the field of quotients F of an integral domain R is the smallest field
containing R in the sense that there does not exist a field K such that R K F.
The field F 0 in Theorem 9.1.6 is called the quotient field of R0 in K. In view of Theorem 9.1.6 and the
comments preceding it, we do not dierentiate between the notation a/b and ab1 for the elements of F.

Worked-Out Exercises
Exercise 1 Let D = { ab Q | 5 does not divide b}. Show that D is a subring of Q with 1. Find the quotient field of D.
Solution: Let a/b, c/d D. Because 5 does not divide b and 5 does not divide d, 5 does not divide bd. Thus,
(ad bc)/bd D and ac/bd D. Hence, D is a subring of Q. Also, 1 = 1/1 D. Because Z D Q and
Q is the quotient field of Z, Q is the quotient field of D.
Exercise 2 Let S be a ring and f a one-one function of S onto a set T. Show that suitable addition and multiplication
can be defined on T so that T becomes a ring isomorphic to S under f.
Solution: Define binary operations + and on T as follows: Let t1 , t2 T. Because f maps S onto T, there exist
s1 , s2 S such that f (s1 ) = t1 and f (s2 ) = t2 . Define
t1 + t2
t1 t2

=
=

f (s1 + s2 ) and
f (s1 s2 ).

First we show that both these binary operations are well defined. Let t1 , t2 , t3 , t4 T be such that t1 = t3
and t2 = t4 . Because f maps S onto T, there exist s1 , s2 , s3 , s4 S such that f (s1 ) = t1 , f (s2 ) = t2 ,
f (s3 ) = t3 , and f (s4 ) = t4 . Therefore, f (s1 ) = f (s3 ) and f (s2 ) = f (s4 ). Because f is one-one, s1 = s3 and
s2 = s4 . Hence, t1 + t2 = f (s1 + s2 ) = f (s3 + s4 ) = t3 + t4 and t1 t2 = f (s1 s2 ) = f(s3 s4 ) = t3 t4 . Thus, +
and are well defined. It is now a routine verification to show that (T, +, ) is a ring. We verify some of the
properties and leave others as an exercise. First we show that + is associative. Now t2 + t3 = f (s2 + s3 )
and t1 + t2 = f(s1 + s2 ). Thus, t1 + (t2 + t3 ) = f (s1 + (s2 + s3 )) = f ((s1 + s2 ) + s3 ) (because + is associative
for S) = (t1 + t2 ) + t3 . Hence, + is associative for T . Also, f (0) + t1 = f (0 + s1 ) = f (s1 ) = f (s1 + 0) =
t1 + f (0). This implies that f(0) is the additive identity. Similarly, we can verify the other properties of
a ring. It is immediate that f is a homomorphism and because f is one-one and f maps S onto T, S is
isomorphic to T.

Exercises
1. Prove the associative, commutative, and distributive laws in Theorem 9.1.4.
2. Let R be an integral domain, which is a subring of a field F. Let F 0 = {ab1 | a, b R, b 6= 0}. Show that
F 0 is a subfield of F. Furthermore, show that F 0 is the smallest subfield of F which contains R.
3. Let R and R0 be integral domains contained in fields. Set F = {ab1 | a, b R, b 6= 0} and F 0 = {a0 b01 |
a0 , b0 R0 , b0 6= 00 }. Suppose f is an isomorphism of R onto R0 . Prove that f has a unique extension to
an isomorphism of F onto F 0 .
4. Prove that any field R is equal to its field of quotients F in the sense that f(R) = F, where f is the
isomorphism defined in Theorem 9.1.4.
5. Prove that isomorphic integral domains have isomorphic fields of quotients.

6. Find the field of quotients of the integral domains Z[i] and Z[ 2].
7. Let R be a ring of characteristic n > 0 and
R Zn = {(r, [m]) | r R and [m] Zn }.

9.1. Embedding of Rings

169

Define + and on R Zn by
(a, [m]) + (b, [t])
(a, [m]) (b, [t])

=
=

(a + b, [m + t]),
(ab, [mt])

for all a, b R, [m], [t] Zn . Prove that

(i) the above two operations are well defined,
(ii) (R Zn , +, ) is a ring with 1,
(iii) (R Zn , +, ) is of characteristic n,
(iv) there exists a monomorphism from R into (R Zn , +, ).

8. Let S and R0 be disjoint rings with the property that S contains a subring S 0 such that there is an
isomorphism f 0 of S 0 onto R0 . Prove that there is a ring R containing R0 and an isomorphism f of S onto
R such that f 0 = f |S 0 .

170

9. Ring Embeddings

Chapter 10

Direct Sum of Rings

In this chapter, we construct some new rings from a given family {Ri | i I} of rings. For this purpose, we
introduce the complete direct sum, the direct sum, and the subdirect sum of this family. The results developed
in this chapter also help us to obtain structure results of rings.

10.1

Complete Direct Sum and Direct Sum

Let {Ri | i I} be a family of rings indexed by a nonempty set I. The Cartesian product {Ri | i I} of the
sets Ri is the set of all functions f : I {Ri | i I} such that f(i) Ri for all i I. Let f, g {Ri | i I}.
Define f + g, fg by
(f + g)(i) = f (i) + g(i)
(f g)(i) = f (i)g(i)
for all i I. Then f + g, f g {Ri | i I}. It can be easily verified that {Ri | i I} together with the above
two operations is a ring. This ring is called the complete direct sum of the family of rings {Ri | i I} and
is denoted by iI Ri . The zero element of iI Ri is the function 0 : I {Ri | i I} defined by 0(i) = 0i ,
the zero element of Ri , for all i I. The additive inverse of f iI Ri is the function f : I {Ri | i I}
defined by (f)(i) = f (i) Ri for all i I. Let f iI Ri and let f (i) = ai Ri for all i I. Usually f is
identified with the image set {ai | i I}. Using this notation, the above two operations can be defined by
{ai | i I} + {bi | i I}
{ai | i I} {bi | i I}

=
=

{ai + bi | i I}
{ai bi | i I}

for all ai , bi Ri for all i I.

Suppose now that I is a finite set, say, I = {1, 2, . . . , n}. In this case, the complete direct sum is denoted by
iI Ri = R1 R2 Rn and an element {ai | i I} is usually written as an n-tuple (a1 , a2 , . . . , an ) .
Definition 10.1.1 The direct sum of a family of rings {Ri | i I}, denoted by iI Ri , is the set
iI Ri

{{ai | i I} iI Ri | ai 6= 0 for at most finitely many i I}.

Theorem 10.1.2 Let {Ri | i I} be a family of rings. Then

(i) iI Ri is a subring of the complete direct sum of rings iI Ri ;
(ii) for all k I, the function ik : Rk iI Ri defined by
ik (a) = {{ai | i I} | ai = 0 for all i 6= k and ak = a}
for all a Rk , is a monomorphism of rings;
(iii) for all k I, ik (Rk ) is an ideal of iI Ri .
Proof. (i) Let {ai | i I} and {bi | i I} be two elements of iI Ri . Because ai 6= 0 for at most
finitely many i I and bi 6= 0 for at most finitely many i I, it follows that ai bi 6= 0 for at most finitely
many i I and ai bi 6= 0 for at most finitely many i I. Hence, {ai | i I} {bi | i I} iI Ri and
{ai | i I}{bi | i I} iI Ri . Thus, iI Ri is a subring.
(ii) Let a, b Rk . Then ik (a + b) = {{ai | i I} | ai = 0 for all i 6= k and ak = a + b} = {{a0i | i I} | a0k = 0
for all i 6= k and a0i = a}+ {{b0i | i I} | b0i = 0 for all i 6= k and b0k = b} = ik (a) + ik (b). Similarly,

171

172

10. Direct Sum of Rings

ik (ab) = ik (a)ik (b). Thus, ik is a homomorphism. By the definition of ik , we find that ik is one-one. Hence, ik
is a monomorphism.
(iii) Because ik is a monomorphism, ik (Rk ) is a subring of iI Ri . Let {bi | i I} iI Ri and {ai | i I}
ik (Rk ). Because ai = 0 for all i 6= k, bi ai = 0 for all i 6= k. Also, for i = k, bk , ak Rk . Therefore, bk ak Rk . Thus,
{bi | i I}{ai | i I} ik (Rk ), proving that ik (Rk ) is a left ideal. Similarly, {ai | i I}{bi | i I} ik (Rk ).
Hence, ik (Rk ) is an ideal.
By Theorem 10.1.2, we find that Rk is isomorphic to the subring ik (Rk ) of iI Ri . Identifying Rk with
ik (Rk ), we can say that iI Ri contains Rk as an ideal.
Let I = {1, 2, . . . , n} and {Ri | i I} be a finite family of rings. From the definition of direct sum, it follows
that the complete direct sum and the direct sum of this family is the same. Hence, by Theorem 10.1.2, we can
say that the direct sum, R1 R2 Rn , contains each of R1 , R2 , . . . , Rn as an ideal.
We now investigate the conditions under which a ring R is isomorphic to a direct sum of a family of ideals
(considering each ideal as a ring) of R.
Definition 10.1.3 Let I be a finite nonempty set, say,S{1, 2, . . . , n}, and {Ai | i I} be a family of ideals of a
ring R. Then the sum of this finite family, denoted by iI Ai , is the set
[
Ai = {a1 + a2 + + an | ai Ai , i = 1, 2, . . . , n}.
iI

If I is empty, then let us take

iI

Ai = {0}.

If I = {1, 2, . . . , n}, then we also use the notation A1 + A2 + + An to denote the sum
We leave the proof of the following theorem as an exercise.

iI

Ai .

Theorem
S 10.1.4 Let {Ai | i I} be a finite family of ideals of a ring R. Then
(i) iI A
i is an ideal of R,
S
(ii) Ai jI Aj for all i I,
S
(iii) if A is an ideal of R such that Ai A for all i I, then iI Ai A.

Definition 10.1.5 Let {AS

i | i I} be a family of ideals of a ring R, where I is finite or infinite. Then the sum
of this family, denoted by iI Ai , is the set
[
[
Ai = {a R | a
Ai for some finite subset I0 of I}.
iI

iI0

S
Theorem 10.1.6 Let {Ai | i I} be a family of ideals of a ring R. Then iI Ai is an ideal of R.
S
S
S
S
R. Then a
Proof.
Because 0
iI Ai ,
iI Ai 6= . Let a, b
iI Ai and r S
iI1 Ai and
S
bS iI2 Ai for some finite subsets I1 and I2 ofS
I. Let I3 = I1 I2 . Then a, b S iI3 Ai . By
Theorem
10.1.4,
S
iI3 Ai is an ideal of R. Hence, a b, ar, ra
iI3 Ai . Thus, a b, ar, ra
iI Ai , so
iI Ai is an ideal
of R.
Definition 10.1.7 Let {Ai | i I} be a finite family of ideals of a ring R. A sum
called a direct sum if for all k I,
[
Ai = {0}.
Ak

iI

Ai of {Ai | i I} is

iI, i6=k

Lemma 10.1.8 Let {Ai | i I} be a finite family of ideals of a ring R. If

a Ak , b Al , k 6= l, ab = 0.

iI

Ai is a direct sum, then for all

Proof.
Let a Ak , S
b Al , and k 6= l. Because Ak andSAl are ideals, ab AS
k and ab Al . Because
S
Al S iI, i6=k Ai , ab
iI, i6=k Ai . Therefore, ab Ak
iI, i6=k Ai . Because
iI Ai is a direct sum,
Ak iI, i6=k Ai = {0}. Hence, ab = 0.
Theorem 10.1.9 Let {Ai | i I} be a family of ideals of a ring R, I = {1, 2, . . . , n}. Then the following
conditions
S are equivalent.
(i) iI Ai is a direct sum.
(ii) a1 + a2 + + an = 0, ai Ai , i I, implies that ai = 0 for all i I.

10.1. Complete Direct Sum and Direct Sum

(iii) Each element a

iI

173

Ai is uniquely expressible in the form

a = a1 + a2 + + an ,

where ai Ai , i I.
Proof. (i)(ii) Let a1 + a2 + + an = 0, ai Ai , i I. Let k I. Now
ak = a1 + a2 + + ak1 + ak+1 + + an Ak

iI, i6=k

Ai = {0}.

Hence, ak = 0.
(ii)(iii) Let a = a1 + a2 + + an = b1 + b2 + + bn , where ai , bi Ai for all i I. Then (a1 b1 ) +
(a2 b2 ) + + (an bn ) = 0. Hence, by (ii), ai bi = 0 for all i I, i.e., ai = bi for all i I.
S
(iii)(i) Let a Ak iI, i6=k Ai . Then there exist ai Ai , i = 1, 2, . . . , n, such that
a = ak = a1 + a2 + + ak1 + ak+1 + + an .

This implies
a1 + a2 + + ak1 + (ak ) + ak+1 + + an = 0.
Also, 0 + 0 + + 0 = 0. Therefore,
by (iii), ai = S
0 for all i I because 0 is uniquely expressible as a sum of
S
elements of Ai . Thus, Ak iI, i6=k Ai = {0}, so iI Ai is a direct sum.
Definition 10.1.10 A ring R is said to be an internal direct sum of a finite family of ideals {A1 , A2 , . . . , An }
if
(i) R = A1 + A2 + + An and
(ii) A1 + A2 + + An is a direct sum.
Theorem 10.1.11 Let R be a ring and {Ai | i I} be a finite family of ideals of R. If R is an internal direct
sum of {Ai | i I}, then
R ' iI Ai .
Proof. Let I = {1, 2, . . . , n}. Suppose R is an internal direct sum of ideals A1 , A2 , . . . , An . Let a R. Then
a is uniquely expressible in the form a = a1 + a2 + + an , where ai Ai , i I. Now (a1 , a2 , . . . , an ) iI Ai .
Define f : R iI Ai by
f (a) = (a1 , a2 , . . . , an ).
Let a, b R. Then there exist ai , bi Ai , i I such that a = a1 + a2 + + an and b = b1 + b2 + + bn .
Now a = b if and only if a1 + a2 + + an = b1 + b2 + + bn if and only if ai = bi for all i I if and only if
(a1 , a2 , . . . , an ) = (b1 , b2 , . . . , bn ) if and only if f (a) S
= f (b). This shows that f is a one-one function. Let (a1 , a2 ,
. . . , an ) iI Ai . Then a = a1 + a2 + + an iI Ai = R and f (a) = (a1 , a2 , . . . , an ). Hence, f is onto
iI Ai . Finally, we show that f is a homomorphism. Because a + b = (a1 + b1 ) + (a2 + b2 ) + + (an + bn ),
we have f (a + b) = ((a1 + b1 ), (a2 + b2 ), . . . , (an + bn )) = (a1 , a2 , . . . , an ) + (b1 , b2 , . . . , bn ) = f (a) + f (b). By
Lemma 10.1.8, for all i, j I, i 6= j, ai bj = 0. From this, it follows that ab = a1 b1 + a2 b2 + + an bn . Thus,
f (ab) = (a1 b1 , a2 b2 , . . . , an bn ) = (a1 , a2 , . . . , an )(b1 , b2 , . . . , bn ) = f (a)f(b). Hence, f is an isomorphism of R
onto iI Ai , proving that R ' iI Ai .
If R is an internal direct sum of ideals A1 , A2 , . . . , An , then we identify R with iI Ai and we usually write
R = A1 A2 An .
Let us now characterize the direct sum of ideals of a ring R with 1 with the help of idempotent elements.
Theorem 10.1.12 Let R be a ring with 1 and {A1 , A2 , . . . , An } be a finite family of ideals of R. Then R =
A1 A2 An if and only if there exist idempotents ei Ai , i = 1, 2, . . . , n, such that
(i) 1 = e1 + e2 + + en ,
(ii) Rei = Ai for all i = 1, 2, . . . , n, and
(iii) ei ej = ej ei = 0 for i 6= j.

174

10. Direct Sum of Rings

Proof. Let R = A1 A2 An . Now 1 R. Thus, there exist ei Ai , i = 1, 2, . . . , n, such that

1 = e1 + e2 + + en . Then ei = e1 ei + e2 ei + + e2i + + en ei . By Lemma 10.1.8, ej ei = 0 for all j 6= i.
Hence, ei = e2i , i.e., ei is an idempotent for all i = 1, 2, . . . , n. Because ei Ai and Ai is an ideal, Rei Ai .
Let a Ai . Then
a = a1 = ae1 + ae2 + + aen = aei Rei

because by Lemma 10.1.8, aej = 0 for all j 6= i. Thus, Ai Rei . Therefore, we find that Rei = Ai .
Conversely, assume that there exist idempotents ei Ai , i = 1, 2, . . . , n, satisfying the given conditions. Let
a R. Then a = a1 = a(e1 +e2 + +en ) = ae1 +ae2 + +aen Re1 +Re2 + +Ren A1 +A2 + +An . Hence,
R = A1 +A2 + +An . Let us now show that this sum is direct. Let a Ai (A1 +A2 + +Ai1 +Ai+1 + +An ).
Then there exist a1 , a2 , . . . , an R such that ai ei = a = a1 e1 + + ai1 ei1 + ai+1 ei+1 + + an en . Thus,
a = ai ei implies that aei = ai e2i = ai ei = a and a = a1 e1 + + ai1 ei1 + ai+1 ei+1 + + an en implies that
aei = a1 e1 ei + + ai1 ei1 ei + ai+1 ei+1 ei + + an en ei = a0 + + a0 = 0 (because by (iii), ei ej = 0 for
i 6= j). Hence, a = 0, proving that R = A1 A2 An .
Let us now consider another type of subring of the complete direct sum iI Ri of a family of rings {Ri | i I}.
For this, let us note that the mapping k : iI Ri Rk defined by
k ({ai | i I}) = ak
is an epimorphism of the ring iI Ri onto the ring Rk . k is called the kth canonical projection.
Definition 10.1.13 A subring T of iI Ri is called a subdirect sum of the family of rings {Ri | i I} if i |T
(the restriction of i to T ) is an epimorphism of T onto Ri . We denote T by siI Ri .
Theorem 10.1.14 A ring S is isomorphic to a subdirect sum of a family {Ri | i I} of rings if and only if S
contains a family of ideals {Ai | i I} such that iI Ai = {0}.
Proof. Suppose S is isomorphic to a subdirect sum of a family {Ri | i I} of rings. Then there exists a
subring T of iI Ri such that S ' T and T = siI Ri . Let be the isomorphism of S onto T. Then i : S Ri
is an epimorphism. Let Ai = Ker i . Then Ai is an ideal of S. Let a iI Ai . Then (i )(a) = 0 for all i I.
Thus, i ((a)) = 0, i.e., the ith component of (a) is 0 for all i I. Hence, (a) = 0. Because is one-one,
a = 0. This proves that iI Ai = {0}.
Conversely, suppose S contains a family of ideals {Ai | i I} such that iI Ai = {0}. Consider the family
{S/Ai | i I} of quotient rings. Let R = iI S/Ai . Define : S R by
(a) = {a + Ai | i I}
for all a S. Then is a homomorphism. Let a S. Now a Ker if and only if (a) = 0 if and only if
a + Ai = 0 for all i I if and only if a Ai for all i I if and only if a iI Ai if and only if a = 0.
Therefore, Ker = {0}. Thus, is a monomorphism. Let (S) = T. Then T is a subring of R and also i |T is
an epimorphism.

Worked-Out Exercises
Exercise 1 An idempotent e of a ring R is called a central idempotent if e C(R).
Let R be a ring with 1 and e be a central idempotent in R. Show that
(a) 1 e is a central idempotent in R;

(b) eR and (1 e)R are ideals of R;

(c) R = eR (1 e)R.

Solution:

(a) (1 e)(1 e) = 1 e e + e2 = 1 e e + e = 1 e. Also, for all a R, a(1 e) = a ae = a ea =

(1 e)a. Hence, 1 e is a central idempotent.

(b) Now eR is a right ideal of R. Let a R. Then a(eR) = (ae)R = (ea)R (because e C(R)) = e(aR)
eR. Hence, eR is also a left ideal. Thus, eR is an ideal of R. Similarly, (1 e)R is an ideal of R.
(c) Let a R. Then a = ea + a ea = ea + (1 e)a eR + (1 e)R. Hence, R = eR + (1 e)R. Suppose
b eR (1 e)R. Then there exist c, d R such that b = ec = (1 e)d. Hence, eb = e2 c = ec = b
and eb = e(1 e)d = (e e2 )d = (e e)d = 0. Thus, b = 0. As a result, R = eR (1 e)R.

Exercise 2 Let A and B be two ideals of a ring R such that R = A B. Show that R/A ' B and R/B ' A.

10.1. Complete Direct Sum and Direct Sum

175

Solution: Let x R. Then x can be uniquely expressed as x = a + b, where a A and b B. Define f : R B

by f (x) = b. Clearly f is well defined. Let b B. Then b = 0 + b A + B. Hence, f (b) = b, which shows
that f is onto B. Let x, y R. Then there exist a1 , a2 A and b1 , b2 B such that x = a1 + b1 and
y = a2 + b2 . Now x + y = a1 + b1 + a2 + b2 = (a1 + a2 )+ (b1 + b2 ) A + B and xy = (a1 + b1 )(a2 + b2 ) =
a1 a2 + a1 b2 + b1 a2 + b1 b2 . Because a1 b2 , b1 a2 A B and A B = {0}, a1 b2 = 0 and b1 a2 = 0. Therefore,
xy = a1 a2 + b1 b2 A + B. Hence, f (x + y) = b1 + b2 = f (x) + f (y) and f (xy) = b1 b2 = f(x)f (y). Thus,
f is an epimorphism. Therefore, by the first isomorphism theorem (Theorem 8.3.14), R/Kerf ' B. Let
x Ker f. Then f(x) = 0. Because x Ker f R, there exist a A and b B such that x = a + b .
Now f (x) = b and this implies that b = 0. Therefore, x = a A, so Ker f A. On the other hand, let
a A. Then a = a + 0 A + B. Therefore, f (a) = 0, so a Ker f. Thus, A Ker f. Hence, A = Ker f,
so R/A ' B. Similarly, R/B ' A.
Exercise 3 Let R = R1 R2 Rn be the direct of sum of rings R1 , R2 , . . . , Rn and 1 R. Show that an element
a = (a1 , a2 , . . . , an ) R is a unit if and only if ai is a unit in Ri for all i = 1, 2, . . . , n.
Solution: Because 1 R = R1 R2 Rn , 1 = (e1 , e2 , . . . , en ), where ei is the identity of Ri for all i = 1, 2, . . . , n.
Suppose a = (a1 , a2 , . . . , an ) R is a unit. Then there exists b = (b1 , b2 , . . . , bn ) R such that ab = 1 = ba.
Thus, (a1 , a2 , . . . , an )(b1 , b2 , . . . , bn ) = (e1 , e2 , . . . , en ) = (b1 , b2 , . . . , bn )(a1 , a2 , . . . , an ). From this, it
follows that ai bi = ei = bi ai for all i = 1, 2, . . . , n. Hence, ai is a unit in Ri for all i = 1, 2, . . . , n.
Conversely, assume that ai is a unit in Ri for all i = 1, 2, . . . , n. Thus, there exists bi Ri such that
ai bi = ei = bi ai for all i = 1, 2, . . . , n. Let b = (b1 , b2 , . . . , bn ). Then ab = 1 = ba, proving that a is a unit.
Exercise 4 Let R be a direct of sum of rings R1 , R2 , . . . , Rn with identity. Let A be an ideal of R. Show that there
exist ideals Ai in Ri , i = 1, 2, . . . , n, such that A = A1 A2 An .
Solution: For all k, 1 k n, define k : Ri Rk by
k ((a1 , a2 , . . . , an )) = ak
for all (a1 , a2 , . . . , an ) Ri . It can be easily verified that k is an epimorphism. Let k (A) = Ak . Then
Ak is an ideal of Rk . We now show that A = A1 A2 An . Let a = (a1 , a2 , . . . , an ) A. Now
k (a) = ak Ak . Therefore, a A1 A2 An , so A A1 A2 An . Suppose now that
b = (b1 , b2 , . . . , bn ) A1 A2 An . Then bk Ak = k (A). Therefore, there exists an element a =
(a1 , a2 , . . . , ak1 , bk , ak+1 , . . . , an ) A. Now (0, 0, . . . , 0, bk , 0, . . . , 0) = (0, 0, . . . , 1, . . . , 0)(a1 , a2 , . . . , ak1 ,
bk , ak+1 , . . . , an ) A for all k = 1, 2, . . . , n. Hence, (b1 , b2 , . . . , bn ) = (b1 , 0, . . . , 0)+ (0, b2 , . . . , 0)+ +
(0, 0, . . . , bn ) A showing that A1 A2 An A. Thus, A = A1 A2 An .
Exercise 5 Let R be a ring with 1. Suppose that A and B are ideals of R such that R = A + B. Show that
R/(A B) ' R/A R/B.
(This result is known as the Chinese remainder theorem for rings.)
Solution: Define f : R R/A R/B by

f (x) = (x + A, x + B)

for all x R. Let x, y R. Then

f (x + y)

=
=
=
=

((x + y) + A, (x + y) + B)
((x + A) + (y + A), (x + B) + (y + B))
(x + A, x + B) + (y + A, y + B)
f(x) + f(y).

Similarly, f (xy) = f (x)f (y). Hence, f is a homomorphism. Now R = A + B implies that 1 = a + b for
some a A and b B. Thus, a + B = (1 b) + B = (1 + B) + (b + B) = 1 + B because b B. Similarly,
b + A = 1 + A. Let (x + A, y + B) R/A R/B. Now xb + ya R. Therefore,
f (xb + ya)

=
=
=
=
=
=
=

((xb + ya) + A, (xb + ya) + B)

((xb + A) + (ya + A), (xb + B) + (ya + B))
((xb + A)+(0 + A), (0 + B)+(ya + B)) (because a A, b B)
((xb + A) , (ya + B))
((x + A)(b + A), (y + B)(a + B))
((x + A)(1 + A), (y + B)(1 + B))
( x + A , y + B ).

176

10. Direct Sum of Rings

Hence, f is an epimorphism. By the first isomorphism theorem (Theorem 8.3.14),
R/Ker f ' R/A R/B.
We now show that Ker f = A B.
Ker f

=
=
=
=
=
=

{x R | f(x) = 0}
{x R | (x + A, x + B) = (A, B)}
{x R | x + A = A and x + B = B}
{x R | x A and x B }
{x R | x A B }
A B.

Consequently, R/(A B) ' R/A R/B.

Exercises
1. Let R = R1 R2 Rn be a direct sum of rings. If Ai is an ideal of Ri , (1 i n), prove that
A = A1 A2 An is an ideal of R.
2. Let R be a direct of sum of rings R1 , R2 , . . . , Rn with 1. Let A be an ideal of R. Show that there exist
ideals Ai of Ri , i = 1, 2, . . . , n, such that A = A1 A2 An and
R/A ' R1 /A1 R2 /A2 Rn /An .
3. Show that the ring Z cannot be expressed as a direct sum of a finite family of proper ideals of Z.
4. If m and n are two positive integers such that gcd(m, n) = 1, prove that Zmn ' Zm Zn .

Chapter 11

Polynomial Rings
The study of polynomials dates back to 1650 B.C., when Egyptians were solving certain linear polynomial
equations. In 600 B.C., Hindus had learned how to solve quadratic equations. However, polynomials, as we
know them today, i.e., polynomials written in our notation, did not exist until approximately 1700 A.D.
About 400 A.D., the use of symbolic algebra began to appear in India and Arabia. Some mark the use of
symbols in algebra as the first level of abstraction in mathematics.

11.1

Polynomial Rings

An important class of rings is the so-called class of polynomial rings. We are all familiar with polynomials. We
may be used to thinking of a polynomial as an expression of the form a0 + a1 x + +an xn , where x is a symbol
and the ai are possibly real numbers, or as a function f (x) = a0 + a1 x + + an xn . However, does one really
know what a polynomial is? What really is the symbol x? Why are two polynomials a0 + a1 x + +an xn and
b0 + b1 x+ + bm xm equal if and only if n = m and ai = bi , i = 1, 2, . . . , n? In this section, we answer these
questions and give some basic properties of polynomials.
Definition 11.1.1 For any ring R, let R[x] denote the set of all infinite sequences (a0 , a1 , a2 , . . .), where ai R,
i = 0, 1, 2, . . . , and where there is a nonnegative integer n (dependent on (a0 , a1 , a2 , . . .)) such that for all integers
k n, ak = 0. The elements of R[x] are called polynomials over R.
We now define addition and multiplication on R[x] as follows:
(a0 , a1 , a2 , . . .) + (b0 , b1 , b2 , . . .)
(a0 , a1 , a2 , . . .) (b0 , b1 , b2 , . . .)

=
=

(a0 + b0 , a1 + b1 , a2 + b2 , . . .)
(c0 , c1 , c2 , . . .),

where
cj =

j
[

ai bji for j = 0, 1, 2, . . .

i=0

We leave it to the reader to verify that (R[x], +, ) is a ring. We do note that (0, 0, . . .) is the additive
identity of R[x] and that the additive inverse of (a0 , a1 , . . .) is (a0 , a1 , . . .). The ring R[x] is called a ring of
polynomials or a polynomial ring over R. It is clear that R[x] is commutative when R is commutative. Also,
if R has an identity 1, then R[x] has an identity, namely, (1, 0, 0, 0, . . .).
The mapping a (a, 0, 0, . . .) is a monomorphism of R into R[x]. Thus, R is embedded in R[x]. Therefore,
we can consider R as a subring of R[x] and we no longer distinguish between a and (a, 0, 0, . . .).
We now convert our notation of polynomials into a notation which is more familiar to the reader.
Let
a = ax0 denote (a, 0, 0, . . .)
ax = ax1 denote (0, a, 0, . . .)
ax2 denote (0, 0, a, . . .)
..
.
Then

177

178

11. Polynomial Rings

(a0 , a1 , a2 , . . . , an , 0, . . .) = (a0 , 0, 0, . . .) + (0, a1 , 0, 0, . . .) + + (0, . . . , 0, an , 0, . . .) = a0 + a1 x + a2 x2 +

+ an xn .
The symbol x is called an indeterminate over R and the elements a0 , a1 , . . . , an of R are called the
coecients of a0 + a1 x + a2 x2 + + an xn .
The reason two polynomials a0 + a1 x + +an xn and b0 + b1 x+ +bm xm are equal if and only if n = m
and ai = bi , i = 1, 2, . . . , n, is that the two sequences (a0 , a1 , . . .) and (b0 , b1 , . . .) are equal if and only if ai = bi ,
i = 1, 2, . . . . (One must recall that an infinite sequence of elements of R is a function from the set of nonnegative
integers into R. Consequently, the concept of an ordered pair is again being used to give a rigorous definition of
a mathematical concept.)
If R has an identity 1, then we can consider x an element of R[x]. We do this by identifying 1x with x, i.e.,
(0, 1, 0, . . .) is called x.
The reader can check that the definitions of addition and multiplication of two polynomials are the familiar
ones. Thus, when R has an identity, ax = (a, 0, 0, . . .)(0, 1, 0, . . .) = (0, a, 0, . . .) = (0, 1, 0, . . .)(a, 0, 0, . . .) = xa.
Theorem 11.1.2 (i) If R is a commutative ring with 1, then R[x] is a commutative ring with 1.
(ii) If R is an integral domain, then R[x] is also an integral domain.
m
be two elementsS
in R[x]. Let
Proof. (i) Let f (x) = a0 + a1 x+ +an xn and g(x) = b0 + b1 x + +bS
mx
j
t
s
f (x)g(x) = c0 +c1 x+ +ct x and g(x)f (x) = d0 +d1 x+ +ds x . Now cj = i=0 ai bji and dj = ji=0 bi aji .
Because R is commutative, cj = a0 bj + a1 bj1 + + aj b0 = b0 aj + b1 aj1 + + bj a0 = dj for all j = 0, 1, 2,
. . . . Thus, R[x] is a commutative ring. Because 1 R, 1 R[x] and 1f(x) = f (x)1 = f (x) for all f (x) R[x].
Hence, R[x] is a commutative ring with 1.
(ii) Let R be an integral domain. Then by (i), R[x] is a commutative ring with 1. Let f(x) = a0 + a1 x +
+an xn and g(x) = b0 + b1 x+ +bm xm be two nonzero polynomials in R[x]. Then there exist ai and bj such
that ai 6= 0, bj 6= 0, ai+t = 0, and bj+t = 0 for all t 1. Consider the polynomial

f (x)g(x) = c0 + c1 x + + cn+m xn+m .

Now
ci+j = a0 bi+j + a1 bi+j1 + + ai bj + + ai+j b0 = ai bj 6= 0
because R is an integral domain. This implies that f (x)g(x) 6= 0. Thus, R[x] is an integral domain.
Definition 11.1.3 Let R be a ring. If f(x) = a0 + a1 x+ + an xn , an 6= 0, is a polynomial in R[x], then n is
called the degree of f (x), written deg f(x), and an is called the leading coecient of f (x). If R has an identity
and an = 1, then f (x) is called a monic polynomial.
The polynomials of degree 0 in R[x] are exactly those elements from R\{0}. 0 R[x] has no degree. We call
the elements of R scalar or constant polynomials.
Theorem 11.1.4 Let R[x] be a polynomial ring and f (x), g(x) be two nonzero polynomials in R[x].
(i) If f (x)g(x) 6= 0, then deg f (x)g(x) deg f(x) + deg g(x).
(ii) If f (x) + g(x) 6= 0, then
deg(f (x) + g(x)) max{deg f (x), deg g(x)}.
Proof. (i) If f(x) = a0 + a1 x+ +an xn and g(x) = b0 + b1 x+ + bm xm , then
f(x)g(x) = a0 b0 + (a0 b1 + a1 b0 )x + + an bm xn+m .
If f(x)g(x) 6= 0, then at least one of the coecients of f (x)g(x) is nonzero. If an bm 6= 0, then
deg(f (x)g(x)) = n + m = deg f (x) + deg g(x).
If an bm = 0 (which can hold if R has zero divisors), then deg(f (x)g(x)) < deg f (x)+ deg g(x).
(ii) If deg f (x) > deg g(x), then deg(f (x) + g(x)) = max{deg f (x), deg g(x)}. If deg f(x) = deg g(x), then
it is possible that f(x) + g(x) = 0 or deg(f (x) + g(x)) < max{deg f (x), deg g(x)}. We leave the details as an
exercise.
From the proof of Theorem 11.1.4(i), it is immediate that if R is an integral domain, then equality holds in
(i).

11.1. Polynomial Rings

179

Example 11.1.5 Consider the polynomial ring Z6 [x]. Let f (x) = [1] + [2]x2 and g(x) = [1] + [3]x. Then
f(x)g(x) = [1] + [3]x + [2]x2 .
Hence, deg (f (x)g(x)) = 2 < 3 = deg f (x) + deg g(x). Let h(x) = [5] + [4]x2 . Then
f(x) + h(x) = [6] + [6]x2 = [0],
so deg(f (x) + h(x)) is not defined.
Theorem 11.1.6 (Division Algorithm) Let R be a commutative ring with 1 and f(x), g(x) be polynomials
in R[x] with the leading coecient of g(x) a unit in R. Then there exist unique polynomials q(x), r(x) R[x]
such that
f (x) = q(x)g(x) + r(x),
where either r(x) = 0 or deg r(x) < deg g(x).
Proof. If f (x) = 0 or deg f(x) < deg g(x), then we take q(x) = 0 and r(x) = f(x). We now assume that
deg f (x) deg g(x) and prove the result by induction on deg f (x) = n. If deg f(x) = deg g(x) = 0, then we have
q(x) = f (x)g(x)1 and r(x) = 0. Make the induction hypothesis that the theorem is true for all polynomials of
degree less than n. Let f(x) = a0 + a1 x+ +an xn have degree n and g(x) = b0 + b1 x + + bm xm have degree
m, where n m. The polynomial
nm
f1 (x) = f(x) (an b1
g(x)
(11.1)
m )x

has degree less than n because the coecient of xn is an (an b1

m )bm = 0. Hence, by the induction hypothesis,
there exist polynomials q1 (x), r1 (x) R[x] such that
f1 (x) = q1 (x)g(x) + r1 (x),

(11.2)

where r1 (x) = 0 or deg r1 (x) < deg g(x). Substituting the representation of f1 (x) in Eq. (11.2) into Eq. (11.1)
and solving for f (x), we obtain
nm
f (x) = (q1 (x) + an b1
)g(x) + r1 (x) = q(x)g(x) + r(x),
m x
nm
where q(x) = q1 (x) + an b1
and r(x) = r1 (x), the desired representation when f (x) has degree n.
m x
The uniqueness of q(x) and r(x) remains to be shown. Suppose there are polynomials q 0 (x) and r0 (x) R[x]
such that
f (x) = q(x)g(x) + r(x) = q 0 (x)g(x) + r0 (x),

where r(x) = 0 or deg r(x) < deg g(x), r0 (x) = 0 or deg r0 (x) < deg g(x). Then
r(x) r0 (x) = (q 0 (x) q(x))g(x).
Suppose r(x) r0 (x) 6= 0. Because the leading coecient of g(x) is a unit,
deg((q 0 (x) q(x))g(x)) = deg(q 0 (x) q(x)) + deg g(x) deg g(x).
This implies that
deg(r(x) r0 (x)) deg g(x),

which is impossible because deg r(x), deg r0 (x) < deg g(x). Thus,

r(x) r0 (x) = 0 or r(x) = r0 (x).

Therefore,
0

0 = (q 0 (x) q(x))g(x).

(11.3)

Because bm is a unit, deg(((q (x) q(x))g(x)) 0 unless q (x) q(x) = 0. Thus, from Eq. (11.3), we see that
q 0 (x) q(x) = 0 must be the case.
The polynomials q(x) and r(x) in Theorem 11.1.6 are called the quotient and remainder, respectively, on
division of f (x) by g(x).
Definition 11.1.7 Let R be a commutative ring with 1 and f (x) = a0 + a1 x + + an xn R[x]. For all r R,
define
f (r) = a0 + a1 r + + an rn .

When f (r) = 0, we call r a root or zero of f (x).

180

11. Polynomial Rings

In Definition 11.1.7, we think of substituting r for x in f(x). The student is used to doing this freely.
However, certain diculties arise when R is not commutative. For instance, let f (x) = a x, g(x) = b x. Set
h(x) = f (x)g(x). Then
h(x) = (a x)(b x) = ab (a + b)x + x2 .

For c R,
while

h(c) = ab (a + b)c + c2 = ab ac bc + c2
f (c)g(c) = (a c)(b c) = ab cb ac + c2 .

Hence, we cannot draw the conclusion that h(c) = f (c)g(c). However, if R is commutative (with identity), then
we can conclude that h(c) = f (c)g(c). Clearly if k(x) = f (x) + g(x), then k(c) = f (c) + g(c).
Definition 11.1.8 Let R be a commutative ring with 1 and f(x), g(x) R[x] be such that g(x) 6= 0. We say
that g(x) divides f (x) or that g(x) is a factor of f (x), and write g(x) | f (x) if there exists q(x) R[x] such
that f(x) = q(x)g(x).
Theorem 11.1.9 (Remainder Theorem) Let R be a commutative ring with identity. For f (x) R[x] and
a R, there exists q(x) R[x] such that
f (x) = (x a)q(x) + f (a).
Proof. By applying the division algorithm with x a = g(x), there exist unique q(x), r(x) R[x] such that
f (x) = (x a)q(x) + r(x), where r(x) = 0 or deg r(x) < 1. Hence, r(x) is a constant polynomial, say, r(x) = d.
By substituting a for x, we obtain f (a) = (a a)q(a) + d = d, which yields the desired result.
Corollary 11.1.10 (Factorization Theorem) Let R be a commutative ring with identity. For f (x) R[x]
and a R, x a divides f (x) if and only if a is a root of f (x).
Proof. Suppose (x a) | f (x). Then there exists q(x) R[x] such that f (x) = (x a)q(x). Hence, f (a) =
(a a)q(a) = 0, so a is a root of f (x). Conversely, suppose a is a root of f(x). Then by the remainder theorem
(Theorem 11.1.9) and the fact that f (a) = 0, we have f (x) = (x a)q(x). Consequently, (x a) | f (x).
Theorem 11.1.11 Let R be an integral domain and f (x) be a nonzero polynomial in R[x] of degree n. Then
f (x) has at most n roots in R.
Proof. If deg f(x) = 0, then f (x) is a constant polynomial, say, f (x) = c 6= 0. Clearly c has no roots
in R. Assume that the theorem is true for all polynomials of degree less than n, where n > 0 (the induction
hypothesis). Suppose deg f (x) = n. If f (x) has no roots in R, then the theorem is true. Suppose r R is a root
of f (x). Then by Corollary 11.1.10, f (x) = (x r)q(x), where deg q(x) = n 1. If there exists any other root
r0 R of f (x), then 0 = f (r0 ) = (r0 r)q(r0 ). Because r0 6= r and R is an integral domain, q(r0 ) = 0, so r0 is a
root of q(x). Therefore, any other root of f (x) is also a root of q(x). Because f (x) = (x r)q(x), any root of q(x)
is also a root of f (x). By the induction hypothesis and the fact that deg q(x) = n 1, there are at most n 1 of
these other roots r0 . Hence, in all, f (x) has at most n roots in R.
We now extend the definition of a polynomial ring from one indeterminate to several indeterminates.
Definition 11.1.12 For any ring R, we define recursively
R[x1 , x2 , . . . , xn ] = R[x1 , x2 , . . . , xn1 ][xn ],
where x1 is an indeterminate over R and xn is an indeterminate over R[x1 , x2 , . . . , xn1 ]. R[x1 , x2 , . . . , xn ] is
called a polynomial ring in n indeterminates.
S
Before describing the ring R[x1 , x2 , . . . , xn ], we introduce some notation. We write in ,...,i1 ri1 ...in xi11 xinn
Skn
Sk1
for in =0 i1 =0 ri1 ...in xi11 xinn , where each ri1 ...in R and k1 , . . . , kn are nonnegative integers.
The ring
[
R[x1 , x2 , . . . , xn ] = {
ri1 ...in xi11 xinn | ri1 ...in R}.
in ,...,i1

We have for n = 2 that

[
R[x1 , x2 ] = R[x1 ][x2 ] = {
si2 xi22 | si2 R[x1 ]}.
i2

11.1. Polynomial Rings

Now each si2 has the form
Thus,

181
S

i1

ri1 i2 xi11 .

R[x1 , x2 ]

=
=
=

S S
{ i2 ( i1 ri1 i2 xi11 )xi22 | ri1 i2 R}
S S
{ i
ri i2 xi1 xi2 | ri1 i2 R}
S 2 i1 1 i1 1i2 2
{ i2 ,i1 ri1 i2 x1 x2 | ri1 i2 R.}.

S
Definition 11.1.13 Let R be a subring of the ring S. Let c1 , c2 , . . . , cn be elements of S. Define R[c1 ] = { i ri
ci1 | ri R} and
R[c1 , c2 , . . . , cn ] = R[c1 , c2 , . . . , cn1 ][cn ].
We say that c1 , c2 , . . . , cn are algebraically independent over R if
[

ri1 ...in ci11 . . . cinn = 0

in ,..,i1

can occur only when each ri1 ...in = 0, where ri1 ...in R.
R[c1 , c2 , . . . , cn ] is a subring of S and equals the set of all finite sums of the form
[

ri1 ...in ci11 . . . cinn ,

in ,..,i1

where ri1 ...in R.

Theorem 11.1.14 Let R be a subring of a commutative ring S such that R and S have the same identity. Let
c S. Then there exists a unique homomorphism of R[x] onto R[c] such that (x) = c and (a) = a for all a
R.

S
S i
S
Proof. Define : R[x] R[c] by ( ai xi ) =
ai c for all
ai xi R[x]. Now a0 +a1 x+ +an xn = b0 +
m
b1 x+ +bm x implies that n = m and ai = bi for i = 1, 2, . . . , n. Thus, a0 +a1 c+ +an cn = b0 +b1 c+ +bn cn ,
so is well defined. By Definition 11.1.13, clearly maps R[x] onto R[c]. Because for any two polynomials f (x),
g(x) R[x], k(x) = f (x) + g(x) implies k(c) = f (c) + g(c) and h(x) = f(x)g(x) implies h(c) = f (c)g(c), it follows
that preserves + and . Therefore, is a homomorphism of R[x] onto R[c]. Clearly (x) = c and (a) = a for
allS
a R. Let
R[x] onto R[c] such that (x) = c and (a) = a for all a R. Then
S be a homomorphism
S i of S
( ai xi ) =
(ai )(x)i =
ai c = ( ai xi ). Thus, = , so is unique.
We emphasize that is well defined in Theorem 11.1.14 because x is algebraically independent over R. We
illustrate this in the following example.

S
S
Example 11.1.15
: Q[ 2] Q[x] by ( ai 2) =
ai xi . Then is not a function because (2) = 2
2 Define
2
2
and (2) = (( 2) ) = x , but 2 6= x .

Worked-Out Exercises
Exercise 1 Let R be a ring with 1. Show that
Solution: Define f : R[x] R by

R[x]/ hxi ' R.

f (a0 + a1 x + a2 x2 + + an xn ) = a0

for all a0 + a1 x + a2 x2 + + an xn R[x]. Suppose that a0 + a1 x + a2 x2 + + an xn = b0 + b1 x + b2 x2 +

+ bm xm . Then a0 = b0 , so f(a0 + a1 x + a2 x2 + + an xn ) = f(b0 + b1 x + b2 x2 + + bm xm ). Thus,
f is well defined. Clearly f is an epimorphism. Now a0 + a1 x + a2 x2 + + an xn Ker f if and only if
f (a0 + a1 x + a2 x2 + + an xn ) = 0 if and only if a0 = 0 if and only if a0 + a1 x + a2 x2 + + an xn hxi .
Therefore, Ker f = hxi . Thus,
R[x]/ hxi ' R.
Exercise 2 Let F be a field and : F [x] F [x] be an automorphism such that (a) = a for all a F. Show that
(x) = ax + b for some a, b F.

182

11. Polynomial Rings

Solution: By the division algorithm, (x) = g(x)x + b for some g(x) F [x] and b F. Because is onto F [x],
there exist h(x), p(x) F [x] such that g(x) = (h(x)) and x = (p(x)). Therefore, (x) = g(x)x +
b = (h(x))(p(x)) + (b) = (h(x)p(x) + b). Thus, x = h(x)p(x) + b because is one-one. Now
deg(x) = deg(h(x)p(x) + b) implies that deg(h(x)p(x)) = 1. Hence, either deg h(x) = 1 and deg p(x) = 0
or deg h(x) = 0 and deg p(x) = 1. Suppose deg p(x) = 0. Then p(x) = c for some c F. This implies that
x = (p(x)) = (c) = c, which is a contradiction. Therefore, deg h(x) = 0 and deg p(x) = 1. Let h(x) = a
for some a F. Thus, (x) = (h(x))x + b = (a)x + b = ax + b.

Exercise 3 Let R be a commutative ring with 1 and f (x) = a0 + a1 x + a2 x2 + + an xn R[x]. If a0 is a unit and
a1 , a2 , . . . , an are nilpotent elements, prove that f (x) is invertible.

Solution: We prove this result by induction on n = deg f (x). If n = 0, then f (x) = a0 . Hence, f (x) is invertible.
Assume that the result is true for all polynomials of the above form and degree < n. Suppose now
f (x) = a0 +a1 x+a2 x2 + +an xn R[x] such that a0 is a unit and a1 , a2 , . . . , an are nilpotent elements and
deg f (x) = n. Let g(x) = a0 +a1 x+a2 x2 + +an1 xn1 . Note that deg g(x) < n. Hence, by the induction
hypothesis, g(x) is invertible. Because an is nilpotent there exists a positive integer m such that am
n = 0.
g(x)(m1) x(m1)n ) = 1.
Then (g(x) + an xn )(g(x)1 an g(x)2 xn + a2n g(x)3 x2n + (1)m1 am1
n
It now follows that f (x) is invertible.

Exercises
1. If I is an ideal of a ring R, prove that I[x] is an ideal of the polynomial ring R[x].
2. Let R be an integral domain. Prove that R and R[x] have the same characteristic.
3. Let R be a commutative ring with 1. Describe, hxi , the ideal of R[x] generated by x.

4. (i) Let f (x) = x4 + 3x3 + 2x2 + 2 and g(x) = x2 + 2x + 1 Q[x]. Find the unique polynomials q(x),
r(x) Q[x] such that f (x) = q(x)g(x) + r(x), where either r(x) = 0 or 0 deg r(x) < deg g(x).
(ii) Let f (x) = x4 + [3]x3 + [2]x2 + [2] and g(x) = x2 + [2]x + [1] Z5 [x]. Find q(x), r(x) Z5 [x] such that
f (x) = q(x)g(x) + r(x), where either r(x) = 0 or 0 deg r(x) < deg g(x).
5. Let f (x) = x5 + x4 + x3 + x + [3], g(x) = x4 + x3 + [2]x2 + [2]x Z5 [x]. Find q(x), r(x) Z5 [x] such that
f (x) = q(x)g(x) + r(x), where either r(x) = 0 or 0 deg r(x) < deg g(x).

6. Let R = Z Z. Show that the polynomial (1, 0)x in R[x] has infinitely many roots in R.

7. Show that the polynomial ring Z4 [x] over the ring Z4 is infinite, but Z4 [x] is of finite characteristic.

8. In the ring Z8 [x], show that [1] + [2]x is a unit.

9. Let R be a commutative ring with 1 and f (x) = a0 + a1 x + + an xn R[x]. If f(x) is a unit in R[x],
prove that a0 is a unit in R and ai is nilpotent for all i = 1, 2, . . . , n.
10. Use the result of Exercise 9 to show that 1 + 5x is not a unit in Z[x].
11. Find all units of Z[x].
12. Find all units of Z6 [x].
13. Let R be an integral domain. Prove that the units of R[x] are contained in R.
14. In Z8 [x], prove the following.
(i) [4]x2 + [2]x + [4] is a zero divisor.
(ii) [2]x is nilpotent.
(iii) [4]x + [1] and [4]x + [3] are units.
15. Let R be a subring of a commutative ring S such that R has an identity.
(i) In the polynomial ring R[x1 , x2 , . . . , xn ], prove that x1 , x2 , . . . , xn are algebraically independent over R.
(ii) Prove that the mapping
: R[x1 , x2 , . . . , xn ] R[c1 , c2 , . . . , cn ]
S
S
i1
defined by ( in ...i1 ri1 ...in x1 . . . xinn ) = in ...i1 ri1 ...in ci11 . . . cinn is a homomorphism of R[x1 , . . . , xn ]
onto R[c1 , . . . , cn ], where c1 , . . . , cn S.
(iii) Prove that the homomorphism in (ii) is an isomorphism if and only if c1 , c2 , . . . , cn are algebraically
independent over R.
16. Let f (x) be a polynomial of degree n > 0 in a polynomial ring K[x] over a field K. Prove that any element
of the quotient ring K[x]/ hf(x)i is of the form g(x) + hf (x)i , where g(x) is a polynomial of degree at most
n 1.

11.1. Polynomial Rings

183

17. For the following statements, write the proof if the statement is true; otherwise, give a counterexample.
(i) If a polynomial ring R[x] has zero divisors, so does R.
(ii) If R is a field, then R[x] is a field.
(iii) In Z7 [x], (x + [1])7 = x7 + [1].

184

11. Polynomial Rings

Chapter 12

Euclidean Domains
We have seen that both rings Z and F [x], F a field, have a Euclidean or division algorithm. Because of the
significance of these rings and the power of this common property, the concept of a division algorithm is worth
abstracting.

12.1

Euclidean Domains

Definition 12.1.1 A Euclidean domain (E, +, , v) is an integral domain (E, +, ) together with a function
v : E\{0} Z# such that
(i) for all a, b E with b 6= 0, there exist q, r E such that a = qb + r, where either r = 0 or v(r) < v(b)
and
(ii) for all a, b E\{0}, v(a) v(ab).
v is called a Euclidean valuation.
The next two results show that the ring Z and the polynomial ring F [x], F a field, are Euclidean domains.
Example 12.1.2 The ring Z of integers can be considered a Euclidean domain with v(a) = |a| , a 6= 0.
Theorem 12.1.3 If F is a field, then the polynomial ring F [x] is a Euclidean domain.
Proof. By Theorem 11.1.2(ii), F [x] is an integral domain. Define
v : F [x]\{0} Z#
by
v(f (x)) = deg f (x)
for all f(x) F [x]\{0}. Since deg f (x) 0, v(f (x)) Z# for all f(x) F [x]\{0}. Let f (x), g(x) F [x], g(x) 6= 0.
By Theorem 11.1.6, there exist q(x), r(x) F [x] such that
f (x) = q(x)g(x) + r(x), where either r(x) = 0 or deg r(x) < deg g(x).
Hence,
f (x)

q(x)g(x) + r(x), where either r(x) = 0 or v(r(x)) < v(g(x)).

Let f (x) = a0 + a1 x + + an xn , an 6= 0 and g(x) = b0 + b1 x + + bm xm , bm 6= 0. Then f (x)g(x) =

a0 b0 + (a0 b1 + a1 b0 )x + + an bm xn+m . Since F is a field and an 6= 0, bm 6= 0, we find that an bm 6= 0. This
implies that deg(f (x)g(x)) = n + m. Thus, v(f (x)) = deg(f (x)) = n n + m = deg(f (x)g(x)) = v(f (x)g(x)).
Hence, F [x] is a Euclidean domain.
Example 12.1.4 Any field can be considered as a Euclidean domain with v(a) = 1 for all a 6= 0. (a = (ab1 )b +
0.)
Definition 12.1.5 The subset Z[i] = {a + bi | a, b Z} of the complex numbers is called the set of Gaussian
integers.

185

186

12. Euclidean Domains

In the next theorem, we show that Z[i] is a subring of C and determine the units of Z[i]. Gauss was the first
to study Z[i] and hence in his honor Z[i] is called the ring of Gaussian integers.
Theorem 12.1.6 The set Z[i] of Gaussian integers is a subring of C. The units of Z[i] are 1 and i.
Proof. It is easily verified that Z[i] is a subring of C. Since C is a field, Z[i] is of course an integral domain.
Suppose a + bi is a unit of Z[i]. Then there exists c + di Z[i] such that (a + bi)(c + di) = 1. This implies that
1 = 1 = (a + bi)(c + di) =(a + bi) (c + di) = (a bi)(c di), where the bar denotes complex conjugate. Thus,
1 = (a2 + b2 )(c2 + d2 ) and therefore 1 = a2 + b2 . Hence, a = 0, b = 1, or a = 1, b = 0, proving that the only
units of Z[i] are 1, i.
Theorem 12.1.7 The ring Z[i] of Gaussian integers becomes a Euclidean domain when we let the function,
N : Z[i]\{0} Z#
defined by N(a + bi) = (a + bi)(a bi) = a2 + b2 for all a, b Z, serve as the function v.
Proof. Clearly N(a+bi) is a positive integer for any nonzero element a+bi Z[i]. Let a+bi, c+di Z[i]\{0}.
Now N((a+bi)(c+di)) = N(acbd+(bc+ad)i) = (acbd)2 +(bc+ad)2 = (a2 +b2 )(c2 +d2 ) = N(a+bi)N(c+di).
From this, it follows that N(a + bi) N((a + bi)(c + di)).
It remains to be shown that for a + bi and c + di 6= 0 in Z[i], there exist q0 + q1 i, r0 + r1 i Z[i] such that
a + bi = (q0 + q1 i)(c + di) + (r0 + r1 i),
where r0 + r1 i = 0 or N(r0 + r1 i) < N(c + di). We work backward in order to see how to choose q0 + q1 i. If such
an element q0 + q1 i exists, then in C
r0 + r1 i

=
=

(a + bi) (c + di)(q0 + q1 i)
(c + di)[(a + bi)(c + di)1 (q0 + q1 i)].

Let (a + bi)(c + di)1 = u + vi, where u and v are rational numbers. Then
r0 + r1 i

=
=
=

Now
N(r0 + r1 i)

=
=

(c + di)[(u + vi) (q0 + q1 i)]

(c + di)[(u q0 ) + (v q1 )i]
[c(u q0 ) d(v q1 )] + [c(v q1 ) + d(u q0 )]i.
[c(u q0 ) d(v q1 )]2 + [c(v q1 ) + d(u q0 )]2
(c2 + d2 )[(u q0 )2 + (v q1 )2 ].

Hence, N(r0 +r1 i) < N(c+di) if (uq0 )2 +(v q1 )2 < 1. We now find an element q0 +q1 i Z[i] so that the latter
inequality holds. Take integers q0 and q1 such that (uq0 )2 14 and (v q1 )2 14 . Then (uq0 )2 +(v q1 )2 < 1.
Let
r0 + r1 i = (a + bi) (c + di)(q0 + q1 i).
Then a + bi = (c + di)(q0 + q1 i) + (r0 + r1 i), where r0 + r1 i = 0 or N(r0 + r1 i) < N(c + di).
We now consider the ideals of a Euclidean domain.
Recall that an ideal I of a ring R is called a principal ideal if I = hai for some a I.

Definition 12.1.8 Let R be a commutative ring with 1. If every ideal of R is a principal ideal, then R is called a
principal ideal ring. An integral domain which is also a principal ideal ring is called a principal ideal domain
(PID).
Theorem 12.1.9 Every Euclidean domain is a principal ideal domain.
Proof. Let E be a Euclidean domain with Euclidean valuation v. We want to show that every ideal of E
is a principal ideal. Let I be an ideal of E. Since E is a commutative ring with 1, it is enough to show that
I = Ea for some a E. If I is the zero ideal, then I = E0. Suppose now I 6= {0}. Then I contains some
nonzero element. Let P = {v(x) | 0 6= x I}. This is a nonempty subset of the nonnegative integers. By the
well-ordering principle, we find that P contains a least element. Therefore, there exists an element a I, a 6= 0
such that v(a) 0 and v(a) v(b) for all b I, b 6= 0. We now show that I = Ea. Since I is an ideal and a I,
it follows that Ea I. Let b I. Since E is a Euclidean domain, there exist q, r E such that b = aq + r, where
r = 0 or v(r) < v(a). Now r = b qa I. If r 6= 0, then v(r) P. This is a contradiction of the minimality of
v(a) since v(r) < v(a). Therefore, r = 0 and so b = qa Ea. This proves that I Ea. Hence, I = Ea.
By Theorem 12.1.9, Z, F [x] (F a field), and Z[i] are principal ideal domains.

12.1. Euclidean Domains

187

Theorem 12.1.10 Let R be a commutative ring with 1. The following conditions are equivalent.
(i) R is a field.
(ii) R[x] is a Euclidean domain.
(iii) R[x] is a PID.
Proof. (i)(ii) Follows from Theorem 12.1.3.
(ii)(iii) Follows from Theorem 12.1.9.
(iii)(i) Let a R and a 6= 0. Consider I = ha, xi , the ideal of R[x] generated by a and x. Since R[x] is a
PID, there exists f(x) R[x] such that I = hf (x)i . Now a, x hf (x)i . Therefore, there exist g(x) and h(x) in
R[x] such that f (x)g(x) = a and f(x)h(x) = x. Since f (x)g(x) = a, we must have deg f (x) = 0 and so f (x) R.
Let f (x) = b. Now bh(x) = x implies that bc = 1 for some c R. Thus, b is a unit and so I = hbi = R[x]. From
this, we have 1 I. Therefore, 1 = af1 (x) + xf2 (x) for some f1 (x), f2 (x) R[x]. This implies that 1 = da for
some d R. Hence, a is a unit in R and so R is a field.
Corollary 12.1.11 Z[x] is not a PID.
Proof. Now Z is a commutative ring with 1. Since Z is not a field, Z[x] is not a PID by Theorem 12.1.10.
We conclude this section with the following remark.

Remark 12.1.12
Consider Z[ 19] = {a + b 19 | a, b Z and a and b are either both even or both odd}.
It is known that Z[ 19] is a principal ideal domain, but not a Euclidean domain. The proof of this result is
beyond the scope of this book. However, the interested reader can find the proof in, J.C. Wilson, A principal
ideal ring that is not a Euclidean ring, Mathematics Magazine 46(1973), 34 38.

Worked-Out Exercises
Exercise 1 Let (E, +, , v) be a Euclidean domain.
(a) Show that v(a) = v(a) for all a E\{0}.

(b) Show that for all a E\{0}, v(a) v(1), where equality holds if and only if a is a unit in E.
(c) Let n be an integer such that v(1) + n 0. Show that the function
vn : E\{0} Z#
defined by vn (a) = v(a) + n for all a E\{0} is a Euclidean valuation.
Solution:

(a) For all a E\{0}, v(a) = v((1)(a)) v(a) = v((1)a) v(a). Hence, v(a) = v(a) for all
a E\{0}.

(b) Let a E\{0}. Now v(a) = v(1a) v(1). Suppose a is a unit. Then there exists an element c E
such that ac = 1. Thus, v(1) = v(ac) v(a). This implies that v(a) = v(1). Conversely, suppose that
v(a) = v(1). Since a 6= 0, there exist q, r E such that 1 = qa + r, where r = 0 or v(r) < v(1). Now
v(r) < v(1) is impossible. Hence, r = 0, showing that 1 = qa. Thus, a is a unit.

(c) Let a E\{0}. Then vn (a) = v(a) + n v(1) + n 0. Hence, vn (a) Z# . Suppose a, b E with
b 6= 0. There exist q, r E such that a = qb + r, where either r = 0 or v(r) < v(b). Now v(r) < v(b)
implies that v(r) + n < v(b) + n. Thus, vn (r) < vn (b). Also, for a, b E\{0}, vn (ab) = v(ab) + n
v(a) + n = vn (a). Therefore, vn is a Euclidean valuation on E.
Exercise 2 Let n be a square free
integer dierent from 0 and
1, which is not divisible by the square of
integer (an
n]
=
{a
+
b
n
|
a,
b

Z}.
Show
that
Z[
n] is an integral domain. Define a function
any integer).
Let
Z[

N : Z[ n] Z# by

N(a + b n) = (a + b n)(a b n) = a2 nb2 .

(a) Let x Z[ n]. Prove that N(x) = 0 if and only if x = 0.

(b) Prove that N(xy) = N(x)N(y) for all x, y Z[ n].

(c) Let x Z[ n]. Prove that N(x) = 1 if and only if x is a unit in Z[ n].

= (a c) + (b d) n Z[ n]
Solution: Let x = a + b n and y = c + d n be two
elements in Z[ n]. Now
x y
and xy = (ac + nbd) + (ad + bc)
n Z[ n]. We have 0 = 0 + 0 n Z[ n] and 1 = 1 + 0 n Z[ n].
Now it is easy to verify that Z[ n] is an integral domain.

188

12. Euclidean Domains

(a) Let x = a + b n. Then N(x) = a2 nb2 . Suppose N(x) = 0. If b = 0, then a = 0. If b 6= 0, then

2
n = ab2 = ( ab )2 , which is a contradiction to the assumption that n is a square free integer. Therefore,
a = 0 and b = 0. Thus, x = 0. The converse is trivial.

N(xy) = [(ac + nbd) + (ad + bc) n][(ac + nbd) (ad + bc) n]

2
2
= (ac + nbd) (ad + bc) n

= a2 c2 + n2 b2 d2 a2 d2 n b2 c2 n
(b) Let x = a+b n and y = c+d n. Now
= (a2 nb2 )(c2 nd2 )
= N(x)N(y).

1 if and only if (a
(c) Let x = a + b n. N(x) =
+ b n)(a b n) = 1 if and only if a + b n divides
1, i.e., if and only if a + b n is a unit in Z[ n].

Exercise 3 Show that Z[ n] is a Euclidean domain for n = 1, 2, 2, 3.

#
Z
Solution: By Worked-Out
Exercise

2 (page 187), Z[ n] is an integral domain. Define v : Z[ n]\{0}

by
v(a
+
b
n)
=
|N(a
+
b
n)|
,
where
N
is
defined
as
in
Worked-Out
Exercise
2.
Let
a
+
b
n,
c
+
d
n

Z[ n]\{0}. Now

v((a + b n)(c + d n)) = |N((a + b n)(c + d n))|

2
2 
= (a2 nb2 )(c
  nd )2 
2
2  2

= (a nb ) (c nd )
(a2 nb2 )
= v((a + b n)).

Let a +
b n, c+ d n Z[ n] with c + d n 6= 0. We want to show that there exist q0 + q1 n,
r0 + r1 n Z[ n] such that

a + b n = (c + d n)(q0 + q1 n) + (r0 + r1 n),

 2
 2



2 
2 


where either r
0 + r1 n = 0 or (r0 nr1 ) < (c nd ) . We work backward in order to see how to

choose q0 + q1 n. If such an element q0 + q1 n exists in Z[ n], then in Q[ n]

r0 + r1 n

=
=

(a + bn) (c +d n)(q0 +

q1 1n)
(c + d n)[(a + b n)(c + d n) (q0 + q1 n)].

Let (a + b n)(c + d n)1 = u + v n, where u and v are rational numbers. Then

r0 + r1 n

Now

=
=
=

(c + dn)[(u + v n) (q0 +
q1 n)]
(c + d n)[(u q0 ) + (v q1 ) n]

[c(u q0 ) + d(v q1 )n] + [c(v q1 ) + d(u q0 )] n.



[c(u q0 ) + d(v q1 )n]2 [c(v q1 ) + d(u q0 )]2 n

 2
2
2
2 

(c2 nd2 )[(u
 q0 ) n(v q1 ) ]
(c nd )





if (u q0 )2 n(v q1 )2  < 1. We now find an element q0 +q1 n Z[ n] such that (u q0 )2 n(v q1 )2  <
1. Take integers q0 and q1 such that (u q0 )2 14 and (v q1 )2 14 . For n = 1 or 2,

For n = 2 or 3,

v(r0 + r1 n)

=
=
<



(u q0 )2 n(v q1 )2  1 + (n) 1 < 1.
4
4

n
1
(u q0 )2 n(v q1 )2 .
4
4




Then (u q0 )2 n(v q1 )2  < 1 for n = 1, 2, 2 or 3. Hence, there exist q0 + q1 n, r0 + r1 n Z[ n]

such that

a + b n = (c + d n)(q0 + q1 n) + (r0 + r1 n),


 


where either r0 + r1 n = 0 or (r02 nr12 ) < (c2 nd2 ) .

= {a + bi 3 | a, b Z}. Show that Z[i 3] is an integral domain.Define v : Z[i 3]\{0} Z#

Exercise 4 Let Z[i 3]
by v(a + bi 3) = a2 + 3b2 . Show that v is not a Euclidean valuation on Z[i 3].

12.2. Greatest Common Divisors

189

Solution: Proceeding as in Worked-Out Exercise 2 (page 187), we

can show that Z[i 3]
is an integral domain.
Supposev is a Euclidean
valuation.
Now
2
and
1
+
i
3
are
elements
of
Z[i
3]. Suppose there exist

q0 + q1 i 3, r0 + r1 i 3 Z[i 3] such that

2 = (1 + i 3)(q0 + q1 i 3) + (r0 + r1 i 3),

where either r0 + r1 i 3 = 0 or r02 + 3r12 < 4. If r0 + r1 i 3 = 0, then

2 = (1 + i 3)(q0 + q1 i 3).
This implies that

4 = v(2) = v((1 + i 3)(q0 + q1 i 3)) = 4(q02 + 3q12 ).

Then q02 + 3q12 = 1, which shows that q0 = 1, q1 = 0. As a result, 2 = 1 + i 3 or 2 = (1 + i 3),

2
2
2
2
a contradiction. Suppose now r0 + 3r1 < 4. Then r0 + 3r1 = 1, 2, or 3. Since r0 and r1 are integers,
r02 + 3r12 6= 2. Suppose r02 + 3r12 = 1. Then r0 = 1, r1 = 0. Thus,

2 = (1 + i 3)(q0 + q1 i 3) + (r0 + r1 i 3),

whence
2 = q0 3q1 + r0
and
0 = q1 + q0 + r1 .
If r0 = 1 and r1 = 0, then q0 3q1 = 1 and q1 + q0 = 0. This implies that 2q1 = 1, which is impossible.
Similarly, for each remaining case we can show a contradiction.
Also, from r02 + 3r12 = 3, we can show a

contradiction. Hence, v is not a Euclidean valuation on Z[i 3].

Exercises
1. Show that the mapping v : Z\{0} N defined by v(a) = |a|n for some fixed positive integer n is a
Euclidean valuation on Z.

2. In Z[ 3], for 9 + 5 3 and 1 + 7 3, find q0 + q1 3, r0 + r1 3 Z[ 3] such that

9 + 5 3 = (q0 + q1 3)(1 + 7 3) + r0 + r1 3,



where either r0 + r1 3 = 0 or r02 3r12  < 146.

3. Consider the integral domain Z[i]. Find q0 + q1 i, r0 + r1 i Z[i] such that

3 + 7i = (q0 + q1 i)(1 + 2i) + r0 + r1 i,
 2

where either r0 + r1 i = 0 or r0 + r12  < 5.

4. Let a = 3 + 8i, b = 2 + 3i Z[i]. Find c, d = x + yi in Z[i] such that a = bc + d, where either d = 0 or

x2 + y 2 < 9.
5. Let f : R S be an epimorphism of rings. If R is a principal ideal ring, prove that S is also a principal
ideal ring.
6. Prove that the ring Zn is a principal ideal ring for all n N.
7. Which of the following statements are true? Justify.
(i) (Z, +, , v) is a Euclidean domain, where v(n) = n2 for all n > 0.
 
 
(ii) (Q, +, , v) is a Euclidean domain, where v( pq ) =  pq  for all pq 6= 0.

(iii) If a ring R is a PID, then every subring of R with identity is a PID.

12.2

Greatest Common Divisors

Definition 12.2.1 Let R be a commutative ring and a, b R be such that a 6= 0. If there exists c R such that
b = ac, then a is said to divide b or a is said to be a divisor of b and we write a | b.

190

b.

12. Euclidean Domains

When we write a | b, we mean that a 6= 0 and a divides b. The notation a - b will mean that a does not divide

Let R be a commutative ring with 1. By Definition 12.2.1, the following results follow immediately. For all
a, b, c R,
(i) a | a, 1 | a and a | 0,
(ii) a is a unit if and only if a | 1,
(iii) if a | b and b | c, then a | c.
Definition 12.2.2 Let R be a commutative ring with 1. A nonzero element a R is said to be an associate of
a nonzero element b R if a = bu for some unit u R.
Example 12.2.3 (i) In Z, 1 and 1 are the only units. For every 0 6= a Z, a and a are associates.
(ii) In Z[i], 1, 1, i, i are the only units. Thus, 1 + i, 1 i, 1 + i, 1 i are all associates of 1 + i.
Example 12.2.4 In the polynomial ring F [x] over a field F, the units form the set F \{0}. A nonconstant
polynomial f (x) has uf (x) for an associate, where u is a unit in F.
Theorem 12.2.5 Let R be a commutative ring with 1 and a, b, c R.
(i) If a is an associate of b, then b is an associate of a.
(ii) If a is an associate of b and b is an associate of c, then a is an associate of c.
(iii) Suppose R is an integral domain. Then a is an associate of b if and only if a | b and b | a.
(iv) Suppose R is an integral domain. Then a and b are associates of each other if and only if hai = hbi .
Proof. (i) This result follows from the fact that the inverse of a unit is also a unit.
(ii) This result follows from the fact that the product of two units is also a unit.
(iii) Suppose a is an associate of b. Then a = bu for some unit u R. This implies that b = au1 . Hence,
a | b and b | a. Conversely, suppose that a | b and b | a. Then there exist q1 , q2 R such that a = q1 b and
b = q2 a. Thus, b = q2 q1 b and so 1 = q2 q1 by cancellation. This implies that q1 and q2 are units and so a and b
are associates.
(iv) The result here follows from (iii) and the fact that hai = {q2 a | q2 R} and hbi = {q1 b | q1 R}.
We now introduce the notion of a greatest common divisor in a commutative ring.
Definition 12.2.6 Let R be a commutative ring and a1 , a2 , . . . , an be elements in R, not all zero. A nonzero
element d R is called a common divisor of a1 , a2 , . . . , an if d | ai for all i = 1, 2, . . . , n. A nonzero element
d R is called a greatest common divisor (gcd) of a1 , a2 , . . . , an if
(i) d is a common divisor of a1 , a2 , . . . , an and
(ii) if c R is a common divisor of a1 , a2 , . . . , an , then c | d.
The greatest common divisor (gcd) of two elements need not be unique. In fact, the gcd of two elements may
not even exist.
Example 12.2.7 Consider the ring Z10 . Then [4] = [4][6] and [6] = [4][4]. This shows that [4] and [6] are
common divisors of each other. Hence, [4] and [6] must be greatest common divisors of [4] and [6]. Now [4] and
[6] are associates since [9] is a unit and [6] = [9][4].
Example 12.2.8 In the ring E of even integers, 2 has no divisor. Hence, 2 and no other even integer can have
a common divisor.
Example 12.2.9 In a field F, a | b and b | a for all a, b F with a 6= 0 and b 6= 0. Thus, every nonzero element
is a gcd of any pair of elements.
The next result shows that in a principal ideal ring, every pair of elements not both zero has a gcd.
Theorem 12.2.10 Let R be a principal ideal ring and a, b R not both zero. Then a and b have a gcd d. For
every gcd d of a and b, there exist s, t R such that d = sa + tb.

12.2. Greatest Common Divisors

191

Proof. The ideal ha, bi of R must be a principal ideal, whence there exists d R such that ha, bi = hdi .
Thus, there exist u, v R such that a = ud and b = vd. Therefore, d is a common divisor of a and b. Since
d ha, bi , there exist s, t R such that d = sa + tb. Now suppose c is any common divisor of a and b. Then
there exist u0 , v 0 R such that a = u0 c and b = v0 c. Thus, d = (su0 + tv 0 )c and so c | d. Hence, d is a gcd of a
and b. Let d0 be any gcd of a and b. Then d | d0 and d0 | d, whence hd0 i = hdi = ha, bi . Thus, there exist s0 , t0 R
such that d0 = s0 a + t0 b.
Corollary 12.2.11 Let R be a Euclidean domain and a, b R, not both zero. Then a and b have a gcd d. For
every gcd d of a and b, there exist s, t R such that d = sa + tb.
Proof. Since every Euclidean domain is a principal ideal ring, the corollary follows by Theorem 12.2.10.
Proceeding as in the proof of Theorem 12.2.10, we can prove a similar result for any finite set of elements
a1 , a2 , . . . , an (not all zero) of a principal ideal ring.
Let R be an integral domain and a1 , a2 , . . . , an R, not all zero. Suppose that a gcd of a1 , a2 , . . . , an exists.
Let d and d0 be two greatest common divisors of a1 , a2 , . . . , an . Then d | d0 and d0 | d. We ask the reader to verify
in Exercise 6 (page 193) that d and d0 are associates. If d is a gcd of a1 , a2 , . . . , an , then any associate of d is also
a gcd of a1 , a2 , . . . , an . Considering this, we can say that the gcd of a1 , a2 , . . . , an is unique in the sense that if
d and d0 are greatest common divisors of a1 , a2 , . . . , an , then d and d0 are associates. Hence, from now on, the
gcd of a1 , a2 , . . . , an is denoted by gcd(a1 , a2 , . . . , an ). This outcome motivates the definition of associates. We
will further motivate this concept when we examine unique factorization in integral domains.
In a Euclidean domain (E, +, , v), we have seen that the gcd(a, b) of two elements a, b E (a, b not both
zero) exists in E. Next we give an algorithm similar to the algorithm of finding the gcd of two integers given in
Chapter 1.
Let a, b E with b 6= 0.
Step 1: Find q1 and r1 in E such that a = q1 b + r1 , where r1 = 0 or v(r1 ) < v(b). If r1 = 0, then b | a and
so gcd(a, b) = b. If r1 6= 0, then gcd(a, b) = gcd(b, r1 ). Thus, we need to find gcd(b, r1 ).
Step 2: Find q2 and r2 in E such that b = q2 r1 + r2 , where r2 = 0 or v(r2 ) < v(r1 ). If r2 = 0, then
gcd(a, b) = gcd(b, r1 ) = r1 . If r2 6= 0, then proceed to find gcd(r1 , r2 ). Since v(b) > v(r1 ) > v(r2 ) > is a
strictly descending chain of nonnegative integers, the above process must stop after a finite number of steps.
Therefore, there exists a positive integer n such that in the nth step there exist elements qn and rn in E such
that rn2 = qn rn1 + rn , where rn = 0. Thus,
gcd(a, b)

=
=
=
..
.
=

gcd(b, r1 )
gcd(r1 , r2 )
gcd(r2 , r3 )
..
.
gcd(rn2 , rn1 )

gcd(rn1 , rn )

(a = q1 b + r1 , v(r1 ) < v(b))

(b = q2 r1 + r2 , v(r2 ) < v(r1 ))
(r1 = q3 r2 + r3 , v(r3 ) < v(r2 ))
..
.
(rn3 = qn1 rn2 + rn1 ,
v(rn1 ) < v(rn2 ))
(rn2 = qn rn1 + rn , rn = 0).

Next we find x, y in E such that gcd(a, b) = ax + by.

rn1

=
=
=
..
.
=

rn3 qn1 rn2

rn3 qn1 (rn4 qn2 rn3 )
rn3 (1 + (qn1 )(qn2 )) + rn4 (qn1 )
..
.
by + ax.

Worked-Out Exercises
Exercise 1 Let E be a Euclidean domain. Let a, b, q, r E be such that b 6= 0, a = qb + r, and r 6= 0. Show that
gcd(a, b) = gcd(b, r).
Solution: Let gcd(a, b) = d and gcd(b, r) = d0 . Now d | a and d | b. Thus, r = a qb implies that d | r. Hence, we
find that d is a common divisor of b and r and so d0 | d. Now d0 | b and d0 | r and so a = qb + r implies
that d0 | a. Therefore, d0 is a common divisor of a and b and so d | d0 . By Theorem 12.2.5(iii), it follows
that d and d0 are associates and so gcd(a, b) = gcd(b, r).
Exercise 2 Let a, b, and c be three nonzero elements of a PID R. Show that there exist x, y R such that ax + by = c
if and only if gcd(a, b) | c.

192

12. Euclidean Domains

Solution: Let gcd(a, b) = d. Suppose there exist x, y R such that ax + by = c. Since d | a and d | b, we find that
d | c. Conversely, suppose that gcd(a, b) | c. Then c = dd0 for some d0 R. Now there exist x0 , y 0 R such
that d = ax0 + by 0 . Then ax0 d0 + by 0 d0 = dd0 = c. Let x = x0 d0 and y = y 0 d0 . Then ax + by = c.

Exercise 3 In the domain Z[i 5], prove the following:

(a) gcd(2, 1 + i 5) = 1,

(b) gcd of 6(1 i 5) and 3(1 + i 5)(1 i 5) does not exist.

the units
are 1 and 1. Let
a + ib 5= gcd(2, 1 + i 5). Then (a + ib 5) | 2. Thus,
Solution: (a) In Z[i 5],
2 = (a + ib 5)(c + id 5) for some c + id 5 Z[i 5]. This implies that
4 = (a2 + 5b2 )(c2 + 5d2 ).
Hence,
a2 + 5b2 = 2,

c2 + 5d2 = 2

(12.1)

a2 + 5b2 = 4,

c2 + 5d2 = 1

(12.2)

a2 + 5b2 = 1,

c2 + 5d2 = 4.

or
or
(12.3)
2

Now Eqs. (12.1) cannot hold for any c, d Z. The only integral solutions of a + 5b = 4 are a = 2
Eqs.
and b = 0 and the only integral solutions of a2 + 5b2 = 1 are a = 1 and
b = 0. Thus, from
(12.2) and Eqs.
(12.3)
we
find
that
gcd(2,
1
+
i
5)
=
1
or
2.
If
gcd(2,
1
+
i
5)
=
2,
then
2
|
(1
+
i
5).

Hence, 1 + i 5 = 2(p + iq 5) for some p + iq 5 Z[i 5]. This implies that

2p
=
1
=
2q.
But
there

do not exist integers p and q such that 2p = 1 = 2q. Therefore, gcd(2, 1 + i 5) = 1.

(b) Suppose
i 5), 3(1 + i 5)(1 i 5)) =
gcd(6(1 i 5),
3(1 + i 5)(1
i 5)) exists.
Then gcd(6(1

3(1 i 5)
i 5). Now (1 + i 5)(1 i 5)
= 3(1
= 6. Hence, 6 is a common divisor

gcd(2, 1 + i 5)
i 5)(1 i 5). Consequently,
6
|
3(1

i
of 6(1 i 5) and 3(1 +

5). This
implies that 2 | (1 i 5),
which is not true in Z[i 5]. Therefore, gcd(6(1 i 5), 3(1 + i 5)(1 i 5)) does not exist.

Exercise 4 In Z[i], find gcd(9 5i, 9 + 13i).

Solution: By Theorem 12.1.7, Z[i] is a Euclidean domain, where the valuation is defined by N(a + bi) = a2 + b2 .
Now N(9 5i) = 106 and N(9 + 13i) = 250.

72i
40
34
Step 1: 9+13i
= (9+13i)(9+5i)
= 8145i+117i65
= 146+72i
= 146
+ 106
= (1 106
) + (1 106
)i =
95i
106
106
106
106
(1 + i) 40+34i
.
106
Thus, 9 + 13i = (1 + i)(9 5i) 40+34i
(9 5i) = (1 + i)(9 5i) 360+306i200i+170
= (1 + i)(9
106
106
5i) 530+106i
=
(1
+
i)(9

5i)
+
(5

i).
Note that N(5 i) < N(9 5i).
106
95i
95i 5+i
45+9i+25i+5
17
7
4
= 5i
= 40+34i
= 20+17i
= 20
Step 2: 5i
5+i =
26
26
13
13 + 13 i = (1 13 ) + (1 + 13 )i =
7+4i
(1 + i) + 13 .
35+7i20i+4
= (1+i)(5i)+ 3913i
=
Thus, 95i = (1+i)(5i)+ 7+4i
13 (5i) = (1+i)(5i)+
13
13
(1 + i)(5 i) + (3 i). Note that N(3 i) < N(5 i).
3+i
= 5i
= 155i3i+1
= 148i
= 74i
= 7
4i
= (1 25 ) (1 15 )i =
Step 3: 5i
3i
3i 3+i
10
10
5
5
5
2+i
(1 i) + 5 .
(3 i) = (1 i)(3 i)+ 6+2i+3i+1
= (1 i)(3 i)+ 5+5i
=
Thus, 5 i = (1 i)(3 i) + 2+i
5
5
5
(1 i)(3 i) + (1 + i). Note that N(1 + i) < N(3 i).
3i
3i
1i
Step 4: 1+i
= (1+i)
= 33i+i1
= 42i
= 2 i.
(1i)
2
2

Thus, 3 i = (2 i)(1 + i) + 0.
Hence, gcd(9 5i, 9 + 13i) = 1 + i.

Exercise 5 In Z[x], find two polynomials f (x) and g(x) such that gcd(f (x), g(x)) = 1, but there do not exist f1 (x)
and g1 (x) in Z[x] such that 1 = f (x)f1 (x)+ g(x)g1 (x).
Solution: x + 6 and x + 4 are elements of Z[x]. The gcd(x + 6, x + 4) = 1. Suppose there exist f1 (x) and g1 (x) in
Z[x] such that
1 = (x + 6)f1 (x) + (x + 4)g1 (x).
(12.4)
The constant term of the right-hand side in Eq. (12.4) is an even integer, whereas in the left-hand side,
the constant term is 1, a contradiction. Hence, there do not exist f1 (x) and g1 (x) in Z[x] such that
1 = (x + 6)f1 (x) + (x + 4)g1 (x).

12.2. Greatest Common Divisors

193

Exercise 6 Let R be a commutative ring with 1 and S denote the set of all infinite sequences {an } of elements from
R. Define + and on S by
{an } + {bn } = {an + bn } and
{an } {bn } = {cn },

where
Show that

cn = a0 bn + a1 bn1 + + an b0 for all n = 0, 1, 2, . . . .

(a) S is a commutative ring with 1;

(b) An element {an } is a unit if and only if a0 is a unit in R;
(c) If R is a field, then S is a PID.

Solution:

(a) It is easy to verify that S is a commutative ring with 1. The sequence {1, 0, 0, . . .} is the identity
element of S.
(b) Let {an } S. Suppose {an } is a unit. Then there exists a sequence {bn } such that {an }{bn } = 1.
Hence, a0 b0 = 1 and so a0 is a unit. Conversely, suppose that a0 is a unit. We now consider the
1
1
1
sequence {bn }, where b0 = a1
0 , b1 = a0 (a1 a0 ), . . . , bk = a0 (a1 bk1 + + ak b0 ), k 2.
1
1
Now a0 b0 = 1, a0 b1 + a1 b0 = a0 (a1
(a
a
))+
a
a
=
0,
.
.
.
, ak b0 + ak1 b1 + + a0 bk =
1 0
1 0
0
(a
b
+

+
a
b
))
=
0.
Therefore,
{a
ak b0 + ak1 b1 + + a0 (a1
1
0
n }{bn } = 1, proving that {an }
k1
k
0
is a unit.
(c) Suppose R is a field. Let I be an ideal of S. If I = {0}, then I is a principal ideal. Suppose I 6= {0}.
Let {an } be a nonzero element of I. We define the order of a nonzero sequence {an } as the first
nonnegative integer n such that an 6= 0, i.e., n is a nonnegative integer such that an 6= 0 and ai = 0
for i < n. There exists a sequence {an } such that order of {an } order of {bn } for all {bn } I.
Suppose order of {an } = k. Let {cn } be a sequence such that ci = ak+i for all i 0. Then {cn }1
exists and {cn }1 {an } = {dn } I. Also, dk = 1 and di = 0 for all i 6= k. We now show that
I = h{dn }i . Clearly h{dn }i I. Suppose {un } I. Let the order of {un } be m. Then m k. Let
{rn } S be such that rmk+i = um+i for all i 0 and ri = 0 for all i m k. It is easy to verify
that {un } = {rn }{dn } h{dn }i . Hence, I = h{dn }i .

Exercises

1. Find all associates of (i) 3 2i in Z[i], (ii) 1 + i 5 in Z[i 5], (iii) [6] in Z10 , (iv) [4] in Z5 , and (v) [2] + x
in Z3 [x].

2. Find all the units of the integral domain Z[i 3].

3. Find all the associates of 2 + x 3x2 in Z[x].
4. Show that [4] and [6] are associates in Z10 .

5. Find all units of the polynomial ring Z7 [x]. Find all associates of x2 + [2] in Z7 [x].
6. Let R be an integral domain and a1 , a2 , . . . , an (n 2) be elements of R not all zero. If d1 and d2 are two
greatest common divisors of a1 , a2 , . . . , an , prove that d1 and d2 are associates.
7. Let (E, +, , v) be a Euclidean domain. Let a, b E be such that a and b are associates. Prove that
v(a) = v(b).
8. Let (E, +, , v) be a Euclidean domain and a, b E. If a | b and v(a) = v(b), prove that a and b are
associates.
9. Let (E, +, , v) be a Euclidean domain and a and b be nonzero elements of E. Prove that v(ab) > v(a) if
and only if b is not a unit.
10. Let E be a Euclidean domain. Let a, a0 , b, b0 , d be nonzero elements of E such that a = a0 d and b = b0 d.
Prove that gcd(a0 , b0 ) = 1 if and only if gcd(a, b) = d.
11. In a PID R, prove that the congruence ax b(mod c), where a, b, c are nonzero elements of R has a
solution in R if and only if gcd(a, c) | b. (Here ax b(mod c) means ax b = cr for some r R.)

12. Let R be an integral domain. Let a, b, and c be nonzero elements of R such that gcd(a, b) and gcd(ca, cb)
exist. Prove that gcd(ca, cb) = c gcd(a, b).

13. In Z[i], find gcd(27i, 2+11i). Also, find x and y in Z[i] such that gcd(27i, 2+11i) = x(27i)+y(2+11i).

194

12. Euclidean Domains

14. Let R be an integral domain and a1 , a2 , . . . , an (n 2) be nonzero elements of R. An element d R is

called a least common multiple (lcm) of a1 , a2 , . . . , an if
(i) ai | d, i = 1, 2, . . . n and
(ii) if c R is such that ai | c, i = 1, 2, . . . n, then d | c.
Prove the following in R.
(i) If d1 and d2 are two least common multiples of a1 , a2 , . . . , an , then d1 and d2 are associates.
(ii) If d is a least common multiple of a1 , a2 , . . . , an , then rd is a least common multiple of ra1 , ra2 , . . . , ran ,
for all r R, r 6= 0.

15. Let I be the set of all nonunits of Z[i]. Is I an ideal of Z[i]? Show that for any nontrivial ideal P of Z[i],
the quotient ring Z[i]/P is a finite ring.

16. Show that Z[ 2] has no unit between 1 and 1 + 2.

17. In thedomain Z[ 2], prove that an element a + b 2 6= 1 is a unit if and only if a + b 2 = (1 + 2)k or
a + b 2 = (1 + 2)k for some positive integer k.
18. An integral domain R is said to satisfy the gcd property if every finite nonempty subset of R has a gcd.
Prove that every PID satisfies the gcd property.

19. Prove that the integral domain Z[ 2] satisfies the gcd property, where the gcd property is defined in
Exercise 18.

12.3

Prime and Irreducible Elements

In this section, we introduce the concepts of prime elements and irreducible elements in a commutative ring with
1. We show that in a PID and hence in a Euclidean domain these two concepts coincide.
Definition 12.3.1 Let R be a commutative ring with 1.
(i) An element p of R is called irreducible if p is nonzero and a nonunit, and p = ab with a, b R implies
that either a or b is a unit. An element p of R is called reducible if p is not irreducible.
(ii) An element p of R is called prime if p is nonzero and a nonunit, and if whenever p | ab, a, b R, then
either p divides a or p divides b.
(iii) Two elements a and b of R are called relatively prime if their only common divisors are units.
Remark 12.3.2 Let p Z. If p is an ordinary prime, then both p and p are irreducible and prime in the sense
of Definition 12.3.1.
From the definition of an irreducible element, it follows that the only divisors of an irreducible element p are
the associates of p and the unit elements of R. The converse of this result does not always hold in a commutative
ring with 1.
Example 12.3.3 The ring Z6 is a commutative ring with 1. In this ring, the unit elements are [1] and [5]. Since
[3] = [3][3] and [3] is not a unit it follows that [3] is not irreducible. But [3] is an associate of [3]. Also, in Z6 ,
it can be verified that [3] is divisible only by associates and the units of Z6 . Next, we show that [3] is a prime
element in Z6 . Let [a], [b] Z6 and [3] | [a][b]. Then there exists [c] Z6 such that [a][b] = [3][c], i.e., [ab] = [3c].
From this, it follows that 6 | (ab 3c). This implies that 3 | (ab 3c). Since 3 | 3c, we must have 3 | ab. Since 3
is prime in Z, 3 | a or 3 | b. Thus, either [3] | [a] or [3] | [b]. Hence, [3] is a prime element in Z6 .
Theorem 12.3.4 Let R be an integral domain and p R be such that p is nonzero and a nonunit. Then p is
irreducible if and only if the only divisors of p are the associates of p and the unit elements of R.
Proof. Suppose the only divisors of p are the associates of p and the unit elements of R. Let p = ab for
some a, b R. Suppose a is not a unit. Then a is an associate of p. Therefore, a = pu for some unit u R. Now
p = pub. Since R is an integral domain, it follows that ub = 1. Hence, b is a unit and so p is irreducible. We leave
the converse as an exercise.
We now consider several examples of prime elements and irreducible elements.
Example 12.3.5 In Z, 1 and 1 are the only units, and therefore 2 is divisible by 1 and 2. It follows that 2
is not divisible by any other integer. Therefore, 2 is an irreducible element. Suppose now 2 | ab and 2 does not
divide a for some a, b Z. Since 2 does not divide a, a is an odd integer and so gcd(2, a) = 1. Therefore, there
exist c, d Z such that 1 = 2c + ad. Thus, b = 2cb + abd. Since 2 | ab and 2 | 2bc, it follows that 2 | b. Hence, 2
is prime.

12.3. Prime and Irreducible Elements

195

Example 12.3.6 The polynomial x2 + 1 is irreducible in R[x], but is reducible in C[x]. If x2 + 1 were reducible
in R[x], then there would exist real numbers a, b, c, d such that
x2 + 1 = (ax + b)(cx + d) = acx2 + (ad + bc)x + bd.
Then ac = 1 = bd and ad + bc = 0. Thus, 1 = (ac)(bd) = (ad)(bc) = (ad)(ad). Hence, 1 = (ad)2 , which is
impossible in R. However, x2 + 1 = (x + i)(x i) in C[x].
Example 12.3.7 The polynomial x2 2 is irreducible in Q[x] and reducible in R[x]. If x2 2 were reducible in
Q[x], then there would exist a, b, c, d Q such that
x2 2 = (ax + b)(cx + d) = acx2 + (ad + bc)x + bd.
2
Thenac = 1, ad + bc = 0, and bd = 2. Thus, (ad)
= (ac)(bd) = 2. This implies
= (ad)(ad) = (ad)(bc)
Q. This
is
a
contradiction
since
2

/ Q. Therefore, x2 2 is irreducible in Q[x]. However,

that 2 = ad

x2 2 = (x 2)(x + 2) in R[x].

Example 12.3.8 The polynomial ax + b is irreducible in F [x], where F is a field and a 6= 0. Suppose ax + b =
f (x)g(x). Then deg(f (x)g(x)) = 1 = deg f (x) + deg g(x). We may assume that deg f (x) = 0 and deg g(x) = 1.
Since deg f (x) = 0, f (x) is a nonzero constant polynomial and thus a unit. Hence, ax + b is irreducible.
Example 12.3.9 Consider the polynomial ring Z[x, y]. Then x and y are irreducible. 2x is not prime since
2x | 2x, but 2x does not divide 2 and 2x does not divide x. Also, 2x is reducible. x2 and y 2 are relatively prime,
but neither is irreducible nor prime.
Theorem 12.3.10 Let R be an integral domain and p be a prime element in R. Then p is irreducible.
Proof. Suppose p = bc for some b, c R. To show p is irreducible, we must show that either b is a unit or c
is a unit. Now p = bc implies that p | bc. Since p is prime, p | b or p | c. If p | b, then b = pq for some q R. Thus,
p = bc = pqc and so p(1 qc) = 0. Since R is an integral domain and p 6= 0, p(1 qc) = 0 and so 1 qc = 0.
Thus, qc = 1, which implies that c is a unit. Similarly, if p | c, then b is a unit. Hence, p is irreducible.
The following example shows that the converse of Theorem 12.3.10 is not true.
Example 12.3.11 Consider the integral domain

Z[i 5] = {a + bi 5 | a, b Z}.

Letus show that 3 = 3 + 0i

5 Z[i 5]is irreducible, but not prime. Suppose 3 = (a + bi 5)(c + di 5) in
Z[i 5]. Then 3 = 3 = (a bi 5)(c di 5). Hence, 9 = (a2 + 5b2 )(c2 + 5d2 ). Since a, b, c, d are integers, the
previous equality implies that
a2 + 5b2 = 3 and c2 + 5d2 = 3
(12.5)
or
a2 + 5b2 = 1 and c2 + 5d2 = 9

(12.6)

a2 + 5b2 = 9 and c2 + 5d2 = 1.

(12.7)

or
Clearly there do not exist integers a, b, c, d satisfying
Eqs. (12.5). The first equation of Eqs. (12.6) implies that

b = 0 and a= 1. Thus, it follows that a + bi 5 is a unit. Similarly, the second

equation
of Eqs. (12.7) implies

that c + di 5 is a unit. Hence,

3
is
irreducible.
Now
3
|
6
and
6
=
(1
+
i
5)(1

i
5).
Suppose
3 | (1 + i 5).

the equation
Then 1 + i 5 = 3(a + bi 5) for some a, b Z. This implies
that 3a = 1, a contradiction, since
3a = 1 has no solution in Z. Hence, 3 does not divide (1 + i 5). Similarly, 3 does not divide (1 i 5). Thus, 3
is not prime.
The following theorem show that the converse of Theorem 12.3.10 holds in a principal ideal ring.
Theorem 12.3.12 Let R be a principal ideal ring and p R. If p is irreducible, then p is prime.
Proof. Suppose p divides ab, where a, b R. Then there exists r R such that pr = ab. Now hp, bi = hdi
for some d R. Therefore, there exists q R such that p = dq. Since p is irreducible, either d or q must be a
unit. If d is a unit, then hp, bi = hdi = R. Hence, 1 = sp + tb for some s, t R. Therefore, a = asp + atb =
asp + tpr = (as + tr)p. This implies that p divides a. If, on the other hand, q is a unit, then d = pq 1 hpi .
Thus, hdi hpi hp, bi = hdi so that hpi = hp, bi . Hence, b hpi and so p divides b.

196

12. Euclidean Domains

Corollary 12.3.13 Let R be a principal ideal domain and p R. Then p is irreducible if and only if p is prime.
Proof. The result follows by Theorems 12.3.10 and 12.3.12.
Corollary 12.3.14 Let R be a Euclidean domain and p R. Then p is irreducible if and only if p is prime.
Proof. Since every Euclidean domain is a principal domain, the result follows from Corollary 12.3.13.
Theorem 12.3.15 Let R be a principal ideal ring and a, b R. If a and b are relatively prime, then there exist
s, t R such that 1 = sa + tb.
Proof. Since the common divisors are units, 1 is a gcd of a and b. The desired result follows from Theorem
12.2.10.
We conclude this section by proving the following theorem, which characterizes irreducible polynomials over
a field.
Theorem 12.3.16 Consider the polynomial ring F [x] over the field F and p(x) F [x]. Then the following
conditions are equivalent.
(i) p(x) is irreducible.
(ii) F [x]/ hp(x)i is an integral domain.
(iii) F [x]/ hp(x)i is a field.
Proof. (i)(iii). Let f (x) F [x]/ hp(x)i be such that f (x) 6= 0, where f(x) denotes the coset f (x) + hp(x)i .
Now up(x) and u, where u F \{0}, are the only elements of F [x] which divide p(x). Since f (x)
/ hp(x)i , f(x)
and p(x) are relatively prime and so there exist s(x), t(x) F [x] such that 1 = s(x)f (x) + t(x)p(x). Thus
1 = s(x)f (x) + t(x)p(x) (in F [x]/ hp(x)i )
and so 1 = s(x) f (x). Hence, f (x) has an inverse, namely, s(x), and so F [x]/ hp(x)i is a field.
(iii)(ii): Immediate.
(ii)(i): If p(x) is a unit, then hp(x)i = F [x] and so F [x]/ hp(x)i = {0}, a contradiction to the hypothesis
that F [x]/ hp(x)i is an integral domain. Therefore, p(x) is not a unit. Suppose p(x) = f (x)g(x). Then 0 =
p(x) = f (x)g(x) = f (x) g(x). Therefore, f (x) = 0 or g(x) = 0. This implies that f(x) hp(x)i or g(x) hp(x)i ,
say, f(x) hp(x)i . Thus, f (x) = q(x)p(x) for some q(x) F [x]. Hence, p(x) = q(x)p(x)g(x) and so by a degree
argument q(x), g(x) F \{0} are units. Thus, the only factorization of p(x) is u1 (up(x)), where u is a unit in
F [x]. Consequently, p(x) is irreducible.

Worked-Out Exercises
Exercise 1 Show that [2] is a prime element in Z10 , but [2] is not irreducible in Z10 .

Solution: In Z10 , [1], [3], [7], and [9] are the only units. Now [2] = [2] [6]. Since neither [2] nor [6] is a unit, [2] is
reducible. Suppose [2] | [a][b]. Then [2] | [ab]. Therefore, [ab] = [k][2] for some [k] Z10 . This implies that
ab 2k = 10r for some r Z, i.e., ab = 2k + 10r = 2(k + 5r). Therefore, 2 | ab. Since 2 is prime in Z, 2 | a
or 2 | b. Hence, [2] | [a] or [2] | [b]. Thus, [2] is prime. Note that Z10 is not an integral domain.

Exercise 2 Let R be an integral domain such that any two elements a, b R, not both zero, have a gcd d expressible
in the form d = ra + tb, r, t R. Let p R. Show that p is prime if and only if p is irreducible.

Solution: Every prime element in an integral domain is irreducible by Theorem 12.3.10. Let us prove the converse.
Suppose p is irreducible. Let p | ab, a, b R. Now gcd(p, a) exists in R. Let d = gcd(p, a). Since d | p and p
is irreducible, it follows that either d is an associate of p or d is a unit. Suppose d is an associate of p. Then
p | d. This implies that p | a, since d | a. Suppose d is a unit. Since 1 is an associate of d, 1 = gcd(p, a).
Thus, there exist s, t R such that 1 = ps + at. This implies that b = psb + abt. Now p | psb and p | abt.
Hence, p | b.

Exercise 3 Let n be a square free

integer (an
integer dierent from 0 and 1, whichis not divisible by the square of
any integer). Let Z[ n] = {a + b n | a, b Z}. Define a function N : Z[ n] Z by

N(a + b n) = (a + b n)(a b n) = a2 nb2 .

Show that if N(x) is a prime integer, then x is irreducible for all x Z[ n].

Solution: Suppose
N(x) = p, where p is a prime integer. Suppose x = (a+b n)(c+d n). Now p = N(a+b n)N(c+

2
2
d n) = (a2 nb2 )(c2 nd2 ) by Worked-Out Exercise
2 (page 187). Hence, either (a nb ) = 1 or
(c2 nd2 ) = 1, i.e., either a + b n is a unit or c + d n is a unit. Thus, x is irreducible.

12.3. Prime and Irreducible Elements

197

Exercises

1. Show that in the integral domain Z[i 5], 2 + i 5 is an irreducible element, but not a prime element.
2. Show that 2 i, 1 + i, and 11 are irreducible elements in Z[i].

3. In Z[i 5], show that 3 is not a prime element.

4. In Z12 , show that [3] is a prime element, but is not irreducible.
5. Is the polynomial x2 + [1] irreducible in Z2 [x]?
6. Let T be the set of all sequences {an } of elements of Z. Prove the following.
(i) T is an integral domain with respect to addition and multiplication defined by for all {an }, {bn } T,
{an } + {bn }
{an } {bn }

=
=

{an + bn }
{cn },

where cn =

Sn

i=0

ai bni .

(ii) T0 = {{an } T | ai = 0 for all but a finite number of indices} is a subring with identity.
(iii) The element (1, 1, 0, . . .) is a unit in T, but not in T0 .
(iv) (2, 3, 1, 0, 0, . . .) is irreducible in T, but not in T0 .
7. Let R be an integral domain. Show that (i) every associate of an irreducible element in R is irreducible
and (ii) every associate of a prime element in R is prime.
8. In Z[i], show that 3 is a prime element, but 5 is not a prime element.
9. What are the prime elements of Z9 ? Are they irreducible?
10. In Z[i], if a + bi is an element such that a2 + b2 is a prime integer, then show that a + bi is a prime element.

11. Let a + bi 3 Z[i 3]. If a2 + 3b2 is a prime integer, show that a+ bi 3 is an irreducible element in Z[i 3].
12. In the following exercises, write the proof if the statement is true; otherwise, give a counterexample.
(i) 13 is an irreducible element in Z[i].
(ii) Every prime element of Z is also a prime element of Z[i].
(iii) In Z18 , every prime element is an irreducible element.
(iv) In Z[i], a + bi is a prime element if and only if a bi is a prime element.
(v) In a PID R, if p and q are two prime elements such that p | q, then p and q are associates.

198

12. Euclidean Domains

Chapter 13

Unique Factorization Domains

13.1

Unique Factorization Domains

In this section, we study those integral domains in which an analogue of the fundamental theorem of arithmetic
holds.
Definition 13.1.1 A nonzero nonunit element a of an integral domain D is said to have a factorization if a
can be expressed as
a = p1 p2 pn ,
where p1 , p2 , . . . , pn are irreducible elements of D. The expression p1 p2 pn is called a factorization of a.
An integral domain D is called a factorization domain (FD) if every nonzero nonunit element has a
factorization.
In Chapter 15, we saw that in an integral domain D every nonzero element a D is always divisible by the
associates of a and the units of D. These are called the trivial factors of a. All other factors (if any) of a are
called nontrivial. For example, 2 and 3 are nontrivial factors of 6 in Z. In the following lemma, we show
that a nonzero nonunit element that has no factorization as a product of irreducible elements can be expressed
as a product of any number of nontrivial factors.
Lemma 13.1.2 Let D be an integral domain. Let a be a nonzero nonunit element of D such that a does not
have a factorization. Then for every positive integer n, there exist nontrivial factors a1 , a2 , . . . , an D of a such
that a = a1 a2 an .
Proof. By the hypothesis, a is not irreducible. Therefore, a = a1 b1 , where a1 , b1 D are nontrivial factors
of a. At least one of a1 or b1 does not have a factorization; otherwise the factorization of a1 and b1 put together
produces a factorization of a. Suppose a1 does not have a factorization. Then a1 is a nonzero nonunit element
and a1 is not irreducible. There exist nontrivial factors a2 , b2 D of a1 such that a1 = a2 b2 . Then a = a2 b2 b1 .
Now at least one of a2 or b2 does not have a factorization. If a2 does not have a factorization, we repeat the
above process with a2 . Proceeding this way, we can find nontrivial factors a1 , a2 , . . . , an D of a such that
a = a1 a2 an .
Theorem 13.1.3 Let D be an integral domain with a function N : D\{0} Z# such that for all a, b
D\{0}, N(ab) N(b), where equality holds if and only if a is a unit. Then D is a FD.
Proof. Suppose D contains a nonzero nonunit element a such that a does not have a factorization. Now
N(a) Z# . Let N(a) = n. By Lemma 13.1.2, a can be expressed as a product of n + 2 nontrivial factors
a1 , a2 , . . . , an+2 D. Then a = a1 a2 an+2 and
n

=
>
>
>
..
.
>
>

N(a)
N(a2 an+2 )
N(a3 an+2 )
N(a4 an+2 )

(since a1 is not a unit)

N(an+1 an+2 )
N(an+2 ).

199

200

13. Unique Factorization Domains

This shows that there exist at least n + 1 distinct nonnegative integers strictly less than n, a contradiction. Thus,
D is a FD.
Example 13.1.4 Consider the integral domain Z[i]. Define
N : Z[i]\{0} Z#
by N(a + bi) = a2 + b2 for all a + bi Z[i]. It is easy to verify that a + bi is a unit if and only if N(a + bi) = 1.
Let a + bi, c + di be two nonzero elements of Z[i]. Then N((a + bi)(c + di)) = N((ac bd) + (ad + bc)i) =
(ac bd)2 + (ad + bc)2 = (a2 + b2 )(c2 + d2 ) (c2 + d2 ) = N(c + di), where the equality holds if and only if
N(a + bi) is a unit. Hence, Z[i] is a FD.
Definition 13.1.5 An integral domain D is said to satisfy the ascending chain condition for principal
ideals (ACCP), if for each sequence of principal ideals, ha1 i , ha2 i , ha3 i , . . . such that
ha1 i ha2 i ha3 i ,
there exists a positive integer n (depending on the sequence) such that han i = hat i for all t n.
Lemma 13.1.6 Every principal ideal domain D satisfies the ACCP.
Proof. Let ha1 i ha2 i ha3 i be a chain of principal ideals in D. It can be easily verified that
I = iN hai i is an ideal of D. Since D is a PID, there exists an element a D such that I = hai . Hence,
a han i for some positive integer n. Then I han i I. Therefore, I = han i . For t n, hat i I = han i hat i .
Thus, han i = hat i for all t n.
Theorem 13.1.7 An integral domain D with the ACCP is a FD.
Proof. Suppose D is not a FD. Then there exists a nonzero nonunit element a such that a does not have a
factorization. Thus, a is not irreducible and so a = a1 b1 , where a1 , b1 D are nontrivial factors of a. At least
one of a1 or b1 must not have a factorization, otherwise the factorization of a1 and b1 put together will produce
a factorization of a. Suppose a1 does not have a factorization. Now a and a1 are not associates. Therefore,
hai ha1 i . Since a1 does not have a factorization, we can express a1 = a2 b2 , where a2 , b2 D are nontrivial
factors of a1 . At least one of a2 or b2 does not have a factorization. Suppose a2 does not have a factorization.
Then hai ha1 i ha2 i . We now repeat the above process with a2 . Thus, we find that there exists an infinite
strictly ascending chain of principal ideals in D, a contradiction. Hence, D is a FD.
Corollary 13.1.8 Every PID is a FD.
Proof. The proof is immediate by Lemma 13.1.6 and Theorem 13.1.7.
Definition 13.1.9 An integral domain D is called a unique factorization domain (UFD) if the following
two conditions hold in D :
(i) every nonzero nonunit element of D can be expressed as
a = p1 p2 pn ,
where p1 , p2 , . . . , pn are irreducible elements of D and
(ii) if a = p1 p2 pn = q1 q2 qm are two factorizations of a as a finite product of irreducible elements of D,
then n = m and there is a permutation of {1, 2, . . . , n} such that pi and q(i) are associates for all i = 1, 2, . . . ,
n.
From the above definition, it follows that an integral domain D is a UFD if and only if D is a FD and every
nonzero nonunit element of D is uniquely expressible (apart from unit factors and order of the factors) as a finite
product of irreducible elements.
Let us first prove the following interesting property of a UFD.
Theorem 13.1.10 In a unique factorization domain, every irreducible element is prime.

13.1. Unique Factorization Domains

201

Proof. Let D be a UFD. Let p be an irreducible element of D and p | ab in D, where a, b D. If a = 0,

then p divides a, and if b = 0, then p divides b. If a is a unit, then p divides b, and if b is a unit, then p divides
a. We now assume that a and b are nonzero and nonunits. Now ab = pc for some c D. Let d = pc = ab. Since
neither a nor b is a unit, it follows that d is not a unit. If c is a unit, then d is irreducible and so either a or b
must be a unit, a contradiction. Therefore, c is not a unit. Since D is a UFD, there exist irreducible elements
c1 , c2 , . . . , cn , a1 , a2 , . . . , am , and b1 , b2 , . . . , br in D such that c = c1 c2 cn , a = a1 a2 am , and b = b1 b2 br .
Hence, d = pc1 c2 cn = a1 a2 am b1 b2 br are two expressions of d as a finite product of irreducible elements.
Since D is UFD, p must be an associate of one of the irreducible elements a1 , a2 , . . . , am , b1 , b2 , . . . , br . If one of
a1 , a2 , . . . , am is an associate of p, then p | a, and if one of b1 , b2 , . . . , br is an associate of p, then p | b. Hence, p
is prime.

Example 13.1.11 Consider the integral domain Z[i 5] = {a + bi 5 | a, b Z}. Define

N : Z[i 5]\{0} Z#
by

N(a + bi 5) = a2 + 5b2 .

We can show that

only if N(a + bi 5) = 1. Let a +
a + bi 5 is a unit
if and
bi 5, c + di 52 be two nonzero
elements of Z[i 5]. Then N((a + bi 5)(c + di 5)) = N((ac 5bd) + i(ad + bc) 5) = (ac 5bd) + 5(ad + bc)2 =
2
(a2 + 5b2 )(c2 + 5d
) (c2 + 5d2 ) = N((c +
di 5)), where equality holds if and only if N((a + bi 5)) = 1, i.e., if
FD by Theorem 13.1.3.In Example 12.3.11,
we showedthat 3
and only if a + bi 5 is a unit. Hence, Z[i
5] is a

is an irreducible
element.
Now 3 | (2 + i 5)(2 i 5). Suppose 3 | (2 + i 5). Then 2 + i 5 = 3(m + ni 5) for
some m +
2 = 3m and 1 = 3n, which
is impossible forintegers m and n. Therefore,

ni 5 Z[i 5]. This implies

3 - (2 + i 5). Similarly, 3 - (2 i 5). Thus, 3 is not prime in Z[i 5]. Hence, Z[i 5] is not a UFD by Theorem
13.1.10.

elements and 2 is not an

In this integral domain, we
that 2, 1 + i 5, 1 i 5are irreducible

can also show

associate of any one of 1 + i 5 and 1 i 5. Hence, 6 = 2 3 = (1 + i 5)(1 i 5) are two factorizations of 6,
but there does not exist any correspondence between the irreducible factors such that the corresponding elements
are associates.
Theorem 13.1.12 A factorization domain D is a UFD if and only if every irreducible element of D is a prime
element.
Proof. Suppose the factorization domain D is a UFD. Then by Theorem 13.1.10, every irreducible element
is a prime element.
Conversely, assume that every irreducible element is a prime element in the FD D. Suppose a = p1 p2 pn =
q1 q2 qm are two factorizations of a as a finite product of irreducible elements. Then p1 p2 pn = q1 (q2 qm )
implies that q1 | (p1 p2 pn ). Since q1 is also prime, at least one of p1 , p2 , . . . , pn is divisible by q1 . Let q1 | p1 .
Now p1 and q1 are both irreducible. Hence, p1 = u1 q1 for some unit u1 . Then u1 q1 p2 pn = q1 q2 qm , from
which it follows by the cancelation property that u1 p2 pn = q2 qm = q2 (q3 qm ). Now q2 | (u1 p2 pn ).
Since q2 is prime, q2 does not divide u1 . Hence, q2 divides one of p2 , . . . , pn , say, q2 | p2 . Then p2 = u2 q2 for some
unit u2 and u1 u2 q2 p3 pn = q2 qm . Canceling q2 from this relation, we obtain u1 u2 p3 pn = q3 qm . If
n > m, then proceeding this way we find that u1 u2 um pm+1 pn = 1, which implies that each of pm+1 , . . . ,
pn is a unit, a contradiction. If n < m, then we find that u1 u2 un = qn+1 qm . This implies that each of qn+1 ,
. . . , qm divides a unit, which is again a contradiction. Thus, n = m. Also, we have shown that the corresponding
irreducible factors pi , qi , i = 1, 2, . . . , n, in the factorizations p1 p2 pn and q1 q2 qn are associates. Hence, D
is a UFD.
Theorem 13.1.13 Every PID is a UFD.
Proof. From Lemma 13.1.6, we find that every PID satisfies ACCP. Hence, by Theorem 13.1.7, every PID
is a FD. Also, by Theorem 12.3.12, every irreducible element is prime in a PID. Thus, by Theorem 13.1.12, it
follows that every PID is a UFD.
By Theorem 12.1.9, every Euclidean domain is a PID and hence by Theorem 13.1.13, every Euclidean domain
is a UFD. This result is one of the important results in factorization theory. Let us prove this result independently.
First we prove the following lemma.
Lemma 13.1.14 Let E be a Euclidean domain and a, b E. If a | b, b 6= 0, and a is neither a unit nor an
associate of b, then v(a) < v(b).

202

13. Unique Factorization Domains

Proof. Since a is not an associate of b, it follows that b | a. Hence, a = bq + r, where r = 0 or v(r) < v(b).
Now b = ac for some c E. This implies that r = a bq = a acq = a(1 cq). If 1 cq = 0, then c is a unit
and so b is an associate of a, a contradiction. Therefore, 1 cq 6= 0. Thus, v(r) = v(a(1 cq)) v(a) and so
v(b) > v(a).
Theorem 13.1.15 A Euclidean domain E is a unique factorization domain.
Proof. Let v denote the Euclidean valuation of the Euclidean domain E. By induction on v(a), we first show
that every nonzero element a of E is either a unit or can be written as a finite product of irreducible elements. If
v(a) = v(1), then a is a unit. Assume that every nonzero element b E is either a unit or expressible as a finite
product of irreducible elements if v(b) < v(a), where v(a) > v(1) (the induction hypothesis). If a is irreducible,
there is nothing to prove. Suppose that a is not irreducible. Then a = bc, where neither b nor c is a unit. Suppose
b is an associate of a. Then b = au for some unit u E. Thus, a = bc = auc and so 1 = uc, i.e., c is a unit, a
contradiction. Therefore, b is not an associate of a. Similarly, c is not an associate of a. By Lemma 13.1.14, it
now follows that v(b) < v(a) and v(c) < v(a). Thus, by our induction hypothesis, b and c are expressible as a
finite product of irreducible elements of E. Hence, so is a.
The uniqueness of the factorization follows as in Theorem 13.1.12
From Theorem 12.1.9, we know that every Euclidean domain is a principal ideal domain. We noted in the
remark on page 187 that the converse of this result is not true. In Theorem 13.1.13, we showed that every
principal ideal domain is a unique factorization domain. The converse of this result is also not true. There is a
class of rings for which the converse is true. Call a complex number an algebraic integer if it is a root of a
monic polynomial p(x) in Z[x]. The set of all algebraic integers in a finite field extension (Chapter 24) of
Q is
such a ring. However, most of these rings are not unique factorization domains. Forexample,the ring Z[i 5] in
Example 13.1.11 is a ring in which there is no unique factorization. Here 6 = (1 i 5)(1 + i 5) = 2 3 are two
factorizations of 6 as a product of two irreducible elements. However, the ideal h6i has a unique (up to order)


2

factorization as a product of prime ideals (defined in Chapter 17), h6i = 3, 1 + i 5 3, 1 i 5 2, 1 + i 5 .
As a matter of fact, the entire class of rings in question has the property that every ideal has a unique factorization
as a product of prime ideals.

Worked-Out Exercises

Exercise 1 Show that the integral domain Z[ 10] = {a + b 10 | a, b Z} is a FD.

Solution: Define N : Z[ 10]\{0} Z# by for all a + b 10 Z[ 10],



N(a + b 10) = a2 10b2  .



Now N(a + b 10) = 1 if and only if a2 10b2  = 1 if and only if (a + b 10)(a b 10) = 1 if

and onlyif a + b
10 is a unit.
b 10, c + d 10 be two nonzeroelements of Z[ 10]. Then
 2 Let2 a +
2
2
2
2
N((a + b 10)(c + d 10)) = a 10b  c 10d  c 10d  = N((c + d 10)), where equality holds

if and only if N((a + b 10)) = 1, i.e., if and only if a + b 10 is a unit. Hence, Z[ 5] is a FD by Theorem
13.1.3.
Exercise 2 Show that in a UFD, every nonzero nonunit has only a finite number of nonassociated nontrivial factors.
Solution: Let D be a UFD. Suppose a is a nonzero nonunit element of D. Then a can be expressed uniquely as
a = pr11 pr22 prkk ,
where p1 , p2 , . . . , pk are distinct primes and r1 , r2 , . . . , rk are positive integers. Let d = pt11 pt22 ptkk , where
0 ti ri , i = 1, 2, . . . , k. Then d is a divisor of a. Now suppose d is any divisor of a and d is a
tm
nonunit. Then d can be expressed uniquely as d = q1t1 q2t2 qm
, where q1 , q2 , . . . , qm are distinct primes
r
and t1 , t2 , . . . , tm are positive integers. Since d | a, for all i = 1, 2, . . . , m, qiti | pj j for some j, 1 j k.
rj
Then qi | pj and so qi | pj . Therefore, qi is an associate of pj . Also, we find that ti rj . Thus, d is
an associate of pl11 pl22 plkk , 0 li ri , i = 1, 2, . . . , k. Consequently, a has only a finite number of
nonassociated nontrivial divisors.
Exercise 3 Let R = {a0 + a1 x + + an xn Q[x] | a0 Z, n Z# }. Show that R is not a UFD.

Solution: Clearly R is a subring of Q[x] and R contains 1. Hence, R is an integral domain. Now any unit of R is also
a unit of Q[x]. In Q[x], the units are the nonzero elements of Q. Since R Q = Z, it follows that 1 and 1
are the only units of R. For any nonnegative integer n, 21n x R and 21n x is not an associate of 21m x when
n 6= m. Now x = 2n ( 21n x) shows that 21n x is a divisor of x. Hence, x has infinite number of nontrivial

13.2. Factorization of Polynomials over a UFD

203

divisors in R. If R is a UFD, then x cannot have an infinite number of nontrivial divisors. Thus, R is not
a UFD.
Exercise 4 In a UFD, show that the gcd of any two nonzero elements exists.

Solution: Let R be a UFD and a, b be nonzero elements of R. If one of a or b is a unit, then gcd(a, b) = 1. Suppose
a and b are nonunits. Then a can be expressed uniquely as
a = pt11 pt22 ptkk ,
where p1 , p2 , . . . , pk are irreducible elements such that pi is not an associate of pj when i 6= j and
t1 , t2 , . . . , tk are positive integers. Similarly, b can be expressed uniquely (up to associates) as
b = q1r1 q2r2 qnrn ,
where q1 , q2 , . . . , qn are irreducible and r1 , r2 , . . . , rn are positive integers. Now if q1 is not an associate of
any of p1 , . . . , pk , then we write a = pt11 ptkk q10 . Next if q2 is not an associate of any of p1 , p2 , . . . , pk ,
then we write a = pt11 pt22 ptkk q10 q20 . But, if q2 is an associate of one of p1 , p2 , . . . , pk , then skip q2 and
consider q3 . Continue the process for q3 , . . . , qn . We do the same thing for b. So we can write
a
b

=
=

nm
1 n2
un
1 u2 um
l1 l2
lm
u1 u2 um ,

where u1 , u2 , . . . , um are irreducible elements such that ui is not an associate of uj when i 6= j and
n1 , n2 , . . . , nm , l1 , l2 , . . . , lm are nonnegative integers. Let d = uk1 1 uk2 2 ukmm , where ki = min{ni , li },
i = 1, 2, . . . , m. Then d | a and d | b. Let c | a and c | b, c R. Since any irreducible divisor of c is an
associate of one of u1 , u2 , . . . , um , it follows that c must be of the form
c = uh1 1 uh2 2 uhmm ,
where hi 0, and hi ni , hi li , i = 1, 2, . . . , m. Thus, hi ki , i = 1, 2, . . . , m. Hence, c | d. Thus,
d = gcd(a, b).

Exercises
1. Show that Z satisfies the ACCP.
2. If the integral domain R satisfies the ACCP, prove that the polynomial ring R[x] satisfies the ACCP.
3. Prove that an integral domain D is a UFD if and only if D satisfies the ACCP and every irreducible
element is prime in D.

4. Show that the integral domains Z[i 6], Z[i 7], and Z[i 10] are factorization domains, but not unique
factorization domains.
5. Let a, b be two nonzero elements of a UFD D. If gcd(a, b) = 1 and a | c, b | c, prove that ab | c in D, where
c D.

6. For the following statements, write the proof if the statement is true; otherwise, give a counterexample.
(i) Any subring of a UFD with identity is also a UFD.
(ii) 1 and 1 are the only units of a UFD.

13.2

Factorization of Polynomials over a UFD

In this section, we show that every polynomial of degree 1 over a UFD R can be uniquely expressed as a
product of irreducible polynomials over R.
Definition 13.2.1 Let f (x) = a0 +a1 x+ +an xn be a nonzero polynomial in R[x]. Then the gcd{a0 , a1 , . . . , an }
is called the content of f(x).
It is known that the gcd of {a0 , a1 , . . . , an } is not unique. If u and v are two gcds of {a0 , a1 , . . . , an }, then u
and v are associates. Hence, if c1 and c2 are two contents of f (x), then c1 and c2 are associates and any associate
of c1 is also a content of f (x). If a and b are two elements of R such that a is an associate of b, then we write
a b.
The content of f(x) is denoted by contf (x).

204

13. Unique Factorization Domains

Definition 13.2.2 A nonzero polynomial f(x) R[x] is called a primitive polynomial if contf (x) is a unit.
Lemma 13.2.3 Let R be a UFD. Let f (x) and g(x) be two primitive polynomials in R[x]. Then f (x)g(x) is also
a primitive polynomial in R[x].
Proof. Let f (x) = a0 +a1 x+ +an xn and g(x) = b0 +b1 x+ + bm xm . Let cf contf(x) and cg contg(x).
Since f (x) and g(x) are primitive, cf and cg are unit elements in R. Suppose that f (x)g(x) is notSa primitive
polynomial. Let f (x)g(x) = c0 + c1 x + + cn+m xn+m , where c0 = a0 b0 , c1 = a0 b1 + a1 b0 , . . . , ci = ij=0 aj bij ,
where aj = 0 if j > n, and bij = 0 if i j > m. Now contf (x)g(x) is not a unit. Let p be a prime element in R
such that p divides contf (x)g(x). Then p divides ci for all i = 0, 1, . . . , n + m. Since cf and cg are unit elements,
p does not divide each of a0 , a1 , . . . , an and also p does not divide each of b0 , b1 , . . . , bm . Let t be the smallest
nonnegative integer such that p does not divide at . Then p divides ai , for i = 0, 1, . . . , t 1, and p does not divide
at . Similarly, let r be the smallest nonnegative integer such that p does not divide br . Then p divides bj , for
j = 0, 1, . . . , r 1, and p does not divide br . Therefore, p does not divide at br . Now ct+r = a0 bt+r + a1 bt+r1 +
+ at1 br+1 + at br + at+1 br1 + + at+r b0 , where bi = 0 if i > m and ai = 0 if i > n. Now p divides ai ,
for i = 0, 1, . . . , t 1, p divides bj , for j = 0, 1, . . . , r 1, and p divides ct+r . Hence, p divides at br , which is a
contradiction. Thus, contf(x)g(x) is a unit and so f (x)g(x) is a primitive polynomial.
Example 13.2.4 In Z[x], 6x2 + 3x 9 = 3(2x2 + x 3). Hence, 6x2 + 3x 9 is not a primitive polynomial.
But 2x2 + x 3 is a primitive polynomial.
Theorem 13.2.5 Let R be a UFD. Let f (x) and g(x) be two nonzero polynomials in R[x]. Then there exists a
unit u R such that
cont(f(x)g(x)) = ucontf (x)contg(x).
Proof. Let cf denote contf(x) and cg denote contg(x). Then f (x) = cf f1 (x) and g(x) = cg g1 (x), where
f1 (x) and g1 (x) are primitive polynomials in R[x]. Now cont(f (x)g(x)) and cont(cf cg f1 (x)g1 (x)) are associates.
Since cf cg is a nonzero element of R, it follows that
cont(cf cg f1 (x)g1 (x))
and
cf cg cont(f1 (x)g1 (x))
are associates. By Lemma 13.2.3, cont(f1 (x)g1 (x)) is a unit. Hence,
cont(f(x)g(x)) = ucf cg
for some unit u.
It is known that the polynomial ring F [x] over a field F is a Euclidean domain, and hence a unique factorization domain. To take advantage of this result, let us extend an integral domain R to its quotient field Q(R)
and establish the relationship between elements of Q(R)[x] and R[x].
In the remainder of the section, we let Q(R) denote the quotient field of R.
Lemma 13.2.6 Let R be a UFD. If f (x) is a nonzero polynomial in Q(R)[x], then there exist nonzero elements
a, b R and a primitive polynomial f1 (x) in R[x] such that f (x) = ab1 f1 (x), where b1 is the inverse of b in
Q(R)[x].
Proof. Let f (x) = c0 + c1 x + + cn xn Q(R)[x] be a nonzero polynomial. Then ci Q(R), i = 0, 1, . . . , n.
1
1
Therefore, there exist ai , bi R such that ci = ai b1
i , bi 6= 0, i = 0, 1, . . . , n. Now f (x) = a0 b0 + a1 b1 x + +
1 n
an bn x . Let b = b0 b1 bn . Then
bf (x) = a0 b1 bn + a1 b0 b2 bn x + + an b0 b1 bn1 xn R[x].
Clearly bf (x) is nonzero. Let a = cont(bf (x)). Then bf(x) = af1 (x), where contf1 (x) is a unit and f1 (x) R[x].
Hence, f (x) = b1 af1 (x), where b, a R and f1 (x) is a primitive polynomial in R[x].
Lemma 13.2.7 Let R be a UFD. Let f (x) be a nonzero polynomial in R[x]. If f (x) = d1 f1 (x) = d2 f2 (x), where
f1 (x) and f2 (x) are primitive polynomials in R[x] and d1 , d2 Q(R), then d1 = ud2 for some unit u R.
Proof. Since d1 , d2 Q(R), we can write d1 = ab1 and d2 = cd1 for some a, b, c, d R. Thus, f (x) =
ab1 f1 (x) = cd1 f2 (x). This implies that adf1 (x) = cbf2 (x). Since f1 (x) and f2 (x) are primitive, ad = ucb for
some unit u R by Theorem 13.2.5. Thus, d1 = ab1 = ucd1 = ud2 .

13.2. Factorization of Polynomials over a UFD

205

Lemma 13.2.8 Let R be a UFD. Let f (x) be a nonconstant primitive polynomial in R[x]. Then f (x) is irreducible
in R[x] if and only if f(x) is irreducible in Q(R)[x].
Proof. Suppose f (x) is irreducible in R[x] and f(x) is not irreducible in Q(R)[x]. Then there exist h(x), g(x)
Q(R)[x] such that f (x) = h(x)g(x), deg h(x) 1, and deg g(x) 1. By Lemma 13.2.6, there exist a, b, c, d R
with b 6= 0, d 6= 0, and primitive polynomials h1 (x), g1 (x) R[x] such that h(x) = ab1 h1 (x) and g(x) =
cd1 g1 (x). Hence, f (x) = ab1 cd1 h1 (x)g1 (x). This implies that bdf (x) = ach1 (x)g1 (x). Now f (x) is primitive
and so contf (x) is a unit. Thus, cont(bdf(x)) = bdu for some unit u. Now
cont(ach1 (x)g1 (x))

=
=
=

vac cont(h1 (x)g1 (x)) for some unit v R

v1 ac cont(h1 (x)) cont(g1 (x)) for some unit v1 R
v1 acv2 v3 for some units v2 , v3 R.

Hence, bd = acw for some unit w R. Thus, f (x) = wh1 (x)g1 (x) for some unit w R. This shows that f (x) is
not irreducible in R[x], which is a contradiction. Therefore, f(x) is irreducible in Q(R)[x]. Conversely, let f(x)
be irreducible in Q(R)[x]. Suppose f (x) is reducible in R[x]. Now f (x) = rg(x), where r R and r is a not a unit
is impossible since f (x) is primitive. Thus, there exist polynomials f1 (x), f2 (x) in R[x] such that deg f1 (x) 1,
deg f2 (x) 1, and f (x) = f1 (x)f2 (x). Now f1 (x) and f2 (x) are also nonconstant polynomials in Q(R)[x]. Hence,
f (x) is not irreducible in Q(R)[x], a contradiction. Consequently, f (x) is irreducible in R[x].
Example 13.2.9 Consider the polynomial 4x + 4 in Q[x]. Now 4x + 4 = 4(x + 1). 4 is a unit in Q[x] and x + 1
is irreducible in Q[x]. Hence, 4x + 4 is irreducible in Q[x]. But 4 is not a unit in Z[x]. Hence, 4x + 4 is not
irreducible in Z[x]. Also, 3 is irreducible in Z[x], but 3 is not irreducible in Q[x].
We are now in a position to prove our main result of this section. Before proving this theorem, let us recall
the following assertions concerning the polynomial ring R[x] so that we can enjoy the beauty and depth of this
theorem.
(i) If R is a commutative ring with 1, then R[x] is a commutative ring with 1.
(ii) If R is an integral domain, then R[x] is an integral domain.
(iii) If R is a field, then R[x] is not a field, but R[x] is a Euclidean domain.
(iv) If R is a PID, then R[x] may not be a PID.
Theorem 13.2.10 Let R be a UFD. Then R[x] is a UFD.
Proof. Let f (x) be a polynomial of degree n 1. Let f (x) = cf f1 (x), where cf is a content of f (x) and
f1 (x) is a primitive polynomial in R[x]. Now Q(R)[x] is a UFD and f1 (x) R[x] Q(R)[x]. Therefore, there
exist irreducible polynomials g1 (x), g2 (x), . . . , gr (x) in Q(R)[x] such that f1 (x) = g1 (x)g2 (x) gr (x). By Lemma
13.2.7, gi (x) = ai b1
i hi (x), ai , bi R, bi 6= 0, and hi (x) is a primitive polynomial in R[x], i = 1, 2, . . . , r. Also,
by Lemma 13.2.8, hi (x) is irreducible in R[x], i = 1, 2, . . . , r. Hence,
1
1
f1 (x) = a1 a2 ar b1
1 b2 br h1 (x) hr (x).

Let a = a1 a2 ar and b = b1 b2 br . Then

bf1 (x) = ah1 (x) hr (x).

(13.1)

By Lemma 13.2.3, h1 (x) hr (x) is primitive. This implies that a = ub for some unit u R and so
f1 (x) = uh1 (x) hr (x).
This shows that
f (x) = ucf h1 (x) hr (x).

(13.2)

Since an associate of an irreducible polynomial is also an irreducible polynomial, it follows that uh1 (x) is irreducible. Thus, for any polynomial f(x) of degree 1, there exist irreducible polynomials g1 (x), . . . , gk (x) in R[x]
such that
f (x) = cf g1 (x) gk (x),
where cf = contf (x). If cf is not a unit, then there exist irreducible elements a1 , a2 , . . . , at R such that
f (x) = a1 a2 at g1 (x) gk (x).

(13.3)

f (x) = a1 a2 at g1 (x) gk (x) = b1 b2 bl h1 (x) hq (x),

(13.4)

Suppose now that

206

13. Unique Factorization Domains

where ai , bj are irreducible elements in R, i = 1, . . . , t, j = 1, . . . , l and

g1 (x), . . . , gk (x), h1 (x), . . . , hq (x)
are irreducible elements in R[x]. Now a1 a2 at and b1 b2 bl are two factorizations as a product of irreducible
elements in R of cf . Therefore, by (13.4)
g1 (x) gk (x) = dh1 (x) hq (x),

(13.5)

where d is a unit in R. Now g1 (x), . . . , gk (x), h1 (x), . . . , hq (x) are primitive and irreducible in R[x]. Hence, these
polynomials are also irreducible in Q(R)[x]. Since Q(R)[x] is a UFD, Eq. (13.5) implies that k = q and there exists
a one-one correspondence between {g1 (x), . . . , gk (x)} and {h1 (x), . . . , hq (x)} such that the corresponding factors
are associates in Q(R)[x] and hence by Lemma 13.2.7, they are also associates in R[x]. Thus, the factorization
(13.4) of f (x) in R[x] is unique. Consequently, R[x] is a UFD.
Corollary 13.2.11 Let R be a UFD. The polynomial ring R[x1 , . . . , xn ] is a UFD.
We see that the polynomial ring F [x, y] is a unique factorization domain. However, F [x, y] is not a Euclidean
domain. This can be verified by showing that F [x, y] is not a principal ideal ring. We ask the reader to show in
the exercises that the ideal hx, yi in F[x, y] is not a principal ideal.
As shown in Example 13.1.11, Z[i 5] is not a UFD. Thus, even though the polynomial ring F [x] is a unique
factorization domain, a ring of the form F [c] need not be one. Thus, the homomorphic image of a unique
factorization domain need not be a unique factorization domain.

Worked-Out Exercises
Exercise 1 Let f (x) be a nonzero polynomial in Z[x]. Show that f (x) can be expressed as a product of two polynomials
g(x) and h(x) of Q[x] with deg g(x) < deg f (x) and deg h(x) < deg f (x) if and only if there exist g1 (x),
h1 (x) Z[x] such that deg g(x) = deg g1 (x), deg h(x) = deg h1 (x), and f (x) = g1 (x)h1 (x).
Solution: Suppose there exist g(x) and h(x) in Q[x] with deg g(x) < deg f (x), deg h(x) < deg f (x), and f (x) =
g(x)h(x). There exist nonzero elements a, b, c, d Z and primitive polynomials g2 (x), h2 (x) Z[x] such
that g(x) = ab1 g2 (x) and h(x) = cd1 h2 (x) by Lemma 13.2.6. Hence, f (x) = ab1 cd1 g2 (x)h2 (x). This
implies that bdf (x) = acg2 (x)h2 (x). Let d1 be the content of f(x). Then we can write f (x) = d1 f1 (x),
where f1 (x) is a primitive polynomial in Z[x]. Hence, bdd1 f1 (x) = acg2 (x)h2 (x). Now g2 (x)h2 (x) is also a
primitive polynomial. Then bdd1 = uac for some unit u Z. This implies bdd1 = ac or bdd1 = ac. Hence,
f1 (x) = g2 (x)h2 (x) or f1 (x) = g2 (x)h2 (x). Let g1 (x) = d1 g2 (x). Now f (x) = d1 f1 (x) = d1 g2 (x)h2 (x) =
g1 (x)h1 (x), where h1 (x) = h2 (x) or f (x) = d1 f1 (x) = d1 g2 (x)h2 (x) = g1 (x)h1 (x), where h1 (x) = h2 (x).
Also, from the construction, it follows that deg g2 (x) = deg g1 (x) = deg g(x) < deg f (x) and deg h2 (x) =
deg h1 (x) = deg h(x) < deg f (x). The converse is trivial.
Exercise 2 Show that Z[x] is a UFD, but not a PID.
Solution: Since Z is a UFD, Z[x] is a UFD by Theorem 13.2.10. (By Corollary 12.1.11, Z[x] is not a PID. However,
here we want to show that Z[x] is not a PID by showing the existence of ideals in Z[x], which are not
principal.) Consider
I = hxi + hni ,
where n Z, n
/ {0, 1, 1}. We claim that I is not a principal ideal. Suppose I = hf (x)i , where
f (x) Z[x]. Then hni hf (x)i . Therefore, n = f (x)g(x) for some g(x) Z. Since deg n = 0, deg f (x) = 0
and hence f (x) Z. Let f (x) = a Z. Now hxi hai . Then x = ah(x) for some h(x) Z[x]. Again by a
degree argument, deg h(x) = 1. Let h(x) = a0 +a1 x, where a0 , a1 Z, a1 6= 0. Then x = a(a0 +a1 x). Hence,
1 = aa1 hai = I = hxi + hni . Thus, 1 = xs(x) + nt(x) for some s(x), t(x) Z[x]. Let t(x) = t0 + t1 x+
+tr xr . Then by comparing coecients in 1 = xs(x) + nt(x), we get 1 = nt0 . Hence, n divides 1, which
is a contradiction. Therefore, I is not a principal ideal.

Exercises
1. Let f (x) Z[x] be irreducible. Prove that f (x) is primitive.
2. Let f (x) be a nonconstant primitive polynomial in Z[x]. Prove that if f(x) is not irreducible in Q[x], then
f (x) is not irreducible in Z[x].
3. Show that the polynomial ring Q[x, y] is a UFD, but not a PID.
4. Let R be a UFD. Let f (x) be a primitive polynomial in R[x]. Show that any nonconstant divisor of f(x)
is also a primitive polynomial.

13.3. Irreducibility of Polynomials

13.3

207

Irreducibility of Polynomials

In the previous section, we proved that any polynomial of degree 1 over a UFD can be expressed as a product
of irreducible polynomials. Thus, irreducible polynomials play an important role in polynomial rings. But it
is not always easy to determine if a polynomial is irreducible over a UFD. In this section, we establish some
criteria for irreducibility of polynomials. We first note that any polynomial of degree 1 over a field F is always
irreducible. If f (x) = ax + b F [x] with a 6= 0, then x = a1 b is a root of f (x) in F. In this connection, let us
point out that a linear polynomial over a UFD D may not be irreducible in D[x]. For example 2x + 4 = 2(x + 2)
is not irreducible in Z[x]. We now consider polynomials of degree 2 and 3. For these polynomials, we can apply
the following test to check irreducibility. Let F denote a field.
Theorem 13.3.1 Let f(x) F [x] be a polynomial of degree 2 or 3. Then f (x) is irreducible over F if and only
if f (x) has no roots in F.
Proof. Suppose that deg f (x) = 3 and f (x) is irreducible. If f (x) has a root in F, say a, then x a
divides f (x) in F [x] and so f(x) is reducible over F. Conversely, suppose f (x) has no roots in F. Assume that
f (x) is reducible. Then f(x) = g(x)h(x) for some g(x), h(x) F [x], deg g(x) 1 and deg h(x) 1. Now
deg(g(x)h(x)) = 3. Therefore, either deg g(x) = 1 and deg h(x) = 2 or deg h(x) = 1 and deg g(x) = 2. To be
specific, let deg g(x) = 1 and deg h(x) = 2. Then g(x) = ax + b for some a, b F, a 6= 0. Now a1 b F and
g(a1 b) = 0. Thus, a1 b is a root of g(x) and hence a1 b is a root of f (x) in F. This is a contradiction to
our assumption that f (x) has no roots in F. Hence, f (x) is irreducible over F. A similar argument can be used
for the case when deg f (x) = 2.
Example 13.3.2 (i) Let f (x) = x2 + x + [1] Z2 [x]. Now
f ([0]) = [0]2 + [0] + [1] 6= [0],
f([1]) = [1]2 + [1] + [1] = [1] 6= [0].

Hence, f (x) has no roots in Z2 . Thus, by Theorem 13.3.1, f (x) is irreducible over Z2 .
(ii) Let g(x) = x3 + [2]x + [1] Z3 [x]. Now
g([0]) = [0]3 + [2][0] + [1] 6= [0],
and

g([1]) = [1]3 + [2][1] + [1] = [4] = [1] 6= [0],

g([2]) = [2]3 + [2][2] + [1] = [13] = [1] 6= [0].

Hence, g(x) has no roots in Z3 . Thus, by Theorem 13.3.1, g(x) is irreducible over Z3 .
Instead of considering polynomials over an arbitrary field, let us now consider polynomials over the field Q
of all rational numbers. By Lemma 13.2.8, a nonconstant primitive polynomial f (x) Z[x] is irreducible in Q[x]
if and only if f (x) is irreducible in Z[x]. It is not dicult to decide whether or not a polynomial is primitive. In
order to decide whether or not f (x) is irreducible, we sometimes consider the corresponding polynomial in Zp [x]
for some prime p.
Theorem 13.3.3 Let f (x) = a0 + a1 x + + an xn Z[x] be of degree n > 1. If there exists a prime p such that
f (x) = [a0 ] + [a1 ]x + + [an ]xn is irreducible in Zp [x] and deg f(x) = deg f (x), then f (x) is irreducible in Q[x].
Proof. Suppose f (x) satisfies the given conditions of the theorem for some prime p. Suppose f (x) is reducible
in Q[x]. Then there exist polynomials g(x) = b0 + b1 x + + bm xm and h(x) = c0 + c1 x + + ck xk in
Z[x], 0 < m < n, 0 < k < n such that f(x) = g(x)h(x) by Worked-Out Exercise 1 (page 206). Thus,
[a0 ] + [a1 ]x + + [an ]xn = ([b0 ] + [b1 ]x + + [bm ]xm ) ([c0 ]+ [c1 ]x + + [ck ]xk ). Since deg f (x) = deg f(x) =
n = k + m, it follows that [bm ][ck ] 6= 0 in Zp . Hence, [bm ] 6= [0] and [ck ] 6= [0]. Consequently, g(x) and h(x) are
nonconstant polynomials in Zp [x]. Since the units of Zp [x] are the nonzero elements of Zp , it follows that g(x)
and h(x) are nonunits. Therefore, f (x) is not irreducible in Zp [x], a contradiction. Hence, f (x) is irreducible in
Q[x].
Example 13.3.4 Consider the polynomial f (x) = 57 x3 12 x + 1 in Q[x]. Then 14f (x) = 10x3 7x + 14. Let
f1 (x) = 10x3 7x + 14. Now in Z3 [x], f1 (x) = [10]x3 [7]x + [14] = x3 x + [2]. Since f1 ([0]) = [2], f1 ([1]) = [2],
f1 ([2]) = [2]3 [2] + [2] = [2], it follows that f1 (x) has no root in Z3 [x]. As a result 14f(x) is irreducible in Q[x].
But 14 is a unit in Q[x]. Hence, f (x) is irreducible in Q[x].

208

13. Unique Factorization Domains

Let f (x) Q[x] and deg f(x) 2. If f (x) has a root in Q, then f (x) is reducible. The following theorem will
help us to see whether a polynomial f (x) Q[x] has a root in Q.
Theorem 13.3.5 Let f (x) = a0 + a1 x + + an xn Z[x] be of degree n and a0 6= 0. Let
f (x), where u and v are relatively prime. Then

u
v

Q be a root of

u | a0 and v | an .
Proof. Since

u
v

is a root of f (x),
u
u
u
0 = f ( ) = a0 + a1 + + an ( )n .
v
v
v

Thus,
0 = a0 v n + a1 uv n1 + + an1 un1 v + an un .
Hence,
v(a0 v n1 + a1 uv n2 + + an1 un1 ) = an un .

This implies that v | an un . Since u and v are relatively prime, v | an . Similarly, u | a0 .

Example 13.3.6 Let f(x) = 2x3 7x + 1 and uv Q be a root of f (x) with gcd(u, v) = 1. Then u | 1 and v | 2.
Hence, u = 1 and v = 1, 2. This implies that uv = 1, 12 . Now f (1) 6= 0, f (1) 6= 0, f ( 12 ) = 14 72 + 1 6= 0,
and f ( 12 ) = 14 + 72 + 1 6= 0. So we find that f(x) has no root in Q. Thus, by Theorem 13.3.1, f (x) is irreducible
in Q[x]. Since f (x)is primitive, f (x) is also irreducible in Z[x].
Let us now give another criterion for irreducibility. This famous criterion is known as Eisensteins irreducibility criterion.
Theorem 13.3.7 (Eisensteins Irreducibility Criterion) Let D be a UFD and Q(D) be its quotient field.
Let
f (x) = a0 + a1 x + + an xn
be a nonconstant polynomial in D[x]. Suppose that D contains a prime p such that
(i) p | ai , i = 0, 1, . . . , n 1,
(ii) p - an , and
(iii) p2 - a0 .
Then f (x) is irreducible in Q(D)[x].
Proof. Case 1. f(x) is a primitive polynomial in D[x]. Under this assumption, if we can show that f (x) is
irreducible in D[x], then by Lemma 13.2.8, it will follow that f (x) is irreducible in Q(D)[x]. Suppose that f(x)
is not irreducible in D[x]. Then there exist polynomials
g(x)
h(x)

=
=

b0 + b1 x + + bt xt
c0 + c1 x + + ck xk

in D[x] such that f (x) = g(x)h(x) and g(x) and h(x) are nonunits in D[x]. Now n = t + k. If t = 0, then
g(x) = b0 , a nonunit element of D. Thus, f(x) = b0 h(x) implies that f (x) is not primitive. Therefore, t 6= 0.
Similarly, k 6= 0. Hence, 0 < t < n and 0 < k < n. Now from f (x) = g(x)h(x), we find that a0 = b0 c0 . Since p
is a prime such that p | a0 and p2 - a0 , it follows that p divides one of b0 , c0 , but not both. Suppose p | b0 and
p - c0 . Since p - an and an = bt ck , p - bt and p - ck . Thus, p | b0 and p - bt . Let m be the smallest positive integer
such that p - bm . Then p | bi for 0 i < m t. Now considering the coecient of xm in f (x) and g(x)h(x), it
follows that
am = b0 cm + b1 cm1 + + bm1 c1 + bm c0 .
Since p | bi , 0 i < m, we find that p | (am bm c0 ). Since m t < n, p | am . Hence, p | bm c0 and so p | bm or
p | c0 since p is prime. This is a contradiction. Therefore, f (x) is irreducible in D[x] and hence in Q(D)[x].
Case 2. f(x) is not a primitive polynomial in D[x]. Let d = gcd{a0 , a1 , . . . , an } in D. Then f (x) = df1 (x),
where f1 (x) is a primitive polynomial in D[x]. Let f1 (x) = d0 + d1 x + + dn xn . Then ai = ddi , for all i =
1, 2, . . . , n. Since p does not divide an , p does not divide d. Therefore, it now follows that p | di , i = 0, 1, . . . , n 1,
p - dn and p2 -d0 . Thus, by Case 1, f1 (x) is irreducible in Q(D)[x]. Now d is a unit in Q(D). Hence, f (x) is
irreducible in Q(D)[x].

13.3. Irreducibility of Polynomials

209

Corollary 13.3.8 Let D be a UFD and f (x) = a0 + a1 x + + an xn be a nonconstant primitive polynomial in

D[x]. Suppose that D contains a prime p such that
(i) p | ai , i = 0, 1, . . . , n 1,
(ii) p - an , and
(iii) p2 - a0 .
Then f (x) is irreducible in D[x].
Corollary 13.3.9 Let f (x) = a0 + a1 x + + an xn be a nonconstant polynomial in Z[x]. If there exists a prime
p such that
(i) p | ai , i = 0, 1, . . . , n 1,
(ii) p - an , and
(iii) p2 - a0 ,
then f (x) is irreducible in Q[x].
Corollary 13.3.10 The cyclotomic polynomial
p (x) = 1 + x + + xp1 =

xp 1
x1

is irreducible in Z[x], where p is a prime.

Proof. Since the content of p (x) is 1, we find that p (x) is a primitive polynomial. Suppose p (x) is not
irreducible in Z[x]. Then there exist nontrivial factors h(x) and g(x) of p (x) such that p (x) = h(x)g(x). This
implies that p (x + 1) = h(x + 1)g(x + 1) is a nontrivial factorization of p (x + 1). However,
p (x + 1)

=
=
=

(x+1)p 1
(x+1)1
xp +pxp1 ++(p
xi ++px
i)
px i1
p + + i x
+ +

pxp2 + xp1

is clearly irreducible by Eisensteins criterion. Hence, p (x) is irreducible in Z[x].

Gauss is said to have placed Eisenstein at the same mathematical level as Newton and Archimedes. However,
Eisensteins influence on mathematics is considered to be small in comparison to that of the giants of mathematics.

Worked-Out Exercises
Exercise 1 Show that f (x) = x3 + [2]x + [4] is irreducible in Z5 [x].
Solution: f ([0]) = [4], f ([1]) = [7] = [2], f ([2]) = [3]+[4]+[4] = [1], f ([3]) = [2]+[1]+[4] = [2], f([4]) = [4]+[3]+[4] =
[1]. Hence, f (x) has no roots in Z5 . Thus, by Theorem 13.3.1, f (x) is irreducible in Z5 [x].
Exercise 2 Let f (x) = x6 + x3 + 1 Z[x]. Show that f(x) is irreducible over Q.

Solution: Now f (x + 1) = x6 + 6x5 + 15x4 + 21x3 + 18x2 + 9x + 3. Let p = 3. Then by Eisensteins criterion, f (x + 1)
is irreducible over Q. Hence, f(x) is irreducible over Q.

Exercise 3 Show that f (x) = x4 5x2 + x + 1 is irreducible in Z[x].

Solution: Let us first show that f (x) is irreducible in Q[x]. If f(x) has a linear factor, then f (x) has a root in Q. Let
a
(a, b are relatively prime) be a root of f(x) in Q. Then b | 1 and a | 1 by Theorem 13.3.5. Hence, ab = 1
b
or 1. But f (1) = 1 5 + 1 + 1 = 2 6= 0 and f (1) = 1 5 1 + 1 = 4 6= 0. Therefore, f (x) has no
linear factors in Q[x]. Let f (x) = (x2 + ax + b)(x2 + cx + d) in Z[x]. Equating coecients of powers of x,
we find that
c + a = 0, d + b + ac = 5, ad + bc = 1, bd = 1.
Now bd = 1 implies that either b = d = 1 or b = d = 1. Suppose b = d = 1. Then a + c = 1. But we
also have a + c = 0, a contradiction. Suppose b = d = 1. Then ad + bc = 1 implies that a + c = 1.
Thus, a + c = 1 and a + c = 0, a contradiction. Hence, we find that there are no integers a, b, c, d such
that f (x) = (x2 + ax + b)(x2 + cx + d). This also implies that f(x) cannot be factored as a product of two
quadratic polynomials in Q[x] (see Worked-Out Exercise 1, page 206). Thus, f (x) is irreducible in Q[x].
Hence, by Lemma 13.2.8, f (x) is irreducible in Z[x].
Exercise 4 Show that f (x) = x5 + 15x3 + 10x + 5 is irreducible in Z[x].

210

13. Unique Factorization Domains

Solution: The content of f (x) is 1. Therefore, f(x) is a primitive polynomial. Now 5 is a prime integer and 5 | 5,
5 | 10, 5 | 0, 5 | 15, 5 - 1, 52 - 5. Hence, by Corollary 13.3.8, f(x) is irreducible in Z[x].

Exercise 5 Give an example of a primitive polynomial which has no root in Q, but is reducible over Z.

Solution: Let f (x) = x4 + 2x2 + 1. This is a primitive polynomial in Z[x]. If possible, let ab be a root of f (x), where
a 6= 0, b 6= 0 and gcd(a, b) = 1. Then a | 1 and b | 1 by Theorem 13.3.5. Hence, ab = 1. But f(1) 6= 0 and
f (1) 6= 0. Therefore, f (x) has no root in Q. Since f (x) = (x2 + 1)(x2 + 1), f (x) is reducible in Z[x].

Exercise 6 Show that x2 + x + [1] is the only irreducible polynomial of degree 2 over Z2 .

Solution: Any polynomial of degree 2 over Z2 is of the form ax2 + bx + c, where a, b, c Z2 = {[0], [1]}. Now a 6= [0].
Therefore, a = [1]. Then x2 , x2 + x, x2 + [1], and x2 + x + [1] are the only polynomials of degree 2 over
Z2 . Now x2 = xx, x2 + x = x(x + [1]), and x2 + [1] = (x + [1])(x + [1]) showing that x2 , x2 + x, and
x2 + [1] are reducible. Let f (x) = x2 + x + [1]. Then f ([0]) = [1] 6= 0 and f([1]) = [3] = [1] 6= 0. Therefore,
f (x) has no root in Z2 . Thus, x2 + x + [1] is irreducible over Z2 .

Exercises
1. Find all irreducible polynomials of degree 2 in Z2 [x]. Is x3 + [1] irreducible in Z2 [x]? If not, then express
it as a product of irreducible polynomials in Z2 [x].
2. Show that the polynomial x5 + x2 + [1] is irreducible in Z2 [x]. Hence, prove that x5 x2 + 9 is irreducible
in Z[x].
3. Show that the polynomial x2 + [2]x + [6] is reducible in Z2 [x] even though x2 + 2x + 6 is irreducible in Z[x].
4. Use Eisensteins criterion to prove that the polynomials x2 + 2x + 6 and 2x4 + 6x3 9x2 + 15 are irreducible
over Z.
5. For f (x) D[x], D a UFD, prove that f (x) is irreducible in D[x] if and only if f (x c) is irreducible in
D[x] for any c D.
6. Show that the polynomials x3 x2 + 1, x3 x + 1, and x3 + 2x2 + 3 are irreducible in Z[x].

7. Show that the polynomial 2x3 x2 + 4x 2 is not irreducible in Z[x].

8. Show that the polynomial x2 + 13 x

2
5

is irreducible in Q[x].

9. Prove that the polynomial f(x) = 1 x + x2 x3 + + (1)p1 xp1 is irreducible in Z[x] for any prime
p.

10. Let D be a UFD and f (x) = a0 + a1 x + + an xn D[x] be of degree n and a0 6= 0. Let uv 1 Q(D)
be a root of f (x), where u, v D and gcd(u, v) = 1. Prove that u | a0 and v | an in D.

11. Show that for any positive integer n > 1, f (x) = xn + 2 is irreducible in Z[x].
12. Find all irreducible polynomials of degree 2 over the field Z3 .

13. If f(x) is an irreducible polynomial over R, prove that either f (x) is linear or f (x) is quadratic.
14. Show that there are only three irreducible monic quadratic polynomials over Z3 .
15. (i) Show that there are only 10 irreducible monic quadratic polynomials over Z5 .
(ii) Let p be a prime. Find the number of irreducible monic quadratic polynomials over Zp .

13.3. Irreducibility of Polynomials

211

Leopold Kronecker (18231891) was born on December 7, 1823, in Liegnitz, Germany, to a wealthy family.
He was provided with private tutoring at home. He later entered Liegnitz Gymnasium, where E. E. Kummer
was his mathematics teacher. Kummer recognized his talent and encouraged him to do independent research.
In 1841, he matriculated at the University of Berlin. There he attended Dirichlets and Steiners mathematics
lectures. He was also attracted to astronomy and in 1843 attended the University of Bonn. He returned to Berlin
in 1845, the year he received his Ph.D. His thesis was on complex units.
On Kummers nomination, Kronecker became a full member of the Berlin Academy in 1861. He was very influential at the Academy and personally helped fifteen mathematicians, including Riemann, Sylvester, Dedekind,
Hermite, and Fuchs, to get various memberships.
Kroneckers primary work is in algebraic number theory. He is believed to be one of the inventors of algebraic
number theory along with Kummer and Dedekind. He was the first mathematician who clearly understood
Galoiss work. He also proved the fundamental theorem of finite Abelian groups.
Briefly Kronecker withdrew from academic life to manage the family business. However, he continued to do
mathematics as a recreation. In 1855, he returned to the academic life in Berlin. In 1880, he became editor of
the Journal fr die reine and angewandte Mathematik.
Kronecker and Weierstrass were good friends. While Weierstrass and Cantor were creating modern analysis,
Kroneckers remark that God himself made the whole numberseverything else is the work of men deeply
aected Cantor, who was very sensitive. His remarks in opposition to Cantors work are believed to be a factor
in Cantors nervous breakdown.
Kronecker died on December 29, 1891.

212

13. Unique Factorization Domains

Chapter 14

Maximal, Prime, and Primary Ideals

14.1

Maximal, Prime, and Primary Ideals

In this section, we introduce certain special ideals. These ideals are motivated in large part by certain arithmetic
properties of the integers. Throughout the section, we assume that the ring R contains at least two elements.
Definition 14.1.1 An ideal P of a ring R is called prime if for any two ideals A and B of R, AB P implies
that either A P or B P.
The following theorem gives a useful characterization of a prime ideal with the help of elements of R. Let us
first recall that if A is a left ideal and B is a right ideal of a ring R, then AB is an ideal of R. Let a R. Then
Ra is a left ideal of R and aR is a right ideal of R. Thus, R(aR) is an ideal of R. We denote R(aR) by RaR.
Also, for a R, aRa = {ara | r R}.
Theorem 14.1.2 An ideal P of a ring R is a prime ideal if and only if for all a, b R, aRb P implies that
either a P or b P.
Proof. Suppose P is a prime ideal and aRb P, where a, b R. Let A = RaR and B = RbR. Then A and
B are ideals of R. Also, AB = (RaR)(RbR) R(aRb)R RP R P. Since P is a prime ideal, it follows that
either A P or B P. Suppose A P. Now hai3 RaR = A P. Since P is a prime ideal, hai P and so
a P. Similarly, if B P, then b P. Thus, either a P or b P. Conversely, suppose that the ideal P satisfies
the given condition of the theorem. Let A and B be two ideals of R such that AB P. Suppose that A 6 P.
Then there exists a A such that a
/ P. Let b B. Now aRb = (aR)b AB P. This implies that a P or
b P. But a
/ P. Therefore, b P. Hence, B P.
Corollary 14.1.3 Let R be a commutative ring. An ideal P of R is a prime ideal if and only if for all a, b R,
ab P implies that either a P or b P.
Example 14.1.4 In the ring Z of integers, the ideal P = {3k | k Z} is a prime ideal. For, ab P if and only
if ab is divisible by 3 if and only if a is divisible by 3 or b is divisible by 3 (since 3 is prime) if and only if a is a
multiple of 3 or b is a multiple of 3 if and only if a P or b P. In Z, the ideal J = {6k | k Z} is not a prime
ideal since 3 2 = 6 J, but 3
/ J and 2
/ J.
Theorem 14.1.5 Let R be a PID and P be a nonzero ideal of R. Then P is prime and P 6= R if and only if P
is generated by a prime element.
Proof. Let R be a PID and P = hpi be a nonzero proper prime ideal of R. Then p 6= 0. Since P 6= R, p is
not a unit. Let a, b R be such that p | ab. Then ab = pc for some c R. Hence, ab P. Since P is a prime
ideal, either a P or b P. Therefore, either p | a or p | b. Thus, p is a prime element. Conversely, suppose that
P = hpi is a nonzero ideal of R such that p is a prime element. Since p is not a unit, P 6= R. Let a, b be two
elements of R such that ab P. Then p | ab. Since p is a prime element, either p | a or p | b. Therefore, either
a P or b P. Hence, P is a prime ideal of R.
As a consequence of Theorem 14.1.5 and Theorem 12.1.9, the prime ideals of Z are precisely those ideals
generated by primes and the ideals {0} and Z. Also, by Theorem 12.3.16, the prime ideals in the polynomial ring
F [x] over a field F are those ideals generated by irreducible polynomials and the ideals {0} and F [x].

213

214

14. Maximal, Prime, and Primary Ideals

Definition 14.1.6 Let R be a ring and M be a (left, right) ideal of R. Then M is called a maximal (left,
right) ideal of R if M 6= R and there does not exist any (left, right) ideal I of R such that M I R.
Theorem 14.1.7 Let R be a commutative ring with 1. Then every maximal ideal of R is a prime ideal of R.
Proof. Let I be a maximal ideal of R and a and b be two elements of R such that ab I and a
/ I.
Now hI, ai = {u + ra | u I, r R} is the ideal generated by I {a}. Since a
/ I, I hI, ai . Also, since
I is a maximal ideal, hI, ai = R. Thus, there exist u I and r R such that 1 = u + ra. This implies that
b = ub + rab I. Hence, I is a prime ideal.
The converse of the above theorem is not true, as shown by the following examples.
Example 14.1.8 In the ring Z of integers, {0} is a prime ideal, but not a maximal ideal.
Example 14.1.9 Let R = {(a, b) | a, b Z}. Then (R, +, ) is a ring, where + and are defined by
(a, b) + (c, d)
(a, b) (c, d)

=
=

(a + c, b + d),
(ac, bd)

for all a, b, c, d Z. Let I = {(a, 0) | a Z}. Then I is a prime ideal of R, but not a maximal ideal since
I hI, (0, 2)i R.
Theorem 14.1.10 Let R be a principal ideal domain. Then a nonzero ideal P (6= R) of R is prime if and only
if it is maximal.
Proof. Suppose P (6= R) is a nonzero prime ideal. By Theorem 14.1.5, P = hpi for some prime element
p R. We now show that there is no ideal I of R such that P I R. Suppose I is an ideal of R such that
P I. Since P 6= I, there exists an element a I such that a
/ P. Then a and p are relatively prime and so
there exist s, t R such that 1 = sa + tp. Since sa I and tp P I, we must have 1 I. This implies that
I = R. Hence, P is maximal.
We now give characterizations of prime ideals and maximal ideals in a commutative ring with identity by the
quotient rings of the ideals.
Theorem 14.1.11 Let R be a commutative ring with 1 and P be an ideal of R such that P 6= R. Then P is a
prime ideal if and only if R/P is an integral domain.
Proof. Let P be a prime ideal of R. Since R is a commutative ring with 1, the quotient ring R/P is also a
commutative ring with 1. Now P 6= R and so the identity element 1 + P of R/P is dierent from the zero element
0 + P. Let us now show that R/P has no zero divisors. Let a + P, b + P R/P, and (a + P )(b + P ) = 0 + P.
Then ab + P = 0 + P, which implies that ab P. Since P is a prime ideal, either a P or b P, i.e., either
a + P = 0 + P or b + P = 0 + P. Thus, R/P has no zero divisors. This implies that R/P is an integral domain.
Conversely, suppose R/P is an integral domain. Let ab P. Then 0 + P = ab + P = (a + P )(b + P ), whence
a + P = 0 + P or b + P = 0 + P. Thus, a P or b P and so P is a prime ideal.
Theorem 14.1.12 Let R be a commutative ring with 1 and M be an ideal of R. Then M is a maximal ideal if
and only if R/M is a field.
Proof. Suppose that M is a maximal ideal. Since R is a commutative ring with 1, R/M is a commutative
ring with 1. For all a R, let a denote the coset a + M in R/M. Let a R/M be such that a 6= 0. Then a
/
M. Hence, the ideal hM, ai generated by M {a} properly contains M. Since M is a maximal ideal, we have
hM, ai = R. This implies that there exist m M and r R such that m + ra = 1. Thus, m + ra = 1 and
so ra = 1. Hence, a has an inverse. This shows that every nonzero element of R/M is a unit and so R/M is
a field. Conversely, suppose R/M is a field. Since R/M is a field, R 6= M. Let I be an ideal of R such that
M I R. There exists a I such that a
/ M. Then a 6= 0 and so there exists r R/M such that ar = 1.
Thus, (a + M)(r + M) = 1 + M, which implies 1 ar M. Hence, 1 = m + ar for some m M. Thus,
1 = m + ar M + I I. This implies that I = R. Therefore, M is maximal.
As a consequence of Theorems 12.1.9 and 14.1.10, the maximal ideals of Z are precisely those ideals generated
by primes. Also, by Theorem 12.3.16, the maximal ideals in the polynomial ring F [x] over a field F are those
ideals generated by irreducible polynomials.

14.1. Maximal, Prime, and Primary Ideals

215

Example 14.1.13 Consider the polynomial ring R[x, y] over an integral domain R. Then R[x, y]/ hxi ' R[y]
and R[x, y]/ hyi ' R[x], which are integral domains. Thus, hxi and hyi are prime ideals. Since R[x, y]/ hxi and
R[x, y]/ hyi are not fields, hxi and hyi are not maximal ideals.
Example 14.1.14 Consider E, the ring of even integers. The ideal h4i is maximal, but not prime in E since
2 2 h4i , but 2
/ h4i . Note that E is commutative without identity.
We now show the existence of maximal ideals in certain rings. In order to accomplish this, we require Zorns
lemma.
Theorem 14.1.15 Let R be a commutative ring with 1. Then every proper ideal of R is contained in a maximal
ideal of R.
Proof. Let I be a proper ideal of R and set A = {J | I J, J is a proper ideal of R}. Since I A, A 6= .
Also, A is a partially ordered set, where the partial order is the usual set inclusion. We now show that any
chain in A has an upper bound in A. Let C = {J | K} be a chain in A. Since I J for all , I J .
Let a, b J . Then a J and b J for some , . Since C is a chain, either J J or J J , say,
J J . Thus, a, b J . Since J is an ideal of R, a b J J . Let r R. Then ra J J ,
whence J is an ideal of R. Now J 6= R else 1 J for some , which is impossible since J 6= R. Hence,
J A, which is clearly an upper bound of C and so by Zorns lemma, A has a maximal element, say, M. We
now show that M is a maximal ideal. If there exists an ideal J of R such that M J R, then J A and so
M is not maximal in A, a contradiction. Thus, no such J exists and so M is a maximal ideal.
Corollary 14.1.16 Let R be a commutative ring with 1 and a R. Then a is in a maximal ideal of R if and
only if a is not a unit.
Proof. Suppose a is not a unit. Then hai R else 1 = ra for some r. By Theorem 14.1.15, there exists a
maximal ideal M such that hai M. Now a hai M. Conversely, suppose a M, where M is a maximal
ideal. If a is a unit, then 1 = a1 a M and so M = R, a contradiction.
Corollary 14.1.17 Let R be a commutative ring with 1. Then R has a maximal ideal.
Proof. In R, {0} is a proper ideal. Hence, by Theorem 14.1.15, there exists a maximal ideal M of R such
that {0} M.
The fundamental theorem of arithmetic says that any integer n has a prime factorization n = pe11 pess ,
where p1 , . . . , ps are primes and e1 , . . . , es are positive integers. The ideals hpi i are prime ideals of Z. The ideals
hpei i i are also special ideals of Z. Their study is motivated in part by the fundamental theorem of arithmetic.
Definition 14.1.18 Let R be a commutative ring and Q be an ideal of R. Then Q is called a primary ideal if
for all a, b R, ab Q and a
/ Q implies that there exists a positive integer n such that bn Q.
From the definition of primary ideal, it follows immediately that every prime ideal in a commutative ring is
a primary ideal. Now in the ring Z, for any prime integer p, the ideal hpn i contains pn but not p, where n is a
positive integer and n 2. Hence, hpn i is not a prime ideal. The following example shows that hpn i is a primary
ideal.
Example 14.1.19 Let p be a prime in Z and n be a positive integer. We show that hpn i is a primary ideal. Let
ab hpn i and a
/ hpn i . Then there exists r Z such that ab = rpn . Since pn does not divide a, p | b and so
b = qp for some q Z. Thus, bn = q n pn and so bn hpn i .
Example 14.1.20 Let p(x) be irreducible in F [x], F a field, and n be a positive integer. Then hp(x)n i is a
primary ideal by an argument entirely similar to the one used in Example 14.1.19.
Definition 14.1.21 Let R be a commutative ring and I be an ideal of R. Then the radical of I, denoted by
is defined to be the set

I = {a R | an I for some positive integer n}.

Theorem
ideal of a commutative ring R. Then
14.1.22 Let Q be an
(i) Q is an ideal of R and Q
Q,
(ii) if Q is a primary ideal, then Q is a prime ideal.

I,

216

14. Maximal, Prime, and Primary Ideals

n m
Proof. (i) Clearly Q Q. Let a,b Q. Then there exist positive integers n, m such
that a , b Q.
n+m
Thus, (a b)
Q and so a b Q. Let r R. Then (ra)n = rn an Q and so ra Q. Hence, Q is
an ideal of R.

(ii) Let a, b R be such that ab Q and a

/ Q. There exists a positive integer n such that an bn =
n
n
(ab) Q. But
/ Q. Since
Q is primary, there exists a positive integer m such that bnm = (bn )m Q.
a

Therefore, b Q and so Q is prime.

Definition 14.1.23 Let Q be a primary ideal of a commutative ring R. Then the radical P = Q of Q is called
the associated prime ideal of Q and Q is called a primary ideal belonging to (or primary for) the prime
ideal P.


Example 14.1.24 Let i be a positive integer. In Z, we show that pi is primary for hpi , where p is a prime.
s
s

It suces to show that hpi = hpi i. Let a hpi i. Then there exists a positive integer n such that an pi .
s
Therefore, an = rpi for some r Z. This implies that p | a and so a hpi . Hence, shpi i hpi . Let a s
hpi .
i
i
i i
i i. Thus, hpi
Then there exists t Z such that a = tp. This implies that
a
and
so
a

=
t
p

p
hp
hpi i.


In F [x] (F a field), a similar argument shows that p(x)i is primary for hp(x)i , where p(x) is irreducible
s
and hp(x)i = hp(x)i i.
Theorem 14.1.25 Let Q and P be ideals of a commutative ring R. Then Q is primary and P =
only if

(i) Q P Q and
(ii) ab Q, a
/ Q implies b P.

Q if and

Proof.
and (ii) hold. Let ab Q, a
/ Q. Then
The necessity of (i) and (ii) is immediate. Suppose (i)
n
b P Q and so there
Q is primary. We now show
exists a positive integer n such that b Q, whence
that P = Q. Let b Q. Then there exists a positive integer n such that bn Q P. Let n be the smallest
n
positive integer such that b
Q. If n = 1, thenb P. So assume that n 2. Then bbn1 Q and bn1
/Q
implies that b P. Hence, Q P and so P = Q.
We now show that every primary ideal I of a commutative ring R can be characterized with the help of some
properties of the quotient ring R/I.
Theorem 14.1.26 Let R be a commutative ring and I be an ideal of R. Then I is a primary ideal if and only
if every zero divisor of R/I is nilpotent.
Proof. First suppose that I is a primary ideal. Let a + I be a zero divisor in R/I. Then there exists an
element b + I R/I, b + I 6= I, such that (a + I)(b + I) = I. Now ab I and b
/ I. Since I is a primary ideal, it
follows that an I for some positive integer n. Hence, (a + I)n = an + I = I, showing that a + I is nilpotent.
Conversely, suppose that every zero divisor of R/I is nilpotent. Let a, b R be such that ab I and a
/ I.
Then a+I 6= I. Now (a+I)(b +I) = ab+I = I. If b +I = I, then b I. Suppose b +I 6= I. This implies that b+I
is a zero divisor and so is nilpotent. Therefore, there exists a positive integer n such that bn + I = (b + I)n = I.
Thus, bn I. Consequently, I is a primary ideal.
Consider Z. For the prime factorization of an integer n, n = pe11 pess , we have

hni = hpe11 i hpess i = hpe11 i hpess i

s e
hpii i = hpi i , i = 1, 2, . . . , s. However, in the polynomial ring Z[x, y], it can be shown that the ideal
and
x2 , xy, 2 is an intersection of primary ideals, but not a product of primary ideals. These concepts involving
prime and primary ideals are used in the study of nonlinear equations. For example, consider the following
nonlinear equations:
x2 y = 0
x2 z = 0.

2

2 
2
2
In the polynomial ring
2R[x, y], let2 I = x y, x z . It can be shown that x y, z and x , y are primary
ideals and that I = x , y x y, z . In fact, it can be shown in any polynomial ring F [x1 , . . . , xn ] over
a field F that every ideal is a finite intersection of primary ideals.
This latter
s
s of fundamental
result is a type
theorem of arithmetic for ideals. It can also be shown that hx2 y, zi = x2 y, z and hx2 , yi = hx, yi .
The solution to the above system of equations is
{(x, x2 , 0) | x R} {(0, 0, z) | z R}.

2
The ideal x y, z corresponds to {(x, x2 , 0) | x R}, while the ideal hx, yi corresponds to {(0, 0, z) | z R}.
We conclude this section by mentioning the following dierences between the ideals of Z and Z[x].

14.1. Maximal, Prime, and Primary Ideals

217

1. In the ring Z, every ideal is a principal ideal, but in Z[x] there exist ideals (for example, hx, 2i), which are
not principal.
2. In the ring Z, a nontrivial ideal is a prime ideal if and only if it is a maximal ideal. In the ring Z[x], there
are prime ideals (for example hxi), which are not maximal.

3. In the ring Z, a nontrivial ideal I is a primary ideal if and only if I = hpn i for some prime p and for some
positive integer n. Hence, in Z, if I is a primary ideal, then I is expressible as some power of its associated
prime ideal. In Z[x], this is not true, as hx, 4i is a primary ideal with hx, 2i as its associated prime ideal,
but hx, 4i 6= hx, 2in for any n 1.

Worked-Out Exercises
Exercise 1 Let R be an integral domain. Prove that if every ideal of R is a prime ideal, then R is a field.

Solution: Let 0 6= a R. Then a2 R is an ideal of R and hence it is a prime ideal. Now a2 a2 R. Since a2 R is a
prime ideal, a a2 R. Thus, a = a2 b for some b R. Then a(1 ab) = 0. Since R is an integral domain
and a 6= 0, 1 ab = 0 and so ab = 1, proving that a is a unit. Hence, R is a field.

Exercise 2 Let R be a commutative ring with 1. Suppose that hxi is a prime ideal of R[x]. Show that R is an integral
domain.
Solution: Since hxi is a prime ideal R[x]/ hxi is an integral domain. Since R[x]/ hxi ' R, R is an integral domain.

Exercise 3 Let R be a commutative ring and I be an ideal of R. Let P be a prime ideal of I. Show that P is an ideal
of R.
Solution: Let a P I and r R. Then rar I. Therefore, a(rar) P and so (ar)2 P. Since P is a prime ideal
of I, ar P. Hence, P is an ideal of R.

Exercise 4 Show that a proper ideal I of a ring R is a maximal ideal if and only if for any ideal A of R either A I
or A + I = R.

Solution: Suppose I is a maximal ideal of R and let A be any ideal of R. If A 6 I, then A + I is an ideal of R such
that I A + I. Since I is maximal, it follows that A + I = R.
Conversely, assume that the proper ideal I satisfies the given condition. Let J be an ideal of R such that
I J. Now J 6 I. Therefore, I + J = R. But I + J = J. Thus, J = R. Hence, I is a maximal ideal of R.
Exercise 5 Let R be a PID which is not a field. Prove that any nontrivial ideal I of R is a maximal ideal if and only
if it is generated by an irreducible element.
Solution: Since R is not a field, there exists an element 0 6= a R such that a is not a unit. Then h0i hai R.
Therefore, h0i is not a maximal ideal. Let I be a maximal ideal of R. Then I 6= {0} and I = hpi for
some p R, where p is irreducible by Theorem 14.1.5 and Corollary 12.3.13. Conversely, let I = hpi and
p be irreducible. Let I J R. Since R is a PID, J = hai for some a R. Since p hai , a divides p.
Thus, p = ab for some b R. Since p is irreducible, either a is a unit or b is a unit. If b is a unit, then
a = pb1 hpi . Thus, J I, which is a contradiction. Hence, a is a unit and so J = R. Thus, I is a
maximal ideal.
Exercise 6 Show that the ideal hxi in Z[x] is a prime ideal, but not a maximal ideal.

Solution: Let f(x) = a0 + a1 x + + an xn and g(x) = b0 + b1 x + + bm xm be two elements in Z[x] such that
f (x)g(x) hxi . Then a0 b0 = 0. Thus, either a0 = 0 or b0 = 0. Hence, either f (x) hxi or g(x) hxi ,
showing that hxi is a prime ideal. Now the ideal hx, 2i of Z[x] is such that hxi hx, 2i Z[x]. Hence, hxi
is not a maximal ideal.

Exercise 7 Let R be a commutative ring with 1. Let A and B be two distinct maximal ideals of R. Show that
AB = A B.

Solution: Since AB A and AB B, AB A B. Since A and B are distinct maximal ideals, there exists b B
such that b
/ A. Then hA, bi = {a + br | a A, r R} is an ideal of R such that A hA, bi . Since A
is maximal, hA, bi = R. This implies that 1 = a + br for some a A and r R. Let x A B. Then
x = x1 = xa + xbr = xa + (xb)r AB. Hence, A B AB. Thus, AB = A B.

Exercise 8 Let f (x) = x5 + 12x4 + 9x2 + 6. Show that the ideal I = hf (x)i is maximal in Z[x].

Solution: I will be a maximal ideal if we can prove that f(x) is an irreducible polynomial in Z[x]. The content of
f (x) is 1. Hence, f (x) is a primitive polynomial in Z[x]. Also, for the prime 3, we find that 3 | 6, 3 | 9,
3 | 12, 3 - 1, 32 - 6. Hence, f(x) is irreducible in Z[x], by Eisensteins criterion.

218
Exercise 9
Solution:

14. Maximal, Prime, and Primary Ideals

(a) Find all maximal ideals of the ring Z6 .
(b) ind all ideals and all maximal ideals of the ring Z8 .
(a) The mapping : Z Z6 defined by (n) = [n] is a homomorphism of Z onto Z6 and Ker = 6Z. If
I is any ideal of Z6 , then there exists a unique ideal A of Z such that Ker A and (A) = I. Now
Z, 2Z, 3Z, and 6Z are the only ideals of Z which contain 6Z. Also, (Z) = Z6 , (2Z) = {[0], [2], [4]},
(3Z) = {[0], [3]}, and (6Z) = {[0]}. Hence, {[0], [2], [4]} and {[0], [3]} are the only maximal ideals
of Z6 since 2Z and 3Z are maximal ideals of Z.

(b) The mapping : Z Z8 defined by (n) = [n] is an epimorphism of rings and Ker = 8Z. Now Z,
2Z, 4Z, and 8Z are the only ideals of Z which contain 8Z. Also, (Z) = Z8 , (2Z) = {[0], [2], [4], [6]},
(4Z) = {[0], [4]}, and (8Z) = {[0]}. Hence, the ideals of Z8 are Z8 , {[0], [2], [4], [6]}, {[0], [4]}, and
{[0]}. Now {[0]} {[0], [4] } {[0], [2], [4], [6]} Z8 . This implies that Z8 has only one maximal
ideal, which is {[0], [2], [4], [6]}.

Exercise 10 Show that x2 is a primary ideal in Z[x] with hxi as its associated prime ideal.

n
m
Solution: Let f(x) = a0 +
 a1 x + + a
n x and g(x) = b0 + b1 x + + bm x be two elements in Z[x] such that
/ x 2 . Then f (x)g(x) = x2 h(x) for some h(x) Z[x]. Hence, a0 b0 = 0 and
f (x)g(x) x2 and f (x)
/ x2 , it follows that either a0 6= 0 or a1 6= 0. If a0 6= 0, then b0 = 0 and
a0 b1 + a1 b0 = 0. Since f (x)

2
b1 = 0 and so g(x) x . If a0 = 0, then a1 6= 0. Hence,
a0 b1 + a1 b0 = 0 shows that b0 = 0. So we find

that b20 = 0, b0 b1 + b1 b0 = 0 and thus (g(x))2 x2 . Hence, x2 is a primary ideal. Now x2 hxi
s

and f (x) hx2 i if and only if (f (x))n x2 for some positive integer n. This is true if and only if the
constant term of f(x) is zero, i.e., if and only if f (x) hxi .

Exercise 11 Show that a commutative ring R with 1 is isomorphic to a subdirect sum of a family of fields if and only
if the intersection of all maximal ideals of R is {0}.
Solution: Suppose R is isomorphic to a subdirect sum of a family of fields {Fi | i I}. Then there exists a subring
T of iI Fi such that T = siI Fi and R ' T. Let : R T be an isomorphism. Then i : R Fi
is an epimorphism for all i I, where i is the ith canonical projection. Proceeding as in the proof of
Theorem 10.1.14, we can show that
iI Ai = {0},

where Ai =Ker i for all i I. Now R/Ai ' Fi . Since Fi is a field, Ai is a maximal ideal for all i I.
If A is the intersection of all maximal ideals of R, then A iI Ai = {0}. Hence, A = {0}. Conversely,
suppose that A = {0}, where A = iJ {Mi | Mi is a maximal ideal of R}. By Theorem 10.1.14, R is
monomorphic to the subdirect sum of a family of rings {R/Mi | i J}. Since each Mi is a maximal ideal,
we find that R/Mi is a field.

Exercises
1. Find all maximal and prime ideals of Z10 .
2. Prove that I = {(5n, m) | n, m Z} is a maximal ideal of ZZ.

3. Find all ideals and maximal ideals of Zpk , where p is a prime and k is a positive integer.
4. Let I = {a0 + a1 x + + an xn Z[x] | 3 divides a0 }. Show that I is a prime ideal of Z[x]. Is I a maximal
ideal?
5. Let I be an ideal of a ring R. Prove that the following conditions are equivalent.
(i) I is a prime ideal.
(ii) If a, b R\I, then there exists c R such that acb R\I.

6. Let R be a finite commutative ring with 1. Show that in R, every prime ideal I 6= R is a maximal ideal.

7. Let R be a Boolean ring. Prove that a nonzero proper ideal I of R is a prime ideal if and only if it is a
maximal ideal.
8. Let R be a ring with 1. Prove that a nonzero proper ideal I of R is a maximal ideal if and only if the
quotient ring R/I is a simple ring.
9. Let I be an ideal of a ring R. If P is a prime ideal of the quotient ring R/I, prove that there exists a prime
ideal J of R such that I J and J/I = P.

10. Let R be a commutative ring with 1. Prove that there exists an epimorphism from R onto some field.

11. Let I be an ideal of a ring R with 1. Prove that the quotient ring R/I is a division ring if and only if I is
a maximal right ideal.

14.1. Maximal, Prime, and Primary Ideals

219

12. For all r R, show that Ir = {f (x) R[x] | f (r) = 0} is a maximal ideal of R[x] and R[x]/Ir ' R. Also,
prove that rR Ir = {0}.
13. Consider the polynomial ring K[x] over a field K. Let a K. Define the mapping a : K[x] K by
a (f (x)) = f (a) for all f (x) K[x]. Show that a is an epimorphism and Ker a is a maximal ideal of
K[x].
14. Let R be a PID.
(i) Prove that every nonzero nonunit element is divisible by a prime element.
(ii) If {In }nN is a sequence of ideals of R such that I1 I2 In , prove that there exists a
positive integer n such that In = In+1 = .
(iii) Prove that every nonzero nonunit can be expressed as a finite product of prime elements.
15. Let {I } be a collection of prime ideals in a commutative ring R such that {I } forms a chain. Prove that
I and I are prime ideals of R.

16. If I1 and I2 are ideals of a commutative ring 1, prove that I1 I2 = I1 I2 .

n
. . . , n, be ideals
17. Let
R be a commutative ring with 1 and Qi , i = 1, 2,
in R. Set Q = i=1 Qi . Prove that if
Qi = P for some ideal P of R, i = 1, 2, . . . , n, then Q = P. If Qi = P, i = 1, 2, . . . , n, and each Qi is
primary, prove that Q is primary.

18. If I is an ideal of a commutative ring R with 1 such that I is a maximal ideal, prove that I is a primary
ideal.

19. In the polynomial ring Z[x], prove the following.

(i) I = {f(x) Z[x] |the constant term of f (x) is divisible by 4} is a primary ideal with J = hx, 2i as its
associated prime ideal.
(ii) The ideal hx, 6i is not a primary ideal.
20. Prove that every prime ideal is a primary ideal in a commutative ring.
21. Let M be an ideal of a commutative ring R. Prove that R/M is a field if and only if M is a maximal ideal
and x2 M implies x M for all x R.
22. Prove that in a PID every nontrivial ideal I can be expressed as a finite product of prime ideals I = P1 Pn
such that P1 , P2 , . . . , Pn are determined uniquely up to order.
23. An ideal P of a ring R is called a semiprime ideal if for any ideal I of R, I 2 P implies that I P.
(i) Prove that an ideal P of R is a semiprime ideal if and only if the quotient ring R/P contains no nonzero
nilpotent ideals.

(ii) If R is a commutative ring with 1, prove that an ideal P of R is a semiprime ideal if and only if P = P.
24. A commutative ring R with 1 is called a local ring if R has only one maximal ideal. Prove the following.
(i) Z8 and Z9 are local rings.
(ii) In a local ring, all nonunits form a maximal ideal.
(iii) In a local ring R, for all r, s R, r + s = 1 implies either r is a unit or s is a unit.

25. Let p be a prime integer and Qp = { ab Q | p does not divide b}. Show that Qp is a local ring under the
usual addition and multiplication of rational numbers.

26. Let R be a field and T be the set of all sequences {an } of elements of R. Then (T, +, ) is a ring, where +
and are defined as in Worked-Out Exercise 6 (page 193). Prove the following.
(i) The set I of all nonunits of T is a maximal ideal of T.
(ii) I is the only maximal ideal of T.
(iii) T is a local ring.
27. Let R = R1 R2 Rn be the direct sum of the finite family of rings {R1 , R2 , . . . , Rn }, where each
Ri contains an identity. Prove the following:
(i) If Mi is a maximal ideal of Ri (1 i n), then R1 R2 Ri1 Mi Ri+1 Rn is a
maximal ideal of R.
(ii) Every maximal ideal M of R is of the form
R1 R2 Ri1 Mi Ri+1 Rn ,
where Mi is a maximal ideal of Ri for some i (1 i n).

220

14. Maximal, Prime, and Primary Ideals

28. Show that the ring Z is isomorphic to a subdirect sum of a family of fields.
29. An ideal I of a ring R is called a minimal ideal if I 6= {0} and there does not exist any ideal J of R such
that {0} 6= J I. If I is a minimal ideal of a commutative ring R with 1, prove that either I 2 = {0} or
I = eR for some idempotent e R.

30. In the following exercises, write the proof if the statement is true; otherwise, give a counterexample.
(i) Let R be a commutative ring with 1 and P be a prime ideal of R such that P 6= R. If the quotient ring
R/P contains a finite number of elements, then R/P is a field.
(ii) In a PID dierent from a field, there exists a prime element.
(iii) In a PID, every proper prime ideal is a maximal prime ideal.
(iv) The intersection of two prime ideals of a ring R is a prime ideal of R.
(v) If I is a prime ideal of a ring R, then I[x] is also a prime ideal of R[x].
(vi) If I is a maximal ideal of a ring R, then I[x] is also a maximal ideal of R[x].
(vii) A commutative ring with 1 and with only a finite number of maximal ideals is a field.
(viii) In the ring Z, the ideal h5i is a maximal ideal, but in the ring Z[i], the ideal h5i is not a maximal
ideal.

Chapter 15

Modules and Vector Spaces

15.1

Modules and Vector Spaces

Our main interest here is to set down only the results of vector spaces which are needed for our study of fields in
the next chapter. We do this in such a way that the reader will become acquainted with the notion of a module.
Definition 15.1.1 Let R be a ring. A commutative group (M, +) is called a left R-module or a left module
over R with respect to a mapping : R M M if for all r, s R and m, m0 M,
(i) r (m + m0 ) = r m + r m0 ,
(ii) r (s m) = (rs) m,
(iii) (r + s) m = r m + s m.
If R has an identity 1 and if 1 m = m for all m M, then M is called a unitary or unital left R-module.
A right R-module can be defined in a similar fashion.
In the above definition, we used the same notation for the addition in the ring R and the addition in the
group M. We also used the same notation for the multiplication in R and the multiplication between the elements
of R and M. It should be clear to the reader by now that there are actually four distinct operations involved.
We write rm for r m.
Example 15.1.2 In a ring R, every left ideal is a left R-module and every right ideal is a right R-module. In
particular, R is a left and right R-module.
Example 15.1.3 Every commutative group M is a module over the ring of integers Z. For n Z and a M,
the element na is defined to be a added to itself n times if n is positive and a added to itself |n| times if n is
negative. 0a is defined to be the zero element of M. Under these definitions, M becomes a unitary left Z-module.
Let M be any commutative group and R be any ring. If we define rm = 0 for all r R, m M, then M
forms a left R-module, called a trivial module.
Since all results that are true for left R-modules are also true for right R-modules, we prove results only for
left R-modules. From now on, unless stated otherwise, by an R-module, we mean a left R-module.
Definition 15.1.4 Let M be an R-module and N be a nonempty subset of M. Then N is called a submodule
of M if N is a subgroup of M and for all r R, a N, we have ra N.
It is clear that a submodule of an R-module is itself an R-module.
Using arguments similar to those used for subgroups and ideals, one can show that the intersection of any
nonempty collection of submodules of an R-module is again a submodule.
Definition 15.1.5 Let X be a subset of an R-module M. Then the submodule of M generated by X is defined
to be the intersection of all submodules of M which contain X and is denoted by hXi . X is called a basis of
hXi if no proper subset of X generates hXi . If M = hXi and X is a finite set, then M is said to be finitely
generated. When X = {x} and M = h{x}i , then M is called a cyclic R-module and in this case we write
M = hxi .

221

222

15. Modules and Vector Spaces

We ask the reader to prove that any finitely generated module has a finite basis.
The proof of the following theorem is similar to that of the corresponding theorem for ideals, Theorem 8.2.9.
Hence, we omit its proof.
Theorem 15.1.6 Let M be an R-module and X be a nonempty subset of M. Then
S
S
hXi = { ki=1 ri xi + lj=1 nj x0j | ri R, nj Z, xi , x0j X,
1 i k, 1 j l, k, l N}.
If M is a unitary R-module, then
k
[
hXi = {
ri xi | ri R, xi X, 1 i k, k N}.
i=1

Example 15.1.7 (i) Q is a Q-module. If N is a submodule of Q, then N is a left ideal of Q. Since Q is a field,
the only left ideals of Q are {0} and Q. Hence the submodules of Q are {0} and Q.
(ii) We know that Q Q is a commutative group. For all x Q and for all (a, b) Q Q, define x(a, b) =
(xa, xb). Then QQ is a Q-module. We now determine all submodules of QQ. Let M be a nonzero Q-submodule
of Q Q.
Case 1: Suppose for all (a, b) M, b = 0. Now there exists (a, 0) M such that a 6= 0. Then (1, 0) =
M. Thus, M = Q {0}.
Case 2: Suppose for all (a, b) M, a = 0. Now there exists (0, b) M such that b 6= 0. Then (0, 1) =
1
b (0, b) M. Thus, M = {0} Q.
Case 3: Suppose there exists (a, b) M such that a 6= 0, b 6= 0.
Case 3a: Suppose M = h(a, b)i . Then M is a cyclic submodule of Q Q generated by (a, b).
Case 3b: Suppose M 6= h(a, b)i . Then h(a, b)i M. Thus, there exists (a0 , b0 ) M\ h(a, b)i . Then a0 6= 0
or b0 6= 0. Suppose that a0 = 0. Then (0, 1) = b10 (0, b0 ) M. Therefore, (a, 0) = (a, b) (0, 1)b M. Hence,
(1, 0) = a1 (a, 0) M. Thus, (1, 0), (0, 1) M. This implies that M = Q Q. Similarly, if b0 = 0, then M = Q Q.
Now suppose that a0 6= 0 and b0 6= 0. If aa0 = bb0 = t (say), then t(a0 , b0 ) = (ta0 , tb0 ) = ( aa0 a0 , bb0 b0 ) = (a, b)
0 qa0
h(a, b)i , which is a contradiction. Therefore, aa0 6= bb0 and so ab0 ba0 6= 0. Let (p, q) Q Q. Choose t = pb
ab0 ba0
qapb
0 0
and s = ab
0 ba0 . Then (p, q) = t(a, b) + s(a , b ) M. Thus, Q Q M. Hence, M = Q Q.
Consequently, if M is a Q-submodule of Q Q, then M is of the following form:
(i) M = {0}, or
(ii) M = {0} Q = h(0, 1)i , or
(iii) M = Q {0} = h(1, 0)i , or
(iv) M = h(a, b)i , a 6= 0, b 6= 0, a, b Q, or
(v) M = Q Q.
This also proves that M is finitely generated.
1
(a, 0)
a

Definition 15.1.8 Let F be a field. A unitary (left) F -module M is called a (left) vector space over F. The
elements of M are called vectors and the elements of F are called scalars. A submodule of M is called a
subspace of M. If X is a subset of M such that M = hXi , then X is said to span or generate M and M is
called the span of X over F.
Example 15.1.9 Let F be any field and F n denote the Cartesian product of F with itself ntimes. Then F n
becomes a vector space over F under the following definitions: For all (a1 , a2 , . . . , an ), (b1 , b2 , . . . , bn ) F n and
aF
(a1 , a2 , . . . , an ) + (b1 , b2 , . . . . , bn ) = (a1 + b1 , +a2 + b2 , . . . , an + bn ),
a(a1 , a2 , . . . , an ) = (aa1 , aa2 , . . . , aan ).
The set
X = {(1, 0, 0, . . . , 0), (0, 1, 0, . . . , 0), . . . , (0, 0, 0, . . . , 1)}

spans F n since for all (a1 , a2 , . . . , an ) F n ,

(a1 , a2 , . . . , an ) = a1 (1, 0, 0, . . . , 0) + a2 (0, 1, 0, . . . , 0) + + an (0, 0, 0, . . . , 1).

When n = 2 or 3 and F is the field of real numbers, then the vector space F n over F is the one usually encountered
in elementary analytical geometry.

15.1. Modules and Vector Spaces

223

By Example 15.1.9, R3 is a vector space over R.

Example 15.1.10 Consider the vector space R3 over R. Let
U = {(a, b, c) R3 | 2a + 3b + 5c = 0}.
Then U is a subspace of V3 (R). Let
U1 = {(a, b, c) R3 | 2a + 3b + 5c = 5}.
/ U1 . Hence, U1 is not a subspace of R3 .
Now (0, 0, 1) and (1, 1, 0) U1 , but (0, 0, 1) + (1, 1, 0)
Example 15.1.11 Let V be a vector space over F. Then {0} and V are subspaces of V. These are called trivial
subspaces of V.
Theorem 15.1.12 Let V be a vector space over F and S be a nonempty subset of V. Then S is a subspace of
V if and only if for all a F and for all x, y S, ax + y S.
Proof. Suppose S is a subspace of V. Then for all a F and for all x, y S, ax S and so ax + y S.
Conversely, suppose for all a F and for all x, y S, ax + y S. Since S 6= , there exists x S. By Exercise
2 (page 226), x = (1)x. Therefore, 0 = x + x = (1)x + x S. Hence, for all x S, x = (1)x + 0 S.
Also, for all x, y S, x + y = 1x + y S. S inherits the associative and commutative laws. Thus, (S, +) is a
commutative group. Now for all a F and for all x S, ax = ax + 0 S. Therefore, S is a vector space over F
since the other properties are inherited.
Theorem 15.1.13 Let V be a vector space over F and {U | I} be any nonempty collection of subspaces of
V. Then I U is a subspace of V.
Proof. First note that 0 U for all I and so 0 I U . Therefore, I U 6= . Let a F
and x, y I U . Then x, y U for all . Since U is a subspace of V, ax + y U for all I and so
ax + y I U . Thus, I U is a subspace of V by Theorem 15.1.12.
Theorem 15.1.14 Let V be a vector space over F and S be a nonempty subset of V. Then
[
hSi = {
ai si | ai F, si S},
S
where
ai si is a finite sum.
S
S
S
S
S
Let U = { ai si | ai F, si S}. Let a F and
ai si ,
bj sj U. Then a( ai si ) + bj sj =
S Proof. S
(aai )si + bj sj U and so U is a subspace of V by Theorem 15.1.12. S
Since for all s S, s = 1s U, U S.
Thus, U hSi since
hSi
is
the
smallest
subspace
of
V
containing
S.
Let
ai si U. Then since si S hSi ,
S
ai si hSi , whence U hSi .
ai si hSi . Thus,

Definition 15.1.15 Let V be a vector space over the field F. A subset X of V is called linearly independent
over F if for every finite number of distinct elements x1 , x2 , . . . , xn X, a1 x1 + a2 x2 + + an xn = 0 implies
that a1 = a2 = an = 0 for any finite set of scalars {a1 , a2 , . . . , an }. Otherwise X is called linearly dependent
over F.
The set X in Example 15.1.9 is linearly independent over F. {0} is linearly dependent over F.
Definition 15.1.16 Let V be a vector space over F. A subset A of V is called a basis for V over F if A spans
V, i.e., V = hAi , and A is linearly independent over F.
Consider the zero vector space, {0}, over the field F. We note that the empty subset, , is linearly independent
over F vacuously and that spans {0}. Hence, is a basis for {0}.
Example 15.1.17 The set
X = {(1, 0, 0, . . . , 0), (0, 1, 0, . . . , 0), . . . , (0, 0, 0, . . . , 1)}
of Example 15.1.9 is a basis for F n . We showed there that X spans F n over F. Suppose
(0, 0, . . . , 0) = a1 (1, 0, 0, . . . , 0) + a2 (0, 1, 0, . . . , 0) + + an (0, 0, 0, . . . , 1).
Then (0, 0, . . . , 0) = (a1 , a2 , . . . , an ). Therefore, we must have ai = 0 for i = 1, 2, . . . , n. Thus, X is linearly
independent.

224

15. Modules and Vector Spaces

Theorem 15.1.18 Let V be a vector space over F and S be a subset of V. If s hSi , then hS {s}i = hSi .
Proof. Clearly hSi ShS {s}i . If S = , then hSi = {0} and soS
s = 0. Hence, hS {s}iS= h{0}i = {0} =
hSi . Suppose S 6= . Let
ai si + as hS {s}i , where si S. Then
ai si , as hSi and so
ai si + as hSi .
Hence, hS {s}i = hSi .
Theorem 15.1.19 Let V be a vector space over F and A = {x1 , x2 , . . . , xr } be a subset of V which spans V.
Let B be any linearly independent set of vectors in V. Then B contains at most r vectors.
Proof. If B contains less than r vectors, the theorem is true. Suppose B contains at least r vectors, say,
y1 , y2 , . . . , yr B. Then since A spans V,
r
[
y1 =
ai1 xi
i=1

and since y1 6= 0, not all ai1 = 0, say, a11 6= 0. Thus,

x1 =

r
[
1
(a1
11 ai1 )xi + a11 y1 .
i=2

This implies that x1 h{y1 , x2 , . . . , xr }i . Hence, h{y1 , x2 , . . . , xr }i = V by Theorem 15.1.18. Assume h{y1 , y2 , . . . , yk , xk+1 , . . . , xr }i =
V, the induction hypothesis. Then
yk+1 h{y1 , y2 , . . . , yk , xk+1 , . . . , xr }i .
Thus,
yk+1 =

k
[

ai,k+1 yi +

i=1

r
[

ai,k+1 xi

i=k+1

and not all ai,k+1 = 0 for i = k + 1, . . . , r, say, ak+1,k+1 6= 0. This implies that
xk+1 =

k
r
[
[
1
(a1
a
)y
+
(a1
i
i,k+1
k+1,k+1
k+1,k+1 ai,k+1 )xi + ak+1,k+1 yk+1 .
i=1

i=k+2

Thus, xk+1 h{y1 , y2 , . . . , yk , yk+1 , xk+2, . . . , xr }i . Hence,

V = h{y1 , y2 , . . . , yk , yk+1 , xk+2, . . . , xr }i
by Theorem 15.1.18.
yr }i = V by induction. If there exists y B such that y 6= yi , i = 1, 2,
S Thus, h{y1 , y2 , . . . ,S
. . . , r, then y = ri=1 ai yi and so 0 = ri=1 ai yi + (1)y and since 1 6= 0, y1 , y2 , . . . , yr , y are not linearly
independent, a contradiction. Therefore, y does not exist and so B = {y1 , y2 , . . . , yr }.
Theorem 15.1.20 Let V be a vector space over F, A = {x1 , . . . , xr }, and B = {y1 , . . . , ys } be two bases for V.
Then r = s.
Proof. Since A spans V and B is linearly independent, s r by Theorem 15.1.19. Similarly, r s.
Definition 15.1.21 Let V be a vector space over F. If V is spanned by a finite set of vectors, then V is called
finite dimensional over F.
Lemma 15.1.22 Let V be a vector space over F and A be a linearly independent subset of V. If x V and
x
/ hAi , then A {x} is linearly independent.
Proof. Let x1 , . . . , xn A. Suppose 0 = a1 x1 + a2 x2 + + ar xr + ax. Suppose a 6= 0. Then
x = (a)1 a1 x1 + + (a)1 ar xr hAi ,
a contradiction. Thus, a = 0. Hence, 0 = a1 x1 + a2 x2 + + ar xr . Since {x1 , x2 , . . . , xr } is linearly independent,
a1 = 0, . . . , ar = 0. Thus, A {x} is linearly independent.
Theorem 15.1.23 Let V be a finite dimensional vector space over F. Then V has a basis.

15.1. Modules and Vector Spaces

225

Proof. If V = {0}, then is a basis for V. We now assume that V 6= {0}. Let x1 V be such that x1 6= 0.
Then x1 is linearly independent. If hx1 i 6= V, then there exists x2 V such that x2
/ hx1 i . By Lemma 15.1.22, x1
and x2 are linearly independent. Suppose x1 , . . . , xk V are linearly independent and h{x1 , . . . , xk }i 6= V. Then
/ h{x1 , . . . , xk }i . Therefore, x1 , . . . , xk , xk+1 are linearly independent.
there exists xk+1 V such that xk+1
Since V is finite dimensional, V is spanned by, say, r vectors. By Theorem 15.1.19, any linearly independent set
of vectors in V cannot have more than r vectors. Hence, if we continue the above process of constructing xi s,
then there must exist a positive integer s such that {x1 , . . . , xs } is linearly independent, h{x1 , . . . , xs }i = V, and
s r. Thus, {x1 , . . . , xs } is a basis of V.
Theorem 15.1.23 gives us a method for constructing a basis for a finite dimensional vector space V of dimension
n over F. We first take any nonzero vector x1 of V. If hx1 i = V, then {x1 } is a basis of V. If hx1 i V, then we
/ hx1 i . Then by Lemma 15.1.22 {x1 , x2 } is linearly independent over F. If h{x1 , x2 }i = V,
take any x2 V, x2
then {x1 , x2 } is a basis for V over F. If h{x1 , x2 }i V, we can choose x3 V, x3
/ h{x1 , x2 }i and so on. In a
finite number of steps, precisely n steps, we must arrive at a basis for V over F.
Definition 15.1.24 Let V be a finite dimensional vector space over F. The dimension V is the number of
elements in a basis for V.
0.

From the statements following Definition 15.1.16, it follows that the zero vector space, {0}, is of dimension

Theorem 15.1.25 Let V be a vector space of dimension n over the field F. Then X = {x1 , x2 , . . . , xn } is a basis
of V if and only if every vector in V is a unique linear combination of x1 , x2 , . . . , xn over F.
Proof. Suppose X is a basis of V over F. Then by Theorem 15.1.14, every vector v V is a linear combination
of x1 , x2 , . . . , xn . Let
v = a1 x1 + + an xn = b1 x1 + + bn xn
be any two linear combinations of x1 , x2 , . . . , xn . Then
0 = (a1 b1 )x1 + + (an bn )xn .
The linear independence of X over F implies that a1 b1 = 0, . . . , an bn = 0. That is, the representation of
v as a linear combination of x1 , x2 , . . . , xn is unique. Conversely, suppose every vector in V is a unique linear
combination of x1 , x2 , . . . , xn over F. Then clearly X generates V over F. Suppose 0 = a1 x1 + +an xn for
ai F. Since also 0 = 0x1 + +0xn , we have ai = 0, i = 1, . . . , n. Thus, X is linearly independent over F. By
definition, X is a basis of V over F.
We now show that every nonzero vector space, not necessarily finite dimensional, has a basis. For this we
prove the following lemma.
Lemma 15.1.26 Let V be a vector space over a field F and X be a nonempty subset of V. Then X is a basis
for V if and only if X is a maximal linearly independent set over F.
Proof. If X is a basis for V, then X is linearly independent over F and hXi = V. Let y V, y
/ X. Then
V = hXi hX {y}i V so that V = hX {y}i . Since the proper subset X of X {y} also generates V, X
{y} cannot be linearly independent over F. Thus, X is a maximal linearly independent set over F. Conversely,
let X be a maximal linearly independent set over F. It suces to show that V = hXi . If V hXi , then
there exists y V, y
/ hXi . By Lemma 15.1.22, X {y} is linearly independent over F, which contradicts the
maximality of X. Thus, V = hXi .
Theorem 15.1.27 Let V be a vector space over the field F. Then V has a basis.
Proof. If V = {0}, then is a basis for V. We now assume that V 6= {0}. Let x be a nonzero element of V.
Then {x} is a linearly independent subset of V. Let T be the set of all linearly independent subsets of V that
contain {x}. Clearly T 6= . T is a poset with respect to the set inclusion relation. By Zorns lemma, we can show
that T has a maximal element, say, X. Then X is a maximal linearly independent subset of V and by Lemma
15.1.26, it follows that X is a basis of V.
Finally, we state the following theorem without proof. The finite dimensional case was proved in Theorem
15.1.20.
Theorem 15.1.28 Let V be a vector space over a field F . If A and B are two bases of V, then |A| = |B| .
From Theorem 15.1.27, we find that a vector space V over a field F has a basis B. If B is a basis for V over
F, then |B| is called the dimension of V over F.

226

15. Modules and Vector Spaces

Worked-Out Exercises
Exercise 1 Let V be a vector space of dimension n. Show that any set of n linearly independent vectors is a basis of
V.
Solution: Let B be a set of n linearly independent vectors. Suppose V 6= hBi . Let y V be such that y
/ hBi .
Then B {y} is a set of n + 1 linearly independent vectors by Lemma 15.1.22, a contradiction to Theorem
15.1.19. Hence, B is a basis of V.
Exercise 2 Let u1 = (0, 1, 1, 0), u2 = (1, 0, 1, 0), and u3 = (1, 2, 0, 0) be three vectors in R4 . Show that {u1 , u2 , u3 }
is a linearly independent set. Extend this set to a basis of R4 .
Solution: Let a1 , a2 , a3 R be such that

a1 u1 + a2 u2 + a3 u3 = 0.

Then a2 a3 = 0, a1 2a3 = 0, and a1 + a2 = 0. From this, it follows that a1 = a2 = a3 = 0. Hence,

{u1 , u2 , u3 } is a linearly independent set. Suppose
(0, 0, 0, 1) h{u1 , u2 , u3 }i .
Then there exists b1 , b2 , b3 R such that
b1 u1 + b2 u2 + b3 u3 = (0, 0, 0, 1).
/
Thus, b2 b3 = 0, b1 2b3 = 0, b1 + b2 = 0, and 1 = 0, a contradiction. Therefore, e4 = (0, 0, 0, 1)
h{u1 , u2 , u3 }i . Hence, {u1 , u2 , u3 , e4 } is a linearly independent set of vectors in R4 . Since the dimension of
R4 is 4, {u1 , u2 , u3 , e4 } is a basis.

Exercise 3 Let V be a nonzero vector space of dimension n. Let X be a finite subset of V such that V = hXi . Show
that X contains a subset Y such that Y is a basis of V.

Solution: Let X = {x1 , x2 , . . . , xt }. Clearly t n. Since V 6= {0}, X contains a nonzero element. Thus, X contains
a linearly independent subset. If X is linearly independent, then X is a basis of V and n = t. Suppose
X is not linearly independent. Then there exists xi , say, xt , such that xt h{x1 , x2 , . . . , xt1 }i . Then
V = h{x1 , x2 , . . . , xt1 i}. Let s = t n 1. By repeating the process finitely many times, we can show
that there are s vectors xi1 , . . . , xis {x1 , x2 , . . . , xt1 } such that
xi1 , . . . , xis h{x1 , x2 , . . . , xt1 }\{xi1 , . . . , xis }i .
Let
Y = {x1 , x2 , . . . , xt1 }\{xi1 , . . . , xis }.

Then Y X, |Y | = n, and V = hY i . If Y is not linearly independent, then there exists y Y such that
y hY \{y}i . Then V = hY \{y}i and |Y \{y}| = n 1, a contradiction to the fact that the dimension of
V is n.
Exercise 4 Let T = {(x, y, z) R3 | 2x + 3y + z = 0}. Show that T is a subspace of V3 (R). Find a basis for T.

Solution: Since (0, 0, 0) T, T 6= . Let (x1 , y1 , z1 ), (x2 , y2 , z2 ) T and r R. Then 2x1 + 3y1 + z1 = 0 and
2x2 +3y2 +z2 = 0. Hence, 2(x1 +x2 )+ 3(y1 +y2 )+ (z1 +z2 ) = 0 and 2rx1 +3ry1 +rz1 = r(2x1 +3y1 +z1 ) = 0.
Therefore, (x1 , y1 , z1 )+ (x2 , y2 , z2 ) T and r(x1 , y1 , z1 ) T. Thus, T is a subspace of V3 (R). Now
2x1 + 3y1 + z1 = 0 implies that (x1 , y1 , z1 ) = (x1 , y1 , 2x1 3y1 ) = x1 (1, 0, 2)+ y1 (0, 1, 3). Since
(1, 0, 2), (0, 1, 3) T and (x1 , y1 , z1 ) is an arbitrary element of T, T = h{(1, 0, 2), (0, 1, 3)}i . It is
easy to verify that {(1, 0, 2), (0, 1, 3)} is a linearly independent set. Hence, {(1, 0, 2), (0, 1, 3)} is a
basis of T.

Exercises
1. For the vector space R3 over R, determine whether or not the sets listed are bases of R3 .
(i) {(1, 1, 0), (1, 1, 1), (1, 0, 0)}.
(ii) {(2, 0, 0), (0, 2, 0), (0, 0, 2)}.
(iii) {(1, 0, 0), (0, 1, 0), (0, 0, 1)}.
(iv) {(1, 0, 0), (1, 1, 0), (1, 1, 1), (0, 1, 0)}.

2. Let M be an R-module, m M and r R. Prove that r0 = 0, 0m = 0, and (rm) = (r)m = r(m).

3. Show that the intersection of two submodules of an R-module M is a submodule.

15.1. Modules and Vector Spaces

227

4. Show that the Z-module Q has no finite set of generators.

5. Find all subspaces of the real vector space R2 . Is is true that for any elements u = (a, b) and v = (c, d) of
R2 , there exists a nontrivial subspace W of R2 such that u, v W ?
6. Let A, B, and C be submodules of an R-module M.
(i) Prove that A + B = {a + b | a A, b B} is a submodule of M.
(ii) If A C, prove that A + (B C) = (A + B) C.

7. Let M be an R-module and a M. Show that T = {ra + na | r R, n Z} is a submodule of M.

8. Let M be a unitary R-module. M is called a simple R-module if M 6= {0} and the only submodules of
M are M and {0}. Prove that M is simple if and only if M is generated by any nonzero element of M.
9. Let N be a submodule of a unitary R-module M and a M. Let
a + N = {a + b | b N}.
Prove the following.
(i) a a + N.
(ii) For all a, b M, a + N = b + N if and only if a b N.
(iii) For all a, b M, either (a + N) (b + N) = or a + N = b + N.

10. Let N be a submodule of an R-module M. Let

M/N = {a + N | a N}.
Define the following operations on M/N
(a + N) + (b + N)
r(a + N)

=
=

(a + b) + N
ra + N

for all a + N, b + N M/N, r R. Prove that M/N is an R-module.

11. Let V be a finite dimensional vector space over F. If U and W are two subspaces of V, prove the following:
(i) U + W = {u + w | u U, w W } is a subspace of V.
(ii) dimU + dimW = dim(U + W ) dim(U W ).

12. Let N be a submodule of an R-module M. N is called a direct summand of M if there exists a submodule
P of M such that M = N + P and N P = {0}. In a finite dimensional vector space V over F, show that
every subspace is a direct summand of V.

13. Write the proof if the statement is true; otherwise give a counterexample.
(i) If {u, v, w} is a linearly independent subset of a vector space V, then {u, u + v, u + v + w} is also a
linearly independent subset.
(ii) If W is a subspace of a finite dimensional vector space V such that dim W = dim V, then W = V.
(iii) Let V be a vector space over a field F. If 0 6= v V, then there exists a basis containing v.
(iv) If S and T are two basis of a vector space V, then S T is a basis of V.

228

15. Modules and Vector Spaces

Chapter 16

Field Extensions
In this chapter, we study a special type of ring called a field. Results about fields have applications in number
theory and the theory of equations. The theory of equations deals with roots of polynomials. It is here that our
main interest lies. This interest leads us to an introduction of Galois theory.
The importance of the concept of a field was first recognized by Abel and Galois in their research on the
solution of equations by radicals. However, the formal definition of a field appeared more than 70 years later.
The works of Dedekind and Kronecker seem to be responsible for the entrance of the concept of a field into
mathematics. However, in 1910, in his paper, Algebraic Theorie der Kperer, Steinitz gave the first abstract
definition of a field. His work freed the concept of a field from the context of complex numbers.

16.1

Algebraic Extensions

Let us recall that the characteristic of a field F is either 0 or a prime p. By Theorem 8.1.9, the intersection of
any collection of subfields of a field F is again a subfield of F. Hence, a field contains a subfield which has no
proper subfield, namely, the intersection of all its subfields.
Definition 16.1.1 A field F is called a prime field if F has no proper subfield.
Theorem 16.1.2 Let F be a field.
(i) If the characteristic of F is 0, then F contains a subfield K such that K ' Q.
(ii) If the characteristic of F is p > 0, then F contains a subfield K such that K ' Zp .
Proof. Define f : ZF by
f (n) = n1
for all n Z, where 1 denotes the identity of F. Then f is a homomorphism.
(i) Suppose the characteristic of F is 0. Then Ker f = {0} and so f is one-one. Define f : Q F by
a
f ( ) = f(a)f (b)1
b
for all ab Q. Let ab , dc Q. Now ab = dc if and only if ad = bc if and only if f(ad) = f (bc) if and only if
f (a)f(d) = f (c)f (b) if and only if f (a)f (b)1 = f (c)f (d)1 if and only if f ( ab ) = f ( dc ). Hence, f is a one-one
function. Now
)
f ( ab + dc ) = f ( ad+bc
bd
= f (ad + bc)f(bd)1
= (f (a)f (d) + f (b)f (c))f (b)1 f (d)1
= f (a)f (b)1 + f (c)f(d)1
= f ( ab ) + f ( dc ).
Also,
f ( ab dc )

=
=
=
=
=

f ( ac
)
bd
f(ac)f(bd)1
f(a)f (c)f (b)1 f (d)1
f(a)f (b)1 f (c)f (d)1
f ( ab )f ( dc ).

229

230

16. Field Extensions

Thus, f is a homomorphism. Hence, Q ' I(f ), where I(f ) is the image of f . Let K = I(f ).
(ii) Suppose the characteristic of F is p > 0. Now
Z/Ker f ' I(f ).
Since the characteristic of F is not zero, I(f ) 6= {0}. Therefore, I(f) is a nontrivial subring with 1 of the field F.
Consequently, I(f) is an integral domain and so Z/Ker f is an integral domain. This implies Ker f is a prime
ideal of Z and Z6= Ker f. There exists a prime q such that Ker f = qZ. Now q1 = 0 implies that p|q and so q = p.
Hence, Z/Ker f ' Zp .
Let L be a subfield of Q. Since L\{0} is a subgroup of Q\{0} under multiplication, 1 L. Hence, Z L and
so Q L. Thus, Q has no proper subfield. Similarly, Zp has no proper subfield, where p is a prime.
Thus, the subfield K of the field F in Theorem 16.1.2 is the prime subfield of F.
The following theorem can be easily verified. We leave its proof as an exercise.
Theorem 16.1.3 Let F be a field and K be a subfield of F. The following conditions are equivalent.
(i) K is the prime subfield of F.
(ii) K is the intersection of all subfields of F.
Let F be a field and K a subfield of F. The field F is called an extension of the field K. We express this by
F/K and call F/K a field extension or an extension field.
Definition 16.1.4 Let F/K be a field extension and C be a subset of F. Define K(C) to be the intersection of
all subfields of F which contain K C. Then the subfield K(C) of F is called the subfield of F generated by C
over K. C is called a set of generators for K(C)/K.
Let K[C] be the smallest subring of F containing K C. Since any subfield of F which contains K C must
contain K[C], we have that K(C) equals the intersection of all subfields which contain K[C]. Now K[C] is an
integral domain since it is a subring (with identity) of a field. Thus, by Theorem 9.1.6,
K(C) = {ab1 | a, b K[C], b 6= 0}.
That is, K(C) is the set of all rational expressions of the elements of K[C]. Hence, K(C) is a quotient field of
K[C].
Let F/K be a field extension and c1 , c2 , . . . , cn F. Considering Definition 16.1.4, it follows that K(c1 , c2 , . . . , cn ) =
K(c1 , c2 , . . . , cn1 )(cn ). Recall that K(c1 ) = {ab1 | a, b K[c1 ], b 6= 0}.
Definition 16.1.5 Let F/K be a field extension. An element a F is said to be algebraic over K if there exist
k0 , k1 , . . . , kn K, not all zero, such that k0 + k1 a+ + kn an = 0; otherwise a is called transcendental over
K.
Let F/K be a field extension and let a F. Then a is algebraic over K if and only if a is a root of a nonzero
polynomial with coecients from K.

Example 16.1.6 The element 2 in R is algebraic over Q since 2 is a root of x2 2 Q[x]. The element
i C is algebraic over R and Q since i is a root of x2 + 1 Q[x].
Example 16.1.7 It can be shown that , e R are transcendentalS
over Q. In the quotient field F (x) of the
n
i
polynomial ring F [x], F a field, x is transcendental over F since
i=0 ai x = 0 if and only if ai = 0 for
i = 0, 1, . . . , n.
Theorem 16.1.8 Let F/K be a field extension and c F. Then c is algebraic over K if and only if c is a root
of some unique irreducible monic polynomial p(x) over K.
Proof. Suppose c is algebraic over K. There exists a nonzero polynomial f (x) K[x] such that c is a root
of f(x) and f (x)
/ K. By Theorem 13.1.15, there exist irreducible polynomials f1 (x), f2 (x), . . . , fm (x) K[x]
such that f (x) = f1 (x)f2 (x) fm (x). Thus,
0 = f (c) = f1 (c)f2 (c) fm (c).

16.1. Algebraic Extensions

231

Since F has no zero divisors, we must have fi (c) = 0 for some i. Thus, there exists an irreducible polynomial
h(x) = b0 + b1 x+ + bm xm , bm 6= 0, such that h(c) = 0. Let p(x) = b1
m h(x). Then p(x) is an irreducible monic
polynomial in K[x] with c as a root.
Let g(x) be any polynomial in K[x], which has c as a root. Let p(x) be a monic polynomial of smallest degree
in K[x], which has c as a root. There exist q(x), r(x) K[x] such that g(x) = q(x)p(x) + r(x), where either
r(x) = 0 or deg r(x) < deg p(x). Now
0 = g(c) = q(c)p(c) + r(c) = q(c) 0 + r(c).
Thus, r(c) = 0, whence r(x) = 0 else we contradict the minimality of the degree of p(x). This implies that
p(x)|g(x) in K[x]. Let s(x) be any irreducible polynomial in K[x], which has c as a root (one such polynomial
is fi (x) for some i, 1 i m). Then p(x)|s(x). Now p(x) is not a constant polynomial in K[x] since it has c
as a root. Thus, since s(x) is irreducible in K[x], p(x) must be irreducible in K[x]. Also, p(x) = ks(x) for some
k K. If we choose s(x) monic, then k = 1 and so we have the desired uniqueness property of p(x). The converse
is immediate.
The proof of Theorem 16.1.8 yields the next result.
Corollary 16.1.9 Let F/K be a field extension and c F be such that c is algebraic over K. Then the unique
monic irreducible polynomial p(x) over K having c as a root satisfies the following properties:
(i) There is no polynomial g(x) K[x] having smaller degree than p(x) and which has c as a root.
(ii) If c is a root of some g(x) K[x], then p(x)|g(x) in K[x].
We call the polynomial p(x), in Corollary 16.1.9, the minimal polynomial of c over K. The degree of p(x)
is called the degree of c over K
Example 16.1.10 By Examples 16.1.6, 12.3.6, and 12.3.7, we have that x2 2 is the minimal polynomial of
over Q and x2 + 1 is the minimal polynomial of i over R.

Theorem 16.1.11 Let F/K be a field extension and c F.

(i) If c is transcendental over K, then K(c) ' K(x), where K(x) is the quotient field of the polynomial ring
K[x].
(ii) If c is algebraic over K, then K[c] ' K[x]/ hp(x)i , where p(x) is the minimal polynomial of c over K.
Proof. Define the mapping : K[x] K[c] by for all f (x) K[x],
(f(x)) = f (c).
Then by Theorem 11.1.14, is a homomorphism of K[x] onto K[c]. Thus,
K[x]/Ker ' K[c].
(i) Now f (x) Ker if and only if f (c) = 0, i.e., if and only if c is a root of f(x). Hence, Ker = {0} if and
only if c is transcendental over K. Thus, c is transcendental over K implies is an isomorphism of K[x] onto
K[c] and so by Exercise 5 (page 168), can be extended to an isomorphism of K(x) onto K(c). Consequently,
if c is transcendental over K, then K(x) ' K(c).
(ii) Suppose c is algebraic over K. Since K[x] is a principal ideal domain, there exists g(x) K[x] such
that Ker = hg(x)i . Now (g(x)) = g(c) = 0. Hence, c is a root of g(x). Thus, p(x)|g(x) and so there exists
q(x) K[x] such that g(x) = q(x)p(x). This implies that g(x) hp(x)i and so
Ker = hg(x)i hp(x)i .
Since p(c) = 0, p(x) Ker . Therefore, hp(x)i Ker . Consequently, Ker = hp(x)i .
Corollary 16.1.12 Let F/K be a field extension and c F. Then
(i) K[c] K(c) if and only if c is transcendental over K,
(ii) K[c] = K(c) if and only if c is algebraic over K.

232

16. Field Extensions

Proof. Since K[c] K(c) always holds, (i) and (ii) are equivalent statements. Hence, we show that (ii)
holds. Suppose c is algebraic over K. Then by Theorem 16.1.11,
K[c] ' K[x]/ hp(x)i
and since p(x) is irreducible, K[x]/ hp(x)i is a field. Thus, K[c] = K(c). Conversely, suppose K[c] = K(c). If
c = 0, then c is the root of the polynomial x K[x]. Suppose that c 6= 0. Then c1 K(c) and so c1 =
k0 + k1 c+ + kn cn for some ki K. This implies that 0 = 1 + k0 c + k1 c2 + + kn cn+1 and so c is algebraic
over K.
Let F/K be a field extension. Under the field operations of F, F can be considered as a vector space over K.
The elements of F are thought of as vectors while those of K are thought of as scalars. Recall that (F, +) is a
commutative group and that for all k1 , k2 K and a1 , a2 F, k1 (a1 +a2 ) = k1 a1 +k1 a2 , (k1 +k2 )a1 = k1 a1 +k2 a1
hold from the distributive laws and that (k1 k2 )a1 = k1 (k2 a1 ) holds from the associative law of multiplication.
Definition 16.1.13 Let F/K be a field extension. The dimension of the vector space F over K is called the
degree or dimension of F/K and is denoted by [F : K]. If the dimension of F/K is finite, then F/K is called
a finite extension.
Theorem 16.1.14 Let F/K be a field extension and c F be algebraic over K. Let p(x) be the minimal
polynomial of c over K. If deg p(x) = n, then {1, c, c2 , . . . , cn1 } is a basis of K(c)/K.
Proof. By Corollary 16.1.12, K[c] = K(c). Let g(c) K[c] and g(x) be the corresponding element in K[x].
There exist q(x), r(x) K[x] such that g(x) = q(x)p(x) + r(x), where either r(x) = 0 or deg
S r(x) <i deg p(x).
Thus, g(c) = q(c)p(c) + r(c) = r(c). Hence, {1, c, c2 , . . . , cn1 } spans K(c)/K. Suppose 0 = n1
i=0 ki c , ki K.
If the ki s are not all zero, then c is a root of a polynomial of degree n 1 < n, a contradiction. Thus, ki = 0
for i = 0, 1, . . . , n 1 and so {1, c, c2 , . . . , cn1 } is linearly independent over K. Hence, {1, c, c2 , . . . , cn1 } is a
basis of K(c)/K.
Corollary 16.1.15 Let F/K be a field extension. If c F is algebraic and of degree n over K, then [K(c) :
K] = n.

Example 16.1.16 The field extension Q( 2)/Q is of degree 2 and {1, 2} is a basis
of Q( 2)
over Q since
2
p(x) = x 2 is the minimal polynomial of 2 over Q by Example 16.1.10. Thus, Q( 2) = {a + b 2 | a, b Q}.

The student may recall from another mathematics course that

a + b 2 = c + d 2 if and only if a = c and
b = d, where a, b, c, d Q. This becomes clear now since 1 and 2 are linearly independent over Q by Theorem
16.1.14.
Example 16.1.17 By Theorem 16.1.14, the field extension R(i)/R is of degree 2 and {1, i} is a basis of R(i)
over R since p(x) = x2 + 1 is the minimal polynomial of i over R. Thus, R(i) = {a + bi | a, b R}. Hence, we
see that R(i) is C, the field of complex numbers.
Theorem 16.1.18 Let F/K be a finite field extension. Then every element of F is algebraic over K.
Proof. Let n be the dimension of F/K. Let c F be such that c 6= 0, c 6= 1. (Clearly 0 and 1 are
algebraic over K.) If the set {1, c, c2 , . . . , cn } does not contain n + 1 distinct elements, then cji = 1 for some i,
j (0 i < j n) and so c is a root of xji 1. Suppose 1, c, c2 , . . . , cn are distinct. Then they must be linearly
dependent since they are more in number than
of the vector space F over K.
there exist
S the dimension
SHence,
n
i
i
k0 , k1 , . . . , kn K not all zero such that 0 = n
i=0 ki c . Thus, c is a root of the polynomial
i=0 ki x over K.

The converse of Theorem 16.1.18 is not true, that is, it is not necessarily the case that if every element of
F is algebraic over K, then F/K is a finite field extension. It can be shown that the set of all elements A of C,
which are algebraic over Q is a field such that [A : Q] is infinite (Theorem 16.1.22 and Example 16.1.25). A is
called the field of algebraic numbers.
Theorem 16.1.19 Let K(c)/K be a field extension. Then K(c)/K is finite if and only if c is algebraic over K.

16.1. Algebraic Extensions

233

Proof. If K(c)/K is finite, then c is algebraic over K by Theorem 16.1.18. If c is algebraic over K, then
K(c)/K is finite by Corollary 16.1.15.
Let F/K be a field extension. A subfield L of F is called an intermediate field of F/K if K L F.
Since a b L for all a, b L and ka L for all k K and a L, it follows that L is a subspace of F over K.
An intermediate field L of F/K is called proper if L 6= F.
Theorem 16.1.20 Let F/K be a field extension and L be an intermediate field of F/K. Then
[F : K] = [F : L][L : K].
Moreover, F/K is a finite extension if and only if F/L and L/K are finite extensions.
Proof. Let V be a basis of F/L and U be a basis of L/K. We show that
W = {uv | u U, v V }
is a basis of F/K. Let c F. Since V is a basis of F/L, there exist v1 , v2 , . . . , vn V and c1 , c2 , . . . , cn L such
that
n
[
c=
cj vj .
(16.1)
j=1

Since U is a basis of L/K, there exist u1 , u2 , . . . , um U and k1j , k2j , . . . , kmj K such that
cj =

m
[

kij ui ,

j = 1, 2, . . . , n.

(16.2)

i=1

Substituting Eq. (16.2) into Eq. (16.1), we obtain

c=

m
n [
[

kij ui vj .

j=1 i=1

Thus, W spans F over K. Suppose

0=

m
n [
[

kij ui vj ,

j=1 i=1

where ui U, vj V, and kij K for all i = 1, 2, . . . , m; j = 1, 2, . . . , n. Then

0=

n [
m
[
(
kij ui )vj
j=1 i=1

and since V is linearly independent over L,

0=

m
[

kij ui , j = 1, 2, . . . , n.

i=1

Thus, kij = 0 for i = 1, 2, . . . , m; j = 1, 2, . . . , n since U is linearly independent over K. Hence, W is linearly

independent over K, whence W is a basis of F over K. Let u, u0 U and v, v 0 V. If v 6= v 0 , then uv 6= u0 v 0
since v and v 0 are linearly independent over L. If v = v 0 , then uv = u0 v 0 if and only if u = u0 . Consequently, for
all u, u0 U and for all v, v 0 V if either u 6= u0 or v 6= v 0 , then uv 6= u0 v 0 . Hence, [F : K] = |U V | = |U | |V | =
[F : L][L : K]. Now if either U or V is infinite, then W is infinite. If U and V are finite sets, then W is a finite
set. Hence, F/K is a finite extension if and only if F/L and L/K are finite extensions.

Example
16.1.21 Consider the field extension Q( 2, 3)/Q. By
Example 16.1.10,
x2 2 is the minimal poly
2
nomial
of 2 over Q. Also, x 3 is the minimal polynomial of 3 over Q( 2). (That x2 3 is irreducible
overQ( 2) follows by
an argument that
one
used in
Worked-Out Exercise 1, page
235.)
is similar to the
Thus,

of
Q(
2)/Q
and
{1,
3}
is
a
basis
of
Q(
2,
3)/Q(
2).
By
Theorem
16.1.20,
{1,
2,
3, 6}
{1, 2} is a basis

is a basis of Q( 2, 3)/Q. [Q( 2, 3) : Q] = 4, [Q( 2, 3) : Q( 2)] = 2, and [Q( 2) : Q] = 2.

Theorem 16.1.22 Let F/K be a field extension. If L is the set of all elements in F, which are algebraic over
K, then L is an intermediate field of F/K.

234

16. Field Extensions

Proof. Any k K is a root of the polynomial x k over K. Thus, L K. Let a and b be elements of L,
where a is of degree m over K and b is of degree n over K. Then K(a)/K is of degree m and K(a, b)/K(a) is
of degree at most n. Hence, by Theorem 16.1.20, K(a, b)/K is a finite extension. By Theorem 16.1.18, every
element of K(a, b) is algebraic over K. Since a b and ab1 (for b 6= 0) are elements of K(a, b), a b and ab1 (for
b 6= 0) are algebraic over K. Thus, a b and ab1 (for b 6= 0) L and so L is a field.
Definition 16.1.23 A field extension F/K is called algebraic if every element of F is algebraic over K; otherwise F/K is called transcendental.
Theorem 16.1.24 Let L be an intermediate field of the field extension F/K. Then F/K is an algebraic extension
if and only if F/L and L/K are algebraic extensions.
Proof. Suppose that F/K is algebraic. Let a F. Then a is a root of a nonzero polynomial p(x) K[x].
Since K L, p(x) L[x]. Thus, a is algebraic over L and so F/L is algebraic. Every element of L is an element
of F. Hence, L/K is algebraic. Conversely, suppose F/L and L/K are algebraic extensions. Let c F. Then c is
a root of some nonzero polynomial c0 + c1 x + + cn xn L[x]. Thus, c is algebraic over K(c0 , c1 , . . . , cn ) whence
K(c0 , c1 , . . . , cn )(c)/K(c0 , c1 , . . . , cn ) is a finite extension. Since c0 , c1 , . . . , cn are algebraic over K, repeated
application of Theorem 16.1.20 yields that K(c0 , c1 , . . . , cn )(c)/K is a finite extension. Therefore, c is algebraic
over K by Theorem 16.1.18. Hence, F/K is an algebraic extension.

Example 16.1.25 Let F = Q({ p | p Z, p is a prime}) R. We show that F/Q is algebraic and [F : Q] = .

Now for any prime p, p

/ Q. Let p1 , . . . , pn be any distinct primes. Suppose p 6= pi , i = 1, 2, . . . , n, and p

is a prime. Assume that p

/ Q( p1 , . . . , pn ), the induction hypothesis. (The case n = 0 is p
/ Q and
this case is described above.) We show that if p1 , . . . , pn+1 are distinct primes and p 6= pi , i = 1, 2, . . . , n + 1,

then p
/ Q( p1 , . . . , pn+1 ). Suppose p Q( p1 , . . . , pn+1 ). Then there exist a, b Q( p1 , . . . , pn ) such

2
that p = a + b pn+1 . If a = 0, then p = b pn+1 , a contradiction since p and pn+1 are distinct primes. If

b = 0, then p = a Q( p1 , . . . , pn ), a contradiction to our induction hypothesis. Suppose a 6= 0 and b 6= 0.

2
2
Then p = a + pn+1 b + 2ab pn+1 . Hence, pn+1 = (p a2 pn+1 b2 )/2ab Q( p1 , . . . , pn ) and so p

/ Q( p1 , . . . , pn+1 ). Thus, by the induction

Q( p1 , . . . , pn ), a contradiction of the hypothesis. Hence, p

hypothesis, we find that for any positive integer k, if p1 , . . . , pk , p are distinct primes, then p
/ Q( p1 , . . . , pk ).
Hence,

Q Q( 2) Q( 2, 3)
is an infinite strictly ascending chain of intermediate fields of F/Q. Hence, F/Q must be of infinite dimension.

Let a F. Then there exist primes p1 , . . . , pn such that a Q( p1 , . . . , pn ). Since Q( p1 , . . . , pn )/Q is

a finite field extension, a is algebraic over Q by Theorem 16.1.18. Hence, F/Q is algebraic. Note that from this
example, it follows that [R : Q] = .
The above example provides us with a field extension F/Q which shows that the converse of Theorem 16.1.18
is not true. Since the field of algebraic numbers A contains F, we have [A : Q] = .
Definition 16.1.26 Let F/K and L/K be field extensions and : F L be a homomorphism. Then is called
a K-homomorphism if (a) = a for all a K.
Let F/K and L/K be field extensions and : F L be a K-homomorphism. Since is a nonzero
homomorphism, Ker 6= F. Therefore, Ker = {0} since the only ideals of F are F and {0}. This implies that
is one-one. Hence, is an isomorphism of F onto (F ). We simply call a K-isomorphism of F into L. If
L = F = (F ) and is a K-isomorphism of F into L, then we call a K-automorphism.
Theorem 16.1.27 Let F/K be an algebraic extension and : F F be a K-homomorphism. Then is an
automorphism.
Proof. As above is one-one. To show is an automorphism, it only remains to be shown that (F ) = F,
i.e., is onto F.Let a F. Let f (x) = a0 + a1 x + + ak xk K[x] be the minimal polynomial of a over K. Let b
be any root of f (x) in F. Then f ((b)) = a0 + a1 (b) + + ak (b)k = (a0 + a1 b + + ak bk ) = 0. Hence, (b)
is a root of f (x). Let F 0 be the subfield of F generated by all roots of f (x) over K that lie in F. Then F 0 /K is a
finite extension. Since maps a root of f (x) to a root of f (x), maps F 0 into F 0 . Since [F 0 : K] = [(F 0 ) : K],
it now follows that [F 0 : (F 0 )] = 1 by Theorem 16.1.20 and so F 0 = (F 0 ). Hence, a F 0 = (F 0 ) (F ).
Thus, is onto F.

16.1. Algebraic Extensions

235

Worked-Out Exercises

Exercise 1: Show that the polynomial x2 7 is irreducible in Q( 3)[x].

2
2
Solution: Suppose
x2 7 = (x (a+ b 3))(x

(c + d 3)), where a, b, c, d Q. Then x 7 = x ((a + c) + (b +

d) 3)x + (ac + 3bd + ad 3 + bc 3). This implies that

(a + c) +(b + d)3 = 0
ac + 3bd + ad 3 + bc 3 = 7.
Since {1,

3} is linearly independent over Q, a + c = 0 and b + d = 0. Hence,

a2 3b2 + (2ab) 3 = 7.

Thus, a2 3b2 = 7 and 2ab = 0. Hence, ab = 0. Suppose a = 0. Then 3b2 = 7. Now b = m

for some
n
integers m and n with gcd(m, n) = 1. Therefore, 3m2 = 7n2 , which contradicts the fundamental theorem
of arithmetic. Suppose b = 0. Then a2 = 7, which again
leads to a contradiction of the fundamental
theorem of arithmetic. Thus, x2 7 is irreducible in Q( 3)[x].

Exercise 2: Find
[Q(
3, 7) : Q( 3)] and [Q( 3) : Q]. Also, find a basis for Q( 3, 7)/Q( 3) and a basis for
Q( 3, 7)/Q.

Solution: By Worked-Out Exercise 1 (page 235), x2 7 is irreducible over Q( 3). Thus,

[Q( 3, 7) : Q( 3)] = deg(x2 7) = 2.

2
By Theorem 16.1.14,
{1, 7} is a basis

for Q( 3, 7)/Q( 3). Since x 3 is irreducible over Q, [Q( 3) :

Q] =2 and {1, 3} is a basis for Q( 3)/Q. Thus,

[Q( 3, 7) : Q] = [Q( 3, 7) : Q( 3)][Q( 3) : Q] = 2 2 = 4.

By Theorem 16.1.20, {1, 3, 7, 21} is a basis of Q( 3, 7)/Q.

Exercise 3: Find an element u R such that Q( 2, 3 7) = Q(u).





2 3 7. Since u =
2 3 7 Q( 2, 3 7),
Q(u)
Q( 2, 3 7).
Now 2 3 7 Q(u) implies
Solution: We claim
u=
that

that
14 2 = ( 2 3 7)3
Q(u).
Hence, 2 Q(u). Since 2, 2 3 7 Q(u), 3 7 Q(u). Therefore, Q( 2,

3
3
7) Q(u). Thus, Q( 2, 7) = Q(u).
Exercise 4:

(a) Let F be a field and a, b be members of a field containing F. Suppose that a and b are algebraic of
degree m and n over F, respectively. Suppose m and n are relatively prime. Show that [F (a, b) :
F ] = mn.
(b) Show that the result in (i) need not be true if m and n are not relatively prime.

Solution:

(a) Let f(x) F [x] be the minimal polynomial of a of degree m. Now f (x) F [x] F (b)[x]. Thus, a
satisfies a polynomial of degree m over F (b). Hence, [F (b)(a) : F (b)] m. Since F (b)(a) = F (a, b),
[F (a, b) : F (b)] m. Now [F (a, b) : F ] = [F (a, b) : F (b)][F (b) : F ] mn. Also,
[F (a, b) : F ] = [F (a, b) : F (b)][F (b) : F ] = [F (a, b) : F (b)]n.
Thus, n|[F (a, b) : F ]. Similarly, m|[F (a, b) : F ]. Since m and n are relatively prime, mn|[F (a, b) : F ].
Therefore, [F (a, b) : F ] mn. Consequently, [F (a, b) : F ] = mn.
1

(b) Let F = Q, a = 2 6 , and b = 2 4 . Then a is algebraic over F of degree 6 and b is algebraic over F
1
1
1
1
1
of degree 4. We claim that F (a, b) = F (2 12 ). Now b = (2 12 )3 F (2 12 ) and a = (2 12 )2 F (2 12 ).
1
1
11
1
1 1
1
Thus, F (a, b) F (2 12 ). Now 2 12 = 2 4 6 = 2 4 (2 6 ) F (a, b). Hence, F (a, b) = F (2 12 ). Since
1
1
x12 2 is the minimal polynomial of 2 12 , [F (2 12 ) : F ] = 12 6= 24 = 4 6.
Exercise 5: Consider the unique factorization domain F [t], where F is a field and t is transcendental over F. Show
that the polynomial x2 + tx + t F (t)[x] is irreducible over F (t). Also, show that x2 + tx + t F (x)[t] is
irreducible over F (x).
Solution:

Now t |/1, t|t, but t2 |/t. Note t is prime in F [t]. Thus, x2 + tx + t F (t)[x] is irreducible over F (t) by
Eisensteins criterion. If we consider x2 +tx+t as a polynomial in t over F (x), then x2 +tx+t = (x+1)t+x2 .
It follows that Eisensteins criterion does not apply. However, since (x + 1)t + x2 is of degree 1 in t, it is
irreducible over F (x).

236

16. Field Extensions

Exercise 6: Let K[u, v] denote the polynomial ring in two algebraic independent indeterminates u, v over the field K.
Let F denote the field of quotients K(u, v) of K[u, v]. Prove that the polynomial x2 + vx + u is irreducible
over F.
Solution:

Suppose x2 + vx + u is reducible over F. Then




f (u, v)
p(u, v)
2
x + vx + u = x +
x+
,
q(u, v)
g(u, v)
where p(u, v), q(u, v), f (u, v), g(u, v) K[u, v]. We may assume that p(u, v) and q(u, v) are relatively
prime in K[u, v] and also f (u, v) and g(u, v) are relatively prime in K[u, v]. Now
uq(u, v)g(u, v) = p(u, v)f (u, v).

(16.3)

Hence, g(u, v) divides p(u, v), p(u, v) divides ug(u, v), q(u, v) divides f (u, v), and f (u, v) divides uq(u, v).
Also,
p(u, v)
f(u, v)
v=
+
.
q(u, v)
g(u, v)
Consequently,
vq(u, v)g(u, v) = p(u, v)g(u, v) + q(u, v)f (u, v).
(16.4)
Therefore, g(u, v) divides q(u, v) and q(u, v) divides g(u, v). Thus,
g(u, v) = kq(u, v)
for some k K. Hence, g(u, v) and p(u, v) are relatively prime. Similarly, q(u, v) and f (u, v) are relatively
prime. Thus, p(u, v) divides u and f (u, v) divides u by Eq. (16.3). Hence,
either p(u, v) = k1 u or p(u, v) = k1 ,

(16.5)

either f (u, v) = k2 u or f (u, v) = k2

(16.6)

for some k1 , k2 K. Suppose that p(u, v) = k1 u and f (u, v) = k2 u. Then substituting into Eq. (16.4) we
obtain
vq(u, v)g(u, v) = k1 ug(u, v) + k2 uq(u, v).
Thus,
vq(u, v)g(u, v) = k1 ukq(u, v) + k2 uq(u, v).
Hence, vg(u, v) = (k1 k + k2 )u. However, this contradicts the algebraic independence of u, v over K.
Substituting the remaining possibilities in Eqs. (16.5) and (16.6) into Eq. (16.4), we also obtain a
contradiction of the algebraic independence of u, v over K. Thus, x2 + vx + u is irreducible over F.
Exercise 7: Let F = K(x, y), where K is a field and x, y are algebraically independent indeterminates over K. Show
that F 6= K(x)K(y), where
S
K(x)K(y) = { i (pi (x)/qi (x))(ui (y)/vi (y)) | pi (x), qi (x) K[x],
ui (y), vi (y) K[y], qi (x) 6= 0, vi (y) 6= 0}.
Solution:

/ K(x)K(y) else
Now x+y
Thus,

1
x+y

S
= ( i (fi (x)gi (y))/h(x)k(y), after obtaining a common denominator.
[
h(x)k(y) = (x + y)( (fi (x)gi (y)).
i

This implies that x + y divides h(x)k(y). Hence, x + y divides h(x) or k(y) since x + y is prime in the UFD
K[x, y], a contradiction of the algebraic independence of x, y over K.

Exercises

1. Show that Q( 3, 3) = Q( 3).

2. Let F/K be a field extension. Show that [F : K] = 1 if and only if F = K.

3. Consider the field extension R/Q.

(i) Show that 2 is transcendental over Q.

(ii) Show that is transcendental over Q.

4. Consider the field extension R/Q. Show that 3 is transcendental over Q.

16.2. Splitting Fields

237

5. Consider the field extension R/Q. Show that is transcendental over Q( 2).

6. Consider the field extension R/Q. Show that + 2 is transcendental over Q.

7. Let F/K be a field extension such that [F : K] < . Let p(x) be an irreducible polynomial in K[x].
Suppose p(c) = 0 for some c F. Prove that deg p(x) divides [F : K].

8. Find [Q( 3 5) : Q].

9. Show that Q( 3 5) = Q( 3, 5). Find [Q( 3 5) : Q].

10. Show that the polynomial x2 5 is irreducible over Q( 2).

s

11. Find the minimal polynomial of 2 + 5 over Q.

12. Let c = 5 3. Show that Q(c) = Q(c2 ).

13. Find [Q( 2, 5) : Q( 2)], [Q( 2, 5) : Q], a basis for Q( 2, 5)/ Q( 2), and a basis for Q( 2, 5)/Q.
14. Let F/K be a field extension and c F be algebraic over K. Let f (x) K[x]. Show that f(c) is algebraic
over K.
15. Prove that if [F : K] = p, p a prime, then F/K has no proper intermediate fields.
16. Let L and M be intermediate fields of the field extension F/K. Suppose that [L : K] is a prime. Prove
that either L M = K or L M.

17. Let F/K be a field extension, f (x) be a nonzero polynomial in K[x], and c F. If f (x) is algebraic over
K, prove that c is algebraic over K.
18. Let F/K be a field extension such that [F : K] = p, p a prime. Prove that if c F, c
/ K, then F = K(c).

19. Let F/K be a field extension and a, b F be algebraic over K. If a has degree m over K and b 6= 0 has
degree n over K, prove that the elements a + b, ab, a b, ab1 have degree at most mn over K.

20. Prove that 2 + 3, 2 3 have degree 4 over Q and that 2 3, 2/ 3 have degree 2 over Q. Find
the minimal polynomials of these elements over Q.
21. Let F/K be a field extension and R be a ring such that K R F. Prove that if every element of R is
algebraic over K, then R is a field.
22. Let F/K be a field extension and u, v F.
(i) Prove that K(u, u + v) = K(u, v).
(ii) If u and u + v are algebraic over K, prove that [K(u, v) : K] is finite and v is algebraic over K.
23. Answer the following statements true or false. If the statement is true, prove it. If it is false, give a
counterexample.
(i) Let F/K be a field extension and L be an intermediate field of F/K. Let V be a basis of F/L such that
1 V and U be a basis of L/K such that 1 U. Then U V is linearly independent over K.
(ii) Let F/K be a field extension and L be an intermediate field of F/K. Let V be a basis of F/L and U
be a basis of L/K. Then U V is a basis of F/K.
(iii) Let F/K be a field extension and c, d F. If K(c, d) = K(c), then d = f (c) for some polynomial
f (x) K[x].

16.2

Splitting Fields

Here we give some results concerning the existence of field extensions which are generated by roots of polynomials.
These results are basic to Galois theory.
Consider the polynomial ring K[x] over the field K. LetS
f (x) K[x]. In the quotient ring K[x]/ hf(x)i ,
i
we let g(x) denote the coset g(x) + hf(x)i . Thus, if g(x) = n
i=0 ki x , then by the definition of addition and
Sn
i
multiplication of cosets, we have that g(x) = i=0 ki x .
Theorem 16.2.1 (Kronecker) Let K be a field. If f (x) is a nonconstant polynomial in K[x], then there exists
a field extension F/K such that F contains a root of f (x).

Proof. Since K[x] is a unique factorization domain, there exist irreducible polynomials f1 (x), . . . , fn (x)
K[x] such that f(x) = f1 (x) fn (x). Thus, a root of any fi (x), i = 1, 2, . . . , n, is a root of f (x). Hence, it suces
to prove the theorem for f (x) irreducible in K[x]. The ideal hf(x)i is maximal in K[x] and so F = K[x]/ hf (x)i
is a field. Let be the natural homomorphism of K[x] onto K[x]/ hf (x)i . Since K hf (x)i = {0}, maps K

238

16. Field Extensions

one-one into F. Thus, say, K F, that is, we identify k K with k in F. Hence, (f (x)) = f(x) = f (x), where
f (x) = f (x) + hf (x)i and x = x + hf (x)i . Now (f(x)) = 0 and so f (x) = 0. Therefore, x is a root of f (x).
The field
F/K
16.2.1
Sextension
Sm in Theorem
Sm hasi some interesting properties. Consider the subring K[x] of
i
i
F. Then ( m
i=0 ki x ) =
i=0 ki x for all
i=0 ki x K[x] and so maps K[x] onto K[x]. Since also maps
K[x] onto F, we have F = K[x] = K(x). Thus, for f (x) irreducible in K[x], we have by Theorem 16.1.14 that
[F : K] = n and {1, x, . . . , xn1 } is a basis of F/K, where n = deg f (x).


2
Example 16.2.2
in R[x]. Now C = R/ x2 + 1 = R[x] = {a + bx | a, b R} is a field,
2x +
 1 is irreducible
2
where x = x + x + 1 . Since x = 1, we may call C the field of complex numbers. We may think of x as i.

4
4
Example 16.2.3
the polynomial
4 Consider

4 x
 3 Q[x]. By Eisensteins criterion, x 3 is irreducible in Q[x].
Set = x + x 3 in the field Q[x]/ x 3 . Then


Q[x]/ x4 3 = Q() = {a + b + c2 + d3 | a, b, c, d Q}

and {1, , 2 , 3 } is a basis of Q() over Q. Let us multiply two elements of Q() and determine the form
a + b + c2 + d3 for their product. Consider (1 + + 3 ) and (1 + 2 ). Then
(1 + + 3 )(1 + 2 ) = 1 + + 2 + 23 + 5 .

Now
1 + x + x2 + 2x3 + x5 = x(x4 3) + 1 + 4x + x2 + 2x3

using the division algorithm. Thus,

1 + + 2 + 23 + 5

=
=

(4 3) + 1 + 4 + 2 + 23
0 + 1 + 4 + 2 + 23 .

Hence,
(1 + + 3 )(1 + 2 ) = 1 + 4 + 2 + 23 .
Let us find (1 + + 3 )1 . Since x4 3 is irreducible over Q, the gcd of x4 3 and x3 + x + 1 is 1. Therefore,
there exist s(x), t(x) Q[x] such that
1 = s(x)(x4 3) + t(x)(1 + x + x3 ).
Thus,
1
1

=
=

s()(4 3) + t()(1 + + 3 )
0 + t()(1 + + 3 ).

Hence, t() = (1 + + 3 )1 . We have not really calculated t(), however. To do this calculation, we must know
the exact form of s(x) and t(x). The method for finding s(x) and t(x) is described below. Now by repeated use of
the division algorithm, we have
x4 3 = x(x3 + x + 1) + (x2 x 3)
x3 + x + 1
x2 x 3
x + 4

=
=
=

Thus, by back substitution, we obtain

23
23

=
=
=
=
=

(x + 1)(x2 x 3) + (x + 4)
(x + 5)(x + 4) + (23)
1
4
( 23
x 23
)(23) + 0.

x2 x 3 (x + 5)(x + 4)
x2 x 3 (x + 5)[x3 + x + 1 (x + 1)(x2 x 3)]
(x2 4x + 6)(x2 x 3) (x + 5)(x3 + x + 1)
(x2 4x + 6)[x4 3 x(x3 + x + 1)] (x + 5)(x3 + x + 1)
(x2 4x + 6)(x4 3) + (x3 + 4x2 7x 5)(x3 + x + 1).

This implies that

1=

1
1
(x2 4x + 6)(x4 3) + ( )(x3 + 4x2 7x 5)(x3 + x + 1).
23
23

Therefore,
t(x) =

5
4
7
1 3
x x2 + x + .
23
23
23
23

16.2. Splitting Fields

239

Consequently,
7
4
5
1
+ 2 3 .
23
23
23
23
Since is a root of x4 3 in Q(), we know by Corollary 11.1.10 that x divides x4 3 over Q(). In fact,
x4 3 = (x)(x3 +x2 +2 x+3 ). We know there exists a field Q()(2 ), where 2 is a root of x3 +x2 +2 x+3
over Q() by Theorem 16.2.1. Over the field Q()(2 ), x3 + x2 + 2 x + 3 factors into (x 2 )q(x), where q(x)
has degree 2. There exists a field Q()(2 )(3 ), where 3 is a root of q(x), and over the field Q()(2 (3 ), q(x)
factors into (x 3 )(x 4 ). Thus,
(1 + + 3 )1 =

x4 3 = (x )(x 2 )(x 3 )(x 4 )

over Q()(2 )(3 )(4 ). In this particular example, we can take 2 = and so Q() = Q()(2 ). Hence,
Q(, 2 , 3 , 4 ) = Q(, 3 ).
Now over Q(),
2

x4 3 = (x )(x + )(x2 + 2 ).

Also, x + is irreducible over Q(), a fact we leave as an exercise. Thus, [Q() : Q] = 4 and [Q()(3 ) :
Q()] = 2. Hence, [Q()(3 ) : Q] = 8.
Example 16.2.3 leads us to believe that given any polynomial f (x) in a polynomial ring K[x] over a field K,
there exists a field extension F/K such that f (x) factors completely into linear factors. This is indeed the case,
as we will presently show.
Definition 16.2.4 Let K be a field. A polynomial f (x) in K[x] is said to split over a field S K if f (x) can
be factored as a product of linear factors in S[x]. A field S containing K is said to be a splitting field for f(x)
over K if f(x) splits over S, but over no proper intermediate field of S/K.
Example 16.2.5 The field of complex numbers C is a splitting field for the polynomial x2 + 1 over R. This
follows since x2 + 1 = (x + i)(x i) in C[x] and C/R has no proper intermediate fields because [C : R] = 2.
(If C L R, where L is an intermediate field of C/R, then 2 = [C : L][L : R] and so either [C : L] = 1 or
[L : R] = 1. Thus, either C = L or L = R.) Note that C is not the splitting field of x2 + 1 over Q since x2 + 1
splits over Q(i) C.
Theorem 16.2.6 Let K be a field and f (x) be a polynomial in K[x] of degree n. Let F/K be a field extension.
If
f (x) = c(x c1 )(x c2 ) (x cn ) in F [x],
then K(c1 , c2 , . . . , cn ) is a splitting field for f (x) over K.

Proof. Since c1 , c2 , . . . , cn are the roots of f (x), f (x) splits over K(c1 , c2 , . . . , cn ). Let L be an intermediate
field of K(c1 , c2 , . . . , cn )/K such that f (x) splits over L. Since K[x] is a UFD, there is only one way f (x) can
split over L, namely, f(x) = c(x c1 )(x c2 ) (x cn ). Thus, c1 , c2 , . . . , cn L, whence L K(c1 , c2 , . . . , cn ).
Hence, K(c1 , c2 , . . . , cn ) is the smallest intermediate field over which f (x) splits.
The field Q(, 3 ) of Example 16.2.3 is a splitting field for x4 3 over Q. We now prove the existence of
splitting fields.
Theorem 16.2.7 Let K be a field and f (x) be a nonconstant polynomial over K. Then there is a splitting field
for f (x) over K.
Proof. If deg f (x) = 1, then K is a splitting field for f(x) over K. Assume the theorem is true for all
polynomials of degree n 1 ( 1). Suppose deg f(x) = n. There exists a field K1 K such that K1 contains a
root c1 of f (x) by Theorem 16.2.1. Thus, f (x) = (x c1 )f1 (x) in K1 [x] and deg f1 (x) = n 1. By the induction
hypothesis, there exists a field extension E/K1 such that f1 (x) splits in E[x]. Thus, f(x) splits in E[x], say,
f (x) = c(x c1 )(x c2 ) (x cn ).
By Theorem 16.2.6, the intermediate field K(c1 , c2 , . . . , cn ) of E/K is a splitting field for f (x) over K.

The intermediate field Q( 4 3, i 4 3) of C/Q is a splitting field for x4 3 over Q. The field
Q(,
3 ) of Example
16.2.3 is also a splitting field for x4 3 over Q. However, we cannot conclude that Q( 4 3, i 4 3) = Q(, 3 ).
Hence, splitting fields for a given polynomial over a field are not unique. We will show, however, that they are
unique up to isomorphism.

240

16. Field Extensions

Theorem 16.2.8 Let be an isomorphism of the field K onto the field K 0 . Let p(x) = k0 +k1 x+k2 x2 + +kn xn
be an irreducible polynomial in K[x] of degree n, c be a root of p(x) in some field extension of K, and p0 (y) =
(k0 ) + (k1 )y + (k2 )y 2 + + (kn )y n be the corresponding polynomial in K 0 [y]. Then p0 (y) is irreducible in
K 0 [y]. If c0 is a root of p0 (y) in some field extension of K 0 , then can be extended to an isomorphism 0 of K(c)
onto K 0 (c0 ) with 0 (c) = c0 . 0 is the only extension of such that 0 (c) = c0 .
Proof. By an argument similar to the one used in the proof of Theorem 11.1.14, can be uniquely extended
to an isomorphism of K[x] onto K 0 [y] so that for every polynomial b0 + b1 x + b2 x2 + + bm xm K[x],
(b0 + b1 x + b2 x2 + + bm xm ) = (b0 ) + (b1 )y + (b2 )y 2 + + (bm )y m .
We leave to the reader the verification that p0 (y) is irreducible in K 0 [y]. Let be the natural homomorphisms
of K[x] onto K[x]/ hp(x)i and 0 be the natural homomorphisms of K 0 [y] onto K 0 [y]/ hp0 (y)i . Then Ker =
Ker 0 . Hence, there exists an isomorphism of K[x]/ hp(x)i onto K 0 [y]/ hp0 (y)i such that 0 = .
By Theorem 16.1.11 and Corollary 16.1.12, there exist isomorphisms and 0 of K[x]/ hp(x)i onto K(c) and
K 0 [y]/ hp0 (y)i onto K 0 (c0 ), respectively. Thus, 0 is the map 0 1 . The situation is described by the
following diagram:

Let 00 be any other extension of to an isomorphism of K(c) onto K 0 (c0 ) such that 00 (c) = c0 . Now
{1, c, . . . , cn1 } is a basis for K(c)/K and {1, c0 , . . . , c0n1 } is a basis for K 0 (c0 )/K 0 . We have that
n1
[

00 (

i=0

00

ki ci ) =

n1
[

00 (ki )00 (ci ) =

i=0

n1
[

n1
[

(ki )c0i = 0 (

i=0

ki ci ).

i=0

Hence, = .
Corollary 16.2.9 Let E/K be a field extension and p(x) be an irreducible polynomial in K[x]. If a, b E are
roots of p(x), then K(a) ' K(b).
Proof. Let K = K 0 and be the identity map.

From Corollary 16.2.9, we have Q( 4 3) ' Q(i 4 3) in Example 16.2.3.

Theorem 16.2.10 Let be an isomorphism from the field K onto the field K 0 . Let
f (x) = k0 + k1 x + k2 x2 + + kn xn
be a polynomial in K[x] and
f 0 (y) = (k0 ) + (k1 )y + (k2 )y 2 + + (kn )y n
be the corresponding polynomial in K 0 [y].
If S is a splitting field for f(x) over K and S 0 is a splitting field for f 0 (y) over K 0 , then can be extended
to an isomorphism 0 of S onto S 0 .

16.2. Splitting Fields

241

Proof. The proof is by induction on deg f (x). If deg f(x) = 1, then K = S and K 0 = S 0 . In this case, we can
take 0 = . Assume the theorem is true for all polynomials of degree less than n (the induction hypothesis).
Suppose deg f (x) = n. Extend to an isomorphism of K[x] onto K 0 [y] as in Theorem 16.2.8. Let p(x) be
an irreducible factor of f (x) and c1 S be a root of p(x). Let c01 S 0 be a root of (p(x)) = p(y). Then by
Theorem 16.2.8, can be extended to an isomorphism 1 of K(c1 ) onto K 0 (c01 ). Extend 1 to an isomorphism
1 of K(c1 )[x] onto K 0 (c01 )[y]. Now f(x) = (x c1 )f1 (x) in K(c1 )[x] and f 0 (y) = (y c01 )f10 (y) in K(c01 )[y], where
f10 (y) = 1 (f1 (x)). Clearly S is a splitting field for f1 (x) over K(c1 ) and S 0 is a splitting field for f10 (y) over
0
0
K (c1 ). Since deg f1 (x) = n 1 = deg f10 (y), 1 can be extended to an isomorphism of S onto S 0 by the induction
hypothesis.
Corollary 16.2.11 Let f(x) K[x]. Any two splitting fields for f (x) over K are isomorphic.
Proof. Let S and S 0 be two splitting fields for f (x) over K. In Theorem 16.2.10, take K = K 0 and the
identity mapping on K.
Definition 16.2.12 Let F/K be a field extension and a, b F. Then a and b are called conjugates if a and b
are roots of the same irreducible polynomial over K.
We ask the reader to prove that the notion of conjugates defines an equivalence relation on F.
Example 16.2.13 Consider the field extension C/R. Let a, b R. Then a + bi and its complex conjugate a bi
are conjugates in the sense of Definition 16.2.12. This is obvious if b = 0. Suppose b 6= 0. Then a + bi
/ R. Let
f (x) = x2 2ax + (a2 + b2 ). Since a + bi
/ R, [R(a + bi) : R] = 2. Now a + bi is a root of f (x) and f(x) must be
irreducible over R. a bi is also a root of f (x).
In certain cases, the following theorem is useful in determining the irreducibility of a polynomial.
Theorem 16.2.14 Let F be a field. Let p be a prime in Z and a F. Then the polynomial xp a is reducible
over F if and only if xp a has a root in F.
Proof. Suppose f (x) = xp a F [x] is reducible. Let f (x) = g(x)h(x) for some g(x), h(x) F [x],
deg g(x) = m, 0 < m < p, and 0 < deg h(x) < p. Since f (x) is monic, we can take g(x) to be monic. By factoring
g(x) as a product of linear factors in a splitting field of g(x) over F, we see that the constant term of g(x) is
(1)m d for some d F. Since gcd(m, p) = 1, there exist integers s and t such that 1 = sm + tp. By Theorem
16.2.1, there is a field extension of F which contains a root of f (x). Let b be such a root of f (x).
Case 1: Suppose the characteristic of F is p. Since b is a root of f (x), bp = a. Thus,
(x b)p = xp bp = xp a
and all the roots of f(x) equal b. Now every root of g(x) is also a root of f (x). Thus, all the m roots of g(x) are
equal to b. Hence, bm = d. Now
ds = bms = b1pt = bbpt = bat .
Hence, b = ds at F and so f (x) has a root in F.
Case 2: Suppose that F is not of characteristic p. Let c be any other root of f (x). Then
cp = a = bp .
Hence, c = bu, where u = cp+1 bp1 and up = 1. From this, it follows that the roots of f (x) are of the form
b, bu1 , . . . , bup1 ,
where upi = 1. As in case 1, we have that the product of the roots of g(x) is
d = bm u1 u2 um1 = bm v,
where v p = 1. Now 1 = sm + tp implies that
bsm = v s ds = b1tp = bat .
Therefore, b = v s ds at . It then follows that
a = bp = (v s ds at )p = v sp (ds at )p = (ds at )p .
Thus, ds at F is a root of f (x).
The converse follows from Corollary 11.1.10.

242

16. Field Extensions

Worked-Out Exercises
Exercise 1: Find a splitting field S of x4 10x2 + 21 over Q. Find [S : Q] and a basis for S/Q.
Solution:

Note
thatx4 10x2 + 21 = (x2 3)(x2 7) over
Q. Therefore, a splitting field S of x4 10x2 + 21 over Q

is Q( 3, 7). Hence, [S : Q] = 4 and {1, 3, 7, 21} is a basis for S/Q, as can be seen from Worked-Out
Exercise 2 (page 235).

Exercise 2: Show that the splitting field of xp 1 over Q is of degree p 1, where p is a prime.
Solution:

Let f (x) = xp 1 Q[x]. Now f (x) = (x 1)g(x), where g(x) = xp1 + xp2 + + x + 1. Also,
g(x) =

xp 1
.
x1

Hence,

# $
#
$
(x + 1)p 1
p p2
p
p1
g(x + 1) =
=x
+
+ +
x
.
x
1
p1
 p 
 
. Therefore, by Eisensteins
Now since p is prime, p| pr for all 1 r p 1. Also, p2 does not divide p1
2i

criterion, g(x + 1) is irreducible over Q. Thus, g(x) is irreducible over Q. Let = e p , where i2 = 1.
Then the roots of f (x) are 1, , 2 , . . . , p1 and the roots of g(x) are , 2 , . . . , p1 . Now the splitting
field of f (x) is S = Q(1, , 2 , . . . , p1 ) = Q(). Also, g(x) is the minimal polynomial of over Q. Hence,
[S : Q] =p 1.

Exercise 3: Find the splitting field of the following polynomials over Q.

(a) x4 + 1.
(b) x6 + x3 + 1.
Solution:

(a) Let f (x) = x4 + 1. Then f (x) = (x2 + 2x + 1)(x2 2x + 1) over Q( 2). Therefore, the roots of
f (x) are

2i 2
2i 2
,
.
2
2

Let S be the splitting field of f (x) over Q. We claim that S = Q( 2, i). Now

2+i 2
2i 2
+
S
2=
2
2
and

2+i 2
2i 2

S.
2
2

This implies that i = i22 S. It now follows that Q( 2, i) S. Clearly S Q( 2, i). Consequently,

S = Q( 2, i). Now x2 2 is the minimal polynomial of 2 over Q andx2 + 1 is the minimal

2
polynomial

of i over Q. In fact, x + 1 is the minimal polynomial of i over Q( 2). Thus, [S : Q] = [S :

Q( 2)][Q( 2) : Q] = 2 2 = 4.

2i =

(b) Let f (x) = x6 + x3 + 1. Now (x9 1) = (x3 1)(x6 + x3 + 1). The roots of (x9 1) are 1, , 2 , . . . , 8
2i
and 1, 3 , 6 are the roots of (x3 1), where = e 9 . Hence, , 2 , 4 , 5 , 7 , 8 are the roots of
6
3
2
4
5
7
8
x + x + 1. Therefore, S = Q(, , , , , ) = Q() is the splitting field of x6 + x3 + 1 over Q.
Since x6 + x3 + 1 is irreducible over Q, [S : Q] =6.

Exercises
1. Prove that the polynomial p0 (y) in Theorem 16.2.8 is irreducible in K 0 [y].
2. Let F/K be an algebraic field extension. Define on F by for all a, b F, a b if and only if a and b
are conjugates. Prove that is an equivalence relation.

3. (i) Show that the polynomials x2 2x 1 and x2 2 have the same splitting field over Q.
(ii) Find a pair of polynomials in Q[x], other than the pair given in (i), which have the same splitting field
over Q.
4. Find a splitting field S of the polynomial x3 3 over Q. Find [S : Q] and a basis for S/Q.

16.3. Algebraically Closed Fields

243

5. Find a splitting field S of the polynomial x2 + x + [1] over Z5 . Find [S : Z5 ] and a basis for S/Z5 .
6. Find a splitting field S of the polynomial x2 + [1] over Z2 . Find [S : Z2 ] and a basis for S/Z2 .
7. Find a splitting field S of the polynomial x4 7x2 + 10 over Q. Find [S : Q] and a basis for S/Q.
8. Prove that Q( 12 +

3
i)
2

is a splitting field of the polynomial x4 + x2 + 1 over Q. Find [Q( 12 +

3
i)
2

: Q].

9. Let f (x) K[x], a polynomial ring over the field K. Let S be a splitting field for f (x) over K. Prove that
for any field L, S L K, S is a splitting field of f(x) over L.
10. Let f (x), g(x), and h(x) K[x], a polynomial ring over the field K. Suppose that S is a splitting field of
f (x) over K and f (x) = g(x)h(x). Prove that S contains a splitting field of g(x) over K.
11. Let f (x), g(x) K[x], a polynomial ring over the field K. Suppose that g(x) = f (ax+b), where 0 6= a, b K.
Prove that f (x) and g(x) have equal splitting fields over K.
12. Prove that if f (x) is a polynomial in K[x] of degree n, then [S : K] n!, where S is a splitting field of
f (x) over K.
13. Let K be a field and f1 (x), f2 (x), . . . , fn (x) K[x] be such that deg fi (x) 1, 1 i n. Show that there
exists a field extension F/K such that each fi (x) has a root in F.
14. Let F be a field of prime characteristic p and a F. Prove that xp x a is reducible over F if and only
if xp x a has a root in F.
15. Answer the following statements, true or false. If the statement is true, prove it. If it is false, give a counter
example.
(i) Let f (x) be an irreducible polynomial of degree n over a field K of characteristic 0. Let S = K(c1 ,
c2 , . . . , cn ) be a splitting field of f (x) over K, where c1 , c2 , . . . , cn are the roots of f (x). Then K(c2 , . . . , cn )
S.
(ii) The polynomial f (x) = x5 x 30 is reducible over Q.
(iii) C is a splitting field of some polynomial over Q.

16.3

Algebraically Closed Fields

The most important result in Steinitzs work in 1910 was his proof of the existence and uniqueness of an algebraic
closure of a field. In this section1 , we present these results.
Definition 16.3.1 A field K is called algebraically closed if for all f (x) K[x] with deg f (x) 1, f (x) has
a root in K.
Theorem 16.3.2 Let K be a field. The following conditions are equivalent.
(i) K is algebraically closed.
(ii) Every irreducible polynomial in K[x] is of degree 1.
(iii) Let f (x) K[x], deg f (x) 1. Then f (x) splits as a product of linear factors over K.
(iv) If F/K is an algebraic field extension, then F = K.
Proof. (i)(ii) Let p(x) K[x] and p(x) be irreducible. By (i), there exists a K such that p(a) = 0. Then
p(x) = (x a)g(x) for some g(x) K[x]. Since p(x) is irreducible, g(x) K. Hence, deg p(x) = 1.
(ii)(iii) Let f (x) K[x] and deg f (x) 1. Let f (x) = p1 (x) ps (x), where pi (x) K[x] is irreducible,
1 i s. Then deg pi (x) = 1, 1 i s. We may write pi (x) = ki (x ai ), where ki , ai K, 1 i s. Let
k = k1 ks . Then f (x) = k(x a1 ) (x as ). Thus, f (x) splits as a product of linear factors over K.
(iii)(iv) Let F/K be an algebraic field extension. Let c F and let p(x) K[x] be the minimal polynomial
of c over K. Since p(x) is irreducible, deg p(x) = 1 by (iii). Therefore, p(x) = ax + b K[x]. Since p(c) = 0,
ac + b = 0. Thus, c = a1 b K. Hence, K = F.
(iv)(i) Let f (x) K[x], deg f (x) 1. There exists a field extension F/K such that F has a root of f (x),
say, a. Then K(a)/K is an algebraic field extension. Therefore, K(a) = K and so a K. Thus, K is algebraically
closed.
We now prove the existence of an algebraically closed field. The following proof is due to Artin.
1 This section may be skipped without any discontinuity. The only place this section is needed is in Exercise 4 (Section
24.1).

244

16. Field Extensions

Theorem 16.3.3 Let K be a field. Then there exists an algebraically closed field F such that K is a subfield of
F.
Proof. We first construct an extension F1 /K such that if f (x) K[x] and deg f (x) 1, then f (x) has
a root in F1 . Let K be the set of all polynomials in K[x] of degree 1. Let S be a set which is in one-one
correspondence with K. For f (x) K, let xf be the corresponding element in S.
Consider the polynomial ring K[S]. Let I be the ideal of K[S] generated by all polynomials f(xf ) in K[S].
We claim that I 6= K[S]. Suppose that I = K[S]. Then there exists gi K[S] such that
g1 f1 (xf1 ) + g2 f2 (xf2 ) + + gn fn (xfn ) = 1.

(16.7)

Write xi = xfi , 1 i n. Since the polynomials gi , 1 i n, involve only a finite number of indeterminates,
say, x1 , x2 , . . . , xm , with m n, we may write Eq. (16.7) as
n
[

gi (x1 , x2 , . . . , xm )fi (xi ) = 1.

(16.8)

i=1

By Exercise 13 (page 243), there exists a finite extension L/K such that each polynomial fi , 1 i n, has a
root in L. Let ci be a root of fi in L, 1 i n. Let ci = 0 for n < i m. Substituting ci for xi , 1 i n, in
Eq. (16.8), we get 0 = 1, a contradiction. Hence, I 6= K[S].
Let M be a maximal ideal of K[S] such that I M. Let F1 = K[S]/M. Then F1 is a field containing an
isomorphic copy (K + M)/M of K. Thus, F1 can be regarded as a field extension of K. Also, if f K[x] and
deg f (x) 1, then xf + M is a root of f in F1 .
By induction, we can form a chain of fields
F1 F2 Fn
such that every polynomial of degree 1 in Fn has a root in Fn+1 . Let F =
i=1 Fn . Then F is a field. Let
f F [x]. Then f Fn [x] for some positive integer n. Thus f has a root in Fn+1 F. Hence, F is algebraically
closed.
Corollary 16.3.4 Let K be a field. Then there exists an algebraic field extension F/K such that F is algebraically closed.
Proof. By Theorem 16.3.3, there exists a field extension E/K such that E is algebraically closed. Let
F = {a E | a is algebraic over K}. Then F/K is an algebraic extension. Let f(x) F [x] and deg f (x) 1.
Then f (x) has a root c in E. Thus, c is algebraic over F. Since F/K is an algebraic extension, c is algebraic over
K. Hence, c F and so F is algebraically closed.
Definition 16.3.5 Let K be a field. A field F K is called an algebraic closure of K if
(i) F/K is algebraic and
(ii) F is algebraically closed.
For any field K, Corollary 16.3.4 guarantees the existence of an algebraic closure of K.
Lemma 16.3.6 Let F and L be fields with L algebraically closed. Let : F L be an isomorphism of F into
L. Let a be an algebraic element over F in some field extension of F. Let f (x) F [x] be the minimal polynomial
of a. Then can be extended to an isomorphism of F (a) into L and the number of such extensions is equal to
the number of distinct roots of f (x).
Proof. Let f (x) = a0 + a1 x + + an xn F [x] and f (x) = (a0 )+ (a1 )x+ + (an )xn L[x]. Since L
is algebraically closed there exists a root b of f (x) in L. Since a is algebraic over F, F (a) = F [a] by Corollary
16.1.12. Thus, if u F (a), then u = c0 + c1 a + + ck ak F [a]. Define : F (a) L by
(c0 + c1 a + + ck ak ) = (c0 ) + (c1 )b + + (ck )bk
for all c0 +c1 a+ +ck ak F (a). Suppose c0 +c1 a+ +ck ak = d0 +d1 a+ +ds as . Let (x) = c0 +c1 x+ +ck xk
and 0 (x) = d0 + d1 x + + ds xs . Then ( 0 )(a) = 0. Hence, f (x) divides ( 0 )(x). Thus, f (x) divides
( 0 )(x). Consequently, ( 0 )(b) = 0 and so (c0 )+(c1 )b+ +(ck )bk = (d0 )+(d1 )b+ +(ds )bs .
Thus, is well defined. Clearly is an isomorphism. The number of distinct roots of f (x) in the algebraic closure
of F is equal to the number of distinct roots of f (x) in L. For any extension : F (a) L, (a) is a root of

16.3. Algebraically Closed Fields

245

f (x). Therefore, the number of such extensions is equal to the number of distinct roots of f (x).
We close this section by showing that the algebraic closure of a field is unique up to isomorphism. Our proof
uses Zorns lemma while Steinitzs original proof used the equivalent concept of the axiom of choice.
Theorem 16.3.7 Let F/K be an algebraic field extension. Let L be an algebraically closed field and be an
isomorphism of K into L. Then there exists an isomorphism of F into L such that |K = .
Proof. Let S = {(E, ) | E is a subfield of F, K E and : E L is an isomorphism such that |K = }.
Since (K, ) S, S 6= . Let (E, ), (E 0 , 0 ) S. Define a relation on S by (E, ) (E 0 , 0 ) if E E 0
and 0 |E = . Then (S, ) is a poset. Let {(Ei , i )}i be a chain in S. Let E = i Ei . Then E is a field
and K E. Define : E L as follows: Let a E. Then a En for some n. Define (a) = n (a). Since
{(Ei , i )}i is a chain, is an isomorphism of E into L. Hence, (E, ) S and (E, ) is an upper bound of
{(Ei , i )}i . Hence, by Zorns lemma, S has a maximal element, say, (T, ). Suppose T 6= F. Let a F \T.
By Lemma 16.3.6, there exists an isomorphism : T (a) L such that |T = . From this, it follows that
(T (a), ) S, a contradiction of the maximality of (T, ). Thus, F = T.
Theorem 16.3.8 Let K be a field. Let F and F 0 be two algebraic closures of K. Then there exists an isomorphism of F onto F 0 such that (a) = a for all a K.
Proof. Let : K F 0 be such that (a) = a for all a K. Then is an isomorphism of K into F 0 . By
Theorem 16.3.7, there exists an isomorphism : F F 0 such that |K = . Now (F ) ' F. Thus, (F ) is
algebraically closed and K (F ). Now K (F ) F 0 . Since F 0 /K is algebraic, F 0 /(F ) is algebraic. Thus,
F 0 = (F ). Hence, F ' F 0 .

Exercises
1. If F is a field with a finite number of elements, prove that F is not algebraically closed.

246

16. Field Extensions

Chapter 17

Multiplicity of Roots
17.1

Multiplicity of Roots

In some cases, an irreducible polynomial p(x) of degree n over a field K does not have n distinct roots in a
splitting field of p(x) over K. In this chapter, we examine this situation.
If f (x) is a polynomial over K and c is a root of f (x) in some field F containing K, then the multiplicity
of c is the largest positive integer m such that (x c)m divides f (x) over F.
Definition 17.1.1 Let K be a field and p(x) be an irreducible polynomial in K[x] of degree n. Then p(x) is called
separable if it has n distinct roots in a splitting field S of p(x) over K; otherwise p(x) is called inseparable
over K. An arbitrary polynomial in K[x] is called separable if each of its irreducible factors in K[x] is separable;
otherwise it is called inseparable.
Definition 17.1.2 Let F/K be a field extension and c be an element of F which is algebraic over K. Then c
is called separable (or separable algebraic) over K if its minimal polynomial over K is separable; otherwise
c is called inseparable over K. If F/K is an algebraic extension, then F/K is called separable (or separable
algebraic) if every element of F is separable over K; otherwise F/K is called inseparable.
Let F/K be a field extension and L be an intermediate field of F/K. Let c F and suppose c is separable
over K. Then c must be separable over L. This follows since if f (x) and p(x) are the minimal polynomials of c
over K and L, respectively, then p(x)|f (x). Hence, c cannot be a multiple root of p(x) since it is not one of f (x).
Example 17.1.3 Consider the field K(t), where K is a field of prime characteristic p and t is transcendental
over K. It follows that the polynomial xp tp is irreducible over K(tp ) by Eisensteins criterion since tp is
irreducible in K[tp ]. Now xp tp factors into
(x t)(x t) (x t) = (x t)p
~}


p tim es

over K(t). Thus, K(t) is a splitting field for x tp over K(tp ) and we see that xp tp has only one root in K(t),
namely, t. (Since t
/ K(tp ), we can also use Theorem 16.2.14 to deduce that xp tp is irreducible over K(tp ).)
p
p
Thus, x t , t, and K(t) are inseparable over K(tp ). Note that t has multiplicity p over K(tp ).
Let K be a field and
f(x) = k0 + k1 x + + kn xn

be a polynomial in K[x]. Then by the formal derivative, f 0 (x), of f (x) we mean the polynomial
f 0 (x) = k1 + + iki xi1 + + nkn xn1 K[x].
Let K be a field and f (x), g(x) K[x]. The following properties of formal derivatives are easily verified:
(f (x) + g(x))0
(f (x)g(x))0
(kf (x))0

=
=
=

f 0 (x) + g 0 (x),
f(x)g 0 (x) + f 0 (x)g(x),
kf 0 (x) for all k K

and if f (x) = x, then f 0 (x) = 1.

247

248

17. Multiplicity of Roots

Theorem 17.1.4 Let K be a field and f (x) K[x], f (x) 6= 0. Let a be a root of f (x) in some extension field F
of K. Then a is a multiple root of f (x) if and only if f 0 (a) = 0.
Proof. Suppose a is a multiple root of f (x). Then (x a)2 divides f (x). Hence,
f (x) = (x a)2 g(x)
for some g(x) F [x]. Now f 0 (x) = (x a){(x a)g 0 (x) + 2g(x)}. Therefore, f 0 (a) = 0. Conversely, suppose
f 0 (a) = 0. Then deg f (x) 2. By the division algorithm,
f (x) = (x a)2 q(x) + h(x)
for some q(x), h(x) F [x], where either h(x) = 0 or deg h(x) 1. Suppose h(x) 6= 0. Since f (a) = 0, h(a) = 0.
Thus, deg h(x) = 1 and a is a root of h(x). Hence, h(x) = b(x a) for some 0 6= b K. This implies that
f (x) = (x a)2 q(x) + b(x a)
and so
Therefore,

f 0 (x) = (x a){(x a)q 0 (x) + 2q(x)} + b

0 = f 0 (a) = b,

a contradiction. Hence, h(x) = 0 and so f (x) = (x a)2 q(x). Consequently, a is a multiple root of f (x).
Theorem 17.1.5 For any field K, an irreducible polynomial p(x) in K[x] is separable if and only if p(x) and
its formal derivative p0 (x) are relatively prime.
Proof. Let d(x) denote the gcd of p(x) and p0 (x). Suppose p(x) is separable. Let c be a root of p(x) in some
field containing K. Then p(x) = (x c)f (x) for some f (x) K(c)[x]. Since p(x) is separable, f (c) 6= 0. Now
p0 (x) = f (x) + (x c)f 0 (x) and so p0 (c) = f (c) + 0 6= 0. Hence, c is not a root of d(x). But every root of d(x)
must be a root of p(x) since d(x)|p(x). Thus, since we have just seen that d(x) and p(x) have no common roots,
d(x) has no roots. Therefore, d(x) = 1.
Conversely, suppose that d(x) = 1. Let c be any root of p(x). Let m denote the multiplicity of c. Then
p(x) = (x c)m f(x)
over K(c) and c is not a root of f (x). Now
p0 (x)

=
=

m(x c)m1 f (x) + (x c)m f 0 (x)

(x c)m1 [mf (x) + (x c)f 0 (x)].

Thus, (x c)m1 is a common divisor of p0 (x) and p(x). Hence,

(x c)m1 |d(x).
Since d(x) = 1, m = 1. Consequently, p(x) has no repeated roots.
Theorem 17.1.6 For any field K, an irreducible polynomial p(x) in K[x] is separable if and only if p0 (x) 6= 0.
Proof. Let d(x) denote the gcd of p(x) and p0 (x). Suppose p(x) is separable. If p0 (x) = 0, then d(x) =
p(x) 6= 1, a contradiction of Theorem 17.1.5. Conversely, suppose p0 (x) 6= 0. Since p(x) is irreducible, the only
common divisors of p(x) and p0 (x) are 1 and p(x). Since 1 deg p0 (x) < deg p(x), 1 is the only common divisor
of p0 (x) and p(x). Hence, d(x) = 1. Thus, p(x) is separable by Theorem 17.1.5.
Corollary 17.1.7 Let K be a field of characteristic 0. Then every nonconstant polynomial in K[x] is separable.
Proof. Let f (x) be any nonconstant polynomial in K[x] and p(x) = k0 +k1 x + k2 x2 + + kn xn be any
irreducible factor of f(x), where n 1. Then there exists i > 0 such that ki 6= 0. Hence, iki 6= 0 since K has
characteristic 0. Thus, p0 (x) 6= 0 and so p(x) is separable by Theorem 17.1.6. Hence, f (x) is separable.
Example 17.1.8 Consider the irreducible polynomial p(x) = xp tp over K(tp ) of Example 17.1.3. Then p0 (x) =
pxp1 = 0. Thus, xp tp is inseparable over K(tp ).

17.1. Multiplicity of Roots

249

Theorem 17.1.9 Let K be a field of characteristic p > 0. Then an irreducible polynomial p(x) = k0 + k1 x +
k2 x2 + + kn xn over K is inseparable if and only if p(x) = q(xp ) for some q(xp ) K[xp ].
Proof. Clearly p0 (x) = 0 if and only if iki = 0 for all i = 1, 2, . . . , n. Thus, p0 (x) = 0 if and only if p|i for
those i such that ki 6= 0. Hence, p0 (x) = 0 if and only if p(x) = q(xp ) for some q(xp ) K[xp ]. The conclusion
now follows from Theorem 17.1.6.
Let K be a field of characteristic p > 0. Let K p = {ap | a K}. The reader is asked to verify in Exercise 7
(page 256) that K p is a subfield of K.
Definition 17.1.10 Let K be a field. Then K is called perfect if every algebraic extension of K is separable.
Example 17.1.11 By Corollary 17.1.7, every field of characteristic 0 is perfect.
The following theorem gives a necessary and sucient condition for a field to be perfect.
Theorem 17.1.12 Let K be a field of characteristic p > 0. Then K is perfect if and only if K = K p .
Proof. Suppose K is perfect. Let a K and F be a splitting field of xp a K[x]. Then F/K is a separable
extension. Let b F be a root of xp a. Then
xp a = (x b)p .
Let p(x) K[x] be the minimal polynomial of b. Then p(x) has distinct roots. If deg p(x) > 1, then since
p(x)|(x b)p , p(x) has multiple roots, a contradiction. Hence, deg p(x) = 1. This implies that b K. Hence,
a = bp K p . Thus, K = K p .
Conversely, suppose K = K p . Let F/K be an algebraic field extension. Let a F and f(x) K[x] be
the minimal polynomial of a. Suppose f (x) is not separable. Then by Theorem 17.1.9, f (x) = g(xp ) for some
g(x) K[x]. Hence,
f (x) = a0 + a1 xp + + ak xpk ,
ai K, 1 i k. Since K = K p , ai = bpi for some bi K, 1 i k. Therefore,
f (x) = (b0 + b1 x + + bk xk )p ,
a contradiction, since f (x) is irreducible over K. Hence, f (x) is separable. Thus, F/K is a separable extension.
Consequently, K is perfect.
Example 17.1.13 Let K be a finite field of characteristic p. Define : K K p by (a) = ap . Then is
a homomorphism. Suppose that (a) = (b). Then ap = bp and so (a b)p = 0. Since K is a field, K has
no nonzero nilpotent elements. Thus, a = b and so is one-one. Hence, |K| = |(K)| |K p | |K| and so
|K| = |K p | . Since K p is a subfield of K and K is finite, K = K p . Hence, K is perfect. We have thus shown that
every finite field is perfect.
If p(x) = k0 + k1 x + k2 x2 + + kn xn is irreducible and inseparable over K in Theorem 17.1.9, then
2
2
p(x) = k0 + kp xp + + kpm (xp )m = q(xp ). It may be the case that p(x) = q(xp ) = s(xp ) in K[xp ]. However,
pe
pe
pe
there exists a largest positive integer e such that p(x) = t(x ) for some t(x ) K[x ]. If n = deg p(x), then
pe |n.
Definition 17.1.14 Let K be a field of characteristic p > 0 and p(x) be an irreducible polynomial in K[x].
e
e
e
Let e be the largest nonnegative integer such that p(x) = q(xp ) for some q(xp ) K[xp ]. Then e is called the
e
exponent of inseparability of p(x) and p is called the degree of inseparability of p(x). If n denotes the
degree of p(x), then n0 = pne is called the degree of separability or reduced degree of p(x) over K.
By Theorem 17.1.9, p(x) in Definition 17.1.14 is separable if and only if e = 0.
Theorem 17.1.15 Let K be a field of characteristic p > 0 and
e

p(x) = kn0 (xp )n0 + + k1 xp + k0

be an irreducible polynomial in K[x], where e is the exponent of inseparability of p(x). Then the polynomial
s(y) = kn0 y n0 + + k1 y + k0 K[y]
is irreducible and separable over K.

250

17. Multiplicity of Roots

e

Proof. If s(y) = f (y)g(y) K[y], then p(x) = f (xp )g(xp ), contrary to the fact that p(x) is irreducible in
e+1
K[x]. Thus, s(y) is irreducible in K[y]. If s(y) = q(y p ) for some q(y p ) K[y p ], then p(x) = q(xp ), contrary to
the maximality of e. Hence, s(y) is separable.
Example 17.1.16 Consider the polynomial p(x) = x2p + txp + t over the field K(t), where K is a field of
characteristic p > 0 and t is transcendental over K. By Eisensteins criterion, p(x) is irreducible over K(t). Now
p(x) = (xp )2 + txp + t K(t)[x] and so p(x) is inseparable over K(t). The inseparability exponent e of p(x) equals
1. Thus, x2 + tx + t is separable over K(t).
Definition 17.1.17 Let F/K be a field extension. F is called a simple extension if F = K(a) for some a F.
Such an element a is called a primitive element.
Theorem 17.1.18 Let K be an infinite field and K(a, b)/K be a field extension with a algebraic over K and b
separable algebraic over K. Then there exists an element c K(a, b) such that K(a, b) = K(c), i.e., K(a, b)/K
is a simple extension.
Proof. Let f (x) and g(x) be the minimal polynomials of a and b over K with degrees n and m and roots
a = a1 , a2 , . . . , an , and b = b1 , b2 , . . . , bm , respectively, in some extension field of K. Since b is separable, all bi s
are distinct. Also, since K is infinite, there exists s K such that a + sb 6= ai + sbj , i.e.,
s 6=

ai a
bbj

for all 1 i n, 1 < j m. Let c = a + sb. Then c sbj 6= ai for all 1 i n, 1 < j m. Also, K(c) K(a, b).
Let h(x) = f (c sx) K(c)[x]. Now
h(b) = f (c sb) = f(a) = 0.
Thus, g(x) and h(x) have the common root b of multiplicity 1 in the field K(a, b). Now
h(bj ) = f(c sbj ) 6= 0
for all 1 < j m. Thus, g(x) and h(x) have only root b in common. Let d(x) K(c)[x] be the greatest common
divisor of g(x) and h(x). Then b is a root of d(x). Every root of d(x) is also a root of g(x) and h(x). Since
g(x) and h(x) have no roots other than b in common in any field and b is of multiplicity 1, d(x) is of degree 1.
Hence, d(x) = x b. But then b K(c). Thus, a = c sb K(c). Therefore, K(a, b) K(c) K(a, b) and so
K(c) = K(a, b).
Corollary 17.1.19 Let K be an infinite field. Let a1 , a2 , . . . , an be elements in some field containing K. Suppose
that a1 is algebraic and a2 , . . . , an are separable algebraic over K. Then there exists an element c K(a1 , . . . ,
an ) such that K(c) = K(a1 , . . . , an ), i.e., K(a1 , . . . , an )/K is a simple extension.
Proof. The result follows by induction on n and Theorem 17.1.18.
Corollary 17.1.20 Let F/K be a field extension and the characteristic of K be 0. Let a1 , a2 , . . . , an F be
algebraic over K. Then K(a1 , . . . , an )/K is a simple extension.
Proof. The proof follows by Corollaries 17.1.7 and 17.1.19.

Example 17.1.21 Consider Q( 2, i). Now 1 6=

Theorem 17.1.18, with s = 1 there.

2( 2)
i(i)

2
i .

Thus, Q( 2, i) = Q( 2 + i) by the proof of

Theorem 17.1.22 (Artin) Let K be an infinite field. Let F/K be a finite field extension. Then F/K is a
simple extension if and only if there are only a finite number of intermediate fields of F/K.
Proof. Suppose F/K is a simple extension. Let F = K(a) for some a F. Let L be an intermediate field
of F/K and f(x) be the minimal polynomial of a over L. Let L0 be the field generated by K and the coecients
of f (x). Then L0 L and f (x) is also the minimal polynomial of a over L0 . Hence,
[F : L] = deg f (x) = [F : L0 ].
Thus, [L : L0 ] = 1 and so L = L0 . Let g(x) be the minimal polynomial of a over K. Then f (x) divides g(x). Now
g(x) has only a finite number of distinct monic factors. Hence, the number of intermediate fields is finite.

17.1. Multiplicity of Roots

251

Conversely, suppose there are only a finite number of intermediate fields of F/K. Let a, b F. We first show
that K(a, b)/K is a simple extension. Let c K and Fc = K(a + cb). Then for all c K, Fc is an intermediate
field of K(a, b)/K. Since the number of intermediate fields is finite and K is infinite, there exists c, d K, c 6= d
such that Fc = Fd . Then
b = (c d)1 (a + cb a db) Fc .
Hence, a = a + cb cb Fc . Thus, K(a, b) = Fc = K(a + cb), i.e., K(a, b)/K is a simple extension. Now for all
a F, K(a) is an intermediate field of F/K. Since [F : K] is finite, [K(a) : K] is finite. Let
A = {[K(a) : K] | a F }.
Then A is a finite subset of Z. Let a F be such that the maximum of A = [K(a) : K]. Suppose F 6= K(a). Let
b F be such that b
/ K(a). Then K(a) K(a, b). There exists c F such that K(a, b) = K(c). Therefore,
K(a) K(c). Hence, [K(c) : K] > [K(a) : K], a contradiction to the maximality of [K(a) : K]. Consequently,
F = K(a), i.e., F/K is a simple extension.
Let F/K be a field extension. In the next chapter, we show that every finite extension of a finite field is a
simple extension (Corollary 18.1.8, page 258). Hence, from this and Theorem 17.1.22, it follows that F/K is a
simple extension if and only if there are only a finite number of intermediate fields of F/K.
We now focus our attention on the study of separable algebraic and purely inseparable extensions.1
e

Theorem 17.1.23 Let K be a field of characteristic p > 0 and f(x) = xp k be a polynomial over K, where e
is a positive integer. Then f(x) is irreducible over K if and only if k
/ Kp.
e1

k0 )p ,
Proof. Suppose f (x) is irreducible over K. If k = k0p K p for some k0 K, then f (x) = (xp
p
p
/ K . Let p(x) be
contrary to the fact that f (x) is irreducible over K. Hence, k
/ K . Conversely, suppose k
a nonconstant monic irreducible factor of f (x) in K[x] and c be a root of p(x). Then c is a root of f (x) and so
e
e
cp = k and f (x) = (x c)p over K(c). Since K(c)[x] is a unique factorization domain, it follows that p(x) is
some power of (xc), say, p(x) = (xc)m . Thus, mn = pe for some n so that m = pr and n = ps for nonnegative
r
r
e
r
s
s
integers r and s. Therefore, p(x) = xp cp in K[x]. If s > 0, then k = cp = (cp )p K p K p , which is
p
contrary to the assumption k
/ K . Thus, s = 0 and so r = e. Hence, p(x) = f (x), i.e., f(x) is irreducible.
Definition 17.1.24 Let F/K be a field extension of characteristic p > 0. Let c F be a root of the irreducible
polynomial p(x) in K[x]. If the degree of separability n0 of p(x) equals 1, then c is said to be purely inseparable
over K. If every element of F is purely inseparable over K, then F/K is called a purely inseparable extension.
e

In Theorem 17.1.15, let c be a root of p(x). Then cp is a root of s(y). We have K(c) K(cp ) K and c is
e
e
e
e
e
e
e
a root of the polynomial xp cp over K(cp ). It follows that xp cp is irreducible over K(cp ), K(c)/K(cp )
pe
is purely inseparable, and K(c )/K is separable.
Theorem 17.1.25 Let F/K be a field extension of characteristic p > 0 and c be an element of F. Then c is
m
purely inseparable over K if and only if cp K for some nonnegative integer m.
Proof. Let c be purely inseparable over K. Then the degree of separability n0 of the minimal polynomial
e
p(x) of c equals 1. Thus, p(x) = xp + k in K[x], where e is the exponent of inseparability of p(x) over K.
e
e
m
Therefore, cp + k = 0 or cp = k K. Hence, we can take m = e. Conversely, suppose cp K. Let e be
e
e
the smallest nonnegative integer such that cp K. Then c is a root of the polynomial xp k over K, where
e
e
k = cp . If xp k is not irreducible over K, then e > 0 and k = k0p for some k0 K by Theorem 17.1.23. In
e1
e1
pe
k0 )p . Thus, (cp
k0 )p = 0 and since a field has no nonzero nilpotent elements,
this case, x k = (xp
e1
e1
e
cp
k0 = 0 or cp
= k0 K. However, this contradicts the minimality of e. Thus, xp k is irreducible over
pe
K. Clearly the degree of separability of x k is 1. Therefore, c is purely inseparable over K.
Corollary 17.1.26 Let F/K be a field extension of characteristic p > 0 and c F.
(i) If c is algebraic over K, then c is purely inseparable over K if and only if the minimal polynomial of c
e
e
e
over K is xp cp , where e is the smallest nonnegative integer such that cp K.
e
(ii) If c is purely inseparable over K, then [K(c) : K] = p for some nonnegative integer e.
(iii) If c is purely inseparable and separable algebraic over K, then c K.
1 The remainder of this section may be skipped without any discontinuity. The only place this material is needed is in
Example 24.2.8.

252

17. Multiplicity of Roots

Proof. The proof of (i) follows from Theorem 17.1.25. Statement (ii) is an immediate consequence of
statement (i). For the proof of statement (iii), we see that since c is purely inseparable over K the minimal
e
polynomial of c over K has the form xp k. Since c is separable algebraic over K, the exponent of inseparability
e
of xp k is 0, i.e., e = 0. Thus, x k is the minimal polynomial of c over K, whence c = k K.
Corollary 17.1.27 Let F/K be a field extension of characteristic p > 0.
(i) If F = K(M) for some subset M of F such that every element of M is purely inseparable over K, then
F/K is a purely inseparable extension.
(ii) Let L be an intermediate field of F/K. Then F/K is purely inseparable if and only if F/L and L/K are
purely inseparable.
(iii) The set of all elements of F which are purely inseparable over K is an intermediate field of F/K.
Proof. (i) Let c be an element of F. Then there exists a finite subset {m1 , m2 , . . . , ms } of M such that
[
c=
ki1 ...is mi11 miss ,
i1 ,...,is

where here we are using the fact that F = K[M] since F/K is necessarily an algebraic extension. Let e =
ei
max{e1 , . . . , es }, where ei is a nonnegative integer such that mpi K, i = 1, . . . , s. Then
[ pe
e
e
e
cp =
ki1 ...is (mp1 )i1 (mps )is K.
i1 ,...,is

Hence, c is purely inseparable over K.

(ii) Suppose that F/K is purely inseparable. Let c F. Then there exists a nonnegative integer e such that
e
e
cp K and so cp L. Thus, F/L is purely inseparable. L/K is purely inseparable since every element of L
is an element of F. Conversely, suppose F/L and L/K are purely inseparable. Let c F. Then there exists a
m
nonnegative integer m such that cp L. Since L/K is purely inseparable, there exists a nonnegative integer n
m
n
m+n
such that (cp )p K. Therefore, cp
K so that c is purely inseparable over K.
(iii) Let J denote the set of all elements of F which are purely inseparable over K. Then K J and so
e
f
J 6= . Let c, d J. Then cp K and dp K for some nonnegative integers e and f. Let n = max{e, f }. Then
n
n
n
n
n
n
(c d)p = cp dp K. Hence, c d J. If d 6= 0, then (cd1 )p = cp (dp )1 K. Thus, cd1 J. Hence,
J is an intermediate field of F/K.
Theorem 17.1.25 and Corollary 17.1.27(i) make it quite easy to construct examples of purely inseparable field
extensions.
Example 17.1.28 Let J be any field of characteristic p > 0; e.g., J = Zp . Let F = J(x, y, z), where x, y, z
2
3
are algebraically independent over J. Set K = J(xp , y p , z p ). Then F/K is purely inseparable since x, y, z are
purely inseparable over K. It can be shown that [F : K] = p6 since x, y, z are algebraically independent over J.
2
3
3
Since xp , y p , z p K, we have F p K.
e

For any field F of prime characteristic p, F/F p is a purely inseparable field extension for any nonnegative
integer e.
The following example is essentially the same as that in Example 17.1.28.
Example 17.1.29 Let J be any field of characteristic p > 0. Let K = J(x, y, z), where x, y, z are algebraically
independent over J. Let F = J(a, b, c), where a is a root of the polynomial tp x over K, b is a root of the
2
3
polynomial tp y over K(a), and c is a root of the polynomial tp z over K(a, b). Then F/K is purely
3
1
2
3
inseparable, [F : K] = p6 , and F p K. One often writes a = xp , b = y p , and c = z p .
Example 17.1.30 Let J be any field of characteristic p > 0. Let K = J(t), where t is transcendental over J. Let
1
2
3
1
2
3
n
F = K(tp , tp , tp , . . .). Then F/K is purely inseparable by Corollary 17.1.27. Since [K(tp , tp , tp , . . . , tp ) :
1
2
3
n+1
K(tp , tp , tp , . . . , tp
)] = p for all positive integers n, [F : K] = . There does not exist a positive
e
integer e such that F p K.
Example 17.1.31 Let J be any field of characteristic p > 0. Let K = J(x1 , x2 , x3 , . . .), where x1 , x2 , x3 , . . .
1
2
3
are algebraically independent over J. Let F0 = K(xp1 , xp2 , xp3 , . . .). Then F0 /K is purely inseparable and
2
2
2
2
[F0 : K] = . Let F1 = K(xp1 , xp2 , xp3 , . . .). Then F1 /K is purely inseparable, [F1 : K] = , and F1p K.

17.1. Multiplicity of Roots

253

We now turn our attention to separable extensions.

Theorem 17.1.32 Let F/K be a field extension of characteristic p > 0. If F/K is separable algebraic, then
F = K(F p ). If [F : K] < and F = K(F p ), then F/K is separable algebraic.
Proof. Suppose F/K is separable algebraic. Now every element of F is purely inseparable over F p and thus
purely inseparable over K(F p ). Every element c of F is separable algebraic over K and thus separable algebraic
over K(F p ). Thus, every element c of F is in K(F p ) by Corollary 17.1.26(iii). Hence, F K(F p ), so that
F = K(F p ). Conversely, suppose [F : K] < and F = K(F p ). Let a be any element of F. Since [F : K] < ,
a is algebraic over K. If a is not separable over K, then the minimal polynomial of a over K has the form
(xp )n + + k1 xp + k0 .
Therefore, 0 = anp + +k1 ap + k0 1 with not all the ki = 0. Hence, 1, ap , . . . , anp are linearly dependent over
K. By Theorem 16.1.14, 1, a, a2 , . . . , an , . . . , anp1 are linearly independent over K, whence 1, a, a2 , . . . , an are
linearly independent over K.
We now show that this is impossible by showing that whenever n elements b1 , . . . , bn of F are linearly
independent over K, then the elements bp1 , . . . , bpn are linearly independent over K. We can assume that b1 , . . . , bn
is a basis of F/K since any linearly independent set over K can be extended to a basis of F/K, in particular, the
linearly independent set {1, a, . . . , an }. By Exercise 7 (page 256), the mapping : F F p defined by (c) = cp
for c F is an isomorphism, which maps K onto K p . Thus, since b1 , . . . , bn is a basis of F/K, bp1 , . . . , bpn is a
basis of F p /K p . Hence, bp1 , . . . , bpn spans F p over K p . Consequently, bp1 , . . . , bpn spans K(F p ) over K; i.e., F over
K. Since F has dimension n over K and the n elements bp1 , . . . , bpn span F over K, the elements bp1 , . . . , bpn must
be a basis for F over K.
The field extension F/K of Example 17.1.30 shows that the finiteness condition [F : K] < cannot be
dropped in the above theorem. We have F = K(F p ), F/K is not separable algebraic, in fact, F/K is purely
inseparable.
Corollary 17.1.33 Let F/K be a field extension of characteristic p > 0.
(i) Let a be an element of F. Then K(a) = K(ap ) if and only if K(a)/K is separable algebraic.
(ii) Let a1 , a2, . . . , an be elements of F. Then K(a1 , . . . , an )/K is separable algebraic if and only if a1 is
separable algebraic over K and ai is separable algebraic over K(a1 , . . . , ai1 ), i = 2, 3, . . . , n.
Proof. (i) If K(a) = K(ap ), then a cannot be transcendental over K and so a must be algebraic over K.
By Theorem 17.1.32, K(a) = K(K(a)p ) if and only if K(a)/K is separable algebraic. We thus have the desired
result since K(K(a)p ) = K(ap ).
(ii) Suppose K(a1 , . . . , an )/K is separable algebraic. Then a1 , . . . , an are separable algebraic over K. By
the discussion following Definition 17.1.2, ai is clearly separable algebraic over K(a1 , . . . , ai1 ), i = 2, 3, . . . , n.
Conversely, suppose a1 is separable algebraic over K and ai is separable algebraic over K(a1 , . . . , ai1 ), i =
2, 3, . . . , n. Then K(a1 ) = K(ap1 ), . . . , K(a1 , . . . , ai1 )(ai ) = K(a1 , . . . , ai1 )(api ), i = 2, 3, . . . , n. Thus, K(a1 ,
. . . , an ) = K(ap1 , . . . , apn ) = K([K(a1 , . . . , ai1 )]p ). The conclusion now holds from Theorem 17.1.32.
Corollary 17.1.34 Let F/K be a field extension of characteristic p > 0.
(i) If F = K(M) for some subset M of F such that every element of M is separable algebraic over K, then
F/K is separable algebraic.
(ii) Let L be an intermediate field of F/K. Then F/K is separable algebraic if and only if F/L and L/K are
separable algebraic.
(iii) The set of all elements of F which are separable algebraic over K is an intermediate field of F/K.
Proof. (i) Let a F. There exists a finite subset {m1 , . . . , ms } of M such that a K(m1 , . . . , ms ). Since each
mi is separable algebraic over K, we have by Corollary 17.1.33(ii) that K(m1 , . . . , ms )/K is separable algebraic.
Hence, a and thus F/K is separable algebraic.
(ii) Suppose F/K is separable algebraic. Then F/L is separable algebraic by the discussion following Definition 17.1.2. L/K is separable algebraic since every element of L is an element of F. Suppose F/L and L/K
are separable algebraic. Let a F. Let c0 , c1 , . . . , cn L be the coecients of the minimal polynomial p(x)
of a over L. Since a is separable algebraic over L, a is separable algebraic over K(c0 , c1 , . . . , cn ). (p(x) is also
the minimal polynomial of a over K(c0 , c1 , . . . , cn ).) Since c0 , c1 , . . . , cn L and L/K is separable algebraic,
K(c0 , c1 , . . . , cn )/K is separable algebraic by Corollary 17.1.33(ii). Thus, a and so F is separable algebraic over
K.

254

17. Multiplicity of Roots

(iii) Let S denote the set of elements of F which are separable algebraic over K. Then S K. Let a, b S.
Then by Corollary 17.1.33(ii), K(a, b)/K is separable algebraic. Since a b K(a, b) and (for b 6= 0) ab1
K(a, b), a b, and ab1 (b 6= 0) are separable algebraic over K and thus are members of S. Hence, S is a field.
Definition 17.1.35 Let F/K be an algebraic field extension of characteristic p > 0. Then the intermediate field
of F/K consisting of all elements of F which are separable algebraic over K is called the separable closure of
K in F or the maximal separable intermediate field of F/K. We denote this field by Ks .
Theorem 17.1.36 Let F/K be an algebraic field extension of characteristic p > 0. Then F/Ks is purely inseparable, where Ks is the separable closure of F/K.
Proof. If F = Ks the theorem is immediate. Suppose F Ks . Let a F, a
/ Ks . Let
e

p(x) = k0 + k1 xp + + (xp )n0

be the minimal polynomial of F/Ks , where e is the exponent of inseparability and n0 is the reduced degree of
e
p(x) over Ks . Now by Theorem 17.1.15, k0 + k1 y + + y n0 is the minimal polynomial of ap over Ks and this
pe
pe
polynomial is separable over Ks . Hence, a is separable over Ks . Thus, Ks (a )/Ks is separable algebraic and so
e
e
Ks (ap )/K is separable algebraic. By the definition of Ks , we have ap Ks . Therefore, a is purely inseparable
over KS .
We can think of field theory as being separated into two parts, namely, that in which the fields are of
characteristic 0 and that in which the fields are of prime characteristic p. It can be shown that for any field
extension F/K, there exists a subset X of F which is algebraically independent over K and which also has the
property that F/K(X) is algebraic. The above theorem shows that the study of algebraic field extensions of
characteristic p > 0 can be separated into two parts, the separable part and the purely inseparable part. Separable
algebraic field extensions of characteristic p > 0 often act entirely similar to field extensions of characteristic 0.
Purely inseparable field extensions have their own distinctive behavior.
Definition 17.1.37 Let F/K be an algebraic field extension of characteristic p > 0. Then the degree [Ks : K]
is called the degree of separability of F/K and is denoted by [F : K]s . The degree [F : Ks ] is called the degree
of inseparability of F/K and is denoted by [F : K]i .
Theorem 17.1.38 Let K be a field of characteristic p > 0 and p(x) an irreducible polynomial in K[x]. Let K(a)
be an extension of K obtained by adjoining a root a of p(x) to K. Then
[K(a) : K]s = n0 ,
[K(a) : K]i = pe ,
where n0 is the reduced degree of p(x) over K and pe is the degree of inseparability of p(x) over K.
S
i
Proof. Let b K(a). Then b = n1
i=0 ki a , where n is the degree of p(x) over K and each ki K. Therefore,
e

bp =

n1
[
i=0

pe

kip (ap )i K(ap ).

e

Thus, b is purely inseparable over K(a ). Hence, K(a)/K(ap ) is purely inseparable. By the definition of the
e
e
degree of inseparability of p(x) over K, K(ap )/K is separable algebraic. Now Ks K(ap ). Let b Ks . We
e
pe
have just seen that b is purely inseparable over K(a ). But b is also separable algebraic over K(ap ). Therefore,
pe
pe
pe
b K(a ) so that Ks = K(a ). By Theorem 17.1.15, the minimal polynomial of a over K is of degree
e
e
e
n0 and so [K(a) : K]s = [K(ap ) : K] = n0 . Thus, n0 pe = [K(a) : K] = [K(a) : K(ap )][K(ap ) : K] =
pe
e
pe
[K(a) : K(a )]n0 . Consequently, p = [K(a) : K(a )] = [K(a) : K]i .
Example 17.1.39 Let K denote the field Zp (u, v), where u and v are algebraically independent over Zp . Let a
be a root of the polynomial x2p + vxp + u over K. By use of Worked-Out Exercise 6 (page 236), one can deduce
that x2p + vxp + u is irreducible over K. Let F be the field K(a). We ask the reader to verify the following
properties of the field extension F/K. Ks = K(ap ), [F : K]i = p, and [F : K]s = 2. Also, the extension F/K has
no elements which are purely inseparable over K ( except those elements which are already in K). Thus, if J is
the intermediate field of F/K consisting of all the elements of F purely inseparable over K, then J = K. Hence,
F/J is not separable algebraic.

17.1. Multiplicity of Roots

255

Worked-Out Exercises
Exercise 1: Determine if the following polynomials are separable or inseparable over the given fields.
(a) x2 6x + 9 over Q;

(b) x4 + x2 + [1] over Z2 .

Solution:

(a) x2 6x + 9 = (x 3)2 over Q. Now x 3 is irreducible over Q. Since x 3 is separable over Q,

x2 6x + 9 is separable over Q.

(b) x4 + x2 + [1] = (x2 + x + [1])2 over Z2 . Now x2 + x + [1] has no roots in Z2 . Hence, x2 + x + [1] is
irreducible over Z2 . Now Dx (x2 + x+ [1]) = [2]x + [1] = [1] 6= [0]. Thus, x2 + x + [1] and so x4 + x2 + [1]
is separable over Z2 .
Exercise 2: Prove that the following polynomials are irreducible over Z3 (t), where t is transcendental over Z3 . Find
the exponent of inseparability and the degree of separability of the polynomials over Z3 (t).
(a) p(x) = x36 + tx18 + t.
(b) q(x) = x24 + tx18 + t.
(c) r(x) = x20 + tx18 + t.
(d) s(x) = x9 + t.
Solution:

Since t|t, t|0, t |/1, t2 |/t, the polynomials p(x), q(x), r(x), s(x) are irreducible over Z3 (t).
2

(a) p(x) = x43 + tx23 + t and so the exponent of inseparability e = 2 and the degree of separability
n0 = 4.
(b) q(x) = x83 + tx63 + t and so the exponent of inseparability e = 1 and the degree of separability
n0 = 8.
(c) Since 3 |/20, e = 0 and n0 = 20.
(d) Here e = 2 and n0 = 1.
Exercise 3: Let f (x) and g(x) be polynomials over the field K.
(a) Does f (c) = g(c) for all c K imply that f (x) = g(x)?
(b) Does f (c) = 0 for all c K imply that f (x) = 0?
Solution:

(a) Let f (x) = [3]x5 [4]x2 Z5 [x] and g(x) = x2 + [3]x Z5 [x]. Now f ([0]) = [0] = g([0]), f ([1]) =
[4] = g([1]), f ([2]) = [0] = g([2]), f ([3]) = [3] = g([3]), f ([4]) = [3] = g([4]). Hence, f (c) = g(c) for all
c Z5 . However, f (x) 6= g(x).

(b) Let f(x) = x2 + x Z2 [x]. Then f (c) = 0 for all c Z2 , but f(x) 6= 0.
2

Exercise 4: Let K = P (x, y, z) and F = K(z p , z p xp + y p ), where P is a perfect field of characteristic p > 0
1
1
and x, y, z are algebraically independent indeterminates over P. Prove that K p F = K(z p ), where
1
1
K p = {kp | k K}.
Solution:

Clearly F K p F K(z p ). Now [F : K] = p3 . Suppose that K p F K(z p ). Then

1
2
2
1
1
1
F = (K p F )(z p ) since z p
/ K p F and [K p F : K] must be p2 . Thus, [F : K p F ] = p. Since
1
1
1
1
1
2
[K p (F ) : K p ] = p, any basis of F/(K p F ) remains a basis of K p (F )/K p . Now Z = {1, z p ,
2
1
. . . , (z p )p1 } is a basis of F/(K p F ). Also,
2

zp

xp

+ yp

p1
[

2 i

ki (z p

),

i=0

where ki K p F, i = 0, 1, . . . , p 1. Since Z remains linearly independent over K p , y p = k0

1
1
1
1
1
K p F and xp = k1 K p F. Therefore, xp , y p F. Thus, [F : K] = p4 , a contradiction.
p1
p1
Hence, K
F = K(z
).

256

17. Multiplicity of Roots

Exercises
1. Let f (x) K[x], a polynomial ring over a field K and c F, where F is an extension field of K. Prove
that (x c)2 |f (x) if and only (x c)|f(x) and (x c)|f 0 (x).
2. Let f (x) K[x], a polynomial ring over a field K. Use Exercise 1 to prove that f(x) has no repeated roots
in any extension field of K if and only if f (x) and f 0 (x) are relatively prime.

3. Let f(x) = xn x K[x], a polynomial ring over a field K. Suppose that n 2 and that either K has
characteristic 0 or a prime p such that p does not divide n 1. Prove that f (x) has no repeated roots in
any extension field F of K.
4. Let f (x) = xp k K[x], a polynomial ring over a field K of characteristic p > 0. Prove that either f(x)
is irreducible over K or that f (x) is a power of a linear polynomial in K[x].
5. Determine if the following polynomials are separable or inseparable over the given field.
(i) x2 4x + 4 over Q.
(ii) x5 + tx + t over Z5 (t), where t is transcendental over Z5 .
6. Prove that the following polynomials are irreducible over Z5 (u), where u is transcendental over Z5 . Find
the exponent of inseparability and the degree of separability of the polynomials over Z5 (u).
(i) p(x) = x250 + ux125 + u.
(ii) g(x) = x128 + ux125 + u.
(iii) s(x) = x125 + u.
e

7. Let F be a field of characteristic p > 0. Prove that for any nonnegative integer e, F p is a subfield of F.
e
e
Prove also that the mapping : F F p defined by (a) = ap is an isomorphism.
8. Prove that a root of the polynomials in Examples 17.1.16 and 17.1.39 is neither purely inseparable nor
separable algebraic over K(t) and K, respectively.
9. Let K(a)/K be a field extension of characteristic p > 0. Prove that (K(a))p = K p (ap ).
10. Let F/K be a finite field extension of characteristic p > 0. If [F : K] is not divisible by p, prove that F/K
is separable.
11. Let F/K be an algebraic field extension and S be an intermediate field of F/K such that F/S is purely
inseparable and S/K is separable algebraic. Prove that S = Ks .
12. Let P be a perfect field of characteristic p > 0. Let P (a)/P be an algebraic field extension. Prove that
P (a)/P is separable and that P (a) is perfect.
13. Let K be any field of characteristic p > 0. Prove that Zp is the smallest subfield of K which is perfect and
pi
is the largest subfield of K which is perfect.

i=0 K
14. Verify the properties of the field extension F/K of Example 17.1.39.

15. Answer the following statements, true or false. If the statement is true, prove it. If it is false, give a
counterexample.
(i) Let F be a field of characteristic p > 0. Since F ' F p and F p F, it follows that F p = F.
(ii) Let F/K be a field extension of characteristic p > 0. Let c F \K. Then it is impossible for c to be
both separable and purely inseparable over K.
(iii) Let F/K be a field extension of characteristic p > 0. Let c F. Then it is impossible for c to be both
separable and inseparable over K.

Chapter 18

Finite Fields
The theory of finite fields has come to the fore in the last 60 years due to newfound applications. The applications
of finite fields are in coding theory, combinatorics, switching circuits, statistics via finite geometries, and certain
areas of computer science.

18.1

Finite Fields

A finite field (or Galois field) is a field with a finite number of elements. If F is a finite field, then F has
prime characteristic p and contains a subfield isomorphic to Zp . Since F has only a finite number of elements,
[F : Zp ] < .
We denote a finite field of n elements by GF(n). We will show in the next result that n must be a power of
p. The result is due to E.H. Moore (18621932). The United States is indebted to Moore for its beginnings in
abstract algebra and for its initial international recognition in research.
Theorem 18.1.1 If F is a finite field of characteristic p and n = [F : Zp ], then F contains pn elements.
Proof. Since [F : Zp ] = n, F/Zp has a basis of n elements, say, b1 , b2 , . . . , bn . Every element a of F is a
linear combination of b1 , b2 , . . . , bn , i.e., a = a1 b1 + a2 b2 + + an bn , where ai Zp , i = 1, 2, . . . , n. Now
Zp has p elements. Hence, F has at most pn elements. Since {b1 , b2 , . . . , bn } is linearly independent over Zp ,
a1 b1 + a2 b2 + + an bn is distinct for every choice of a1 , a2 , . . . , an . Thus, F has exactly pn elements.
Theorem 18.1.2 Every element of a finite field F of characteristic p and of pn elements is a root of the
n
n
polynomial xp x Zp [x]. Moreover, F is a splitting field of xp x over Zp .
n

Proof. First note that (F \{0}, ) is a commutative group of order pn 1. Thus, for all a F \{0}, ap 1 = 1,
n
n
n
whence ap = a. Clearly 0p = 0. Since F contains all the roots of xp x, F contains a splitting field S of
n
n
xp x over Zp . However, F is exactly the set of all the roots of xp x and so F = S.
In the following result, we once again use a positive integer and the concept of an isomorphism to completely
characterize an algebraic structure.
Corollary 18.1.3 Any two finite fields of pn elements are isomorphic, where p is a prime and n is a positive
integer.
n

Proof. If F and F 0 are finite fields with pn elements, then they are splitting fields of the polynomial xp x
over Zp . Hence, F ' F 0 .
The next theorem can be used to show that there exists an irreducible polynomial of arbitrary degree n over
Zp . (See Exercise 8, page 260.) Even though its proof is not constructive in nature, it is informative for certain
applications. Exercises 5 and 6 can be used to actually count the irreducible polynomials of a given degree.
There is an algorithm which can be used to test the irreducibility of a polynomial over a finite fieldnamely,
Berlekamps algorithm. This algorithm is discussed in Isaacs.
Theorem 18.1.4 For any prime p, there exists a field extension F/Zp of arbitrary finite degree n.

257

258

18. Finite Fields

n

Proof. Let S be the splitting field of the polynomial f (x) = xp x over Zp . Let a S be a root of f(x) of
multiplicity m. Then
f(x) = (x a)m g(x),
where a is not a root of g(x). Now
1 = f 0 (x) = (x a)m1 [mg(x) + (x a)g 0 (x)].
This implies that (x a)m1 divides 1, whence m 1 = 0. Thus, every root of f (x) in S has multiplicity 1.
Hence, f (x) has pn distinct roots in S. Let F denote the subset of S, which consists of all roots of f (x). Let a,
n
n
n
b F. Then (a b)p = ap bp = a b. Therefore, a b F. For b 6= 0,
n

(ab1 )p = ap (bp )1 = ab1

and so ab1 F . Thus, F is a subfield of S. Since F contains all the roots of f (x) and S is generated by the
roots of f (x) over Zp , F = S. By Exercise 6 (page 260), [F : Zp ] = n.
Theorem 18.1.5 Let F be a field and G be a finite subgroup of the multiplicative group F = F \{0}. Then G
is cyclic.
Proof. Since G is a finite Abelian group, G is a direct product of cyclic subgroups C1 , C2 , . . . , Ck , where
|Ci | = ni , n1 > 1, and ni |ni+1 , 1 i < k. From this it follows that g nk = 1 for all g G. Thus, every element of
G is a root of xnk 1 F [x]. Since xnk 1 has at most nk distinct roots in F, |G| nk . Now Ck is a subgroup
of G and |Ck | = nk . Hence, G = Ck and so G is cyclic.
The following corollary is an immediate consequence of Theorem 18.1.5.
Corollary 18.1.6 The multiplicative group of a finite field is cyclic.
Theorem 18.1.7 Let F be a finite field and F (a, b)/F a field extension with a, b algebraic over F. Then there
exists c F (a, b) such that F (a, b) = F (c), i.e., F (a, b) is a simple extension.
Proof. Since F (a, b)/F is algebraic, [F (a, b) : F ] < . Thus, F (a, b) is a finite field since F is a finite field.
Since F (a, b)\{0} is a cyclic group with some generator, say, c by Theorem 18.1.5, it follows that F (a, b) = F (c).

Corollary 18.1.8 Every finite extension of a finite field is simple.

Worked-Out Exercises
Exercise 1: Prove that x3 + x + [1] is irreducible in Z2 [x]. Write out the addition and multiplication tables for the field


Z2 [x]/ x3 + x + [1] .

Find a splitting field S1 for x3 + x + [1] over Z2 . Find a basis for S1 /Z2 and [S1 : Z2 ].
Solution:

x3 + x + [1] is irreducible over Z2 if and only if Z2 contains no root of x3 + x + [1]. Since [0]3 + [0] + [1] 6= [0]
and [1]3 + [1] + [1] 6= [0] in Z2 , Z2 contains no roots of x3 + x + [1] over Z2 . Hence, x3 + x + [1] is irreducible
over Z2 . By Theorem 16.1.11,


Z2 [x]/ x3 + x + [1] = Z2 (),


where denotes the coset x + x3 + x + [1] . By Theorem 16.1.14,
Z2 () = {[0], [1], , 2 , [1] + , [1] + 2 , + 2 , [1] + + 2 }.

Now
x3 + x + [1] = (x + )(x2 + x + [1] + 2 )
and 2 and + 2 are the roots of x2 + x + [1] + 2 . Since 2 , + 2 Z2 (), Z2 () is a splitting field of
x3 + x + [1] over Z2 . Let S1 = Z2 (). Then {[1], , 2 } is a basis for S1 /Z2 and [S1 : Z2 ] = 3. Let denote

18.1. Finite Fields

259

[1] + + 2 . The addition table for Z2 () is given below.

+
[0]
[1]

2
[1]+
[1]+2
+2

[0]
[0]
[1]

2
[1]+
[1]+2
+2

[1]
[1]
[0]
[1]+
[1]+2

+2

[1]+
[0]
+2
[1]

2
[1]+2

2
2
[1]+2
+2
[0]

[1]

[1]+

[1]+
[1]+

[1]

[0]
+2
[1]+2
2

[1]+2
[1]+2
2

[1]
+2
[0]
[1]+

+2
+2

[1]+2
[1]+
[0]
[1]

+2
[1]+2
[1]+
2

[1]
[0]

For the multiplication table, we make a few entries, such as ([1] + )([1] + ) = [1] + 2 and ([1] + +
2 )([1] +2 ) = [1] + + 3 +4 . We now reduce [1] + + 3 +4 to the form a + b + c2 , where a, b, c Z2 .
We divide x4 + x3 + x + [1] by x3 + x + [1] to obtain x4 + x3 + x + [1] = (x + [1])(x3 + x + [1]) + x2 + x. Thus,
4 + 3 + + [1] = ( + [1])(3 + + [1]) + 2 + = [0] + 2 + . Hence, ([1] + + 2 )([1] + 2 ) = + 2 .
Exercise 2: Prove that x3 + x2 + [1] is irreducible in Z2 [x]. Write out the addition and multiplication tables for the
field


Z2 [x]/ x3 + x2 + [1] .
Find a splitting field S2 for x3 + x + [1] over Z2 . Find a basis for S2 /Z2 and [S2 : Z2 ]. Compare your results
with those in Worked-Out Exercise 1.

Solution:

Since [0]3 + [0]2 + [1] 6= [0] and [1]3 + [1]2 + [1] 6= [0] in Z2 , Z2 contains no roots of x3 + x2 + [1] over Z2 .
Hence, x3 + x2 + [1] is irreducible over Z2 . By Theorem 16.1.11,


Z2 [x]/ x3 + x2 + [1] = Z2 (),


where denotes the coset x + x3 + x2 + [1] . By Theorem 16.1.14,
Z2 () = {[0], [1], , 2 , [1] + , [1] + 2 , + 2 , [1] + + 2 }.

Now x3 +x2 +[1] = (x+)(x2 +([1]+)x++2 ) and 2 and [1]++2 are the roots of x2 +([1]+)x++2 .
Since 2 , [1] + + 2 Z2 (), Z2 () is a splitting field of x3 + x2 + [1] over Z2 . Let S2 = Z2 (). Then
{[1], , 2 } is a basis for S2 /Z2 and [S2 : Z2 ] = 3. The addition table for Z2 () is determined in a manner
similar to that in Exercise 1. In fact, one may obtain the addition table by substituting for in the
addition table of Z2 (). We now consider multiplication. We note that ([1]+)([1]+) = [1]+2 . However,
([1] + + 2 )([1] + 2 ) = [1] + + 3 +4 = [1]. Hence, we note the first algebraic dierence between Z2 ()
and Z2 ().
Exercise 3: Show that there exists an isomorphism f of Z2 () onto Z2 () considered as vector spaces over Z2 such that
f is the identity on Z2 and f () = , f (2 ) = 2 , where and are as defined in Worked-Out Exercises
1 and 2, respectively.
Solution:

{[1], , 2 } is a basis for Z2 () over Z2 and {[1], , 2 } is a basis for Z2 () over Z2 . Hence, there exists a
unique linear transformation f of Z2 () onto Z2 () such that f ([1]) = [1], f() = , and f (2 ) = 2 . This
linear transformation is given by
f (a[1] + b + c2 ) = a[1] + b + c2 ,
where a, b, c Z2 . Since {[1], , 2 } is linearly independent, f is one-one.

Exercise 4: Show that Z2 () and Z2 () are isomorphic as fields, where and are as defined in Worked-Out Exercises
1 and 2, respectively.
Solution:

Since |Z2 ()| = 23 = |Z2 ()| , Z2 () and Z2 () are splitting fields of x8 x over Z2 and thus are isomorphic.

Exercise 5: Factor the polynomial x8 x over Z2 .

Solution:

x8 x = x(x + [1])(x6 + x5 + x4 + x3 + x2 + x + [1]). Now x2 + x + [1] is the only irreducible quadratic

polynomial over Z2 . But x2 +x+[1] does not divide x6 +x5 +x4 +x3 +x2 +x+[1]. We have that x3 +x+[1] and
x3 +x2 +[1] are irreducible polynomials over Z2 and x6 +x5 +x4 +x3 +x2 +x+[1] = (x3 +x+[1])(x3 +x2 +[1]).
Hence, x8 x = x(x + [1])(x3 + x + [1])(x3 + x2 + [1]).

Exercise 6: Find the roots of x3 + x2 + [1] in Z2 (), where is as defined in Worked-Out Exercise 1.
Solution:

[0] is a root of x, [1] is a root of x + [1], and , 2 , + 2 are roots of x3 + x + [1]. Hence, [1] + , [1] + 2 ,
and [1] + + 2 are roots of x3 + x2 + [1].

260

18. Finite Fields

Exercise 7: Find the roots of x3 + x + [1] in Z2 (), where is as defined in Worked-Out Exercise 2.
Solution: [0] is a root of x, [1] is a root of x + [1], and , 2 , [1] + + 2 are roots of x3 + x2 + [1]. Hence,
[1] + , [1] + 2 , and + 2 are roots of x3 + x + [1].
Exercise 8: Show that there exists an isomorphism g of Z2 () onto Z2 ([1] + ) such that g() = [1] + , where and
are as defined in Worked-Out Exercises 1 and 2, respectively.
Solution:

The result here follows immediately by Corollary 16.2.9.

Exercise 9: Show that there does not exist an isomorphism h of Z2 () onto Z2 () such that h() = , where and
are as defined in Worked-Out Exercises 1 and 2, respectively.
Solution:

Suppose there exists an isomorphism h of Z2 () onto Z2 () such that h() = . Then [0] = h([0]) =
h(3 + + [1]) = 3 + + [1]. Also, [0] = 3 + 2 + [1]. Hence, 3 + + [1] = 3 + 2 + [1]. Thus, 2 = .
Therefore, = [1], a contradiction.

Exercises
1. Let F be a finite field. A generator for F = F \{0} is called a primitive element for F. Find a primitive
element for the following fields.
(i) Z7 .
(ii) Z11 .
(iii) F, where F Z2 and [F : Z2 ] = 8.
2. Construct a field with 9 elements.

3. Construct a field with 27 elements.

4. Suppose that F is a finite field of characteristic p. If c is a primitive element of F, prove that cp is a
primitive element of F.
5. Let F be a finite field of characteristic p. If n = [F : Zp ], prove that there exists c F such that c is
algebraic of degree n over Zp and F = Zp (c).
6. If F is a finite field of pn elements, p a prime and n a positive integer, prove that [F : Zp ] = n.
2

7. Describe the splitting field of x3 x over Z3 .

8. Prove that there exists an irreducible polynomial of arbitrary degree n over Zp .

9. If F is a subfield of GF (pn ), prove that F ' GF (pm ), where m|n.

10. Show that if m and n are positive integers such that m|n, then GF (pn ) contains a unique subfield GF (pm ),
m
n
m
n
pm 1 divides pn 1, whence xp 1 1 divides xp 1 1 and so xp x divides xp x.

11. Let F be a field containing Zp and f (x) be a polynomial over Zp . If c F is a root of f (x), prove that cp
is also root of f (x).

12. Let f (x) = xp x [1] Zp [x]. Show that a splitting field of f (x) over Zp is Zp (c), where c is a root of
f (x).
13. Let F be a field and G and H be subgroups of F . If G and H have order n, prove that G = H.
14. If F is a field such that F is cyclic, prove that F is finite.

References
1. Aschbacher, M. The classification of finite simple groups. Mathematics Intelligencer, 3(2), 5965, 1981.
2. Barnes, W. E. Introduction to Abstract Algebra. Boston: D.C. Heath and Company, 1963.93.
3. Bell, E. T. Men of Mathematics. 2d ed. New York: Simon and Schuster, 1962.
4. Burton, D. M. Elementary Number Theory. Boston: Allyn & Bacon, 1980.
5. Edwards, H. M. The genesis of ideal theory. Arch. History Exact Sci. 23, 321378, 1980.
6. Edwards, H. M. Dedekinds invention of ideals. In Studies in the History of Mathematics, E.R. Phillips,
ed. The Mathematical Association of America, 1987.
7. Gillispie, C. C., ed. Dictionary of Scientific Biography, Vols. 1-14. New York: Charles Scribners Sons.
8. Halmos, P. R. Naive Set Theory. New York: Springer Verlag, 1974.
9. Hardy, G. H., and Wright, E. M. An Introduction to the Theory of Numbers, 4th ed. Oxford, England.
Clarendon Press, 1960.
10. Herstein, I. N. Topics in Algebra. 2d ed. New York: Wiley, 1975.
11. Hungerford, T. W. Algebra. New York: Holt, Reinhart and Winston, 1974.
12. Kleiner, I. The evolution of group theory: A brief survey. Mathematics Magazine 59(4), 195215, 1986.
13. Kleiner, I. A sketch of the evolution of (noncommutative) ring theory. LEnseignement Mathmatique, 33,
227267, 1987.
14. Malik, D.S., Mordeson, J.N., and Sen, M.K., Fundamentals of Abstract Algebra, McGraw Hill, 1997.
15. McCoy, N. H. The Theory of Rings. New York, Chelsea Publishing Company, 1973.
16. Rotman, J. J. An Introduction to the Theory of Groups. Iowa, Wm. C. Brown, 1988.
17. Rotman, J. J. Galois Theory. New York: Springer Verlag, 1990.
18. Van der Waerden, B. L. A History of Algebra. New York: Springer Verlag, 1985.
19. Zariski, O., and Samuel, P. Commutative Algebra, Vol. 1. New Jersey: D. Van Nostrand Co. Inc., 1960.

261

INDEX
degree of a polynomial, 178
degree of inseparability, 249, 254
degree of separability, 249, 254
DeMorgans law, 6, 7
dihedral group, 113
dimension, 225
direct sum of rings, 171
direct summand, 227
disjoint permutations, 63
divide, 10, 189
division algorithm, 179
division ring, 134
divisor, 10, 189
common, 10, 190
greatest common, 10, 190
domain
Euclidean, 185
factorization, 199
integral, 135
principal ideal, 186
unique factorization, 200
double coset, 88

Abel, Niels Henrik, 58

Abelian group, 37
action of groups, 116
algebraic closure, 244
algebraic element, 230
algebraic field extension, 234
algebraically closed field, 243
algebraically independent, 181
alternating group, An , 67
ascending chain condition for principal ideals, 200
associate, 190
associated prime ideal, 216
automorphism
of groups, 100
of rings, 158
basis, 221
binary operation, 32
associative, 32
closed under, 32
commutative, 32
Boolean ring, 140
Burnside theorem, 119

Eisensteins irreducibility criterion, 208

element
algebraic, 230
associate, 190
centralizer of an, 77
conjugate, 241
fixed, 119
idempotent, 138
identity, 33, 36, 131
image of, 24
inseparable, 247
integral power of an, 45
inverse of an, 36
invertible, 133
irreducible, 194
order of an, 46
preimage of, 24
prime, 194
primitive, 250
purely inseparable, 251
reducible, 194
regular, 141
relatively prime, 194
separable, 247
separable algebraic, 247
transcendental, 230

Cartesian product, 5, 28
Cauchy, Augustin-Louis, 70
Cayleys theorem, 102
Cayley, Arthur, 121
center
of groups, 73
of rings, 130
characteristic
of a ring, 137
subgroup, 112
Chinese remainder theorem, 23
Chinese remainder theorem for rings, 175
commutative
group, 37
ring, 130
complete direct sum of rings, 171
composition of functions, 25
congruence modulo n, 18
conjugate element, 241
content of a polynomial, 203
coset
left, 81
right, 81
cyclic module, 221
cyclic structure, 69

262

INDEX
unit, 133
epimorphism
of groups, 98
of rings, 158
equivalence
class, 18
relation, 17
Euclidean domain, 185
Euclidean valuation, 185
Euler -function, 14
even permutation, 67
exponent of inseparability, 249
factor, 10
nontrival, 199
trivial, 199
factorization, 199
factorization domain, 199
factorization theorem, 180
field, 134
algebraic closure of, 244
algebraically closed, 243
extension, 230
finite, 257
Galois, 257
intermediate, 233
maximal separable intermediate, 254
of complex numbers, 134
of quotients, 167
of rational numbers, 134
of real numbers, 134
perfect, 249
prime, 229
primitive element for, 260
quotient, 167
separable closure of, 254
simple extension, 250
splitting, 239
field extension, 230
algebraic, 234
degree of, 232
finite, 232
purely inseparable, 251
transcendental, 234
finite dimensional vector space, 224
finite extension, 232
finite field, 257
finitely generated left ideal, 151
finitely generated module, 221
fixed element, 119
formal derivative, 247
function, 24
composition of, 25
extension of, 28
invertible, 27
left invertible, 27
one-one, 25
onto, 25
restriction of, 28

263
right invertible, 27
single valued, 24
well defined, 24
fundamental theorem
of arithmetic, 12
G-set, 116
Galois
field, 257
Gaussian integers, 185
gcd property in a ring, 194
generator, 230
group
special linear group of degree 2, 55
group, groups, 36
Abelian, 37
action of , 116
alternating group, An , 67
automorphism of, 100
cancellation law, 40
Cayleys theorem for, 102
center of, 73
commutative, 37
correspondence theorem of, 108
cyclic, 78
dihedral, 75, 113
epimorphism of, 98
external direct product, 124
finite, 45
first isomorphism theorem, 106
fundamental theorem of homomorphisms, 105
homomorphic image of, 98, 106
homomorphism of, 97
identity element of, 36
infinite, 46
inner automorphism of, 110
internal direct product, 124
inverse of an element in, 36
isomorphic, 100
isomorphism of, 100
isotropy, 117
Klein 4-group, 78
Lagranges theorem, 84
monomorphism of, 98
natural homomorphism of, 99
noncommutative, 37
of symmetries of square, 49
orbit, 117
order, 45
permutation, 59
quaternion, 114
quotient, 90
second isomorphism theorem, 107
set of generators for, 73
symmetric group, Sn , 62
third isomorphism theorem, 108
torsion, 47
torsion-free, 47
trivial subgroup of, 72

264
homomorphic image, 98
homomorphism
fundamental theorem of, 105, 161
kernel of, 98, 159
natural, 99, 160
of groups, 97
of rings, 158
trivial, 97
ideal, 149
annihilator of, 158
associated prime, 216
direct sum of, 172
finitely generated, 151
generated by, 150
internal direct sum of, 173
left, 149
maximal, 214
minimal, 220
nil, 155
nilpotent, 155
nontrivial, 149
primary, 215
primary for, 216
prime, 213
principal, 151
product of, 153
proper, 149
radical of, 215
right, 149
semiprime, 219
sum of, 153
trivial, 149
idempotent, 138
central, 174
identity
left, 139
of a ring, 131
of group, 36
right, 139
identity map, 25
indeterminate, 178
infinite order of an element, 46
inner automorphism, 110
inseparable element, 247
inseparable polynomial, 247
integer, integers
algebraic, 202
division algorithm, 9
prime, 12
relatively prime, 12
integer, integers, 7
integral domain, 135
intermediate field, 233
irreducible element, 194
isomorphism
of groups, 100
of rings, 158

INDEX
K-automorphism, 234
k-cycle, 62
K-homomorphism, 234
K-isomorphism, 234
kernel, 159
kernel of a homomorphism, 98
Kronecker, Leopold, 211
Lagrange, Joseph Louis, 95
least common multiple, 16, 194
left cancellation law, 136
left ideal, 149
generated by, 150
principal, 151
linearly dependent, 223
linearly independent, 223
local ring, 219
mapping, 24
mathematical system, 32
maximal ideal, 214
meaningful product, 42
minimal ideal, 220
minimal polynimial, 231
module
cyclic, 221
finitely generated, 221
left, 221
right, 221
simple, 227
unital left, 221
monic polynomial, 178
monomorphism
of groups, 98
of rings, 158
multiplicity, 247
nil ideal, 155
nilpotent ideal, 155
noncommutative
group, 37
ring, 130
normal subgroup, 89
odd permutation, 67
one-one correspondence, 26
one-one function, 25
onto function, 25
orbits of a group, 117
order
of a group, 45
of an element, 46
ordered n-tuples, 28
ordered pair, 5
partition
of a set, 19
perfect field, 249
permutation, 59
conjugate, 63

INDEX
cyclic structure of, 69
disjoint, 63
even, 67
odd, 67
two-row notation of, 60
polynomial, 177
coecients of, 178
constant, 178
content of, 203
cyclotomic, 209
degree of, 178
factor of, 180
formal derivative of, 247
inseparable, 247
leading coecient of, 178
minimal, 231
monic, 178
primitive, 204
root of, 179
separable, 247
split over a field, 239
zero of, 179
polynomial ring, 177
polynomial ring in n indeterminates, 180
power set, 5
primary ideal, 215
primary ideal belonging to, 216
prime element, 194
prime field, 229
prime ideal, 213
primitive element, 250
primitive polynomial, 204
principal ideal domain, 186
principle of mathematical induction, 8
principle of well-ordering, 7
projection, 174
quaternion group, 114
quotient, 10, 179
group, 90
ring, 154
set, 30
radical, 215
reduced degree, 249
reducible element, 194
regular ring, 141
relation, 17
binary, 17
composition of, 20
congruence, 94
domain, 17
equivalence, 17
image, 17
inverse, 20
range, 17
reflexive, 17
symmetric, 17
transitive, 17

265
transitive closure of, 23
relatively prime elements, 194
remainder, 10, 179
remainder theorem, 180
right cancellation law, 136
right ideal, 149
generated by, 150
principal, 151
ring, 130
automorphism of, 158
Boolean, 140
center of, 130
characteristic of, 137
commutative, 130
complete direct sum of, 171
correspondence theorem, 161
direct sum, 140
direct sum of, 171
division, 134
embedding of, 165
epimorphism of, 158
finite, 136
first isomorphism theorem of, 161
fundamental theorem of homomorphisms, 161
identity element, 131
infinite, 136
isomorphic, 159
isomorphism of, 158
left identity of, 139
local, 219
monomorphism of, 158
natural homomorphism of, 160
nilpotent element in, 138
noncommutative, 130
of Gaussian integers, 186
of integers, 130
of integers mod n, 130
of real quaternions, 134
polynomial, 177
principal ideal, 186
quotient, 154
regular, 141
regular element of, 141
right identity of, 139
simple, 152
subdirect sum of, 174
with identity, 131
zero, 136
zero divisor in, 134
zero element of, 130
root of a polynomial, 179
scalar, 178, 222
Schrder-Bernstein, 31
semigroup, 43
commutative, 43
idempotent element in, 43
noncommutative, 43
semiprime ideal, 219

266
separable element, 247
separable polynomial, 247
set, sets, 3
Cartesian cross product, 5, 28
complement, 6
dierence of, 5
disjoint, 4
empty, 3
equal, 3
equipollent, 26
finite, 3
image of, 29
index, 4
infinite, 3
intersection of, 4
null, 3
partition of, 19
power, 5
proper subset of, 3
relative complement of, 5
subset of, 3
symmetric dierence of, 7
union of, 4
simple extension, 250
simple ring, 152
skew-field, 134
span, 222
splitting field, 239
stabilizer, 117
standard product, 42
subfield, 145
generated by, 230
subgroup, 71
characteristic, 112
double coset of, 88
generated by, 73
index of a, 83
invariant, 89
left coset of a, 81
normal, 89
product of, 74
right coset of a, 81
submodule, 221
generated by, 221
subring, 145
subspace, 222
trivial, 223
symmetric group, Sn , 62
torsion group, 47
torsion-free group, 47
transcendental element, 230
transcendental field extension, 234
transposition, 62
unique factorization domain, 200
vector, 222
vector space, 222

INDEX
dimension of, 225
finite dimensional, 224
left, 222
Weber, Heinrich, 128
zero divisor, 134
zero of a polynomial, 179
zero ring, 136