MA136

Introduction to Abstract Algebra

Samir Siksek
Mathematics Institute
University of Warwick
 Contents

Chapter I. Prologue 1
I.1. Who Am I? 1
I.2. A Jolly Good Read! 1
I.3. Proofs 2
I.4. Acknowledgements and Corrections 2

Chapter II. Algebraic Reorientation 3

II.1. Sets 3
II.2. Binary Operations 4
II.3. Vector Operations 5
II.4. Operations on Polynomials 5
II.5. Composition of Functions 6
II.6. Composition Tables 7
II.7. Commutativity and Associativity 7
II.8. Where are the Proofs? 9
II.9. The Quaternionic Number System (do not read) 10

Chapter III. Matrices—Read On Your Own 13

III.1. What are Matrices? 13
III.2. Matrix Operations 14
III.3. Where do matrices come from? 16
III.4. How to think about matrices? 17
III.5. Why Column Vectors? 19
III.6. Multiplicative Identity and Multiplicative Inverse 20
III.7. Rotations 26

Chapter IV. Groups 27

IV.1. The Definition of a Group 27
IV.2. First Examples (and Non-Examples) 27
IV.3. Abelian Groups 29
IV.4. Symmetries of a Square 30

Chapter V. First Theorems 35

V.1. Getting Relaxed about Notation 36
V.2. Additive Notation 38

Chapter VI. More Examples of Groups 39

VI.1. Matrix Groups I 39
VI.2. Congruence Classes 40
i
ii CONTENTS

Chapter VII. Orders and Lagrange’s Theorem 43

VII.1. The Order of an Element 43
VII.2. Lagrange’s Theorem—Version 1 46

Chapter VIII. Subgroups 47

VIII.1. What Were They Again? 47
VIII.2. Criterion for a Subgroup 47
VIII.3. Roots of Unity 55
VIII.4. Matrix Groups II 56
VIII.5. Differential Equations 57
VIII.6. Non-Trivial and Proper Subgroups 58
VIII.7. Lagrange’s Theorem—Version 2 59

Chapter IX. Cyclic Groups and Cyclic Subgroups 61

IX.1. Lagrange Revisited 64
IX.2. Subgroups of Z 65

Chapter X. Isomorphisms 67

Chapter XI. Cosets 69

XI.1. Geometric Examples 70
XI.2. Solving Equations 72
XI.3. Index 74
XI.4. The First Innermost Secret of Cosets 74
XI.5. The Second Innermost Secret of Cosets 75
XI.6. Lagrange Super-Strength 76

Chapter XII. Quotient Groups 79

XII.1. Congruences Modulo Subgroups 79
XII.2. Congruence Classes and Cosets 81
XII.3. R/Z 82
XII.4. R2 /Z2 83
XII.5. R/Q 84
XII.6. Well-Defined and Proofs 84

Chapter XIII. Symmetric Groups 87

XIII.1. Motivation 87
XIII.2. Injections, Surjections and Bijections 88
XIII.3. The Symmetric Group 91
XIII.4. S n 91
XIII.5. A Nice Application of Lagrange’s Theorem 94
XIII.6. Cycle Notation 95
XIII.7. Permutations and Transpositions 99
XIII.8. Even and Odd Permutations 100

Chapter XIV. Rings 107

XIV.1. Definition 107
XIV.2. Examples 108
 CONTENTS iii

XIV.3. Subrings 110

XIV.4. The Unit Group of a Ring 112
XIV.5. The Unit Group of the Gaussian Integers 115
Chapter XV. Fields 119
Chapter XVI. Congruences Revisited 121
XVI.1. Units in Z/mZ 121
XVI.2. Fermat’s Little Theorem 122
XVI.3. Euler’s Theorem 123
XVI.4. Vale Dicere 124
 CHAPTER I

Prologue

I.1. Who Am I?
I Samir Siksek have the immense pleasure of introducing you to three
heroes of abstract algebra: groups, rings and fields. I am not an alge-
braist, but I have nothing but love, admiration and enthusiasm for the
subject. Some of my best friends are algebraists.

I.2. A Jolly Good Read!

Abstract algebra is about patterns. You see one pattern repeating itself
across mathematics and you try to extract the essential elements of that
pattern and turn them into a definition. This process gives you groups,
rings, fields, vector spaces, etc. You then study each of these new alge-
braic objects and become familiar with it. After that, when you spot one
of these patterns in a new context, you’ll say ‘Aha! I know what that is,
and what to do with it’.

three tips Abstract algebra is incredibly useful, but to get any benefit from it you
need to develop three essential habits:
(i) Study as many different examples as you can. The examples are
as important as the theorems and definitions. There is abso-
lutely no use in knowing the definition of a group if you’re not
familiar with the standard examples.
(2) Do calculations. Use calculations with matrices, permutations,
symmetries, etc. to test your ideas. Calculations will lead you to
counterexamples that can correct any erroneous ideas that you
have. But also with practice, you will find that calculations often
contain the germ of the proof you’re looking for.
(c) Think geometrically and draw pictures. The true meaning of
most mathematical concepts is geometric. If you spend all your
time manipulating symbols (i.e. doing algebra) without under-
standing the relation to the geometric meaning, then you will
have very little in terms of mathematical insight.
The three habits will not only help you learn the subject and apply it, you
will develop great mathematical taste.
1
2 I. PROLOGUE

I.3. Proofs
When I was a student I found it very hard to follow proofs in books and
lectures. So when I read a theorem, I would put down the book and try out
a few examples. After that I would try to prove the theorem myself. After I
finished (or if I failed) I would look at the proof in the book and compare.
I heartily recommend this strategy. You’ll gain a great understanding of
the subject. You’ll also get really good practice for the exam, where you
may asked to prove statements that you haven’t seen before.

I.4. Acknowledgements and Corrections

I thank Jonathan Addison, Alex Best, George Christofi, Jenny Coo-
ley, John Cremona, Edward Day, Harry Graham, Darij Grinberg, Christian
Fieldhouse, Giles Hutchings, Roderick Mansel, Dave McCormick, Joseph
Miller, Xiao Lin, Ghaleo Tsoi Kwok-Wing, Joe O’Sullivan, James Soffe and
Esther Turner for suggesting corrections to previous versions of these
notes.
Please email me your comments, misprints and corrections. My ad-
dress is [email protected].
 CHAPTER II

Algebraic Reorientation

II.1. Sets
Sets are a basic notation for most of modern pure mathematics, but
life is too short to spend too much time on them. A set is simply a col-
lection of objects. We use curly brackets to denote sets. For example, if I
write
A = {2, 5, 13},
then I’m saying that the set A consists of the elements 2, 5, 13. This is
one way of specifying a set; we simply list all its elements between curly
brackets. The notation x ∈ S means x is a member of the set S and the
notation x ∉ S means x is not a member of the set S. For the set A above,
we know 13 ∈ A but 11 ∉ A.
We can also specify some infinite sets in this fashion; for example, the
set of all integers
Z = {. . . , −3, −2, −1, 0, 1, 2, 3, . . . }.
This is absolutely standard notation: when you see Z, you’re expected to
know that it’s the set of integers. The set of natural numbers is
N = {0, 1, 2, 3, 4 . . . }.
Again this is standard notation (but not all mathematicians include 0 in
the natural numbers).
Here is an example of another way of specifying a set:
B = {x ∈ Z : x 2 = 16}.
This is saying that B is the set of all integers x satisfying the equation
x 2 = 16. Of course, another way of specifying the same set would be to
write B = {−4, 4}. If we write
C = {x ∈ N : x 2 = 16},
then C = {4}.
If we write
D = {u ∈ Z : u 3 = 2},
then D is the set of integers u satisfying u 3 = 2. There are no integers
satisfying this equation, so D is the empty set. We denote the empty set
by ;, so we can write D = ;. Here are a couple more examples of empty
sets:
{w ∈ N : w ≤ −1} = ;, {v ∈ Z : 3.01 ≤ v ≤ 3.99} = ;.
3
4 II. ALGEBRAIC REORIENTATION

To get more practice with this notation, observe that another way of
specifying the natural numbers is to write
N = {x ∈ Z : x ≥ 0}.
Yet another correct—although admittedly stupid way—is to write
N = {x ∈ Z : x ≥ −0.5}.
Here are some other sets that you’re meant to know:
(1) Q is the set of rational numbers. We can write this as
na o
Q= : a, b ∈ Z, b 6= 0 .
b
Examples of elements of Q are 0, 5, −7/11, 3/2, 6/4 (the last two
being pthe same element). From Foundations you should know
that
p 2 is irrational. You can write this statement in set notation:
2 ∉ Q. Other examples of irrational numbers are e and π.
(2) R is the set of real numbers. It isn’t possible to write R in straight-
forward way as for the sets above, but you can think of the ele-
ments of R as pointsp on the real line. Examples of elements of R
are −7, 3/5, 3.85, 7, (π + 1)/2, sin 5.
(3) C is the set of complex numbers. You have seen complex num-
bers in your Further Mathematics A-Level. Recall that i is a sym-
bol that satisfies i 2 = −1. We can write the set of complex num-
bers as
C = { a + bi : a, b ∈ R }.

While we’re on the subject of notation, compare the following two state- Tip
ments:
• Some positive real numbers are irrational.
• ∃x ∈ R s.t. x > 0 ∧ x ∉ Q.
The two statements say exactly the same thing. A professional mathe-
matician prefers the first, and an amateur prefers the second. Use only as
much mathematical notation as needed to make your ideas transparent
and precise, but no more 1.

II.2. Binary Operations

Let S be a set. A binary operation on S is a rule which for every two
elements of S gives another element of S. For example, addition is a bi-
nary operation on R, because given any two real numbers, their sum is
a real number. One way mathematicians like to say this is, “R is closed
under addition”. All that means is that the sum of two real numbers is a
real number.
Addition is also a binary operation on C, Q, Z and N. Likewise, multi-
plication is a binary operation on N, Z, Q, R, C.
1Coming soon to a supervision area near you: B, C, and F .
 II.4. OPERATIONS ON POLYNOMIALS 5

Is subtraction a binary operation? This question does not make sense

because we haven’t specified the set. Subtraction is a binary operation
on Z, Q, R, C. Subtraction is not a binary operation on N; for example 1,
2 ∈ N but 1 − 2 = −1 ∉ N. Thus N is not closed under subtraction.
Is division a binary operation on R? No, because 1, 0 are real numbers
but 1/0 is not defined. Thus R is not closed under division. Let us define
R∗ to be the set of non-zero real numbers:

R∗ = { x ∈ R : x 6= 0 }.

Now division is a binary operation on R∗ . But notice that addition is no

longer a binary operation on R∗ ; for example 5, −5 ∈ R∗ but 5 + (−5) = 0 ∉
R∗ .

II.3. Vector Operations

We define Euclidean n-space as

Rn = {(x 1 , x 2 , . . . , x n ) : x 1 , x 2 , . . . , x n ∈ R}.

Thus R2 is the set of vectors in the plane, and R3 is the set of vectors
in 3-space. Addition is a binary operation on Rn , and so is subtraction.
What about multiplication by a scalar? If λ is a scalar (i.e. λ ∈ R) and
x = (x 1 , x 2 , . . . , x n ) ∈ Rn is a vector, we define

λx = (λx 1 , λx 2 , . . . , λx n ).

Notice that the result is in Rn , but still multiplication by a scalar is not a

binary operation on Rn , because we’re not ‘combining’ two elements of
Rn , but one element of R which is λ, and one element of Rn which is x.
What about the dot product and the cross product? The dot product
is defined on Rn for all n. If x = (x 1 , . . . , x n ) and y = (y 1 , . . . , y n ) we define
their dot product to be

x · y = x1 y 1 + x2 y 2 + · · · + xn y n .

Notice that the result is in R, not Rn , so the dot product is not a binary
operation. The cross product is defined on R3 only. If x, y ∈ R3 the x × y is
again in R3 . So the cross product is a binary operation on R3 .

II.4. Operations on Polynomials

We shall write R[x] for the set of polynomials in x with real coeffi-
cients, C[x] for the set of polynomials in x with complex coefficients, Q[x]
for the set of polynomials in x with rational coefficients, and Z[x] for the
set of polynomials in x with integer coefficients. All these are closed un-
der addition, multiplication and subtraction, but not division; for exam-
ple x/(x + 1) is not a polynomial.
6 II. ALGEBRAIC REORIENTATION

II.5. Composition of Functions

Let S 1 , S 2 and S 3 be sets and f , g be functions

f : S1 → S2, g : S2 → S3.

We can define the composition g ◦ f : S 1 → S 3 by the rule: (g ◦ f )(x) =

g ( f (x)). I.e. g ◦ f is the function obtained by substituting f into g .
Example II.1. Let
f : R → R, f (x) = x 2 − 5

and
g : R → R, g (x) = 3x + 2.

Then

( f ◦ g )(x) = f (g (x)) = f (3x + 2) = (3x + 2)2 − 5 = 9x 2 + 12x − 1,

(g ◦ f )(x) = g ( f (x)) = g (x 2 − 5) = 3(x 2 − 5) + 2 = 3x 2 − 13.

The order matters here: f ◦g is the result of substituting g into f , and g ◦ f

is the result of substituting f into g . ♦
Note that in the example we started with functions R → R and com-
posed to obtain functions R → R. Likewise, in the above definition, if
S 1 = S 2 = S 3 = S say, so that f and g are functions S → S then g ◦ f is a
function S → S. In this case (i.e. when the domains and codomains are
equal) ◦ is a binary operation. It is not a binary operation on S, because it
doesn’t take two elements of S and give us another element. It is a binary
operation on the set of functions from S to itself.
The following lemma might look silly and useless, but it one of the
most important results we shall meet in this module, and we shall use it
again and again.
Lemma II.2. Let S 1 , S 2 , S 3 , S 4 be sets and let f , g , h be functions

h : S1 → S2, g : S2 → S3, f : S3 → S4.

Then f ◦ (g ◦ h) = ( f ◦ g ) ◦ h.

P ROOF. To stop ourself from getting muddled, let k = g ◦ h and ` = f ◦ g .

Note that k(x) = g (h(x)) and `(x) = f (g (x)). So

( f ◦ (g ◦ h))(x) = ( f ◦ k)(x) = f (k(x)) = f (g (h(x))).

Also
(( f ◦ g ) ◦ h)(x) = (` ◦ h)(x) = `(h(x)) = f (g (h(x))).

So f ◦ (g ◦ h) = ( f ◦ g ) ◦ h. 
 II.7. COMMUTATIVITY AND ASSOCIATIVITY 7

II.6. Composition Tables

Recall our definition of a binary operation on a set S: it is simply a
rule which for any pair of elements of S produces a third element. This
binary operation does not have to be ‘natural’, whatever that means. It
does not have to be something we met before, like addition, multiplica-
tion, composition of functions, etc. We can simply invent a set S and
binary operation on it. If the S is finite, this is easy by means of a com-
position table which tells us for any pair of elements of S what the third
element is.
Example II.3. Let S = {a, b, c}. Let ◦ be the binary operation on S with the
following composition table:
◦ a b c
a b c a
b a c a
c b b c
The result of the composition a ◦ b, is found at the intersection of the
row headed by a with the column headed by b. In other words, for com-
position tables, the first element determines the row and the second de-
termines the column. Thus for the composition table above,
a ◦ b = c, b ◦ a = a, c ◦ b = b, a ◦ a = b, . . . .
You might think that this example is somewhat contrived, and you’re ab-
solutely right. But later on we’ll meet more natural composition tables
that arise from studying groups, permutations, etc. ♦

II.7. Commutativity and Associativity

Definition. Let S be a set and ◦ a binary operation 1 on S. We say that the
binary operation ◦ is commutative on S if a ◦ b = b ◦ a for all a, b ∈ S. We
say that the binary operation ◦ is associative on S if (a ◦ b) ◦ c = a ◦ (b ◦ c)
for all a, b, c ∈ S.
Example II.4. Addition and multiplication on R (or C or R[x] or . . . ) are
both commutative and associative. When operations are commutative
and associative, order and bracketing do not matter:
e + ((c + b) + (d + a)) = a + b + c + d + e, e · ((c · b) · (d · a)) = a · b · c · d · e.
Of course subtraction is neither commutative nor associative (write some
examples). ♦
Example II.5. Addition is commutative and associative on Rn . The cross
product is not commutative on R3 . You should know that if x, y ∈ R3 then
y × x = −x × y.
1Here ◦ doesn’t have to be composition of functions. Simply any binary operation
on any set.
8 II. ALGEBRAIC REORIENTATION

We say that the cross product is anti-commutative. ♦

Example II.6. Let S = {a, b, c} and let ◦ be the binary operation given by
the composition table in Example II.3. Then ◦ is not commutative; for
example
a ◦ b = c, b ◦ a = a.
It is also not associative; for example
(a ◦ b) ◦ c = c ◦ c = c, a ◦ (b ◦ c) = a ◦ a = b.
♦
Example II.7. Composition of functions from a set A to itself is associa-
tive but not commutative. We know that it is associative from Lemma II.2.
We know that it isn’t commutative by Example II.1. When a binary oper-
ation is associative bracketing doesn’t matter. For example,
(a ◦ b) ◦ ((c ◦ d ) ◦ e) = (a ◦ (b ◦ c)) ◦ (d ◦ e).
As long as we keep a, b, c, d , e in the same order from left to right, then
the order in which we do the compositions does not matter. Thus there
would be no ambiguity in writing
(a ◦ b) ◦ ((c ◦ d ) ◦ e) = a ◦ b ◦ c ◦ d ◦ e.
This fact that bracketing doesn’t matter as long as we keep the same order
is called the general associativity theorem. For a proper formulation and
proof see
https://proofwiki.org/wiki/General_Associativity_Theorem/Formulation_2/Proof_1 ♦

Example II.8. Are there binary operations that are commutative but not
associative? Yes but it isn’t easy to come up with ‘natural’ examples. How-
ever it is easy to invent a finite set and a composition table that is commu-
tative but not associative. Let S = {a, b, c}. Let ◦ be the binary operation
on S with the following composition table:
◦ a b c
a b c a
b c c a
c a a c
Note that ◦ is commutative; you can see this by noting that the table
is symmetric about the diagonal from the top left corner to the bottom
right corner. But it isn’t associative. For example,
(b ◦ c) ◦ a = a ◦ a = b, b ◦ (c ◦ a) = b ◦ a = c.
♦
Exercise II.9. In the following, is ◦ a binary operation on A? If so, is it
commutative? Is it associative? In each case justify your answer.
(a) A = R is the set of real numbers and a ◦ b = a/b.
(b) A = {1, 2, 3, 4, . . . } is the set of positive integers and a ◦ b = a b .
 II.8. WHERE ARE THE PROOFS? 9

(c) A = {. . . , 1/8, 1/4, 1/2, 1, 2, 4, 8, . . . } is the set of powers of 2 and a ◦

b = ab.
(d) A = C is the set of complex numbers and a ◦ b = |a − b|.

II.8. Where are the Proofs?

You might be somewhat perturbed by the cavalier way I’m stating
things without proving them. In mathematics we have to start with some
assumptions (sometimes called axioms) and then prove things from there.
A reasonable starting point is the properties of the real numbers. These
we assume. What are they?
For all real numbers a, b, c
(i) a + b = b + a (addition is commutative)
(ii) (a + b) + c = a + (b + c) (addition is associative)
(iii) a + 0 = a (0 is the additive identity element)
(iv) there is a real number −a (the additive inverse of a) such that
a + (−a) = 0.
(v) ab = ba (multiplication is commutative)
(vi) (ab)c = a(bc) (multiplication is associative)
(vii) a(b + c) = ab + ac (multiplication distributes over addition)
(viii) a · 1 = a (1 is the multiplicative identity element)
(ix) if a 6= 0, there is a real number denoted by a −1 (the multiplicative
inverse of a) such that a · a −1 = 1.
We have not exhausted the properties of real numbers. For example, we
can add
(x) If a ≥ b then a + c ≥ b + c.
(xi) If a ≥ b and c > 0 then ac ≥ bc. If a ≥ b and c < 0 then ac ≤ bc.
One particularly important property that we will not write down, but which
you will come to admire in the analysis courses is ‘The Completeness Ax-
iom’.
These properties are a reasonable starting point. We should be able
to prove all the facts that we have been stating starting from here. For
example, let us prove that multiplication of complex numbers is com-
mutative. In other words, we want to show that if α and β are complex
numbers then αβ = βα. So suppose that α and β are complex numbers.
Write α = a + bi and β = c + d i where a, b, c, d are real numbers. Then by
the definition of multiplication

αβ = (ac − bd ) + (ad + bc)i , βα = (c a − d b) + (d a + cb)i .

But ac = c a, bd = d b, ad = d a, bc = cb. How do we know this; isn’t

this the same as what we want to prove? No, not really. We know this
because a, b, c, d are real numbers and we are using the commutativity
of multiplication for real numbers which we already decided to assume.
It follows that αβ = βα which we wanted to prove.
10 II. ALGEBRAIC REORIENTATION

Exercise II.10. You know that if a, b ∈ R and ab = 0 then either a = 0 or

b = 0. Explain how this follows from property (ix) above.

II.9. The Quaternionic Number System (do not read)

This section is non-examinable; do not read it. It is here for the ben-
efit of those who believe that the above discussion of commutativity of
complex numbers is overly pedantic. “Why should multiplication not be
commutative? After all, it is just multiplication. You are wasting time on
contrived pedanticisms”. For your benefit I will briefly exhibit the quater-
nionic number system where multiplication is not commutative. Quater-
nions were fashionable in the late 19th century and had substantial phys-
ical applications. Eventually it was discovered that vectors do a better job
of just about anything you could do with quaternions, and they fell out of
fashion.
Remember that the complex numbers are of the form a + bi where a,
b are real and i is a symbol satisfying i 2 = −1. Well, quaternions are of
the form a + bi + c j + d k where a, b, c, d are real and i , j , k are symbols
satisfying
i 2 = j 2 = k 2 = −1, i j = − j i = k, j k = −k j = i , ki = −i k = j.
You can already see that quaternionic multiplication is not commutative,
since i j 6= j i . You might also calculate (1 + i )(1 + j ) and (1 + j )(1 + i ).
Here is a standard description of the discovery of quaternions, which
I’ve copied and pasted from Wikipedia:
[Sir William Rowan] Hamilton knew that the complex
numbers could be viewed as points in a plane, and he
was looking for a way to do the same for points in space.
Points in space can be represented by their coordinates,
which are triples of numbers, and for many years Hamil-
ton had known how to add and subtract triples of num-
bers. But he had been stuck on the problem of multi-
plication and division: He did not know how to take the
quotient of two points in space.
The breakthrough finally came on Monday 16 Oc-
tober 1843 in Dublin, when Hamilton was on his way
to the Royal Irish Academy where he was going to pre-
side at a council meeting. While walking along the tow-
path of the Royal Canal with his wife, the concept be-
hind quaternions was taking shape in his mind. Hamil-
ton could not resist the impulse to carve the formulae
for the quaternions
i 2 = j 2 = k 2 = i j k = −1
into the stone of Brougham Bridge as he passed by it
. . . Since 1989, the Department of Mathematics of the
 II.9. THE QUATERNIONIC NUMBER SYSTEM (DO NOT READ) 11

National University of Ireland, Maynooth has organized

a pilgrimage, where scientists (including physicists Mur-
ray Gell-Mann in 2002, Steven Weinberg in 2005, and
mathematician Andrew Wiles in 2003) take a walk from
Dunsink Observatory to the Royal Canal bridge where,
unfortunately, no trace of Hamilton’s carving remains.
You see, even though the quaternions have been consigned to the com-
post heap of algebra, Hamilton’s graffiti became history’s most celebrated
act of mathematical vandalism. There is a great moral to this, but I can’t
find it.
 CHAPTER III

Matrices—Read On Your Own

You almost certainly met matrices during A-Levels, and you’ll see them
again in Linear Algebra. In any case you need to know about matrices for
this module. In this chapter I summarize what you need to know. We’ll
not cover this chapter in the lectures; I expect you to read it on your own.
Even if you think you know all about matrices I advise you to read this
chapter: do you know why matrix multiplication is defined the way it is?
Do you know why matrix multiplication is associative?

III.1. What are Matrices?

Let m, n be positive integers. An m × n matrix (or a matrix of size
m × n) is a rectangular array consisting of mn numbers arranged in m
rows and n columns:
 
a 11 a 12 a 13 . . . a 1n
 a 21 a 22 a 23 . . . a 2n 
..  .
 
 .. .. ..
 . . . . 
a m1 a m2 a m3 . . . a mn
Example III.1. Let
   
µ ¶ 3 −2 3 1 5
1 −2 0
A= , B = −1 8  , C = −6 −8 12 .
−1 7 14
2 5 2 5 0

A, B , C are matrices. The matrix A has size 2×3 because it has 2 rows and
3 columns. Likewise B has size 3 × 2 and C has size 3 × 3. ♦
Displaying a matrix A by writing
 
a 11 a 12 a 13 . . . a 1n
 a 21 a 22 a 23 . . . a 2n 
A =  .. ..  .
 
.. ..
 . . . . 
a m1 a m2 a m3 . . . a mn

wastes a lot of space. It is convenient to abbreviate this matrix by the

notation A = (a i j )m×n . This means that A is a matrix of size m × n (i.e. m
rows and n columns) and that we shall refer to the element that lies at the
intersection of the i -th row and j -th column by a i j .
13
14 III. MATRICES—READ ON YOUR OWN

Example III.2. Let A = (a i j )2×3 . We can write A out in full as

µ ¶
a 11 a 12 a 13
A= .
a 21 a 22 a 23
Notice that A has 2 rows and 3 columns. The element a 12 belongs to the
1st row and the 2nd column. ♦
Definition. M m×n (R) is the set of m × n matrices with entries in R. We
similarly define M m×n (C), M m×n (Q), M m×n (Z), etc.
Example III.3.
½µ ¶ ¾
a b
M 2×2 (R) = : a, b, c, d ∈ R .
c d
♦

III.2. Matrix Operations

Definition. Given matrices A = (a i j ) and B = (b i j ) of size m×n, we define
the sum A + B to be the m × n matrix whose (i , j )-th element is a i j + b i j .
We define the difference A − B to be the m × n matrix whose (i , j )-th ele-
ment is a i j − b i j .
Let λ be a scalar. We define λA to be the m × n matrix whose (i , j )-th
element is λa i j .
We let −A be the m × n matrix whose (i , j )-th element is −a i j . Thus
−A = (−1)A.
Note that the sum A + B is defined only when A and B have the same
size. In this case A +B is obtained by adding the corresponding elements.
Example III.4. Let
   
µ ¶ 4 3 −4 2
2 −5
A= , B =  1 0 , C =  0 6 .
−2 8
−1 2 9 1
Then A + B is undefined because A and B have different sizes. Similarly
A +C is undefined. However B +C is defined and is easy to calculate:
     
4 3 −4 2 0 5
B +C =  1 0 +  0 6 = 1 6 .
−1 2 9 1 8 3
Likewise A − B and A −C are undefined, but B −C is:
     
4 3 −4 2 8 1
B −C =  1 0 −  0 6 =  1 −6 .
−1 2 9 1 −10 1
Scalar multiplication is always defined. Thus, for example
   
µ ¶ 8 6 −6 3
−2 5
−A = , 2 B =  2 0 , 1.5C =  0 9 .
2 −8
−2 4 13.5 1.5
 III.2. MATRIX OPERATIONS 15

♦
Definition. The zero matrix of size m × n is the unique m × n matrix
whose entries are all 0. This is denoted by 0m×n , or simply 0 if no con-
fusion is feared.
Definition. Let A = (a i j )m×n and B = (b i j )n×p . We define the product
AB to be the matrix C = (c i j )m×p such that
ci j = ai 1 b1 j + ai 2 b2 j + ai 3 b3 j + · · · + ai n bn j .
Note the following points:
• For the product AB to be defined we demand that the number
of columns of A is equal to the number of rows of B .
• The i j -th element of AB is obtained by taking the dot product of
the i -th row of A with the j -th column of B .
Example III.5. Let
µ ¶ µ ¶
1 2 5 −3
A= , B= .
−1 3 0 −2
Both A and B are 2 × 2. From the definition we know that A × B will be a
2 × 2 matrix. We see that
µ ¶ µ ¶
1×5+2×0 1 × −3 + 2 × −2 5 −7
AB = = .
−1 × 5 + 3 × 0 −1 × −3 + 3 × −2 −5 −3
Likewise
µ ¶ µ ¶
5 × 1 + −3 × −1 5 × 2 − 3 × 3 8 1
BA= = .
0 × 1 − 2 × −1 0 × 2 + −2 × 3 2 −6
We make a very important observation: AB 6= B A in this example. So
matrix multiplication is not commutative. ♦
Example III.6. Let A be as in the previous example, and let
µ ¶
2 1 3
C= .
3 −4 0
Then µ ¶
8 −7 3
AC = .
7 −13 −3
However, C A is not defined because the number of columns of C is not
equal to the number of rows of A. ♦
Remark. If m 6= n, then we can’t multiply two matrices in M m×n (R). How-
ever, matrix multiplication is defined on M n×n (R) and the result is again
in M n×n (R). In other words, multiplication is a binary operation on M n×n (R).
Exercise III.7. Commutativity—What can go wrong?
• Give a pair of matrices A, B , such that AB is defined but B A isn’t.
• Give a pair of matrices A, B , such that both AB and B A are de-
fined but they have different sizes.
16 III. MATRICES—READ ON YOUR OWN

• Give a pair of matrices A, B , such that AB and B A are defined

and of the same size but are unequal.
• Give a pair of matrices A, B , such that AB = B A.

III.3. Where do matrices come from?

No doubt you have at some point wondered where matrices come
from, and what is the reason for the weird definition of matrix multipli-
cation. It is possible that your A-Level teachers didn’t want to tell you.
Because I am a really sporting kind of person and I love you very much, I I ♥ you
am telling you some secrets of the trade.
Matrices originate from linear substitutions. Let a, b, c, d be fixed
numbers, x, y some variables, and define x 0 , y 0 by the linear substitutions
x 0 = ax + b y
(III.1)
y 0 = c x + d y.
The definition of matrix multiplication allows us to express this pair of
equations as one matrix equation
µ 0¶ µ ¶µ ¶
x a b x
(III.2) = .
y0 c d y
You should multiply out this matrix equation and see that it is the same
as the pair of equations (III.1).
Now suppose moreover that we define new quantities x 00 and y 00 by
x 00 = αx 0 + βy 0
(III.3)
y 00 = γx 0 + δy 0 ,
where α, β, γ, δ are constants. Again we can rewrite this in matrix form
as
α β x0
µ 00 ¶ µ ¶µ ¶
x
(III.4) = .
y 00 γ δ y0
What is the relation between the latest quantities x 00 , y 00 and our first
pair x, y? One way to get the answer is of course to substitute equations
(III.1) into (III.3). This gives us
x 00 = (αa + βc)x + (αb + βd )y
(III.5)
y 00 = (γa + δc)x + (γb + δd )y.
This pair of equations can re-expressed in matrix form as
αa + βc αb + βd x
µ 00 ¶ µ ¶µ ¶
x
(III.6) = .
y 00 γa + δc γb + δd y
Another way to get x 00 , y 00 in terms of x, y is to substitute matrix equation
(III.2) into matrix equation (III.4):
α β a b x
µ 00 ¶ µ ¶µ ¶µ ¶
x
(III.7) = .
y 00 γ δ c d y
 III.4. HOW TO THINK ABOUT MATRICES? 17

If the definition of matrix multiplication is sensible, then we expect that

matrix equations (III.6) and (III.7) to be consistent. In other words, we
would want that
α β a b αa + βc αb + βd
µ ¶µ ¶ µ ¶
= .
γ δ c d γa + δc γb + δd
A quick check using the definition of matrix multiplication shows that
this is indeed the case.

III.4. How to think about matrices?

Let A ∈ M 2×2 (R). In words, A is a 2 × 2 matrix with real entries. Write
µ ¶
a b
A= .
c d
For now, think of the elements of R2 as column vectors: any u ∈ R2 can be
written as µ ¶
x
u=
y
with x, y real numbers. Thus we’re thinking of the elements of R2 as 2×1-
matrices. Note that in equation
µ 0¶ (III.2), the matrix A ‘converts’ the vector
x
u to another vector u0 = 0 .
y
Some mathematicians would think that the word ‘converts’ is not very
mathematical. Instead they would think of the matrix A as defining a
function
T A : R2 → R2 , T A (u) = Au.
Other (less pedantic) mathematicians would not distinguish between the
matrix and the function it defines. One of the points of the previous sec-
tion is that if C = B A then TC = TB ◦ T A , so that matrix multiplication is
really an instance of composition of functions.
Let us look at some examples of these functions T A .
Example III.8. Let
µ ¶ µ ¶ µ ¶
0 1 2 0 0 0
A= , B= , C= .
1 0 0 1 0 1
Then A defines a function T A : R2 → R2 given by T A (u) = Au. Let us cal-
culate T A explicitly:
µ ¶ µ ¶µ ¶ µ ¶
x 0 1 x y
TA = = .
y 1 0 y x
We note that, geometrically speaking, T A represents reflection in the line
y = x. µ ¶ µ ¶
x 2x
Similarly TB = , which geometrically represents stretching by
y y
a factor of 2 in the x-direction.
18 III. MATRICES—READ ON YOUR OWN
µ ¶ µ ¶
x 0
Also TC = . Thus geometrically, TC represents projection onto
y y
the y-axis.
Again, if we choose not to distinguish between the matrix and the
function it defines we would say that A represents reflection in the line
y = x, B represents stretching by a factor of 2 in the x-direction, and C
represent projection onto the y-axis.
Now is a good time to revisit the non-commutativity of matrices. Let non-commutativity
us see a geometric example of why matrix multiplication is not commu- seen geometrically
tative. Consider the matrices AB and B A where A, B are the above ma-
trices. Notice (AB )u = A(B u). This means stretch u by a factor of 2 in the
x-direction, then reflect it in the line y = x. And (B A)u = B (Au), which
means reflect u in the line y = x and then stretch by a factor of 2 in the
x-direction. The two are not the same as you can see from Figure III.1.
Therefore AB 6= B A. ♦

F IGURE III.1. Non-commutativity of matrix multiplica-

tion. The matrix A represents reflection in the line y = x
and the matrix B represents stretching by a factor of 2 in
the x-direction. On the top row we apply B first then A;
the combined effect is represented by AB . On the bottom
we apply A first then B ; the combined effect is represented
by B A. It is obvious from comparing the last picture on the
top row and the last one on the bottom row that AB 6= B A.
 III.5. WHY COLUMN VECTORS? 19

Remark. Matrices don’t give us all possible functions R2 → R2 . You will

see in Linear Algebra that they give us what are called the linear transfor-
mations. For now, think about
µ ¶ µ ¶
2 2 x x
S :R →R , S = .
y y +1
Geometrically, S translates a vector by 1 unit in the y-direction. Can we
get S from a matrix A? Suppose we can, so S = T A for some matrix A.
What this means is that Su = T A u for all u ∈ R2 . But T A u = Au. So Su =
Au. Now let u = 0. We see that
µ ¶
0
Su = , Au = 0
1
which contradicts Su = Au. So we can’t get S from a matrix, and the rea-
son as you’ll see in Term 2 is that S is not a linear transformation.

III.4.1. Why is matrix multiplication associative? At the end of Ex-

ample III.8 there was a slight-of-hand that you might have noticed 1. We
assumed that matrix multiplication is associative when we wrote (AB )u =
A(B u); here we’re thinking of u as a 2 × 1-matrix. In fact matrix multipli-
cation is associative whenever it is defined.
Theorem III.9. Let A be an m × n matrix, B be an n × p matrix and C a
p × q matrix, then
(III.8) (AB )C = A(BC ).

P ROOF. In terms of functions, (III.8) is saying

(T A ◦ TB ) ◦ TC = T A ◦ (TB ◦ TC ).
This holds by Lemma II.2 2. 
Don’t worry too much if this proof makes you uncomfortable! When
you do Linear Algebra in Term 2 you will see a much more computational
proof, but in my opinion the proof above is the most enlightening one.
For now, you should be pleased if you have digested Example III.8.

III.5. Why Column Vectors?

You will have noticed that early on in these notes we were thinking
of the elements of Rn as row vectors. But when we started talking about
matrices as functions, we have taken a preference for column vectors as
opposed to row vectors. Let us see how things are different if we stuck
with row vectors. So for the moment think of elements of Rn , Rm as row
vectors. Let A be an m × n matrix.
1“well-done” if you did notice, and “learn to read more critically” if you haven’t
2Here T , T , T respectively are functions Rn → Rm , Rp → Rn , Rq → Rp .
A B C
20 III. MATRICES—READ ON YOUR OWN

If u is a (row) vector in Rn or Rm then Au is undefined. But we find

that uA is defined if u is a (row) vector in Rm and gives a (row) vector in
Rn . Thus we get a function
S A : Rm → Rn
given by S A (u) = uA. It is now a little harder to think of the matrix A as a
function since we have written it on the right in the product uA (remem-
ber that when we thought of vectors as columns we wrote Au).
Some mathematicians (particularly algebraists) write functions on the
right, so instead of writing f (x) they will write x f . They will be happy to
think of matrices as functions on row vectors because they can write the
matrix on the right 1. Most mathematicians write functions on the left.
They are happier to think of matrices as functions on column vectors be-
cause they can write the matrix on the left.
Many of the abstract algebra textbooks you will see in the library write left versus right
functions on the right. Don’t be frightened by this! If you’re uncomfort-
able with functions on the right, just translate by rewriting the theorems
and examples in your notation.

III.6. Multiplicative Identity and Multiplicative Inverse

We have mostly been focusing on 2×2 matrices, and we will continue
to focus on them. One natural question to ask is what is the multiplica-
tive identity for 2 × 2 matrices? You might be wondering what I mean by
the multiplicative identity? You of course know that a · 1 = 1 · a = a for all
real numbers a; we say that 1 is the multiplicative identity in R. Likewise
the multiplicative identity for 2 × 2 matrices will be a 2 × 2 matrix, which
we happen to call I 2 , satisfying AI 2 = I 2 A = A for all 2 × 2 matrices A. An-
other natural question is given a 2 × 2 matrix A, what is its multiplicative
inverse A −1 ? Does it even have an inverse? It is likely that you know the
answers to these questions from school. If not don’t worry, because we’re
about to discover the answers. If yes, please unremember the answers,
because we want to work out the answers from scratch. We want to im- temporary amnesia
merse ourselves in the thought process that went into discovering these required
answers.
The first question is about the multiplicative identity. We haven’t yet
discovered what the multiplicative identity is, but let us denote it by I 2 .
What is the geometric meaning of I 2 ? Clearly we want I 2 to have the geo-
metric meaning of ‘do nothing’, as opposed to reflect, stretch, project, etc.
In symbols we want a 2 × 2 matrix I 2 so that I 2 u = u for all u ∈ R2 . Since I 2

1Algebraists have many idiosyncrasies that distinguish them from other mathe-
maticians. I find most of these bewildering. However, they do have a very good point
in the way they write functions. We said that B A means do A first and then B , because
(B A)u = B (Au). However if you use row vectors then u(B A) = (uB )A, so B A means do B
first then A, which seems entirely natural.
 III.6. MULTIPLICATIVE IDENTITY AND MULTIPLICATIVE INVERSE 21

is a 2 × 2 matrix we can write

µ ¶
a b
I2 = ,
c d
where a, b, c, d are numbers. Let us also write
µ ¶
x
u= .
y
We want µ ¶µ ¶ µ ¶
a b x x
= .
c d y y
We want this to be true for all values of x, y, because we want the matrix
I 2 to mean ‘do nothing to all vectors’. Multiplying the two matrices on the
right and equating the entries we obtain
ax + b y = x, c x + d y = y.
We instantly see that the choices a = 1, b = 0, c = 0, d = 1 work. So the
matrix µ ¶
1 0
I2 =
0 1
has the effect of ‘do nothing’. Let’s check algebraically that I 2 is a multi-
plicative identity for 2 × 2 matrices. What we want to check is that
(III.9) AI 2 = I 2 A = A
for every 2 × 2 matrix A. We can write
α β
µ ¶
A= .
γ δ
Now multiplying we find
α β 1 0 α×1+β×0 α×0+β×1 α β
µ ¶µ ¶ µ ¶ µ ¶
AI 2 = = = = A.
γ δ 0 1 γ×1+δ×0 γ×0+δ×1 γ δ
In exactly the same way, you can do the calculation to show that I 2 A = A,
so we’ve established (III.9).
Before moving on to inverses, it is appropriate to ask in which world
does the identity (III.9) hold? What do I mean by that? Of course A has to
be a 2×2 matrix, but are its entries real, complex, rational, integral? If you
read the above again, you will notice that we’ve used properties common
to the number systems R, C, Q, Z. So (III.9) holds for all matrices A in
M 2×2 (R), M 2×2 (C), M 2×2 (Q), M 2×2 (Z).

Now what about inverses? Let A be a 2 × 2 matrix, and let A −1 be ‘its

inverse’ whatever that means. If A represents a certain geometric oper-
ation then A −1 should represent the opposite geometric operation. The
matrix A −1 should undo the effect of A. The product A −1 A, which is the
result of doing A first then A −1 , should now mean ‘do nothing’. In other
words, we want A −1 A = I 2 whenever A −1 is the inverse of A. Another way
of saying the same thing is that if v = Au then u = A −1 v.
22 III. MATRICES—READ ON YOUR OWN

Should there be such an inverse A −1 for every A. No, if A = 02×2 then

A −1 A = 02×2 6= I 2 . The zero matrix is not invertible, which is hardly sur-
prising. Are there any others? Here it is good to return to the three matri-
ces in Example III.8 and test if they’re invertible.
Example III.10. Let
µ ¶ µ ¶ µ¶
0 1 2 0 0 0
A= , B= , C= .
1 0 0 1 0 1
Recall that A represents reflection in the line y = x. If we repeat a reflec-
tion then we end up where we started. So we expect that A · A = I 2 (or
more economically A 2 = I 2 ). Check this by multiplying. So A is its own
inverse.
The matrix B represents stretching by a factor of 2 in the x-direction.
So its inverse B −1 has to represent stretching by a factor of 1/2 (or shrink-
ing by a factor of 2) in the x-direction. We can write
µ ¶
−1 1/2 0
B = .
0 1
Check for yourself that B −1 B = I 2 . Also note that
µ ¶ µ ¶
−1 x x/2
B =
y y
which does what we want: B −1 really is the inverse of B .
Finally recall that C represents projection onto the y-axis. Is there
such as thing as unprojecting from the y-axis? Note that
µ ¶ µ ¶ µ ¶ µ ¶ µ ¶ µ ¶ µ ¶ µ ¶
1 0 2 0 3 0 4 0
C = , C = , C = , C = ,....
0 0 0 0 0 0 0 0
Let’s assume that C has an inverse and call it C −1 . One of the things we
want is for v = C u to imply u = C −1 v. In other words, C −1 is the opposite
of C . If there was such an inverse C −1 then
µ ¶ µ ¶ µ ¶ µ ¶ µ ¶ µ ¶ µ ¶ µ ¶
−1 0 1 −1 0 2 −1 0 3 −1 0 4
C = , C = , C = , C = ,....
0 0 0 0 0 0 0 0
This is clearly absurd! Therefore, C is not invertible 1. For a more graphic
illustration of this fact, see Figure III.2.
The matrix C is non-zero, but it still doesn’t have an inverse. This
might come as a shock if you haven’t seen matrix inverses before. So let’s
check it in a different way. Write
µ ¶
−1 a b
C = .
c d
1In Foundations, one of the things you’ll learn (or have already done) is that a func-
tion is invertible if and only if it is bijective. To be bijective a function has to be injective
and surjective. We have shown that the ‘function’ C is not injective, therefore it is not
bijective, therefore it is not invertible. If this footnote does not make sense to you yet,
return to it at the end of term.
 III.6. MULTIPLICATIVE IDENTITY AND MULTIPLICATIVE INVERSE 23

y
projection onto the y-axis

L
S1 S2 S3
x

F IGURE III.2. Some non-zero matrices don’t have in-

verses. The matrix C represents projection onto the y-axis.
Note that C sends the three ‘smileys’ S 1 , S 2 , S 3 to the line
segment L. If C had an inverse, would this inverse send L
to S 1 , S 2 or S 3 ? We see that C −1 does not make any sense!

We want C −1C = I 2 . But

µ ¶µ ¶ µ ¶
−1 a b 0 0 0 b
C C= = .
c d 0 1 0 d
We see that no matter what choices of a, b, c, d we make, this will not
equal I 2 as the bottom-left entries don’t match. So C is not invertible. ♦

III.6.1. Discovering Invertibility. We will now work in generality. Let

A be a 2 × 2 matrix and write
µ ¶
a b
A= .
c d
Suppose that A is invertible and write
µ ¶
−1 x y
A = .
z w

We want to express A −1 in terms of A, or more precisely, x, y, z, w in

terms of a, b, c, d . We want
µ ¶µ ¶ µ ¶
x y a b 1 0
= .
z w c d 0 1
Multiplying and equating entries we arrive at four equations:
(III.10) ax + c y = 1
(III.11) bx + d y = 0
(III.12) az + c w = 0
(III.13) bz + d w = 1.
24 III. MATRICES—READ ON YOUR OWN

We treat the first two equations as simultaneous equations in x and y.

Let’s eliminate y and solve for x. Multiply the first equation by d , the
second by c and subtract. We obtain (ad − bc)x = d . By doing similar
eliminations you’ll find that
(
(ad − bc)x = d , (ad − bc)y = −b,
(III.14)
(ad − bc)z = −c, (ad − bc)w = a.
Let’s assume that ad − bc 6= 0. Then, we have
d −b −c a
x= , y= , z= , w= .
ad − bc ad − bc ad − bc ad − bc
Thus µ ¶
−1 1 d −b
A = .
ad − bc −c a
Now check by multiplying that A −1 A = I 2 and also A A −1 = I 2 .
What if ad − bc = 0? We assumed that A has an inverse A −1 and de-
duced (III.14). If ad − bc = 0 then (III.14) tells us that a = b = c = d = 0
and so A = 02×2 which certainly isn’t invertible. This is a contradiction.
Thus if ad − bc = 0 then A is not invertible. We’ve proved the following
theorem.
µ ¶
a b
Theorem III.11. A matrix A = is invertible if and only if ad − bc 6=
c d
0. If so, the inverse is
µ ¶
−1 1 d −b
A = .
ad − bc −c a

Definition. Let A be a 2 × 2 matrix and write

µ ¶
a b
A= .
c d
We define the determinant of A, written det(A) to be
det(A) = ad − bc.
Another common notation for the determinant of the matrix A is the fol-
lowing
¯ ¯
¯a b ¯
¯ c d ¯ = ad − bc.
¯ ¯

From Theorem III.11 we know that a 2 × 2 matrix A is invertible if and

only if det(A) 6= 0.
Theorem III.12. (Properties of Determinants) Let A, B be 2 × 2 matrices.
(a) det(I 2 ) = 1.
(b) det(AB ) = det(A) det(B ).
1
(c) If A is invertible then det(A) 6= 0 and det(A −1 ) = .
det(A)
 III.6. MULTIPLICATIVE IDENTITY AND MULTIPLICATIVE INVERSE 25

P ROOF. The proof is mostly left as an exercise for the reader. Parts (a), (b)
follow from the definition and effortless calculations (make sure you do
them). For (c) note that

det(A −1 A) = det(I 2 ) = 1.

Now applying (ii) we have 1 = det(A −1 ) det(A). We see that det(A) 6= 0 and
det(A −1 ) = 1/ det(A). 

Exercise III.13. (The Geometric Meaning of Determinant) You might be

wondering (in fact should be wondering) about the geometric meaning
of the determinant. This exercise answers your question. Let A be a 2 × 2
matrix and write
µ ¶
a b
A= .
c d
µ ¶ µ ¶
a b
Let u = and v = ; in other words, u and v are the columns of A.
c d
Show that |det(A)| is the area of the parallelogram with adjacent sides u
and v (See Figure III.3). This tells you the meaning of |det(A)|, but what
about the sign of det(A)? What does it mean geometrically? Write down
and sketch a few examples and see if you can make a guess. Can you
prove your guess?

F IGURE III.3. If u and v are the columns of A then the

shaded area is |det(A)|.

µ ¶ µ ¶
a b
Exercise III.14. Suppose u = and v = are non-zero vectors, and
c d
µ ¶
a b
let A be the matrix with columns u and v; i.e. A = . Show (al-
c d
gebraically) that det(A) = 0 if and only if u, v are parallel. Explain this
geometrically.
26 III. MATRICES—READ ON YOUR OWN

III.7. Rotations
We saw above some examples of transformations in the plane: reflec-
tion, stretching, projection. In this
µ ¶section we take a closer look at rota-
x
tions about the origin. Let P = be a point in R2 . Suppose that this
y
point is rotated anticlockwise about the µorigin
0
¶ through an angle of θ. We
x
want to write down the new point P 0 = 0 in terms of x, y and θ. The
y
easiest way to do this to use polar coordinates. Let the distance of P from
−−→
the origin O be r and let the angle OP makes with the positive x-axis be
φ; in other words the polar coordinates for P are (r, φ). Thus
x = r cos φ, y = r sin φ.
Since we rotated P anticlockwise about the origin through an angle θ to
obtain P 0 , the polar coordinates for P 0 are (r, φ + θ). Thus
x 0 = r cos(φ + θ), y 0 = r sin(φ + θ).
We expand cos(φ + θ) to obtain
x 0 = r cos(φ + θ)
= r cos φ cos θ − r sin φ sin θ
= x cos θ − y sin θ.
Similarly
y 0 = x sin θ + y cos θ.
We can rewrite the two relations
x 0 = x cos θ − y sin θ, y 0 = x sin θ + y cos θ,
in matrix notation as follows
cos θ − sin θ x
µ 0¶ µ ¶µ ¶
x
= .
y0 sin θ cos θ y
Thus anticlockwise rotation about the origin through an angle θ can be
achieved by multiplying by the matrix 1
cos θ − sin θ
µ ¶
Rθ = .
sin θ cos θ

Exercise III.15. You know that R θ represents anticlockwise rotation about

the origin through angle θ. Describe in words the transformation associ-
ated to −R θ . (Warning: don’t be rash!)

Gutted that the chapter on matrices is coming to an end? Cackle.

You’ll get to gorge (binge?) on them in Linear Algebra.

cos π4 − sin π4
x ¶

µ ¶µ ¶
1Is this clever . . . or lame: x
µ

= ?
sin π4 cos π4 y
y
 CHAPTER IV

Groups

IV.1. The Definition of a Group

A group is a pair (G, ◦) where G is a set and ◦ is a binary operation on
G, such that the following four properties hold 2
(i) (closure) for all a, b ∈ G, a ◦ b ∈ G;
(ii) (associativity) for all a, b, c ∈ G,
a ◦ (b ◦ c) = (a ◦ b) ◦ c;
(iii) (existence of the identity element) there is an element e ∈ G such
that for all a ∈ G,
a ◦ e = e ◦ a = a;
(iv) (existence of inverses) for every a ∈ G, there is an element b ∈ G
(called the inverse of a) such that
a ◦ b = b ◦ a = e.
Remark. If ◦ is a binary operation then (i) automatically holds. So why
did I list (i) in the definition? I’ve put it in for good measure! When you
suspect an operation gives you a group the first thing you should check is
that the operation is really a binary operation.

IV.2. First Examples (and Non-Examples)

Example IV.1. (R, +) is a group. We know already that addition is a binary
operation on R, so ‘closure’ holds. We know addition of real numbers
is associative. What is the identity element? We want an element e ∈ R
so that a + e = e + a = a for all a ∈ R. It is clear that e = 0 works and
is the only possible choice. Moreover, the (additive) inverse of a is −a:
a + (−a) = (−a) + a = 0. ♦
Example IV.2. Recall our definition of the natural numbers:
N = {0, 1, 2, . . . }.
Is (N, +) a group? Conditions (i), (ii) are satisfied. For condition (iii) we
can take the identity element to be 0 (again the only possible choice).
299% of mathematicians call (i)–(iv) the “group axioms”, and you can call them that
if you wish. I call them the “defining properties of a group” since I think that the word
axiom should be reserved for statements of ‘universal truth’. An example of an axiom is:
a + b = b + a for any two integers a, b.
27
28 IV. GROUPS

But (iv) does not hold. For example, if we take a = 1, there is no b ∈ N

such that a + b = b + a = 0. Thus (N, +) is not a group. ♦
Example IV.3. (Z, +), (Q, +) and (C, +) are groups. ♦
Example IV.4. Recall we defined
R∗ = {α ∈ R : α 6= 0}.
Then (R∗ , ·) is a group, where of course · means multiplication. Again
closure and associativity are obvious. If e is the identity element then it
has to satisfy α · e = e · α = α for all α ∈ R. Thus e = 1 and this is the only
choice possible. Then the inverse of α is α−1 .
We can define C∗ and Q∗ in the same way and obtain groups (C∗ , ·)
and (Q∗ , ·).
Can we obtain from Z a group with respect to multiplication? In view
of the above, the obvious candidate is
U = {α ∈ Z : α 6= 0}.
But (U , ·) is not a group. It is true that (i), (ii) and (iii) hold with 1 being
the identity element. But, for example, 2 ∈ U does not have an inverse:
there is no b ∈ U such that b · 2 = 2 · b = 1. So (U , ·) is not a group. But
the answer is not no; all we’ve done is shown that the obvious choice for
a group (Z∗ , ·) made up of integers does not work. We’ll return to this
question and answer it fully in Section XIV.4. ♦
Example IV.5. (R2 , +) is a group. Let’s prove this. We’re allowed to assume
the usual properties of the real numbers (see Section II.8). The elements
of R2 are pairs (a 1 , a 2 ) where a 1 , a 2 are real numbers. Addition is defined
by
(a 1 , a 2 ) + (b 1 , b 2 ) = (a 1 + b 1 , a 2 + b 2 ).
Note that the entries a 1 + b 1 and a 2 + b 2 are real numbers, and so (a 1 +
b 1 , a 2 + b 2 ) is a pair of real numbers. Hence (a 1 + b 1 , a 2 + b 2 ) is in R2 . In
other words, R2 is closed under addition, which shows that (R2 , +) satis-
fies condition (i). Next we want to prove associativity of addition. Con-
sider a, b, c in R2 . We can write
a = (a 1 , a 2 ), b = (b 1 , b 2 ), c = (c 1 , c 2 ).
Here a 1 , a 2 , b 1 , b 2 and c 1 , c 2 are real numbers. Note that
(a + b) + c = ((a 1 + b 1 ) + c 1 , (a 2 + b 2 ) + c 2 ) .
Likewise,
a + (b + c) = (a 1 + (b 1 + c 1 ), a 2 + (b 2 + c 2 )) .
Because addition of real numbers is associative, we know that
(a 1 + b 1 ) + c 1 = a 1 + (b 1 + c 1 ), (a 2 + b 2 ) + c 2 = a 2 + (b 2 + c 2 ).
Hence
(a + b) + c = a + (b + c).
2
This shows that (R , +) satisfies (ii).
 IV.3. ABELIAN GROUPS 29

Next we need an identity element, and the obvious candidate is 0 =

(0, 0). Then
(a 1 , a 2 ) + (0, 0) = (a 1 + 0, a 2 + 0) = (a 1 , a 2 ),
and
(0, 0) + (a 1 , a 2 ) = (0 + a 1 , 0 + a 2 ) = (a 1 , a 2 ).
Thus (iii) is satisfied.
Finally we want an inverse. If a = (a 1 , a 2 ) is in R2 then the inverse
we choose (there’s no other choice) is b = (−a 1 , −a 2 ). This is in R2 and
satisfies
a + b = b + a = (0, 0).
Hence (iv) is satisfied and so (R2 , +) is a group.
If you got bored reading this then you have my sympathy; I was bored
typing it. What matters is that you realize that the properties of addition
no magic yet in R2 are not there by divine covenant nor by magic; they simply follow
from the definition of addition in R2 and corresponding properties of the
real numbers. I can write down similar proofs for Examples IV.6, IV.7, and
IV.8, but I daren’t try your patience with the interminable tedium. ♦
Example IV.6. (Rn , +) is a group for any n ≥ 2. ♦
Example IV.7. (R[x], +) is a group. ♦
Example IV.8. (M m×n (K ), +) are groups for K = C, R, Q, Z, with 0m×n the
identity element. ♦
Example IV.9. All the groups we have met so far are infinite. Here is an
example of a finite group. Let A = {+1, −1}. Then (A, ·) is a group (where
of course · is multiplication). ♦
p
Example IV.10. Let B = {1, i , −1, −i }, where i = −1. Then (B, ·) is an-
other example of a finite group. ♦
Example IV.11. Let C = {1, i }. Then (C , ·) is not a group since it isn’t closed;
for example i · i = −1 ∉ C . ♦

IV.3. Abelian Groups

We say that a group (G, ◦) is abelian if (in addition to the defining
properties (i)–(iv) of a group) it also satisfies
(v) (commutativity) for all a, b ∈ G,
a ◦ b = b ◦ a.
Example IV.12. All the groups we have seen above are actually abelian:
(R, +), (C, +), (Q, +), (R[x], +), (Rn , +), (R∗ , ·), (C∗ , ·), (M m×n (R), +), . . . ♦
Are there any non-abelian groups? There are many, but perhaps not
ones that you’re used to thinking about. In the next section we give an
example of a non-abelian group.
30 IV. GROUPS

IV.4. Symmetries of a Square

In many ways the examples above are misleading for three reasons:
• Most of the examples of groups we have met above have addi-
tional structure. For example, in R we can add, but we can also
multiply and we can divide by non-zero numbers. In fact R is an
example of a field. Like in R2 we have addition and scalar mul-
tiplication, so R2 is an example of a vector space. This doesn’t
stop (R, +) and (R2 , +) from being groups, but if you want to test
your own ideas in group theory, it is best to also look at examples
where there aren’t any of these additional structures.
• The groups above are abelian. The theory of abelian groups is
rather close in flavour to linear algebra. Many of the most inter-
esting groups that you’ll come across during your degree will be
non-abelian.
• All the groups above, except for Example IV.9, are infinite. Al-
though infinite groups are important and interesting, most the-
orems we will do in this course will apply only to finite groups.
Thus it is essential to become familiar with examples of finite
groups.
The group of symmetries of a square is a great example of a group; it is shameless hard sell
finite, non-abelian and there is no additional structure. In future, this will
be an excellent example to test any ideas you have on groups. Consider
a square as in Figure IV.1 with vertices labelled 1, 2, 3, 4 (note the vertex
numbering goes in an anticlockwise direction). Let O be the centre of the
square.
The symmetries of the square are the rotations and reflections that
keep the square occupying exactly the same position (but might change
where the vertex numbers are). Let ρ 0 , ρ 1 , ρ 2 , ρ 3 be anticlockwise rota-
tions of the square about O by 0◦ , 90◦ , 180◦ and 270◦ . In effect, ρ 0 means
“do nothing”. We think of two symmetries as being the same if they have
the same effect on the square. A rotation about O of 360◦ has the same
effect as ρ 0 . A clockwise rotation about O of 90◦ has the same effect as
ρ 3 . We quickly see that ρ 0 , ρ 1 , ρ 2 , ρ 3 are the only rotations that keep the
square in exactly the same position.
What about reflections? There are four reflections that keep the square
occupying exactly the same position, as in Figure IV.1:
• σ0 the reflection about the diagonal joining the top-right vertex
to the bottom-left vertex;
• σ1 the reflection about the line joining the midpoint of top side
and the midpoint of bottom side;
• σ2 the reflection about the diagonal joining top-left vertex and
the bottom-right vertex;
• σ3 the reflection about the line joining the midpoint of the left
side and the midpoint of the right side.
 IV.4. SYMMETRIES OF A SQUARE 31

σ1
σ2 σ0
2 1

σ3
O

3 4

F IGURE IV.1. Left: the square with vertices labelled 1, 2, 3,

4. Right: the reflections σ0 , σ1 , σ2 , σ3 .

Thus the symmetries of a square form a set which we shall denote by

D 4 = {ρ 0 , ρ 1 , ρ 2 , ρ 3 , σ0 , σ1 , σ2 , σ3 }.

We talked about a group of symmetries, so it is not enough to just list

the symmetries, but we have to specify a binary operation. If we have
two symmetries, how do we combine them? In other words, if α, β ∈ D 4 ,
what is α ◦ β? We define α ◦ β to be the symmetry we obtain by doing
β first then α (the order is very important and we’ll talk more about this
later). Thus ρ 1 ◦ ρ 2 is anticlockwise rotation about O of 180◦ followed by
anticlockwise rotation about O of 90◦ . This has the same effect as ρ 3 .
Thus we write ρ 1 ◦ ρ 2 = ρ 3 .
Let’s try another composition: ρ 1 ◦ σ0 . In other words, first reflect in
the diagonal joining 1 and 3, then rotate anticlockwise about O by 90◦ .
This has the same effect at σ1 and we can write ρ 1 ◦ σ0 = σ1 . Also (draw
pictures) σ1 ◦ σ0 = ρ 1 and σ2 ◦ σ2 = ρ 0 (note that ρ 0 means do nothing).
Now we know how to do composition we can write out a composition
table:
◦ ρ0 ρ1 ρ2 ρ3 σ0 σ1 σ2 σ3
ρ0 ρ0 ρ1 ρ2 ρ3 σ0 σ1 σ2 σ3
ρ1 ρ1 ρ2 ρ3 ρ0 σ1 σ2 σ3 σ0
ρ2 ρ2 ρ3 ρ0 ρ1 σ2 σ3 σ0 σ1
ρ3 ρ3 ρ0 ρ1 ρ2 σ3 σ0 σ1 σ2
σ0 σ0 σ3 σ2 σ1 ρ0 ρ3 ρ2 ρ1
σ1 σ1 σ0 σ3 σ2 ρ1 ρ0 ρ3 ρ2
σ2 σ2 σ1 σ0 σ3 ρ2 ρ1 ρ0 ρ3
σ3 σ3 σ2 σ1 σ0 ρ3 ρ2 ρ1 ρ0
32 IV. GROUPS

It is not worth your while to check every entry in the table, but make
sure you check four or five entries at random to get an idea of how to
compose symmetries, and let me know if I’ve made any mistakes!
I want to convince you that (D 4 , ◦) is a group. The first thing we should
ask about is closure. This is clear from the table; when you compose two
elements of D 4 you get an element of D 4 . Let’s skip associativity for now
and talk about the existence of the identity element. For this we don’t
need the table. It is clear that ρ 0 (=do nothing) is an identity element.
It is also (geometrically) clear that every element has an inverse which
does belong to D 4 . If you reflect twice in the same line you end up where
you started, so σi ◦ σi = ρ 0 ; in other words, σi is its own inverse for i =
0, 1, 2, 3. The inverse of an anticlockwise rotation around O by 90◦ is an
anticlockwise rotation around O by 270◦ . We find that the inverses of ρ 0 ,
ρ 1 , ρ 2 and ρ 3 respectively are ρ 0 , ρ 3 , ρ 2 and ρ 1 .
What’s left is to prove associativity. So we have to prove that ( f ◦g )◦h =
f ◦(g ◦h) for all f , g , h ∈ D 4 . But there are 512 triples f , g , h, so that’s a lot
of checking. Is there a clever way? Yes there is, and it relies on thinking
about the elements of D 4 as functions 1. Remember that we labelled the
vertices of the square with 1, 2, 3, 4 as in Figure IV.1. Now ρ 1 takes vertex
1 to where vertex 2 was and vertex 2 to where vertex 3 was and vertex 3 to
where vertex 4 was and vertex 4 to where vertex 1 was. We can think of ρ 1
as a function from {1, 2, 3, 4} to itself as in Figure IV.2.

ρ1 σ0
1 1 1 1

2 2 2 2

3 3 3 3

4 4 4 4

F IGURE IV.2. ρ 1 (left) and σ0 (right) considered as func-

tions from {1, 2, 3, 4} to itself.

In fact, this way every element of D 4 can be thought of as a function a key trick
from {1, 2, 3, 4} to itself. If we think like this, ◦ simply becomes composi-
tion of functions from D 4 to itself. We know that composition of func-
tions is associative by Lemma II.2. Thus the binary operation ◦ on D 4 is

1You’ve seen this idea before: in the proof of Theorem III.9 we showed that matrix
multiplication is associative by thinking of matrices as functions and matrix multiplica-
tion as composition of functions.
 IV.4. SYMMETRIES OF A SQUARE 33

associative. We have now checked all the conditions (i)–(iv) for a group,
so (D 4 , ◦) is a group!

Remarks.
Moral • Notice that by changing the way we looked at the elements of D 4 ,
we saved ourselves from excruciatingly checking 512 laborious
cases. This is a recurring theme in algebra, where a conceptual
outlook saves you from much trouble.
• You might have found our definition of composition in D 4 strange:
α◦β means apply β first then α. The reason for this choice is that
we want to sometimes think of the elements of D 4 as functions,
and when we do that we want composition in D 4 to agree with
the usual composition of functions. Recall that f ◦g means apply
g first then f .
non-abelian group • D 4 is our first example of a non-abelian group. To check that it
isn’t abelian all we have to do is give a pair of symmetries that
don’t commute. For example,
σ0 ◦ ρ 1 = σ3 , ρ 1 ◦ σ0 = σ1 .
IV.4.1. Subgroups of D 4 . The set D 4 contains rotations and reflec-
tions. Let us now look at the rotations on their own and the reflections on
their own:
R = {ρ 0 , ρ 1 , ρ 2 , ρ 3 }, S = {σ0 , σ1 , σ2 , σ3 }.
For now let us look at the part of the composition table that involves only
rotations:
◦ ρ0 ρ1 ρ2 ρ3
ρ0 ρ0 ρ1 ρ2 ρ3
ρ1 ρ1 ρ2 ρ3 ρ0
ρ2 ρ2 ρ3 ρ0 ρ1
ρ3 ρ3 ρ0 ρ1 ρ2
Notice from the table that if we compose two rotations we obtain a ro-
tation. We didn’t really need the table for this; it’s geometrically obvious.
Thus ◦ is a binary operation on R (as well as being a binary operation
on D 4 ). We can ask whether (R, ◦) is a group, and it is easy to see that
the answer is yes (with the same reasoning as before). We have an inter-
esting phenomenon, which is a group (R, ◦) contained in another group
(D 4 , ◦). We say that (R, ◦) is a subgroup of (D 4 , ◦). We will discuss sub-
groups at length later. It is also interesting to note that (R, ◦) is abelian.
An algebraic way of seeing the (R, ◦) is abelian is to note that its compo-
sition table is symmetric about the leading diagonal. But you should also
see geometrically that if you compose two rotations (centred at the same
point) then the order does not matter. So (R, ◦) is an abelian subgroup of
the non-abelian group (D 4 , ◦).
What about (S, ◦)? Do the reflections of the square form a group? By
looking at the composition table the first thing we notice is that S is not
34 IV. GROUPS

closed under composition. So (S, ◦) is not a group. Are there any other
subgroups inside (D 4 , ◦) besides (R, ◦)? Yes. See Figure IV.3 for a complete
list.

(D 4 , ◦)

({ρ 0 , ρ 2 , σ0 , σ2 }, ◦) ({ρ 0 , ρ 1 , ρ 2 , ρ 3 }, ◦) ({ρ 0 , ρ 2 , σ1 , σ3 }, ◦)

({ρ 0 , σ0 }, ◦) ({ρ 0 , σ2 }, ◦) ({ρ 0 , ρ 2 }, ◦) ({ρ 0 , σ1 }, ◦) ({ρ 0 , σ3 }, ◦)

({ρ 0 }, ◦)

F IGURE IV.3. The figure shows the subgroups of (D 4 , ◦)

and how they fit inside each other.

Again, check that a couple of these are subgroups. Don’t waste time
checking there aren’t other subgroups of (D 4 , ◦); when you know a lot
more about groups and subgroups you can come back to this question,
but even then it will still be a little tedious!
Exercise IV.13. In this exercise you will write out the composition table
for the group D 3 which is the group of symmetries of an equilateral tri-
angle. Sketch an equilateral triangle and label the vertices 1, 2, 3 in an-
ticlockwise order. Label the centre of the triangle with O. Let ρ 0 , ρ 1 , ρ 2
denote anticlockwise rotations about O through angles 0, 2π/3 and 4π/3.
Let σ1 , σ2 , σ3 denote reflections about the lines respectively joining ver-
tices 1, 2, 3 to O. Let
D 3 = {ρ 0 , ρ 1 , ρ 2 , σ1 , σ2 , σ3 }.
Write down a composition table for D 3 and explain why it is a group 1. Is
it abelian? It has six subgroups; write them down.
Exercise IV.14. Write down the symmetries of a triangle that is isoceles
but not equilateral and a composition table for them. Do they form a
group?

1More generally, D denotes the group of symmetries of a regular polygon with n

n
sides. These are called the dihedral groups. Some mathematicians denote D n by D 2n
because it has 2n elements. Mysteriously, they don’t denote S n by S n! .
 CHAPTER V

First Theorems

Our first two theorems deal with subconscious assumptions. One of

the defining properties of a group is the ‘existence of the identity element’
(property (iii)). The word ‘the’ contains a hidden assumption; how do we
know there is only one identity element? Shouldn’t we be talking about
the ‘existence of an identity element’?
Theorem V.1. Let (G, ◦) be a group. Then (G, ◦) has a unique identity ele-
ment.

P ROOF. Suppose that e and e 0 are identity elements. Thus, for all a ∈ G
we have

(V.15) a ◦ e = e ◦ a = a,

and

(V.16) a ◦ e 0 = e 0 ◦ a = a.

Now let us try evaluating e ◦ e 0 . If we let a = e and use (V.16) we find

e ◦ e 0 = e.

But if we let a = e 0 and use (V.15) we find

e ◦ e 0 = e 0.

Thus e = e 0 . In other words, the identity element is unique. 

Theorem V.2. Let (G, ◦) be a group and let a be an element of G. Then a

has a unique inverse.

P ROOF. Our proof follows the same pattern as the proof of Theorem V.1,
and you’ll see this pattern again and again during your undergraduate
key trick! career. Almost all uniqueness proofs follow the same pattern: suppose
that there are two of the thing that we want to prove unique; show that
these two must be equal; therefore it is unique.
For our proof we suppose that b and c are both inverses of a. We want
to show that b = c. By definition of inverse (property (iv) in the definition
of a group) we know that

a ◦ b = b ◦ a = e, a ◦ c = c ◦ a = e,
35
36 V. FIRST THEOREMS

where e is of course the identity element of the group. Thus

b = b ◦e by (iii) in the definition of a group
= b ◦ (a ◦ c) from the above a ◦ c = e
= (b ◦ a) ◦ c by (ii) in the definition of a group
= e ◦c from the above b ◦ a = e
=c by (iii) again.
Thus b = c. Since any two inverses of a must be equal, we see that the
inverse of a is unique. 

V.1. Getting Relaxed about Notation

It is quite tedious to keep writing ◦ for the group operation. If (G, ◦) is
a group and a, b ∈ G, we shall write ab for a ◦ b, unless there is reason to
fear confusion. For example if (G, ◦) = (R, +) then it is stupid to write ab
for a + b because the usual meaning for ab is “a × b”. But it is OK most of
the time, and when it is OK we will do it. Moreover, we shall often say “let
G be a group”, without giving an explicit name to the binary operation.
When we talk of the groups R, R2 , R[x], R∗ , etc. we shall mean the groups
(R, +), (R2 , +), (R[x], +), (R∗ , ·), etc.
If G is a group, and we’re writing ab for a ◦ b, then it makes sense to
use 1 to denote the identity element instead of e. We write a −1 for the
(unique) inverse of a. Now
aa −1 = a −1 a = 1,
which looks familiar. Moreover, if n is a positive integer we shall write
a n = |aa{z
· · · a} .
n times
0 −n n −1
We let a = 1 and a = (a ) . Again we should reflect a little to make
sure we’re not being reckless. Does a 3 mean (a ◦ a) ◦ a or a ◦ (a ◦ a)? It
doesn’t matter because of the associativity property of a group.
Example V.3. Let ◦ be the binary operation on S = {a, b, c} in Example II.3.
Note that (S, ◦) most definitely is not a group, as ◦ is not associative. Now
you can check that
(a ◦ a) ◦ a = a, a ◦ (a ◦ a) = c.
3
Thus writing a in this context does not make any sense. ♦
Let’s get back to groups. It’s the associativity which makes it OK for us
to write a n , and you can convince yourself quickly that
(a m )n = a mn , a m a n = a m+n .
You should realize that all this is happening inside the group G that con-
tains the element a. In particular, a n ∈ G for all n ∈ Z. How do we know
this? For a start, G is closed under composition, so because a ∈ G, so
is a 2 = a ◦ a. Now that we know that a and a 2 are in G, we know that
 V.1. GETTING RELAXED ABOUT NOTATION 37

a 3 = a ◦ a 2 ∈ G and so on. You can use induction to show that a n ∈ G for

n = 1, 2, 3, . . . . Also 1 ∈ G (because 1 is the symbol we’re using for the iden-
tity element of G). And we’ve adopted the convention a 0 = 1, so a 0 ∈ G.
We also want to check that a −1 , a −2 , . . . are in G. But a −n = (a n )−1 , and
since a n is already in G for positive n, so is its inverse.
an algebraic booby What about (ab)n = a n b n ? Does this identity hold too? Let us think
trap about this with n = 2. Now in the old notation 1
(ab)2 = a ◦ b ◦ a ◦ b
and
a 2 b 2 = a ◦ a ◦ b ◦ b.
Do these have to be the same? No, because the order of the middle two is
different and since we’re not assuming that our group is abelian we have
no right to assume that b ◦ a = a ◦ b.
Example V.4. In D 4 you can check that
ρ 21 σ20 = ρ 2 , (ρ 1 σ0 )2 = ρ 0 ,
and so ρ 21 σ20 6= (ρ 1 σ0 )2 . ♦
Let us summarize our findings.
Theorem V.5. Let G be a group, and let a ∈ G. Then a n ∈ G for all n ∈ Z.
Moreover, if m, n are integers then
(a m )n = a mn , a m a n = a m+n .
Further, if the group G is abelian, a, b ∈ G and n an integer then
(ab)n = a n b n .
Here is a crucial result that you should get used to.
Theorem V.6. Let G be a group and a, b ∈ G. Then
(ab)−1 = b −1 a −1 .
Notice that we reverse the order when taking inverse. You have prob-
ably seen this before when you did matrices at school.
P ROOF. We’re being asked to prove that b −1 a −1 is the inverse of ab. So we
want to show that
(b −1 a −1 )(ab) = 1 = (ab)(b −1 a −1 ).
Now
(b −1 a −1 )(ab) = b −1 (a −1 a)b by associativity
−1
=b 1b
= 1,
1Perhaps you think that I should write (ab)2 = (a ◦ b) ◦ (a ◦ b), but because of asso-
ciativity the bracketing does not matter. Whichever bracketing you apply to a ◦ b ◦ a ◦ b
you will get the same result.
38 V. FIRST THEOREMS

and similarly (ab)(b −1 a −1 ) = 1. 

Never write a/b unless the group is abelian. This notation is ambigu- pitfall
ous; does a/b mean b −1 a or ab −1 ? The two aren’t the same in the non-
abelian world.
Exercise V.7. Use D 4 to give counterexamples to the following:
• b −1 a = ab −1 ,
• (ab)−1 = a −1 b −1 ,
• a −1 ba = b.
Exercise V.8. Let G be a group satisfying a 2 = 1 for all a in G. Show that G
is abelian.

V.2. Additive Notation

For some groups the binary operation is ‘addition’ (whatever that means).
These include (R, +), (Z, +), (R[x], +), (R2 , +) etc. An important conven-
tion is that additive notation is only ever used for abelian groups. A mul-
tiplicative group can be abelian, such as (R∗ , ·), and can be non-abelian,
such as (D 4 , ◦).
You need to rephrase statements appropriately when using additive
notation. For example, instead of speaking of
a n = |aa{z
· · · a},
n times
you need to talk about
na = |a + a +
{z· · · + a} .
n times
−1
Instead of b write −b. We will mostly state and prove theorems in mul-
tiplicative notation, but it’s up to you to translate these into additive no-
tation for groups where the binary operation is addition. Let’s do this for
Theorem V.5. Here is the translation.
Theorem V.9. Let G be an (abelian) group with addition as the binary op-
eration, and let a ∈ G. Then na ∈ G for all n ∈ Z. Moreover, if m, n are
integers then
m(na) = (mn)a, ma + na = (m + n)a.
Further, if a, b ∈ G and n an integer then
n(a + b) = na + nb.
 CHAPTER VI

More Examples of Groups

Examples are an integral part of abstract algebra, and give it meaning

and life. They are as important as the definitions and theorems. For this
reason, I’ve crammed these notes with examples. Don’t just flick through
them saying, “yeah, yeah, that’s obvious”. Make a serious effort to study
exam tip! them, and know them for the exam. And enjoy them.

VI.1. Matrix Groups I

We saw that (M 2×2 (R), +) is a group. This in fact is not an interest-
ing group, because addition of matrices is not a very interesting opera-
tion. Multiplication of matrices is a far more interesting and natural op-
eration; as we saw, if A, B represent certain geometric operations (e.g.
scaling, reflection, rotation, etc.) then B A is the operation that one ob-
tains from doing A first then B ; if this doesn’t sound familiar look again
at Section III.4 and in particular at Example III.8. Can we obtain a group
out of (say) 2 × 2 matrices under multiplication?
To answer, let’s look back to Example IV.4. There we obtained a mul-
tiplicative group from the real numbers by removing 0. Of course we re-
moved 0 because it doesn’t have a multiplicative inverse. It will not be
enough for us to exclude the zero matrix, simply because there are non-
zero matrices that do not have an inverse—see for example III.10. What
if we exclude all non-invertible matrices; do we get a group under multi-
plication?
Define
½µ ¶ ¾
a b
GL2 (R) = : a, b, c, d ∈ R and ad − bc 6= 0 .
c d
Recall that ad − bc is the determinant of the 2 × 2-matrix ac db , and the
¡ ¢

matrix is invertible if and only if this determinant is non-zero (Theo-

rem III.11). So GL2 (R) contains all the invertible 2 × 2 matrices (with real
entries) and none of the non-invertible ones.
Theorem VI.1. GL2 (R) is group under multiplication of matrices.
We call GL2 (R) the general linear group.
P ROOF. The first thing to check is that GL2 (R) is closed under multiplica-
tion. If A and B are in GL2 (R) then AB is a 2 × 2 matrix with real entries.
Also, we know that det(AB ) = det(A) det(B ) (by Theorem III.12). Because
A and B have non-zero determinants, so does AB . So AB is in GL2 (R).
39
40 VI. MORE EXAMPLES OF GROUPS

Next we want to show associativity. But we already know that matrix

multiplication is associative thanks
¡ ¢ to Theorem III.9.
The identity matrix I 2 = 10 01 is in GL2 (R) and is the multiplicative
identity element; it satisfies AI 2 = I 2 A = A for any 2 × 2 matrix A. Finally,
we should ask if every matrix in GL2 (R) has an inverse. We cooked up
GL2 (R) so every element is invertible, but we need to make sure that the
inverse is also in GL2 (R). If A ∈ GL2 (R) then det(A) 6= 0. We know by The-
orem III.12 that det(A −1 ) 6= 0 and indeed det(A −1 ) = 1/ det(A). Moreover,
A −1 is a 2 × 2 matrix with real entries. Hence A −1 ∈ GL2 (R). 
We can define GL2 (Q) and GL2 (C) in a similar way and show that they
are groups. However, as this very important exercise shows. . .
Exercise VI.2. Show that
½µ ¶ ¾
a b
: a, b, c, d ∈ Z and ad − bc 6= 0
c d
is not a group with respect to multiplication.
It turns out that there is a natural definition for a group GL2 (Z). We’ll
return to this in Example XIV.25.

VI.2. Congruence Classes

Let m ≥ 2 be an integer. By Z/mZ we mean the set of congruence
classes modulo m. In Foundations this is denoted by Z/m and in most
algebra textbooks by Zm . Our notation is the least economical, but also Zm vs. Z/mZ
the least arbitrary. I have an excellent reason for writing Z/mZ instead of
Z/m and Zm . I want you to get used to the notation of quotient groups
which we’ll cover in Chapter XII.
If a is an integer, we shall write a for the congruence class of a modulo
m. Thus
a = {. . . , a − 3m, a − 2m, a − m, a, a + m, a + 2m, a + 3m, . . . }.
In otherwords, a consists of all integers congruent to a modulo m. From
Foundations you know that
Z/mZ = {0, 1, 2, . . . , m − 1}
and that the classes 0, 1, . . . , m − 1 are distinct, so Z/mZ consists of ex-
actly m classes. You know how addition and multiplication is defined on
Z/mZ:
a + b = a + b, a · b = ab.
Example VI.3. The addition and multiplication tables for Z/6Z are in Ta-
ble VI.1.
♦

Exercise VI.4. Write down the addition and multiplication tables for Z/4Z
and Z/5Z.
 VI.2. CONGRUENCE CLASSES 41

+ 0 1 2 3 4 5 × 0 1 2 3 4 5
0 0 1 2 3 4 5 0 0 0 0 0 0 0
1 1 2 3 4 5 0 1 0 1 2 3 4 5
2 2 3 4 5 0 1 2 0 2 4 0 2 4
3 3 4 5 0 1 2 3 0 3 0 3 0 3
4 4 5 0 1 2 3 4 0 4 2 0 4 2
5 5 0 1 2 3 4 5 0 5 4 3 2 1
TABLE VI.1. The addition and multiplication tables for Z/6Z.

Theorem VI.5. Let m be an integer satisfying m ≥ 2. Then (Z/mZ, +) is an

abelian group.
P ROOF. To show that Z/mZ a group, we want to check that Z/mZ is closed
under addition, that addition is associative, that there is an identity ele-
ment, and that every element has an additive inverse.
We defined Z/mZ to be the set of congruence classes modulo m. We
defined the sum of classes a and b to be a + b which is a congruence class
modulo m. So Z/mZ is closed under addition. Let’s prove associativity.
Note
(a + b) + c = a + b + c
= (a + b) + c
= a + (b + c) addition in Z is associative
= a +b +c
= a + (b + c).
Thus addition in Z/mZ is associative. Obviously 0 is the additive identity.
What about the additive inverse? Note that a + −a = 0 so every class has
an additive inverse 1.
Thus (Z/mZ, +) is a group. We leave the proof that it is abelian as an
easy exercise. 

1Perhaps you prefer the inverse of a where 0 ≤ a < m to be of the form b where b
also satisfies 0 ≤ b < m. In this case, if 0 < a < m, then observe that 0 < m − a < m, and
a + m − a = 0, since a + (m − a) ≡ 0 (mod m). Moreover −0 ≡ 0 (mod m), thus −0 = 0.
 CHAPTER VII

Orders and Lagrange’s Theorem

We return to using multiplicative notation. In Theorem V.5 we ob-

served that if G is a group containing an element a, then a n is also in G
for all integers n. It seems at first sight that this makes every group infi-
nite: just pick an element a and you have an infinite list of elements

. . . , a −4 , a −3 , a −2 , a −1 , 1, a, a 2 , a 3 , a 4 , a 5 , . . . .

The group D 4 is finite, so what goes wrong? Take a = ρ 1 ∈ D 4 which rep-

resents anti-clockwise rotation by 90◦ . Then a 4 = 1. Thus the seemingly
infinite list above simply becomes

. . . , 1, a, a 2 , a 3 , 1, a, a 2 , a 3 , 1, . . . .

In reality the list consists of exactly four elements 1, a, a 2 , a 3 .

VII.1. The Order of an Element

The above discussion leads us to the following definition.
Definition. The order of an element a in a group G is the smallest posi-
tive integer n such that a n = 1. If there is no such positive integer n, we
say a has infinite order.
Example VII.1. The order of ρ 1 is D 4 is 4. The order of ρ 2 is 2. The order
of ρ 0 is 1. What are the orders of the other elements? ♦
Example VII.2. In (R∗ , ·), the element 1 has order 1 and the element −1
has order 2. What is the order of 7? Is there a positive integer n such that
7n = 1? No. Thus 7 has infinite order.
What are the elements of finite order in R∗ . These are the non-zero
real numbers a such that a n = 1 for some positive integer n. You should
know that the only such real numbers are 1 and −1. So the only elements
of finite order in R∗ are 1 and −1 and all the other elements have infinite
order. ♦
Example VII.3. When you saw the equation a n = 1 in the above example,
I’m sure you immediately remembered the n-th roots of unity! The n-th
roots of unity don’t all live in R; they live in C. In fact, they live in C∗ .
For concreteness we take n = 3. You will know from Foundations that
there are three cube roots of unity. These are 1, ζ, ζ2 , where ζ = e 2πi /3 . See
Figure VII.1. Let us think of these inside the group C∗ . Then ζ and ζ2 have
43
44 VII. ORDERS AND LAGRANGE’S THEOREM

order 3. Let’s check this for ζ2 . We note

(ζ2 )1 = ζ2 , (ζ2 )2 = ζ4 = ζ · ζ3 = ζ, (ζ2 )3 = (ζ3 )2 = 12 = 1.
So the least positive integer n such that (ζ2 )n = 1 is n = 3, so ζ2 has order
3. Don’t forget that 1 has order 1. So there are three cube roots of unity.
Two have order 3 and one has order 1.
Now let us think briefly about the fourth roots of unity. These are
1, i , i 2 , i 3 . Again see Figure VII.1. Note that i 2 = −1 and i 3 = −i . Of the
four, only two have order 4 and these are i and i 3 (check). Of course, −1
has order 2 and 1 has order 1. ♦

i
ζ

1 −1 1

ζ2
−i

F IGURE VII.1. On the left, the three cube roots of unity:

here ζ = e 2πi /3 . On the right, the four fourth roots of unity.
Note that e 2πi /4 = e πi /2 = i , so the fourth roots of unity are
1, i , i 2 = −1, and i 3 = −i .

Exercise VII.4. Write down and sketch the sixth roots of unity. What are
their orders? Repeat with the eighth roots of unity.
Exercise VII.5. C∗ has plenty of elements of infinite order. Write down a
few.
Exercise VII.6. Let G = GL2 (R). Show that
µ ¶ µ ¶
0 1 1 1
A= , B=
−1 0 0 1
belong to G. Determine their orders.
Whilst reading the above examples and working out your own, the fol-
lowing observations will have dawned on you (given here in multiplica-
tive notation).
Lemma VII.7. Let G be a group and g be an element of G.
(i) g has order 1 if and only if g is the identity element.
(ii) Let m be a non-zero integer. Then g m = 1 if and only if g has
finite order d with d | m.
 VII.1. THE ORDER OF AN ELEMENT 45

P ROOF. Let G be a group. Suppose g has order 1. By definition of order,

g 1 = 1. Thus g = 1 which is the identity element of G. Conversely, the
identity element clearly has order 1. This proves (i).
Part (ii) is an ‘if and only if’ statement. Suppose that g has order d and
d | m. Then g d = 1 and m = qd where q is an integer. So g m = (g d )q = 1.
Let us prove the converse. Suppose g m = 1 where m is a non-zero integer.
Then g |m| = 1, and |m| is a positive integer. Thus g has finite order, which
we denote by d . Using division with remainder, we may write

m = qd + r, q, r ∈ Z and 0 ≤ r < d .

Now g d = 1 by definition of order, so 1 = g m = (g d )q · g r = g r . But 0 ≤ r <

d . As d is the order, it is the least positive integer such that g d = 1. So
g r = 1 is possible with 0 ≤ r < d if and only if r = 0. This happens if and
only if m = qd which is the same as d | m. 

Exercise VII.8. Let G be an abelian group. Suppose a, b are elements

of orders m and n. Let d = lcm(m, n). Show that (ab)d = 1, ensuring
that you point out where you have used the fact the G is abelian. Give
a counterexample to show that this does not have to be true if G is non-
abelian. Hint: Look at D 3 .
Now we return to our examples. We’ve looked at various multiplica-
tive groups, but what about additive groups? If (G, +) is a group where
the binary operation is addition, what is the order of an element a? Of
course, it is the smallest positive integer n such that na = 0. If there is no
such positive integer that a has infinite order.
Example VII.9. In (R, +), (Z, +), (R[x], +), (C, +), the only element of finite
order is 0, which has order 1. All other elements have infinite order.
How do we know this. Look at the equation na = 0 with a in the group
and n a positive integer. We can divide both sides by n and obtain a = 0.
♦
You’re probably wondering if in every additive group, the identity el-
ement 0 is the only one of finite order. The following example shows that
this isn’t true.
Example VII.10. Observe that in (Z/mZ, +), every element a has finite
order. Indeed, ma ≡ 0 (mod m) and so ma = 0. This does not mean that
every element has order m, since the order of a is defined to be the least
positive integer n such that na = 0. However, we do know by Lemma VII.7
that the order n is a divisor of m.
Let us look at the elements of (Z/6Z, +) and determine their orders.
We quickly find that 0 has order 1 (as usual); 1 and 5 have order 6; 2 and 4
have order 3; and 3 has order 2. ♦
Exercise VII.11. Find the orders of the elements of (Z/4Z, +) and (Z/5Z, +).
46 VII. ORDERS AND LAGRANGE’S THEOREM

VII.2. Lagrange’s Theorem—Version 1

Mathematics is unique in that supreme beauty goes hand in hand
with tremendous power. Lagrange’s Theorem is one of the loveliest exam-
ples of such a combination of qualities, and we’re almost ready to meet
it. I know you’re brimming with excitement, but please be a little patient; self-control advised
we need one more definition.
Definition. Let G be a group. The order of G is the number of elements
that G has. We denote the order of G by |G| or #G.
Theorem VII.12. (Lagrange’s Theorem—Version 1) Let G be a finite group,
and let g be an element of G. The order of g divides the order of G.
The proof of Lagrange’s Theorem will have to wait till Chapter XI.
Here’s a useful corollary.
Corollary VII.13. Let G be a finite group of order n, and let g be an ele-
ment of G. Then g n = 1.
P ROOF. Let d be the order of g . By definition of the order of an element,
g d = 1. By Lagrange’s Theorem, d divides n. Thus n = kd for some integer
k. Now
g n = (g d )k = 1k = 1,
which is what we set out to prove. 
Example VII.14. Lagrange’s Theorem applies to finite groups of which
you haven’t seen many examples yet. One example of a finite group is D 4
which has order 8. So every element of D 4 must have order dividing 8. In
fact the elements of D 4 have orders 1, 2 and 4. ♦
Example VII.15. The set {1, i , −1, −i } forms a group of order 4 under mul-
tiplication (convince yourself that this is true). Then 1 has order 1; −1 has
order 2; i and −i have order 4. This is all consistent with Lagrange’s The-
orem. ♦
 CHAPTER VIII

Subgroups

Excruciating pain It will be a long time before you come to appreciate and enjoy groups.
precedes orgasmic Abstract algebra goes from being mind-numbingly boring to being an ac-
pleasure! quired taste and then an exhilarating experience and finally—if you’re not
careful—a hopeless addiction. We’re still in the mind-numbingly boring
part of the journey; you should see this part as an initiation rite that can’t
be skipped.

VIII.1. What Were They Again?

We met subgroups in the last chapter when we discussed the group
D 4 . Let us write down the formal definition.
Definition. Let (G, ◦) be a group. Let H be a subset of G and suppose that
(H , ◦) is also a group. Then we say that H is a subgroup of G (or more
formally (H , ◦) is a subgroup of (G, ◦)).
For H to be a subgroup of G, we want H to a group with respect to the
same binary operation that makes G a group.
a trivial but Example VIII.1. R∗ is a subset of R and both are groups. But R∗ is not a
important example subgroup of R, since the operation that makes R∗ a group is multiplica-
tion and the operation that makes R a group is addition. ♦
Example VIII.2. Z is a subgroup of R (or more formally, (Z, +) is a sub-
group of (R, +)); because Z is a subset of R and both are groups with re-
spect to the same binary operation which is addition. ♦
Example VIII.3. R is a subgroup of R[x] since any real number can be
viewed as a polynomial of degree 0. ♦
Example VIII.4. (;, +) is not a subgroup of (R, +), simply because (;, +)
is not a group; a group has to be non-empty since it has to contain an
identity element. ♦

VIII.2. Criterion for a Subgroup

Theorem VIII.5. Let G be a group. A subset H of G is a subgroup if and
only if it satisfies the following three conditions
(a) 1 ∈ H ,
(b) if a, b ∈ H then ab ∈ H ,
(c) if a ∈ H then a −1 ∈ H .
Let’s delay the proof until after we’ve tried out the theorem.
47
48 VIII. SUBGROUPS

Example VIII.6. Let’s take G = R∗ and H the subset of positive real num-
bers:
H = {a ∈ R∗ : a > 0}.
Let’s show that H is a subgroup of G. First, 1 is positive, so 1 ∈ H . Hence
condition (a) is satisfied.
To check (b), suppose that a, b are in H . Thus a and b are positive,
and so their product ab is also positive. Hence ab ∈ H and we know that
(b) is satisfied.
Finally, we want to check condition (c). Suppose a is an element of H .
Then a is positive, and so a −1 is positive. Hence a −1 is also an element of
H . It follows that condition (c) is satisfied.
By Theorem VIII.5, H is a subgroup of R∗ . ♦
Example VIII.7. Let
2Z = {2a : a ∈ Z} = {. . . , −6, −4, −2, 0, 2, 4, 6, . . . }.
In other words, 2Z is the set of even integers. Now 2Z is a subset of Z,
but is it a subgroup of Z? We should check the three conditions in the
theorem, where G = Z and H = 2Z. Condition (a) is “1 ∈ H ”. What does
that mean in our context? 1 is not the number 1. The 1 in the theorem is
the identity element for the group operation on Z. The group operation
on Z is addition. The identity element is 0. As 0 is an even number (after
all 0 = 2 × 0) we have 0 ∈ 2Z. Thus condition (a) is satisfied.
Let’s move on to condition (b). This says “if a, b ∈ H then ab ∈ H ”.
Again ab doesn’t always mean the product of a and b; it is shorthand for
a ◦ b where ◦ is the binary operation on G. Here G = Z and the binary
operation on Z is +. So to check (b) what we must check is the following
“if a, b ∈ 2Z then a + b ∈ 2Z”. In words this just says “the sum of two even
integers is even”, which is true so (b) holds.
Finally we have to interpret (c) in our context. Here a −1 is the inverse
of a with respect to addition, so it just means −a. Thus to check (c) we
want to check that “if a is an even integer then −a is also even”. Again this
is true, so (c) holds.
It follows from Theorem VIII.5 that 2Z is a subgroup of Z.
By contrast, the set of odd integers
{. . . , −5, −3, −1, 1, 3, 5, . . . }
is not a subgroup of Z. For example, it does not contain the identity ele-
ment 0, so does not satisfy (a). ♦
Example VIII.8. In Subsection IV.4.1, we listed the ten subgroups of D 4 .
Go back to that list, and use Theorem VIII.5 to verify that a couple of them
are indeed subgroups. ♦
Example VIII.9. Let
V = {(a, a) : a ∈ R}.
 VIII.2. CRITERION FOR A SUBGROUP 49

In other words V is the subset of R2 where the x-coordinate equals the

y-coordinate. Thus V is the line y = x in R2 . It is geometrically obvious
that V contains the origin, which is the identity element of R2 ; that if we
add two vectors belonging to it the result also belongs to it; and that if
we multiply any vector belonging to this diagonal by −1 the result also
belongs to V . Figure VIII.1 will help you visualise this. But at this stage in
your academic career, you are expected to write a proof in symbols. Let
us do that:
First note that 0 = (0, 0) ∈ V . Secondly, suppose u ∈ V and v ∈ V . By
definition of V , u = (a, a) and v = (b, b) for some a, b ∈ R. Thus u + v =
(a + b, a + b) which again belongs to V . Finally, suppose that v ∈ V . By
definition of V , v = (a, a) for some a ∈ R. So −v = (−a, −a) which is in V .
This shows that V is a subgroup of R2 . ♦

y
V

F IGURE VIII.1. The set V = {(a, a) : a ∈ R} is the line y =

x. It contains the identity element (0, 0), is closed under
addition and negation. Therefore it is a subgroup of R2 .

Example VIII.10. This time we take W = {(a, a) : a ∈ R, a ≥ 0}. The set W

is not all the line y = x but a ‘ray’ as in Figure VIII.2. Note that W does
satisfy the first two conditions (a), (b) for being a subgroup. However, it
does not satisfy condition (c); for example, v = (1, 1) belongs to W but
−v = (−1, −1) does not. Hence W is not subgroup of R2 .
To show that W is not a subgroup, we gave a counterexample. This
means that we gave an example to show that at least one of the require-
ments in the theorem is not always satisfied. ♦
Example VIII.11. Let
V = {(a, a) : a ∈ R}, V 0 = {(−a, a) : a ∈ R}.
union of subgroups You know from Example VIII.9 that V is a subgroup of R2 (and is the line
not (always) y = x). It is just as easy to show that V 0 (which happens to be the line y =
subgroup −x) is also a subgroup of R2 . What about their union U = V ∪V 0 ? You can
check that U satisfies conditions (a) and (c) of Theorem VIII.5. However,
50 VIII. SUBGROUPS

y W

(1, 1)

(−1, −1)

F IGURE VIII.2. The ray W = {(a, a) : a ∈ R, a ≥ 0} is not

a subgroup of R2 . It contains the identity element (0, 0)
and is closed under addition. The problem is with the exis-
tence of additive inverses; e.g. (1, 1) is in W but its inverse
(−1, −1) isn’t in W .

(1, 1) and (−1, 1) are in U but their sum (0, 2) is not in U . So U does not
satisfy (b), and is therefore not a subgroup of R2 . See Figure VIII.3.
On the other hand, the intersection V ∩ V 0 = {(0, 0)} is a subgroup of
R2 . ♦
y
0
V (0, 2) V

(−1, 1) (1, 1)

F IGURE VIII.3. The lines y = x and y = −x are subgroups

of R2 . Their union is not.

Exercise VIII.12. Let G be a group and let H1 , H2 be subgroups. Show

that H1 ∩ H2 is also a subgroup of G.
Example VIII.13. Let’s take
C = {(a, a 3 ) : a ∈ R}.
Clearly C is a subset of R2 ; in fact it is the graph y = x 3 (see Figure VIII.4).
But is it a subgroup? It contains the identity element (0, 0). Moreover,
−(a, a 3 ) = (−a, (−a)3 ). So C satisfies condition (c) for subgroups. But
it doesn’t satisfy condition (b). To show this we give a counterexample.
Note that (1, 1) is in C but (1, 1) + (1, 1) = (2, 2) is not in C . ♦
 VIII.2. CRITERION FOR A SUBGROUP 51

y
C

(a, a3 )

(−a, −a3 )

F IGURE VIII.4. The set C = {(a, a 3 ) : a ∈ R} is the graph y =

x 3 . It satisfies conditions (a) and (c) for subgroups but not
condition (b).

Example VIII.14. Z2 is a subgroup of R2 . ♦

Example VIII.15. In Example VIII.9 we saw that the line y = x in R2 gives
us a subgroup. In this example we would like to think about planes in R3
and whether they give us subgroups of R3 . One way to specify a plane in
R3 is via the point-normal equation which you should’ve met at A-Level,
but which we revise now. Let Π be a plane in R3 . Let n be a vector nor-
mal to Π (by normal to Π we simply mean perpendicular to Π) as in Fig-
−−→
ure VIII.5. Choose and fix a point Q on the plane Π and let u = OQ be
the position vector of Q. Suppose now that P is any point on Π and let
−−→ −−→
x = OP be its position vector. Note that the vector QP = x − u is parallel
to the plane and so perpendicular to n. Hence n · (x − u) = 0. This is the
point-normal equation for the plane:

(VIII.17) Π : n · (x − u) = 0.

Here n is any (non-zero) vector normal to the plane, and u is the position
vector of any point on the plane.
The plane Π in (VIII.17) defines a set

VΠ = {x ∈ R3 : n · (x − u) = 0}.

This is the set of points on the plane. It is a subset of the group R3 . Is

VΠ a subgroup? Of course to be a subgroup it has to contain the identity
element of R3 which is 0. So we can choose u = 0. This doesn’t mean that
our original Q was the origin. We’re free to choose Q anywhere we like on
Π, and if Π goes through the origin then we choose it to be the origin, and
so take u = 0. With this choice, we can simplify VΠ to obtain

VΠ = {x ∈ R3 : n · x = 0}.
52 VIII. SUBGROUPS

Q x−u
P

u x

F IGURE VIII.5. The point-normal equation of a plane.

Here n is normal to the plane Π, Q is a fixed point on Π
and u is its position vector. If P is any point on Π with po-
sition vector x, then x − u is parallel to the plane, and so
n · (x − u) = 0.

Let’s check that this is indeed a subgroup of VΠ . If x1 , x2 ∈ VΠ then n·xi = 0

so
n · (x1 + x2 ) = n · x1 + n · x2 = 0 + 0 = 0.
Thus x1 + x2 ∈ VΠ . Also
n · (−x1 ) = −n · x1 = −0 = 0.
Thus −x1 ∈ VΠ . Hence VΠ is a subgroup of R3 .
Conclusion: a plane defines a subgroup of R3 if and only if it passes
through the origin. ♦
Exercise VIII.16. Which lines in R2 define a subgroup? Justify your an- an important
swer. geometric exercise
Example VIII.17. Recall that
C∗ = {α ∈ C : α 6= 0}.
Geometrically, C∗ is the whole complex plane minus the origin. We have
observed before that C∗ is a group (where the binary operation is multi-
plication of complex numbers). Let
S = {α ∈ C : |α| = 1}.
The set S is the set of all points in the complex plane with distance 1 from
the origin. Of course this is just the unit circle (the circle centred at the
origin with radius 1) as in Figure VIII.6. Let us check that S is a subgroup
of C∗ ; it is clearly a subset. Of course the unit element of C∗ is 1 and
 VIII.2. CRITERION FOR A SUBGROUP 53

i
S

1 −1 1

−i

F IGURE VIII.6. On the left, the group S which is just the

unit circle. On the right, the subgroup of the fourth roots
of unity.

|1| = 1 so 1 ∈ S, which proves (a). Suppose α, β ∈ S. Then |α| = 1 and

|β| = 1. From the properties of the absolute value 1 we have
|αβ| = |α||β| = 1.
Thus αβ ∈ S. This proves (b).
To check (c), suppose α ∈ S, so that |α| = 1. Then, again from the
properties of the absolute value,
1
|α−1 | = = 1,
|α|
so α−1 ∈ S. By Theorem VIII.5, S is indeed a subgroup of C∗ .
We shall call S the circle group. Notice that S is an infinite subgroup of
C∗ . But C∗ has plenty of finite subgroups too. An example is {1, i , −1, −i }.
This is the set of solutions to the equation x 4 = 1 (check). The solutions
to x 4 = 1 are called the fourth roots of unity. Check for yourself that
{1, i , −1, −i } is a subgroup of C∗ (and in fact a subgroup of S). Can you
find a finite subgroup of C∗ that isn’t a subgroup of S? We’ll return to
roots of unity later. ♦
Exercise VIII.18. In the following, is H a subgroup of the group G? Give
full justification. Before you start answering: You might be wondering
why I don’t specify the binary operation on G. Mathematicians generally
don’t; you’re expected to figure it out from the context 2.
(i) G = R, H = R∗ .
1At school you probably called |α| the modulus of α. Most mathematicians call |α|
the absolute value of α.
2Devil’s advocate: “Yes, I know that addition makes R into a group, and multipli-
cation doesn’t. But are there really no other binary operations on R that make it into a
group?”
In maths it is good to play the rôle of the devil’s advocate, but not to the extent of
renouncing good taste and common sense. Yes, there are binary operations other than
54 VIII. SUBGROUPS

(ii) G = R∗ , H = {1, −1}.

(iii) G = C, H = 2Z.
(iv) G = C, H = {a + ai : a ∈ R}.
(v) G = C∗ , H = {α ∈ C∗ : α3 = 1}.
(vi) G = Z, H = Z/2Z.
(vii) G = R[x], H = Z[x].
(viii) G = R[x], H = { f ∈ R[x] : f (0) = 0}.
(ix) G = R[x], H = { f ∈ R[x] : f (0) = 1}.
(x) G = Z/10Z, H = {0, 5}.
1
Exercise VIII.19. Show that the subgroups of Z/4Z are {0}, {0, 2} and
Z/4Z.
Exercise VIII.20. Show that the only subgroups of Z/3Z are {0} and Z/3Z.
Exercise VIII.21. Let
D = {α ∈ C∗ : |α| ≤ 1}.
Sketch D. Show that D is not a subgroup of C∗ .
Exercise VIII.22. Let r be a positive real number. Let

Sr = {α ∈ C∗ : |α| = r }.

What does Sr represent geometrically? For what values of r will Sr be a

subgroup of C∗ ?

P ROOF OF T HEOREM VIII.5. The theorem has an “if and only if” state-
ment. It usual when proving an “if and only if” statement to break it up if and only if
into an “if” part, and an “only if” part, and prove each part separately.
This is what we will do here. The “if” part says: “if H is a subset of G that
satisfies (a),(b),(c) then it is a subgroup of G”. The “only if” part says: “if
H is a subgroup of G then H satisfies (a), (b), (c)”.
Let us do the “if” part of the proof first. We have a group G and a
subset H of G. All we have been told is that H satisfies conditions (a), (b),
(c) in the statement of the theorem. We want to show that H is a group,
where the binary operation on H is the same as the binary operation on
G. This means that we have to show that H satisfies properties (i), (ii),
(iii), (iv) in the definition of a group.
Property (i) is ‘closure’: we want that if a, b ∈ H then ab ∈ H . But this
is what (b) is saying. So (i) is satisfied.
Property (ii) is associativity. We want to show that for all a, b, c ∈ H ,
we have (ab)c = a(bc). But if a, b, c are elements of H then they are also

addition that make the set of real numbers into a group. But if I wanted anything other
than the usual or obvious operation I’d have told you so.
1When answering a maths question, you should always be careful about what is
being asked. Here you’re being asked to show two things. The first is that the three listed
sets are indeed subgroups. The second is that there aren’t any other subgroups.
 VIII.3. ROOTS OF UNITY 55

elements of G. We know that associativity holds in G: (ab)c = a(bc). So

(ii) holds 1.
Property (iii) is the existence of the identity element in H . But (a)
tells us that 1 ∈ H . This 1 is the identity element of G and so satisfies
a1 = 1a = a for all a in G. Since every a in H is also in G we have that
a1 = 1a = a for all a in H so 1 is the identity element of H , and so (iii)
holds.
Finally, property (iv) asserts the existence of an inverse for every a ∈
H . This follows from (c). Hence H is a group contained in G and so a
subgroup. We have now finished the proof of the “if” part.
Next we do the “only if”part of the proof. But I’m already bored typing
this proof, so I’ll leave this part as a (mandatory) exercise. 

VIII.3. Roots of Unity

Let n be a positive integer. Let ζ = e 2πi /n . The n-th roots of unity are
the solutions in C to the equation x n = 1. Recall that there are exactly n
of them:
1, ζ, ζ2 , . . . , ζn−1 .
See Figure VII.1 for the roots of unity when n = 3 and n = 4 and note how
they’re distributed on the unit circle. Write
Un = {1, ζ, ζ2 , . . . , ζn−1 }.
That is, Un is the set of n-th roots of unity.
Lemma VIII.23. Un is a subgroup of C∗ of order n.

P ROOF. Clearly Un is a subset of C∗ containing 1. Suppose a, b ∈ Un . We

want to check that ab ∈ Un . But since a n = b n = 1 we know that (ab)n =
a n b n = 1. So ab is also an n-th root of unity and so ab ∈ Un . Likewise,
(a −1 )n = (a n )−1 = 1. So a −1 is an n-th root of unity and so a −1 ∈ Un . Thus
Un is indeed a subgroup of C∗ . Since it has n elements, it has order n. 

L Notation Warning. The notation Un is not standard. Why do I point this

out? You must always be careful with notation: do other people under-
stand you? If you write C∗ then this is standard notation and every math-
ematician will know what you mean. If you write Un , others (e.g. your
tutor and supervisor) will not know what you mean. They will of course
know that the n-th roots of unity are a subgroup of C∗ , but they will not
know that you’re denoting this subgroup by Un . If you write Un , even in
your homework, then you have to say what it is.
Exercise VIII.24. Is U2 ∪U3 a subgroup of C∗ ?
1There is a subtle point here that is camouflaged by our notation, and that is that
the binary operation we’re using on H is precisely the same one as the binary operation
we’re using on G. If it was different we would have no right to say: because associativity
holds in G it holds in H .
56 VIII. SUBGROUPS

VIII.4. Matrix Groups II

In Section VI.1 you met the general linear group
GL2 (R) = {A ∈ M 2×2 (R) : det(A) 6= 0}.
This is group where the operation is multiplication of matrices. In this
section we’ll meet some subgroups of it.
Exercise VIII.25. Let
SL2 (R) = {A ∈ M 2×2 (R) : det(A) = 1}.
Show that SL2 (R) is a group 1 (with respect to multiplication). This is
known as the special linear group 2.
Exercise VIII.26. Show that
{A ∈ M 2×2 (Z) : det(A) 6= 0}
is not a group under multiplication. Let
SL2 (Z) = {A ∈ M 2×2 (Z) : det(A) = 1}.
Show that SL2 (Z) is a group. This is known as the modular group 3.
Now is a good time to revise Section III.7 on rotation matrices. Recall
that the matrix
cos θ − sin θ
µ ¶
Rθ =
sin θ cos θ
represents anticlockwise rotation about the origin through an angle θ. It
is geometrically clear that if compose two rotations about the origin we
obtain a rotation about the origin. So it is natural to expect that rotations
form a subgroup of GL2 (R), and indeed this is the case. We define
SO2 (R) = {R θ : θ ∈ R}.
This is called the special orthogonal group.
Theorem VIII.27. SO2 (R) is a subgroup of GL2 (R).
P ROOF. First we have to check that SO2 (R) is a subset of GL2 (R). In other
words, we want to check that every matrix R θ has non-zero determinant.
Note det(R θ ) = cos2 θ + sin2 θ = 1. Hence SO2 (R) is contained in GL2 (R).
Also 4 I 2 = R 0 , so SO2 (R) contains the identity element of GL2 (R).
Next we have to show that SO2 (R) is closed under multiplication. Con-
sider two elements of SO2 (R) and call them R θ and R φ . Now R θ and R φ
represent anticlockwise rotation about the origin through angles θ and
1Recall, the easiest way to show that a set is a group is to show that it is a subgroup
of a something you already know to be a group.
2If you’ve done Exercise III.13 then you’ll see that SL (R) consists of the matrices
2
that preserve area and orientation.
3The modular group is probably the most interesting group in mathematics. Google
it!
4In geometric terms, both I and R mean ‘do nothing’, so they must be equal.
2 0
 VIII.5. DIFFERENTIAL EQUATIONS 57

φ. Thus R θ R φ represents the combined effect of rotations through angles

φ then θ. Clearly, from this geometric reasoning R θ R φ = R θ+φ , but let’s
check this algebraically:

cos θ − sin θ cos φ − sin φ

µ ¶µ ¶
Rθ Rφ =
sin θ cos θ sin φ cos φ
cos θ cos φ − sin θ sin φ − cos θ sin φ − cos φ sin θ
µ ¶
=
cos θ sin φ + cos φ sin θ cos θ cos φ − sin θ sin φ
cos(θ + φ) − sin(θ + φ)
µ ¶
=
sin(θ + φ) cos(θ + φ)
= R θ+φ .

Thus SO2 (R) is closed under multiplication.

Finally we must check that the inverse of every matrix in SO2 (R) is
again in SO2 (R). Geometrically, it’s easy to see that the inverse of R θ is
R −θ ; I’ll leave it to you to check this algebraically. This completes the
proof. 

Remark. It is clear (at least geometrically) that R θ R φ = R φ R θ . Thus SO2 (R)

is an abelian subgroup of the non-abelian group GL2 (R). We saw this
phenomenon before: the group D 4 is non-abelian, but its subgroup of
rotations is abelian.

VIII.5. Differential Equations

Let C be the set of infinitely differentiable real functions. This proba-
bly sounds scary, but to reassure you I’ll just point out that C contains all
polynomials, as well as sin t , cos t , e t , e −t . It is a fact that C is an additive
group. Don’t worry about the proof; it depends on properties of differen-
tiability that you’ll see eventually in analysis. Addition in C is done in a
common sense way. For example, if f (t ) = t 2 + sin(t ) and g (t ) = 2t 2 − e t
then f (t ) + g (t ) = 3t 2 + sin(t ) − e t . The identity element is 0.
Let’s dive straight into an example. We define the following subset
dx
½ ¾
H = x(t ) ∈ C : t − 2x = 0 .
dt
In other words, H is the set of infinitely differentiable functions x(t ) that
satisfy the differential equation
dx
(VIII.18) t − 2x = 0.
dt
The function x(t ) = 0 (which is the identity element of C ) clearly satisfies
(VIII.18) and so belongs to H . Suppose x 1 (t ) and x 2 (t ) are in H . Thus
d x1 d x2
t − 2x 1 = 0, t − 2x 2 = 0.
dt dt
58 VIII. SUBGROUPS

Let x(t ) = x 1 (t ) + x 2 (t ). By the properties of differentiation,

d x d x1 d x2
= + .
dt dt dt
Thus
dx d x1 d x2
µ ¶
t − 2x = t + − 2(x 1 + x 2 )
dt dt dt
d x1 d x2
=t − 2x 1 + t − 2x 2
dt dt
= 0.
Therefore x(t ) ∈ H . Similarly, using the properties of differentiation, you
can show that if x 1 (t ) ∈ H then −x 1 (t ) ∈ H (easy exercise). So H is a sub-
group of C .
Note that we didn’t have to solve the differential equation to know
that its set of solutions is a group; we merely used the properties of dif-
ferentiation. But in fact it is easy to solve this particular equation using
separation of variables. If you do that (try it) you’ll find that
H = {at 2 : a ∈ R}.
Now check again that H forms an additive group.
Exercise VIII.28. Which of the following differential equations define sub-
groups of C ?
dx
(i) t − 2x = t 3 .
dt
d2 x dx
(ii) −5 + 6x = 0.
dt2 dt
dx
(iii) − x 2 = 0.
dt

VIII.6. Non-Trivial and Proper Subgroups

It’s very easy for you to prove the following proposition.
Proposition VIII.29. Let G be a group. Then G and {1} are subgroups.
Here, of course, {1} is the subset containing the identity element of G.
We call {1} the trivial subgroup of G; any other subgroup is called non-
trivial. A subgroup of G that is not equal to G is called proper. The sub-
groups {1} and G are boring, since they’re always there. The interesting
subgroups are the proper non-trivial subgroups.
Example VIII.30. The trivial subgroup of Z is {0}. Examples of a non-
trivial subgroups are Z and 2Z. The subgroup 2Z is proper and non-
trivial. ♦
Example VIII.31. Consider the group U4 which is the group of fourth
roots of unity. Thus U4 = {1, i , −1, −i }; of course the binary operation is
 VIII.7. LAGRANGE’S THEOREM—VERSION 2 59

multiplication. The trivial subgroup is {1}. We note that U2 = {1, −1} is a

non-trivial proper subgroup. Are there any others? Suppose H is another
non-trivial proper subgroup of U4 . Then 1 ∈ H , as subgroups always con-
tain the identity element. Since H is non-trivial, and H 6= {1, −1}, it must
contain either i or −i . Suppose H contains i . Then H contains i 2 = −1
and i 3 = −i . Therefore H = U4 , which contradicts the assumption that H
is proper. Similarly if H contains −i then H = U4 (check). Therefore the
only non-trivial proper subgroup of U4 is U2 = {1, −1}. ♦
Exercise VIII.32. For what values of m does Z/mZ have non-trivial proper
subgroups? Try out a few examples and see if you can make a conjecture.
Can you prove your conjecture?

VIII.7. Lagrange’s Theorem—Version 2

Here is another version of Lagrange’s Theorem. The relation between
this version and the earlier one (Theorem VII.12) will be explained once
we have studied cyclic groups.
Theorem VIII.33. (Lagrange’s Theorem—Version 2) Let G be a finite group,
and H a subgroup of G. Then the order of H divides the order of G.
Example VIII.34. We saw in Example VIII.31 that U4 , the group of 4-th
roots of unity, contains U2 , the group of square-roots of unity. Now U2
has order 2, U4 has order 4. Lagrange’s Theorem tells that the order of U2
must divide the order of U4 which is correct. ♦
Example VIII.35. Recall that D 4 has order 8. In Figure IV.3 we listed the
ten subgroups of D 4 . These have orders 1, 2, 4 and 8. This is consistent
with Lagrange’s Theorem. ♦
Exercise VIII.36. Let G be a group, and suppose the order of G is p where
p is a prime. Show that the only subgroups of G are {1} and G.

Still hurting?
 CHAPTER IX

Cyclic Groups and Cyclic Subgroups

Cyclic groups are the simplest groups to understand.

Theorem IX.1. Let G be a group, and let g be an element of G. Write 〈g 〉
for the set

〈g 〉 = {g n : n ∈ Z} = {. . . , g −2 , g −1 , 1, g , g 2 , g 3 , . . . }.

Then 〈g 〉 is a subgroup of G.

P ROOF. This is very easy to prove using Theorem VIII.5. Have a go! 

Definition. We call 〈g 〉 the cyclic subgroup generated by g . If G = 〈g 〉 then

we call G a cyclic group, and we say that g is a generator of G.
Example IX.2. As roots of unity are fresh in your mind, let’s start with
them. The group of n-th roots of unity Un is cyclic, since every element
is a power of ζ = e 2πi /n ; indeed the elements of Un are precisely

ζ0 = 1, ζ, ζ2 , . . . , ζn−1 .

Thus Un = 〈ζ〉 and ζ is a generator.

Let’s consider U6 , and calculate the cyclic subgroup generated by each
element. Write ζ = e 2πi /6 . Note that ζ6 = 1. Consider for example h = ζ2 .
The powers of h are 1, h, h 2 . Indeed, note that h 3 = ζ6 = 1. Thus

h 4 = h, h 5 = h 2 , h 6 = 1, h 7 = h, . . . .

What about h −1 . We know that h 3 = 1; multiplying both sides by h −1 we

deduce that h −1 = h 2 . Thus

h −2 = h, h −3 = 1, h −4 = h 2 , h −5 = h, . . . .

Thus the distinct powers of h are 1, h, h 2 , which are 1, ζ2 , ζ4 . We can’t write

all the elements of U6 as powers of h; therefore h is not a generator of U6 .
However, let us consider g = ζ5 . We can write the powers of g and
simplify them using the fact that ζ6 = 1. For example,

g 2 = ζ10 = ζ6 ζ4 = ζ4 .

We find that 1, g , g 2 , g 3 , g 4 , g 5 are respectively, 1, ζ5 , ζ4 , ζ3 , ζ2 , ζ. Since ev-

ery element of U6 is a power of g = ζ5 , we see that g is also a generator of
U6 . Table IX.1 lists the elements of U6 and the subgroups they generate.
♦
61
62 IX. CYCLIC GROUPS AND CYCLIC SUBGROUPS

g 〈g 〉
1 {1}
ζ {1, ζ, ζ2 , ζ3 , ζ4 , ζ5 }
ζ2 {1, ζ2 , ζ4 }
ζ3 {1, ζ3 }
ζ4 {1, ζ2 , ζ4 }
ζ5 {1, ζ, ζ2 , ζ3 , ζ4 , ζ5 }
TABLE IX.1. The six elements of U6 and the cyclic sub-
groups they generate.

Example IX.3. For each element of the group Z/mZ, we write down the
cyclic group it generates. Note that since Z/mZ is an additive group, the
subgroup generated by g is 〈g 〉 = {ng : n ∈ Z}. That is, it is the set of
multiples of g rather than the set of powers of g . See Table IX.2. ♦

a 〈a〉
0 {0}
1 {0, 1, 2, 3, 4, 5}
2 {0, 2, 4}
3 {0, 3}
4 {0, 2, 4}
5 {0, 1, 2, 3, 4, 5}
TABLE IX.2. The six elements of Z/6Z and the cyclic sub-
groups that they generate.

Example IX.4. Recall the group D 4 of the symmetries of the square. It

has 8 elements. It’s easy to write down the subgroup generated by each
element (see Section IV.4 to remind yourself of the notation):
g 〈g 〉
1 {1}
ρ 1 {1, ρ 1 , ρ 2 , ρ 3 }
ρ 2 {1, ρ 2 }
ρ 3 {1, ρ 1 , ρ 2 , ρ 3 }
σ0 {1, σ0 }
σ1 {1, σ1 }
σ2 {1, σ2 }
σ3 {1, σ3 }
 IX. CYCLIC GROUPS AND CYCLIC SUBGROUPS 63

None of the elements of D 4 generates it. We see that D 4 is not a cyclic

group. ♦
Theorem IX.5. Cyclic groups are abelian.
P ROOF. Let G be a cyclic group generated by g . Let a, b be elements of
G. We want to show that ab = ba. Now, a = g m and b = g n for some
integers m and n. So, ab = g m g n = g m+n and ba = g n g m = g n+m . But
m + n = n + m (addition of integers is commutative). So ab = ba. 

Whilst working through the above examples, you will have noticed a
pattern about 〈g 〉, which we state in the following theorem.
Theorem IX.6. Let G be a group and let g be an element of finite order n.
Then
〈g 〉 = {1, g , g 2 , . . . , g n−1 }.
In particular, the order of the subgroup 〈g 〉 is equal to the order of g .
P ROOF. Observe that 〈g 〉 is a set, and {1, g , . . . , g n−1 } is a set. We want to
a fundamental show that these sets are the same. Whenever you have two sets, A and
principle B , and you want to prove that they’re equal, one way to do this is to show
that every element of A belongs to B and every element of B belongs to A.
You will see this principle again and again throughout your undergradu-
ate career.
Let’s apply this principle in our situation. By definition,
〈g 〉 = {g n : n ∈ Z} = {. . . , g −2 , g −1 , 1, g , g 2 , g 3 , . . . }.
That is 〈g 〉 is the set of all powers of g . It is obvious that every element
of {1, g , . . . , g n−1 } belongs to 〈g 〉. What about the other way round. Sup-
pose that h is an element of 〈g 〉. We want to show that h is an element
of {1, g , . . . , g n−1 }. We can write h = g m where m is an integer (positive or
negative). We want to show that h = g r where r is one of 0, 1, 2, . . . , n − 1.
The division For this we will use the division algorithm which you met in Foundations.
algorithm is one of We can write
the most powerful m = qn + r, q, r ∈ Z, 0 ≤ r < n.
ideas in algebra. Here we simply divided m by n; the integers q, r are respectively the quo-
tient and the remainder. Thus
h = g m = g qn+r = (g n )q · g r .
However, g n = 1 since g has order n. So h = g r . Since 0 ≤ r < n, we
see that r is one of 0, 1, . . . , n − 1. Therefore h is in {1, g , . . . , g n−1 }. By our
principle, we see that 〈g 〉 = {1, g , . . . , g n−1 }. 
Exercise IX.7. In each of the following groups G, write down the cyclic
subgroup generated by g .
(a) G = S, g = exp(2πi /7).
(b) G = Z/12Z, g = 8.
¡ 0 1¢
(c) G = GL2 (R), g = −1 0 .
64 IX. CYCLIC GROUPS AND CYCLIC SUBGROUPS

Exercise IX.8. Which of the following groups G are cyclic? Justify your
answer for each, and if G is cyclic then write down a generator.

(a) G = kZ (where k is a non-zero integer).

(b) G = Z/mZ (where m is a positive integer).
(c) D 3 .
Exercise IX.9. In this exercise, you will show using contradiction that R∗
is not cyclic. Suppose that it is cyclic and let g ∈ R∗ be a generator. Then
R∗ = 〈g 〉. In particular, |g |1/2 ∈ R∗ and so |g |1/2 = g m for some integer
m. Show that the only solutions to this equation are g = ±1. Where’s the
contradiction?
Exercise IX.10. In this exercise you’ll show that Q is not cyclic. Let a, b
be integers with b 6= 0. Let p be a prime that does not divide b. Show that
1/p cannot be written in the form na/b with n an integer. Deduce that Q
is not cyclic.
Exercise IX.11. Show that S is not cyclic.

IX.1. Lagrange Revisited

You saw two versions of Lagrange’s Theorem:
Theorem IX.12. (Lagrange’s Theorem—Version 1) Let G be a finite group,
and let g be an element of G. The order g divides the order of G.
Theorem IX.13. (Lagrange’s Theorem—Version 2) Let G be a finite group,
and H a subgroup of G. Then the order of H divides the order of G.
In fact Version 2 implies Version 1. Let us prove that.
Proposition IX.14. Version 2 of Lagrange’s Theorem implies Version 1 of
Lagrange’s Theorem.

P ROOF. We assume Version 2 and deduce Version 1. Let G be a finite

group and g an element of G. Suppose g has order n. By Theorem IX.6,
the cyclic subgroup generated by g , denoted 〈g 〉, also has order n. By
Version 2, the order of the subgroup 〈g 〉 divides the order of G. Hence n
divides the order of G, which is what we wanted to prove. 

This doesn’t mean that we’ve proved Version 1 of Lagrange’s Theorem.

It does mean that once we prove Version 2, then we will have also proved
Version 1.
Exercise IX.15. Let G be a group of order p, where p is a prime number.
Let H be a subgroup. Show that H must either equal G or the trivial sub-
group {1}. Deduce that if g ∈ G is not the identity element, then G = 〈g 〉.
 IX.2. SUBGROUPS OF Z 65

IX.2. Subgroups of Z
Trust me, I’m a I’m feeling particularly inarticulate at the moment, so I can’t explain
doctor. why subgroups of Z are important. But nevertheless they are important
and so we’ll do them.
The first thing to note about Z is that it is cyclic. Does that mean
that all elements of Z are powers of some element. No, because it is an
additive group. If G is an additive group, and g is an element of G then
〈g 〉 = {ng : n ∈ Z} = {. . . , −2g , −g , 0, g , 2g , 3g , . . . }.
Thus Z = 〈1〉 and so it is cyclic. In fact, it is infinite and cyclic, unlike for
example, Un .
Lemma IX.16. Let k be an integer. Write
kZ = {ka : a ∈ Z}.
Then kZ is a subgroup of Z.
P ROOF. You can prove this in a similar way to Example VIII.7. However,
it is quicker to note that kZ = 〈k〉, and so is a subgroup by Theorem IX.1.

Note that 0Z = {0} has only the identity element. Also
(−k)Z = {. . . , −2(−k), −(−k), 0, −k, 2(−k), . . . }
= {. . . , 2k, k, 0, −k, −2k, . . . }
= {. . . , −2k, −k, 0, k, 2k, . . . }
= kZ
because the order of elements in a set does not matter. In other words,
−Z = Z, −2Z = 2Z, −3Z = 3Z, . . . .
So we have an infinite list of subgroups
{0}, Z, 2Z, 3Z, 4Z, . . .
and we want to know if they’re all the subgroups of Z. The following the-
orem tells us that they are.
Theorem IX.17. Any subgroup of Z has the form kZ for some non-negative 1
integer k. In particular, all subgroups of Z are cyclic.
P ROOF. Let H be a subgroup of Z. We want to show that there is a non-
negative integer k such that H = kZ. We divide the proof into two cases.
The first case is when H is the subgroup {0}. Then H = 0Z and we’ve done
what we wanted.
So let’s look at the second case where H has non-zero elements. If a is
a non-zero element of H then because H is a(n additive) group, −a is also
a non-zero element of H but it has a different sign. So we know for sure
1The non-negative integers are 0, 1, 2, 3, . . . .
66 IX. CYCLIC GROUPS AND CYCLIC SUBGROUPS

that H has some positive elements. Let k be the smallest positive element
of H . We will prove that H = kZ.
Whenever you have two sets, A and B , and you want to prove that Is this familiar?
they’re equal, one way to do it is to show that every element of A belongs
to B and every element of B belongs to A.
As k is in H , we know by Theorem V.9 that all the multiples of k be-
long to H . Thus every element of kZ belongs to H . We must show the
converse: every element of H is a multiple of k.
Let a be an element of H . By the division algorithm which you have
met in Foundations, we can write
a = qk + r, q, r are integers and 0 ≤ r < k.
To remind: here q is the quotient of dividing a by k and r is the remainder.
Now a is in H ; qk is in H because it is a multiple of k ∈ H . So r = a − qk punchline
is also in H . But 0 ≤ r < k, and k is the smallest positive element of H . If
r > 0 then it would be an even smaller positive element of H giving us a
contradiction. So r = 0. Hence a = qk is a multiple of k.
Thus we’ve also shown that every element of H is a multiple of k and
so belongs to kZ. Hence H = kZ, as required. 
Exercise IX.18. The subgroups of Z2 are harder to describe. Write down
a few.
Thrilled? Aren’t you glad you trusted me?
 CHAPTER X

Isomorphisms

You must’ve noticed that there’s a lot in common between the group
of m-th roots of unity Um , and the group Z/mZ. If not, take another look
Tables IX.1 and IX.2. In fact the groups Um and Z/mZ are isomorphic.
What does this mean?
Definition. Let (G, ◦) and (H , ∗) be groups. We say that the function φ :
G → H is an isomorphism if it is a bijection and it satisfies
φ(g 1 ◦ g 2 ) = φ(g 1 ) ∗ φ(g 2 )
for all g 1 , g 2 in G. In this case we say that (G, ◦) and (H , ∗) are isomorphic.
Isomorphic groups may look different, but in essence are the same.
An isomorphism is a way of relabeling the elements of one group to ob-
tain another group, as the following examples will make clear.
Example X.1. Define φ : Z/mZ → Um by the simple rule
φ(a) = ζa , a = 0, 1, . . . , m − 1.
Then φ is a bijection and satisfies the magical property

φ(a + b) = φ(a + b) = ζa+b = ζa · ζb = φ(a)φ(b).

So φ is an isomorphism. ♦
Example X.2. Recall that the matrix
cos θ − sin θ
µ ¶
Rθ =
sin θ cos θ
represents anticlockwise rotation around the origin through an angle θ.
The identity
R θ+φ = R θ R φ .
turns addition into multiplication, and so it should remind you of the
identity e θ+φ = e θ e φ . In fact, a more accurate analogy is identity

e i (θ+φ) = e i θ e i φ .
The reason is because multiplying a complex number by e i θ rotates it
about the origin anticlockwise through the angle θ (prove this using the
exponential form for complex numbers).
Now that you know that R θ and e i θ are analogues, you will expect that
the groups SO2 (R) and S are isomorphic. Recall that SO2 (R) is the special
67
68 X. ISOMORPHISMS

orthogonal group (Theorem VIII.27) defined by

SO2 (R) = {R θ : θ ∈ R},
and S is the circle group (page 53) given by
S = {α ∈ C : |α| = 1} = {e i θ : θ ∈ R}.
You can satisfy yourself that the map
φ : SO2 (R) → S, φ(R θ ) = e i θ
is an isomorphism. You should have no trouble guessing what the matrix
analogues of the n-th roots of unity are. If we let
µ ¶
cos (2π/n) − sin (2π/n)
Z = R 2π/n = ,
sin (2π/n) cos (2π/n)
then I 2 , Z , . . . , Z n−1 all satisfy the relationship A n = I 2 . ♦
Exercise X.3. Let Z = R 2π/6 . Show that {1, Z , . . . , Z 5 } is a subgroup of
SO2 (R). Write down the orders of its elements and check that they are
consistent with Lagrange’s Theorem.
Exercise X.4. Suppose groups G and H are isomorphic. Show that G and
H have the same order. Show that G is abelian if and only if H is abelian.
Show that G is cylic if and only if H is cyclic.

Tragically, the powers that be (who shall remain nameless) decided

that this Introduction to Abstract Algebra should be a half-module, and
so we won’t have the time to explore the manifold pleasures of isomor-
phisms.
“SAMIR, WHY HAVE THEY DENIED US THE CATS TO Indignant?
EXPERIENCE THESE PLEASURES ? D ON ’ T THEY LOVE Paranoid? Seething
US LIKE YOU DO ? D O THEY WANT US TO FAIL AND with self-righteous
TURN TO THE BOTTLE ?” rage?
Don’t be a drama queen—there must be a perfectly innocent explanation.
 CHAPTER XI

Cosets

Cosets are what we get when we ‘shift’ a subgroup by the elements of

the group.
Definition. Let G be a group and H a subgroup. Let g be an element of
G. We call the set
g H = {g h : h ∈ H }
a left coset of H in G and the set
H g = {hg : h ∈ H }
a right coset of H in G.
Example XI.1. Let’s take G to be D 4 and R the subgroup made up of ro-
tations:
R = {1, ρ 1 , ρ 2 , ρ 3 }.
Revisit Section IV.4 to remind yourself of the notation. Let’s compute σ1 R.
By definition,
σ1 R = {σ1 · 1, σ1 ρ 1 , σ1 ρ 2 , σ1 ρ 3 }
= {σ1 , σ0 , σ3 , σ2 }
= {σ0 , σ1 , σ2 , σ3 }.
Let’s try another coset.
ρ 2 R = {ρ 2 · 1, ρ 2 ρ 1 , ρ 2 ρ 2 , ρ 2 ρ 3 }
= {ρ 2 , ρ 3 , 1, ρ 1 }
= {1, ρ 1 , ρ 2 , ρ 3 }.
We see that ρ 2 R is equal to R, and σ1 R isn’t equal to R. In fact, σ1 R isn’t
even a subgroup of D 4 ; why? You can carry on computing all eight left
cosets, and you’ll find
1 · R = ρ 1 R = ρ 2 R = ρ 3 R = {1, ρ 1 , ρ 2 , ρ 3 }
and
σ0 R = σ1 R = σ2 R = σ3 R = {σ0 , σ1 , σ2 , σ3 }.
♦
Exercise XI.2. Recall that H = {1, σ2 } is also a subgroup of D 4 . Compute
its left cosets. Check that σ1 H 6= H σ1 .
69
70 XI. COSETS

Of course, for an additive group G, a subgroup H , a left coset would

be of the form
g + H = {g + h : h ∈ H }
for some g in G.
Example XI.3. Z is an additive group. The set of even integers 2Z is a
subgroup. What are its cosets? Let’s compute a few:
0 + 2Z = {. . . , 0 + (−4), 0 + (−2), 0 + 0, 0 + 2, 0 + 4, . . . } = {. . . , −4, −2, 0, 2, 4, . . . };
1 + 2Z = {. . . , 1 + (−4), 1 + (−2), 1 + 0, 1 + 2, 1 + 4, . . . } = {. . . , −3, −1, 1, 3, 5, . . . };
2 + 2Z = {. . . , 2 + (−4), 2 + (−2), 2 + 0, 2 + 2, 2 + 4, . . . } = {. . . , −4, −2, 0, 2, 4, . . . };
3 + 2Z = {. . . , 3 + (−4), 3 + (−2), 3 + 0, 3 + 2, 3 + 4, . . . } = {. . . , −3, −1, 1, 3, 5, . . . }.
You’ll quickly discover that
· · · = −4 + 2Z = −2 + 2Z = 2Z = 2 + 2Z = 4 + 2Z = . . .
and
· · · = −3 + 2Z = −1 + 2Z = 1 + 2Z = 3 + 2Z = . . . .
So 2Z has two cosets in Z, which happen to be 2Z itself, and 1 + 2Z which
is the set of odd integers. ♦
Exercise XI.4. You know that Z2 is a group. Let
2Z2 = {(2a, 2b) : a, b ∈ Z}.
In otherwords, 2Z2 is the set of vectors in Z2 with both coordinates even.
Check that 2Z2 is a subgroup of Z2 , having four cosets. What are they?
Exercise XI.5. Let R+ be the subset of R∗ consisting of the positive num-
bers. Show that R+ is a subgroup and that it has exactly two cosets in
R∗ .

XI.1. Geometric Examples

Long ago (page 1) I told you:
You should get used to thinking geometrically, and to draw-
ing pictures. The true meaning of most mathematical con-
cepts is geometric. If you spend all your time manipulating
symbols (i.e. doing algebra) without understanding the re-
lation to the geometric meaning, then you will have very
little in terms of mathematical insight.
No doubt you have taken my advice on board and so there is no need for
me to repeat it.
Example XI.6. You’ll recall the circle group S which is the subgroup of
C∗ consisting of all elements of absolute value 1; see Example VIII.17 if
you need to refresh your memory. Let’s study the cosets of S in C∗ . Of
course C∗ is abelian, and so there is no distinction between left and right
cosets; they’re the same. A coset of S in C∗ has the form αS where α is in
 XI.1. GEOMETRIC EXAMPLES 71

C∗ (i.e. α is a non-zero complex number). As such, we can write α = r e i θ ,

where r is positive (it is the absolute value of α), and θ is the argument of
α. Consider e i θ S. Multiplying any complex number by e i θ simply rotates
anticlockwise through angle θ about the origin. So e i θ S = S. Now αS =
r S. What does multiplying by r do? It scales the circle S by a factor of
r . Two different positive real numbers r 1 6= r 2 will give different cosets
r 1 S 6= r 2 S, since the first has radius r 1 and the second has radius r 2 . See
Figure XI.1.

1.5S

0.5S
1

F IGURE XI.1. S and its cosets 0.5S and 1.5S in C∗ .

So S has as many cosets in C∗ as there are positive real numbers.

Summary: S is the circle centred at the origin of radius 1, and its cosets
in C∗ are the circles centred at the origin (of positive radius). ♦
Example XI.7. In Exercise VIII.16 I asked you the following question: which
lines in R2 define a subgroup? Let’s go back to this question and answer
it again, and this time for lines that define a subgroup we want to deter-
mine the cosets too.
One convenient way of a specifying a line L in R2 is as follows. Let Q
be a point on L, with position vector w. Let v be a vector parallel to L.
Then L has the parametric form
L : x = w + t v.
This is a (slightly clumsy) school way of saying things. What it means
is that the points with position vector w + t v are on the line, where t is
any ‘scalar’ (i.e. real number). A much better way is to just write L in set
notation:
L = {w + t v : t ∈ R}.
72 XI. COSETS

Now L is a subset of R2 and we want to know if it defines a subgroup. Of

course, if L does not pass through the origin, then it does not contain the
identity element, and so cannot be a subgroup. So, let’s suppose L passes
through the origin. The point Q was any point on the line; we will choose
Q to be the origin, and so its position vector is w = (0, 0). Now we have
L = {t v : t ∈ R}.
Is L a subgroup of R2 ? It is straightforward to ‘see’ geometrically that if we
add two vectors in L then the sum is in L. Let’s check that algebraically. If
v1 and v2 are in L then they have the form v1 = t 1 v and v2 = t 2 v. So
v1 + v2 = (t 1 + t 2 )v
which in L. Also, −v1 = (−t 1 )v1 is in L. Hence L is a subgroup of R2 .
What are the cosets of L in R2 ? They have the form
w + L = {w + t v : t ∈ R}
where w is a vector in R2 . This is the line with parametric form w + t v.
Note that both L and its coset w + L are parallel to v. See Figure XI.2

w+L

y L

v
w

F IGURE XI.2. A line L defines a subgroup of R2 if and only

if it passes through the origin. In that case, its cosets are
the lines parallel to it.

Conclusion: A line in R2 is a subgroup if and only if it passes through the

origin. If it does, then its cosets are the lines parallel to it. ♦

XI.2. Solving Equations

Cosets arise naturally when solving certain types of equations. It’s
difficult to make this precise at present. Instead I’ll show you some ex-
amples so that you can see what I mean.
 XI.2. SOLVING EQUATIONS 73

Example XI.8. If you did matrices at school, then you will probably know
that a system of m linear equations in n variables can be written as a
single matrix equation

(XI.19) Ax = b

where A is an m ×n matrix, b is a vector in Rm and x is an unknown vector

in Rn .
Let
K = {x ∈ Rn : Ax = 0}.
That is, K is the set of solutions x of the equation Ax = 0. It is easy to show
that K is a subgroup of Rn (exercise!). We call K the kernel of A. What
is the relation between K and the solutions of (XI.19)? If (XI.19) has no
solutions then there is no relation. So let’s suppose it has some solutions,
and let’s take x0 to be one of them. Let x be any other solution. Then

Ax = b, Ax0 = b.

Subtracting we find
A(x − x0 ) = 0.
So the difference x − x0 belongs to the subgroup K . Thus x belongs to the
coset x0 + K . In fact, the set of solutions to (XI.19) is precisely the coset
x0 + K . ♦
Example XI.9. In the Differential Equations module, one of things you’ll
look at are linear second order differential equations. For example, you’ll
see equations of the form
d 2x dx
(XI.20) a 2
+b + c x = f (t ),
dt dt
with a, b, c constants (again, it is likely that you’ve seen these at school).
To solve this you look at the corresponding homogeneous equation

d 2x dx
(XI.21) a 2
+b + c x = 0.
dt dt
Convince yourself that the solutions to the homogeneous equation (XI.21)
form a group K with respect to addition (revise Section VIII.5 if you need
to). In some textbooks on differential equations (and some old A-Level
maths textbooks), K is called the kernel. Now we ask the same question
as in the previous example: what is the relation between the solutions to
(XI.20) and K ? Again, if (XI.20) does not have a solution then there is no
relation. Suppose it has solutions, and let x 0 (t ) be one of them. In your
Differential Equations module, x 0 (t ) is called ‘a particular integral’. If x(t )
is any other solution to (XI.20), then you can check that x(t ) − x 0 (t ) is a
solution to the homogeneous equation (XI.21), and so is an element of K .
It follows that the set of solutions to (XI.20) is the coset x 0 (t ) + K . ♦
74 XI. COSETS

Are the similarities between the above two examples a coincidence?

No, they are instances of a recurrent theme in mathematics. This theme is
formalized in the First Isomorphism Theorem, which you’ll meet in Al-
gebra II. A lot of maths students never understand the First Isomorphism a fate almost
Theorem. They somehow don’t realize that they’ve been using it for years worse than death
when solving linear equations (and linear differential equations). Don’t
let that happen to you; after you meet the First Isomorphism Theorem,
come back and review these two examples again.

XI.3. Index
Definition. Let G be a group and H be a subgroup. We shall define the
index of H in G, denoted by [G : H ], to be the number of left cosets of H
in G.
Example XI.10. In Example XI.1, we computed the left cosets of R = {1, ρ 1 , ρ 2 , ρ 3 }
in D 4 and found exactly two of them: namely
{1, ρ 1 , ρ 2 , ρ 3 } and {σ0 , σ1 , σ2 , σ3 }.
So the index [D 4 : R] = 2. ♦
Example XI.11. In Example XI.3 we found that the cosets of 2Z in Z are
2Z itself, and 1 + 2Z, so the index [Z : 2Z] = 2. If you’ve done Exercise XI.4
then you’ll know that [Z2 : 2Z2 ] = 4. ♦
Example XI.12. In Example XI.6, we found that the cosets of the circle
group S in C∗ are the circles centred at the origin. So the index [C∗ : S] =
∞.
Example XI.13. Now let’s look at the index of the trivial group {0} as a
subgroup of Z. Note that
a + {0} = {a}.
So the cosets of {0} in Z are
. . . , {−2}, {−1}, {0}, {1}, {2}, . . .
Clearly [Z : {0}] = ∞. ♦
Exercise XI.14. Let G be a finite group. Let {1} be the trivial subgroup
consisting only of the identity element. Explain why [G : {1}] = |G|.

XI.4. The First Innermost Secret of Cosets

Apart from the definition, you need to know two facts about cosets.
The first is that a coset of a subgroup has the same size as the subgroup.
Lemma XI.15. Let G be a group and H a finite subgroup. If g ∈ G then g H
and H g have the same number of elements as H .
P ROOF. We’ll just prove the lemma for left cosets. The proof for right
cosets is nearly the same. Let g be an element of G. We want to show
that H and g H has the same number of elements. The sets H and g H are
 XI.5. THE SECOND INNERMOST SECRET OF COSETS 75

A priceless tip! finite. The best way to show that two finite sets have the same number
of elements is to set up a bijection between them. Let

φ : H → g H, h 7→ g h.

From the definition of g H it is clear that φ(h) is in the coset g H whenever

h is in the subgroup H . So the map φ makes sense. To check that it is a
bijection we need to show that it is injective and surjective.
Injectiveness: Suppose two elements h 1 , h 2 map to the same element in
g H . In otherwords, φ(h 1 ) = φ(h 2 ). We want to show that h 1 = h 2 . But
φ(h 1 ) = φ(h 2 ) means
g h1 = g h2 .
Now we can’t say, ‘divide by g ’. If you’ve forgotten why, see the pitfall on
page 38. By we can say multiply both sides on the left by g −1 , to obtain

g −1 (g h 1 ) = g −1 (g h 2 ).

Thus h 1 = h 2 .
Surjectiveness: Suppose k is an element of the coset g H . We want to
show that k is of the form φ(h) for some element h of the subgroup H . But
by definition, g H = {g h : h ∈ H }, so k = g h = φ(h) for some h in H . 

A Highbrow Remark for the Cognoscenti. Note that the proof that φ :

n
H → g H is a bijection did not assume the finiteness of H ; it is true for any
subgroup H whether finite or infinite. The finiteness is used to conclude

w
that the number of elements of H and the number of elements of g H are

, a
the same. What happens if H is infinite? Mathematicians still think of H

y
and g H as having the same number of elements, even though they are
infinite, simply because there is a bijection between them. Thus |2Z| =

w n
|1 + 2Z|, and |S| = |2S|. However, |2Z| 6= |S|, because 2Z is countable and
S is uncountable. If you find this interesting, have a look at cardinalities

ya
on Wikipedia. But only a brief look; trust me, set theory is as boring as
hell. In any case, feel free to ignore this remark.

Example XI.16. Now is a good time to revisit the examples at the begin-
ning of the chapter and make sure that Lemma XI.15 holds for them.

XI.5. The Second Innermost Secret of Cosets

Lemma XI.17. Let G be a group and H be a subgroup. Let g 1 , g 2 be ele-
ments of G. Then the cosets g 1 H , g 2 H are either equal or disjoint 1.

1Two sets A, B are disjoint if they have no members in common. Another way of
saying the same thing is: two sets A, B are disjoint if A ∩B = ;. I’m now confused—have
I said it in another way, or in the same way but with more notation?
76 XI. COSETS

Example XI.18. Look again at Example XI.6 and in particular Figure XI.1.
There we looked the cosets of the circle subgroup S inside C∗ . We found Geometric Epiphany
that the cosets are the circles centred at the origin of positive radius. It is I
obvious that two such circles are either equal or disjoint. ♦
Example XI.19. In Example XI.7, we saw that a line L in R2 passing through Geometric Epiphany
the origin defines a subgroup. The cosets of L are the lines parallel to it. II
Again it is clear that two lines parallel to L are either equal or disjoint. ♦

P ROOF OF L EMMA XI.17. Suppose g 1 H and g 2 H are not disjoint. We want

to show that they’re equal. If you look again at the examples you’ll see that
g 1 H = g 2 H doesn’t necessarily mean that g 1 = g 2 .
As g 1 H and g 2 H are not disjoint, they must have a common element.
The elements of g 1 H have the form g 1 h 1 and the elements of g 2 H have
the form g 2 h 2 where h 1 , h 2 are in H . Thus there is a pair h 1 , h 2 in H so
that g 1 h 1 = g 2 h 2 . In particular
(XI.22) g 1 = g 2 h 2 h 1−1 .
We want to show that g 1 H = g 2 H . You’ll no doubt recall that to prove
two sets are equal we have to show that every element in either set is an
element of the other set. Take an element of g 1 H . This must have the
form g 1 h for some h in H . We want to show that g 1 h is also an element
of g 2 H . Now note
g 1 h = (g 2 h 2 h 1−1 )h by (XI.22)
= g 2 (h 2 h 1−1 h).

However, h 2 h 1−1 h is a product of elements of the subgroup H and there-

fore an element of H . Hence we’ve written g 1 h in the form g 2 h 0 where
h 0 = h 2 h 1−1 h is an element of H . Thus every element of g 1 H is again an
element of g 2 H . Similarly, every element of g 2 H is an element of g 1 H .
Hence g 1 H = g 2 H . 

XI.6. Lagrange Super-Strength

I’ve stated Lagrange’s Theorem a very long time ago, and kept you
waiting for the proof ever since. Surely you consider this delay a delib- shedding bitter tears
erate act of unspeakable cruelty. It was indeed deliberate; I thought the of remorse and
prolonged wait would heighten the anticipation and make you appreci- penantly pleading for
ate and enjoy the proof even more. Alas, through an act of infinite self- forgiveness
lessness, I’ve sacrificed my popularity to intensify your infatuation with
the subject.
We now state an even stronger version of Lagrange’s Theorem.
Theorem XI.20. (Lagrange’s Theorem—Version 3) Let G be a finite group
and H a subgroup. Then
|G| = [G : H ] · |H |.
 XI.6. LAGRANGE SUPER-STRENGTH 77

This version is saying more than Version 2 of Lagrange’s Theorem

(Theorem VIII.33). Version 2 says that |H | divides |G|. Version 3 tells us
that not only does |H | divide |G|, but that the ratio is the index [G : H ]. So
if we prove this version of Lagrange’s Theorem then we have also proved
Version 2.
P ROOF OF T HEOREM XI.20. Let g 1 H , g 2 H , . . . , g m H be the distinct left co-
sets of H . As they are distinct, we know by Lemma XI.17 that they are
disjoint. Suppose now that g is an element of G. Then g H must equal one
of the g i H . But g ∈ g H , since 1 ∈ H . Hence the cosets g 1 H , g 2 H , . . . , g m H
are not only disjoint, but every element of G belongs to exactly one of
them. Hence
|G| = |g 1 H | + |g 2 H | + · · · + |g m H |.
Now by Lemma XI.15,
|g 1 H | = |g 2 H | = · · · = |g m H | = |H |.
Hence
|G| = m · |H |.
What is m? It is the number of left cosets of H in G. We defined this to be
the index of H in G, so m = [G : H ]. This completes the proof. 

Pure ecstacy? Of course! Maths is about delayed gratification.

 CHAPTER XII

Quotient Groups

Taking quotients is one of the most powerful concepts in mathemat-

ics. It should also be one of the least painful to assimilate. Instead of
revelling in quotients, most Warwickers go through three or four miser-
allow me to end your able years of being terrorised by them. The difficulties with quotients are
suffering purely psychological. To overcome them, you just need to study and vi-
sualize a good number of examples. What are we waiting for?

uotients! I lo love quotie I

ve ve
q n n ts!
I lo
ts

ie
qu

ot
!

! I love qu
o
tients! I lov

nts
eq

tie

uo o
tie qu
nt ve
s!
Il I lo
ov s!
ent
ti
e
qu

o
qu
ot
ien

ve
I lo
!ts

F IGURE XII.1. Nurture a positive attitude to quotients—it

won’t let you down!

XII.1. Congruences Modulo Subgroups

Let (G, +) be an abelian group, where the binary operation is ‘addi-
tion’. For example, G could be R, R2 , C, Z, R[x] etc. Let H be a subgroup.
Let a, b be elements of G. We say that a, b are congruent modulo H if
a − b ∈ H . In this case we write a ≡ b (mod H ).
Example XII.1. Let m ≥ 2 be an integer. We know that mZ is the sub-
group of Z consisting of the multiples of m. Let a, b ∈ Z. Then a ≡ b
(mod mZ) if and only a − b is a multiple of m. In other words, a ≡ b
79
80 XII. QUOTIENT GROUPS

(mod mZ) if and only if a ≡ b (mod m). The concept of congruence mod-
ulo subgroups is a generalization of the earlier concept of congruence mod-
ulo integers. ♦
Example XII.2. Z is a subgroup of R. Two real numbers a, b are con-
gruent modulo Z if and only if a − b ∈ Z. This means that their differ-
ence is an integer. So, for example 1437.14 ≡ 0.14 (mod Z). It may seem
that congruence modulo Z is a stupid idea. After all, we’re concentrating
on the small fractional part of number and ignoring the bigger integer
part. However in some situations, the fractional part is the important
one. Let’s see one of those situations. In Example VIII.17 we defined the
circle group
S = {α ∈ C : |α| = 1}.
Let
f : R → S, f (θ) = e 2πi θ .
What happens to f (θ) as θ changes? If we start with θ = 0 ∈ R and increase
the value of θ, then f (θ) starts at 1 ∈ S and moves anticlockwise. When θ
reaches 1 ∈ R then f (θ) will have done a complete circle and returned to
1 ∈ S. By the time θ reaches 2 ∈ R, f (θ) will have done another complete
circle and returned again to 1 ∈ S. Of course, you want me to be less
clumsy and say that f is periodic with period 1. Indeed f (φ) = f (θ) if and
only if φ = θ + n where n is an integer. Now Z is a subgroup of R. So we
can rewrite that fact as f (φ) = f (θ) if and only if φ ≡ θ (mod Z). ♦
Example XII.3. Let X = {(a, 0) : a ∈ R}. It’s easy to show that X is a sub-
group of R2 , which is simply the x-axis. What does it mean for two points
to be congruent modulo X ? Suppose (a 1 , b 1 ) and (a 2 , b 2 ) are in R2 . Then
(a 1 , b 1 ) ≡ (a 2 , b 2 ) (mod X ) if and only if (a 1 −a 2 , b 1 −b 2 ) belongs to X . This
happens if and only if b 1 − b 2 = 0. So two points are congruent modulo X
if and only if they have the same y-coordinate. ♦
Example XII.4. Let G = R[x]. Let H = { f ∈ R[x] : f (0) = 0}. It is an easy
exercise to show that H is a subgroup of R[x]. Now let’s understand what
it means for two polynomials to be congruent modulo H . Suppose g ,
h ∈ R[x]. Write 1

g = a0 + a1 x + · · · + an x n , h = b0 + b1 x + · · · + bn x n ,

where a 0 , . . . , a n and b 0 , . . . , b n are real numbers. Let f = g −h. Then g ≡ h

(mod H ) if and only if f (0) = 0, which means a 0 −b 0 = 0. Therefore g and
h are congruent modulo H if and only their constant terms are equal. ♦

1It seems that we’re writing f and g both as polynomials of the same degree n; this
looks wrong as there is no reason to suppose that g and h have the same degree. But
looks can be misleading. Here we’re in fact writing g and h as polynomials of degree at
most n. For example, if g = 2 + 7x and h = 4 − 3x + 2x 3 then we can take n = 3 and let
a 0 = 2, a 1 = 7, a 2 = a 3 = 0, and b 0 = 4, b 1 = −3, b 2 = 0, b 3 = 2
 XII.2. CONGRUENCE CLASSES AND COSETS 81

Exercise XII.5. Let (G, +) be an abelian group. We know that {0} and G
are subgroups of G. What does it mean for a and b to congruent modulo
{0}? What does it mean for a and b to be congruent modulo G?

XII.2. Congruence Classes and Cosets

Let (G, +) be an additive abelian group and H a subgroup. Let a ∈ G.
We shall denote by a the congruence class of a modulo H ; this is defined
by
a = {b ∈ G : b ≡ a (mod H )}.

In words, the congruence class of a modulo H is the set of all elements of

G that are congruent to a modulo H .
Example XII.6. If G = Z and H = mZ, then a is simply the congruence
class of a modulo m:

a = {. . . , a − 2m, a − m, a, a + m, a + 2m, a + 3m, . . . }.

In Foundations, a is denoted by [a]. ♦

Lemma XII.7. Let (G, +) be an additive abelian group and H a subgroup.
Let a ∈ G let a be the congruence class of a modulo H . Then

a = a + H.

P ROOF. We know a is the set of b ∈ G that are congruent to a modulo

H . But b ≡ a (mod H ) is the same as saying b − a ∈ H or b ∈ a + H . So
a = a + H. 

We made the set of congruence classes in Z modulo mZ into a group

Z/mZ, and in the same way we can form a group out of the set of congru-
ence classes in an additive abelian group G modulo a subgroup H .
Definition. Let (G, +) be an additive abelian group and H a subgroup.
We define the quotient group (G/H , +) to the set of congruence classes
(or the set of cosets)
G/H = {a : a ∈ G}

with addition being defined by

(XII.23) a + b = a + b.

As usual, we need to prove that (G/H , +) is a group (in fact it is abelian).

There is a more serious point which is that we need to show that the oper-
ation (XII.23) is well-defined. These details will disrupt the flow of things
and I’ve relegated them to Section XII.6. For now, we want to focus on
understanding quotient groups and how to think about them.
82 XII. QUOTIENT GROUPS

XII.3. R/Z
In Example XII.2 we looked at congruences in R modulo Z. We now
want to understand the group R/Z. Note that every real number is con-
gruent modulo Z to a unique number in the half-open interval

[0, 1) = {x ∈ R : 0 ≤ x < 1}.

Therefore
R/Z = {a : a ∈ [0, 1)}.
So when we add a + b, we take the result of a + b, and and simplify by
subtracting an integer if necessary to obtain c ∈ [0, 1), and then letting
a + b = c. For example,

0.7 + 0.2 = 0.9, 0.7 + 0.4 = 0.1, 0.3 − 0.5 = 0.8.

If we go back to the map

f : R → S, f (θ) = e 2πi θ ,

we can define a similar map,

³ ´
fˆ : R/Z → S, f θ = e 2πi θ .
ˆ

Check that fˆ is a bijection that satisfies

³ ´ ³ ´ ³ ´
fˆ θ + φ = fˆ θ fˆ φ .

In essence what this is saying is that the two groups (R/Z, +) and (S, ·) are
‘essentially the same’. Indeed, they’re isomorphic. Now is a good time to
look again at Chapter X.
Here is how you should think about R/Z. We identified it with the
interval [0, 1). Think of starting at 0.95 and moving up in small steps of
0.01:
0.95, 0.96, 0.97, 0.98, 0.99, 0.00, 0.01, 0.02, 0.03, . . .
So we should really think of R/Z as the interval [0, 1) with one end joined
to the other. If you take a string and join one end to the other you ob-
tain a loop, or a ‘circle’. This is what fˆ is doing. It is showing that R/Z is
isomorphic to the unit circle S. Indeed, the 2π in the formula for fˆ is a
‘stretching factor’, since the interval [0, 1) of length 1 has to be ‘stretched’
around the unit circle of perimeter 2π.
Exercise XII.8. R/Z has four elements of order 5. Find them.
Exercise XII.9. Let α ∈ [0, 1). Show that α has finite order in R/Z if and
only if α is rational.
Exercise XII.10. Show that fˆ takes rational numbers to roots of unity.
 XII.4. R2 /Z2 83

XII.4. R2 /Z2
After R/Z you shouldn’t have any trouble imagining R2 /Z2 . You’re al-
lowed to shift any point in R2 by an integer multiple of i and an integer
multiple of j. So you end up in the unit square:
(XII.24) {(x, y) : 0 ≤ x < 1, 0 ≤ y < 1}.
But you should think of this square as having the top side glued to the
bottom side, and the right side glued to the left side! See Figure XII.2

(0, 1) (1, 1)

Peb

ahhh—psychedelic visions of ponies and glitter

bles Bub
bles

(0, 0) (1, 0)

F IGURE XII.2. R2 /Z2 is really just the unit square with the
top side glued to the bottom side, and the right side glued
to the left side.

Example XII.11. In this example, we’ll find the elements of order 2 in

R2 /Z2 . Suppose (x, y) is such an element, where x, y belong to the in-
terval [0, 1). Then (2x, 2y) = 0 and so 2x, 2y are integers. Hence
2x = . . . , −1, 0, 1, 2, 3, . . . , 2y = . . . , −1, 0, 1, 2, 3, . . . .
Therefore,
1 1 3 1 1 3
x = · · · , − , 0, , 1, , · · · , y = · · · , − , 0, , 1, , · · · .
2 2 2 2 2 2
As x and y belong to the interval [0, 1), we see that x = 0 or 1/2 and y = 0
or 1/2. Hence
(x, y) = (0, 0), (1/2, 0), (0, 1/2), (1/2, 1/2).
However, the first of these has order 1. So the elements of order 2 in R2 /Z2
are
(x, y) = (1/2, 0), (0, 1/2), (1/2, 1/2).
84 XII. QUOTIENT GROUPS

♦
Exercise XII.12. Find all elements of order 3 in R2 /Z2 (there are eight of
them).
Exercise XII.13. In Z2 we let i = (1, 0) and j = (0, 1) as usual. Write 2Z2 =
{(2a, 2b) : a, b ∈ Z}. Convince yourself that 2Z2 is a subgroup of Z2 of
index 4 and that
Z2 /2Z2 = {0, i, j, i + j}.
Write down an addition table for Z2 /2Z2 .
Exercise XII.14. How would you describe C/Z[i ]? Is it really different
from R2 /Z2 ?
Exercise XII.15. How would you describe C/Z? Find all elements of order
2.

XII.5. R/Q
In this section, we shall briefly think about R/Q. In R/Z, we treat the
integers as ‘zero’. In R/Q, we treat the rationals as ‘zero’. This is a much
trickier quotient group. The trickiness does not come from the defini-
tion; there is no difficulty there. We can add in R/Q using the defini-
tion (XII.23). The problem is with ‘simplifying’ the result. Let’s try some
p
numerical examples
p so that you see what I mean. If we take a = 1 + 2
and b = 2/3 − 2, then
a + b = a + b = 5/3 = 0,
because 5/3 − 0 = 5/3 ∈ Q. However, if we take a = π and b = e (where
these have their usual values) then
a + b = π + e.
Can we simplify this? For example, is this equal to 0? It is if and only if
π + e is a rational number. No one knows if the number π + e is rational
or not (but we know that both π and e are irrational). So we don’t know if
the result of the above calculation is equal to 0 or not.

XII.6. Well-Defined and Proofs

We know that in Z/mZ, not only does addition make sense, but also
multiplication makes sense. In Z/mZ we defined multiplication by
(XII.25) a · b = ab.
Now, you might ask why we don’t define multiplication on R/Z in the
same way? OK, let’s try using the same definition for multiplication on
R/Z and see what happens:
0.5 × 0.5 = 0.25, 1.5 × 0.5 = 0.75.
There is a problem: in R/Z, the classes 1.5 and 0.5 are equal, but the
classes 0.75 and 0.25 aren’t. Multiplication doesn’t make sense on R/Z.
 XII.6. WELL-DEFINED AND PROOFS 85

The problem comes from the ‘definition’ for multiplication in (XII.25).

We’re trying to define the product of a and b in terms of the representa-
tives a, b of these classes. But each class has many representatives. For
a definition such as this to make sense, the result must be independent
of the choice of representatives. Now you might be wondering why mul-
tiplication in Z/mZ makes sense. This was actually done in Foundations
but it is worth looking at the proof again.
Lemma XII.16. Let m ≥ 2 be an integer. Let a, a 0 , b, b 0 satisfying
a = a0, b = b0,
in Z/mZ. Then
ab = a 0 b 0 .
We say that multiplication is well-defined on Z/mZ. This means that
the result of a product does not depend on the choice of representatitives,
even though it defined in terms of those representatives.

P ROOF. As a = a 0 and b = b 0 we know that

a 0 = a + km, b 0 = b + `m,
where k and ` are integers. So
a 0 b 0 = ab + m(kb + `a + mk`).
But kb + `a + mk` is an integer as it is a sum of products of integers. So
a 0 b 0 ≡ ab (mod mZ),
which means
ab = a 0 b 0 .

Are we sure addition that addition is well-defined in R/Z and the other
quotient groups that we’ve been working with? The following lemma
checks that.
Lemma XII.17. Let (G, +) be an additive abelian group and H a subgroup.
Let a, a 0 , b, b 0 be elements of G such that in G/H we have
a = a0, b = b0,
then
a + b = a0 + b0.
P ROOF. Suppose a = a 0 and b = b 0 in G/H . Then a−a 0 = h 1 and b−b 0 = h 2
where h 1 , h 2 ∈ H . Thus
(a + b) − (a 0 + b 0 ) = (a − a 0 ) + (b − b 0 ) = h 1 + h 2 .
As H is a subgroup containing h 1 and h 2 , we know that the sum h 1 + h 2
belongs to H . Thus the classes a + b and a 0 + b 0 are equal. 
86 XII. QUOTIENT GROUPS

To wrap up this chapter, we need to check one thing: that G/H is in-
deed a group.
Theorem XII.18. Let (G, +) be an additive abelian group and H a sub-
group. Then (G/H , +) is an abelian group.
P ROOF. All we have to do is check the defining properties for abelian
groups. I’ll just check that addition is commutative and leave the rest
to you. Suppose a, b ∈ G. Then
b+a =b+a from the definition of addition in G/H
(XII.26) = a +b b + a = a + b as G is abelian
= a +b from the definition of addition in G/H .

We’ve only looked at quotients of abelian additive groups. For general
groups, things are more tricky. At the heart of the trickiness is that in the
non-abelian setting the binary operation on cosets might not be well-
defined. For now, if you’ve got to grips with Z/mZ, R/Z and R2 /Z2 then
you’ve made an excellent start with quotients.
 CHAPTER XIII

Symmetric Groups

The coolest groups have elements that are functions. Matrix groups
are examples, and symmetric groups are other examples. It turns out that
every finite group is a subgroup of one of the symmetric groups. So if we
understand symmetric groups completely, then we’ll understand finite
groups completely. Have I given you hope that you’ll have a complete
understanding of finite groups by the end of the chapter? Sorry, I was
saying ‘if . . . ’

XIII.1. Motivation
Let A be a set, and let f , g be functions from A to itself. We know that
we can compose f , g to obtain f ◦ g which is also a function from A to
itself. We shall write Map(A) for the set of functions from A to itself. Then
◦ is a binary operation on Map(A). And it’s natural to ask if this makes
Map(A) into a group. After all, we know by Lemma II.2 that composition
of functions is associative. The following example will help clarify these
ideas.
Example XIII.1. Let A = {1, 2}. You will quickly convince yourself that
there are only four functions from A to itself, which are given in Fig-
ure XIII.1.

1 1 1 1

2 2 2 2
f1 f2

1 1 1 1

2 2 2 2
f3 f4

F IGURE XIII.1. f 1 , f 2 , f 3 and f 4 are the four functions from

{1, 2} to itself.

87
88 XIII. SYMMETRIC GROUPS

Thus Map(A) = { f 1 , f 2 , f 3 , f 4 }. Is Map(A) a group with respect to com-

position of functions? Here is the composition table for Map(A):
◦ f1 f2 f3 f4
f1 f1 f2 f3 f4
f2 f2 f1 f4 f3
f3 f3 f3 f3 f3
f4 f4 f4 f4 f4
Make sure you understand the table. The entry for f i ◦ f j is at the in-
tersection of the i -th row and j -th column. As always, f i ◦ f j means apply
f j first then f i . We know that composition of functions is associative by
Lemma II.2. Moreover, it is clear from the table that f 1 is the identity ele-
ment. But f 3 and f 4 don’t have inverses; we can’t combine either of them
with any of the four functions to obtain the identity f 1 .
But if you look carefully at the table, you will see a group with respect
to composition. It is the subset: { f 1 , f 2 }. If you’ve been paying attention
in Foundations you will know why f 1 , f 2 have inverses (which in this case
happen to be f 1 and f 2 respectively), and f 3 , f 4 don’t: the functions f 1 , f 2
are bijections and f 3 and f 4 are not.
Now is a good time for you to revise Example III.10. There you saw a
that non-invertible matrix (which represented projection onto the y-axis)
was also a non-bijective ‘function’ R2 → R2 . ♦

XIII.2. Injections, Surjections and Bijections

In this section we revise some stuff that you’ve done in Foundations.
For the proofs you should refer back to your Foundations lecture notes.
Definition. Let A, B be sets, and let f : A → B be a function from A to
B (also called a map or a mapping from A to B ). We call A the domain
of f and B the codomain of f or the range of f . We say f is injective if
whenever a 1 , a 2 ∈ A satisfy a 1 6= a 2 then f (a 1 ) 6= f (a 2 ). In other words,
distinct elements of A are mapped to distinct elements of B .
We say f is surjective if for every b ∈ B , there is some element a ∈ A
such that f (a) = b. In other words, every element of B is in the image.
We say f is bijective if it is injective and surjective.
Example XIII.2. See Figure XIII.2. Here f 1 is not a function, since f 1 (2) =
b and f 1 (2) = c. A function takes one elementp of the domain to exactly
one element of the codomain. If you write 4 = ±2, then you’re thinking
p
of as a multifunction and not a function. In terms of pictures as in
the figure, for f to be a function, exactly one arrow originates at any one
element of the domain.
f 2 is injective since there are exactly two distinct elements in its do-
main which are 1 and 2 and these get mapped to distinct elements c and
a. However, f 2 is not surjective, since b is in the codomain, but f (1) 6= b
 XIII.2. INJECTIONS, SURJECTIONS AND BIJECTIONS 89

1 a 1 a
b b
2 c 2 c
f1 f2

B C
a 1 a u
b b v
c 2 c w
f3 f4

F IGURE XIII.2. f 1 is not a function, f 2 is injective but not

surjective, f 3 is surjective but not injective and f 4 is a bi-
jection.

and f (2) 6= b. In terms of pictures, surjective means that every element of

the codomain is the end point of at least one arrow.
f 3 is surjective but not injective. In terms of pictures, injective means
that no two arrows share the same end point.
bijection is f 4 : B → C is a bijection. A bijection is merely an act of relabelling.
relabelling The sets B and C are the same if we relabel a as v, b as u and c as w. ♦
Example XIII.3. Let f 1 , f 2 , f 3 , f 4 be the functions {1, 2} → {1, 2} in Fig-
ure XIII.1. Then f 1 and f 2 are bijections. However, f 3 and f 4 are neither
injective nor surjective. ♦
Remarks.

(i) Instead of saying that a function is injective, mathematicians some-

times say that it is one-to-one (also written 1 − 1).
(ii) The definition of injective is often given in the contrapositive
form: if f (a 1 ) = f (a 2 ) then a 1 = a 2 . The way we’ve phrased the
definition is more helpful for this chapter, but you should get
used to both forms.
(iii) Instead of saying that a function is surjective, mathematicians
sometimes say that it is onto. I found this jarring when I first saw
it. But I quickly got used to it.
(iv) A bijection is also called a one-to-one correspondence.

Here is a theorem you have seen in Foundations where it was proba-

bly called ‘the pigeon-hole principle’.
Theorem XIII.4. Let A be a finite set and let f be a function from A to
itself. Then f is injective if and only if f is surjective.
90 XIII. SYMMETRIC GROUPS

Example XIII.5. Look back at the functions in Figure XIII.1 for a very ba-
sic illustration of Theorem XIII.4. ♦
Example XIII.6. Theorem XIII.4 is true for finite sets only. For infinite
sets it might or might not hold. Let f 1 : N → N be given by f 1 (x) = x +
1. Then f 1 is not surjective since 0 is not in the image. However, f 1 is
injective. See Figure XIII.3. By contrast, let f 2 : Z → Z be also given by

N N Z Z
0 0 .. ..
. .

1 1 −2 −2

2 2 −1 −1

3 3 0 0

4 4 1 1

5 5 2 2

.. .. .. ..
. . . .
f1 f2

F IGURE XIII.3. f 1 : N → N and f 2 : Z → Z are both given by

x 7→ x + 1. The function f 1 is injective but not surjective.
The function f 2 is injective and surjective; therefore it is
a bijection. The pigeon-hole principle holds only for finite
sets!

x 7→ x + 1. Then f 2 is a bijection. ♦
The following theorems collect together key results regarding bijec-
tions. Again you know all of this from Foundations.
Theorem XIII.7. Let f : A → B and g : B → C be bijections. Then g ◦ f is a
bijection A → C .
Definition. Let A be a set. The identity map on A is the map id A : A → A
satisfying id A (x) = x for all x ∈ A.
Let f : A → A be a function on A. We say that f is invertible if there
exists a function g : A → A such that f ◦ g = g ◦ f = id A . If such a g exists
then we call it the inverse of f and denote it by f −1 .
Of course you know from Foundations that if f has an inverse then it
is unique.
Theorem XIII.8. Let f : A → A be a function. Then f is invertible if and
only if f is a bijection. If f is invertible then f −1 is also a bijection.
 XIII.4. S n 91

XIII.3. The Symmetric Group

Let A be a set. We shall denote the set of bijections from A to itself by
Sym(A).
Example XIII.9. In Example XIII.1 we wrote down all the functions from
A = {1, 2} to itself and found that exactly two of these are bijections. These
were called f 1 and f 2 in Figure XIII.1. Hence Sym(A) = { f 1 , f 2 }. Note that
f 1 = id A . In that example, we noted that { f 1 , f 2 } is a group under compo-
sition with f 1 being the identity element. Check this again, and note that
the group is abelian. ♦
Theorem XIII.10. Let A be a set. Then (Sym(A), ◦) is a group with id A as
the identity element.
We call Sym(A) the symmetric group on A.
P ROOF. By Theorem XIII.7, Sym(A) is closed under composition. More-
over, composition of functions is associative by Lemma II.2.
Clearly id A is a bijection and so is in Sym(A). We want to check that
id A is the identity for composition, which means that for any f ∈ Sym(A)
we want f ◦ id A = id A ◦ f = f . Note
( f ◦ id A )(x) = f (id A (x)) = f (x), (id A ◦ f )(x) = id A ( f (x)) = f (x).
Thus f ◦ id A = id A ◦ f = f holds.
Finally we want every element of Sym(A) to have an inverse in Sym(A).
This is true by Theorem XIII.8. 
Example XIII.11. Let f 1 , f 2 be as in Example XIII.6. Note that f 1 ∉ Sym(N)
since it is not a bijection. However f 2 ∈ Sym(Z). What is f 2−1 ? It is simply
the function Z → Z given by x 7→ x − 1.
Let’s calculate g = f 23 = f 2 ◦ f 2 ◦ f 2 . Then
g (x) = f 2 ( f 2 ( f 2 (x))) = f 2 ( f 2 (x + 1)) = f 2 (x + 2) = x + 3.
It will be easy for you to show, for any integer n, that f 2n is the function
Z → Z satisfying x 7→ x + n. In particular, f 2n 6= id A for n 6= 0. Thus f 2 is an
element of infinite order in the group Sym(Z). ♦
Exercise XIII.12. Let f : Z → Z and g : R → R be given by x 7→ 2x. Show
that f ∉ Sym(Z) but g ∈ Sym(R). Write down g n for integers n.
Exercise XIII.13. Let f : C → C, g : C → C, h : C → C be given by f (z) =
z + 1, g (z) = z + i , h(z) = i z. Describe f , g , h geometrically. Show that f ,
g , h are in Sym(C). Show that f and g commute. What about f and h or
g and h? What are the orders of f , g and h?

XIII.4. S n
We define S n to be the group Sym({1, 2, . . . , n}). We call S n the n-th
symmetric group. In Example XIII.9 we found that S 2 is a group of order
2.
92 XIII. SYMMETRIC GROUPS

Theorem XIII.14. S n has order n!.

P ROOF. S n is the set of bijections from {1, 2, . . . , n} to itself. So we want to
count these bijections. Since the set {1, 2, . . . , n} is finite, Theorem XIII.4
tells us that a bijection from the set to itself is the same as an injection. So
let’s count the injections. Let f be an injection from {1, 2, . . . , n} to itself.
Then f (1) can be any of 1, 2, . . . , n; that is, there are n choices for f (1). If
we fix f (1) then f (2) 6= f (1). So there are n − 1 choices for f (2) once we’ve
chosen f (1). Likewise there are n − 2 choices for f (3) once we’ve chosen
f (1) and f (2). It is now clear that the number of injections is
n × (n − 1) × · · · × 1 = n!.

The elements of S n are called permutations. One way of represent-
ing permutations is to use diagrams such as those for f 1 , f 2 ∈ S 2 in Fig-
ure XIII.1. The following is a more economical way. Let a 1 , a 2 , . . . , a n be
the numbers 1, 2, . . . , n in some order. Then
µ ¶
1 2 ··· n
a1 a2 · · · an
represents the unique permutation in S n that sends 1 to a 1 , 2 to a 2 , . . . ,
and n to a n .
Example XIII.15. S 2 has two elements:
µ ¶ µ ¶
1 2 1 2
, .
1 2 2 1
These are respectively the same as f 1 , f 2 in Figure XIII.1. The first of these
is the identity element. We noted in Example XIII.9 that S 2 = Sym({1, 2})
is abelian. ♦
Example XIII.16. We know from Theorem XIII.14 that S 3 has 6 elements.
These are
µ ¶ µ ¶ µ ¶
1 2 3 1 2 3 1 2 3
, ,
1 2 3 2 3 1 3 1 2
µ ¶ µ ¶ µ ¶
1 2 3 1 2 3 1 2 3
, , .
1 3 2 2 1 3 3 2 1
Again, the first of these is the identity element. It is important that you
know what the notation means and how to multiply two permutations
written in this notation, so let’s have some practice. Let
µ ¶ µ ¶
1 2 3 1 2 3
ρ= , µ= .
3 1 2 1 3 2
Never forget that these are bijections from {1, 2, 3} to itself. To find out
what ρ does, look at the columns. ρ is the function that sends 1 to 3, 2 to
1 and 3 to 2. Thus
(XIII.27) ρ(1) = 3, ρ(2) = 1, ρ(3) = 2.
 XIII.4. S n 93

Likewise,
µ(1) = 1, µ(2) = 3, µ(3) = 2.

Now let us compute ρµ. As always, this means apply µ first then ρ. So

(ρµ)(1) = ρ(µ(1)) = ρ(1) = 3;

(ρµ)(2) = ρ(µ(2)) = ρ(3) = 2;
(ρµ)(3) = ρ(µ(3)) = ρ(2) = 1.

Thus
µ ¶
1 2 3
ρµ = .
3 2 1

Similarly,
µ ¶
1 2 3
µρ = .
2 1 3

Note that µρ 6= ρµ, so S 3 is non-abelian. How do we compute ρ −1 ? From

(XIII.27) we find

1 = ρ −1 (3), 2 = ρ −1 (1), 3 = ρ −1 (2).

We rearrange this:

ρ −1 (1) = 2, ρ −1 (2) = 3, ρ −1 (3) = 1.

Hence
µ ¶
−1 1 2 3
ρ = .
2 3 1

♦
Exercise XIII.17. Write down a multiplication table for S 3 and determine
the orders of all six elements checking that your answers are consistent
with Lagrange’s Theorem.
Exercise XIII.18. Let ρ and τ be the following permutations:
µ ¶ µ ¶
1 2 3 4 5 1 2 3 4 5
ρ= , τ= .
2 3 5 1 4 3 1 2 5 4

Compute ρ −1 , ρτ, τ2 .
Exercise XIII.19. Show that S n is non-abelian for n ≥ 3.
Exercise XIII.20. Recall (page 32) that we interpreted elements of D 4 as
functions from {1, 2, 3, 4} to itself. Go back and check that these are bijec-
tions. Thus D 4 is a subgroup of S 4 .
94 XIII. SYMMETRIC GROUPS

XIII.4.1. What’s Special About S n ? We started the chapter by look-

ing at symmetry groups of arbitrary sets A. Then we restricted ourself to
S n = Sym({1, 2, . . . , n}). This is not as big a restriction as it looks. Suppose
the set A is finite, and let n = |A|, the number of elements of A. Then
Sym(A) is isomorphic to S n . One way of seeing this is convince ourselves
that every permutation of {1, 2, . . . , n} gives us a permutation of A. For
example, suppose A = {a 1 , a 2 , a 3 }. Then the permutation {1, 2, 3} given by
µ ¶
1 2 3
2 3 1
corresponds to the permutation of A given by
µ ¶
a1 a2 a3
.
a2 a3 a1
Understanding Sym(A) with |A| = n is the same as understanding S n .

XIII.5. A Nice Application of Lagrange’s Theorem

Let n, m be integers with 0 ≤ m ≤ n. You met before the binomial
coefficient à !
n n!
= .
m m!(n − m)!
It is not all obvious from this formula that the binomial coefficient is an
integer. As an application of Lagrange’s Theorem, we show that it is. Re-
call that S n has order n!. If we can find a subgroup H of S n of order
m!(n − m)!, then à !
n |S n |
= .
m |H |
We know by Lagrange’s Theorem that the right-hand side is simply [S n :
H ], the number of cosets of H in S n , and is therefore an integer. All
we have to do now is give the subgroup H of order m!(n − m)!. Let H
be the subset of S n consisting of permutations σ such that σ permutes
{1, 2, . . . , m} and permutes {m + 1, m + 2, . . . , n}. What do we mean by this?
Write
A = {1, 2, . . . , m}, B = {m + 1, m + 2, . . . , n}.
Note that
{1, 2, . . . , n} = A ∪ B.
The elements of S n are the bijections from the set {1, 2, . . . , n} to itself. The
elements of H are those elements σ of S n that satisfy σ(a) ∈ A for all a ∈
A, and σ(b) ∈ B for all b ∈ B . It’s an easy exercise to show that H is a
subgroup of S n . We want to check that its order is really m!(n − m)!. We
count the elements of H in a similar way to the argument in the proof
of Theorem XIII.14. Let σ be an element of H . Then σ(1) can be any of
1, 2, . . . , m. Once we’ve chosen σ(1), we know σ(2) can be any of 1, 2, . . . , m
except for σ(1). Thus there are m choices for σ(1), m − 1 choices for σ(2)
 XIII.6. CYCLE NOTATION 95

and so on. Until we reach σ(m + 1). This can be any element of B , and so
there are n − m choices for σ(m + 1). etc. You see that the order of H is
m(m − 1) · · · 1 · (n − m)(n − m − 1) · · · 1 = m!(n − m)!
Don’t you just love maths?
Exercise XIII.21. To make sure you’ve understood the argument above,
let m = 2 and n = 4, so that
A = {1, 2}, B = {3, 4}.
Now write down all permutations σ in S 4 that satisfy σ(a) ∈ A for all a ∈ A
and σ(b) ∈ B for all b ∈ B , and convince yourself that these form a group.

XIII.6. Cycle Notation

Let a 1 , a 2 , . . . , a m be distinct elements of the set {1, 2, . . . , n}. By the no-
tation
(XIII.28) (a 1 , a 2 , . . . , a m )
we mean the element of S n that takes a 1 to a 2 , a 2 to a 3 , . . . , a m−1 to a m
and a m back to a 1 , and fixes all other elements of {1, 2, . . . , n}. The permu-
tation (XIII.28) is called a cycle of length m. A cycle of length 2 is called a
transposition.
Example XIII.22. Let µ = (1, 4, 5) ∈ S 5 . The cycle µ is of length 3 and is
illustrated in Figure XIII.4.

2 3
5

F IGURE XIII.4. The cycle (1, 4, 5) ∈ S 5 .

We can write (1, 4, 5) using our old notation:

µ ¶
1 2 3 4 5
(1, 4, 5) = .
4 2 3 5 1
Notice that (1, 4, 5) = (4, 5, 1) = (5, 1, 4). However, (1, 4, 5) 6= (1, 5, 4).
The transposition (1, 5) ∈ S 5 is given in Figure XIII.5.
In our old notation, the transposition (1, 5) is written as follows:
µ ¶
1 2 3 4 5
(1, 5) = .
5 2 3 4 1
Note that (1, 5) = (5, 1).
96 XIII. SYMMETRIC GROUPS

5 1

2 3 4

F IGURE XIII.5. The transposition (1, 5) ∈ S 5 . This merely

swaps 1 and 5, and fixes all other elements.

Finally (1) is the cycle that takes 1 to itself and fixes all the other el-
ements. Clearly (1) = (2) = (3) = (4) = (5) = id is nothing other than the
identity permutation. ♦
I hope that the above example has convinced you that cycle notation
is simultaneously more concise and and more transparent than the old
notation. If so, the following theorem will come as a pleasant surprize.
Theorem XIII.23. Every permutation can be written as a product of dis-
joint cycles.
What does disjoint mean? Two cycles (a 1 , a 2 , . . . , a n ) and (b 1 , b 2 , . . . , b m )
are said to be disjoint if a i 6= b j for all i , j . What does product mean? The
product of two permutations is of course their composition as functions.
Before we prove the theorem, let’s see an example where we write down
a permutation as a product of cycles.
Example XIII.24. Let
µ ¶
1 2 3 4 5 6 7 8
ρ= .
5 7 1 4 8 2 6 3
Write ρ as a product of disjoint cycles.
Answer: We start with 1 are repeatedly apply ρ to it:
1 7→ 5 7→ 8 7→ 3 7→ 1.
Therefore ρ contains the cycle (1, 5, 8, 3). Now we start with an element of
the set {1, 2, . . . , 8} that is not contained in the cycle (1, 5, 8, 3). For example
start with 2 and repeatedly apply ρ to it:
2 7→ 7 7→ 6 7→ 2.
So ρ also contains the cycle (2, 7, 6). Note that the cycles (1, 5, 8, 3) and
(2, 7, 6) are disjoint, and ρ contains the product (or composition) (1, 5, 8, 3)(2, 7, 6).
There still remains one element of the set {1, 2, . . . , 8} that does not appear
as either of the two cycles (1, 5, 8, 3) and (2, 7, 6) and this is 4. Applying ρ
to 4 we find:
4 7→ 4.
So
ρ = (1, 5, 8, 3)(2, 7, 6)(4)
 XIII.6. CYCLE NOTATION 97

as a product of disjoint cycles. Recall that (4) is just the identity, so it is

usual to omit it and write,
ρ = (1, 5, 8, 3)(2, 7, 6).
You might be wondering why we wrote ρ as above and not ρ = (2, 7, 6)(1, 5, 8, 3).
This does not matter since disjoint cycles commute; more on this below.
♦
Example XIII.25. Let
σ = (1, 3, 10, 9)(2, 5, 6), τ = (4, 3, 10)(1, 5, 8).
Express στ and σ−1 as a product of disjoint cycles.
Answer: We start with 1 and follow the same procedure as the above ex-
ample. Note that στ1 means apply τ first to 1 and then apply σ to the
result. Now τ1 = 5 and σ5 = 6. So στ1 = 6. Next we apply στ to 6. The
permutation τ does not have 6 in its cycle decomposition, so τ6 = 6. So
στ6 = σ6 = 2. We keep applying στ until we return to 1:
1 7→ 6 7→ 2 7→ 5 7→ 8 7→ 3 7→ 9 7→ 1.
Thus στ has the cycle (1, 6, 2, 5, 8, 3, 9) in its decomposition as a product
of disjoint cycles. We note that this cycle has no 4 in it. So we apply στ
repeatedly starting with 4:
4 7→ 10 7→ 4.
Hence στ has the product (1, 6, 2, 5, 8, 3, 9)(4, 10) in its decomposition as
a product of disjoint cycles. Finally, note that of the elements of the set
{1, 2, . . . , 10}, the only one not appearing in the product (1, 6, 2, 5, 8, 3, 9)(4, 10)
is 7. However στ7 = 7. So
στ = (1, 6, 2, 5, 8, 3, 9)(4, 10)
as a product of disjoint cycles.
You may have noticed that we were tacitly assuming that σ and τ are
elements of S 10 and computed the product under that assumption. In
fact, we would have obtained the same result had σ and τ been elements
of S 11 , S 12 , . . . . Indeed viewed as elements of S 11 , the permutations σ and
τ, and the cycles (1, 6, 2, 5, 8, 3, 9) and (4, 10) all fix 11.
To compute σ−1 we start with σ = (1, 3, 10, 9)(2, 5, 6) and reverse the
arrows:
σ: 1 7→ 3 7→ 10 7→ 9 7→ 1, 2 7→ 5 7→ 6 7→ 2;
−1
σ : 1 ←[ 3 ←[ 10 ←[ 9 ←[ 1, 2 ←[ 5 ←[ 6 ←[ 2.
Therefore σ−1 = (1, 9, 10, 3)(2, 6, 5). Check for yourself that σσ−1 is indeed
the identity permutation. ♦
Exercise XIII.26. Let ρ and τ be as given in Exercise XIII.18. Write ρ and
τ as products of disjoint cycles.
Exercise XIII.27. Which of the following pairs of permutations are equal
elements of S 6 ?
98 XIII. SYMMETRIC GROUPS

(i) (1, 2, 3)(4, 6) and (6, 4)(2, 3, 1)(5).

(ii) (4, 5, 6)(1, 2, 3) and (3, 1, 2)(5, 4, 6).
Exercise XIII.28. Let ρ = (1, 2, 3)(4, 5) and τ = (1, 2, 3, 4). Write the follow-
ing in cycle notation (i.e. as a product of disjoint cycles): ρ −1 , τ−1 , ρτ,
τρ 2 .
Lemma XIII.29. Disjoint cycles commute. as promised
P ROOF. Let σ and τ be disjoint cycle in S n and write
σ = (a 1 , a 2 , . . . , a k ), τ = (b 1 , b 2 , . . . , b ` ).
Since σ and τ are disjoint a i 6= b j for i = 1, . . . , k and j = 1, . . . , `.
We want to show that στ = τσ. This means that στx = τσx for all
x ∈ {1, 2, . . . , n}. We subdivide into three cases:
Case 1: x does not equal any of the a i or b j . Then τx = x and σx = x.
Therefore
στx = σx = x = τx = τσx.
Case 2: x = a i for some i = 1, . . . , k. Thus x does not equal any of the b j ,
and so τx = x. Hence στx = σx = σa i = a i +1 ; here a k+1 is interpreted as
being a 1 . Let’s compute τσx. This is τσa i = τa i +1 = a i +1 since a i +1 does
not equal any of the b j . Hence στx = τσx.
Case 3: x = b j for some j = 1, . . . , `. This is similar to Case 2.
We conclude that στ = τσ as required. 
P ROOF OF T HEOREM XIII.23. Let ρ be an element of S n . Consider the
sequence
1, ρ1, ρ 2 1, ρ 3 1, . . .
Every term in this infinite sequence is contained in the finite set {1, 2, . . . , n}.
Thus the sequence must contain repetition. Let ρ u 1 be the first term
in the sequence that has already appeared. Thus ρ u 1 = ρ v 1 for some
0 ≤ v < u. Apply ρ −v to both sides. We obtain ρ u−v 1 = 1. Note that
0 < u − v ≤ u. If u − v < u, then ρ u−v 1 is in fact the first term in the
sequence that has already appeared, which contradicts our assumption.
Therefore, u − v = u and so v = 0. Hence ρ u 1 = 1, and 1, ρ1, . . . , ρ u−1 1 are
distinct.
Let µ1 be the cycle of length u
µ1 = 1, ρ1, ρ 2 1, . . . , ρ u−1 1 .
¡ ¢

It is clear that µ1 has the same effect as ρ on the elements 1, ρ1, . . . , ρ u−1 1.
Now let a be the first element of the set {1, 2, . . . , n} not appearing in
the list 1, ρ1, . . . , ρ u−1 1. Repeat the above argument with the sequence
a, ρa, ρ 2 a, ρ 3 a, . . . .
We deduce the existence of a cycle
µ2 = a, ρa, . . . , ρ v−1 a
¡ ¢
 XIII.7. PERMUTATIONS AND TRANSPOSITIONS 99

such that µ2 and ρ have the same effect on the elements a, ρa, . . . , ρ v−1 a.
Let us show that µ1 and µ2 are disjoint. Suppose otherwise. Then ρ i 1 =
ρ j a for some 0 ≤ i < u and 0 ≤ j < v. Now apply ρ v− j to both sides to
obtain ρ k 1 = a where k = i +v − j . This contradicts our assumption that a
does not appear in the list 1, ρ1, . . . , ρ u−1 1. Hence the cycles µ1 and µ2 are
disjoint. Now the product µ1 µ2 has the same effect as ρ on the elements
1, ρ1, . . . , ρ u−1 1, a, ρa, . . . , ρ v−1 a.
We repeat the process, starting with the first element of {1, 2, . . . , n} not
appearing in either cycle µ1 , µ2 to construct a µ3 that is disjoint from both
µ1 and µ2 , etc. As the set {1, 2, . . . , n} is finite, this process must terminate
eventually with some µr . The product of disjoint cycles µ1 µ2 . . . µr has the
same effect on {1, . . . , n} as ρ. Therefore

ρ = µ1 µ2 · · · µr .

Exercise XIII.30. We will shortly meet the Alternating Groups, one of which
is
A 3 = {id, (1, 2, 3), (1, 3, 2)}.
Verify that A 3 is a subgroup of S 3 , and write down its left cosets.
Exercise XIII.31. Verify that H = {id, (1, 2)} is a subgroup of S 3 , and write
down its left cosets.
Exercise XIII.32. (i) Use Lagrange’s Theorem to show that S 4 does
not have an element of order 5.
(ii) Let σ = (a 1 , a 2 , . . . , a m ) be a cycle of length m in S n . Explain why
σ has order m.
(iii) Now let ρ = σ1 σ2 . . . σk where the σi are disjoint cycles of lengths
m i in S n . Explain carefully why ρ has order lcm(m 1 , m 2 , . . . , m k ).
(iv) Show that S 4 does not have elements of order 6. Could you have
shown this using Lagrange’s Theorem?

XIII.7. Permutations and Transpositions

Lemma XIII.33. Every permutation can be written as a product of trans-
positions.
Note the absence of the word ‘disjoint’.

P ROOF. We know that every permutation can be written a product of cy-

cles. So it is enough to show that a cycle can be written as a product of
transpositions. Check for yourself that

(XIII.29) (a 1 , a 2 , . . . , a m ) = (a 1 , a m ) · · · (a 1 , a 3 )(a 1 , a 2 ).


100 XIII. SYMMETRIC GROUPS

Example XIII.34. Equation (XIII.29) gives a recipe for writing any cycle
as a product of transpositions. For example,
(1, 5, 9) = (1, 9)(1, 5).
Note that these transpositions are not disjoint and so they don’t have to
commute. Check that
(1, 9)(1, 5) 6= (1, 5)(1, 9).
One thing to be careful about is that decomposition of a permutation as
a product of transpositions is not in any way unique. For example, using
(XIII.29) we have
(1, 2, 3, 4) = (1, 4)(1, 3)(1, 2).
However, you can also check that
(1, 2, 3, 4) = (2, 3)(1, 3)(3, 5)(3, 4)(4, 5).
So we can write (1, 2, 3, 4) as a product of 3 transpositions and as a prod-
uct of 5 transpositions. Can we write it as a product of 4 transpositions?
Spend no more and no less than five minutes thinking about this. ♦

XIII.8. Even and Odd Permutations

Let n ≥ 2 be an integer. Let x 1 , x 2 , . . . , x n be variables, and let P n be the
polynomial
Y
Pn = (x i − x j ).
1≤i < j ≤n

The polynomial P n is called the n-th alternating polynomial. It will help

us to discover an important subgroup of S n called the alternating group
and denoted by A n . Let us write down the first three alternating polyno-
mials:
P 2 = x1 − x2 , P 3 = (x 1 − x 2 )(x 1 − x 3 )(x 2 − x 3 ),
P 4 = (x 1 − x 2 )(x 1 − x 3 )(x 1 − x 4 )(x 2 − x 3 )(x 2 − x 4 )(x 3 − x 4 ).
If σ ∈ S n then define
σ(P n ) =
Y
(x σi − x σ j ).
1≤i < j ≤n

Example XIII.35. Let σ = (1, 2) ∈ S 3 . Then

σ(P 3 ) = (x σ1 − x σ2 )(x σ1 − x σ3 )(x σ2 − x σ3 )
= (x 2 − x 1 )(x 2 − x 3 )(x 1 − x 3 )
= −P 3 .
We obtain the equality in the final step of the calculation by compar-
ing the factors of P 3 with the factors of σ(P 3 ), and not by expanding!
Note that the first factor of P 3 changed sign and the last two factors are
swapped. So σ(P 3 ) = −P 3 .
 XIII.8. EVEN AND ODD PERMUTATIONS 101

Now let τ = (1, 2, 3) ∈ S 3 . Then

τ(P 3 ) = (x τ1 − x τ2 )(x τ1 − x τ3 )(x τ2 − x τ3 )
= (x 2 − x 3 )(x 2 − x 1 )(x 3 − x 1 )
= P3.
Again we obtain equality by comparing factors. Write down ρ(P 3 ) for the
other four elements ρ ∈ S 3 . ♦
Lemma XIII.36. Let τ ∈ S n be a transposition. Then τ(P n ) = −P n .
The proof of the lemma is not hard. But it is very easy to get muddled
in this proof. So before you read on, try out lots of examples and drink
lots of coffee. Sit where no one can see you and try to prove it. If you
manage it, feel free to jump up and down from excitement—you deserve
it.
The examples we’ve done are quite basic, so let’s do a more serious
one.
Example XIII.37. Let τ = (2, 4) ∈ S 5 . We want to check that τ(P 5 ) = −P 5 .
Some factors of P 5 are unaffected. For example, τ(x 1 − x 3 ) = x τ1 − x τ3 =
x 1 −x 3 . The ones that aren’t affected are the ones that don’t contain either
of x 2 or x 4 . These are,
x1 − x3 , x1 − x5 , x3 − x5 .
We will split the other factors of P 5 into four groups 1:
(I) x1 − x2 , x1 − x4 ,
(II) x2 − x3 , x3 − x4 ,
(III) x2 − x5 , x4 − x5 ,
(IV) x2 − x4 .

Let’s study what τ does to each group. Note that

τ(x 1 − x 2 ) = x 1 − x 4 , τ(x 1 − x 4 ) = x 1 − x 2 .
Thus τ swaps the factors in group (I) whilst keeping their signs the same.
But
τ(x 2 − x 3 ) = x 4 − x 3 = −(x 3 − x 4 ), τ(x 3 − x 4 ) = x 3 − x 2 = −(x 2 − x 3 ).
Thus τ swaps the factors in group (II) and changes the sign of each. More-
over,
τ(x 2 − x 5 ) = x 4 − x 5 , τ(x 4 − x 5 ) = x 2 − x 5 .
So τ swaps the factors in group (III) whilst keeping their signs the same.
Finally,
τ(x 2 − x 4 ) = x τ2 − x τ4 = x 4 − x 2 = −(x 2 − x 4 ).
1The word “groups” here is used in its English language sense, not in its mathemat-
ical sense.
102 XIII. SYMMETRIC GROUPS

So the one factor in group (IV) simply changes sign. We see that τ(P 5 ) has
the same factors as P 5 with three sign changes: τ(P 5 ) = (−1)3 P 5 = −P 5 . ♦
P ROOF OF L EMMA XIII.36. Let τ = (`, m). The transposition (`, m) swaps
` and m, and keeps everything else fixed. In particular (`, m) = (m, `).
So we can suppose that ` < m. Any factor x i − x j where neither i nor j
is equal to ` nor m, is unaffected by τ. We pair off the other factors as
follows:


 x1 − x` , x1 − xm ,

x 2 − x ` ,

x2 − xm ,
(I) . .

 .. ..


x `−1 − x ` , x `−1 − x m ,



 x ` − x `+1 , x `+1 − x m ,

x ` − x `+2 , x `+2 − x m ,

(II) .. ..


 . .

x ` − x m−1 , x m−1 − x m ,



 x ` − x m+1 , x m − x m+1 ,

x ` − x m+2 , x m − x m+2 ,

(III) .. ..


 . .

x` − xn , xm − xn ,

n
(IV) x` − xm .

Now τ swaps each pair in (I), keeping the signs the same; it swaps each
pair in (II) and changes the sign of each; it swaps each pair in (III), keep-
ing the signs the same; it changes the sign of x ` −x m . So τ(P n ) has exactly
the same factors as P n , up to a certain number of sign changes. How
many sign changes? The number of sign changes is:
2(m − ` − 1) + 1.
The 1 is for changing the sign of x ` −x m . There are 2 sign changes coming
from each pair in (II). The number of such pairs is m − ` − 1. Since the
number of sign changes is odd, we see that τ(P n ) = −P n . 
Lemma XIII.38. If σ ∈ S n then σ(P n ) = ±P n . More precisely, if σ is a prod-
uct of an even number of transpositions then σ(P n ) = P n and if σ is a prod-
uct of an odd number of transpositions then σ(P n ) = −P n .
P ROOF. Recall, by Lemma XIII.33, that we can write every permutation as
a product of transpositions. Every transposition changes the sign of P n .
The lemma follows. 
Example XIII.39. We have noted in Example XIII.34 that the way we ex-
press a permutation as a product of transpositions is not unique. Indeed
 XIII.8. EVEN AND ODD PERMUTATIONS 103

we saw that
(1, 2, 3, 4) = (1, 4)(1, 3)(1, 2), (1, 2, 3, 4) = (2, 3)(1, 3)(3, 5)(3, 4)(4, 5).
So we can write (1, 2, 3, 4) as a product of 3 transpositions and as a prod-
uct of 5 transpositions. We asked the question of whether (1, 2, 3, 4) can
be written as a product of 4 transpositions? Write σ = (1, 2, 3, 4). From the
above lemma, we see that σ(P n ) = −P n . If we’re able to write σ as a prod-
uct of an even number of transpositions then σ(P n ) = P n . We would then
have P n = −P n which is a contradiction. Therefore we cannot write σ as
a product of 4 transpositions. ♦
You should now have no trouble in proving the following theorem.
Theorem XIII.40. Every permutation in S n can be written as a product of
either an even number of transpositions, or an odd number of transposi-
tions but not both.
We shall call a permutation even if we can write it as a product of an
even number of transpositions, and we shall call it odd if we can write it
as a product of an odd number of transpositions.
Example XIII.41. (1, 2, 3, 4) is an odd permutation because we can write
it as the product of 3 transpositions:
(1, 2, 3, 4) = (1, 4)(1, 3)(1, 2).
Indeed, a cycle of length n can be written as product of n − 1 transposi-
Beware! tions by (XIII.29). So a cycle of length n is even if n is odd, and it is odd if
n is even!
The permutation (1, 2, 3)(4, 5) is the product of an even permutation
which is (1, 2, 3) and an odd permutation which is the transposition (4, 5).
Thus (1, 2, 3)(4, 5) is an odd permutation.
What about the identity element id? Note that id(P n ) = P n , so id must
be even. We must be able to write it as a product of an even number of
transpositions. A mathematician would say that the identity element is
the product of zero transpositions, so it is even. If you find that kind of
reasoning disturbing, you have my sympathy. Instead, note that
id = (1, 2)(1, 2),
which does allow us to check that id is indeed even.
We now come to define a very important group. Let n ≥ 2. We define
the n-th alternating group to be
A n = {σ ∈ S n : σ is even}.
As usual, all we’ve done is specify a subset of S n which we’ve denoted by
A n and we must indeed show that A n is a group.
Theorem XIII.42. A n is a subgroup of S n .
P ROOF. We’ve already seen that the identity element id is even, so id ∈ A n .
If σ, ρ ∈ A n then we can write each as an even number of transpositions.
104 XIII. SYMMETRIC GROUPS

Therefore the product σρ can be written as an even number of transpo-

sitions (even+even=even). Hence σρ ∈ A n .
Finally we must show that the inverse of an even permutation is even.
Suppose σ is even. We can write
σ = τ1 τ2 . . . τ m
where the τi are transpositions, and m is even. Now
σ−1 = (τ1 τ2 · · · τm )−1
= τ−1 −1 −1
m τm−1 · · · τ1
= τm τm−1 · · · τ1 .
Here you should convince yourself that τ−1 = τ for any transposition τ.
Since m is even, we find that σ−1 is even and so σ−1 ∈ A n .
Hence A n is a subgroup of S n . 
Example XIII.43. Recall that S 2 = {id, (1, 2)}. We see that A 2 = {id} is the
trivial subgroup. ♦
Example XIII.44. Recall that S 3 has 3! = 6 elements:
S 3 = {id, (1, 2), (1, 3), (2, 3), (1, 2, 3), (1, 3, 2)}.
Then
A 3 = {id, (1, 2, 3), (1, 3, 2)}.
Note that S 3 is non-abelian, but you can check that A 3 is abelian. ♦
In the above examples we saw that A n has half the number of ele-
ments of S n for n = 2, 3. In fact, this pattern continues.
Theorem XIII.45. Let n ≥ 2. Then A n has order
1 n!
|A n | = |S n | = .
2 2
P ROOF. We know by Lagrange’s Theorem that
|S n | = [S n : A n ]|A n |.
To prove the theorem it is sufficient to show that the index [S n : A n ] = 2.
Fix a transposition τ (e.g. τ = (1, 2)). We shall show that the distinct cosets
of A n in S n are A n and τA n . It will then follow that the index [S n : A n ] = 2,
completing the proof.
We know that A n is the subset (indeed subgroup) of S n consisting of
all the even permutations. Thus τA n consists only of odd permutations.
Does τA n contain all the odd permutations? Suppose σ is odd. Then τσ
is even and is hence in A n . Therefore τ(τσ) is in the coset τA n . But
τ(τσ) = τ2 σ = σ,
since transpositions have order 2, and so σ ∈ τA n .
We have now shown that τA n is the set of all odd permutations, and
we know that A n is the set of all even permutations. Are there any other
cosets? If there were any they would have to overlap with either A n or
 XIII.8. EVEN AND ODD PERMUTATIONS 105

τA n , and we know that cosets are either disjoint or equal (Lemma XI.17).
So there aren’t any other cosets and the proof is complete. 
Exercise XIII.46. Let ρ and τ be as given in Exercise XIII.18. Write ρ and
τ as products of transpositions and state if they’re even or odd.
Exercise XIII.47. Write down the elements of A 3 and check that it is cyclic
(and hence abelian). Show that A n is non-abelian for n ≥ 4.
Exercise XIII.48. Let f be a polynomial in variables x 1 , x 2 , . . . , x n . Let σ be
a permutation in S n . We define σ( f ) to be the polynomial f (x σ1 , x σ2 , . . . , x σn ).
For example, if f = x 1 + x 22 + x 3 x 4 and σ = (1, 4)(2, 3) then σ swaps x 1 and
x 4 , and swaps x 2 and x 3 ; thus σ( f ) = x 4 + x 32 + x 2 x 1 . Compute σ( f ) for the
following pairs f , σ:
(i) f = x 12 − x 2 x 3 , σ = (1, 2, 3).
(ii) f = x 1 x 2 + x 3 x 4 , σ = (1, 3)(2, 4).
Exercise XIII.49. Let f be a polynomial in variables x 1 , . . . , x n .
(a) Let H be a subgroup of S n . We say that f is H -invariant if it
satisfies the property that σ( f ) = f for all σ ∈ H . We say that f is
symmetric if it is S n -invariant. Find a polynomial in x 1 , x 2 , x 3 , x 4
that is D 4 -invariant but not symmetric.

craving my daily algebra fix

(b) Define Fix( f ) = {σ ∈ S n : σ( f ) = f }. Show that Fix( f ) is a sub-
group of S n . Write down Fix( f ) for the following polynomials in
x1 , . . . , x4 :
(i) x 42 + x 1 x 2 x 3 .
(ii) x 1 x 2 + x 3 x 4 .

Exercise XIII.50. Let ρ and τ ∈ S n . Show that τ is even if and only if ρ −1 τρ

is even. (Hint: It will help to show that if ρ = c 1 c 2 · · · c m as a product of
transpositions, then ρ −1 = c m c m−1 . . . c 1 ).
Exercise XIII.51. This exercise concerns the 15-tile puzzle. The puzzle
consists of 15 square tiles (numbered 1, 2, . . . , 15) arranged in a 4×4 square
with one position blank. The initial arrangement of the tiles is as follows:

1 2 3 4

5 6 7 8

9 10 11 12

13 15 14

You can slide any tile adjacent to the blank into the position of the
blank. So starting from the initial arrangement there are two possible
moves:
106 XIII. SYMMETRIC GROUPS

1 2 3 4 1 2 3 4

5 6 7 8 5 6 7 8

9 10 11 9 10 11 12

13 15 14 12 13 15 14

In the 1880s—as a marketing ploy to improve the sales of the puzzle—

Sam Lloyd (an amateur mathematician) offered $1000 to anyone who can
reach:

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15

Show that this is impossible! You might want to follow these hints and a mathematical scam
tips:
(i) Think of the blank as a tile numbered 16. This way every re-
arrangement is a permutation on 1, . . . , 16 and so an element of
S 16 .
(ii) Observe that every move is a transposition involving 16.
(iii) Observe that to go from the initial arrangement to the desired
final arrangement, tile 16 must make the same number of moves
down as up, and the same number of moves right as left.
Can you use your knowledge of maths to think of other ways of ripping applied maths!
people off? This of course is a purely intellectual exercise. As citizens of
Warwick plc you are fine upright human beings who would not dream
of putting such ideas into practice. But DON’T share these ideas with
friends from less scrupulous universities. I don’t have a particular two
universities in mind.

Riveted?
 CHAPTER XIV

Rings

If groups took your breath away, wait till you meet rings.

XIV.1. Definition
A ring is a triple (R, +, ·), where R is a set and +, · are binary operations
on R such that the following properties hold
(i) (closure) for all a, b ∈ R, a + b ∈ R and a · b ∈ R;
(ii) (associativity of addition) for all a, b, c ∈ R

(a + b) + c = a + (b + c);

(iii) (existence of an additive identity element) there is an element

0 ∈ R such that for all a ∈ R,

a + 0 = 0 + a = a.

(iv) (existence of additive inverses) for all a ∈ R, there an element,

denoted by −a, such that

a + (−a) = (−a) + a = 0;

(v) (commutativity of addition) for all a, b ∈ R,

a + b = b + a;

(vi) (associativity of multiplication) for all a, b, c ∈ R,

a · (b · c) = (a · b) · c;

(vii) (distributivity) for all a, b, c ∈ R,

a · (b + c) = a · b + a · c; (b + c) · a = b · a + c · a;

(viii) (existence of a multiplicative identity) there is an element 1 ∈ R

so that for all a ∈ R,

1 · a = a · 1 = a.

Moreover, a ring (R, +, ·) is said to be commutative, if it satisfies the fol-

lowing additional property:
(ix) (commutativity of multiplication) for all a, b ∈ R,

a · b = b · a.
107
108 XIV. RINGS

Note that the word ‘commutative’ in the phrase ‘commutative ring’ refers
to multiplication. Commutativity of addition is part of the definition of
ring. Some textbooks omit property (viii) from the definition of a ring.
Those textbooks call a ring satisfying (viii) a ring with unity. We shall
always assume that our rings satisfy (viii).
Observe, from properties (i)–(v), if (R, +, ·) is a ring, then (R, +) is an
abelian group.

XIV.2. Examples
Example XIV.1. You know lots of examples of rings: Z, Q, R, C, R[x], etc.
All these examples are commutative rings.
Example XIV.2. Let
½µ ¶ ¾
a b
M 2×2 (R) = : a, b, c, d ∈ R .
c d
This is the set of 2 × 2 matrices with real entries. From the properties of
matrices it is easy to see that M 2×2 (R) is a ring with the usual addition
and multiplication of matrices. The additive identity is the zero matrix,
and the multiplicative identity is I 2 . The ring M 2×2 (R) is an example of a
non-commutative ring, as matrix multiplication is non-commutative.
Similarly we define M 2×2 (C), M 2×2 (Z), M 2×2 (Q). These are all non-
commutative rings. ♦
Theorem XIV.3. Let m be an integer satisfying m ≥ 2. Then Z/mZ is a
ring.
P ROOF. We really mean that (Z/mZ, +, ·) is a commutative ring. We’ve al-
ready seen that Z/mZ is closed under addition and multiplication, and
that (Z/mZ, +) is an abelian group. I leave you to ponder why the re-
maining properties (vi)–(ix) must be true. 
Example XIV.4. You’re familiar with the following two binary operations
on R3 : addition and the cross product (also known as the vector product).
Is (R3 , +, ×) a ring? No. First the cross product is not associative. For
example,
i × (j × j) = 0, (i × j) × j = −i.
We only need one of the properties (i)–(viii) to fail for us to conclude that
(R3 , +, ×) is not a ring. We know that (vi) fails. It is interesting to note that
(viii) fails too, as we now show. Indeed,
(XIV.30) a × b = −b × a.
Suppose 1 is a vector in R that satisfies
3

a×1 = 1×a = a
for all a ∈ R3 . From (XIV.30) we see that a = −a for all a ∈ R3 . This gives a
contradiction. Therefore (viii) fails too. ♦
 XIV.2. EXAMPLES 109

Example XIV.5. Consider (R[x], +, ◦), where ◦ is composition of polyno-

mials. Is this a ring? No. It is easy to see that all the required properties
hold except for distributivity (the “multiplicative identity” is the polyno-
mial f (x) = x). Let us give a counterexample to show that distributivity
fails. Let
f (x) = x 2 , g (x) = x, h(x) = x.
Then
f ◦ (g + h) = f (2x) = 4x 2 ; f ◦ g + f ◦ h = x 2 + x 2 = 2x 2 .
♦
Example XIV.6. The zero ring is the ring with just one element {0}. In
this ring 1 = 0, and there is only one possible definition of addition and
multiplication: 0 + 0 = 0, 0 · 0 = 0. The zero ring is not interesting.
Let R be a ring in which 1 = 0. Then a = a · 1 = a · 0 = 0 for all a ∈ R
and so R is the zero ring. To summarise a ring is the zero ring if and only
if 1 = 0.
Example XIV.7. Let’s step back a little and think about R2 . We know that
(R2 , +) is an abelian group. Is there a way of defining multiplication on R2
so that we obtain a ring? We will define two different multiplications that
make R2 into a ring. The first is rather obvious: we define
(a 1 , a 2 ) × (b 1 , b 2 ) = (a 1 b 1 , a 2 b 2 ).
With this definition, you can check that (R2 , +, ×) is a ring, where the mul-
tiplicative identity is 1 = (1, 1).
The other way is more subtle: we define
(XIV.31) (a 1 , a 2 ) × (b 1 , b 2 ) = (a 1 b 1 − a 2 b 2 , a 1 b 2 + a 2 b 1 ).
Where does this definition come from? Recall that R2 is represented geo-
metrically by the plane, and C is represented geometrically by the plane.
If we’re thinking of points in the plane as elements of R2 then we write
them as ordered pairs of real numbers: (a, b). If we’re thinking of points
in the plane as elements of C then we write them in the form a +i b where
again a, b are real numbers. We multiply in C using the rule
(XIV.32) (a 1 + i a 2 ) × (b 1 + i b 2 ) = (a 1 b 1 − a 2 b 2 ) + i (a 1 b 2 + a 2 b 1 ).
Notice that definitions (XIV.31), (XIV.32) are exactly the same at the level
of points on the plane. We’ve used the multiplicative structure of C to
define multiplication on R2 . With this definition, (R2 , +, ×) is a ring. What
is the multiplicative identity? It’s not (1, 1). For example (1, 1) × (1, 1) =
(0, 2). Think about the multiplicative identity in C. This is simply 1 = 1 +
0i . So the multiplicative identity in (R2 , +, ×) (with multiplication defined
as in (XIV.31)) is (1, 0). Check for yourself that
(a 1 , a 2 ) × (1, 0) = (1, 0) × (a 1 , a 2 ) = (a 1 , a 2 ).
This example is NOT IMPORTANT. Don’t lose any sleep over it. ♦
110 XIV. RINGS

XIV.3. Subrings
Just as we have subgroups, so we have subrings.
Definition. Let (R, +, ·) be a ring. Let S be a subset of R and suppose that
(S, +, ·) is also a ring with the same multiplicative identity. Then we say
that S is a subring of R (or more formally (S, +, ·) is a subring of (R, +, ·)).
For S to be a subring of R, we want S to a ring with respect to the same
two binary operations that makes R a ring, and 1R ∈ S where 1R is the
multiplicative identity of R.
Example XIV.8. Z is a subring of R; Q is a subring of R; Z is a subring of
Q; R is a subring of R[x]. ♦
Theorem VIII.5 gave a criterion for a subset of a group to be a sub-
group. As you’d expect we have a similar criterion for a subset of a ring to
be a subring.
Theorem XIV.9. Let R be a ring. A subset S of R is a subring if and only if
it satisfies the following conditions
(a) 0, 1 ∈ S (that is S contains the additive and multiplicative identity
elements of R);
(b) if a, b ∈ S then a + b ∈ S;
(c) if a ∈ S then −a ∈ S;
(d) if a, b ∈ S then ab ∈ S.
P ROOF. First re-read the proof of Theorem VIII.5. Then prove this theo-
rem on your own. It won’t take you long. 
Example XIV.10. In Example VIII.7, we saw that the set of even integers
2Z is a subgroup of Z. Strictly speaking, (2Z, +) is a subgroup of (Z, +).
Now we know that (Z, +, ·) is a ring. Is (2Z, +, ·) a subring? From Theo-
rem XIV.9 we see that it isn’t because 1 ∉ 2Z. ♦
Example XIV.11. In view of the previous example, let’s try to discover if Z
has any subrings other than itself. Let S be a subring of Z. We know that
0, 1 ∈ S. Also, by (b) we know that 2 = 1 + 1 ∈ S. Repeating the argument,
3 = 2 + 1 ∈ S and so on. By induction we know that 0, 1, 2, . . . are all in S.
But by (c), if a ∈ S then −a ∈ S. So . . . , −3, −2, −1 are also in S. Hence Z is
contained in S. But S is a subset of Z. So they must be equal: S = Z.
Therefore, the only subring of Z is Z itself. By contrast, in Section IX.2
we saw that Z has infinitely many subgroups. ♦
Exercise XIV.12. Let m be an integer satisfying m ≥ 2. Show that the only
subring of Z/mZ is Z/mZ itself.
Remark. The easiest way to show that a set is a ring is to show that it is A very important
a subring of a known ring. If you do this, you only have four properties tip!
to check (a),(b),(c),(d). If you don’t do this, you’ll have eight properties to
check (i)–(viii). The following two examples will help you appreciate this
principle.
 XIV.3. SUBRINGS 111

Example XIV.13. Let

Z[i ] = {a + bi : a, b ∈ Z}.
The set Z[i ] is called the set of Gaussian integers. Show that Z[i ] is a ring.
Answer. We can try checking the eight defining properties of a ring. How-
ever, we note that Z[i ] is contained in C. Indeed, it is the set of complex
numbers where the real and imaginary parts are integers. So let’s prove
that Z[i ] is a subring of C.
Now 0 = 0+0i , 1 = 1+0i are clearly in Z[i ]. Suppose α, β ∈ Z[i ]. Write
α = a1 + a2 i , β = b1 + b2 i ,
where a 1 , a 2 , b 1 , b 2 are integers. To apply Theorem XIV.9 we need to
check that α + β, −α and αβ are in Z[i ]. We note that
α + β = (a 1 + b 1 ) + (a 2 + b 2 )i , −α = −a 1 + (−a 2 )i ,
and
αβ = (a 1 a 2 − b 1 b 2 ) + (a 1 b 2 + a 2 b 1 )i .
Since we want to show that α+β, −α and αβ are in Z[i ], we want to show
that their real and imaginary parts are integers. Now as a 1 , a 2 , b 1 , b 2 are
integers, so are
a1 + b1 , a2 + b2 , −a 1 , −b 1 , a1 a2 − b1 b2 , a1 b2 + a2 b1 .
Hence α + β, −α and αβ are in Z[i ]. By Theorem XIV.9, we see that Z[i ] is
a subring of C. Since Z[i ] is a subring, it is a ring! ♦
Exercise XIV.14. Let S be a subring of Z[i ]. Suppose i ∈ S. Show that
S = Z[i ].
Example XIV.15. Let
na o
S= : a, r ∈ Z, r ≥ 0 .
2r
Show that S is a ring.
Answer. We shall follow the same strategy as the previous example. First
think of a ring that contains S. The elements of S are rational numbers
whose denominator is a power of 2; for example
7 −1 15 15
7= 0, , = 3
2 2 8 2
are elements of S. An obvious choice of a ring that contains S is Q, the
ring of rational numbers. So let’s show that S is a subring of Q. Clearly
0 = 0/20 and 1 = 1/20 are in S. Suppose α, β are elements of S. We can
write
a b
α= r , β= s,
2 2
where a, b, r , s ∈ Z and r , s ≥ 0. We want to check that α + β, −α and αβ
are in S. Note that
−a ab
−α = r , αβ = r +s .
2 2
112 XIV. RINGS

Clearly −α, αβ are in S, since −a, a +b, r , r +s are integers and r , r +s ≥ 0.

Now for the sum, we’ll assume without loss of generality that r ≥ s. Then
a + 2r −s b
α+β = .
2r
Now since a, b, r , s are integers and r ≥ s, we have a + 2r −s b is also an in-
teger. Clearly, α+β is in S. By Theorem XIV.9, S is a subring and therefore
a ring. ♦
Exercise XIV.16. Let
Z[2i ] = {a + 2bi : a, b ∈ Z}.
Show that Z[2i ] is a subring of Z[i ]. Is {2a + 2bi : a, b ∈ Z} a subring of
Z[i ]?
Exercise XIV.17. Which of the following are subrings of M 2×2 (R)? If so,
are they commutative?
½µ ¶ ¾
a b
(i) : a, b, c ∈ R .
0 c
½µ ¶ ¾
a b
(ii) : a, b ∈ R .
0 0
½µ ¶ ¾
a b
(iii) : a, b ∈ R .
0 1
½µ ¶ ¾
a 0
(iv) : a ∈ R, b ∈ Z .
0 b
½µ ¶ ¾
a b
(v) : a, b ∈ R .
−b a
(vi) {A ∈ M 2×2 (R) : det(A) = 1}.

XIV.4. The Unit Group of a Ring

Recall that we defined R∗ , Q∗ , C∗ be removing from R, Q, C the zero
element; e.g.
R∗ = {a ∈ R : a 6= 0}.
We found that R∗ is group with respect to multiplication. In Example IV.4
we tried to do the same with Z and failed to obtain a group. Note that R,
Q, C are rings and so is Z. Given a ring, is there a naturally defined subset
that is a group with respect to multiplication? It turns out that the answer
is yes, and that for R, Q and C we obtain R∗ , Q∗ , C∗ as we’d expect. To
define this subset, we need the concept of a unit.
Definition. Let R be a ring. An element u is called a unit if there is some
element v in R such that uv = vu = 1. In other words, an element u of R
is a unit if it has a multiplicative inverse that belongs to R.
Example XIV.18. In any non-zero ring, 0 is a non-unit. ♦
Example XIV.19. In R, Q, C, every non-zero element has a multiplicative
inverse. So the units are the non-zero elements. ♦
 XIV.4. THE UNIT GROUP OF A RING 113

Example XIV.20. What are the units in Z? Suppose u is a unit in Z. Then

there is some v ∈ Z such that uv = vu = 1. This means that 1/u is an
integer. The only integers u such that 1/u is also an integer are ±1. So the
units in Z are ±1. ♦
Example XIV.21. Recall that R[x] is the ring of polynomials with real co-
efficients. Then x is not a unit, since 1/x is not a polynomial. However, 2
is a unit, since 1/2 is a polynomial in R[x] with real coefficients:
1 1
= + 0x.
2 2
♦
We can now answer the question posed above.
Definition. Let R be a ring. We define the unit group of R to be the set 1
(XIV.33) R ∗ = {a ∈ R : a is a unit in R}.
Just because we’ve called R ∗ the unit group of R doesn’t get us out of
checking that it is really a group.
Lemma XIV.22. Let (R, +, ·) be a ring and let R ∗ be the subset defined in (XIV.33).
Then (R ∗ , ·) is a group.

P ROOF. We must first show that R ∗ is closed under multiplication. Sup-

pose u 1 , u 2 ∈ R ∗ . Thus u 1 , u 2 are units of R, and so there are v 1 , v 2 ∈ R
such that
(XIV.34) u 1 v 1 = v 1 u 1 = 1, u 2 v 2 = v 2 u 2 = 1.
We want to show that u 1 u 2 is a unit. Note that v 2 v 1 ∈ R since R is closed
under multiplication (it’s a ring after all). Moreover,
(u 1 u 2 )(v 2 v 1 ) = u 1 (u 2 v 2 )v 1 associativity of multiplication
= u1 · 1 · v 1 since u 2 v 2 = 1
=1 since u 1 v 1 = 1.

Similarly (v 2 v 1 )(u 1 u 2 ) = 1. Thus u 1 u 2 is a unit 2 in R, and so u 1 u 2 ∈ R ∗ .

We’ve proved that R ∗ is closed under multiplication.

1Functorially-inclined mathematicians write R × instead of R ∗ . I happen to be

functorially-disinclined, but I do use their notation when I’m feeling pretentious.
2Start again. We have u , u are units and so satisfy (XIV.34) for some v , v in R. We
1 2 1 2
want to show that u 1 u 2 is a unit. What is wrong with the following argument?

 (u 1 u 2 )(v 1 v 2 ) = (u 1 v 1 )(u 2 v 2 ) = 1 · 1 = 1.

Similarly (v 1 v 2 )(u 1 u 2 ) = 1. Thus u 1 u 2 is a unit.

offence intended Do I distress you by repeatedly exhibiting such offences against mathematical de-
cency? Do you feel that these notes are degenerating into page after page of perver-
sion and blasphemy? I am sorry; I simply want you to join me in condemning these
abominations.
114 XIV. RINGS

We want to show that multiplication is associative in R ∗ . But multi-

plication is associative in R since R is a ring. Therefore it is associative in
R ∗.
Since 1 · 1 = 1, 1 is a unit and so 1 ∈ R ∗ .
Finally we want to show that every element in R ∗ has a multiplicative
inverse that belongs to R ∗ . Suppose u ∈ R ∗ . Then uv = vu = 1 for some
v ∈ R. Note that this makes v also a unit, and so v ∈ R ∗ . Thus u has a
multiplicative inverse in R ∗ . This completes the proof that R ∗ is a group.

Example XIV.23. Note that R∗ , C∗ , Q∗ have exactly the same meaning as
before. ♦
Example XIV.24. We showed that the units of Z are ±1. Therefore the
unit group of Z is
Z∗ = {1, −1}.
♦
Example XIV.25. Recall that M 2×2 (R) is the ring of 2×2 matrices with real
entries. It is clear from the definition of a unit, that the units of M 2×2 (R)
are the invertible matrices. In other words, they are the ones having non-
zero determinant. Thus
(M 2×2 (R))∗ = GL2 (R).
Similarly,
(M 2×2 (Q))∗ = GL2 (Q), (M 2×2 (C))∗ = GL2 (C).
What about the unit group of M¡2×2 (Z)? ¢ This is more complicated. For
3 1
example, consider the matrix A = 1 1 . The matrix A is invertible, and
A −1 = −1/2
¡ 1/2 −1/2 ¢
3/2 . Although A is in M 2×2 (Z), its inverse is not in M 2×2 (Z),
but it is in M 2×2 (Q) and M 2×2 (R). Thus A is a unit in M 2×2 (Q), and M 2×2 (R)
but not in M 2×2 (Z). The problem is clear: when calculating the inverse of
a matrix, we must divide by its determinant, and the result does not have
to be an integer.
Let’s go back to the definition of a unit. Suppose A ∈ M 2×2 (Z) is a unit.
Then there is a matrix B ∈ M 2×2 (Z) such that
AB = B A = I 2 .
Taking determinants, are recalling that det(AB ) = det(A) det(B ) we find
that
det(A) det(B ) = 1.
Now det(A) and det(B ) are integers because A and B have integer entries.
Thus
det(A) = det(B ) = 1, or det(A) = det(B ) = −1.
Conversely if A ∈ M 2×2 (Z) has determinant ±1, then its inverse will have
integer entries and so A is a unit. We deduce that
(M 2×2 (Z))∗ = {A ∈ M 2×2 (Z) : det(A) = ±1} .
 XIV.5. THE UNIT GROUP OF THE GAUSSIAN INTEGERS 115

We define the group GL2 (Z) by

GL2 (Z) = {A ∈ M 2×2 (Z) : det(A) = ±1} ;
then (M 2×2 (Z))∗ = GL2 (Z). In fact, for a commutative ring R we define
GL2 (R) = A ∈ M 2×2 (R) : det(A) ∈ R ∗ .
© ª

You will easily see that this is consistent with the earlier definitions of
GL2 (R), GL2 (C), GL2 (Q) and GL2 (Z), and that moreover, (M 2×2 (R))∗ = GL2 (R).
Example XIV.26. Let
½µ ¶ ¾
a b
S= : a, b, c ∈ Z .
0 c
Show that S is a ring under the usual addition and multiplication of ma-
trices. Compute S ∗ .
Answer: To show that S is a ring it is enough to show that it is a subring
of M 2×2 (Z). We leave that as an easy exercise.
Let us compute the unit group. Suppose A = a0 bc is in S. To be unit
¡ ¢

it is not enough for this matrix to be invertible, we also want the inverse
to belong to S. So we require the determinant ac to be non-zero and we
want µ ¶ µ ¶
−1 1 c −b 1/a −b/ac
A = =
ac 0 a 0 1/c
to belong to S. Thus we want the integers a, b, c to satisfy
1 1 b
ac 6= 0, , ,− ∈ Z.
a c ac
This happens precisely when a = ±1 and c = ±1. Thus
½µ ¶ ¾
∗ ±1 b
S = :b∈Z .
0 ±1
♦
Exercise XIV.27. In Example XIV.15, we showed that
na o
S = r : a, r ∈ Z, r ≥ 0
2
is a ring. Find its unit group.

XIV.5. The Unit Group of the Gaussian Integers

The Gaussian integers Z[i ] resemble the usual integers Z in many
ways. For example, you know that every non-zero integer can be written
r r
as ±1 · p 11 . . . p nn where the p i are distinct primes, and this representation
is unique (up to reordering the primes). This is the Unique Factoriza-
tion Theorem. The Gaussian integers have their own Unique Factoriza-
tion Theorem, which we don’t have time to cover, but you can look for-
a profound example ward to doing this in Algebra II. For now, we want to determine the unit
116 XIV. RINGS

group of Z[i ]. The most elegant way of doing this is via the norm map.
We define the norm map N : Z[i ] → Z by
N (a + bi ) = a 2 + b 2 , a, b ∈ Z.
The norm map is multiplicative:
Lemma XIV.28. Let α, β ∈ Z[i ]. Then N (αβ) = N (α)N (β).
P ROOF. α and β are complex numbers, and you can see that N (α) = |α|2 .
From the properties of the absolute value you know that |αβ| = |α| · |β|.
The lemma follows. 
Theorem XIV.29. The unit group of Z[i ] is {1, −1, i , −i }.
In other words, (Z[i ])∗ = U4 , the group of fourth-roots of unity.
P ROOF. We want the units of Z[i ]. Let α be a unit. Then there is some
β ∈ Z[i ] such that 1 αβ = 1. Applying the norm map, and recalling that it
is multiplicative, we see that
N (α)N (β) = N (αβ) = N (1) = 1.
Now N (α) and N (β) are in Z (go back to the definition of the norm map
to see this), and they multiply to give 1. So
N (α) = N (β) = 1, or N (α) = N (β) = −1.
Write α = a + bi where a, b are in Z. Then a 2 + b 2 = N (α) = ±1. Of course
−1 is impossible, so a 2 + b 2 = 1. But a, b are integers. So (a, b) = (±1, 0) or
(0, ±1). Hence α = a + bi = ±1 or ±i . Clearly ±1, ±i are units. So the unit
group is
Z[i ]∗ = {1, −1, i , −i }.

Remark. Compare the above proof to our determination of the unit group
of M 2×2 (Z) in Example XIV.25. I hope you agree that the similarities are
striking!
Exercise XIV.30. In Exercise XIV.16 you met the ring Z[2i ]. Find its unit
group. (Hint: Show first that any unit in Z[2i ] is a unit in Z[i ].)
p p p
Exercise XIV.31. pLet Z[ 2] = {a + b 2 : a, b ∈ Z}. Show that Z[ 2] is a
ring
p and that 1 + 2 is a unit. What is its order as an element of the group
Z[ 2]∗ ?
Exercise XIV.32. Let ζ = e 2πi /3 (this is a cube root of unity). Check that
ζ = ζ2 . Let Z[ζ] = {a + bζ : a, b ∈ Z}.
(i) Show that ζ2 ∈ Z[ζ] (Hint: the sum of the cube roots of unity is
. . . ).
(ii) Show that Z[ζ] is a ring.
1We could have written αβ = βα = 1. But Z[i ] is a commutative ring, so writing
αβ = 1 is enough.
 XIV.5. THE UNIT GROUP OF THE GAUSSIAN INTEGERS 117

(iii) Show that ±1, ±ζ and ±ζ2 are units in Z[ζ].

(iv) (Harder) Show that Z[ζ]∗ = {±1, ±ζ, ±ζ2 }.
(v) Show that this group is cyclic.

Algebra: a legal high.

 CHAPTER XV

Fields

A field (F, +, ·) is a commutative ring such that every non-zero element

is a unit. Thus a commutative ring F is a field if and only if its unit group
is
F ∗ = {a ∈ F : a 6= 0}.
Example XV.1. R, C, Q are fields. ♦
Example XV.2. Z is not a field, since for example 2 ∈ Z is non-zero but
not a unit. ♦
Example XV.3. R[x] is not a field, since for example x ∈ R[x] is non-zero
but not a unit. ♦
Example XV.4. Show that
Q[i ] = {a + bi : a, b ∈ Q}
is a field.
Answer: First we have to show that Q[i ] is a commutative ring. For this
it is enough to show that Q[i ] is a subring of C. It is clearly a subset of C
that contains 0 and 1. Suppose α, β ∈ Q[i ]. We want to show that α + β,
αβ, −α are all in Q[i ]. Write
α = a + bi , β = c + di
where a, b, c, d ∈ Q. Then
α + β = (a + c) + (b + d )i .
Since Q is closed under addition, a + c and b + d ∈ Q. So α + β ∈ Q[i ].
Similarly, check for yourself that αβ and −α are in Q[i ]. Thus Q[i ] is a
subring of C and so a ring 2.
Finally we have to show that every non-zero element of Q[i ] is a unit.
Suppose α is a non-zero element of Q[i ]. We can write α = a +bi where a,
b ∈ Q, and not both zero. We want to show that existence of some β ∈ Q[i ]
such that αβ = βα = 1. In other words, we want to show that 1/α is in
Q[i ]. But we know how to compute 1/α. Recall that to divide complex
2We could’ve made the proof more tedious by writing

r u k m
α= + i, β = + i,
s v ` n
where r , s, u, v, k, `, m, n are integers and s, v, `, n are non-zero. This would’ve worked,
but why do it? Get used to thinking of rational numbers as numbers in their own right!
119
120 XV. FIELDS

numbers we multiply the numerator and denominator by the conjugate

of the denominator:
1 1
=
α a + bi
1 a − bi
= ·
a + bi a − bi
a − bi
= 2
a + b2
a b
= 2 2
− 2 i.
a +b a + b2
As a, b are rationals, so are a/(a 2 + b 2 ) and b/(a 2 + b 2 ). So 1/α is in Q[i ].
Therefore Q[i ] is a field. ♦
p p p
Exercise XV.5. Let Q[ 2] = {a +b 2 : a, b ∈ Q}. Show that Q[ 2] is a field.
Exercise XV.6. Let
a : a, b ∈ R .
©¡ a b ¢ ª
F = −b
(a) Show that F is a field (under the usual addition and multiplica-
tion of matrices). (Hint: Begin by showing that F is a subring of
M 2×2 (R). You need to also show that F is commutative and that
every non-zero element has ¡ aanbinverse in F .)
(b) Let φ : F → C be given by φ −b a = a +bi . Show that φ is a bijec-
¢

tion that satisfies φ(A + B ) = φ(A) + φ(B ) and φ(AB ) = φ(A)φ(B ).

(c) Show that
F 0 = −b a : a, b ∈ C
©¡ a b ¢ ª

is not a field.
 CHAPTER XVI

Congruences Revisited

We saw that there are two binary operations defined on Z/mZ, ad-
dition and multiplication. These make (Z/mZ, +, ·) a commutative ring,
and (Z/mZ, +) a cyclic group of order m. We want to know about the unit
group of Z/mZ.

XVI.1. Units in Z/mZ

Example XVI.1. Find the unit groups of Z/mZ for m = 2, 3, 4, 5, 6.
Answer: You don’t have be very clever here! Just look at the multiplication
table for Z/6Z in Example VI.3 and you’ll see that
(Z/6Z)∗ = 1, 5 .
© ª

In the same way you’ll find that

(Z/2Z)∗ = 1 , (Z/3Z)∗ = 1, 2 ,
© ª © ª

(Z/4Z)∗ = 1, 3 , (Z/5Z)∗ = 1, 2, 3, 4 .
© ª © ª

In particular, Z/2Z, Z/3Z and Z/5Z are fields and Z/4Z, Z/6Z are not
fields. Can you make a general guess as to which Z/mZ are fields and
which aren’t? Can you prove your guess? ♦
Theorem XVI.2. Let a ∈ Z/mZ. Then a is a unit in Z/mZ if and only if
gcd(a, m) = 1. Thus
(Z/mZ)∗ = a : 0 ≤ a ≤ m − 1 and gcd(a, m) = 1 .
© ª

P ROOF. Suppose a is a unit in Z/mZ. Then there is some b in Z/mZ so

that ab ≡ 1 (mod m). Thus, there is some k ∈ Z such that ab − 1 = km.
Write g = gcd(a, m). Then g | a and g | m. So g | (ab − km) = 1. But this
means that g = 1.
Conversely, suppose gcd(a, m) = 1. By Euclid’s Algorithm, we know
that we can write 1 = ba + cm for some integers b, c ∈ Z. Thus ab ≡ 1
(mod m). Hence a is a unit. 
Exercise XVI.3. Redo Example XVI.1 using Theorem XVI.2.
Example XVI.4. By Theorem XVI.2, we know that 19 is invertible in Z/256Z.
But the statement of the theorem does not tell us how to find the inverse.
It would take us a very long to run through the elements u ∈ Z/256Z and
crucial point check to see if 19u ≡ 1 (mod 256). However, the proof of the theorem
does give us a recipe for finding the inverse. We know by factoring that
121
122 XVI. CONGRUENCES REVISITED

1
gcd(19, 256) = 1, but let’s use Euclid’s Algorithm to write 1 as a linear
combination of 19 and 256:
256 = 13 × 19 + 9
19 = 2 × 9 + 1.
Thus
1 = 19 − 2 × 9 = 19 − 2 × (256 − 13 × 19) = (1 − 2 × −13) × 19 − 2 × 256,
so
1 = 27 × 19 − 2 × 256.
Hence 27 × 19 ≡ 1 (mod 256), so 27 is the inverse of 19 in Z/256Z. ♦

XVI.2. Fermat’s Little Theorem

Through the computations you’ve done so far, you’ve probably con-
jectured the following.
Theorem XVI.5. Let p be a prime. Then Z/pZ is a field. Therefore,
(Z/pZ)∗ = {1, 2, . . . , p − 1}.
P ROOF. We already know that Z/mZ is a commutative ring for any integer
m ≥ 2. Now to show that Z/pZ is a field, we must show that any non-
zero a ∈ Z/pZ is invertible. But if a ∈ Z/pZ is non-zero, then a is one of
1, 2, . . . , p −1. Clearly a is not divisible by p. Since p is prime, gcd(a, p) = 1.
Hence by Theorem XVI.2, a is invertible in Z/pZ. This shows that Z/pZ
is a field. 
Exercise XVI.6. Prove the converse of Theorem XVI.5: if Z/mZ is a field
then m is prime.
Theorem XVI.7. (Fermat’s Little Theorem) Let p be a prime and a an in-
teger such that p - a. Then
(XVI.35) a p−1 ≡ 1 (mod p).
P ROOF. We know that a ≡ b (mod p) where b is one of 0, 1, 2, . . . , p − 1.
Now as p - a, we see that b 6= 0. By Theorem XVI.5, b is in the unit group
of Z/pZ which is
(Z/pZ)∗ = {1, 2, . . . , p − 1}.
The order of the group (Z/pZ)∗ is clearly p − 1. By Corollary VII.13 (the
corollary to Lagrange’s Theorem),
p−1
b = 1.
p−1
Thus b ≡ 1 (mod p). Since a ≡ b (mod p), we obtain (XVI.35) 
1It is easy to get muddled in the substitutions involved in Euclid’s Algorithm. One
way to reduce the muddle is to somehow distinguish the numbers you started with, here
256 and 19, and the remainders from the quotients. I did the distinguishing by writing
the numbers we started with and the remainders in boldtype. In your calculations, you
can underline them.
 XVI.3. EULER’S THEOREM 123

Here’s a fun application of Fermat’s Little Theorem.

Example XVI.8. Compute 21000 (mod 13).
Answer: Since 13 is prime and 13 - 2, we know by Fermat’s Little Theorem
that 212 ≡ 1 (mod 13). Now by the Division Algorithm,
1000 = 83 × 12 + 4.
Therefore,
21000 = 283×12+4 = (212 )83 × 24 ≡ 183 × 16 ≡ 3 (mod 13).
♦

XVI.3. Euler’s Theorem

Definition. Let m ≥ 1. We denote the order of the group (Z/mZ)∗ by
ϕ(m). The function ϕ is called Euler’s ϕ-function.
Example XVI.9. We know that if p is a prime, then (Z/pZ)∗ = {1, 2, . . . , p − 1},
and so ϕ(p) = p − 1. ♦
Example XVI.10. We know that
(Z/6Z)∗ = {1, 5},
and so ϕ(6) = 2. ♦
Example XVI.11. Let n ≥ 1. Then (Z/2n Z)∗ consists of a with a in the
range 0 ≤ a ≤ 2n − 1 that are coprime to 2n . These are the odd integers a
in the range 0 ≤ a ≤ 2n − 1. Thus
(Z/2n Z)∗ = {1, 3, . . . , 2n − 1}.
Hence ϕ(2n ) = 2n−1 . ♦
Theorem XVI.12. (Euler’s Theorem) Let m be an integer satisfying m ≥ 2.
Let a be an integer such that gcd(a, m) = 1. Then
a ϕ(m) ≡ 1 (mod m).
P ROOF. This has almost the same proof as Fermat’s Little Theorem. I’ll
leave the necessary modifications as an easy exercise. 
You’re probably wondering if there is a formula for ϕ(m), and in fact
there is.
Proposition XVI.13. Write
r r
m = p 11 · · · p kk
where p 1 , . . . , p k are distinct primes and r 1 , . . . , r k are positive integers. Then
r r −1 r r −1
ϕ(m) = (p 11 − p 11 ) · · · (p kk − p kk ).
The proof of Proposition XVI.13 is not difficult, but it is a little long
and we shall skip it.
Exercise XVI.14. Use Euler’s Theorem to compute 21000 (mod 63).
124 XVI. CONGRUENCES REVISITED

Exercise XVI.15. It is known that (Z/mZ)∗ is cyclic if m = 2, 4, p a or 2p a

where p is an odd prime. For all other m ≥ 2, the unit group (Z/mZ)∗ is
not cyclic. For more on this, do Number Theory in term 3. For now, check
that (Z/7Z)∗ is cyclic, but (Z/8Z)∗ is not cyclic.
Exercise XVI.16. Use Lagrange’s Theorem to show that ϕ(m) is even for
m ≥ 3.

Vale Dicere
XVI.4.
With tear-filled eyes I say goodbye, and begin to suffer the heart-rending a sincere outpouring
pangs of separation . . . of grief

Exercise XVI.17. Write a 5000 word essay on how abstract algebra has
changed your outlook on life, detailing the insights you have gained into
the great challenges facing/menacing humanity 1.
Exercise XVI.18. “Veneration of abstract algebra is at the root of contem-
porary mathematical decline.” Discuss.

1Here’s a good way to start your essay: “Unbeknownst to me, I misspent the first 18
years of my life wallowing in a cesspool of intellectual stagnation, until week six of term
1 when the Abstract Algebra lectures started . . . ”.