Algebra-2: Groups and Rings

Dmitriy Rumynin∗

January 6, 2011

How to use these notes

The lecture notes are split into 27 sections. Each section will be discussed
in one lecture, making every lecture self-contained. This means that the
material in a section may be reshuffled or even skipped for the lecture,
although the numeration of propositions will be consistent. The remaining
(up to 3) lectures will be spent on revisions and exercises including past
exams.
These written notes is an official curriculum: anything in them except
vistas can appear on the exam. Each section contains exercises that you
should do. To encourage you doing them, I will use some of the exercises in
the exam.
Vista sections are not assessed or examined in any way. Skip them if
you are allergic to nuts or psychologically fragile! The vistas are food for
further contemplation. A few of them are sky blue, but most are second year
material that we don’t have time to cover. You are encouraged to expand
one of them into your second year essay.
The main recommended book is Concrete Abstract Algebra by Lauritzen.
It is reasonably priced (£25 new, £11 used on Amazon), mostly relevant
(except chapter 5) and quite thin. The downside of the book is brevity of
exposition and some students prefer more substantial books. An excellent
UK-style textbook is Introduction to Algebra by Cameron (from £17 on
Amazon). Another worthy book is Algebra by Artin (£75 on Amazon for
the new edition but you can get older editions for around £45). This one
will have all details you need and much more than you can bear. Most
of algebraists I asked have started with this book and absolutely love it.
A similarly priced (and better according to some) alternative to Artin is
∗
c
Dmitriy Rumynin 2007

1
Abstract Algebra by Dummit and Foote (£55). My own first book was
Algebra by van der Waerden (£30 for each volume on Amazon) but it appears
that most of the mathematicians who hate Algebra in their later life have
started with it.
An alternative strategy is to get two books: one for rings and one for
groups. Virtually any pair of books will cover all the topics in these lecture
notes, although some interaction between subjects will be missing.
If you see any errors, misprints, oddities of my English, send me an
email. If you think that some bits require better explanation, write me as
well. All the contributions will be acknowledged. I would like to thank sec-
ond year students of 2007 Rupesh Bhudia, Iain Embrey, Alexander Illing-
worth, Matthew Hutton, Philip Jackson, Sebastian Jorn, Karl Pountney,
Jack Shaw-Dunn, Gareth Speight, Mohamed Swed and Jason Warnett for
valuable suggestions how to improve these lecture notes.

Contents
1 Groups and subgroups 3

2 Rings and subrings 8

3 Isomorphisms and symmetric groups 12

4 Generators, cyclic groups, quaternionic group 16

5 Orthogonal and dihedral groups 22

6 Equivalence relations and cosets 27

7 Lagrange’s theorem and applications 31

8 Normal subgroups 34

9 Homomorphisms 38

10 Quotient groups 44

11 Ideals and quotient rings 49

12 Chinese remainder theorem 53

2
13 Actions, stabilisers and alternating group 57

14 Orbits 62

15 Fixed points 66

16 Conjugacy classes 69

17 Conjugacy classes in Sn and An 73

18 Domains, divisibility and PID-s 76

19 Euclidean domains 80

20 Unique factorisation domains 84

21 Polynomials over fields 87

22 Gaussian primes 90

23 Fractions and Gauss lemma 92

24 Polynomials over UFD 97

25 Algebras and division rings 101

26 Quaternions, algebraic properties 104

27 Quaternions and spinors 111

1 Groups and subgroups

We define groups and establish their elementary properties. We learn
how to define a group by a multiplication table and how to build new groups
by using direct products and subgroups.
1.1 Definition of a group
Definition. A group is a set G together with a binary operation ◦ : G×G →
G that satisfies the following properties:
(i) (Closure) For all g, h ∈ G, g ◦ h ∈ G;
(ii) (Associativity) For all g, h, k ∈ G, (g ◦ h) ◦ k = g ◦ (h ◦ k);
(iii) There exists an element e ∈ G such that:

3
 (a) (Identity) for all g ∈ G, e ◦ g = g; and
(b) (Inverse) for all g ∈ G there exists h ∈ G such that h ◦ g = e.
(Actually Property (i) does not really need stating, because it is implied
by the fact that ◦ : G × G → G is a binary operation on G. But it is
traditionally the first of the four group axioms, so we have included it here!)
The number of elements in G is called the order of G and is denoted by
|G|. This may be finite or infinite.
An element e ∈ G satisfying (iii) of the definition is called an identity
element of G, and for g ∈ G, an element h that satisfies (iii)(b) of the
definition (h ◦ g = e) is called an inverse element of g.
We shall immediately prove two technical lemmas, which are often in-
cluded as part of the definition of a group.

Lemma 1.1 Let G be a group, let e ∈ G be an identity element, and for

g ∈ G, let h ∈ G be an inverse element of g. Then g ◦ e = g and g ◦ h = e.
Proof: We have h ◦ (g ◦ e) = (h ◦ g) ◦ e = e ◦ e = e = h ◦ g. Now let h′ be
an inverse of h. Then multiplying the left and right sides of this equation
on the left by h′ and using associativity gives (h′ ◦ h) ◦ (g ◦ e) = (h′ ◦ h) ◦ g.
But (h′ ◦ h) ◦ (g ◦ e) = e ◦ (g ◦ e) = g ◦ e, and (h′ ◦ h) ◦ g = e ◦ g = g, so we
get g ◦ e = g.
We have h ◦ (g ◦ h) = (h ◦ g) ◦ h = e ◦ h = h, and multiplying on the left
by h′ gives (h′ ◦ h) ◦ (g ◦ h) = h′ ◦ h. But (h′ ◦ h) ◦ (g ◦ h) = e ◦ (g ◦ h) = g ◦ h
and (h′ ◦ h) = e, so g ◦ h = e. 2

Lemma 1.2 Let G be a group. Then G has a unique identity element, and
any g ∈ G has a unique inverse.
Proof: Let e and f be two identity elements of G. Then, e ◦ f = f , but
by Lemma 1.1, we also have e ◦ f = e, so e = f and the identity element is
unique.
Let h and h′ be two inverses for g. Then h ◦ g = h′ ◦ g = e, but by
Lemma 1.1 we also have g ◦ h = e, so

h = e ◦ h = (h′ ◦ g) ◦ h = h′ ◦ (g ◦ h) = h′ ◦ e = h′

and the inverse of g is unique. 2

Definition. A group is called abelian or commutative if it satisfies the ad-
ditional property:
(Commutativity) For all g, h ∈ G, g ◦ h = h ◦ g.
We shall now proceed to change notation. The groups in this course will
either be:

4
 • multiplicative groups, where we omit the ◦ sign (g ◦h becomes just gh),
we denote the identity element by 1 rather than by e, and we denote
the inverse of g ∈ G by g−1 ; or
• additive groups, where we replace ◦ by +, we denote the identity ele-
ment by 0, and we denote the inverse of g by −g.
If there is more than one group around, and we need to distinguish
between the identity elements of G and H say, then we will denote them by
1G and 1H (or 0G and 0H ).
Additive groups will always be commutative, but multiplicative groups
may or may not be commutative. The default will be to use the multiplica-
tive notation.
The proof of the next lemma is not in the lecture. Try proving it yourself
before reading my proof. From now on, this result will be used freely and
without explicit reference.

Lemma 1.3 Let g, h be elements of the multiplicative group G. Then (gh)−1 =

h−1 g−1 .
Proof: Let us confirm by a calculation that h−1 g−1 is the inverse of gh.
Indeed, (h−1 g−1 ) · (gh) = h−1 (g−1 · g)h = h−1 h = e. 2
1.2 Examples – from Algebra-1
Actually, you already know a plenty of examples of groups from Algebra-
1. A vector space V is an additive group under addition.
Finitely generated abelian groups are defined using generators and rela-
tions in Algebra-1. One particular group is < x|nx = 0 >, the cyclic group
of order n. We denote the cyclic group Cn and always write in the default
multiplicative notation, so that Cn =< x|xn = 1 >. The infinite cyclic
group is denoted by C∞ .
1.3 Direct product of groups
In Algebra-1, you have used the operation of direct product of abelian
groups. In particular, the fundamental theorem of finitely generated abelian
groups presented any such group as a product of cyclic ones.
Direct product of any groups is defined in the same way. It is a useful
method of building new groups from already known ones.
Definition. Let G and H be two (multiplicative) groups. We define the
direct product G × H of G and H to be the set {(g, h) | g ∈ G, h ∈ H} of
ordered pairs of elements from G and H, with the obvious component-wise
multiplication of elements (g1 , h1 )(g2 , h2 ) = (g1 g2 , h1 h2 ) for g1 , g2 ∈ G and
h1 , h2 ∈ H.

5
 It is straightforward to check that G× H is a group under this operation.
Note that the identity element is (1G , 1H ), and the inverse of (g, h) is just
(g−1 , h−1 ).
1.4 Multiplication Table
A convenient way to describe a group is by writing its multiplication
table. For instance, the Klein four group K4 =< a, b|a2 , b2 > is the set
{1, a, b, c = ab} with the multiplication table:

1 a b c
1 1 a b c
a a 1 c b
b b c 1 a
c c b a 1
However, if the group is infinite or finite but large, the multiplication
table approach is not quite practical. For instance, the famous big monster
group has approximately 1052 elements. Do you think it is a good idea
to write its multiplication table? However, this group can be explicitly
described as a subgroup of a larger group, which can be well understood.
1.5 Definition of a subgroup
Definition. A subset H of a group G is called a subgroup of G if it forms a
group under the same operation as that of G.

Lemma 1.4 If H is a subgroup of G, then the identity element 1H of H is

equal to the identity element 1G of G.
Proof: Clearly, 1h · 1h = 1h . Multiplying by the inverses in G 1−1
h · 1h · 1h =
1−1
h · 1h gives desired 1h = 1G . 2
This lemma implies that 1G ∈ H, in particular, H is non-empty. Indeed,
the empty set is not a subgroup because it is not a group. The identity
axiom fails!

Proposition 1.5 Let H be a nonempty subset of a group G. Then H is a

subgroup of G, if and only if
(i) h1 , h2 ∈ H ⇒ h1 h2 ∈ H; and
(ii) h ∈ H ⇒ h−1 ∈ H.
Proof: H is a subgroup of G if and only if the four group axioms hold in
H. Two of these, ‘Closure’, and ‘Inverses’, are the conditions (i) and (ii)
of the lemma, and so if H is a subgroup, then (i) and (ii) must certainly

6
be true. Conversely, if (i) and (ii) hold, then we need to show that the
other two axioms, ‘Associativity’ and ‘Identity’ hold in H. Associativity
holds because it holds in G, and H is a subset of G. Since we are assuming
that H is nonempty, there exists h ∈ H, and then h−1 ∈ H by (ii), and
hh−1 = 1 ∈ H by (i), and so ‘Identity’ holds, and H is a subgroup. 2

Proposition 1.6 Let H be a nonempty subset of a group G. Then H is a

subgroup of G, if and only if h, g ∈ H ⇒ hg −1 ∈ H.
Proof: Let us build on Proposition 1.5. If its condition holds then g −1 ∈ H
and, consequently, hg −1 ∈ H.
On the other hand if h, g ∈ H ⇒ hg −1 ∈ H then h−1 = (hh−1 )h−1 ∈ H.
Hence, h1 h2 = h1 (h−1
2 )
−1 ∈ H. 2

Corollary 1.7 The intersection of any set of subgroups of G is itself a

subgroup of G.
Proof: Let X = {H} be a set of subgroups of G and T = ∩H∈X H. If
h, g ∈ T then h, g ∈ H for all H ∈ X. By Proposition 1.6, hg −1 ∈ H for all
H ∈ X and consequently to its intersection T . 2
1.6 Examples – Trivial Subgroups
There are two standard subgroups of any group G: the whole group G
itself, and the trivial subgroup {1} consisting of the identity alone. Sub-
groups other than G are called proper subgroups, and subgroups other than
{1} are called nontrivial subgroups.
1.7 Elementary Properties – the Cancellation Laws
Proposition 1.8 Let G be any group, and let g, h, k ∈ G. Then
(i) gh = gk ⇒ h = k; and
(ii) hg = kg ⇒ h = k.
Proof: For (i), we have gh = gk ⇒ g−1 gh = g−1 gk ⇒ h = k, and (ii) is
proved similarly by multiplying by g−1 on the right. 2
1.8 Exercises
(i) Prove that elements of the form (g, 1H ) form a subgroup in G × H.
(ii) Suppose G is a group, x1 , . . . xn ∈ G. Prove that if x1 x2 · · · xn = 1
then x2 x3 · · · xn x1 = 1.
(iii) Give an example showing that a union of subgroups is not necessarily
a subgroup.
(iv) Let G be a group with a subgroup Gi for each natural i. Prove that if
Gi ⊆ Gi+1 then the union H = ∪∞ n=1 Gn is a subgroup.

7
1.9 Vista: other products of groups
The direct products are not the only products of groups available. Its
close relative is a semidirect product. You can twist direct and semidirect
product by a cocycle to get twisted and crossed products. There are also
bycrossed products and knit products. Amazingly enough, all of them are
various group structures on the set G × H for a pair of groups.
If you are willing to consider more general sets, I have further products
up my sleeve. If you can solve Rubik’s cube, you have an experience with
wreath product! It is used to build Rubik’s cube group out of two symmetric
and two abelian groups. Topologists use free products, HNN-extensions and
amalgams. Any of these constructions could be a nice topic for a second
year essay.

2 Rings and subrings

We define rings. We establish that each ring gives two groups: additive
and multiplicative. We also discuss methods of obtaining new rings from
old ones.
2.1 Definition of a ring
Definition. A ring is a set R together with two binary operations
+, · : R × R → R that satisfy the following properties:
(i) (Group under addition) (R, +) is an abelian group.
(ii) (Associativity) For all a, b, c ∈ R, (ab)c = a(bc);
(iii) (Distributivity) For all a, b, c ∈ R, (a + b)c = ac + bc and a(b + c) =
ab + ac.
(iv) (Identity) There exists an element 1 ∈ R such that for all a ∈ R,
1a = a1 = a.
As in the usual arithmetic, multiplication takes precedence over addition.
The identity element in a ring behaves slightly differently from the iden-
tity element in the group. As a start, 1a = a does not formally imply a1 = a.
Nevertheless, the identity is unique.

Lemma 2.1 Let R be a ring. Then R has a unique identity element.

Proof: Let 1 and 1′ be two identity elements of R. Then, 1 = 11′ = 1′ . 2
Besides, various books will treat the identity axiom differently. Some
would skip this axiom identity axiom all together. This would allow the
following degenerate example to be called a ring. Take an abelian group A
and define xy = 0A for all x, y ∈ A. The multiplication is identically zero!! If

8
such things show up in the course, we would call them rings without identity.
A natural question is whether 1 should be different from 0.

Lemma 2.2 Let R be a ring such that 0 = 1. Then R = {0}.

Proof: For all x, x = x1 = x0 = 0. 2
The ring {0} a ring is called the zero ring. It is not a useful ring. All
other rings will be called nonzero rings.
Definition. A ring R is commutative if it satisfies
(v) (Commutativity) For all a, b ∈ R, ab = ba.
2.2 Examples – Numbers
Standard number systems Z, Q, R, and C are commutative rings under
usual addition and multiplication. Quaternions H, which you will see later
in the term, is a ring, which is no longer commutative.
Another important example coming from number is the ring of residues
modulo n. Pick a positive integer n. Then Zn = {0, 1, . . . n − 1} with
multiplication and addition coming from usual ones modulo n. For instance,
in Z6 , 4 + 5 = 3 (residue of 9 modulo 6), 4 · 5 = 2 (residue of 20 modulo 6),
4 · 3 = 0 (residue of 12 modulo 6).
2.3 Matrices over a ring
If R is a ring then the set Mn (R) of n×n-matrices with coefficients in R is
another ring. The multiplication and addition is the same as you P have learnt
in linear algebra: (aij ) + (bij ) = (aij + bij ) and (aij ) · (bij ) = ( k aik bkj ).
Watch out for the order of multipliers if the ring R is no longer commutative!!
For the zero ring R, the ring Mn (R) is also zero. For a non-zero ring
Mn (R) is commutative if and only if R is commutative and n = 1.
2.4 Polynomials over a ring
Let R be a ring, X1 , . . . Xk independent variables. The polynomials in
Xi -s with coefficients in R form the polynomial ring R[X1 , . . . Xn ] under the
usual addition and multiplication of polynomials.
A multi-index notation is a convenient way to describe addition and
multiplication in the polynomial rings. A multi-index α = (α1 , . . . , αn )
is an n-tuple of non-negative integers. They can be added and compared
component-wise. A monomial is written as XP α = X α1 X α2 · · · X αn . A poly-
1 2 n
nomial is a linear combination of monomials r X α with r ∈ R, all zero
P P α α P α
except many. The (Pα rP
finitelyP α α α
P α X ) + ( α tα X ) = α (rα + tα )X and
α α
( α rα X ) · ( α tα X ) = α ( β≤α rβ tα−β )X . α

9
 Observe that R[X1 , . . . Xn ] is commutative if and only if R is commuta-
tive.
2.5 Subrings
Definition. A subset S of a ring R is called a subring of R if it forms a ring
under the same operation as that of R with the same identity element.
The identity element in the rings gives us a trouble again. It is possible
for a subset to be a ring under the same operations but with a different
identity element (see section 2.9 for an example). As a logician would put
it, we consider rings with identity in the signature.

Proposition 2.3 Let S be an abelian subgroup of a ring R. Then S is a

subring of R, if and only if
(i) a1 , a2 ∈ S ⇒ a1 a2 ∈ S; and
(ii) 1R ∈ S.
Proof: S is an abelian subgroup of R, closed under the multiplication
and containing the identity. Thus, S has two operations and identity for
multiplication. All ring axioms of S easily follow from the corresponding
axioms of R. 2

Lemma 2.4 The intersection of any set of subrings of R is itself a subring.

2.6 Complex numbers as matrices

The ring of complex numbers C is a subring of M2 (R). By i we denote
the imaginary
 unit.
 Then we can define a complex number α + βi as the
α β
matrix .
−β α
Actually we have to be more careful because one may choose a different
definition of C, for instance as a set of formal symbols α + βi or pairs (α, β)
with suitably defined operations. Then the subring and C are no longer
equal but rather isomorphic. The precise statement should be that the ring
of complex numbers C is isomorphic to a subring of M2 (R). Isomorphisms
is a topic of the next lecture.
2.7 Additive and multiplicative group of a ring
A ring R gives rise to its additive group R+ . As a set R+ = R, and the
addition is the same. All what happens is that the multiplication and the
unit are completely forgotten. For instance, the additive group Z+ n is the
cyclic group of order n, usually denoted Cn in the multiplicative notation.
Similarly, the additive group Z+n is the infinite cyclic group C∞ .

10
Definition. An element x of a ring R is called a unit if there exists an
element x′ ∈ R such that xx′ = x′ x = 1R .

Lemma 2.5 All the units in a ring R form a group under multiplication.
Proof: Let us denote R× the set of all units in R. The product on R× is
associative because the product on R is associative. The identity element of
R× is 1R and the inverse of x is x′ . 2
′ −1
In particular, x is unique and will be denoted x to be consistent with
the rest of the notation. The group R× is called the group of units of the
ring R or the multiplicative group of R. For example, the multiplicative
group Mn (R)× of the matrix ring Mn (R) is called the general linear group
and denoted GLn (R).
2.8 Fields
Definition. A field is a commutative ring K such that K × = K \ {0}. A
subfield is subring of a field, which is a field under the same operations.
Let us look at some familiar rings. For integers, Z× = {±1} = 6 Z \ {0},
so Z is not a field. For complex numbers, C× = C \ {0}, so C is a field. For
the zero ring, Z×1 = {0} =6 Z \ {0} = ∅, so Z1 is not a field. Thus, a field has
at least two distinct elements 0 6= 1.
2.9 Direct product of rings
Direct products of rings are very similar to direct products of groups.
Definition. Let R and S be two rings. We define the direct product R × S
of R and S to be the set {(r, s) | r ∈ R, s ∈ S} of ordered pairs of elements
from R and S, with the obvious component-wise addition and multiplication
(r1 , s1 ) + (r2 , s2 ) = (r1 + r2 , s1 + s2 ), (r1 , s1 )(r2 , s2 ) = (r1 r2 , s1 s2 ) for r1 , r2 ∈
R and s1 , s2 ∈ S.
It is straightforward to check that R × S is a ring under these operations.
Note that the identity element is (1R , 1S ) but R and S are not subrings of
R × S, in general. Indeed, R can be thought of as elements of the form
(r, 0S ) but it does not contain the identity element of R × S.
2.10 Exercises
(i) Which of the rings Z2 , Z3 , Z4 , Z5 , Z6 , Z7 , Z8 are fields?
(ii) Prove that Zn is a field if and only n is prime.
(iii) Find R+ and R× for the direct product ring Z2 × Z2 .
(iv) What is the intersection of the subgroups R× and {z | |z| = 1} of the
multiplicative group of the complex numbers C× ?

11
 (v) Consider a ring R without identity element. Define R b = R × Z with
operations given by (r, n) + (s, m) = (r + s, n + m) and (r, n) · (s, m) =
(rs + mr + ns, mn). Prove that this is a ring.
(vi) An element p ∈ R such that p2 = p is called an idempotent. Prove that
a field contains exactly two distinct idempotents.
(vii) Describe all idempotents in M2 (R).
(viii) Consider the set C(R, R) of all functions from real numbers R to
real numbers R, continuous at all but finitely many points. Using the
fact from Analysis that the sum and the product of two continuous
functions is continuous, prove that C(R, R) is a ring.
(ix) Which of the following subsets form a subring of C(R, R): smooth func-
tions C ∞ (R, R), compactly supported functions Cc (R, R), polynomial
functions f (X) such that f ′ (0) = 0?
(x) Let V = U ⊕ W be three vector spaces over R such both V and U
are of countable dimension. We consider the ring R = LR (V ) of all
linear maps V → V with the composition of maps as a multiplication.
Choose a linear isomorphism f : V → U that becomes an element of
R now. Prove that there infinitely many x ∈ R such that xf = 1R .
Conclude that f is not a unit and there is no y ∈ R such that
f y = 1R .
2.11 Vista: pass rings
Besides the ring of commutative polynomials R[X1 , . . . Xn ] there is also
a ring of noncommutative polynomials R < X1 , . . . Xn >. Inside this ring
X1 X2 6= X2 X1 . This ring is of crucial importance in Physics and Engineer-
ing as its elements are tensors!
The general algebraic object is the tensor ring of a bimodule. Importance
subclass are pass algebras or quiver algebras1 . Its elements are formal linear
combinations of passes in a directed graph (quiver). The product of two
passes is its concatenation if one pass ends where the other one starts or
zero if not. For instance, R < X1 , . . . Xn > is the pass algebra of the graph
with 1 vertex and n loops at this vertex.

3 Isomorphisms and symmetric groups

Algebraic structures like rings and groups are rarely equal but often
isomorphic. We discuss the notion of isomorphism2 an apply it to symmetric
1
see http://www.amsta.leeds.ac.uk/˜pmtwc/quivlecs.pdf for more info
2
Later on in Section 9, we shall be considering the more general notion of a homomor-
phism

12
groups.
3.1 Isomorphisms
Later on, we shall be considering the more general case of homomor-
phisms, but for now we just introduce the important special case of isomor-
phisms.
Definition. An isomorphism φ : G → H between two groups G and H is a
bijection from G to H such that φ(g1 g2 ) = φ(g1 )φ(g2 ) for all g1 , g2 ∈ G. Two
groups G and H are called isomorphic if there is an isomorphism between
them. In this case we write G ∼
= H.
Isomorphic groups may be considered to be essentially the same group
- H can be obtained from G simply be relabelling the elements of G. The
ring isomorphism is defined in the same way.
Definition. An isomorphism φ : R → T between two rings R and T is
a bijection from R to T such that φ(r1 r2 ) = φ(r1 )φ(r2 ) and φ(r1 + r2 ) =
φ(r1 ) + φ(r2 ) for all r1 , r2 ∈ R. Two rings R and T are called isomorphic if
there is an isomorphism between them. In this case we write R ∼ = T.
In section 2.6, we saw that the ring of complex numbers C is isomorphic
to a subring of M2 (R). Here is our first example of a group isomorphism.
The groups C2 × R+ and R× are isomorphic. To write it explicitly, it is
convenient to think of C2 as its isomorphic group Z× = {1, −1}. The map
φ : C2 × R+ → R× defined by φ(ε, x) = εex is an isomorphism.
The isomorphism φ : C2 × R+ → R× illustrates the fact that often it is
important that isomorphic groups (or rings) are not equal. Restricted to a
subgroup φ is an exponential function : ex : R 7→ R>0 . You won’t get far in
your analysis exam if you think of it as an equality!
3.2 Elementary Properties – Orders of Elements
First some more notation. In a multiplicative group G, we define g2 = gg,
g3 = ggg, g 4 = gggg, etc. Formally, for n ∈ N, we define g n inductively,
by g1 = g and g n+1 = gg n for n ≥ 1. We also define g 0 to be the identity
element 1, and g−n to be the inverse of gn . Then g x+y = gx g y for all x, y ∈ Z.
In an additive group, gn becomes ng, where 0g = 0, and (−n)g = −(ng).
Definition. Let g ∈ G. Then order of g, denoted by |g|, is the least n > 0
such that gn = 1, if such an n exists. If there is no such n, then g has infinite
order, and we write |g| = ∞.
Note that if g has infinite order, then the elements gx are distinct for
distinct values of x, because if gx = g y with x < y, then gy−x = 1 and g has
finite order.

13
 Similarly, if g has finite order n, then the n elements g0 = 1, g1 =
g, . . . , gn−1 = g−1 are all distinct, and for any x ∈ Z, gx is equal to exactly
one of these n elements. Proofs of the next three lemmas are left as exercises
for the reader.

Lemma 3.1 |g| = 1 ⇔ g = 1.

Lemma 3.2 If |g| = n then, for x ∈ Z, g x = 1 ⇔ n|x.

(Recall notation: for integers x, y, x|y means x divides y.)
The following result is often useful. It is the first manifestation of the
principle that isomorphic groups have the same algebraic properties.

Lemma 3.3 If φ : G → H is an isomorphism, then |g| = |φ(g)| for all

g ∈ G.

3.3 Symmetric groups

Let X be any set, and let Sym(X) denote the set of permutations of X;
that is, the bijections from X to itself. The set Sym(X) is a group under
composition of maps. It is known as the symmetric group of X.
The proof that Sym(X) is a group uses results from Foundations. Note
that the composition of two bijections is a bijection, and that composition
of any maps obeys the associative law. The identity element of the group is
just the identity map X → X, and the inverse element of a map is just its
inverse map.
Let us recall the cyclic notation for permutations. If a1 , . . . , ar are dis-
tinct elements of X, then the cycle (a1 , a2 , . . . , ar ) denotes the permutation
of φ ∈ X with
(i) φ(ai ) = ai+1 for 1 ≤ i < r.
(ii) φ(ar ) = a1 , and
(iii) φ(b) = b for b ∈ X \ {a1 , a2 , . . . , ar }.
When X is finite, any permutation of X can be written as a product (=
composite) of disjoint cycles. Note that a cycle (a1 ) of length 1 means that
φ(a1 ) = a1 , and so this cycle can (and normally is) omitted.
For example, if X = {1, 2, 3, 4, 5, 6, 7, 8} and φ maps 1, 2, 3, 4, 5, 6, 7, 8 to
5, 8, 6, 4, 3, 1, 2, 7, respectively, then φ = (1, 5, 3, 6)(2, 8, 7), where the cycle
(4) of length 1 has been omitted. We will denote the identity permutation
in cyclic notation by ().
Remember that a composite φ2 φ1 of maps means φ1 followed by φ2 , so,
for example, if X = {1, 2, 3}, φ1 = (1, 2, 3) and φ2 = (1, 2), then φ1 φ2 =

14
(1, 3), whereas φ2 φ1 = (2, 3). This example shows that Sym(X) is not in
general a commutative group. (In fact it is commutative only when |X| ≤ 2.)
The inverse of a permutation can be calculated easily by just reversing all
of the cycles. For example, the inverse of (1, 5, 3, 6)(2, 8, 7) is (6, 3, 5, 1)(7, 8, 2),
which is the same as (1, 6, 3, 5)(2, 7, 8). (The cyclic representation is not
unique: (a1 , a2 , . . . , ar ) = (a2 , a3 , . . . , ar , a1 ), etc.)

Proposition 3.4 Let X and Y be two sets with |X| = |Y |. Then the groups
Sym(X) and Sym(Y ) of all permutations of X and Y are isomorphic.
Proof: Let ψ : X → Y be a bijection. The map Sym(X) → Sym(Y )
defined by f 7→ ψf ψ −1 is an isomorphism. 2
The notation Sym(n) or Sn is standard for the symmetric group on a set
X with |X| = n. By default, we take X = {1, 2, 3, . . . , n}.
3.4 Exercises
(i) Show that the relationship between groups of being isomorphic satisfies
the conditions of an equivalence relation; that is, G ∼
= G, G ∼ =H ⇒
H∼ = G, and G ∼
= H, H ∼
= K ⇒ G ∼
= K.
(ii) Prove Lemma 3.3.
(iii) If X is finite, what is the order of Sym(X) as a function of |X|?
(iv) Now let n be a positive integer, Hn = {z ∈ C | z n = 1}. Prove that
Hn is a subgroup of C× , isomorphic to Cn .
(v) A matrix M ∈ GLn (R) is a permutation matrix if in each row and
column n − 1 entries are 0R and the remaining entry is 1R . Prove that
permutation matrices form a subgroup of GLn (R) isomorphic to Sn .
3.5 Vista: linear groups
A group G is called linear if it is isomorphic to a subgroup of GLn (F )
for some field. A few things could prevent a group from being linear. For
instance, Sym(X) is not linear if X is infinite. Observe that Sym(X) has at
least countably many commuting (i.e. xy = yx) elements of order 2. Let us
assume that 1 + 1 6= 0 ∈ F which implies that any matrix of order 2 can be
diagonalised. Now finitely many diagonalisable commuting matrices can be
simultaneously diagonalised. But GLn (F ) can accommodate at most 2n − 1
diagonal matrices of order 2.
There are other obstacles. In 1902 Burnside proved that a finitely gen-
erated linear periodic3 must be finite. In 1964 Golod and Shafarevich con-
3
periodic mean every element has finite order

15
structed an infinite finitely generated periodic group4 . These groups cannot
be linear by Burnside’s theorem.
Given an infinite group, it can be a non-trivial challenge to establish
whether it is linear. For instance, it was a long standing problem whether
the braid group5 is linear. It was solved affirmatively in 2000 by Warwick
mathematician Daan Krammer.

4 Generators, cyclic groups, quaternionic

group
We discuss generators of groups. We put on sound footing the discussion
about cyclic groups. We use them in Diffie-Hellman Key Exchange. We
finish with introducing the group Q8 .
4.1 Generators
Let G be a group, H its subgroup.
Definition. The elements {g1 , g2 , . . . , gr . . .} of a group G are said to gen-
erate H (or to form a set of generators for H) if every element of H can be
obtained by repeated multiplication of the gi and their inverses.
This means that every element of H can be written as an expression like
g22 g2−1 g1 g4 g3−1 g1 g2−1 in the gi and gi−1 , which is allowed to be as long as you
like. Such an expression is also called a word in the generators gi .
We write < g1 , g2 , . . . , gr . . . > for the subgroup generated by elements
gi . If < g1 , g2 , . . . , gr . . . >= G, we say that they generate the group. A
group G is called finitely generated if it can be generated by at least one
finite set.
Examples. 1. Z+ =< −1 >, indeed each n ∈ Z can be written as (−n)(−1).
Also, Z+ =< 2, 3 >, indeed each n ∈ Z can be written as n3 − n2. However,
Z+ 6=< 2 >= 2Z, since only even numbers are additive powers of 2.
2. The following property of Sn was established in Linear Algebra in
connection with the definition of the determinant. A cycle of length two
is called a transposition. Since an arbitrary cycle like (1, 2, 3, . . . , n) can
be written as a product of transpositions (for example, (1, 2, 3, . . . , n) =
(1, 2)(2, 3)(3, 4) . . . (n − 1, n) ), we conclude that any permutation on X can
be written as a product of transpositions. Let us state this fact in the
language of generators.
4
see http://www.jstor.org/pss/2324085 which contains an easier than Golod-
Shafarevich’s example
5
its wiki http://en.wikipedia.org/wiki/Braid group is quite good

16
Lemma 4.1 Sn is generated6 by all transpositions.
3. You have seen in Linear Algebra that any invertible matrix is a
product of elementary matrices. This means that the group GLn (K) for a
field K is generated by elementary matrices.
4.2 Cyclic Groups
Definition. A group G is called cyclic, if it is generated by one element. A
cyclic subgroup of G is a subgroup < g > generated by any g ∈ G.
Essentially, a cyclic group G consists of the integral powers of a single
element. In other words, there exists an element g in G with the property
that, for all h ∈ G, there exists x ∈ Z with g x = h. We call the element g a
cyclic generator of G.
We have already seen cyclic groups Cn = Z+ n and C∞ = Z
+ in Sec-

tion 1.2. Are there any other cyclic groups? Let us look at generators first.

Lemma 4.2 In an infinite cyclic group, any generator g has infinite order.
In a finite cyclic group of order n, generators are exactly elements of order
n.
Proof: Let G be a cyclic group, g ∈ G a generator. If |g| = k < ∞ then
gm = gm+k = gm+tk for all t, m ∈ Z. Hence, gm = g (m)k and {gm | m ∈ Z}
contains at most k elements. This proves the first statement.
For the second statement, we use Lemma 3.2 to conclude that the set
m
{g | m ∈ Z} contains exactly k elements. The second statement follows.
2
Let us use the additive notation to describe generators in the cyclic
groups we know. The group Z+ has infinitely many elements of infinite
order but only two generators: 1 and −1. An element x ∈ Z+ n is a generator
if and only it has order n, i.e., the smallest positive integer m such that
n|mx is n. This means that x is a generator if and only if x and n are
coprime. The number7 of possible generators of Z+ n is denoted ϕ(n). The
function ϕ : N → N is called Euler’s totient function. We will compute it in
Section 12.2.
The following fact is an easy corollary of Lemma 4.2, which we state for
future reference. Its proof is left as an exercise.

6
In fact there are much smaller generating sets for Sn , see the exercise below.
7
it is equal to the number of coprime numbers to n between 0 and n − 1 as we have
just proved

17
Proposition 4.3 The order |g| of an element g ∈ G is equal to the order
| < g > | of the cyclic subgroup < g > generated by g.
Now we are ready to establish an identification test for an infinite cyclic
group.

Proposition 4.4 Let G be a group of infinite order generated by an element

g. Then G ∼= Z+ .
Proof: Observe that G = {gx | x ∈ Z}. We saw in Subsection 3.2 that the
elements gx of G are distinct for distinct x ∈ Z, and so the map φ : Z+ → G
defined by φ(k) = gk for all k ∈ Z+ is a bijection, and it is easily checked to
be an isomorphism. 2
Thus, any two infinite cyclic groups are isomorphic. This means that
the notation C∞ is unambiguous. The following identification test for Cn
means that the notation Cn is unambiguous.

Proposition 4.5 Let G be a group of order n generated by an element g.

Then G ∼
= Z+
n.
Proof: Cyclicity of G means that G = {g x | x ∈ Z}. By Proposition 4.3,
|g| = n. By Lemma 3.2, gx = g y if and only if 1 = gx−y if and only if
n|(x − y). Thus, G = {gx | x ∈ Zn } and the map φ : Z+ n → G defined
by φ(k) = gk for all k ∈ Z+
n is a bijection. It is easily checked to be an
isomorphism. 2
4.3 Diffie-Hellman key exchange
Let us consider Alice and Bob who want to communicate secretly over
insecure channel listened to by Eve. Alice and Bob want to exchange a
secret key over the channel and to encode all future communications with
their secret key. This is not a spy novel, this happens every time you use
your credit card on Internet.
Amazingly enough, the elegant, yet most efficient solution uses cyclic
subgroups. Over the channel, Alice and Bob agree on a group G and an
element g ∈ G. Alice secretly picks a power n, computes a = gn and sends
a to Bob over the channel. Bob secretly picks a power m, computes b = gm
and sends b to Alice. The key K = am = bn = gmn is now available to both
Bob and Alice.
Eavesdropping Eve has access to a, b, g, G. To compute K she needs
exponents m or n and there is no (known) way of finding them quickly. In
the standard implementationthe group G is Z× p where p is a large prime
(approximately 1000 binary bits or so), We will prove later (Corollary 21.4)

18
that Z× ∼
p = Cp−1 . Known sneaky ways of finding n and m will need prime
factorisation of p − 1, which is known to be computationally hard.
Suppose a computer can make one group multiplication in G every mi-
crosecond (10−6 seconds). How long will it take to compute gn ? If n = 2m is
even, one uses gn = gm · gm . If n = 2m + 1 is odd, one uses gn = gm · gm · g.
Hence, one computes gn using between log2 n and 2 log2 n operations. If
n ≈ 21000 , i.e. has 1000 digits in binary representation, one needs at most
2000 multiplications. Thus, Alice and Bob will make their computations in
at most 2 milliseconds (2 · 10−3 seconds).
Let us consider how Eve can try to break the key in the most straight-
forward way. She needs to solve the equation gn = a in n that can be
done compute all |g| powers of g until she hits a. Repeating gx = ggx−1 ,
every power requires one multiplication. As |g| ≈ 21000 , Eve needs at most
21000 = (210 )100 ≈ (103 )100 = 10300 multiplications, that can be done in at
most 10294 seconds. It is worse mentioning at this point that 108 seconds
constitute approximately 3 years, so Eve would need to wait for 3 × 10286
years for the code to be broken.
4.4 Pauli’s matrices and quaternionic group
The following matrices in GL2 (C) are known as Pauli’s matrices (we use
i to denote the imaginary unit):
     
0 1 0 −i 1 0
σx = , σy = , σz = .
1 0 i 0 0 −1

In Physics they are used to describe spin, but we will need their scalar
multiples in the inverse order:
     
i 0 0 1 0 i
I = iσz = , J = iσy = , K = iσx = .
0 −i −1 0 i 0

Definition. The quaternionic group Q8 is < I, J, K >, the subgroup of

GL2 (C), generated by I, J and K.
Notice that K = IJ and Q8 is actually generated by I and J. The
following proposition describes the group structure completely.

Proposition 4.6 Q8 has 8 elements

       
1 0 −1 0 i 0 −i 0
1= , -1= , I= , -I= ,
0 1   0 −1  0 −i 0 i 
0 1 0 −1 0 i 0 −i
J= , -J= , K= , -K= .
−1 0 1 0 i 0 −i 0

19
and its multiplication table is
1 −1 I −I J −J K −K
1 1 −1 I −I J −J K −K
−1 1 −1 −I I −J J −K K
I I −I −1 1 K −K −J J
−I −I I 1 −1 −K K J −J
J J −J −K K −1 1 I −I
−J −J J K −K 1 −1 −I I
K K −K J −J −I I −1 1
−K −K K −J J I −I 1 −1
Proof: Computing with matrices, we need to establish that I 2 = J 2 = −1
and IJ = K and JI = −K. Using matrices, we observe that −1 is a scalar
matrix so it commutes with any other matrix: (−1)X = X(−1) = X.
To locate all the elements we start with 1, I, J, K ∈ Q8 , then −1 = I 2 ∈
Q8 , then −I = (−1)I, −J = (−1)J, −K = (−1)K ∈ Q8 . To show that
it is indeed all of Q8 , we have to prove that these 8 matrices are closed
under multiplication (closeness under inverses follows because they have
finite order). We do this by filing the multiplication table. So far we know
the part of the table:
1 −1 I −I J −J K −K
1 1 −1 I −I J −J K −K
−1 1 −1 −I I −J J −K K
I I −I −1 1 K −K
−I −I I 1 −1 −K K
J J −J −K K −1 1
−J −J J K −K 1 −1
K K −K
−K −K K

The rest follows from the following formal calculations: IK = I(IJ) =

I 2 J = −J, KI = (−JI)I = −JI 2 = J, JK = J(−JI) = −J 2 I = I,
KJ = (IJ)J = IJ 2 = −I, and K 2 = (−JI)(IJ) = −JI 2 J = J 2 = −1. 2
The group Q8 is called quaternionic because of the obvious connection
with quaternions. We postpone the discussion about it until we introduce
quaternions in Section 27. But we need an identification test for the group.

Proposition 4.7 Let G be a group of order 8 generated by two elements

a and b that satisfy the equations a4 = 1, b2 = a2 and ba = a−1 b. Then
G∼= Q8 .

20
Proof: Since G is generated by a and b, any element of G can be written as
a product of the generators a, b, a−1 , b−1 . Since a4 = 1 and b4 = (a2 )2 = 1,
a−1 = a3 and b−1 = b3 , so any element of G is a product of several a
and b. Furthermore, we can use the equation ba = a−1 b = a3 b to move
all occurrences of a in the product to the left of the expression, so that
G = {ak bl | k, l ∈ Z}. Since z = a2 = b2 commute with both a and b
(indeed, zb = b3 = bz and the same for a) and z 2 = a4 = 1, we get a
description G = {1, a, b, ab, z, za, zb, zab} with all the elements distinct since
|G| = 8.
An isomorphism φ : G → Q8 is uniquely determined by φ(a) = I and
φ(b) = J, for instance, φ(z) = φ(a2 ) = φ(a)2 = −1 and φ(ab) = φ(a)φ(b) =
K. One way to finish the proof now is to fill out the multiplication table of
G and see that they are the same.
A more elegant way is to observe that we filled the multiplication table
of Q8 formally starting with relations: I 2 = J 2 = −1, IJ = K, JI = −K
and (−1)X = X(−1) = X and, without explicitly mentioning (−1)2 = 1.
In G we know the same relations: a2 = b2 = z, ba = a−1 b = a3 b = zab, z
commutes with everything, and z 2 = 1. Hence, we are bound to arrive at
the same multiplication table. 2
4.5 Exercises
(i) Prove that < g1 , g2 , . . . , gr > is equal to the intersection of all sub-
groups containing every gi .
(ii) Is Q+ a finitely generated group? Justify your answer.
(iii) Prove an improved version of Lemma 4.1: Sn is generated n − 1 trans-
positions (k, k + 1) for 1 ≤ k ≤ n − 1.
(iv) Prove Proposition 4.3.
(v) Prove that Q8 has 6 elements of order 4 and one element of order 2.
(vi) Prove that any subgroup of a cyclic group is cyclic itself.
4.6 Vista: free groups
In Algebra-1 you have seen the free abelian group Ab < X > on a set X.
It consists of all formal Z-linear combinations of elements of X. Similarly
to this idea, there is a free (nonabelian) group on a set X. Let us consider
(finite) words in alphabet {x, x−1 | x ∈ X}. For instance, the empty word ∅,
xx−1 y −1 y, xxxx are all words but x2 is not. A word w is irreducible if words
xx−1 or x−1 x are not subwords of w. Now the free group Gr < X > consists
of all irreducible words in the alphabet. The multiplication is concatenation
followed by reduction, for instance the product of xyzx−1 and xz −1 y is not
xyzx−1 xz −1 y since it is not irreducible but its reduction xyy.

21
 If |X| = n, one writes Fn for Gr < X >. We have seen a couple of
them: F0 is the trivial group C1 , F1 is the infinite cyclic group C∞ . The
next group F2 is brand new and exciting. Its algebraic properties are crucial
to establishing the Banach-Tarski paradox8 where one cuts a 3-sphere into
4 disjoint pieces and makes two identical 3-spheres by combining 2 pairs of
pieces.

5 Orthogonal and dihedral groups

We introduce orthogonal groups and pinpoint the structure of O2 (R).
We introduce the dihedral group as a subgroup of O2 (R). We also realize it
as a subgroup of the symmetric group.
5.1 Orthogonal matrices
We recall that the transpose of a matrix A = (ai,j ) ∈ Mn (R) is the
matrix AT = (bi,j ) where bi,j = aj,i .
Definition. A matrix is A =∈ Mn (R) is orthogonal if it is invertible and
A−1 = AT .
We denote the set of orthogonal matrices by On (R). This set is of interest
only if the coefficient ring R is commutative.

Lemma 5.1 If R is commutative then On (R) is a subgroup of GLn (R).

Proof: Clearly, On (R) is non-empty (I ∈ On (R)) and closed under inverses
((A−1 )T = (AT )T = A = (A−1 )−1 ). By the two-step test (Proposition 1.5),
it suffices to establish that AB ∈ On (R) whenever A, B ∈ On (R). Thus,
(AB)−1 = B −1 A−1 = B T AT = (AB)T finishes the proof. 2
Notice that the very last equality B T AT = (AB)T requires commutativ-
ity of R. If R is not commutative, it breaks down even for 1 × 1-matrices.
The entries of the identity matrix I = (δi,j ) are called Dirak delta-
symbol. In other words, δi,j = 1 if i = j and δi,j = 0 otherwise.

Lemma 5.2 Let F be a field, A ∈ Mn (F ) with columns c1 , c2 . . . cn . Then

A ∈ On (F ) if and only if cTi cj = δi,j for all i and j.
Proof: Notice that cTi cj is the (i, j)-entry of AT A. Thus, cTi cj = δi,j
is equivalent to AT A = I. It remains to notice that this implies (Linear
Algebra) that A is invertible and A−1 = A−T . 2
It is worse pointing out that Lemma 5.2 holds for A ∈ Mn (R) where
R is a commutative ring. Unfortunately, we are not ready to prove it in
8
Its wiki http://en.wikipedia.org/wiki/Banach-Tarski paradox is quite good

22
this generality. We lack the standard properties of the determinants for
matrices over commutative rings, i.e. multiplicative property det(AB) =
det(A) det(B) and the minor formula AM (A) = M (A)A = det(A)I where
M (A) is the matrix of minors in A. Notice that the latter imply that
A ∈ GLn (R) if and only if det(A) ∈ R× .
5.2 Orthogonal group of size 2 over real numbers
We would like to pinpoint the structure of the group O2 (R). Let us
consider the following two matrices
   
cos α − sin α cos α sin α
Rα = and Sα = .
sin α cos α sin α − cos α

Proposition 5.3 O2 (R) = {Sα , Rα | α ∈ R}.

Proof: The matrices Sα and Rα belong to O2 (R) by the direct matrix
calculation (or by Lemma 5.2). In the opposite direction, if X ∈ O2 (R), by
Lemma 5.2, its columns c1 , c2 form an  orthonormal
 basis of O2 (R). Any
cos α
length 1 vector in R2 has a form vα = for some α ∈ R. Let c1 = vα
sin α
for some fixed α. Then either c2 = vα+π/2 and X = Rα or c2 = vα−π/2 and
X = Sα . 2
It follows from the proof that Sα is never equal to Rβ . Another way to
observe this is by computing determinants: det(Sα ) = −1 while det(Rα ) =
1. Let us now describe these transformations geometrically. Observe, using
standard trigonometric identities, that
   
cos α cos β − sin α sin β cos(α + β)
Rα vβ = = = vα+β ,
− sin α cos β + cos α sin β cos(α + β)

hence Rα is a clockwise rotation by α. Similarly,

 
cos α cos β + sin α sin β
S α vβ = = vα−β .
sin α cos β − cos α sin β

In particular, Rα vα/2 = vα/2 and Rα vα/2+π/2 = vα/2−π/2 = −vα/2+π/2 .

Hence, it is a reflection across the line spanned by vα/2 . Using these calcu-
lations, one can now write the multiplication table for O2 (R):

Rβ Sβ
Rα Rα+β Sα+β
Sα Sα−β Rα−β

23
Observe that rotations form a subgroup SO2 (R) (called special orthogonal
group) that looks suspiciously similar to R+ , indeed, Rα Rβ = Rα+β . There
is a natural map R : R+ → SO2 (R) that sends α to Rα , which is not an
isomorphism, because Rα = Rβ whenever α − β = 2nπ for some n ∈ Z. The
map R is an example of a homomorphism, a concept discussed in Section 9.
5.3 Dihedral group
Let n ∈ N with n ≥ 2. The dihedral group D2n is the group generated
by S0 and R2π/n . Unfortunately, some books denote this group by Dn and
others by D2n , which can be confusing! We prefer D2n because it tells us
the number of its elements:

Lemma 5.4 The order of D2n is 2n and all its elements are S2kπ/n , R2kπ/n
for k ∈ Z, 0 ≤ k ≤ n − 1.
Proof: Using the multiplication table, we can easily find elements R2kπ/n =
k
R2π/n and S2kπ/n = R2kπ/n S0 = R2π/nk S0 in D2n =< R2π/n , S0 >.
Let G = {S2kπ/n , R2kπ/n }. It remains to observe that G a subgroup. G is
−1
obviously nonempty. G is closed under inverse because R2kπ/n = R2(n−k)π/n
−1
and S2kπ/n = S2kπ/n . Finally, G is closed under multiplication because
G = {S2kπ/n , R2kπ/n | k ∈ Z} and 2kπZ is a subgroup of R+ (look at the
multiplication table). 2
In fact, it is convenient at this point to denote rotations ak = R2kπ/n
where a ∈ Zn . If we denote the reflection b = S0 , a typical reflection becomes
ak b = S2kπ/n and multiplication table in D2n could be written using addition
in Zn :
al al b
ak a k+l ak+l b
ak b ak−l b ak−l
We need a recognition test for the dihedral group.

Proposition 5.5 Let G be a group of order 2n generated by two elements

a and b that satisfy the equations an = 1, b2 = 1 and ba = a−1 b. Then
G∼= D2n .
Proof: Since G is generated by a and b, any element of G can be written
as a product of the generators a, b, a−1 , b−1 . Since an = 1 and b2 = 1,
a−1 = an−1 and b−1 = b, so any element of G is a product of several a
and b. Furthermore, we can use the equation ba = a−1 b = an−1 b to move
all occurrences of a in the product to the left of the expression, and we

24
end up with a word in the form ak bl . Using an = b2 = 1 again, we can
assume that 0 ≤ k < n and 0 ≤ l < 2. This leaves us with precisely 2n
different words ak bl , and since we are told that |G| = 2n, these words must
all represent distinct elements of G. We have now shown that G = {ak | 0 ≤
k < n} ∪ {ak b | 0 ≤ k < n}, exactly as in D2n .
Using ba = a−1 b twice, we get ba2 = (ba)a = a−1 ba = a−1 a−1 b = a−2 b,
and similarly bak = a−k b for all k ≥ 0, and since a−k = an−k , we have
bak = an−k b for 0 ≤ k < n. Now formally we recover the multiplication
table of G that turns out to be the same as of D2n :
(i) (ak )(al ) = ak+l (k + l < n) or ak+l−n (k + l ≥ n);
(ii) (ak )(al b) = ak+l b (k + l < n) or ak+l−n b (k + l ≥ n);
(iii) (ak b)(al ) = ak an−l b = ak+n−l b (k < l) or ak−l b (k ≥ l);
(iv) (ak b)(al b) = ak an−l bb = ak+n−l (k < l) or ak−l (k ≥ l).
Hence φ(ak bl ) = ak bl is the required isomorphism (from G to D2n or in the
opposite direction - it does not matter). 2
5.4 Dihedral group as a subgroup of symmetric group
Let n ∈ N with n ≥ 2 and let P be a regular n-sided polygon in the
plane whose vertices are vk = v2kπ/n , k ∈ Z. It is clear from the formulas
Sα vβ = vα−β and Rα vβ = vα+β in the previous section that XP = P for
any X ∈ D2n . In fact, the opposite is true: if XP = P for some X ∈ O2 (R)
then X ∈ D2n . Thus, D2n is the group of symmetries of a regular n-gon.
It is instructive to stop thinking of P as a 2-dimensional figure. Let us
think of P as a set of its vertices. This allows us to think of D2n as a subgroup
of Sn . More precise, for each X ∈ D2n there exists a unique σX ∈ Sn such
that Xvk = vσX (k) . For instance, σa = (1, 2, 3, . . . , n) rotates the vertices
counterclockwise. Similarly, b is the reflection through the bisector of P that
passes through the vertex v0 = vn . Then b interchanges the vertices v1 and
vn−1 that are adjacent to vn , and similarly it interchanges v2 and vn−2 , v2
and vn−3 , etc., so we have σb = (1, n − 1)(2, n − 2)(3, n − 3) . . .. For example,
when n = 5, b = (1, 4)(2, 3) and when n = 6, b = (1, 5)(2, 4). Notice the
difference between the odd and even cases. When n is odd, b fixes no vertex
other than v0 , but when n is even, b fixes one other vertex, namely vn/2 . We
summarize this discussion in the following proposition.

Proposition 5.6 The function X 7→ σX is an isomorphism from D2n to a

subgroup of Sn .

25
5.5 Exercises
(i) Verify the multiplicative property det(AB) = det(A) det(B) and the
minor formula AM (A) = M (A)A = det(A)I for 2 × 2-matrices over a
commutative ring.
(ii) Let R be a commutative ring. Prove that A ∈ GL2 (R) if and only if
det(A) ∈ R× .
(iii) Compute the order of the general linear groups GL2 (Z2 ) and GL2 (Z4 ).
(iv) Compute the order of the orthogonal groups O2 (Z2 ) and O2 (Z4 ).
(v) What is the subgroup of O2 (R) generated Sα and S0 ?
(vi) Prove that D4 is isomorphic to K4 .
(vii) Prove that D6 is isomorphic to S3 .
(vii) Prove that D8 has 5 elements of order 2 and 2 elements of order 4.
Conclude that D8 is not isomorphic to Q8 .
(viii) Go to the online Magma calculator http://magma.maths.usyd.edu.au/calc/
and play around with groups. For instance, try the code
G<a,b,c> := Group < a,b,c | aˆk, bˆl, cˆm, a*b*c >; G;
Order(G); for some particular k ≥ l ≥ m ≥ 2. Determine experi-
mentally by running the code which of these groups are finite.
5.6 Vista: defining relations and Burnside’s problem
The equations {an = 1, b2 = 1, ba = a−1 b} are called defining relations
for D2n . One formally writes D2n = Gr < a, b | an = b2 = 1, ba = a−1 b >.
This is a great way of describing groups. Roughly it means that D2n is the
largest group generated by two elements a and b that satisfy these equations.
More precisely, it means that D2n is a quotient group of the free group
F2 = Gr < a, b | ∅ > by a subgroup determined by these relations.
Another example of a group determined by relations is von Dyck group
Dk,l,m = Gr < a, b, c | ak = bl = cm = abc = 1 > with which you worked in
the last exercise. Working out properties of a group from its presentation by
generators and relations is far from straightforward: apart from MAGMA
there packages GAP and SAGE that can help you with it. In Warwick Derek
Holt is actively involved with them and has actually written a code for many
of the functions.
Here is your another chance to get immediate recognition as a mathe-
matical genius on par Galois, Gauss and Perelman. The group B(n, m) =
Gr < x1 , x2 , . . . xn | wm = 1 > is know as Burnside group. By w here I
mean all possible words in a and b. By Proposition 7.6, B(n, 2) is abelian,
of order 2n . Using MAGMA, you can observe that B(2, 3) ∼ = Gr < a, b |
a3 = b3 = (ab)3 = (ab2 )3 = 1 > has order 27.
What is about B(2, 5)? Beware that your laptop is not of big help here.

26
If this group is finite, it has exactly 534 (approximately 6 × 1023 ) elements.
But I would bet that it is infinite. If any betting agency accepts, please, let
me know.
This problem is known as Burnside’s problem9 . In 1984, E. Zelmanov
received Fields Medal for proving that B(k, n) admits a unique maximal
finite quotient B0 (k, n). This was known as restricted Burnside’s problem.
In particular, the order of B0 (2, 5) is 534 .

6 Equivalence relations and cosets

We define equivalence relations and equivalence classes. We introduce
cosets, an important example of equivalence classes.
6.1 Binary Relations
Let X be a set.
Definition. A binary relations on X is a subset R of X × X.
We write xRy whenever (x, y) ∈ R. For instance, let X = R. The
relation greater is a subset {(x, y) ∈ R2 |x > y}.
There is a certain shift of paradigm when we talk about binary relations
in this way. This abstract approach can be pushed to other objects, for
instance, binary operations! We can think of a multiplication on a group G
as a subset {(a, b, c) ∈ G3 |ab = c} of G3 .
Definition. Let R be a binary relation on a set X. We say that R is
(i) symmetric if ∀x, y ∈ X xRy =⇒ yRx;
(ii) reflexive if ∀x ∈ X xRx;
(iii) transitive if ∀x, y ∈ X xRy & yRz =⇒ xRz.
For instance, the relation greater is transitive but neither reflexive, nor
symmetric.
6.2 Equivalence Relations
The binary relation equal is reflexive, transitive and symmetric. The
following definition generalises this situation.
Definition. A binary relation is an equivalence relation if it is reflexive,
transitive and symmetric.
You have already seen the following examples of non-trivial equivalence
relations. It is a good exercise to check the axioms but we will not do it
here.
9
Gupta, On groups in which every element has finite order. Amer. Math. Monthly 96
(1989), 297–308 is an accessible source. This is the first maths paper I have read.

27
Examples. 1. Let X = Z, n ∈ Z, n 6= 0. We say that x ≡n y if n
divides x − y. This equivalence relation, called congruent modulo n, appear
in Foundations.
2. We say that x ∼ y ∈ R+ if Rx = Ry ∈ SO2 (R). Clearly, x ∼ y if and
only if x − y ∈ 2πZ.
3. Let X = F n×m be the set of all n × m-matrices over a field F . The
equivalence relation equivalent appears in Linear Algebra. Let us recall that
A ∼ B if and only if A and B have the same rank if and only if A can
be transformed to B by elementary row and column transformations if and
only if there exist P ∈ GLn (F ), Q ∈ GLm (F ) such that P AQ = B if and
only if A and B represent the same linear map f : F m → F n in different
bases of the two vector spaces.
4. Let X = F n×n be the set of all n × n-matrices over a field F . The
equivalence relation similar appears in Linear Algebra. Let us recall that
A ∼ B if and only if A and B have the same Jordan normal form if and only
if there exists P ∈ GLn (F ) such that P AP −1 = B if and only if A and B
represent the same linear map f : F n → F n in different bases of the vector
space.
5. Let X = S(Rn×n ) be the set of all symmetric n × n-matrices over the
real numbers R. This equivalence relation without a special name appears
in Algebra-1. In this relation A ∼ B if and only if A and B have the same
signature if and only if there exists P ∈ GLn (F ) such that P AP T = B if and
only if A and B represent the same quadratic form q : Rn → R in different
bases of the two vector spaces.
Definition. Given an equivalence relation R on X and a ∈ X, the equiva-
lence class of a is the following set [a] = {x ∈ X | xRa}.

Proposition 6.1 The following are equivalent for a, b ∈ X and an equiva-

lence relation R:
(i) a ∈ [b];
(ii) [a] = [b];
(iii) aRb.
Proof: (iii) implies (i) by definition of [b].
Assume (i). Then aRb, hence bRa by symmetricity. Pick an arbitrary
x ∈ [a] so that xRa. Using transitivity, xRb, hence x ∈ [b]. We proved that
[a] ⊆ [b]. Pick an arbitrary x ∈ [b] so that xRb. Using transitivity, xRb,
hence x ∈ [a]. We proved that [a] = [b].
Finally assume (ii). Then a ∈ [a] = [b] and aRb. 2

28
Corollary 6.2 Two equivalence classes [a] and [b] are either equal or dis-
joint. Hence, the equivalence classes form a partition of X.
Proof: If [a] and [b] are not disjoint, then there exists an element c ∈ [a]∩[b].
So by Proposition 6.1 [a] = [c] = [b]. 2

Corollary 6.3 The equivalence relation can be uniquely recovered from its
partition into equivalence classes.
Proof: This follows immediately from Proposition 6.1 as aRb if and only
if they belong to the same class. 2
Finally, we define the quotient set X/R as a collection of equivalence
classes. We will see the further usefulness of this later on but here are the
first three examples which you already saw in various subjects.
Examples. 6. Let ≡n be the congruence modulo n we saw in examples
today. The quotient set Z/ ≡n is the ring Zn of residues modulo n.
7. Let X be the set of all Cauchy sequences in Q. Recall that a sequence
(an ) is Cauchy if for any ε > 0 there exists N such that |am − an | < ε for
all m, n > N . Two Cauchy sequences (an ) and (bn ) are equivalent if their
difference an −bn tends to zero. The significance of this equivalence relations
is that the quotient set X/ ∼ is the set of real numbers.
8. Various function spaces in analysis also quotient sets where one iden-
tifies functions different by a “negligible” function. For instance, let X
Rbe1 the set of all function f : [0, 1] → R such that the Lebesgue integral
|f (x)|2 dx is well-defined and finite. Two functions f and g in X are
0 R1
equivalent if 0 |f (x) − g(x)|2 dx = 0. The quotient set X/ ∼ is the L2 space
L2 ([0, 1], R).
6.3 Cosets
Given a group G and a subgroup H, we define a binary relation ∼H on
G. We set x ∼H y if there exists h ∈ H such that x = hy. Notice that for
G = Z and H = nZ, this is the congruence modulo n.

Proposition 6.4 The relation ∼H is an equivalence relation. Moreover,

x ∼H y if and only if xy −1 ∈ H.
Proof: Since x = 1 · x the relation is reflexive. If x = hy then y = h−1 x,
so the relation is symmetric. If x = hy and y = gz then y = hgz, so the
relation is transitive.
The last statement follows from the fact that x = hy if and only if
h = xy −1 . If x = hy then y = h−1 x, so the relation is symmetric. 2

29
 Similarly to x ∼H y, there is an equivalence relation x H ∼ y. We say
that x H ∼ y if and only if xh = y for some h ∈ H if and only if x−1 y ∈ H.
Definition. Let g ∈ G, H ≤ G. The right coset of H containing g is the
equivalence class H [g] of ∼H . Similarly, the left coset is the equivalence class
[g]H of H ∼
The rightness of a right coset is the position of g with respect to H.
Notice that H [g] = {hg | h ∈ H} = Hg. In fact, Hg is the standard
notation used in most of the books. We will use both notations since they
are convenient in different situations.
Similarly, [g]H is the left coset gH = {gh | h ∈ H}. If H and the side
are clear from the context, we just write [g]. In the case of additive groups,
the standard notation becomes H + g rather than Hg but we may still use
[g].
Example. If V is a vector space and W is a subspace, the cosets of W are
affine subspaces v + W parallel to W .
Example. Let G = S3 be the symmetric group. Then G consists of the 6
permutations (), (1,2,3), (1,3,2), (1,2), (1,3), (2,3), where () represents the
identity permutation.
Let us first choose H = {(), (1, 2, 3), (1, 3, 2)} to be the cyclic subgroup
generated by a = (1, 2, 3). If we put b = (2, 3), then we find that H [b] =
Hb = {(1, 2), (1, 3), (2, 3)}. In fact any right coset of H is equal to either
H itself or to H [b] = Hb = G \ H. Furthermore, [b]H = H [b], and indeed
H [g] = [g]H , for all g ∈ G, so the right and left cosets are the same in this
example.
Now let us choose H = {(), (2, 3)} to be the cyclic subgroup generated
by b = (2, 3). With a = (1, 2, 3), we have H [a] = Ha = {(1, 2, 3), (1, 3)} and
2 2
H [a ] = {(1, 3, 2), (1, 2)}, but [a]H = aH = {(1, 2, 3), (1, 2)} and [a ]H =
2
a H = {(1, 3, 2), (1, 3)}, so the right and left cosets are not the same in this
case.
The following two corollaries are immediate consequences of Corollary 6.2
and Proposition 6.4

Corollary 6.5 Two right cosets H [g1 ] and H [g2 ] of H in G are either
equal or disjoint.

Corollary 6.6 The right cosets of H in G partition G.

6.4 Exercises
(i) Find a binary relation on the set R that is both reflexive and transitive
but not symmetric.

30
 (ii) Find a binary relation on the set R that is both symmetric and tran-
sitive but not reflexive.
(iii) Find a binary relation on the set R that is both reflexive and symmetric
but not transitive.
(iv) Describe and draw the cosets of R+ in C+ .
(v) Describe and draw the cosets of R× in C× .
(vi) Describe and draw the cosets of H = {z ∈ C||z| = 1} in C× .
(vii) Prove that, if |G : H| is finite, then |G : H| is also equal to the number
of distinct left cosets of H in G. This is clear if G is finite, because
both numbers are equal to |G|/|H|, but it is not quite so easy if G is
infinite.
(viii) Let H be a subgroup of a group G and let Hg be a right coset of H
in G. Prove that the set {k−1 | k ∈ Hg} is a left coset of H in G,
and deduce that there is a bijection between the sets of left and right
cosets of H in G.
(iX) Consider solutions of a homogeneous system of linear equations: H =
{X ∈ Rm |AX = 0}. Show that H is a subgroup.
Now consider solutions of a non-homogeneous system of linear equa-
tions: W = {X ∈ Rm |AX = B}. Show that W is a coset of H.
6.5 Vista: groupoids
The notion of groupoid is a common generalisation of a group and an
equivalence relation. If one develops the theory of groupoids first, then
one can say that a group is a groupoid with one object and an equivalence
relation is a groupoid where homs have at most one element. Find out more
by exploring references on its wiki http://en.wikipedia.org/wiki/Groupoid

7 Lagrange’s theorem and applications

We prove Lagrange’s theorem. Using it, we classify groups of order 4
and groups of exponent 2. We also prove some number theoretic facts.
7.1 Lagrange’s Theorem
We have already observed that cosets form a partition of the group G.

Proposition 7.1 If the subgroup H is finite, then all right cosets have ex-
actly |H| elements.
Proof: Since h1 g = h2 g ⇒ h1 = h2 by the cancellation law, it follows that
the map φ : H = H [1] → H [g] defined by φ(h) = hg is a bijection, and the
result follows. 2

31
 Of course, all of the above results apply with appropriate minor changes
to left cosets.
Corollary 6.6 and Proposition 7.1 together imply:

Theorem 7.2 (Lagrange’s Theorem) Let G be a finite group and H a sub-

group of G. Then the order of H divides the order of G.

Definition. The number of distinct right cosets of H in G is called the

index of H in G and is written as |G : H|.
If G is finite, then we clearly have |G : H| = |G|/|H|.

Proposition 7.3 Let G be a finite group. Then for any g ∈ G, the order
|g| of g divides the order |G| of G.
Proof: Let |g| = n. The powers {gx | x ∈ Z} of g form a subgroup
H of G, and we saw in Subsection 3.2 that the distinct powers of g are
{gx | 0 ≤ x < n}. Hence |H| = n and the result follows from Lagrange’s
Theorem. 2
We finish with the following technical lemma for future use.

Lemma 7.4 If x, y ∈ G then |y| = |xyx−1 | for all g ∈ G.

Proof: It follows immediately from Lemma 3.3 and the fact that g 7→ xgx−1
is an isomorphism G → G. 2
7.2 Groups of prime order
A large part of group theory consists of classifying groups with various
properties. This means finding representatives of the isomorphism classes of
groups with these properties. As an application, we can now immediately
classify all finite groups whose order is prime.

Proposition 7.5 Let G be a group having prime order p. Then G is cyclic;

that is, G ∼
= Cp .
Proof: Let g ∈ G with 1 6= g. Then |g| > 1, but |g| divides p by Propo-
sition 7.3, so |g| = p. But then G must consist entirely of the powers gk
(0 ≤ k < p) of g, so G ∼
= Cp by Proposition 4.4. 2
7.3 Groups of exponent 2
The exponent of a group G is the least common multiple of orders of its
elements. In other words, it is the smallest n such that xn = 1 for all x ∈ G.
We will not use this notion much but we need the following description of
groups of exponent 2.

32
Proposition 7.6 Let G be group where every element has order 2 or 1.
Then G ∼
= (C2 )n and |G| = 2n for some n ∈ N.
Proof: Since x2 = y 2 = (yx)2 = 1 for all x, y ∈ G, xy = y 2 xyx2 =
y(yx)(yx)x = yx, so the group is abelian. Moreover, it has a natural struc-
ture of a vector space of Z2 : 1 · x = x and 0 · x = 1G for all x ∈ G. Since G is
finite, it admits a finite basis (as a vector space over Zn2 ) and the statement
follows. 2
Proposition 7.6 gives a method for classifying groups of order 4:

Proposition 7.7 There are two groups of order 4 up to an isomorphism:

C4 and K4 .
Proof: These two groups are non-isomorphic by Lemma 3.3: C4 has an
element of order 4 but K4 hasn’t.
Now if G is a group of order 4, then by Proposition 7.3 its non-identity
elements have order 2 or 4. If G admits an element a of order 4, then
elements 1, a, a2 , a3 are distinct and G = {1, a, a2 , a3 } is a cyclic group.
If G has no such element, all non-identity elements have order 2 and
∼
G = C2 × C2 = K4 by Proposition 7.6. 2
7.4 Euler’s theorem and Fermat’s little theorem
If we use Lagrange’s theorem to multiplicative groups of some finite rings,
we arrive at some celebrated number theoretical facts. Let us first describe
the group.

Lemma 7.8 Let x ∈ Zn . Then x ∈ Z×

n if and only if x and n are coprime.
Proof: If x and n are not coprime then d = gcd(x, n) > 1. Hence d divides
xy for all y ∈ Zn . Hence xy is never 1 and x is not a unit.
If x and n are coprime then x generates Z+ n as discussed in Section 4.2.
Thus, there exists m ∈ Z such that mx = 1 ∈ Z+ n or mx = an + 1 in Z. Let
k = (m)n , the residue of m modulo n. This means that kx = bn + 1 in Z
and kx = 1 in Z× n. 2
Once again we came across he function ϕ : N → N is called the Euler’s
totient function ϕ : N → N, introduced in Section 4.2. This time |Z× n| =
ϕ(n). We will derive a formula for φ(n) in Section 12.2

Corollary 7.9 Zm is a field if and only if m is prime.

Proof: Every divisor of m will be a non-unit in Zm . So Zm is not a field
unless m is prime. If m is prime it is a field by Lemma 7.8. 2
We are ready for Euler’s theorem.

33
Theorem 7.10 (Euler’s Theorem) Let a and n be coprime integers. Then
n|(aϕ(n) − 1).
Proof: Let b = (a)n be the residue. Since the numbers are coprime, b ∈ Z× n.
By Proposition 7.3, |b| divides ϕ(n). Hence b ϕ(n) ×
= 1 in Zn . Consequently,
aϕ(n) − 1 = (aϕ(n) − bϕ(n) ) + (bϕ(n) − 1) is divisible by n in Z. 2
The following fact follows easily.

Corollary 7.11 (Fermat’s little theorem) Let p be a prime number, a an

integer. Then p|(ap − a).
Proof: Notice that ap − a = a(ap−1 − 1) = a(aϕ(p) − 1). If p|a then the
divisibility comes from the first multiplicand. If not it comes from the second
one by Euler’s theorem. 2
7.5 Exercises
(i) The following groups have order 4: Z× × × ×
5 , Z8 , Z10 , Z12 . Determine
whether each of them is K4 or C4 .
(ii) Find an explicit isomorphism between Z× ×
5 and Z10 .
×
(iii) Let n ∈ N be odd. Find an explicit isomorphism between Z× n and Z2n .
Conclude that ϕ(2n) = ϕ(n). Q Q
(iv) Prove that in a finite abelian group G, x∈G x = |x|=2 x.
(v) Use exercise (iv) to prove Wilson’s theorem: a prime p divides (p −
1)! + 1.
(vi) Prove that any two countable groups of exponent 2 are isomorphic.
7.6 Vista: Sylow’s Theorem
Let us ask the inverse question to Lagrange’s theorem. Suppose n divides
|G|. Does G admit a subgroup of order n. The answer to this naive question
is no and we are going to see an example later in the course. However, there
is a partial positive answer if n is a prime power. It is given by a series of 4
Sylow’s theorems. Unfortunately, we don’t have enough time to cover them
in this course. You can learn about them in the third year Group Theory or
you can write your second year essay on this topic.

8 Normal subgroups
We introduce normal subgroups. Using these tools, we classify groups of
order up to 8.
8.1 Normal Subgroups
We need the notion of normal subgroup to carry on.

34
Definition. A subgroup H of a group G is called normal in G if the left
and right cosets [g]H = gH and H [g] = Hg are equal for all g ∈ G.
The standard notation for “H is a normal subgroup of G” is H
G or
H  G. (H
G is sometimes but not always used to mean that H is a proper
normal subgroup of G – i.e. H 6= G.)
Examples. 1. The two standard subgroups G and {1} of any group G are
both normal.
2. Any subgroup of an abelian group is normal.
3. In the example G = D6 in Subsection 7.1, we saw that the sub-
group {(), (1, 2, 3), (1, 3, 2)} (= {1, a, a2 }) is normal in G, but the subgroup
{(), (2, 3)} = {1, b} is not normal in G.
4. More generally, SO2 (R) is a normal subgroup of O2 (R). From its
multiplication table, one coset consists of rotations Rα and another coset
consists of reflections Sα .
In examples 3 and 4, the normal subgroup has index 2, for instance,
H = {1, a, a2 } and |G|/|H| = 6/3 = 2 in example 3.

Proposition 8.1 If G is any group and H is a subgroup with |G : H| = 2,

then H is a normal subgroup of G.
Proof: Assume that |G : H| = 2. Then there are only two distinct right
cosets of G, one of which is H, and so by Corollary 6.6, the other one must
be G \ H. The same applies to left cosets. Hence, for g ∈ G, if g ∈ H
then gH = Hg = H and if g 6∈ H then gH = Hg = G \ H. In either case
gH = Hg, so H
G. 2
The following result often provides a useful method of testing a subgroup
for normality.

Proposition 8.2 Let H be a subgroup of the group G. Then H is normal

in G if and only if ghg −1 ∈ H for all g ∈ G and h ∈ H.
Proof: Suppose that H  G, and let g ∈ G, h ∈ H. Then Hg = gH, and
gh ∈ gH, so gh ∈ Hg, which means that there exists h′ ∈ H with gh = h′ g.
Hence ghg −1 = h′ ∈ H.
Conversely, assume that ghg −1 ∈ H for all g ∈ G, h ∈ H. Then for
gh ∈ gH, we have ghg−1 ∈ H, so gh = h′ g for some h′ ∈ H; i.e. gh ∈ Hg,
and we have shown that gH ⊆ Hg. For hg ∈ Hg, we have g−1 hg ∈ H
(because g−1 hg = ghg −1 where g = g −1 ), so, putting h′ = g −1 hg, we have
hg = gh′ ∈ gH, and so Hg ⊆ gH. Thus gH = Hg, and H  G. 2

35
8.2 Groups of order 8
Proposition 8.3 Let G be a group of order 8. Then G is isomorphic to
one of C8 , C4 × C2 , C2 × C2 × C2 , D8 and Q8 .
Proof: By Proposition 7.3, nonidenity elements of G may have order 2, 4
and 8. If all non-identity elements have order 2, then G ∼
= C2 × C2 × C2 by
Proposition 7.6.
Otherwise, there is an element a ∈ G of order 4, for instance, if |x| = 8
then a = x2 is of order 4. By Proposition 8.1, N =< a >= {1, a, a2 , a3 } is a
normal subgroup. Pick any b ∈ G\N . Since bab−1 ∈ N and |bab−1 | = |a| = 4
by Lemma 7.4, bab−1 must be either a or a−1 .
Also we cannot have b2 ∈ N b, as this would imply b2 = nb for some
n ∈ N and b = n ∈ N . Hence, b2 ∈ N since G = N ∪ N b.
Before we analyse eight possibilities, we observe that G is generated by
a and b since < a, b > properly contains N , hence, it has index strictly less
then 2. In particular, this implies that bab−1 = a makes G abelian while
bab−1 = a3 makes it nonabelian.
Let us assume bab−1 = a3 and analyse four nonabelian possibilities:

(i)b2 = 1, then G ∼ = D8 by Proposition 5.5,

(ii)b = a is impossible as this means that a3 = bab−1 = bb2 b−1 = b2 = a,
2

(iii) b2 = a2 , then G ∼
= Q8 by Proposition 4.7
(iv) b = a is impossible as this means that a = (a3 )3 = b6 and a3 =
2 3

bab−1 = bb6 b−1 = b6 = a.

Let us finally assume bab−1 = a and analyse four abelian possibilities
using the matrix technique from Algebra-1:
(i) b2 = 1, then G is a quotient of Ab < x, y | 4x, 2y >= C4 × C2 , as they
both have the same order, G ∼ = C4 × C2 ,
(ii) b2 = a, then G is a  of Ab< x,y | 4x,
quotient
 x −  2y >, reducing

4 1 4 1 0 1 8 0
the matrix ; ; ; gives
0 −2 8 0 8 0 0 1
Ab < x, y | 8x, y >∼ = C8 , hence G ∼= C8 by the order equality, and
2 2
(iii) b = a , then G is a quotient
  of Ab <  x, y | 4x, 2x
 − 2y >, reducing
4 2 4 0 4 0
the matrix ; ; gives Ab < x, y |
0 −2 0 −2 0 2
4x, 2y >= C4 × C2 , hence G ∼ = C4 × C2 by the order equality,
(iv) b2 = a3 , then
 G is a quotient
  of Ab <x, y  | 4x, 3x −2y >,reducing

4 3 1 3 1 3 1 0
the matrix ; ; ;
0 −2 2 −2 0 −8 0 8
gives Ab < x, y | x, 8y >∼ = C8 , hence G ∼= C8 by the order equality.

36
 2
It is worse pointing out why these five groups of order 8 are noniso-
morphic. Three of them are abelian, two are nonabelian. The nonabelian
D8 and Q8 have different number of elements of order 2: 5 and 1 corre-
spondingly. The abelian are distinct by the fundamental theorem of abelian
groups from Algebra-1. Alternatively, you can just count the number of
elements of order in each of them.
8.3 Groups of order 2p
Let p be a prime. We know two groups of order 2p: the cyclic group C2p
and the dihedral group D2p . It is worse pointing out what happens at small
primes. If p = 2 then D4 ∼= K4 is abelian, non-isomorphic to C4 . If p = 3
then D6 is nonabelian, isomorphic to the symmetric group S3 .

Proposition 8.4 Let G be group of order 2p. Then G ∼ = C2p or G ∼ = D2p .

Proof: By Proposition 7.3 the orders of elements g ∈ G can be 1,2,p or 2p.
If there is a g with |g| = 2p, then G ∼ = C2p , so assume not. If all elements
had order 1 or 2, then, by Proposition 7.6, |G| = 2m , hence p = 2 and
G∼ = K4 ∼ = D4 .
Hence we assume that there is an element a of order p. Then the sub-
group H =< a >= {1, a, a2 . . . ap−1 } has index 2 in G, so normal by Propo-
sition 8.1. Choose b ∈ G\H. Under assumptions we made the order of b is 2
or p. If |b| = p then the size of the union < b > ∪ < a > is 1+2(p−1) = 2p−1
because < b > ∩ < a >= {1}. Hence there exists c ∈ G \ (< b > ∪ < a >)
and < c > must be {1, c} since there are no more elements left. In partic-
ular, |c| = 2. If |b| = 2 then we set c = b. In both cases we have found
c ∈ G \ H with c2 = 1.
The subgroup < a, c > properly contains H, hence has index less than 2,
hence G =< a, c >. Since H is normal, cac−1 = an for some n. Since c2 = 1,
2
we conclude that a = ccac−1 c−1 = can c−1 = (cac−1 )n = (an )n ) = an . Thus
p divides n2 − 1 = (n − 1)(n + 1). So p must divide either n − 1 or n + 1.
If p divides n − 1, cac−1 = an = a, so G is abelian. If p is odd, then
|ca| = 2p, contradiction. Hence p = 2 and G ∼ = Ab < x, y | 2x, 2y >∼= K4 .
−1 n −1 −1
If p divides n + 1, then cac = a = a and ca = a c. By Proposi-
tion 5.5, G ∼ = D2p . 2
8.4 Exercises
(i) Show that GL2 (Z2 ) is a group of order 6 and determine which of the
groups in Proposition 8.4 it is isomorphic to.

37
 (ii) Show that O2 (Z3 ) is a group of order 8 and determine which of the
groups in Proposition 8.3 it is isomorphic to.
(iii) For a commutative ring R we denote Tn (R) the group of triangular
n × n-matrices with coefficients in R, i.e. Tn (R) = {(aij ) ∈ GLn (R) |
aij = 0 whenever i > j}. Show that T3 (Z2 ) is a group of order 8 and
determine which of the groups in Proposition 8.3 it is isomorphic to.
(iv) Let STn (R) be the subgroup of Tn (R) of matrices with determinant
1. Show that T2 (Z3 ) is a group of order 6 and determine which of the
groups in Proposition 8.4 it is isomorphic to.
(v) Show that ST2 (Z4 ) is a group of order 8 and determine which of the
groups in Proposition 8.3 it is isomorphic to.
8.5 Vista: classification of groups
We have advanced quite far in classification of groups. To facilitate the
discussion, let f (n) be the number of non-isomorphic groups of order n. We
know so far that if p is prime then f (p) = 1, f (2p) = 2, f (8) = 5. Later
one, we are going to classify groups of order p2 : they are all abelian, hence
f (p2 ) = 2.
The next interesting order is 12, f (12) = 5, only 2 are abelian. Out
of 3 nonabelian, we know only D12 so far, but we are introducing the two
remaining one later in this course, albeit we are not proving that the list is
exhaustive. One needs Sylow’s theorem to handle 15, f (15) = 1. The next
number 16 is the first case where I would not know what to do: f (16) = 14
and I don’t know most of them10 . The next interesting case is 18: f (18) = 5.
Four of the groups are straightforward: C18 , C3 × C6 , D18 , C3 × D6 but the
remaining one requires semidirect products. In general, if n have no large
prime powers, f (n) is easy to handle. f (p3 ) = 5 with all 5 groups easy to
construct: three are abelian, U T3 (Zp ) (subgroup of T3 consisting of matrices
with 1 on the main diagonal) and Prime powers11 behave badly: f (32) = 51,
f (64) = 267, . . . , f (1024) = 49487365422.
While classifying all finite groupsPappears hopeless,
P it may be possible
to understand generating functions n f (n)z n or n f (pn )z n .

9 Homomorphisms
We introduce the notion of a homomorphism, its image and its kernel.
10
although see M. Wild, Groups of order 16 made easy, American Mathematical
Monthly, 112 (2005), 20–31.
11
Check out the book with exciting title The groups of order 2n (n equal to 6) by Hall
and Senior.

38
9.1 Definition and examples of homomorphisms
Definition. Let G, H be groups, R, S rings. A group homomorphism φ
from G to H is a function φ : G → H such that φ(g1 g2 ) = φ(g1 )φ(g2 ) for all
g1 , g2 ∈ G. A ring homomorphism φ from R to S is a function φ : R → S
such that φ(1R ) = 1S , φ(r1 r2 ) = φ(r1 )φ(r2 ), and φ(r1 + r2 ) = φ(r1 ) + φ(r2 )
for all r1 , r2 ∈ R.
Notice that the ring homomorphism requires an extra condition concern-
ing the identity. For instance, the natural map R → R × S, r 7→ (r, 0S ) is
not a ring homomorphism. Such surprises don’t happen with groups.

Lemma 9.1 Let φ : G → H be a homomorphism. Then φ(1G ) = 1H and

φ(g−1 ) = φ(g)−1 for all g ∈ G.
Proof: (Recall that 1G and 1H are the identity elements of G and H.) Let
φ(1G ) = h. Then

1H h = h = φ(1G ) = φ(1G 1G ) = φ(1G )φ(1G ) = hh,

so h = 1H by the cancellation law. Similarly, if g ∈ G and φ(g) = h, then

φ(g−1 )φ(g) = φ(g −1 g) = φ(1G ) = 1H = h−1 h = φ(g)−1 φ(g)

so φ(g−1 ) = φ(g)−1 by the cancellation law. 2

In some sources the words monomorphism and epimorphism are used to
describe injective and surjective homomorphisms. The reason for avoiding
this terminology is clear from exercises to this discussion. On the other
hand, we happily use the word isomorphism that describes a bijective ho-
momorphism.
Examples. 1. If H is a subgroup of G, then the map φ : H → G defined by
φ(h) = h for all h ∈ H is an injective homomorphism. It is an isomorphism
if H = G.
2. Similarly, if R is a subring of S, then the map φ : R → S defined by
φ(h) = h for all h ∈ R is an injective homomorphism. It is an isomorphism
if R = S.
3. If G is an abelian group and r ∈ Z, then (gh)r = gr hr for all g, h ∈ G,
so the map φ : G → G defined by φ(g) = g r is a homomorphism.
Warning. This only works when G is abelian.
4. If V and W are vector spaces over the same field F then they are
abelian groups as well. Any linear map ψ : V → W is a group homomor-
phism.
Warning. The opposite is true only for very special fields (for instance, Q).

39
Otherwise, consider the complex conjugation x 7→ x∗ , C → C. It is a group
homomorphism but not a linear map of vector spaces over C, although it is
a linear map of vector spaces over R.
5. Let R be a commutative ring of characteristic p, that is, px = 0 for
any x ∈ R where p is a prime number. The ring R admits a Frobenius
homomorphism, F : R → R defined by F (x) = xp . The tricky part is the
preservation of addition. The identity (x + y)p = xp + y p is sometimes called
freshman’s dream binomial formula. It holds because the commutativity of
x and y implies that
p−1
X p!
(x + y)p = xp + y p + xk y p−k
k!(p − k)!
k=1

and all binomial coefficients in the sum are divisible by p.

6. Let k be an element of a group G. Then, for g, h ∈ G, we have
kghk−1 = kgk −1 khk−1 , so the map γk : G → G defined by γk (g) = kgk −1 is
a homomorphism. In fact it is an isomorphism, because kgk −1 = khk −1 ⇒
g = h by the cancellation laws, and each h ∈ G is equal to γk (k−1 hk).
Alternatively, we can observe that γk−1 = γk−1 .
Notice that if G is abelian, then whatever k we choose, we always get
φk (g) = g for all k, g so, these example is interesting for nonabelian groups
only. We have used γk in Lemma 7.4 already.
The elements g and γk (g) = kgk −1 are called conjugate elements. We
shall be come back to study this relationship later on.
7. Similarly for a ring R, pick k ∈ R× . The map γk : R → R defined
by γk (r) = krk−1 is a homomorphism. It is an isomorphism for the same
reason as in groups.
8. Let G = {1, a, b, c} be a Klein Four Group. Define φ : G → G by
φ(1) = 1, φ(a) = b, φ(b) = c, φ(c) = a. It is straightforward to check that
φ is an isomorphism.
9. If K is a commutative ring, by the determinant of a product rule,
det(AB) = det(A) det(B), it follows that the map φ : GL(n, K) → K ×
defined by φ(g) = det(g) is a homomorphism. The determinant of a product
rule is subtle to prove: you have seen a proof for a field in Linear Algebra-1.
We will not attempt the proof for a general K in this module/
10. There is an injective group homomorphism Ω : Sn → GL(n, R).
In words, Ω(σ) is a linear transformation of Rn permuting elements of the
standard basis ei . In formulas, Ω(σ)i,j = 1 whenever i = σ(j) and zero
otherwise.

40
 The composition det ◦Ω is particularly interesting homomorphism, called
sign homomorphism. Since σ m = 1 for some m > 1, det ◦Ω(σ)m = det ◦Ω(σ m ) =
1. There are only two real numbers 1 and −1 that have finite order, hence
we have a homomorphism sign : Sn → {−1, 1} ∼ = C2 which distinguishes
odd and even permutations. The former have sign(σ) = −1 and the latter
satisfy sign(σ) = 1.
Please, note that there is a risk of circular argument here. It all depends
on
P how you define determinant!! If you define it algebraically by det(ai,j ) =
σ∈Sn sign(σ)aσ(1),1 aσ(2),2 . . . aσ(n),n then you are in real trouble since you
are using determinant to define sign of a permutation and vise versa. We
will break this circle in Section 13.3.
9.2 Image
The image im(φ) of a homomorphism is just its image as a function, and
the following propositions are straightforward to prove.

Proposition 9.2 Let φ : G → H be a group homomorphism. Then im(φ)

is a subgroup of H.

Proposition 9.3 Let φ : R → S be a ring homomorphism. Then im(φ) is

a subring of S.

9.3 Kernels
Definition. Let φ : G → H be a homomorphism. Then the kernel ker(φ)
of φ is defined to be the set of elements of G that map onto 1H ; that is,

ker(φ) = {g | g ∈ G, φ(g) = 1H }.

In the case of additive group or rings this becomes

ker(φ) = {g | g ∈ G, φ(g) = 0H }.

Note that by Lemma 9.1 above, ker(φ) always contains 1G .

The following proposition explains the connection between normal sub-
groups and homomorphisms. Together with Proposition 10.3, it says that
the set of normal subgroups of G is equal to the set of kernels of group
homomorphisms with domain G.

Proposition 9.4 Let φ : G → H be a group homomorphism. Then ker(φ)

is a normal subgroup of G.

41
Proof: Checking that K = ker(φ) is a subgroup of G is straightforward,
using Proposition 1.5. If g ∈ G then

gK = φ−1 (φ(g)) = Kg,

so K is normal. 2
Examples. 10. Here is an example of a homomorphism from an additive
group to a multiplicative group. Let us define φ : C+ → C× by φ(g) =
exp(g). Then φ(g1 +g2 ) = φ(g1 )φ(g2 ), which says that φ is a homomorphism.
In fact φ is a surjective but not injective. The kernel of φ is 2πiZ since
exp(x + iy) = exp(x)(cos(y) + i sin(y)) for x, y ∈ R.
11. A close relative of example 10 is the homomorphism R : R+ →
O2 (R), essentially defined in Section 5.2. Recall that R(α) = Rα , the rota-
tion by α matrix. It is a homomorphism since Rα Rβ = Rα+β . It is neither
surjective, nor injective. Its image is SO2 (R) and its kernel 2πZ.
12. Let G = H = D12 , the dihedral group of order 12. We saw in
Subsection 5.3 that G = {ak | 0 ≤ k < 6} ∪ {ak b | 0 ≤ k < 6}. We define
φ : G → H by φ(ak ) = a2k and φ(ak b) = a2k b for 0 ≤ k < 6. We claim
that φ is a homomorphism. It seems at first sight as though we need to
check that φ(gh) = φ(g)φ(h) for all 144 ordered pairs g, h ∈ G, but we can
group these tests into the four distinct types listed in Subsection 5.3. We
will make free use of the fact that am = 1 when 6|m.
(i) φ(ak al ) = φ(ak+l ) or φ(ak+l−6 ) = a2(k+l) or a2(k+l−6) = a2k a2l =
φ(ak )φ(al );
(ii) φ(ak (al b)) = φ(ak )φ(al b) – this is similar to (i);
(iii) φ((ak b)al ) = φ(ak−l b) or φ(ak−l+6 b) = a2(k−l) b or a2(k−l+6) b = a2k a−2l b =
a2k ba2l = φ(ak b)φ(al )
(iv) φ((ak b)(al b)) = φ(ak b)φ(al b) – this is similar to (iii).
So φ really is a homomorphism. We can check that the only elements
of G with φ(g) = 1 are g = 1 and g = a3 , so ker(φ) = {1, a3 }, which is
the normal subgroup that we considered in Example 3 of Subsection 10.2.
im(φ) consists of the 6 elements 1, a2 , a4 , b, a2 b, a4 b of G.
In general, if φ : G → H is a homomorphism and J is a subset of
H, then we define the complete inverse image of J under φ to be the set
φ−1 (J) = {g ∈ G | φ(g) ∈ J}. It is easy to check, using Proposition 1.5,
that if J is a subgroup of H, then φ−1 (J) is a subgroup of G.
Here is a final statement, which will be useful later.

Proposition 9.5 Let φ : G → H be a homomorphism. Then φ is injective

if and only if ker(φ) = {1G } (or {0} in the case of rings or additive groups).

42
Proof: Since 1G ∈ ker(φ), if φ is injective, then we must have ker(φ) =
{1G }. Conversely, suppose that ker(φ) = {1G }, and let g1 , g2 ∈ G with
φ(g1 ) = φ(g2 ). Then 1H = φ(g1 )−1 φ(g2 ) = φ(g1−1 g2 ) (by Lemma 9.1), so
g1−1 g2 ∈ ker(φ) and hence g1−1 g2 = 1G and g1 = g2 . So φ is injective. 2
9.4 Exercises
(i) Let V and W be vector spaces over the field Q of rational numbers.
Show that any group homomorphism ψ : V → W is a linear map.
(ii) A homomorphism φ : A → B is called an epimorphism if for any pair
of homomorphisms α, β : B → C the equality αφ = βφ implies that
α = β. Prove that any surjective homomorphism is an epimorphism.
(iii) Prove that the natural embedding Z → Q is an epimorphism of rings
but not surjective.
(iv) Let G be a Klein Four Group. How many distinct homomorphisms
φ : G → G are there? How many of these are isomorphisms?
(v) Let φ : R → S be a ring homomorphism. Prove that if K is a subring
of S, then φ−1 (K) is a subring of R.
Prove that if A is a subring of R, then φ(A) is a subring of S.
Prove that if I is an ideal of S, then φ−1 (I) is an ideal of R.
Give an example where J is an ideal of R but φ(J) is not an ideal of
S.
(vi) Let us consider the following setup. For each natural number n we are
given a group Gn and a group homomorphism φn : Gn → Gn+1 Prove
that
∞
Y
G∞ = {(x1 , x2 , . . .) ∈ Gn |∀iφi (xi ) = xi+1 }
n=1
Q∞
is a subgroup of n=1 Gn .
(viii) Let p be a prime number. Let Gn = Cpn be the cyclic group of order
pn with a generator xn . We define φn : Gn → Gn+1 by φ(xan ) = xpa
n+1 .
Using the above construction we obtain a group G, called a quasicyclic
group and usually denoted Cp∞ . Prove that Cp∞ is isomorphic to H
n
where H = {z ∈ C× | ∃n z p = 1}.
(ix) Describe all distinct group homomorphisms from Cn to Cm and com-
pute their number. Which of them are ring homomorphisms from Zn
to Zm ?
9.5 Vista: Jacobian conjecture
Let R = C[x1 , . . . xn ] be the ring of complex polynomials. All C-linear
ring homomorphisms φ :→ R are easy to describe. Such a homomorphism
gives n polynomials fi = φ(xi ). In the other direct, any n-tuple of polyno-

43
mials fi define a linear ring homomorphism φ(F (x1 , . . . xn )) = F (f1 , . . . fn ).
The fun starts when if we want to decide which of them are isomorphisms
and n ≥ 2. Let us consider the Jacobian of φ:

Jφ = (∂fi /∂xj ) ∈ Mn (R).

If φ is an isomorphism then φφ−1 = I where φ−1 is the inverse homo-

morphism and I is the identity homomorphism given by fi = xi . By the
product rule in Analysis JφJφ−1 = JI but the latter is identity matrix.
Hence, Jφ ∈ GLn (R) and consequently det(Jφ) ∈ R× = C× . Indeed, the
units of R are non-zero polynomials of degree zero (prove it).
It is natural to suggest that the opposite holds: det(Jφ) ∈ R× should
imply φ being an isomorphism. This is one of the famous open problems in
Algebraic Geometry called Jacobian Conjecture. It has had a good share of
false proofs! Try to see where the difficulty12 is.

10 Quotient groups
We introduce quotient groups and prove the isomorphism theorem. We
use it to prove Cayley’s theorem.
10.1 Quotient Groups
Definition. If A and B are subsets of a group G, then we define their
product AB = {ab | a ∈ A, b ∈ B}.
The definition of quotient group depends on the following technical re-
sult.

Lemma 10.1 If N is a normal subgroup of G and [g] = N g, [h] = N h are

cosets of N in G, then [g][h] = [gh].
Proof: Let n1 g ∈ [g] = N g and n2 h ∈ [h] = N h. By normality of N ,
[g] = N g = gN , and so gn2 is equal to some element n3 g ∈ N g. Hence
(n1 g)(n2 h) = n1 (gn2 )h = n1 (n3 g)h = (n1 n3 )gh ∈ N gh = [gh], which proves
[g][h] ⊆ [gh]. Finally, ngh = (ng)(1h) ∈ (N g)(N h), so [gh] = N gh ⊆
(N g)(N h) = [g][h], and we have equality. 2
We are taking advantage of N being normal and don’t distinguish right
and left cosets.

12
read http://sbseminar.wordpress.com/2009/05/27/how-not-to-prove-the-jacobian-
conjecture/ before you write one of these http://arxiv.org/abs/0912.1924v1

44
Proposition 10.2 Let N be a normal subgroup of a group G. Then the set
G/N of cosets [g] = N g of N in G forms a group under multiplication of
sets.
Proof: We have just seen that [g][h] = [gh], so we have closure, and asso-
ciativity follows easily from associativity of G. Since [1][g] = [1g] = [g] for
all g ∈ G, [1] is an identity element, and since [g−1 ][g] = [g−1 g] = [1], [g−1 ]
is an inverse to [g] for all cosets [g]. Thus the four group axioms are satisfied
and G/N is a group. 2
Definition. The group G/N is called the quotient group (or the factor
group) of G by N .
Notice that if G is finite, then |G/N | = |G : N | = |G|/|N |. Let us finish
with the following fact.

Proposition 10.3 Let N be a normal subgroup of a group G. Then the map

φ : G → G/N defined by φ(g) = [g] is a surjective group homomorphism with
kernel N .
Proof: It is straightforward to check that φ is a surjective group homo-
morphism, and φ(g) = 1H ⇔ [g] = [1G ] = N ⇔ g ∈ N , so ker(φ) = N .
2
10.2 Examples of Quotient Groups
1. Let G be the infinite cyclic group (Z, +), and let N = nZ be its
subgroup generated by a fixed positive integer n – let’s take n = 5 just to be
specific. Now, by using Lemma 6.4 (after changing from multiplicative to
additive notation!), we see that the cosets [k] = 5Z+k and [j] = 5Z+j of N in
G are equal if and only if k ≡ j (mod 5). So there are only 5 distinct cosets,
namely [0] = N = N +0, [1] = N +1, [2] = N +2, [3] = N +3, [4] = N +4.
It is now clear that G/N is isomorphic to the group Z5 = {0, 1, 2, 3, 4} (see
Subsection 4.2) via the isomorphism [i] 7→ i for 0 ≤ i < 5.
2. Now let G = hgi be finite cyclic, and suppose that |g| = lm is composite.
Let N be the normal subgroup hgm i. Using methods of Subsection 3.2, we
can see that N has order l and consists of the elements {g mk | 0 ≤ k <
l}. Since all cosets have the form N gk for some k ∈ Z, it is clear that
G/N is cyclic and is generated by [g] = N g. We can calculate its order
as |G|/|N | = m. To see this directly, note that (using Lemma 6.4 again)
[gk ] = [gj ] ⇔ gk−j ∈ N ⇔ m|(k − j), and so the distinct cosets are [gk ] for
0 ≤ k < m. In particular, [g]m = [gm ] = N 1 is the identity element of G/N ,
and |[g]| = m.
3. For a more complicated example, we take Example 4 of Subsection 8.1,

45
namely G = D12 and N = {1, a3 }. Then |G/N | = |G|/|N | = 6. Since
a3 ∈ N , we have
[a]3 = [a3 ] = N a3 = N = [1]
is the identity of G/N . We also have [b]2 = [1] and [b][a] = [a−1 ][b], because
these relations are inherited from the corresponding relations of G. Thus
G/N is a group of order 6 satisfying the three relations [a]3 = 1, [b]2 = 1,
[b][a] = [a]−1 [b], and, by Proposition 5.5 G/N ∼= D6 .
It might be helpful in understanding this example to see the full multipli-
cation table of G (cf. Subsection 5.3), with the elements arranged according
to their cosets. Notice that all elements in each 2 × 2 block of this table lie
in the same coset of N in G. We can then see the multiplication table of
G/N by regarding these 2 × 2 blocks as single elements (i.e. cosets) in the
quotient group.

N Na N a2 Nb N ab N a2 b
1 a3 a a4 a2 a5 b a3 b ab a4 b a2 b a5 b
N 1 1 a3 a a4 a2 a5 b a3 b ab a4 b a2 b a5 b
a3 a3 1 a4 a a5 a2 a3 b b a4 b ab a5 b a2 b
Na a a a4 a2 a5 a3 1 ab a4 b a2 b a5 b a3 b b
a4 a4 a a5 a2 1 a3 a4 b ab a5 b a2 b b a3 b
Na 2 a2 a2 a5 a3 1 a4 a a2 b a5 b a3 b b a4 b ab
a5 a5 a2 1 a3 a a4 a5 b a2 b b a3 b ab a4 b
Nb b b a3 b a5 b a2 b a4 b ab 1 a3 a5 a2 a4 a
a3 b a3 b b a2 b a5 b ab a4 b a3 1 a2 a5 a a4
N ab ab ab a4 b b a3 b a5 b a2 b a a4 1 a3 a5 a2
a4 b a4 b ab a3 b b a2 b a5 b a4 a a3 1 a2 a5
N a2 b a2 b a2 b a5 b ab a4 b b a3 b a2 a5 a a4 1 a3
a5 b a5 b a2 b a4 b ab a3 b b a5 a2 a4 a a3 1

10.3 The isomorphism theorem

Sometimes it is also called the first isomorphism theorem, see the exer-
cises for the second and the third ones.

Theorem 10.4 (Isomorphism theorem for groups) Let φ : G → H be a

group homomorphism with the kernel K. Then G/K ∼ = im(φ). More pre-
cisely, there is an isomorphism φ : G/K → im(φ) defined by φ( K [g]) = φ(g)
for all g ∈ G.
Proof: The trickiest point to understand in this proof is that we have to
show that φ( K [g]) = φ(g) really does define a map from G/K to im(φ).

46
The reason that this is not obvious is that we can have K [g] = K [h]
with g 6= h, and when that happens we need to be sure that φ(g) = φ(h).
This is called checking that the map φ is well-defined. In fact, once you
have understood what needs to be checked, then doing it is quite easy,
because K [g] = K [h] ⇒ g = kh for some k ∈ K = ker(φ), and then
φ(g) = φ(k)φ(h) = φ(h).
Clearly im(φ) = im(φ), and it is straightforward to check that φ is a
homomorphism. Finally,

φ( K [g]) = 1H ⇐⇒ φ(g) = 1H ⇐⇒ g ∈ K ⇐⇒ K [g] = K = 1G/K ,

and so φ is a monomorphism by Proposition 9.5. Thus φ : G/K → im(φ) is

an isomorphism, which completes the proof. 2
A particular value of the first isomorphism theorem is that it tells us the
structure of any homomorphism φ : G → H. Indeed, φ is composition of
three other homomorphism: the quotient homomorphism G → G/ ker(φ),
the isomorphism φ and the embedding im(φ) → H. For instance, R : R+ →
O2 (R) from Example 12 in Section 9 becomes a composition of three maps
p ψ ι
→ R+ /2πR −
R : R+ − → SO2 (R) −
→ O2 (R)

the quotient map p(α) = [α], the isomorphism ψ([α]) = Rα and the embed-
ding ι(T ) = T .
Let us illustrate this theorem using Example 12 from Section 9. Note
that the elements of G = D12 are listed in two separate columns in the
diagram, in different orders, once for the domain and once for the codomain
of φ. The elements of imφ are printed slightly to the left of those not in
im(φ) in the codomain column.

47
 φ

1
a3 } = [1] - 1
a
a
a4 } = [a] - a2
a3
a2
a5 } = [a ]
2 - a4
a5
b
a3 b } = [b] - b
ab
ab
a4 b } = [ab] - a2 b
a3 b
a2 b
a5 b } = [a b]
2 - a4 b
a5 b
10.4 Cayley Theorem
As an application, let us prove Cayley’s theorem.

Theorem 10.5 (Cayley’s Theorem) Every group G is isomorphic to a per-

mutation group. (That is, to a subgroup of Sym(X) for some set X.) If G
is finite, the set X can be chosen finite.
Proof: Let X = G. Define a homomorphism φ by φ(x) : y 7→ xy. By
Theorem 10.4, G/ ker(φ) is isomorphic to a subgroup of Sym(X). Let us
compute the kernel. If x ∈ ker(φ) then φ(x) = IdG . Hence, x = φ(x)(1) =
Id(1) = 1. So the kernel is trivial and we are done. 2
Why don’t mathematicians study permutation groups instead of groups?
Well, they do both! Permutation group is not really just a group, but a group
embedded in Sn , there could be different embeddings. You can consider Sn
in Sn or Sym(Sn ) = Sn! .
10.5 Exercises
Exercise (iii) below is often referred to as the second isomorphism the-
orem, while exercise (iv) is referred to as the third isomorphism theorem.

(i) A group G is simple if it has exactly two normal subgroups: G and 1.

Prove that a simple abelian group is a cyclic group of prime order.

48
 (ii) Prove that if G is a simple group and H is a subgroup of index is simple
if it has exactly two normal subgroups: G and 1. Prove that a simple
abelian group is a cyclic group of prime order.
(iii) Let N G. Prove that the subgroups of G/N are precisely the quotient
groups I/N , for subgroups I of G that contain N .
(iv) Let H be any subgroup and let K be a normal subgroup of a group G.
Then H ∩ K is a normal subgroup of H and H/(H ∩ K) ∼ = HK/K.
(v) Let K ⊆ H ⊆ G, where H and K are both normal subgroups of G.
Then (G/K)/(H/K) ∼ = G/H.
10.6 Vista: simple groups
Classification of simple finite groups was both a major success and a
major tragedy of the 20-th century mathematics. While the theorem is
beautiful, the proof has spread over 30,000 journal pages. And up until now
we have not got an acceptably written proof. You can read more about
this topic on wikipedia, search for “Classification of finite simple groups”.
Currently, mathematicians work on second and third generation proofs. Inna
Capdeboscq in Warwick is actively involved in both projects.
If you are thinking of writing an essay, it may be a good idea to describe
all simple groups of order up to 1000. Roger Carter used to teach MA4**
level module with exactly this topic (and title). Besides 168 cyclic groups
of prime order, there are only five nonisomorphic groups: the alternating
groups A5 of order 60 and A6 of order 360 as well as linear groups GL3 (F2 )
of order 168, P GL2 (F8 ) of order 504 and P SL2 (F11 ) of order 660. While
for the most numbers n, it is relatively easy to show that there is no simple
group of order n, numbers 120, 540 and 720 will require special attention13 .

11 Ideals and quotient rings

We introduce ideals and discuss quotient rings. We state, prove and use
the isomorphism theorem for rings.
11.1 Ideals
Definition. An additive subgroup I of a ring R is called an ideal in R if
xI ⊆ I ⊇ Ix for any x ∈ R. One writes I  R.
To clarify the notation, xI = {xr | r ∈ I}. Thus, for any r ∈ I both
xr and rx must be in I for any x ∈ R. For a commutative ring R these
13
cf. http://mathoverflow.net/questions/41958/no-simple-groups-of-order-720 and
http://sci.tech-archive.net/Archive/sci.math/2006-12/msg07456.html

49
two properties are the same. For a non-commutative ring, one sometimes
introduces left ideals satisfying xI ⊆ I and right ideals satisfying Ix ⊆ I
Examples. 1. Any ring R has two boring ideals: zero ideal {0} and the
ring R itself.
2. For any n ∈ Z, the numbers in Z divisible by n form an ideal nZ. Notice
that 1Z = Z and 0Z = {0}.
3. Let R be a non-zero ring, n ≥ 2. Let I be the set of all matrices in
Mn (R) that vanish outside one particular row. Then I is a right ideal but
not an ideal. In particular, I is not a left ideal. Likewise, let J be the set
of all matrices in Mn (R) that vanish outside one particular column. Then
J is a left ideal but not an ideal. In particular, J is not a right ideal.
We won’t take any interest in this course in the left and right ideals.
They will be studied in the third year module Rings and Modules.

Proposition 11.1 Let φ : R → S be a ring homomorphism. Then ker(φ)

is an ideal in R.
Proof: By Proposition 9.4, K = ker(φ) is an additive subgroup of R. If
r ∈ K, x ∈ R then φ(xr) = φ(x)φ(r) = 0S φ(r) = 0S . Hence xr ∈ K.
Similarly, rx ∈ K and K is an ideal. 2
Example. 4. For rings R, S let us consider the projection R × S → S given
by (r, s) 7→ s. It is a ring homomorphism whose kernel is R. Thus, R sitting
inside R × S is an ideal (but not a subring as we saw earlier).
11.2 Generators of an ideal
Proposition
P 11.2 Let R be a ring, x1 , x2 . . . ∈ R. The subset (x1 , x2 . . .) =
{ k,j rk xj sk |rk , sk ∈ R} is an ideal in R.
Proof: The subset is clearly non-empty. Let us clarify P that the subset
consists of the finite sums. P The differenceP of two sums k,j rk xj sk is also
such
P a sum. Finally,
P a( r x s
k,j k j k ) = k,j (ar )x s
k j k ∈ (x1 , x2 . . .) and
( k,j rk xj sk )a = k,j rk xj (sk a) ∈ (x1 , x2 . . .). 2
If I  R and I = (x1 , x2 . . .) we say x1 , x2 . . . are generators of the ideal
I. If I = (x1 , x2 . . . , xn ), I is called finitely generated. If I = (x), I is called
principal. We have already seen the principal ideals (n) = nZ in Z.
P P
Examples. 5. If R is commutative nk=1 rk xsk = x( nk=1 rk sk ), so (x) =
xR = {xr|r ∈ R}.
6. In a noncommutative ring the principal ideals are trickier. For instance,
(E1,1 ) = Mn (R) where Ei,j is a matrix with all zeroes except a single en-
try 1R , on the intersection of the i-th row and the j-th column. The key

50
calculation X
(rij ) = (rk,l Ek,1 )E1,1 E1,l
k,l

writes an arbitrary matrix (rij ) as an element of the principal ideal.

This example shows that the following lemma fails for a non-commutative
ring. In the case of commutative rings, the lemma provides a useful charac-
terisation of units.

Lemma 11.3 Let R be a commutative ring, x ∈ R. Then x ∈ R× if and

only if (x) = R.
Proof: x ∈ R× if and only if ∃y xy = 1R if and only if 1R ∈ (x) if and only
if (x) = R. The last forward implication is clear because if 1R belongs to an
ideal I then a = a · 1 ∈ I for all a ∈ R. 2
11.3 Quotient rings
The following proposition defines the quotient ring R/I for a ring R and
its ideal I.

Proposition 11.4 The additive cosets of an ideal I  R form a ring under

addition [a] + [b] = [a + b] and multiplication [a] · [b] = [ab].
Proof: Recall that the additive cosets are [a] = a + I. The form an abelian
quotient group R+ /I with zero [0] = 0 + I = I. We need to check that the
multiplication is well defined: Let [a] = [x] and [b] = [y]. This means that
a− x ∈ I ∋ b− y. Then ab = ab− ay + ay − xy + xy = a(b− y)+ (a− x)y + xy.
Hence ab − xy ∈ I and [ab] = I + ab = I + xy = [xy].
The ring axioms (associativity, distributivity and unity) follow from the
ring axioms of R. For instance, [1] = 1 + I is the unit. 2
Example. 7. The quotient ring Z/(n) is isomorphic to the ring Zn of the
residues modulo n. The isomorphism Zn → Z/(n) is just m 7→ m + (n).
Remember that we have never checked formally that Zn is a ring. This
would do it. In fact, Z/(n) should be thought of as the definition of Zn .

Proposition 11.5 Let I be an ideal of a ring R. Then the map φ : R →

R/I defined by φ(r) = [r] is a surjective ring homomorphism with kernel I.
Proof: By definition of R/I, φ is a ring homomorphism. Since [x] = φ(x),
φ is surjective. Finally, φ(x) = 0 ⇔ [x] = I + x = I ⇔ x ∈ I, so ker(φ) = I.
2

51
11.4 The isomorphism theorem
The following theorem is also called the first isomorphism theorem, see
the exercises for the second and the third ones.

Theorem 11.6 (Isomorphism theorem for rings) Let φ : R → S be a ring

homomorphism with the kernel I. Then R/I ∼ = im(φ). More precisely, there
is an isomorphism φ : R/I → im(φ) defined by φ([x]I ) = φ(x) for all x ∈ R.
Proof: By Theorem 11.6, φ is a well-defined isomorphism of abelian groups
under addition. It remains to see that this is a ring homomorphism, which
follows from φ being a ring homomorphism. 2
Let us look at two examples where the isomorphism theorem is useful.
Examples. 8. Let R = Z[i]/(1 − 2i). Let us denote J = (1 − 2i), its
additive cosets x + J = [x]. The map ψ : Z → R, ψ(n) = [n] is obviously a
ring homomorphism. It is surjective because (1 − 2i)i = i − 2 ∈ J implies
[i] = [−2]. Hence, [a + bi] = [a] + b[i] = [a] + b[−1] = [a − 2b] = ψ(a − 2b)
is in the image. If n ∈ ker(ψ) then [n] = [0] and n ∈ J. This means that
n = (x + yi)(1 − 2i) = (x + 2y) + (y − 2x)i for some x, y ∈ Z. Hence, y = 2x
and n = 5x Hence, ker(ψ) = (5) and, by the isomorphism theorem, R ∼ = Z5 .
2
9. We want to compute the ring Ra = R[x]/(x − a) for some a ∈ R.
Every real (or complex) number α defines an evaluation homomorphism
fα : R[x] → R, fα (F (x)) = F (α) (or fα : R[x] → C correspondingly).
The idea is to realize J = (x2 − a) as the kernel of a suitable evaluation
homomorphism.
√
If a < 0, we choose the square root b = a and consider the homomor-
phism fb : R[x] → C. It is useful to observe that besides a ring homomor-
phism fb is an R-linear map, in particular, its image is an R-vector subspace.
Since fb (1) = 1(b) = 1 and fb (x) = b is imaginary, fb is surjective. Now
fb (x2 − a) = b2 − a = 0, hence x2 − a ∈ ker fb . Since the kernel is ideal,
J ⊆ ker fb . It remains to observe that both J and ker fb have codimension
2. Hence, J = ker fb and Ra ∼ = C by the isomorphism theorem, which also
tells you explicit isomorphism ψ = fb : Ra → C, ψ([F (x)]) = F (b).
If a > 0, we need two evaluations to cook a ring homomorphism ψ =
(f−√a , f√a ) : R[x] → R × R. The same argument will lead to J = ker ψ and
Ra ∼= R × R.
11.5 Exercises
Exercise (ii) often referred to as the second isomorphism theorem for
rings, while exercise (iii) is referred to as the third isomorphism theorem for
rings.

52
 P in R. Show that I ∩ J, I + J = {x + y | x ∈ I, y ∈ J}
(i) Let I, J be ideals
and IJ = { i xi yi | xi ∈ I, yi ∈ J} are all ideals of R and that
IJ ⊆ I ∩ J ⊆ I + J.
(ii) Let S be any subring and let I be an ideal in a ring R. Then S ∩ I is
an ideal in S and S/(S ∩ I) ∼
= (S + I)/I.
(iii) Let I ⊆ J ⊆ R, where I and J are both ideals in R. Then J/I is an
ideal in R/I and (R/I)/(J/I) ∼ = R/J.
(iv) Compute the ring Z[i]/(1 + 3i).
(v) Compute the ring Z[i]/(5 + 3i).
(vi) Let Ra = R[x]/(x2 − a). Show that R0 , R1 and R−1 are pairwise
non-isomorphic rings.
(vii) Rf = R[x]/(f (x)) where f (x) is a quadratic polynomial. Show that
Rf is isomorphic to either R0 , or R1 , or R−1 .
(viii) Let R be a ring with an ideal Ii for each natural i. Prove that if
Ii ⊆ Ii+1 then the union J = ∪∞ n=1 In is an ideal.

11.6 Vista: simple rings

Similarly to a group, a ring R is called simple if it contains exactly two
ideals 0 and R. In particular, the zero ring is not simple. A field is a simple
ring. More generally, Mn (F ) is a simple ring if F is a field.
Similarly to groups, one can inquire about classification of simple rings.
In groups it is unreasonable to expect to classify all simple groups and the
right question is to classify finite simple groups. Finite simple rings can be
classified: they are all Mn (Fq ) where Fq is a finite field. In fact, there is
a more general class of artinian rings where simple rings can be classified:
they are all Mn (D) where D is a division ring. This result is called Artin-
Wedderburn Theorem and is covered in the 3rd year Rings and Modules.

12 Chinese remainder theorem

We prove Chinese Remainder Theorem (CRT). We use it to compute
Euler’s function and for public key encoding.
12.1 CRT, abstract form
Let R be a ring, Ik a set of ideals. For each ideal in the set we have the
quotient homomorphism Q ψk : R → R/Ik , ψk (x) = x + Ik . We can fuse them
together into ψ : R → k R/Ik , ψ(x) = (x + Ik ).
Q
Lemma 12.1 The map ψ : R → k R/Ik is a ring homomorphism with a
kernel ∩k Ik .

53
Proof: It follows easily from the properties of the quotient maps that ψ is
a ring homomorphism. Now x ∈ ker ψ ⇔ ∀k x + Ik = 0 + Ik ⇔ ∀k x ∈ Ik ⇔
x ∈ ∩ k Ik 2
The isomorphism theorem tells us thatQthere is an induced injective
homomorphism of rings ψ : R/ ∪k Ik → k R/Ik . In general, Chinese
Remainder Theorem (CRT) is any statement that concludes that ψ is an
isomorphism. We are going to prove for the ring Z.

Theorem 12.2 (CRT, abstract form) Let ni , i = 1, . . . t be natural numbers

such that each pair ni , nj , i 6= j is coprime.QLet N = n1 · . . . · nt . Then
∩i (ni ) = (N ) and the natural map ψ : ZN → i Zni is a ring isomorphism.
Proof: The ideal ∩i (ni ) consists of all k divisible by all ni . This is equiv-
alent to divisibility by N since ni are pairwise coprime.
By Lemma 12.1, ψ is an injective ring homomorphism. It is an isomor-
phism since both rings have N elements. 2
Theorem 12.2 fails if there are infinitely many ideals (see exercises). On
the other hand, it holds for any commutative ring if one replaces coprime
generators by comaximal ideals. The proof gets much longer and you can
follow the proof in the exercises. The noncommutative rings require a com-
pletely different statement that you can also find in the exercises.
12.2 Computation of Euler’s function
Recall that the Euler totient function ϕ : N → N is defined as ϕ(m) =
|Zm × | (cf. Lemma 7.8).

Lemma 12.3 If R and S are rings, the groups (R × S)× and R× × S × are
isomorphic.
Proof: Both groups are subsets of the ring R × S. Let us observe that
these subsets are equal: (r, s) ∈ (R × S)× if and only if ∃(a, b) ∈ R ×
S (a, b)(r, s) = (1, 1) if and only if ∃a ∈ R, b ∈ S ar = 1, bs = 1 if and only if
r ∈ R× & s ∈ S × if and only if (r, s) ∈ R× × S × . The required isomorphism
is the identity map, which clearly preserves the multiplication. 2
We are ready to compute it now.

Corollary 12.4 If m and n are coprime then ϕ(mn) = ϕ(m)ϕ(n)

Proof: It follows immediately from Theorem 12.2 and Lemma 12.3. 2

Corollary 12.5 If m = pa11 · · · pakk where pi are distinct primes then ϕ(m) =
Qk ai ai −1
i=1 (pi − pi ).

54
 Q
Proof: By corollary 12.4, ϕ(m) = ki=1 ϕ(pai i ). By Lemma 12.3, m ∈ Zpai
i
is a unit if and only if m is coprime to pai i . The latter is equivalent to not
being divisible by pi . The remainders divisible by pi are of the form xpi for
some x, so there are exactly piai −1 of them. Hence ϕ(pai i ) = pai i − piai −1 . 2
12.3 RSA
Suppose that you run a website so popular with customers, that every
minute thousands of customers are queueing up to pay you with their credit
cards. The key exchange with each of them will slow you down and you
need an asymmetric keys solution instead. You need two keys: a private
key used and known only to you and a public key that you would give to
any customers, including hackers trying to compromise the system. More
serious industrial application of asymmetric keys is the PGP system, which
allows you to exchange messages securely with any number of people on the
internet by giving away your public key.
The source of the key is two large (approximately 1000 bits or so) primes
p and q. The public key consists of the product n = pq and the exponent e
such that gcd(e, ϕ(n)) = 1. Popular choices of public exponent are 65537 =
216 + 1 or 17 = 24 + 1. The private key consists of the private exponent d
such that (ed)ϕ(n) = 1. This number is precomputed once and stored.
The public key is available to anybody shopping online. The credit card
number is padded with bits upfront to form a message m. Padding ensures
the security condition me ≫ n but making sure m is not divisible by p an
q, usually by p > m < q. The encoded message x = (me )n is send over the
internet. The choice of public exponent usually enables fast calculation. For
instance, m65537 is computed using 17 multiplications.
The vendor receives the message and decrypts it by m = (med )n = (xd )n
with the first equality thanks to Euler’s theorem. It is more computationally
intensive but doable: d could have up to 1,000,000 binary bits, so xd may
require up to 2,000,000 multiplications.
A hacker can easily collect the following ingredients: x, e, n. To get to the
credit card number m, the hacker needs d or ϕ(n). He(she) will need either
to decompose n into the product of p and q, which gives ϕ(n), or compute
d directly in the group Z× n by some other means. With such a large n any
known method is going to take thousands of years. Mathematicians know
at least one way to do it faster by using quantum computers. It is engineers’
turn to figure out how to build them.

55
12.4 CRT, elementary form
We recall that one writes x ≡n y if n divides x − y. The Chinese Re-
mainder Theorem can be formulated on the level of systems of comparisons.

Theorem 12.6 (CRT, elementary form) Let ni , ki ∈ Z, i = 1, . . . t such

that each pair ni , nj , i 6= j is coprime. The system of comparisons x ≡ni ki
admits a solution in Z. Any two solutions are different by a multiple of
N = n1 n2 · · · nt .
Proof:
Q Existence of solution is surjectivity of the natural map ψ : Z →
i Z/(ni ), ψ(x) = (x + (n1 ), . . . , x + (nt )), established in Theorem 12.2 Let
x ∈ Z be a solution. Now y ∈ Z is a solution if and only if x − y ∈ ker ψ. It
remains to notice that ker psi = (N ) by Lemma 12.1
Y
ψ:Z→ Z/(ni ), ψ(x) = (x + (n1 ), . . . , x + (nt )).
i

2
Our proof of CRT is not constructive. We do not know from the proof
how to solve the system of comparisons. On the other hand, one can easily
derive a constructive proof from the solutionQmethod we are about to de-
scribe. The key is the number Ni = N/ni = k6=i nk . It is divisible by any
nj , j 6= i and coprime to ni . So, it is a generator of Cni inside Z+ ∼ Q Cn .
N = j j
Let us look at a concrete example. Let us solve x ≡7 6, x ≡11 5,
x ≡13 4. In this case, N = 7 · 11 · 13 = 1001 and the generators are
N1 = 143 ≡7 3, N2 = 91 ≡11 3, N3 = 77 ≡13 12 ≡13 −1. In Z7 we have
6 = 2 · 3, in Z11 we have 5 = 9 · 3 and in Z13 we have 4 = 9 · (−1). Hence,
x = 2N1 + 9N2 + 9N3 = 2 · 143 + 9 · 91 + 9 · 77 = 286 + 819 + 693 = 1798
is a solution. Any other other solution is 1798 + 1001k. In particular.
1798 − 1001 = 797 is the minimal positive solution.
12.5 Exercises
Exercises (v)-(x) and above take you through two general versions of
CRT.
(i) Let pk be the k-th prime
Q number, Ik = (pk ). Prove that the natural
map ψ : Z/ ∩k Ik → k Z/Ik is not an isomorphism. Conclude that
CRT fails for infinite sets of ideals.
(ii) Explain how to compute m65537 using 17 multiplications.
(iii) Explain why the security condition me ≫ n is necessary.
(iv) Find the smallest positive x such that x ≡7 3, x ≡13 4, x ≡19 5.

56
 (v) Let I, J  R. We say that I and J are comaximal if I + J = R. Prove
that the natural map ψ : R/I ∩ J → R/I × R/J is an isomorphism if
and only if I and J are comaximal.
(vi) Let I1 , . . . In be a collection of ideal in R. Suppose Ik and ∩j<k Ij are
comaximal for each k = 2, 3, . . . n. Using Q induction, prove that the
natural homomorphism ψ : R/ ∩k Ik → k R/Ik is an isomorphism.
(vii) Let R be commutative, I, J  R. Prove that (I + J)(I ∩ J) ⊆ IJ.
(viii) Let R be commutative, I, J  R comaximal. Prove that I ∩ J = IJ.
(ix) Let R be commutative, Ik , k = 1, . . . n pairwise comaximal ideals of
R. Using induction, prove that In and ∩k<n Ik are comaximal.
(x) Let R be commutative, Ik , k = 1, . . . n pairwise
Q comaximal ideals of R.
Prove that the natural map ψ : R/∩k Ik → k R/Ik is an isomorphism.
(xi) Let I, J  R be comaximal, R commutative. Prove that I n and J m
are comaximal for all n and m.
12.6 Vista: prime factorisation
Read more about prime numbers and their factorisation in Lauritzen.
The now defunct RSA challenge14 used to pay you money for factoring one
of the numbers of the form pq. This is all related to the P=NP problem, one
of the millennium problems with one million dollars bounty15 . You may also
want to find out more about quantum computers. Imagine notoriety you
can achieve if you find a way of factoring large numbers: you may become
the top criminal of the century by compromising all secure internet traffic
with payment information.

13 Actions, stabilisers and alternating group

We introduce the notion of actions, and discuss stabilisers. We use to
prove that the parity of a permutation is well defined.
13.1 Definition and action
Definition. Let G be a group and X a set. An action of G on X is a map
· : G × X → X, which satisfies the properties:
(i) (gh) · x = g · (h · x) for all g, h ∈ G, x ∈ X;
(ii) 1G · x = x for all x ∈ X.
In this definition, the image of (g, x) under the map · is denoted by g · x.
If G acts on a set X, we would refer to X as a G-set.
14
http://www.rsasecurity.com/rsalabs/node.asp?id=2093
15
http://en.wikipedia.org/wiki/Complexity classes P and NP

57
Proposition 13.1 Let · be an action of the group G on the set X. For g ∈
G, define the map φ(g) : X → X by φ(g)(x) = g · x. Then φ(g) ∈ Sym(X),
and φ : G → Sym(X) is a homomorphism.
Proof: Property (ii) in the definition says that φ(1G ) is the identity map
IX : X → X, and then Property (i) implies that φ(g)φ(g −1 ) = φ(gg−1 ) =
IX , and similarly φ(g−1 )φ(g) = IX . So φ(g) and φ(g −1 ) are inverse maps,
which proves that φ(g) : X → X is a bijection. Hence φ(g) ∈ Sym(X), and
then Property (i) implies immediately that φ is a homomorphism. 2
The opposite is also true: a homomorphism φ : G → Sym(X) defines an
action g · x = φ(g)(x). In fact, it gives a bijection between the set of actions
G × X → X and the set of homomorphisms φ : G → Sym(X) (see exercises
and Example 1).
The kernel of an action · of G on X is defined to be the kernel K = ker(φ)
of the homomorphism φ : G → Sym(X) defined in Proposition 13.1. So

K = {g ∈ G | g · x = x for all x ∈ X}.

The action is said to be faithful if K = {1}. In this case, Theorem 10.4

says that G ∼
= G/K ∼ = im(φ), which we state as a proposition, which can be
thought of as a generalisation of Cayley’s Theorem (Theorem 10.5).

Proposition 13.2 If · is a faithful action of G on X, then G is isomorphic

to a subgroup of Sym(X).

Examples. 1. If G is a subgroup of Sym(X), then we can define an action

of G on X simply by putting g · x = g(x) for x ∈ X. This action is faithful.
2. Let P = {vi | i ∈ Z6 } be a regular hexagon, G =< a = R2π/6 , b =
S0 >= D12 its group of isometries. We use notation of Section 5.4 here.
The action of G on the vertex set gives a homomorphism φ : G → S6 where
φ(a) = (v0 , v1 , v2 , v3 , v4 , v5 ) and φ(b) = (v1 , v5 )(v2 , v4 ).
There are some other related actions however. We could instead take X
to be the set E = {e1 , e2 , e3 , e4 , e5 , e6 } of edges of P , where e1 is the edge
joining v6 and v1 , e2 joins v1 and v2 , etc. The map φ of the action of G on
E is then given by φ(a) = (e1 , e2 , e3 , e4 , e5 , e6 ), φ(b) = (e1 , e6 )(e2 , e5 )(e3 , e4 ).
(Notice that any homomorphism is fully specified by the images of a set of
group generators, because the images of all other elements in the group can
be calculated from these.) This action is still faithful.
As a third possibility, let D = {d1 , d2 , d3 } be the set of diagonals of P ,
where d1 joins vertices v1 and v4 , d2 joins v2 and v5 , and d3 joins v3 and
v0 . Then map φ of the action of G on D is defined by φ(a) = (d1 , d2 , d3 ),

58
and φ(b) = (d1 , d2 ). This action is not faithful, and its kernel is the normal
subgroup {1, a3 } of G that we have already studied. The image is isomorphic
to D6 .
3. There is a faithful action called the regular action, defined for any
group G. The regular G-set X is the set of G as a set. The action g · x
is defined to be gx for all g ∈ G, x ∈ X. Conditions (i) and (ii) of the
definition obviously hold, so we have defined an action. If g is in the kernel
K of the action, then gx = x for all x ∈ X, which implies g = 1 by the
cancellation law, so the action is faithful. From Proposition 13.2, we can
deduce Cayley’s Theorem (Theorem 10.5).
4. Similarly to the regular G-set, there exists antiregular G-set for any
group G. The antiregular G-set X = G, as a set. The action is defined via
g · x = xg −1 for all g ∈ G, x ∈ X.
13.2 Stabilisers
Definition. Let X be a G-set, x ∈ X. The stabiliser of x in G, denoted by
Gx or StabG (x), is {g ∈ G | g · x = x}.
The proof of the following proposition is left as an exercise.

Proposition 13.3 Let G act on X and x ∈ X. Then

(i) StabG (x) is a subgroup of G;
(ii) ∩x∈X StabG (x) is the kernel of the action of G on X.
13.3 Action, rings and alternating groups
When a group G acts on a set X with additional structure, we usually
want the bijections x 7→ g · x to preserve this structure.
Definition. Let G be a group and R a ring. An action of G on the ring
R is an action on the set · : G × R → R such that x 7→ g · x is a ring
homomorphism.
Notice that since the action of g is a bijection, this ring homomorphism
must be a ring isomorphism. We recall explicitly the properties of a homo-
morphism.
(i) g·(ab) = (g·a)(g·b) and g·(a+b) = (g·a)+(g·b) for all g ∈ G, a, b ∈ R;
(ii) g · 1R = 1R for all g ∈ G.
One can define the automorphism group

Aut(R) = {φ ∈ Sym(R) | φ is a ring homomorphism}.

Its elements, isomorphisms R → R, are called automorphisms. See Exercise

(iii) for an adaptation of Proposition 13.1 to the situation of rings.

59
 Let us consider the action of the symmetric group Sn on the set R =
Z[X1 , . . . Xn ] defined by the obvious formula σ·F (X1 , . . . Xn ) = F (Xσ(1) , . . . Xσ(n) ).
For example, if σ = (1, 2, 3) and F = 1 + X1 + X23 + X3 X4 then σ · X1 = X2 ,
σ · X2 = X3 , σ · X3 = X1 , σ · X4 = X4 and σ · F = F (X2 , X3 , X1 , X4 ) =
1 + X2 + X33 + X1 X4 . Let point it out that the action of Sn on the set R is
actually an action on the ring R.
Definition. TheQalternating group An is the stabiliser of the Vandermonde
polynomial Ω = i>j (Xi − Xj ) ∈ Z[X1 , . . . Xn ].
We need the following calculation.

Proposition 13.4 If σ = (i, j) is a transposition then σ · Ω = −Ω.

Proof: Without loss of generality, i > j. We represent Ω as a product of
5 functions Ω = ω1 ω2 ω3 ω4 ω5 and compute all the actions separately. First
ω1 = Xi − Xj and σ · ω1 = Xj − Xi = −ω1 . Then
Y
ω2 = (Xt − Xs )
t>s,t6=i,t6=j,s6=i,s6=j

in which no Xi or Xj appear. Consequently, σ · ω2 = ω2 . The remaining

three functions are similar, as they pair Xi and Xj with Xt in one of 3
regions:
Y Y
ω3 = (Xt − Xi )(Xt − Xj ), σ · ω3 = (Xt − Xj )(Xt − Xi ) = ω3 ,
t>i t>i
Y Y
ω4 = (Xi − Xt )(Xt − Xj ), σ · ω4 = (Xj − Xt )(Xt − Xi ) = ω4 ,
i>t>j i>t>j
Y Y
ω5 = (Xi − Xt )(Xj − Xt )), σ · ω5 = (Xj − Xt )(Xi − Xt ) = ω5 .
t<j t<j

Since Sn acts on the ring R, σ · Ω = (s · ω1 )(s · ω2 )(s · ω3 )(s · ω4 )(s · ω5 ) = −Ω.

2
Proposition 13.4 allows to describe An in the standard terms. We say
that a permutation σ ∈ Sn is even if it is a product of even number of trans-
positions and that σ is odd if it is a product of odd number of transpositions.
Apriori, it is not clear why a permutation cannot be both odd and even.

Corollary 13.5 Each permutation σ ∈ Sn is either odd or even. The al-

ternating group An consists of even permutations in An .

60
Proof: Observe that Ω 6= −Ω since 2Ω 6= 0. If σ = τn · · · τ1 then σ · Ω =
(−1)n Ω by Proposition 13.4. Hence, σ is even (odd) if and only if σ · Ω = Ω
(σ · Ω = −Ω correspondingly). 2
The next two corollaries are immediate.
σ·Ω
Corollary 13.6 The sign function sign : Sn → Z× defined by sign(σ) = Ω
is a well-defined group homomorphism.

Corollary 13.7 If n ≥ 2 then |An | = n!/2.

Proof: Since the sign function is a surjective homomorphism, Sn /An ∼
=
Z× ∼= C2 by the isomorphism theorem. Hence, |Sn |/|An | = 2 and |An | =
n!/2. 2
13.4 Exercises
(i) Let G be a group, X a set. Prove that there is a bijection between the
set of actions G × X → X and the set of homomorphisms φ : G →
Sym(X).
(ii) Prove Proposition 13.3
(iii) Let G be a group, R a ring. Prove that there is a bijection between
the set of actions of G on the ring X and the set of homomorphisms
φ : G → Aut(X).
(iv) In the case of n = 2, express the discriminant Ω2 via functions φ1 =
X1 + X2 and φ2 = X1 X2 . How does it help to solve quadratic equa-
tions?
(v) Let G = Sym(X), Y ⊆ X. Show that the subset GY of G defined by
GY = {g ∈ G | g(y) ∈ Y and g−1 (y) ∈ Y for all y ∈ Y } is a subgroup
of G. If X is finite, what is the order of GY as a function of |Y | and
|X|?
13.5 Vista: Grassmann ring and determinants
The sign function is used to define the determinant. In the opposite
direction each σ ∈ Sn defines permutation matrix Aσ and sign(σ) = det(Aσ ).
There is an alternative way of building two of them together at the expense
of using a slightly more exotic ring.
Let K be a commutative ring. The Grassmann ring S = ΛK < X1 , . . . Xn >
as an additive group is a subgroup of the additive group of the polynomial
ring R = K[X1 , . . . Xn ]. It consists of all polynomials where each Xi enters
with degree at most 1. In particular, if K is a field, S is 2n -dimensional
vector space over K.
The multiplication in S is different: it is K-bilinear but on monomials one
uses rules Xi Xj = −Xj Xi and Xi2 = 0. The linear group G = GLn (K) acts

61
 P
on the ring S: (aij )·Xk = i aik Xi extends to a unique ring automorphism.
Similarly, Sn acts via σ · Xj = Xσ(j) . Now, the key is the top degree
element Ω = X1 X2 . . . Xn . It gives both the sign and the determinant by
(aij ) · Ω = det(aij )Ω and σ · Ω = sign(σ)Ω.
The Grassmann ring is not commutative in general but always supercom-
mutative. Find out more about superalgebras, supergroups and supervector
spaces and how they are used in Physics.

14 Orbits
We introduce the quotient G-set G/H and prove the orbit-stabiliser the-
orem. We look at Riemann sphere and Hopf fibration.
14.1 Homomorphisms of G-sets
Let X and Y be G-sets. A function φ : X → Y is a homomorphism of
G-sets if φ(g · x) = g · φ(x) for all g ∈ G, x ∈ X. A bijective homomorphism
will be referred to as an isomorphism. The G-sets X and Y are called
isomorphic if there exists an isomorphism between them.
Examples. 1. For any G-set X and any x ∈ X, the orbit map βx : G → X
defined by βx (g) = g · x is a homomorphism from the regular G-set to X.
Indeed, βx (h · g) = βx (hg) = hg · x = h · (g · x) = h · βx (g).
2. Let X = G be the antiregular G-set. The orbit map β1 is the inverse
map: β1 (g) = g · 1 = 1g−1 = g−1 . Thus, regular and antiregular G-sets are
isomorphic.
14.2 Orbits
Let · be an action of G act on X. We define a relation ∼ on X by x ∼ y
if and only if there exists a g ∈ G with y = g · x. Then ∼ is an equivalence
relation – the proof is left as an exercise.
Definition. The equivalence classes of ∼ are called the orbits of G on X.
In particular, the orbit of a specific element x ∈ X, which is denoted by
G · x or by OrbG (x) is

{ y ∈ X | ∃g ∈ G with g · x = y }.

Observe that the orbit OrbG (x) is the image of the orbit map βx .
Similarly, to the general equivalence relations it is instructive to consider
the quotient set X/ ∼, i.e. the set of orbits. As the equivalence relation is
carried out by G, we denote this quotient set X/G. If there is a single orbit,
the action (as well the G-set) is called transitive.

62
Example. 3. Let X = G be the antiregular G-set, H a subgroup of G. We
restrict the antiregular action: G is an H-set under the antiregular action:
h · g = gh−1 . The orbits of this action are left cosets [g]H = gH. T he
quotient set G/H is the set of all left cosets. In particular, if H is normal in
G, the quotient set admits a group structure which we called the quotient
group. If H is not necessarily normal, the quotient set G/H still carries
an action of G (or a G-set structure): g · [a]H = [ga]H . The axioms of the
action are apparent but one needs to verify that this action is well defined:
[a]H = [b]H =⇒ ∃h ∈ H a = bh =⇒ ga = gbh =⇒ g · [a]H = [ga]H = [gb]H =
g · [b]H . See also exercise (iii).
4. Smith’s normal form in Linear Algebra deals with the action of G =
GLn (K) × GLm (K) on the set X = K n×m of n × m matrices: (g, h) · x =
gxh−1 . Observe that x ∈ OrbG (y) if and only if x and y are equivalent if
and only if they can be moved one to another by a sequence of elementary
row and column transformations if and only if x and y have the same rank.
Recall that the elementary transformation is just a multiplication by an
elementary matrix, who together generate the general linear group. Thus,
|X/G| = min(n, m) + 1 while the Smith’s normal form of x is a particularly
nice element in OrbG (x).
5. Jordan normal form story is slightly more complicated: G = GLn (C)
acts on X = Mn (C) via g · x = gxg−1 . Similarly, x ∈ OrbG (y) if and only if
x and y are similar if and only if they have the same Jordan form. The set
X/G is infinite but its precise structure can be pinpointed.
6. Classification of quadratic forms (over R) involves two groups G =
GLn (R) ≥ H = On (R) acting on the same set X of real symmetric n × n
matrices: g · x = gxg T . A G-orbit is determined by the rank and the
signature of the form and admits a diagonal representative with P 0, ±1 on the
diagonal. Since there are r+1 distinct forms of rank r, |X/G| = nr=0 r+1 =
(n + 1)(n + 2)/2. Each G-orbit is a union of H-orbits. The latter are
determined by the eigenvalues and also admits a diagonal representative
with the eigenvalues on the diagonal.
14.3 Orbit-Stabiliser Theorem
The next theorem is fundamental in group theory: it is an analogue of
the isomorphism theorem for G-sets.

Theorem 14.1 (The Orbit-Stabiliser Theorem) Let a group G act on X,

x ∈ X, Gx the stabiliser. Then the orbit map βx defines an isomorphism of
G-sets between G/Gx and OrbG (x). In particular, |OrbG (x)| = |G : Gx |.

63
Proof: We consider a function ψ : G/Gx → OrbG (x) defined by ψ([g]) =
βx (g) = g · x. Let us observe that this is well defined: [g] = [h] =⇒ ∃a ∈
Gx ga = h =⇒ ψ([h]) = h · x = ga · x = g · (a · x) = g · x = ψ([g]).
It is a homomorphism of G-sets: ψ(g · [h]) = ψ([gh]) = gh · xg · (h · x) =
g · ψ([h]).
For any y ∈ OrbG (x) there exists a g ∈ G with g · x = y. Hence
ψ([g]) = y. For an element g ′ ∈ G, we have

g′ · x = y ⇐⇒ g ′ · x = g · x ⇐⇒ g−1 g′ · x = x ⇐⇒ g −1 g′ ∈ Gx ⇐⇒ g ′ ∈ gGx .

So the elements g′ with ψ(g′ ) = y are precisely the elements of the coset
gGx . Hence, ψ is a bijection. 2
Examples. 7. In Example 2 of Section 13.1, D12 acts transitively on sets
P , E and D of vertices, edges and diagonals of the regular hexagon. Let
us see what the orbit-stabiliser does for the vertices. Let us pick a ver-
tex v0 . Its stabiliser is H = {1, S0 = b}. The quotient set is D12 /H =
{[1]H , [a]H , [a2 ]H , [a3 ]H , [a4 ]H , [a5 ]H } with [ak ]H = {ak , ak b}. Now the orbit-
stabiliser theorem gives a bijection ψ : D12 /H → P given explicitly by
ψ([ak ]H ) = ak · v0 = vk .
If one chooses another point the picture changes slightly. For instance,
the stabiliser of v2 is A = {1, S2 = a2 b}. The quotient set is D12 /A =
{[1]A , [a]A , [a2 ]A , [a3 ]A , [a4 ]A , [a5 ]A } with [ak ]A = {ak , ak+2 b}. The orbit-
stabiliser theorem bijection ψ ′ : D12 /A → P changes to ψ ′ ([ak ]A ) = ak · v2 =
v2+k .
8. Let F be a field. The projective n-space X = P F n consists lines F a in
F n+1 . The group G = GLn+1 (F ) acts on X transitively by A · F a = F (Aa).
The action has a kernel Z
G that consists of scalar matrices. The quotient
group P GLn+1 (K) = G/Z, called projective linear group acts faithfully and
transitively. The stabiliser of F e1 in G is the group of triangular matrices
Tn+1 (F ). Thus, we have a bijection between P GLn+1 (K)/Tn+1 (F ) and X.
14.4 Exercises
(i) Consider the identity map I : G → G, I(g) = g. Show that I is
a homomorphism of G-sets from the regular G-set to the antiregular
G-set if and only if G is a group of exponent 2.
(ii) Prove that ∼, defined in Section 14.2, is an equivalence relation.
(iii) Let X be a G × H-set. Prove that X/H is a G-set with the action
g · [x] = [g · x] for all g ∈ G, [x] ∈ X/H. How does it apply to the
construction of the quotient set G/H.
(iv) Prove that the stabiliser of [x]H ∈ G/H is xHx−1 .

64
 (v) Let G be a simple group (i.e. it has exactly two normal subgroups), H
its proper subgroup of finite index. Prove that G is finite (hint: what
is the kernel of the action on G/H).
(vi) Derive a precise estimate in (v): if |G : H| = n then |G| ≤ n!/2.
(vii) Prove that the quotient G-sets G/A and G/B are isomorphic if and
only if there exists x ∈ G such that xAx−1 = B.
(viii) Considering action on P F13 , prove that P GL2 (F3 ) is isomorphic to
S4 .
14.5 Vista: Riemann Sphere and Mobius Group
In 1872 Klein proposed Erlangen Program where he suggested to classify
geometries according to their groups of symmetries. Let us take a look how
it works on Riemann Sphere.
The Riemann sphere is the projective 1-space over complex numbers
S = P C1 . Geometrically, it is a 2-sphere. It is customary to represent
2

S 2 = PC1as a union C ∪ {∞} where a complex

  number z represents the
z 1
line C while ∞ represents the line C . In this representation, the
1 0
natural action of GL2 (C) is given by Mobius transformations
 
α11 α12 α11 z + α12
·z = .
α21 α22 α21 z + α22

The action
  of GL2 (C) is not faithful: the kernel consists of scalar matrices
α 0
. The group P GL2 (C) acting faithfully is called Mobius group. The
0 α
transformations z 7→ A·z for A ∈ GL2 (C) are called Mobius transformations.
Mobius transformations are conformal, i.e. they preserve angles between
curves. In fact, P GL2 (C) is the group of all conformal transformations of
P C1 and P GL2 (C) underlines conformal geometry of P C1 .
Recall that a matrix A = (aij ) is unitary if and only if A−1 = A∗ where
A∗ = (a∗ji ) is the conjugate matrix. Unitary matrices form the unitary group
Un (C), while unitary matrices with determinant 1 form the  special∗ unitary

α −β
group SUn (C). The group SU2 (C) consists of matrices with
β α∗
|α|2 + |β|2 = 1. Hence geometrically, SU2 (C) is the unit 3-sphere S 3 .
The Mobius transformations from SU2 (C) preserve spherical distances.
This corresponds to the metric geometry of P C1 in Klein program. The
orbit map βx : SU2 (C) → P C1 is a Hopf fibration βx : S 3 → S 2 with all
βx−1 (y) being unit spheres.

65
15 Fixed points
We introduce fixed points. We prove three counting formulae and discuss
their applications to combinatorics.
15.1 Fixed points
Definition. Let T ⊆ G be a subset of G. The fixed points (or the fixed
point set) is defined as X T = {x ∈ X|∀g ∈ T g · x = x}. In particular, we
are interested in X g = X {g} for g ∈ G and X G .
Notice that in the above example X g = ∅ unless g = 1, in which case
1
X = X. Such actions are called fixed points free or simply free.
15.2 Formulae
We would like to establish three useful formulae underlining combina-
torics of the group action. The first is an immediate consequence of Theo-
rem 14.1

Proposition 15.1 (The Orbit-Stabiliser Formula) Let G be a finite group

acting on a finite set X. The for any x ∈ X

|G| = |OrbG (x)||StabG (x)|.

Proposition 15.2 (The Counting Formula) Let G be a finite group acting

on a finite set X. Then
X
|X| = |X G | + |G|/|StabG (x)|
x

where the sum is taken over the representatives of all orbits containing more
than 1 element.
Proof: X is a disjoint union of orbits. P One element orbits G
P form X . The
number of elements in the larger orbits is x |OrbG (x)| = x |G|/|StabG (x)|
using Proposition 15.1. 2

Theorem 15.3 (The Burnside Formula) Let G be a finite group acting on

a finite set X. Then
1 X g
|X/G| = |X |.
|G|
g∈G
Proof: Let A = {(g, x) ∈ G × X|g · x = x}. The formula is obtained by
counting the size of A in two different ways. On one hand,
X
|A| = |X g |.
g∈G

66
On the other hand,
X X X X
|A| = |StabG (x)| = |G|/|OrbG (x)| = |G| 1/|OrbG (x)| =
x∈X x∈X orbits x∈ an orbit
X
= |G| 1 = |G||X/G|.
orbits
2
15.3 Necklaces and bracelets
The formulae allow us to compute necklaces and bracelets. We want to
make a necklace of k beads. We have n different types of beads with at least
k beads of each type. How many necklaces can we make?
Mathematically, we consider the set Pk of vertices of regular k-gon. The
dihedral group D2k acts on Pk . Now we consider the set X = Xk,n of all
untied necklaces, i.e. functions F : Pk → Zn . The dihedral group D2k and
its rotation subgroup Ck act on Xk,n : (g · F )(v) = F (g −1 · v) where g ∈ D2k ,
F ∈ X, v ∈ Pk . The cardinality of Xk,n is nk .
Definition. A necklace is a Ck -orbit in Xk,n . A bracelet is a D2k -orbit in
Xk,n .
Let us count bracelets and necklaces for k = 5. With necklaces we count
the number of C5 -orbits. Observe that 1 fixes every function f ∈ X while a
non-trivial rotation fixes only the constant functions. Hence, by Burnside’s
formula |C/C5 | = (n5 + n + n + n + n)/5 = (n5 + 4n)/5.
With bracelets we count the number of D10 -orbits. A reflection R fixes
functions that are constant on the orbits of the reflection, i.e. f (Rv) = f (v)
for each vertex v. A reflection has 3 orbits, there are n3 such functions.
Hence, by Burnside’s formula |C/D10 | = (n5 + 4n + 5n3 )/10.

Proposition 15.4 The number of necklaces in Xk,n is

1X
N (k, n) = ϕ(d)nk/t
k
t|k

and the number of bracelets is

 1 1 t
2 N (k, n) + 2 n if k = 2t − 1 is odd
B(k, n) = 1 1 t 1 t+1
2 N (k, n) + 4 n + 4 n if k = 2t is even
Proof: Pick g = am ∈ Ck . If d is the greatest common divisor of m and
k, we can write m = dm′ with k and m′ coprime. It follows that |g| = k/d

67
 g
and when g acts on Pk it has d orbits each of size k/d. Thus, |Xk,n | = nd .
Burnside’s formula gives
1 X k/|g|
N (k, n) = n
k
g∈G

′
Since g = (ad )m , it follows that < g >⊇< ad >. Since | < g > | = k/d =
| < ad > |, it follows that < g >=< ad > and there is a single subgroup
< ad > of order t = k/d. It has ϕ(t) elements of order t. Hence, Ck contains
ϕ(t) elements of order t for each t|k and the formula for N (k, n) follows.
If k = 2t − 1 is odd then each reflection fixes one vertex and has further
g
t − 1 orbits of two vertices each. Hence, |Xk,n | = nt for each of k reflections.
The formula for B(2t − 1, n) follows.
If k = 2t is even then a reflection through a vertex fixes two vertices
g
and has further t − 1 orbits of two vertices each. Hence, |Xk,n | = nt+1 for
t reflections. A reflection through the middle of an edge has t orbits of
g
two vertices each. Hence, |Xk,n | = nt for the remaining t reflections. The
formula for B(2t, n) follows. 2
If the supply of a certain type of beads is limited or any other restrictions
is imposed, one can still do the count but one has to deal with a subset of
Xk,n . Some of these situations are covered in exercises and the vista section.
15.4 Exercises
(i) Check that the alternative formula g · F (v) = F (g · v) with g ∈ D2k ,
F ∈ X, v ∈ Pk does not define an action.
(ii) Let a group G act on a ring R. Prove that the set of fixed points RG
is a subring.
(iii) Write an explicit formula derived from Proposition 15.4 for the num-
bers of bracelets and necklaces if k = p2 (p is prime), n is arbitrary.
(iv) Count the number of necklaces and bracelets one can make from 4
identical white and 3 identical black beads. (All the beads are used.)
(v) Count the number of necklaces and bracelets one can make from 4
identical white and 4 identical black beads. (All the beads are used.)
15.5 Vista: aperiodic necklaces
You must have seen the exponential identity

α2 z 2 α3 z 3 αz n
1 + αz + + + . . . = lim (1 + )
2! 2! n→∞ n

68
where α ∈ R, z is a variable. Have you seen its close relative, the cyclotomic
identity
Y∞
1
1 + αz + α2 z 2 + α3 z 3 + . . . = ( )A(k,α)
1 − kz
k=1
1 P k/t
where A(k, α) = k t|k µ(d)α and µ is the Mobius function?
We say that a necklace in Xk,n is aperiodic if its stabiliser in Ck is
trivial. The function A(k, n) counts the number of aperiodic necklaces.
This could be fused together into a nice second year essay as there are sev-
eral different proofs (http://en.wikipedia.org/wiki/Cyclotomic identity and
http://www.stat.wisc.edu/˜callan/notes/cyclotomic/cyclo.pdf).

16 Conjugacy classes
We use G-action on G to study group theory..
16.1 Definition
In Examples 3 and 4 of Subsection 13.1, a group G was made to act on
the set of its own elements by the regular g · x = gx and antiregular actions
g · x = xg −1 for g, x ∈ G.
There is a third important action of G on X = G, which is defined by
g · x = gxg −1 for g, x ∈ G.
It is easy to check that conditions (i) and (ii) of the definition hold, so this
does indeed define an action. This action is called conjugation. The orbits
of the action are called the conjugacy classes of G, and elements in the same
conjugacy class are said to be conjugate in G. So g, h ∈ G are conjugate if
and only if there exists f ∈ G with h = f gf −1 . We will write ClG (g) for the
orbit of g; that is the conjugacy class containing g. We have seen already in
Proposition 7.4 that conjugate elements have the same order.
What is StabG (g) for this action? By definition it consists of the elements
f ∈ G for which f · g = g; that is, f gf −1 = g, or equivalently f g = gf .
In other words, it consists of those f that commute with g. It is called the
centraliser of g in G and is written as CG (g). Notice that the fixed point set
of g also consists of all f such that gf g−1 = f , i.e. commute with g. Hence
Gg = CG (g).
Applying the formulae 15.1, 15.2, 15.3 from the last lecture we get:

Proposition 16.1 Let G be a finite group, g ∈ G. The following three

formulae hold, in the second one the summation is taken over representatives
of all conjugacy classes, not in the centre.

69
 (i) |ClG (g)| = |G|/|C
PG (g)|
(ii) |G| = |Z(G)| P+ x |G|/|CG (x)|
1
(iii) |G/G| = |G| g∈G |CG (g)|

The kernel K of the action consists of those f ∈ G that fix and hence
commute with all g ∈ G. This is called the centre of G and is denoted by
Z(G). So we have

Z(G) = {f ∈ G | f g = gf ∀g ∈ G}.

Note that g ∈ Z(G) if and only if ClG (g) = {g}.

It is high time that we worked out some examples!
Example. 1. Let G be an abelian group. Then Z(G) = G, CG (g) = G and
ClG (g) = {g} for all g ∈ G.
2. Q8 has 5 conjugacy classes: {1}, {−1}, {±I}, {±J}, {±K}. Observe
that IJI −1 = −J. Observe also how the formulas work: I has a centraliser
of order 4: < I >, hence the conjugacy class must have size 8/4 = 2.
3. The conjugacy classes in GLn (F ) have been studied in Linear Algebra
and Algebra-I. Two matrices are in the same conjugacy class if and only if
they are similar.
16.2 Conjugacy classes in dihedral groups
Using multiplication table of SO2 (R) from Section 5.2, let us draw its
conjugation table. We tabulate xyx−1 on the intersection of the row con-
taining x and the column containing y:
Rβ Sβ
Rα Rβ S2α+β
Sα R−β S2α−β

The conjugation table for D2n = {ak | 0 ≤ k < n} ∪ {ak b | 0 ≤ k < n}

follows:
al al b
ak a l a2k+l b
ak b a−l a2k−l b
The cases when n is odd and even are different. Suppose first that n is
odd. Then, by (i) and (ii), a and a−k = an−k are conjugate for all k, and
we have the distinct conjugacy classes {ak , an−k } for 1 ≤ k ≤ (n − 1)/2, all
of which contain just two elements. By (iii), we see that b is conjugate to
a2l b for 0 ≤ l < n, and when n is odd, this actually includes all elements

70
al b for 0 ≤ l < n. (For example, ab = a2l b with l = (n + 1)/2.) So the set
{ak b | 0 ≤ k < n} forms a single conjugacy class. Geometrically, this is not
surprising, because these n elements are all reflections that pass through one
vertex and the centre of the polygon P of which G is the group of isometries.
Now suppose that n is even. Then, when k = n/2, we have ak = a−k ,
and so {an/2 } is a conjugacy class of size 1 (and hence an/2 ∈ Z(G)). We
also have the classes {ak , an−k } of size 2 for 1 ≤ k ≤ (n − 2)/2. In this case,
the reflections ak b split up into two conjugacy classes of size n/2, namely
{a2k b | 0 ≤ k < n/2} and {a2k+1 b | 0 ≤ k < n/2}. Geometrically these
correspond to the two different types of reflections: those about lines that
pass through two vertices of the regular 2k-gon P and those about lines that
bisect two edges of P .
16.3 Classification of groups up to order 11
As an application we extend our classification of groups to the order 11.
The only outstanding order is 9.

Proposition 16.2 A group of order pn , p is prime has a non-trivial centre.

Proof: By Formula (1) of Proposition 16.1, sizes of conjugacy classes are
powers of p. By Formula (2) of Proposition 16.1, p must divide Z(G). Hence,
the centre is non-trivial. 2

Proposition 16.3 Let p be a prime number. There are two groups of order
p2 up to an isomorphism: Cp × Cp and Cp2 .
Proof: These two groups are non-isomorphic by Lemma 3.3: Cp2 has an
element of order p2 but Cp × Cp hasn’t.
Let us start by proving that G of order p2 is abelian. By Proposition 16.2,
Z(G) 6= 1. By Lagrange’s Theorem, |Z(G)| is either p, or p2 . In the latter
case G is abelian. Suppose the former case. Pick x ∈ G \ Z(G) and consider
CG (x). It clearly contains Z(G) and x. Thus |CG (x)| is bigger than p, hence
it is p2 . Thus CG (x) = G. It is a contradiction as we conclude x ∈ Z(G).
If G admits an element a of order p2 , G is a cyclic group.
If G has no such element, all non-identity elements have order p. As in
the homework problem, G admits a vector space structure over the field Zp .
Choosing a basis, forces an isomorphism G ∼ = Cp × Cp . 2
16.4 Exercises
(i) Show that the centre of a group is an abelian subgroup.
(ii) Find the centre of D2n .
(iii) Find centraliser of each element of D2n .

71
 (iv) Let p be an odd prime. In the series of the next exercises, we
describe conjugacy classes in GL2 (Zp ). Prove that |GL2 (Zp )| = (p2 −
1)(p2 − p).
(v) Thanks to Jordan normal form, we know that if a matrix A ∈
GL2 (Zp ) has an eigenvalue then it is conjugate to one of the matrices
     
α 0 α 0 α 1
, ,
0 α 0 β 0 α
with α 6= β ∈ Zp . Compute the centraliser of each of these matri-
ces. Using these computation, list corresponding conjugacy classes
and their sizes. Verify that these classes contain (p − 1)p2 (p + 1)/2
elements overall.
(vi) Consider a matrix A ∈ GL2 (Zp ) without eigenvalues in Zp with
characteristic polynomial z 2 − αz + β. Verify that for a sufficiently
general v ∈ Zp , A looks like
 
0 −β
B= ,
1 α
in the basis v, Av. Compute the centraliser of B and the size of the
conjugacy class of A.
(vii) How many conjugacy classes of matrices without eigenvalues are
there16 .
16.5 Vista: Sylow’s Theorems
The Norwegian mathematician Sylow proved a number of theorems about
subgroups of groups of prime power order in 1872. The first Sylow’s theorem
tells us that if a prime power pn divides |G| then G has a subgroup of order
pn .
Of particular interest are the maximal prime powers dividing |G|. If
|G| = pn m with p and m coprime. The subgroups of order pn are called Sy-
low’s p-subgroups. The second Sylow’s theorem states that any two Sylow’s
p-subgroups are conjugate.
This tells us that the set of Sylow’s p-subgroups is a G-set under conju-
gation with a single orbit. The stabiliser of a point is the normaliser of the
Sylow’s subgroup: it contains the group itself. In particular, the number sp
of Sylow’s p-subgroups divide m. The first Sylow’s theorem tells us that sp
is also 1 modulo p.
16
It is equal to the number of quadratic equations z 2 − αz + β = 0 without solutions in
Zp . Every such equation has two distinct solutions in the finite field Fp2 of p2 elements.
This field is unique, so the number of such equations is (|Fp2 | − |Zp |)/2 = (p2 − p)/2.

72
 These theorems are very powerful for dealing with finite groups. For
instance, let G be a group of order 15. In such a group s3 should divide 5
and be 1 modulo 3. Hence s3 = 1 and G contains unique normal subgroup
of order 3. Similarly, s5 should divide 3 and be 1 modulo 5. Hence s5 = 1
and G contains unique normal subgroup of order 5. It follows that G ∼ =
∼
C3 × C5 = C15 .

17 Conjugacy classes in Sn and An

We describe conjugacy classes in symmetric and alternating groups. We
use it to prove that A5 is simple.
17.1 Conjugacy classes in symmetric groups
Let G = Sym(X) and let f, g ∈ G. Let us write g in cyclic notation,
and suppose that one of the cycles of g is (x1 , x2 , . . . , xr ). Then g(x1 ) = x2 ,
and so f g(x1 ) = f (x2 ) and hence f gf −1 (f (x1 )) = f (x2 ). Similarly, we
have f gf −1 (f (xi )) = f (xi+1 ) for 1 ≤ i < r and f gf −1 (f (xr )) = f (x1 ).
Hencef gf −1 has a cycle (f (x1 ), f (x2 ), . . . , f (xr )), and we have:

Proposition 17.1 Given a permutation g in cyclic notation, we obtain the

conjugate f gf −1 of g by replacing each element x ∈ X in the cycles of g by
f (x).
For example, if X = {1, 2, 3, 4, 5, 6, 7}, g = (1, 5)(2, 4, 7, 6) and f =
(1, 3, 5, 7, 2, 4, 6), then f gf −1 = (3, 7)(4, 6, 2, 1).
In general, we say that a permutation has cycle-type 2r2 3r3 . . ., if it has
exactly ri cycles of length i, for i ≥ 2. So, for example,

(1, 15)(2, 4, 6, 8, 7)(5, 9)(3, 11, 12, 13, 10)(14, 15, 16)

has cycle-type 22 31 52 . By Proposition 17.1, conjugate permutations have

the same cycle-type, and conversely, it is easy to see that if g and h have the
same cycle-type, then there is an f ∈ Sym(X) with f gf −1 = h. For exam-
ple, if X = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}, g = (1, 5, 9)(2, 4, 6, 8)(7, 10) and h =
(1, 5)(2, 10, 9)(3, 6, 8, 7), then we can choose f to map 1, 5, 9, 2, 4, 6, 8, 7, 10, 3
to 2, 10, 9, 3, 6, 8, 7, 1, 5, 4, respectively, so f = (1, 2, 3, 4, 6, 8, 7)(5, 10). (f is
not unique; can you find some other possibilities?) Hence we have:

Proposition 17.2 Two permutations of Sym(X) are conjugate in Sym(X)

if and only if they have the same cycle-type.

73
 For example, S3 has three conjugacy classes, corresponding to cycle-types
1, 21 , 31 , and S4 has five conjugacy classes, corresponding to cycle-types 1,
21 , 22 , 31 , 41 .
17.2 Conjugacy classes in subgroups
If H is a subgroup of G, the conjugacy classes of H are obviously subsets
of conjugacy classes of G. We would like to make two observations on their
interaction.

Proposition 17.3 Let G be a finite group, H its subgroup of index 2, x ∈

H. One of the following two mutually exclusive statements holds.
(1) There exists g ∈ G\H such that gx = xg. In this case, ClG (x) = ClH (x).
(2) For all g ∈ G \ H such that gx 6= xg. In this case, ClG (x) is a union of
ClH (x) and ClH (y) for any y ∈ ClG (x) \ ClH (x). Moreover, |ClG (x)|/2 =
|ClH (x)| = |ClH (y)|.
Proof: By Proposition 8.1, H is normal subgroup of G. Hence, ClG (x) ⊆
H.
In the case (1), CH (x) is a proper subgroup of CG (x). Hence, by La-
grange’s theorem |CG (x)| ≥ 2|CH (x)|. Using Proposition 16.1, |ClG (x)| =
|G|/|CG (x)| ≤ 2|H|/2|CH (x)| = |ClH (x)|. Hence, ClG (x) = ClH (x).
In the case (2), CH (x) = CG (x). Using Proposition 16.1, |ClG (x)| =
|G|/|CG (x)| = 2|H|/|CH (x)| = 2|ClH (x)|. Pick g ∈ ClG (x) \ ClH (x). it suf-
fices to observe that |ClG (x)|/2 = |ClH (g)|. But g = axa−1 for some a ∈ G.
Consequently, CH (g) = aCH (x)a−1 . In particular, |CH (g)| = |aCH (x)a−1 |
and the calculation above shows that |ClG (x)| = 2|ClH (g)|. 2
The next lemma is a criterion for normality.

Lemma 17.4 A subgroup H of a group G is normal in G if and only if H

consists of a union of conjugacy classes of G.
Proof: By Proposition 8.2, H  G if and only if ghg−1 ∈ H for all g ∈
G, h ∈ H. But this is just saying that H  G if and only if h ∈ H ⇒
ClG (h) ⊂ H, and the result follows. 2
For example, consider the subgroup {(), (1, 2)(3, 4), (1, 3)(2, 4), (1, 4)(2, 3)}
of S4 . This is the union of the conjugacy classes of S4 containing elements
of cycle-types 1 and 22 , and so it is a normal subgroup. Incidentally, since
this subgroup lies in A4 , it is also normal in A4 , and so A4 is not simple.
17.3 The Simplicity of A5
In Subsection 10.1, we defined a group G to be simple if its only normal
subgroups are {1} and G, and we saw that the only abelian simple groups

74
are the cyclic groups of prime order. There are also infinitely many finite
nonabelian simple groups. These were eventually completely classified into
a number of infinite families, together with 26 examples known as sporadic
groups, that do no belong to an infinite family. The work on this proof went
on for decades, the completion was announced in 1981 but a complete proof
is yet to appear.
One of the infinite families of finite nonabelian simple groups consists of
the alternating groups An for n ≥ 5. The aim of this section will be to prove
that A5 is simple.
The conjugacy classes of An can be described using Proposition 17.3 We
need is for A5 . The classes of S5 correspond do cycle-types 1, 21 , 22 , 31 , 21 31 , 41 , 51 ,
and of these, the permutations of cycle-types 1, 22 , 31 and 51 are even per-
mutations and hence lie in A5 .
There is 1 permutation of cycle-type 1, 15 of type 22 , 20 of type 31 , and
24 of type 51 , making 60 elements in total.
The problem is that these are classes in Sn , and two permutations could
conceivably be conjugate in Sn but not in An , in which case the correspond-
ing class would split up into more than one conjugacy class in An .
In fact, the 15 permutations of cycle-type 22 forms a single class in An .
Using Proposition 17.3, g = (x1 , x2 )(x3 , x4 )(x5 ) commutes with h = (x1 , x2 ).
Similarly, the 20 permutations of cycle-type 31 are all conjugate in An ,
because g = (x1 , x2 , x3 )(x4 )(x5 ) commutes with h = (x4 , x5 ).
However, for the cycle-type 51 , if g = (1, 2, 3, 4, 5) does not commute
with odd permutations. The size of its conjugacy class is 4! = 24, so the size
of its centraliser is 120/24 = 5. It already commutes with 1, g, g2 , g3 , g 4 , so
it cannot commute with anything else.
Alternatively, you can argue that 24 does not divide |A5 | = 60, so the
S5 -conjugacy class must split into two A5 -conjugacy classes.
Summing up, we have:

Lemma 17.5 A5 has 5 conjugacy classes, of sizes 1, 15, 20, 12, 12.

Theorem 17.6 A5 is a simple group.

Proof: By Lemma 17.4, a normal subgroup N of A5 would be a union of
conjugacy classes of A5 . But no combination of the numbers 1, 15, 20, 12,
12 that contains 1 adds up to a divisor of 60 other than 1 or 60, and so the
result follows by Lagrange’s Theorem (7.2). 2

75
17.4 Exercises
(i) Verify that Proposition 17.3 holds for Cn inside D2n . Find precisely
which conjugacy classes in D2n split into two and which one don’t.
(ii) Show that if a permutation f ∈ An contains an independent cycle
of even length then f commutes with an odd permutation.
(iii) Show that if a permutation f ∈ An contains two independent
cycles of the same length then f commutes with an odd permutation.
(iv) Show that if a permutation f ∈ An contains independent cycles
pairwise distinct odd length then f does not commutes with odd per-
mutations.
(v) Count the number of conjugacy classes in Sn and An for 1 ≤ n ≤ 9.
Compute the sizes of the conjugacy classes
17.5 Vista: groups of order 12
This would be too little for a second year essay but you may consider
classifying groups up to order 30 for an essay. As far as order 12 is concerned,
you already know 4 of the five groups: C12 , C6 × C2 , D12 ∼ = D6 × C2 , A4 .
For the fifth group, consider embeddings D6 ≤ SO2 (R) ≤ SO3 (R) and the
surjection ψ : SU2 (C) → SO3 (R) from Vista Section 14.5. The last group
BD12 , called binary dihedral group, is ψ −1 (D6 ). It is different
 from the
−1 0
other four groups because it has a single element of order 2: is
0 −1
the only element of order 2 in SU2 (C).
One can sort them out using Sylow’s theorem. By Sylow’s theorem, s2
is either 1 or 3, while s3 is either 1 or 4. Now Sylow’s 2-subgroup could be
either C4 or K4 . You can do the classification by considering the following
cases:
(s2 = 1, s3 = 1, C4 ) the group is C3 × C4 ∼
= C12 ,
(s2 = 1, s3 = 1, K4 ) the group is C3 × K4 ∼= C6 × C2 ,
(s2 = 1, s3 = 4, C4 ) no group, C3 cannot act nontrivially on C4 ,
(s2 = 1, s3 = 4, K4 ) the group is A4 ,
(s2 = 3, s3 = 1, C4 ) the group is BD12 ,
(s2 = 3, s3 = 1, K4 ) the group is D12 ,
(s2 = 3, s3 = 4) no group, Sylow’s 3-subgroups contain (3−1)·4+1 = 9
elements leaving space only for one Sylow’s 2-subgroup.

18 Domains, divisibility and PID-s

Algebra-2 takes a major turn now. We have finished group theory and
would like to study ring theory for the rest of the module. We start by

76
discussing divisibility in domains. We introduce and motivate principal ideal
domains.
18.1 Domains
Definition. Two non-zero elements a, b in a ring R such that ab = 0R are
called zero divisors. A domain is a non-zero commutative ring without zero
divisors.
There are good reasons not to call the zero ring a domain. Unfortunately,
I could not think of a short convenient definition that automatically excludes
it. Could you?
The following proposition has a straightforward proof but is useful for
producing domains.

Proposition 18.1 (1) A field is a domain.

(2) A subring of a domain is a domain
(3) A polynomial ring over a domain is a domain.

18.2 Divisibility
We are working in a domain R. Let us try to replicate techniques known
to you in Number Theory.
Definition. Let x, y ∈ R we say that x divides y and write x|y if y = xr
for some r ∈ R.
The following lemma is obvious.

Lemma 18.2 The following statements are equivalent for all x, y ∈ R.

(i) x|y.
(ii) y ∈ (x).
(iii) (x) ⊇ (y).

Definition. Let x, y ∈ R. We say that x and y are associate (write x ∼ y)

if both x|y and y|x.

Lemma 18.3 The following statements are equivalent

(i) x ∼ y
(ii) (y) = (x)
(iii) There exists q ∈ R× such that x = qy
Proof: (i =⇒ iii) It is clear if x = 0. Without loss of generality we may
assume that x 6= 0 6= y. There exist r, t ∈ R such that x = ry and y = tx.

77
Then x = ry = r(tx) and (1 − rt)x = 0. Because R is a domain, 1 − rt = 0
and q = r ∈ R× .
The other implications are obvious. 2
In Z divisibility is usual: x ∼ y if and only if x = ±y since Z = ×

{1, −1}. In a general domain, divisibility properties are invariant under the
equivalence relation of being associate. In other words, if x satisfies a certain
divisibility property then so is any such y that x ∼ y. For instance, it is easy
to observe (and left as an exercise) that any two greatest common divisors
are associate.
Definition. Let x, y ∈ R. The greatest common divisor gcd(x, y) is such
d ∈ R that d|x, d|y, and if z|x and z|y then z|d. The least common multiple
lcm(x, y) is such l ∈ R that x|l, y|l, and if x|z and y|z then l|z.
Uniqueness of lcm(x, y) and gcd(x, y) (up to an associate element) are
established in the exercises. Existence is a bit trickier. In general, they may
not exist.
18.3 PID-s
Recall that an ideal I
R is principal if I = (a) = aR for some a ∈ R
(see Section 11.2).
Definition. A domain R is called a principal ideal domain (abbreviated
PID) if any ideal of R is principal.
Example. 1. Integers Z is PID. Any ideal I
Z is a subgroup of Z+ and
any subgroup of Z is cyclic. Hence, I =< n > as an additive group, implying
that I = (n) as an ideal.
In the next lecture we will give more examples of PID-s but for now we
will use their properties.

Proposition 18.4 If R is PID then lcm(x, y) and gcd(x, y) exist for any
pair of elements x, y ∈ R.
Proof: Pick d, l ∈ R such that (d) = (x) + (y) and (l) = (x) ∩ (y). We
claim that d is the greatest common divisor and l is the least common
multiple. Indeed, (x) ⊆ (d) ⊇ (y) and whenever (x) ⊆ (z) ⊇ (y) it follows
that (z) ⊇ (x) + (y) = (d). Similarly, (x) ⊇ (l) ⊆ (y) and whenever
(x) ⊇ (z) ⊆ (y) it follows that (z) ⊆ (x) ∩ (y) = (l). 2
Note that (x) + (y) = {rx + sy}, hence gcd(x, y) = rx + sy for some
r, s ∈ R as soon as R is a PID.

78
18.4 Prime and irreducible elements
There are two different ways to say what a prime number is. We are
going to see that these lead to two different notions in an arbitrary domain
R.
Definition. Let us consider r ∈ R \ (R× ∪ {0}). We say that r ∈ R is
irreducible if and r = ab implies that a ∈ R× or b ∈ R× . We say that p ∈ R
is prime if p ∈ R \ (R× ∪ {0}) and p|xy implies that p|x or p|y.

Proposition 18.5 A prime element r is irreducible.

Proof: Let p = ab. Then p|a or p|b since p|p = ab. Without loss of
generality, p|a. Hence p ∼ a and p = aq with q ∈ R× . The domain condition
implies that q = b. 2
Notice that we strategically excluded OR out of the consideration. If we
try to include it, it would be prime but not irreducible causing all sorts of
havoc.

Proposition 18.6 If R is a PID, an irreducible element r is prime.

Proof: This proof is quite tricky (at least for me personally) and we have
to use the fact that R is PID twice. Let r be irreducible, r|ab. The element
ã = gcd(r, a) exists by Proposition 18.4. Then r = ãt for some t ∈ R. Since
r is irreducible, ã or t is a unit. We consider both cases.
Let t be unit. Then r ∼ ã and it must divide a.
Now let ã be a unit. Using the description of the greatest common divisor
in a PID, (ã) = (1) = (a) + (r). Hence, 1 = xa + yr for some x, y ∈ R.
Finally, r divides xab = (1 − yr)b = b − yrb. Hence, b = (b − yrb) + ybr ∈ (r).
2
√ √
Example. 2. Let √ R = √Z[i 5] = {a + ib 5 | a, b ∈ Z}. In this ring,
6 = 2 · 3 = (1 + i 5)(1 − i 5). √ We claim that 2 is irreducible
√ but not prime.
√ not divide 1 ± i 5 because 2x = 1 ± i 5 implies that x =
2 does
1/2 ± i 5/2, which is not an element of R. Hence, 2 is not prime. √
Let
√ us show that 2 is2 irreducible. If 2 = ab with a = x + yi 5, b =
s + ti 5 ∈ R then 4 = |a| |b| = (x +√5y 2 )(s2 + 5t2 ). Clearly, |a|2 , |b|2 ∈ N.
2 2

If |a|2 = 1 then a−1 = a∗ /|a|2 = x−yi 5 ∈ R and a is a unit in R. Similarly,

if |a|2 = 4 then |b|2 = 1 and b is a unit in R. Finally, |a|2 = 2 leads to a
contradiction. Indeed, |a|2 = x2 + 5y 2 = 2 leads to (x + (5))2 = 2 ∈ Z5
which is impossible.
√
Corollary 18.7 Z[i 5] is not PID.

79
18.5 Exercises
(i) Prove Lemma 18.2.
(ii) Let d and d′ be both the greatest common divisor gcd(x, y). Prove
that d and d′ are associate.
(iii) Let l and l′ be both the least common multiple lcm(x, y). Prove
that l and l′ are associate.
(iv) Let p be prime. Show that if p|a1 · a2 · · · an then p divides ai .
(v) Prove Proposition 18.1.
(vi) Show that if R is a domain then R[x1 . . . xn ] is a domain.
18.6 Vista: noetherian rings and group rings
A natural generalisation of principal ideal domains are noetherian do-
mains. A ring R (not necessarily commutative) is noetherian if every ideal
is finitely generated. Hilbert’s basis theorem states that R[X] is noetherian
whenever R is noetherian. The quotient ring of a noetherian is noetherian
(but not a subring, in general). This gives a plenty of examples of noetherian
rings.
Here is another chance for you to become a famous algebraist next week-
end. All you need to do is to figure out when the group ring is noetherian.
Let G be a group, F a field. The group ring FG is a vector space with basis
Eg , g ∈ G. The multiplication is F-bilinear with Eg · Eh = Egh on the basis
elements. When is FG noetherian?
A group G is called polycyclic if it admits a finite chain of subgroups
Gk ≤ G, k = 0, . . . n such that Gn = G, G0 = {1}, Gk
Gk+1 and Gk+1 /Gk
is cyclic for all k. A group G is called virtually polycyclic if it has a polycyclic
subgroup of finite index. It is known (the level of hard exercise) that if
G is virtually polycyclic then FG is noetherian. It is one of the biggest
conjectures in ring theory that the reverse statement holds: is it true that
if FG is noetherian then G is virtually polycyclic.

19 Euclidean domains
First, we discuss what prime elements do on the level of quotient rings.
Then we introduce Euclidean domains and give new examples of PID-s.
19.1 Quotient rings and primes
Proposition 19.1 Let R be a domain. A nonzero element p ∈ R is prime
if and only if R/(p) is a domain.
Proof: We write [x] for the coset x + (p). Now [x] 6= 0 translates into p 6 |x.
Thus, [x][y] = 0 =⇒ [x] = 0 ∨ [y] = 0 translates into the prime element

80
definition p | xy =⇒ p | x ∨ p | y. 2
The following proposition tells us a bit more.

Proposition 19.2 Let R be a PID. If p ∈ R is prime then R/(p) is a field.

Proof: By Proposition 19.1, R/(p) is a domain. We need to find an inverse
for a nonzero element [x] = x + (p) ∈ R/(p). Since [x] 6= 0, p does not divide
x. Since p is irreducible (Proposition 18.5) and gcd(x, p) | p, gcd(x, p) = 1.
Since (x) + (p) = (gcd(x, p)), there exist a, b ∈ R such that 1 = ax + bp.
Hence, [x]−1 = [a] in the quotient ring R/(p). 2
Proposition 19.2 and Theorem 19.3 tell us how to construct new fields:
if F is a field and f = f (X) ∈ F [X] is irreducible then F [X]/(f ) is a field.
19.2 Euclidean domains
Definition. A euclidean domain (ED) is a domain R that admits a norm
function ν : R \ {0} → N such that
(i) ν(ab) ≥ ν(b) for all a, b ∈ R,
(ii) ν(ab) = ν(b) if and only if a ∈ R× ,
(iii) ∀a, b ∃q, r such that a = qb + r and either r = 0 or ν(b) > ν(r).

Examples. 1. Integers Z form a euclidean domain. The norm function

is an absolute value, that is, ν(x) = |x|. The elements q and r in the last
property come from the division with a remainder. Notice that already in
this case the elements q and r are not unique. Let a = 13, b = 5. Both
13 = 5 · 2 + 3 and 13 = 5 · 3 − 2 are acceptable.
2. The ring K[X] of polynomials in one variable over a field K is a
euclidean domain. The norm function is the degree of a polynomial. Notice
that K[X]× consists of nonzero constant polynomials that ensures the second
part of the definition. The elements q and r in the last property come from
the polynomial division with a remainder.
It is essential toPmaster the polynomial division with remainder. Dividing
n−1
byP f (X) = X n + k=0 ak X k , can be done using a rewriting rule X n ;
n−1
− k=0 ak X k . For instance, let us divide X 7 + 1 by X 3 − X + 1. The
rewriting rule X 3 ; X − 1. We apply it to the top degree term and write
what goes into the result on top:

X4 X2
X 7 +1 ; X 4 (X−1)+1 = X 5 −X 4 +1 ; X 2 (X−1)−X 4 +1 = −X 4 +X 3 −X 2 +1
−X 1
; −X(X − 1) + X 3 − X 2 + 1 = X 3 − 2X 2 + X + 1 ; (X − 1) − 2X 2 + X + 1
Hence, X 7 + 1 = (X 4 + X 2 − X + 1)(X 3 − X + 1) + (−2X 2 + 2X).

81
19.3 ED and PID
Theorem 19.3 A euclidean domain is a principal ideal domain.
Proof: Let I be an ideal in an euclidean domain R. Choose b ∈ I \ {0}
with the smallest possible norm. Obviously, (b) ⊆ I. Let us now prove the
opposite inclusion. For an arbitrary a ∈ I we can write a = bq + r with
either r = 0 or ν(b) > ν(r). If r 6= 0 then r = a − bq ∈ I and has a smaller
norm than b. This contradiction
P proves that a = bq ∈ (b). 2
Rings Z[α] = { k ak αk | ak ∈ Z} for α ∈ C are domains but other
properties are harder to predict. We will explain the following examples
later in the course. For now you should take my word on them.
√
Examples. 3. The domain √ Z[(1 + −19)/2] is PID but not ED.
4. The domain Z[ −5] is not PID as shown in Section 18.4. Hence, it
is not ED either.
5. Gaussian integers Z[i] = {a + bi ∈ C|a, b ∈ Z form a subring of
C. Hence, it is a domain. It is euclidean with the norm function ν(x) =
|x|2 . The first property is clear. The second property follows from the fact
that Z[i]∗ = {1, −1, i, −i}, which follows from q −1 = q ∗ /|q|2 where q ∗ =
Re(q) − Im(q)i is the conjugate number of q. For the third property√choose
the Gaussian integer q nearest to a/b. Observe that |q − a/b| ≤ 1/ 2. Let
r = a−qb. As soon as r 6= 0, ν(r) = |a−qb|2 = |q−a/b|2 |b|2 ≤ ν(b)/2 < ν(b).
The last calculation seems to make sense even for r = 0. What is about
our exclusive disjunction “either . . . or”? The answer to this question is that
the function ν is not defined at zero17 .
19.4 Minimal polynomials
Principal ideals have several applications, which you may have seen al-
ready. The idea is always the same. Let us start with the minimal polynomial
of a matrix. Let K be a field, A ∈ Mn (K) P a matrix. It defines
P a ring ho-
momorphism fA : K[X] → Mn (K) by fA ( n αn X ) = n n
n αn A . This
ring homomorphism is sometimes called evaluation homomorphism because
fA (F (X)) = F (A). The homomorphism fA is a linear map from an infi-
nite dimensional vector space to a finite dimensional one. Hence, the kernel
is non-zero. Since K[X] the kernel is an ideal (mA ) for some polynomial
mA ∈ K[X]. Multiplying mA by a scalar does not change the ideal, thus,
without loss of generality, mA is monic (the highest degree term has a co-
efficient 1). This, mA is called the minimal polynomial of A. The kernel of
17
Alternatively, one needs to set ν(0) = −∞ to keep the precious property ν(xy) =
ν(x) + ν(y).

82
fA consists of all polynomials F (X) such that F (A) = 0. Thus, mA is the
monic polynomial of minimal degree such that F (A) = 0.
In a similar way, a complex number α ∈ C a matrix defines an evaluation
ring homomorphism fα : Q[X] → C by fα (F (X)) = F (α). The kernel is
(mα ). If mα = 0 the number α is called transcendental: it does no satisfy
any polynomial with rational coefficients. If mα 6= 0 the number α is called
algebraic. Unique monic mα is called the minimal polynomial of α.
The last application of this sort is the characteristic of a ring. Any ring
R has a natural homomorphism fR : Z → R defined by fR (n) = n1R . The
kernel of this homomorphism is (n) for some n ≥ 0. This number n is the
characteristic of R.
19.5 Exercises
(i) Show that the ring Qp = {x/pn | x ∈ Z} ≤ Q, where p is prime, is
Euclidean domain.
(ii) Divide X 8 + X ∈ Q[X] by X 4 − X − 1 with remainder.
(iii) Divide X 6 − X 5 + 1 ∈ Q[X] by X 3 −√ X 2 + 1 with remainder.
(iv) Find the minimal polynomial of (1 + D)/2 over Q where D is a
square-free integer.
(v) Prove that the characteristic of a domain is always a prime number.
(vi) Describe all rings of characteristic 1.
√
19.6 Vista: −19 and quadratic integers √
You may be left surprised by the lack of details why Z[(1 + −19)/2] is
non-euclidean PID. It is actually easy to see that the usual norm ν(x) = |x|2
fails axiom (iii) of Euclidean domain. It is slightly harder to see a certain
weaker version of axiom (iii). This weaker axiom still ensures that an ideal
I is generated by√a smallest non-zero element. It is somewhat trickier to
show that Z[(1 + −19)/2] has no other euclidean norm18 .
√ It is more interesting to try to understand what is so special √ about
−19. Let D 6= 1 be a square-free integer. The quadratic field Q[ D] has
√ OD of all algebraic√integers. Observe that if D ≡4 1 then
a natural subring
OD = Z[(1 + D)/2], and OD = Z[ D], otherwise. Which of these rings
are ED or PID? The following statements summarize what is known about
imaginary (i.e. D < 0) quadratic integers:
(i) ν(x) = |x|2 is Euclidean norm on OD if and only if
D ∈ {−1, −2, −3, −7, −11},
(ii) OD is PID if and only if D ∈ {−1, −2, −3, −7, −11, −19, −43, −67, −163},
18
O. A. Campoli, The American Mathematical Monthly, Vol. 95 (9), 1988, pp. 868–871
contains full details and elementary treatment

83
 (iii) if D ∈ {−19, −43, −67, −163} then OD is not PID.
Less is known about real (D > 1) quadratic integers:
(i) u(x) = |x|2 is Euclidean norm on OD if and only if
D ∈ {2, 3, 5, 6, 7, 11, 13, 17, 19, 21, 29, 33, 37, 41, 57, 73},
(ii) for D < 100, OD is PID if and only if D ∈ {2, 3, 5, 6, 7, 11, 13, 14, 17,
19, 21, 22, 23, 29, 31, 33, 37, 38, 41, 43, 46, 47, 53, 57, 59, 61, 62, 67, 69,
71, 73, 77, 83, 86, 89, 93, 94, 97},
(iii) it is conjectured by Gauss (open as of 2010) that there are infinitely
many PID-s among OD ,
(iv) if extended Riemann’s hypothesis holds, then OD is PID implies
that OD is ED.
Thus, O−19 is the “smallest” PID, not ED!
If you want to know more, you should consider taking MA3A6, Alge-
braic Number Theory where you will learn that all OD are all Dedekind
domains and a certain finite group, called the ideal class group of OD , con-
trols whether OD is PID (this group must be trivial)!

20 Unique factorisation domains

We introduce the notion of factorisation (into irreducible elements) and
its uniqueness. We go on to prove a deep and difficult theorem that every
PID admits a unique factorisation.
20.1 Factorisation
Definition. A domain R is FD19 (factorisation domain) if each x ∈ R \
(R× ∪ {0} admits a factorisation x = r1 · r2 · · · rn where ri are irreducible
elements.
An FD R is UFD20 (unique factorisation domain) if for any two fac-
torisations of an element x = r1 · r2 · · · rn = s1 · s2 · · · sm (all ri and si are
irreducible), m = n and there exists σ ∈ Sn such that ri ∼ sσ(i) for all i.

Proposition 20.1 Let R be an FD. Then R is a UFD if and only if every

irreducible element is prime.
Proof: For the only if part we consider an irreducible element x such that
x|ab. Factorising a = r1 · · · rk and b = rk+1 · · · rn , we get a factorisation
ab = r1 · · · rn . On the other hand, ab = xy. Factorising y = s1 · · · st , we get
19
You won’t find this notion in the literature as it follows from being noetherian.
20
Don’t’ confuse with UFO.

84
another factorisation ab = xr · s1 · · · st . By the UFD property, x is associate
to ri for some i. If i ≤ k then x|a. If i > k then x|b.
The if part follows by a standard induction on n, the length of one of
factorisations x = r1 · r2 · · · rn = s1 · s2 · · · sm . If n = 1 then x = r1 is
irreducible and everything follows. If we are done for n − 1, we observe that
rn |s1 · s2 · · · sm . Since rn is prime it divides some si . Hence, r1 = qsi for
some unit q. Now we use the induction assumption on (qr1 ) · r2 · · · rn−1 =
s1 · · · si−1 · si+1 · · · sm . 2
20.2 Principal ideals give unique factorisation
Theorem 20.2 A PID is a UFD.
Proof: Using Propositions 18.6 and 20.1, it suffices to show that R is F D.
We have to factorise an arbitrary x ∈ R \ (R∗ ∪ {0}). If x is irreducible then
we are done. If not we can write x = x1,1 · x1,2 where x1,i are not units.
We are going to repeat this step over and over again. The step n + 1
starts with x = xn,1 · xn,2 · · · xn,k where none of xn,i are units. If xn,i is
irreducible for all i, we have arrived to factorisation of x. We terminate
the process. If not pick all of xn,i which are not irreducible, write them as
a product of two non-units xn,i = xn+1,j · xn+1,i+1 . In this case, we write
x = xn+1,1 · xn+1,2 · · · xn+1,t and continue with the process.
If this process terminates for all x, we are done: R is FD. Now suppose
the process does not terminate for some particular x and we are after some
sort of contradiction. The process goes on forever and produces a set of
decompositions x = xn,1 · xn,2 · · · xn,k , one decomposition for each natural
number n. The latter statement seems to be obvious but there is a set the-
oretic issue: we use recursion, which is some sort of induction, to construct
a set. Why can we do? The answer is because of Recursion Theorem in Set
Theory. Let us not get any further into this now.
The next step requires some abstract thinking. To facilitate it, think of
all this decompositions as a binary tree. The root of the tree is the element
x. The nodes at level n are elements xn,i for all i. If xn,i is irreducible, it does
not have any upward edges. If xn,i = xn+1,j · xn+1,i+1 , it has two upward
edges going to xn+1,j and xn+1,i+1 . Since the process has not terminated,
the tree is infinite. This means there is an infinite path in this tree starting
from the root and going upward. Let yn = xn,i be the element of this infinite
path at level n. In particular, y0 = x. Observe that . . . yn+1 |yn . . . y1 |y0 .
We have done all this hard work to obtain the ascending chain of ideals
. . . (yn+1 ) ⊃ (yn ) . . . (y1 ) ⊃ (y0 ) with all the inclusions proper. The trick is
that their union I = ∪∞ n=1 (yn ) is an ideal. This is true because all of the ideal
conditions could be checked at one particular (yn ) (do the exercises below if

85
you have difficulties with it). Since R is a PID, I = (d) for some particular
d ∈ R. Then d ∈ (yn ) for some n. This implies that I = (d) ⊆ (yn ).
Consequently, I = (d) = (yn ) = (yn+1 ) = . . . = (yn+i ) that contradicts all
the ideal inclusions being proper. 2
Examples. √ 1. All our ED-s Z, Z[i], F [X] are UFD-s.
2. Z[i 5] is FD√ but not√UFD. We have seen that it is not UFD since
6 = 2 · 3 = (1 + i 5)(1 − i 5) are two distinct factorisations. We won’t
prove in this course that this ring is FD.
3 Z[X] is UFD, which will be proved later, but not PID: (2, X) is not
principal.
20.3 Exercises
(i) A ring is noetherian if every ideal is finitely generated. Prove that
R is noetherian if and only if every ascending chain of ideals in R
terminates.
(ii) Prove that every noetherian domain is FD.
(iii) Prove that (2, X)
Z[X] is not principal
20.4 Vista: the birth of ring theory
The ring theory appeared as a result of an accident, Lame’s 1847 mis-
take (see http://www.mathpages.com/home/kmath447.htm ). Let ωp =
exp(2πi/p) wherePp > 2 is a prime number. Lame has essentially proved
that if Z[ωp ] = { p−1 i
i=0 ai ω ∈ C | ai ∈ Z} is a UFD then the Fermat Last
Theorem holds for p, i.e. the equation xp + y p = z p have no nontrivial inte-
gral solutions. Lame has not given enough thought to the issue and just used
the UFD property of Z[ωp ]. Kummer has corrected this mistake and given
a criterion in terms of Bernoulli numbers for Z[ωp ] to be UFD. A prime p is
called regular (correspondingly irregular) if Z[ωp ] is UFD (correspondingly
not UFD). Looking at small primes, it appears that all are regular. In fact,
the first irregular prime is 37; then 59, 67, 101, 103, 131, 149 are irregular.
On the other hand, it has been proved that there are infinitely many irreg-
ular primes. It is expected that irregular primes constitute about 39% of all
the primes but it is still an open problem whether there are infinitely many
of them.
The ring O−3 = Z[ω3 ] is called Eisenstein integers. If you were thinking
about Gaussian primes for your second year essay, consider switching to
Eisenstein primes.

86
21 Polynomials over fields
We study F [X] over a field F . Our knowledge of this ring leads to a
number of nontrivial observations about the field F itself.
21.1 Remainder theorem
Proposition 21.1 (Remainder Theorem) Let f = f (X) ∈ F [X]. If f (a) =
0 for some a ∈ F then X − a divides f .
Proof: Divide f (X) by X − a with a remainder:
f (X) = g(X)(X − a) + r.
Notice that r must have degree less than 1, so r ∈ F (a constant polynomial).
Substituting X = a, we arrive at 0 = f (a) = r. 2
Definition. A field F is algebraically closed if for any f (X) ∈ F [X] of degree
at least 1 there exists a ∈ F such that F (a) = 0.
When we want to classify primes in a ring R, we are after a complete
lists of primes, which means that each element on the list is prime and
any other prime is associate to exactly one prime on the list. For example,
X + a ∼ bX + ab and we list just one of them.

Proposition 21.2 If F is an algebraically closed field then {X − a | a ∈ F }

is a complete list of primes in F [X].
Proof: The element X − a is irreducible because any of its divisors must
have degree 1 or 0. If it is 0, the divisor is a unit. If it is 1, the divisor is
associate to X − a.
To show that they are pairwise non-associate, notice that F [X]× = F × .
Hence, X − a is associate only to bX − ab for all b ∈ F × .
Finally, if we have a prime f ∈ F [X] then f has degree at least 1.
Since F is algebraically closed, there is a ∈ F such that f (a) = 0. By
Proposition 21.1, X − a divides f . Hence, f is associate to X − a. 2
21.2 Finite subgroups of fields
Theorem 21.3 Let F be a field. A finite subgroup of F × is cyclic.
Proof: Suppose G ≤ F × is not cyclic of order N . By the classification of
finite abelian groups (from Algebra-1), G is isomorphic to Cm1 ×. . .×Cmn , a
product of cyclic groups of orders n1 |n2 . . . |nm and N = n1 ·n2 · · · nm . Since
(x1 , . . . , xm )n = (xn1 , . . . , xnm ) ∈ Cm1 × . . . × Cmn , we deduce that gn = 1
for any g ∈ G where n = nm < N . This provides F (X) = X n − 1 with N
roots, hence with N > n pairwise non-associate prime divisors X − a for
each a ∈ G. This contradicts the UFD condition for F [X]. 2
The following corollary is immediate.

87
Corollary 21.4 Z×
p is a cyclic group of order p − 1.

21.3 Imaginary units in finite fields

Imaginary unit i ∈ C is a 4-th primitive root of unity, i.e., i4 = 1 while
i2 = −1 6= 1. In a general field F , a primitive n-th root of unity is an
element a ∈ F × of order n. An imaginary unit is a 4th primitive root of
unity.

Proposition 21.5 Zp admits a primitive n-th root of unity if and only if

p ≡n 1.
Proof: By Corollary 21.4, Z× ∼
p = Cp−1 . It admits an element of order n if
and only if n | (p − 1). In particular, if n | (p − 1) and t is a generator of Z×
p
then t(p−1)/n has order n. 2
The following corollary gives as a polynomial prime.

Corollary 21.6 X 2 + 1 is prime in Zp [X] if and only if p ≡4 3.

Proof: If p = 2, then X 2 + 1 = (X + 1)2 is not prime. If p ≡4 1, then X 2 +
1 = (X + i)(X − i) is not prime where i ∈ Zp is an imaginary unit. If p ≡4 3,
then X 2 +1 is prime because any decomposition of X 2 +1 = (aX −b)(cX −d)
gives an imaginary unit b/a. Indeed, (b/a)2 + 1 = (ab/a − b)(cb/a − d) = 0,
so that (b/a)2 = −1 6= 1. 2
21.4 Algebraic closure
The following two theorems will neither be proved, nor examined in this
module because this would lead too far from the material we study. Algebraic
closure of a field F is an algebraically closed field F which contains F as a
subring so that each a ∈ F is algebraic over F .

Theorem 21.7 (Existence and uniqueness of algebraic closure) Every field

admits an algebraic closure. If F and Fe are two algebraic closures of F
then there exists a ring isomorphism ψ : F → Fe such that ψ(x) = x for all
x ∈ F.
In MA3D5 Galois Theory you will learn a standard proof via Zorn’s
lemma. Zorn’s Lemma is equivalent to the axiom of choice. Play with
website http://consequences.emich.edu that deals with various consequences
of the axiom of choice. In particular, Axiom of Choice is form 1, existence
of F is form 69, uniqueness of F is form 233, and Ultrafilter Theorem (a
more high-tech way of proving Theorem 21.7) is form 14.

Theorem 21.8 (The fundamental theorem of algebra) C = C

88
 In other words, complex numbers form an algebraically closed field.
There are numerous proofs using Algebra, Analysis or Topology including
the one done in Foundations along the lines of the original Gauss’ argument.
My two favourite proofs are via Liouville’s Theorem (MA3B8 Complex Anal-
ysis) or Open Mapping Theorem
21.5 Derivatives and square-free elements
If R is a domain, r ∈ R is square-free if r is not a unit and if x2 | r then
r is a unit. I don’t know how to determine whether r ∈ Z is square-free
except as to decompose r into primes and see.
Let D : F [X] → F [X] be an F -linear map defined on the monomials
by D(X n ) = nX n−1 . You must have recognized the usual derivative except
that we do them over any field and do not use limits. We denote D(f ) = f ′
to play on our usual intuition.

Proposition 21.9 If f ∈ F [X] and gcd(f, f ′ ) = 1 then f is square-free.

Proof: Suppose h2 |f , then f = gh2 .
Let us observe the product rule: (ab)′ = ab′ + a′ b. On monomials
(X n X m )′ = (n + m)X n+m−1 = X n (X m )′ + (X n )′ X m . By biliniearity it
holds for any a, b ∈ F [X].
Now f ′ = (gh2 )′ = g(hh)′ + g′ h2 = (2gh′ + g′ h)h. Hence h|gcd(f, f ′ ) = 1
forcing h to be a unit. 2
Amazingly enough, the inverse statement actually fails. Let F = Zp (t)
be the field of rational functions with coefficients in Zp . Consider f (X) =
X p − t. One can use Eisenstein’s criterion (Proposition 24.4 and example 1
afterwards) to show that f (X) is prime and consequently square-free. On
the other hand, f ′ = pX p−1 − 0 = 0, so gcd(f, f ′ )√= f . √
What actually happens is that f (X) = (X − p t)p but the element21 p t
is not in F . However, the following holds true while the proof is left as an
exercise.

Corollary 21.10 Let f ∈ F [X] be a polynomial of degree n. Then gcd(f, f ′ ) =

1 if and only if f has n distinct roots in F .

21.6 Exercises
(i) Let F be a field, f = f (X) ∈ F [X] irreducible. Prove that X +(f ) ∈
F [X]/(f ) is a root of the polynomial f (X).
(ii) Improve Theorem 21.3 by showing that any two subgroups in F ×
of order N are equal.
21
The key word is separable here.

89
 (iii) Prove that {X − a, X 2 + bX + c | a, b, c ∈ R, b2 − 4c < 0} is a
complete list of primes in R[X].
(iv) For which p is X 2 + X + 1 prime in Zp [X]?
(v) For which p is X 2 − X + 1 prime in Zp [X]?
(vi) For which p is X 2 + X + 2 ∈ Zp [X] square free?
(vii) For which p is X p + X + 1 ∈ Zp [X] square-free?
(viii) Prove that if f ∈ F [X] is square-free and F is algebraically closed
then gcd(f, f ′ ) = 1.
(ix) Prove Corollary 21.10 .
21.7 Vista: Artin’s conjecture
Let D > 1 be a square-free integer. Let S(D) be the set of primes p such
that |[D]| = p−1 where [D] = D +(p) ∈ Z× p . Emil Artin conjectured in 1927
that this set was infinite. A positive answer follows from Generalised Rie-
mann Hypothesis (which you should never confuse with Extended Riemann
Hypothesis). Find out more in M. Ram Murty, Mathematical Intelligencer
10 (4), 1988, 59-67.

22 Gaussian primes
We classify primes in Z[i] = Z[ω4 ] = O−1 and derive some consequences.
22.1 Preliminary observations
Primes in Z[i] are called gaussian primes. Let us recall that ν(x) = |x|2 .
It is useful to remember that if x|y in Z[i] then ν(x)|ν(y) in Z.

Proposition 22.1 If x ∈ Z[i] and ν(x) is prime then x is gaussian prime.

Proof: Using Proposition 20.1 and the fact that Z[i] is a UFD, it suffices
to check irreducibility of x. Suppose y|x. Hence, ν(y)|ν(x) = p in Z that
forces ν(y) to be p or 1. If ν(y) = p then y is associate to x. If ν(y) = 1
then y is a unit. 2

Proposition 22.2 Let p ∈ Z be a prime. Then either p is gaussian prime

or p = xx∗ where x is a gaussian prime.
Proof: We obtain a proof by turning around the previous proof. If p
is not gaussian prime, there exists a gaussian prime x such that p = xy
and neither x, nor y is a unit. Hence, ν(x)ν(y) = ν(p) = p2 . This forces
ν(x) = ν(y) = p, which makes x and y prime by Proposition 22.1. Finally,
x∗ = x−1 · ν(x) = (y/p) · p = y. 2

90
Proposition 22.3 Let q ∈ Z[i] be a gaussian prime. Then either ν(q) is a
prime or a square of a prime.
Proof: Let n = ν(q) = qq ∗ . Take decomposition of n into primes in Z, say
n = p1 · · · pt . Then q|pj in Z[i] for some j. Thus, n = ν(q) | ν(pj ) = p2j . 2

Proposition 22.4 Z[i]× = {±1, ±i} ∼

= C4 .
22.2 Gaussian primes
Lemma 22.5 For each p ∈ Z we have an isomorphism of rings Z[i]/(p) ∼ =
Zp [X]/(X 2 + 1).
Proof: Since Z[i] ∼
= Z[X]/(X 2 + 1) both rings are quotient rings of of Z[X].
It remains to notice that the kernel of both natural maps Z[X] → Z[i]/(p)
and Z[X] → Zp [X]/(X 2 + 1) is (p, X 2 + 1). Thus, both rings are isomorphic
to Z[X]/(p, X 2 + 1). 2
Using this technical lemma, we are ready to tackle the main theorem.

Theorem 22.6 The prime elements in Z[i] are obtained from the prime
elements Z. Each prime p ∈ Z, congruent 3 modulo 4 is a gaussian prime.
The prime p = 2 gives rise to a gaussian prime q such that 2 ∼ q 2 . Each
prime p ∈ Z congruent 1 modulo 4 gives rise to two nonassociate gaussian
primes q and q ∗ such that p = qq ∗ .
Proof: By Proposition 19.1, a prime p ∈ Z is a gaussian prime if and only
if Z[i]/(p) ∼= Zp [X]/(X 2 + 1) (Lemma 22.5) is a domain. By Corollary 21.6,
this is equivalent to p ≡4 3.
Now using Proposition 22.3, we can distinguish the two cases for a gaus-
sian prime q. In the first case, q is a gaussian prime such that p2 = ν(q)
for a prime p. Hence, q|p in Z[i]. Pick s ∈ Z[i] such that p = qs. Then
|s| = |p|/|q| = 1 and s is a unit (s−1 = s∗ /|s|2 ). Hence q is associate to p
and p is forced to be 3 modulo 4.
In the second case, q is a gaussian prime such that p = ν(q) is a prime.
As x 7→ x∗ is a ring automorphism, q ∗ is also a prime. Thus we observe
two primes q, q ∗ such that p = qq ∗ = ν(q). As p is not gaussian prime, p is
forced to be 1 or 2 modulo 4.
Now q = x + yi ∼ q ∗ = x − yi if and only if |x| = |y| or x = 0 or
y = 0. The latter two cases are impossible and in the former case, we
must have |x| = |y| = 1. Thus, we get 4 associate primes ±1 ± i and
p = ν(1 + i) = 12 + 12 = 2.
If p is 1 modulo 4, we get two groups of associate primes {q = x+yi, −y+
xi, −x − yi, y − xi} and {q ∗ = x − yi, y + xi, −x + yi, −y − xi}. The primes
in the different groups are not associate. 2

91
22.3 Applications
Corollary 22.7 (Fermat) Every prime p congruent 1 modulo 4 is a sum of
two integer squares in a unique way.
Proof: Theorem 22.6 provides existence: p = qq ∗ for a prime q = x + iy,
hence p = qq ∗ = x2 + y 2 . If p = x2 + y 2 = a2 + b2 then p = (x + iy)(x − iy) =
(a + ib)(a − ib) are two prime decompositions in Z[i]. Everything follows
from the UFD property of Z[i]. 2

Corollary 22.8 n ∈ N is a sum of integer squares if and only if n is not

divisible by any prime congruent 3 modulo 4.

Corollary 22.9 There are infinitely many primes congruent 1 modulo 4.

Proof: Suppose there are only finitely many of them in Z, say p1 , . . . , pn .
Let p0 = 2, q0 = 1 + i, pj = qj qj∗ a prime decomposition of pi . Let us
consider a prime decomposition of x = 2p1 · p2 · · · pn + i ∈ Z[i]. No prime
p ∈ Z, congruent 3 modulo 4 divides x because x/p has 1/p as the coefficient
at i, so it is not in Z[i]. Hence, one of the gaussian primes qj (if it is qj∗
swap the notation between qj and qj∗ ) divides x. Hence, pj = ν(qj )|ν(x) =
4p21 · p22 · · · p2n + 1, which is a contradiction since x has residue 1 modulo all
pj . 2
22.4 Exercises
(i) Prove Proposition 22.4.
(ii) Prove Corollary 22.8
(iii) Decompose 20, 30, 91 and 1001 into a product of gaussian primes.
(iv) Compute gcd(8 + 6i, −1 + 3i)
22.5 Vista: primes in arithmetic progressions
You may notice that there are infinitely many primes congruent 3 modulo
4. The proof is straightforward (hint: x = 4p1 · p2 · · · pn − 1) and no gaussian
primes are required. Dirichlet has proved in general that an arithmetic
progression a+nb with coprime a and b contains infinitely many primes. You
can find the original paper online22 , although I doubt that it was Dirichlet
who submitted it. This result is a cornerstone not only in Number Theory
but also in Representation Theory and Harmonic Analysis.

23 Fractions and Gauss lemma

We introduce the field of fractions and prove Gauss’ Lemma.
22
http://arxiv.org/abs/0808.1408

92
23.1 Fields of Fractions
Let R be a domain. We consider the set W = R × (R \ {0}) = {(x, y) ∈
R × R|y 6= 0}. It admits an equivalence relation where (a, b) ∼ (c, d) when-
ever ad = bc. I leave it as an exercise to show that this is, indeed, an
equivalence relation. An equivalence class of (a, b) is called a fraction and
denoted a/b. Let Q = Q(R) be the set of all the equivalence classes on W .

Proposition 23.1 If R is a domain then Q(R) is a field under the opera-

tions
a c ad + bc a c ac
+ = , · = ,
b d bd b d bd
and π : R → Q(R), π(r) = r/1 is an injective ring homomorphism
Proof: We have to show that these operations are well-defined. Then we
have to establish all the axioms of a field. Finally we have to show that π
is an injective ring homomorphism.
To show that the operations are well defined, we need to notice that
the denominators of the results are non-zero because R is a domain. It
remains to prove that the result is independent of the representative of the
equivalence class. Given a/b = x/y and c/d = u/w, we need to show that
ac/bd = xu/yw and ad + bc/bd = xw + yu/yw. The first equality requires
acyw = bdxu that easily follows from ay = bx and cw = du. The second
equality requires
adyw + bcyw = bdxw + bdyu.
Rewriting it, we get

adyw − bdxw = bdyu − bcyw.

This obviously holds because

adyw − bdxw = dw(ay − bx) = 0 and bdyu − bcyw = by(du − cw) = 0.

The list of axioms of the field is long and we have to go and check
them all. But we are in a good shape because we know that the operations
are well-defined, so we can use our usual intuition about fractions. The
associativity of addition is probably the hardest axiom to check
a c e ad + bc e adf + (bcf + bde) a cf + de a c e
( + )+ = + = = + = +( + ).
b d f bd f bdf b df b d f
The commutativity of addition is easier:
a c ad + bc c a
+ = = + .
b d bd d b

93
The zero and the additive inverse are usual: 0 = 0/1 and −(a/b) = (−a)/b
with all the checks routine. The associativity of multiplication is straight-
forward
a c e ac e ace a ce a c e
( · )· = · = = · = ·( · )
b d f bd f bdf b df b d f
as well as the commutativity:
a c ac c a
· = = · .
b d bd d b
The unity and the multiplicative inverse are usual: 1 = 1/1 and (a/b)−1 =
b/a with all the checks routine. It is worth noticing though why a 6= 0.
Indeed, a = 0 if and only if a · 1 = b · 0 if and only a/b = 0/1 = 0. Finally, we
have to check distributivity but it suffices to do it on one side only because
the multiplication is commutative:
a c e ad + bc e ade + bce ade bce a e c e
( + )· = · = = + = · + · .
b d f bd f bdf bdf bdf b f d f

The map is a ring homomorphism because π(1R ) = 1/1 = 1Q ,

xy x y x+y x y
π(xy) = = · = π(x) · π(y), π(x + y) = = + = π(x) + π(y).
1 1 1 1 1 1
Finally, x is in the kernel if and only if x/1 = 0/1 if and only if x · 1 = 0 · 1
if and only if x = 0. 2
Definition. Q = Q(R) is called the field of fractions of a domain R.
Examples. 1. Q(Z) = Q.
2. Q(F [X]) = F (X), the field of rational functions in one variable X.
3. Q(Z[α]) = Q(α)
23.2 Gauss Lemma
We are concerned with the polynomial ring R[X] over a UFD R.
Definition. A polynomial f (X) is called monic if the coefficient of the
highest degree term is 1. It is called primitive if the greatest common divisor
of all the coefficients of f (X) together is 1.

Theorem 23.2 Let R be a UFD with a field of fractions Q = Q(R). If

f = gh ∈ R[X] for some g, h ∈ Q[X] then there exist a, b ∈ Q such that
gb = ag ∈ R[X] ∋ b
h = bh and f = gbb
h.

94
Proof: Let a1 be the least common multiple of all the denominators of the
coefficients of g(X), a2 the greatest common divisor of all the coefficients
of a1 g(X), a = a1 /a2 ∈ Q(R). We define e g = ag ∈ R[X]. Similarly,
e e
h = bh ∈ R[X]. Notice that ge and h are primitive. Consequently,
u e
f= ge
geh and vf = ueh
v
for some u, v ∈ R. Moreover, the greatest common divisor of u and v is 1.
As soon as we prove that v is unit in R we conclude by setting qb = ueg
and bh = v −1 e
h. Let us suppose that it is not a unit. Then there exists a
prime element p ∈ R that divides v. Let us consider a ring homomorphism
X X
π : R[X] → R/(p)[X], π( ak X k ) = (ak + (p))X k .
k k

Since π(v) = 0, we conclude that

g )π(e
0 = π(v)π(f ) = π(u)π(e h).

As R/(p)[X] is a domain, one of the multiplicands must be zero. This is

a contradiction. First, π(u) 6= 0 since p does not divide u because u and
v have no common prime divisors. Second, π(e g ) 6= 0 6= π(e
h) because these
polynomials are primitive. 2
23.3 Exercises
(i) Prove that the relation ∼ on the W is an equivalence relation.
Let R be a commutative ring, S ⊂ R a denominator set, that is, a sub-
set closed under multiplication, containing 1. Repeat the construction
of the ring of fractions starting with the set W = R × S. The resulting
ring QS (R) is called the partial ring of fractions.
(ii) Show that if R is a domain, p ∈ R is prime then S = R \ (p) is a
denominator set.
(iii) Describe QS (Z) where S = Z \ (p) for some prime p.
(iv) Describe QS (Z) where S = {pn | n ∈ Z} for some prime p.
(v) Determine the kernel of the natural homomorphism R → QS (R),
r 7→ r/1. In particular, show that the kernel is the whole ring R if and
only if 0 ∈ S.
23.4 Vista: fractions over ED
A subtle algebraic property of fractions over ED is useful in Analysis,
namely for integration and interpolation. If R is ED then any element of

95
Q(R) can be represented as
X rj
r+ n where r, rj , pj ∈ R, nj ∈ N, pj are pairwise non-associate prime.
j
pj j

Try to prove it yourself but let us see some example:

Examples. 1. 5/6 = 1/ + 1/3.
2. 101/24 = 4 + 5/24 = 4 + 1/3 − 1/8.
R 3. Here is3 a typical computation of the integral of a rational function
(X + 1)/(X + X) dX.

X +1 aX + b c cX 2 + c + aX 2 + bX
= + =
X3 + X X2 + 1 X X3 + X
Hence, b = c = 1, a = −1 and
Z Z Z Z Z
X +1 −X + 1 1 1 dX 2 dX dX
3
dX = ( 2 + )dX = − 2
+ 2
+ =
X +X X +1 X 2 X +1 X +1 X
1
= − ln(X 2 + 1) + arctan(X) + ln(X) + C
2
Does it ring a bell?
4. Let f = (X − a1 ) · · · (X − an ) ∈ C[X] where aj 6= ak for j 6= k and
g ∈ C[X] of degree less than n. Trying to guess the coefficients, we write
n
X tj
g
=
f X − aj
k=1

and
n
X n
X
tj f (X)
g(X) = = tj (X − a1 ) · · · (X − aj−1 ) · (X − aj+1 ) · · · (X − an ).
X − aj
k=1 k=1

Substituting X = aj we get the answer,

′ g(aj )
g(aj ) = tj (aj −a1 ) · · · (aj −aj−1 )·(aj −aj+1 ) · · · (aj −an ) = tj f (aj ) or tj = ′ .
f (aj )
Turning this calculation around gives Lagrange’s interpolation polynomial,
′
that is, given aj and sj we use f = (X − a1 ) · · · (X − an ) and tj = sj /f (aj )
to define the interpolation polynomial
n
X
g(X) = tj (X − a1 ) · · · (X − aj−1 ) · (X − aj+1 ) · · · (X − an )
k=1

96
 n
X (X − a1 ) · · · (X − aj−1 ) · (X − aj+1 ) · · · (X − an )
= sj .
(aj − a1 ) · · · (aj − aj−1 ) · (aj − aj+1 ) · · · (aj − an )
k=1

It will be the polynomial of the smallest possible degree such that g(aj ) =
sj .

24 Polynomials over UFD

We discuss polynomials with coefficients in UFD. We prove Eisenstein’s
criterion and introduce cyclotomic polynomials.
24.1 Corollaries of Gauss’ Lemma
The following proposition is a corollary of Gauss’ lemma.

Proposition 24.1 If R is UFD then there are two kinds of primes in R[X]:
prime elements in R; primitive elements in R[X] that are prime in Q[X].
Moreover, R[X] is UFD.
Proof: It immediately follows from Theorem 23.2 that all elements listed
are irreducible. Let us establish that any f ∈ R[X] can be factorised into
them, We can factorise f = f1 · · · fn in Q[X]. Getting rid of denominators
and common divisors of numerators, we get f = afe1 · · · fen for some a ∈
Q,fej = aj fj primitive in R[X]. Factorising a in R, we arrive at the required
factorisation of f . Thus, every irreducible element of R[X] is associate to
either a prime in R or a primitive element in R[X], prime in Q[X].
Now we proceed to prove that this factorisation is unique. Let us consider
two factorisations

f = p1 · · · pk f1 · · · fn = q1 · · · qt g1 · · · gm ∈ R[X], pj , qj ∈ R, fj , gj 6∈ R

into irreducible elements. Without loss of generality fj , gj are primitive.

Using the UFD property of Q[X], n = m and fj associate to gσ(j) for some
permutation σ. Since R[X] is a domain, we can cancel all associate elements:

αp1 · · · pk = βq1 · · · qt ∈ R

for some units α, β ∈ R× . Using the UFD property of R establishes the

uniqueness of the factorisation.
Finally, by Proposition 20.1, every irreducible is prime. 2
The following two corollaries provide new examples of UFD-s.

Corollary 24.2 If F is a field then F [X1 , . . . Xn ] is UFD.

97
Proof: We proceed by induction on n. If n = 1 then F [X] is ED, hence
PID, hence UFD. If we have proved it for n − 1 we observe that

F [X1 , . . . Xn ] ∼
= F [X1 , . . . Xn−1 ][Xn ]

is also UFD by Proposition 24.1. 2

Corollary 24.3 Z[X1 , . . . Xn ] is UFD.

24.2 Eisenstein’s Criterion

As we have just observes, the irreducibility of polynomials is the same
over R[X] and Q[X]. However, determining whether a particular polynomial
is irreducible is often subtle. The following is a powerful tool for producing
some of examples.

Proposition 24.4 (Eisenstein’s Criterion) Let R be a UFD,

n
X
f (X) = ak X k ∈ R[X].
k=0

We assume that there exists a prime p ∈ R such that p divides all ak for
k < n but does not divide an and p2 does not divide a0 . If the greatest
common divisor of all the coefficients is 1 then f (X) is irreducible in R[X].
Proof: A factorisation with one polynomial of zero degree is impossible
because the coefficients have no common divisors. Suppose
n
X Xm Xt
k k
f (X) = ak X = ( bk X )( ck X k )
k=0 k=0 k=0
P
with both polynomials of non-zero degree. Then ak = r+s=k br cs for all k
Since p|a0 = b0 c0 , it divides either b0 or c0 but not both since p2 does not
divide a0 . Without loss of generality, p divides b0 but not c0 .
This serves as a basis of induction. We prove that p divides bj for each
0 ≤ j ≤ m < n. Suppose we have done for all j < l. Then

bl c0 = al − (bl−1 c1 + bl−2 c2 + . . .).

Since p divides every term in the right hand side, it divides bl c0 . Since it
does not divide c0 , it divides bl .
Hence, p divides an = bm ct which is a contradiction. 2

98
 Notice that if f (X) admits p as in Eisenstein’s criterion but its coef-
ficients are not relatively prime then f (X) is not irreducible in R[X] but
irreducible in Q[X] where Q = Q(R).
Examples. 1. If p is prime in R then X n + p is prime in R[X] for any
n. In particular, f (X) = X p − t is prime in R[X] where R = F [t] (F is a
field) and, consequently, by Gauss’ lemma, prime in Q[X] = F (t)[X]. See
Section 21.5 where it was used.
2. If p is prime in Z then f (X) = X p−1 + X p−2 + . . . X + 1 is prime in
Z[X]. Consider the shift automorphism

sh−1 : Z[X] → Z[X], sh−1 (g) = g(X + 1)

Since it is a ring automorphism, f is prime if and only if sh−1 (f ) is prime.

Now
p−1
X
p p−1 p!
sh−1 (f ) = ((X + 1) − 1)/(X + 1 − 1) = X + X k−1 .
k!(p − k)!
k=1

is irreducible by Eisenstein’s criterion as p (but not p2 ) divides all the lower

coefficients.
24.3 Cyclotomic polynomial
If n is not prime in Z then f (X) = X n−1 + X n−2 + . . . X + 1 is no longer
irreducible in Z[X]. It is a product of cyclotomic polynomials
Y Y
f= Φk , Φk = (X − e2πid/k ).
k6=1,k|n 0<d≤k,gcd(d,k)=1

Proposition 24.5 For each n, Φn (x) is monic with coefficients in Z.

Proof: We proceed by induction on n. If n = 1 then Φ1 (X) = X − 1. Let
us suppose that Φd is monic for d < n. By definition,
Y
Xn − 1 = Φk = h(X)Φn (X) ,
k|n

where h(X) is the product of all Φd (X) over divisors of n other than n
itself. By the inductive hypothesis, h(X) is monic and has coefficients in Z.
So Φn (X) is the result of dividing X n − 1 by h(X).
The process of dividing one polynomial by another would consist of
rewritings X k ; (X k − h(X)) where k is the degree of h(X). Every time
αX m , α ∈ Z is rewritten, αX m−k goes to the result and αX m−k (X k −h(X))

99
appears. Both have integer coefficients, hence the quotient Φn (X) = (X n −
1)/h(X) is monic with integer coefficients. 2
The polynomial Φk is actually irreducible in Z[X] and you can find more
about it in the vista section below. It is an interesting fact that an early
version of a manual for the computer system Maple has stated that all
coefficients of Φk are ±1 and 0. The smallest counterexample is Φ105 (X) =
X 48 + X 47 + X 46 − X 43 − X 42 − 2X 41 − X 40 − X 39 + X 36 + X 35 + X 34 +
X 33 + X 32 + X 31 − X 28 − X 26 − X 24 − X 22 − X 20 + X 17 + X 16 + X 15 +
X 14 + X 13 + X 12 − X 9 − X 8 − 2X 7 − X 6 − X 5 + X 2 + X + 1
24.4 Exercises
(i) Prove Corollary 24.3
(ii) Prove that F [X1 , . . . Xn ] is not PID if n > 1.
(iii) Prove that Z[X1 , . . . Xn ] is not PID.
(iv) Use Eisenstein’s criterion to produce 6 new irreducible polynomials
in Z[X].
(v) Compute the cyclotomic polynomial Φk for all k < 16.
(vi) Pick two of your favourite polynomials in Z[X] and find the great-
est common divisor.
P k
(vii) Let f =P k αk Xk ∈ R[X] be monic, R any domain, I
R. Show
that if f = k [αk ]X ∈ R/I[X] is irreducible then f is irreducible.
24.5 Vista: irreducibility of cyclotomic polynomials
Irreducibility of the cyclotomic polynomial Φn (X) was part of this mod-
ule last year. You can find it in the lecture notes on Mathstuff and can use
it for your essay. The idea is to try to reduce the coefficients modulo some
prime p, coprime to n. If Φn (x) ∈ Zp [X] is irreducible then so is Φn (x)
(Exercise (vii)). Unfortunately, the life is not so simple: Φn (x) ∈ Zp [X]
is irreducible if and only if |p + (n)| = ϕ(n) in the group Z× n (prove it
as a part of your essay - very good advanced exercise). This means that
Z×n must be cyclic for Φn (x) to have a shot at irreducibility. Look at
Z× ∼ × × ∼
15 = Z3 × Z5 = C2 × C4 . Hence, Φ15 (x) ∈ Zp [X] is not irreducible
for any prime p!
The trick is to use several different primes. Look up the details. A
beautiful consequences of irreducibility of Φn is an elementary proof that
there are infinitely many primes which are 1 modulo n.

100
25 Algebras and division rings
We introduce algebras fusing rings and vector spaces. We discuss division
rings and prove little Wedderburn’s theorem.
25.1 Algebras, their homomorphisms and ideals
Definition. An algebra is a pair (R, F) such that F is a field, R is both a
ring and a vector space over F such that these two structures share the same
addition and α(ab) = (αa)b = a(αb) for all α ∈ F, a, b ∈ R.
It is common to say that A is an algebra over F or simply F -algebra.
Many of the rings we introduced are algebras.
Examples. 1. Any field K is an algebra over any subfield F ≤ K.
2. If F is a field, Mn (F) and F[X] are algebras over F.
3. If R is an algebra over K and F ≤ K is a subfield then R is an algebra
over F.
Algebras have analogues of subrings, ideals and homomorphisms. A
subalgebra of (R, F) is a subring S ≤ R, which is also a vector subspace. An
algebra ideal of (R, F) is an ideal I
R, which is also a vector subspace. An
algebra homomorphism from A = (R, F) to B = (S, F) where both algebras
are over the same field F is an F-linear ring homomorphism from R to S
which is also a linear map. Isomorphism theorem holds for algebras.

Proposition 25.1 (Isomorphism theorem for algebras)

(i) If φ : (R, F) → (S, F) is an algebra homomorphism then ker(φ) is
an algebra ideal of (R, F) and im(φ) is a subalgebra of (S, F).
(ii) If I
(R, F) is an algebra ideal then the quotient ring R/I is an
algebra and the quotient map R → R/I is an algebra homomorphism.
(iii) Let φ : (R, F) → (S, F) be an algebra homomorphism with the
kernel I. Then R/I ∼ = im(φ) as algebras.
So far the algebras seem to be pretty much as rings. The following
examples should raise your alarms.
Examples. 4. Complex numbers C is a C-algebra in two different ways:
(C, C) with α · x = αx and (C, C)′ with α · x = α∗ x. They are isomorphic as
algebras but the identity map I(x) = x is not an algebra homomorphism.
The complex conjugation x 7→ x∗ is an algebra isomorphism.
5. Since C is a subring of M2 (R), M2 (R) is a C-vector space:
 
x −y
(x + yi) · A = A
y x
but it is not C-algebra: α(ab) = a(αb) fails.

101
25.2 Algebras and centres
To understand example 5, we need to recall the notion of the centre.
The centre of a ring R is Z(R) = {a ∈ R | ∀x ∈ R xa = ax}. Similarly to
groups, the centraliser of x ∈ R is C(x) = CR (x) = {a ∈ R | xa = ax}.

Proposition 25.2 Let R be a ring, x ∈ R. Then CR (x) and Z(R) are

subrings of R. If R is an F-algebra, then CR (x) and Z(R) are subalgebras.
Proof: Since 0, 1 ∈ C(x), C(x) is not empty. It is an additive subgroup
since ax = xa, bx = xb imply that (a − b)x = ax − bx = xa − xb = x(a − b).
It is a subring since ax = xa, bx = xb imply that (ab)x = a(bx) = a(xb) =
(ax)b = (xa)b = x(ab). If R is an F-algebra, C(x) is a subalgebra since
ax = xa, α ∈ F imply that (αa)x = α(ax) = α(xa) = x(αa).
All the statements hold for the centre since Z(R) = ∩x∈R CR (x). 2
Centres shed some light on algebra structures (cf. Exercise (vi)).

Proposition 25.3 Let R be a ring, F a field. A ring homomorphism φ :

F → R defines an F-algebra structure on R by f ⋆ r = φ(f )r, f ∈ F, r ∈ R.
Proof: The vector space axioms follow from the homomorphism properties
of φ: (f + f˜) ⋆ (r + r̃) = φ(f + f˜)(r + r̃) = φ(f )r + φ(f )r̃ + φ(f˜)r + φ(f˜)r̃ =
f ⋆ r + f ⋆ r̃ + f˜⋆ r + f˜⋆ r̃ and (f f˜) ⋆ r = φ(f f˜) ⋆ r = φ(f )(φ(f˜)r) = f ⋆ (f˜⋆ r).
Finally the first part of the algebra axiom follows from associativity f ⋆(rr̃) =
(φ(f )r)r̃ = (f ⋆ r)r̃ and the second part follows from the fact that the image
lies in the centre: (φ(f )r)r̃ = (αa)b = a(αb). 2
Now we are ready to explain mysterious Example 5: Z(M2 (R)) = R, thus
the homomorphism C → Z(M2 (R)) does not have an image in the centre.
In fact, there are no C-algebra structures on Z(M2 (R)) because there are
no ring homomorphisms C → R (see Exercises (iii), (iv) and (vi)).
Example. 6. A ring D is a division ring if D × = D \ {0}. A division ring is
not necessarily a field because a field must be commutative while a division
ring does not23 . If ax = xa then xa−1 = a−1 axa−1 = a−1 xaa−1 = a−1 x,
thus all CD (x) are division rings too and Z(D) is a field. By Proposition 25.3,
a division ring D is a Z(D)-algebra. In the next lecture we give an explicit
example of a division algebra, which is not a field. But now we will explain
where not to look.
25.3 Finite division rings
Let us make a general observation about finite algebras.
23
Sometimes division rings are called skew fields.

102
Proposition 25.4 Let (R, F) be an algebra such that R is finite but nonzero.
Then |F| = pm for some prime p and positive integer m while |R| = |F|k for
some positive integer k.
Proof: Since ψ : F → R, ψ(f ) = f 1R is a nonzero ring homomorphism, its
kernel must be zero (F is a field). Thus, F is finite.
The characteristic of F (see Section 19.4) must be prime p. Hence Zp is
a subring of F. Natural action a · x = ax, a ∈ Zp , x ∈ F makes F into a
vector space over Zp . Since F is finite, the vector space is finite dimensional,
say of dimension m. Hence, |F| = |Znp | = pn .
Similarly, if k is the dimension of R over F then |R| = |Fk | = |F|k . 2
Now we are ready for little disappointment.

Theorem 25.5 (Little Wedderburn’s Theorem) A finite division ring is a

field.
Proof: Let R be a finite division ring with centre Z, which is a field and R
is a Z-algebra. By Proposition 25.4, |Z| = q = pm and |R| = q k . It suffices
to prove that k = 1.
Consider the action of G = R× on R by conjugation: g · x = gxg−1 . The
fixed points of this action is the centre Z. The stabiliser StabG (x) consists
of all non-zero elements in the centraliser CR (x) which is a Z-subalgebra of
dimension d(x). Thus, the counting formula (Proposition 15.2) gives us that
X X qk − 1
q k = |R| = |Z| + |G|/|StabG (x)| = q +
x x
q d(x) − 1

where the sum is over representatives of orbits with at least 2 elements, so

d(x) < k. Thus, the cyclotomic polynomial Φk (z) divides z k − 1 but not
z d(x) − 1. Let us rewrite the counting formula as
X qk − 1
q − 1 = (q k − 1) − .
x
q d(x) − 1

The integer Φk (q) divides the right hand side, hence, it divides q − 1. We
claim that |Φk (q)| > q − 1 for k > 1. Indeed, if ξ = e2πi/k then the set of
primitive k-th roots of unity is {ξ t | t ∈ Z×k }. Thus,
Y Y
|Φk (q)|2 = |q − ξ t |2 = [(q − Re(ξ t ))2 + Im(ξ t )2 ]
t t

and since |ξ| = 1, each real part is certainly between −1 and +1, so
q − Re(ξ t ) > q − 1 unless Re(ξ t ) = 1, which happens only if ξ t = 1, which
can happens only if k = 1. 2

103
25.4 Exercises
(i) Prove Proposition 25.1.
(ii) Prove that any two F-algebra structures on a field F are isomorphic
as algebras.
(iii) Prove that if R is a commutative ring then Z(Mn (R)) = {αIn |
α ∈ R}.
(iv) Prove that there are no ring homomorphisms C → R. (Hint: where
should i go?)
(v) Prove that if V is a vector space over a field F then the set of all
linear operators EF (V ) is an F-algebra with S Ṫ = ST and αT : v 7→
α(T v) = T (αv). Prove that Z(EF (V )) = {αIV | α ∈ F}.
(vi) Let R be a ring, F a field. Prove that there is a bijection between
the set A = {(R, F) | (R, F) is an algebra } of algebra structures and
the set B of algebra homomorphisms from F to Z(R).
25.5 Vista: finite fields
To understand finite division rings, it remains to describe finite fields.
We already know that a finite field must have pn elements for some prime
power pn . In fact, for each prime power pn there exists a unique (up to an
isomorphism) field Fpn of order pn . Existence follows from Theorem 21.7.
Let Zp be the algebraic closure of Zp . Then Fpn is the subset of Zp that
n
consists of roots of z p −z. You need the freshman’s dream binomial formula
(Section 9.1) to prove that Fpn is a subfield of Zp .
Uniqueness follows from the UFD property of Zp [z]. Let f (z) be any
n
prime factor of Φpn−1 (z) in Zp [z]. If F is a field of order pn then both z p − z
and Φpn −1 (z) split over F into linear factors. Let ξ ∈ F be a primitive root
of Φpn −1 (z). The evaluation homomorphism Zp [z] → F, h(z) 7→ h(ξ) gives
rise to an isomorphism Zp [z]/(f ) ∼ = F, proving that any two subfields are,
indeed isomorphic.

26 Quaternions, algebraic properties

We study the algebraic properties of quaternions. As an application we
describe Hopf fibration.
26.1 Hamilton quaternions
In Section 4.4 introducing the quaternionic group Q8 , we used the fol-
lowing matrices in M2 (C):
       
1 0 i 0 0 1 0 i
1= , I= , J= , K= .
0 1 0 −i −1 0 i 0

104
Proposition 26.1 The R-span of 1, I, J and K is an R-subalgebra of
M2 (C).
Proof: The space is an R-vector subspace containing 1. It suffices to check
that it is closed under multiplication. We have already seen how to multiply
these matrices (Proposition 4.6):

1 I J K
1 1 I J K
I I −1 K −J
J J −K −1 I
K K J −I −1
P P
We are done because the multiplication is bilinear: ( i αi Ei )( j βj Ej ) =
P
i,j αi βj Ei Ej . 2
Definition. The Hamilton24 quaternions H is the R-algebra described in
Proposition 26.1.
Notice that H is not a C-subalgebra of M2 (C). The C-span of these
elements is the whole M2 (C). Moreover, H is not a C-algebra at all because
Z(H) = R and there are no ring homomorphisms C → R (see exercises).
26.2 Real and imaginary quaternions
A quaternion α1 is called real. A quaternion αI + βJ + γK is called
imaginary. Imaginary quaternions form a three-dimensional subspace H0 ,
while real quaternions form a subalgebra R of H. For each quaternion x =
α1 + βI + γJ + δK ∈ H, analogously to complex numbers we define its
conjugate
x∗ = α1 − βI − γJ − δK,
its real part
Re(x) = (x + x∗ )/2 = α1 ∈ R,
its imaginary part

Im(x) = (x − x∗ )/2 = βI + γJ + δK ∈ H0 ,

and its norm

ν(x) = α2 + β 2 + γ 2 + δ2 .
24
He discovered the quaternions walking along a canal in Dublin. He got so excited that
he vandalised the first bridge with the formulas. Brougham Bridge now carries a plaque
saying Here as he walked by on the 16th of October 1843 Sir William Rowan Hamilton in
a flash of genius discovered the fundamental formula for quaternion multiplication i2 =
j 2 = k2 = ijk = −1 & cut it on a stone of this bridge.

105
We treat the space of the imaginary quaternions H0 as the standard 3-space
with the standard dot and crossed products:

(αI + βJ + γK) • (α′ I + β ′ J + γ ′ K) = αα′ + ββ ′ + γγ ′ ,

(αI+βJ+γK)×(α′ I+β ′ J+γ ′ K) = (βγ ′ −β ′ γ)I+(γα′ −γ ′ α)J+(αβ ′ −α′ β)K.

Theorem 26.2 If x = α + a, y = β + b ∈ H with α, β ∈ R and a, b ∈ H0

then xy = (αβ − a • b) + (αb + βa + a × b).
Proof: By R-bilinearity, xy = αβ + αb + βa + ab. It remains to write
a = α1 I + α2 J + α3 K, b = β1 I + β2 J + β3 K and compute the product
ab = −α1 β1 − α2 β2 − α3 β3 + (α2 β3 − α3 β2 )I + (α3 β1 − α1 β3 )J + (α1 β2 −
α2 β1 )K = −a • b + a × b. 2
Let us draw several corollaries.

Corollary 26.3 If x ∈ H then xx∗ = ν(x)1.

Proof: We write x = α + a with α ∈ R, a ∈ H0 . Then x∗ = α − a and
xx∗ = (αα − a • (−a)) + (αa − αa − a × a) = α2 + a • a = ν(x)1, using the
fact that a × a = 0. 2

Corollary 26.4 If x, y ∈ H then (xy)∗ = y ∗ x∗ .

Proof: Again writing x = α + a, y = β + b with α, β ∈ R, a, b ∈ H0 , we
get y ∗ x∗ = (βα + b • a) + (−αb − βa + b × a) = (xy)∗ , using the fact that
b × a = −a × b. 2

Corollary 26.5 If x, y ∈ H then ν(xy) = ν(x)ν(y).

Proof: ν(xy)1 = xy(xy)∗ = xyy ∗ x∗ = x(ν(y)1)x∗ = xx∗ ν(y)1 = ν(x)ν(y)
2
The vector spaces pH and H0 are euclidean spaces. The euclidean norm
of a vector is ||x|| = ν(x). The inequality ||x|| · ||y|| ≥ |x • y| is called
Schwarz’s inequality. It allows us to define the angle between two nonzero
vectors as θ = arccos (x • y/||x|| · ||y||).

Corollary 26.6 If a, b ∈ H0 and θ is the angle between a and b then

||a × b|| = ||a||||b|| sin θ.
Proof: Since ab = −a • b + a × b, a • b is real and a × b is imaginary,
ν(ab) = ν(a • b) + ν(a × b) = ||a||2 ||b||2 (cos θ)2 + ||a × b||2 and ν(ab) =
||a||||b|| = ||a||2 ||b||2 (cos θ)2 + ||a||2 ||b||2 (sin θ)2 proving the statement. 2

106
26.3 Multiplicative group of quaternions
H is not a field because IJ 6= JI but it is a division ring.

Proposition 26.7 H is a division ring.

Proof: It follows from Corollary 26.3 that x−1 = ν(x)−1 x∗ for each x ∈
H \ {0}. 2
×
Let us analyze the multiplicative group H = H\{0}. Since a quaternion
is naturally a matrix:
 
α + βi γ + δi
x = α1 + βI + γJ + δK = ,
−γ + δi α − βi

H× is a subgroup of GL2 (C). It has several interesting subgroups on its own.

Real quaternions form a subgroup R× that has a further subgroup R× + of
positive real quaternions. Another subgroup is U (H) = {x ∈ H | ν(x) = 1}.

Proposition 26.8 H× ∼ = R× + × U (H).

Proof: The multiplication in H defines a homomorphism φ : R× + × U (H) →
H× , i.e. φ(α, q) = αq. It is injective because αq = 1 imply that Im(αq) =
αIm(q) = 0, i.e. q = β is real. Moreover αβ = 1 while α > 0 and |β| = 1,
hence α = q = 1.
It is surjective because q = ν(q)ν(q)−1 q = φ(ν(q), ν(q)−1 q). 2
×
Writing a quaternion x as αq, α ∈ R+ , q ∈ U (H) is a polar form of the
quaternion.
26.4 Complex numbers and quaternions
Let us look at elements of order 2 in H× .

Proposition 26.9 The following statements hold for x ∈ H× .

(i) x2 ∈ R if and only if x ∈ R ∪ H0 .
(ii) x ∈ R ∪ H0 if and only if x2 ≥ 0.
(iii) x ∈ R ∪ H0 if and only if x2 ≤ 0.
(iv) |x| = 2 if and only if x = −1
(v) |x| = 4 if and only if q(x) = 1 and x ∈ H0
Proof: We write x = α + a with α ∈ R, a ∈ H0 . Then x2 = (αα − a • a) +
(αa + αa + a × a) = α2 − a • a + 2αa, proving (i), (ii) and (iii). Now x2 = 1
if and only if a = 0 and α2 = 1. Since |1| = 1, −1 is the only element of
order 2 in H× . Finally |x| = 4 if and only if x2 = −1 if and only if ν(a) = 1
and α = 0. 2

107
 Thus every imaginary quaternion a ∈ H0 of norm 1 is an imaginary unit.
Imaginary units form a 2-sphere S 2 ⊆ H0 . Each imaginary unit q ∈ S 2
defines a homomorphism C → H, α + βi 7→ α + βq. Thus, in a strange
interplay between Algebra and Geometry the set of all homomorphisms from
C to H is a 2-sphere.
26.5 Hopf fibration
Hopf fibration has been described in Vista Section 14.5. Before we give
a simpler description of it, we need to understand 3D-rotations. Let Rβx
be the anticlockwise rotation by the angle β in the plane orthogonal to x:

x (w) =
Lemma 26.10 If a = cos θ+x sin θ for some imaginary unit x then R2θ
−1
awa for each w ∈ H0 .
Proof: Choose y, z ∈ H0 so that x, y, z is a positive oriented orthonormal
basis. From Theorem 26.2, it follows that x2 = y 2 = z 2 = −1, xy = −yx =
z, yz = −zy = x and zx = −xz = y.
It suffices to check the proposition on the basis because both parts of the
equality R2θ x (w) = awa−1 are results of linear maps applied to w. Notice

that a −1 = cos θ − x sin θ. Let us calculate. First, axa−1 = xaa−1 =

x = R2θ x (x). Then aya−1 = (cos θ + x sin θ)y(cos θ − x sin θ) = (y cos θ +

z sin θ)(cos θ − x sin θ) = ((cos θ)2 − (sin θ)2 )y + (2 cos θ sin θ)z = y cos 2θ +
z sin 2θ = R2θ x (y) and finally aza−1 = (cos θ + x sin θ)z(cos θ − x sin θ) =

(z cos θ − y sin θ)(cos θ − x sin θ) = ((cos θ)2 − (sin θ)2 )z − (2 cos θ sin θ)y =
z cos 2θ − y sin 2θ = R2θx (z). 2
3
The sphere S is the group of norm 1 quaternions. The sphere S is 2

the set of imaginary norm 1 quaternions. The action map S 3 × S 2 → S 2

is written using the multiplication in the quaternions: g · x = gxg −1 . It is
well-defined since S 2 is the set of elements of order 4 in H× and |x| = |gxg −1 |.

108
Theorem 26.11 The U (H)-set S 2 has one orbit. The stabiliser of x ∈ S 2
is U (H) ∪ R(x).
Proof: If x, y ∈ S 2 then R2θ z (x) = y where z ∈ H is any unit vector
0
orthogonal to both x and y and 2θ is the angle between x and y. By
Lemma 26.10, a · x = axa−1 = y where a = cos θ + x sin θ ∈ S 3 = U (H).
Thus, the U (H)-set S 2 has one orbit.
To compute the stabiliser of x ∈ S 2 , observe that an arbitrary ele-
ment a ∈ U (H) can be written as a = cos θ + y sin θ where y ∈ S 2 . By
y
Lemma 26.10, a ∈ Stab(x) if and only if a · x = R2θ (x) = x. For this to
happen we need x and y to be parallel (then y = ±x and a ∈ U (H) ∪ R(x))
or 2θ = 2nπ for some n ∈ Z (then θ = nπ and a = cos nπ + y sin nπ = ±1.
2
Geometrically, the stabiliser U (H) ∪ R(q) is the unit circle in R(x) = C.
Choosing a particular quaternion x ∈ S 2 , its orbit map βx : U (H) → S 2 ,
βx (g) = gxg −1 is the Hopf fibration S 3 → S 2 : the inverse image βx−1 (y) is a
coset of the stabiliser U (H) ∪ R(x), i.e. geometrically a circle.
26.6 Exercises
(i) Prove that CH (K) = R + RK = R(K).
(ii) Prove that Z(H) = R.
(iii) Prove that U (H) = SU2 (C) as subgroups of GL2 (C) (See Vista
Section 14.5 for the definition of SU2 (C).
(iv) Prove that a × b = (ab − ba)/2 for all a, b ∈ H0 .
(v) Using Exercise (iv), prove Jacobi’s identity a × (b × c) + b × (c ×
a) + c × (a × b) = 0 for all a, b, c ∈ H0 .
(vi) Prove Schwarz’s inequality ||x|| · ||y|| ≥ |x • y|.
26.7 Vista: from multiplication tensors to superstring the-
ory
We have successfully used a multiplication table to describe quaternionic
multiplication. Pushing this through for a general algebra leads to tensors.
Let (R, F) be an algebra. Pick elements ei ∈ R constituting a basis of R
as a vector space. We define multiplication for basis elements
X
ei · ej = mki,j ek (1)
k

for uniquely determined mki,j ∈ F. These numbers are called structure con-
stants. Together they form a (2,1)-tensor on the vector space R.
We extend this formula by bilinearity, so the multiplication in R is dis-

109
 P P P
tributive. Notice that if a = i αi ei , b = j β j ej , c = k γ k ek then
X X
(ab)c = αi β j γ k (ei · ei ) · ek , a(bc) = αi β j γ k ei · (ei · ek ).
i,j,k i,j,k

This implies that it is sufficient to check associativity on the basis,

X X
(ei · ej ) · ek = msi,j es · ek = msi,j mts,k et .
s s,t

Similarly, X X
ei · (ej · ek ) = msj,k ei · es = msj,k mti,s et .
s s,t

Thus, associativity is equivalent to the system of quadratic equations on the

structure constants X X
msi,j mts,k = msj,k mti,s (2)
s s
for all possible i, j, k, and s.
P At the end we should not forget to ensure an identity element 1R =
i ui ei . Usually, it is rather straightforward if 1R exists.
Let us try to cook up the structure constants in a very naive way. Let
U be an open subset of Rn . Let us pick a three times differentiable function
Φ : U → R and a point y ∈ U . Structure constants
∂3Φ
mki,j = (y)
∂xi ∂xj ∂xk
define an R-algebra structure on Rn via formula (1) as soon as this multi-
plication is associative and unitary. Equation (2) becomes rewritten as
X ∂3Φ ∂3Φ X ∂3Φ ∂3Φ
(y) (y) = (y) (y).
s
∂xi ∂xj ∂xs ∂xs ∂xk ∂xt s
∂x j ∂x k ∂x s ∂x i ∂x s ∂x t

Should you require now to obtain an associative multiplication at every

point y ∈ U , you end up with a system of non-linear third order differential
equations for each i, j, k, t
X ∂3Φ ∂3Φ X ∂3Φ ∂3Φ
= . (3)
s
∂xi ∂xj ∂xs ∂xs ∂xk ∂xt s
∂xj ∂xk ∂xs ∂xi ∂xs ∂xt

System (3) is known as WDVV-equation in modern physics. It is an equation

for potential in Superstring Theory. It is not known how to solve WDVV-
equation in general.

110
27 Quaternions and spinors
Using quaternions, we describe 3D and 4D spinors.
27.1 Exponents of quaternions
In Algebra-I, you have studied matrix exponents. Since H is an R-
subalgebra of M2 (C), one can formally define
∞
X
a 1 n
e = a = 1 + a + a2 /2! + . . .
n!
n=0

While computation of matrix exponents require some subtle techniques such

as Jordan forms or Lagrangian interpolation, quaternionic exponents are
more straightforward.

Proposition 27.1 (Quaternionic Euler’s formula) If a = α + βx for α, β ∈

R and some imaginary unit x ∈ U (H) ∩ H0 then ea = eα (cos β + x sin β).
PK
Proof: Since a ∈ R(x) ∼
= C, all partial sums an /n! belong to R(x) ∼
n=0 =
C and quaternionic Euler’s formula follows from the usual Euler’s formula
for R(x). 2
Since quaternions do not commute, ea+b 6= ea eb , in general. On the other
hand, if ab = ba then both a and b lie in the same R(x) and ea+b = ea eb =
eb ea ∈ R(x) (cf. exercises (i) and (ii)). Since powers of a single element
commute, the usual De Moivre’s Formula holds for any x ∈ U (H) ∩ H0 :

(cos β + x sin β)n = cos nβ + x sin nβ .

The following proposition is immediate.

Proposition 27.2 The element e2πx/n has order n in H× for any imaginary
unit x ∈ U (H) ∩ H0 .

27.2 3D spinors
Let us play more with the action of U (H) on H0 given by a · x = axa−1 .
In the last lecture we have realized that the orbit map of this action is a
Hopf fibration. Now we would like to study the action map.

Theorem 27.3 The action map φ : U (H) → SO(H0 ) ∼

= SO3 (R) is a sur-
jective two-to-one group homomorphism.
Proof: Apriori, the action map is a group homomorphism φ : U (H) →
S(H0 ). If a ∈ U (H) then a = cos θ + x sin θ for some imaginary unit x and

111
 x (w) = awa−1 for each w ∈ H by Lemma 26.10. Hence, the action map
R2θ 0
is a group homomorphism φ : U (H) → SO(H0 ).
If A ∈ SO(H0 ) ∼ = SO3 (R), the characteristic polynomial χA (z) have
degree 3, so A has a real eigenvalue λ with an eigenvector v. Since ||Av|| =
||v||, |λ| = ±1. Moreover, A preserves the orthogonal complement v ⊥ and
A|v⊥ is orthogonal. If λ = 1 then det(A|v⊥ ) = 1 and A|v⊥ is a rotation.
Thus, A is a rotation Rθv by some angle θ. If λ = −1 then det(A|v⊥ ) = −1
and A|v⊥ is a reflection. Thus, A has two more eigenvectors, including an
eigenvector w with eigenvalue 1. Hence, A is a rotation Rπw . It follows that
φ : U (H) → SO(H0 ) is surjective.
x (w) =
Finally, a = cos θ + x sin θ is the kernel of φ if and only if R2θ
−1
awa = w for each w ∈ H0 if and only if 2θ = 2nπ if and only if θ = nπ if
and only if a = ±1. 2
Using the action homomorphism any SO3 (R)-set becomes a U (H)-set.
The opposite is not true: a U (H)-set is an SO3 (R)-set if and only if −1 lies
in the kernel of the action. A 3D-spinor is an element of a U (H)-set25 which
is not an SO3 (R)-set.
Examples. 1. Any quaternion x ∈ H is a spinor. The action is left multi-
plication in H: a · x = ax.
2. Physicists like illustrating the spinors using the following device.

Let A be the set of all positions of this device such that the centre of
the cork remains in the centre of the cube. We say that two positions are
equivalent if one can be moved to another by adjusting elastic thread only.
The quotient set X = A/ ∼ is a U (H)-set: an element a acts on X by
25
Wikipedia (http://en.wikipedia.org/wiki/Spinor) defines a spinor as an element of a
representation, a U (H)-vector space rather than merely a U (H)-set. This difference is
immaterial: a vector space is a set and any set is a subspace of a vector space, the space
of formal linear combinations of the elements.

112
φ(a) rotating the cork. It is not a SO3 (R)-set because a 360-degree rotation
tangles the thread. It is less obvious that a 720-degree rotation does not
tangle the thread (see http://www.youtube.com/watch?v=O7wvWJ3-t44).
3. A variation of example 2 is Feynman dance:

If we call position 1 x then position 2 is eπI/2 · x, position 3 is eπI · x and

position 4 is e3πI/2 · x. Observe that the further 180-degrees rotation from
position 4 returns everything into position 1, mathematically eπI/2 · (e3πI/2 ·
x) = x.
27.3 Binary dihedral group
We have seen that the natural homomorphism φ : U (H) → SO(H0 )
is two-to-one, i.e. the inverse image φ−1 (A) of each point consists of two
elements. Any subgroup G ≤ SO3 (R) gets its binary analogue φ−1 (G).
In particular, the binary dihedral group BD4n is the subgroup of U (H),
generated by J = e2πJ/4 and x = eπI/n . Notice that if n = 4 then x = I and
BD8 is our old friend the quaternionic group Q8 .

Proposition 27.4 BD4n is a group of order 4n and φ(BD4n ) ∼

= D2n .
Proof: All we need to observe is how φ(J) and φ(x) act on J − K-plane.
φ(x) acts via rotation R2π/n , while φ(J) is the reflection fixing J and sending
K to −K. Thus, φ(BD4n ) ∼ = D2n .
It remains to observe that −1 = J 2 ∈ BD4n . Consequently, BD4n =
−1
φ (φ(BD4n )) has order n. 2
As JxJ −1 −1 −1
= x (see Exercise (iii)), Jx = x J and every element of
BD4n can be written as xk J m . Since J 2 = −1 = xn , BD4n = {xk , xk J | k ∈
Zn } looking surprisingly similar to D4n . Similarly to Section 5.3, we write
the multiplication table of BD4n using addition in Zn :

113
 xl xl J
xk xk+l xk+l J
xk J xk−l J xk−l+n
Nevertheless, these groups are not isomorphic. For instance, BD4 ∼
= C4
while D4 ∼
= K4 or BD8 ∼ = Q8 .

Proposition 27.5 BD4n is not isomorphic to D4n for any n ≥ 1.

Proof: −1 is the only order 2 element of BD4n , while D4n has at least 2
different reflections. 2
Proposition 27.5 gives us the remaining group of order 12: BD12 cannot
be isomorphic to A4 either as A4 has 3 elements of order 2.
27.4 4D spinors
Amazingly 4D-spinors can described in a similar way to 3D-spinors. The
group U (H) × U (H) on H by (a, b) · x = axb−1 . Clearly, these are orthogonal
transformations of H.

Theorem 27.6 The action map φ : U (H) × U (H) → SO(H) ∼

= SO4 (R) is
a surjective two-to-one group homomorphism.
It is prudent at this stage to stop giving complete proofs. The main
issue in this theorem is surjectivity. One can do it in a straightforward way
but it is more elegant to argue using (unproved in this module) Coxeter’s
theorem that any orthogonal transformation in dimension n is a product of
at most n + 1 reflections. A reflection of a euclidean space V is a linear
transformation
hx, yi
Sx : V → V, Sx (y) = y − 2 x.
hx, xi
for some x ∈ V \{0}. Reflection Sx fixes the plane x⊥ while sending x to −x.
In particular, it has determinant −1. Using Coxeter’s theorem, elements of
SO(H) are products of two or four reflections.

Lemma 27.7 If x ∈ U (H) then Sx (y) = −xy ∗ x for each y ∈ H.

One can establish this lemma by a direct calculation. It implies Sx Sy =
φ(xy, yx), proving Theorem 27.6.
Now 4D-spinors are elements of a U (H)×U (H)-set, which is not SO4 -set.
In contrast to 3D-spinors, quaternions are not 4D-spinors because (−1, −1)
acts trivially on H. We have no intention of giving meaningful examples
of 4D-spinors. Instead, in preparation to the grand finale, we observe the

114
structure of SO(H). An element x ∈ U (H) gives rise to a left scroll Lx (y) =
xy and a right scroll Rx (y) = yx−1 . Right and left scrolls commute: Lx Ry =
Ry Lx . Finally, every element of SO(H) is a composition of left and right
scroll (surjectivity of φ in Theorem 27.6).
27.5 Regular polytopes
You probably know that there exists 5 regular polyhedrons (3D-polytopes),
often called platonic solids. It is a drastic contrast with regular polygons
(2D-polytopes) whom there are infinitely many. What is about regular nD-
polytopes? The answer is surprising: if n ≥ 5 there are 3 regular nD-
polytopes, but there are 6 regular 4D-polytopes26 . Our aim is to sketch
construction of the higher dimensional regular polytopes.
Let us start with the three that exist in any dimension. n-hypercube is
the easiest one to imagine: its 2n vertices have coordinates (±1, ±1 . . . , ±1).
n-hypercube is the convex hull of them.
The next one is n-simplex: it is a convex hull of n + 1 points. 2-simplex
is a regular triangle and 3-simplex is a regular tetrahedron.
The last universal one is n-orthoplex. It is the dual27 polyhedron of the
n-cube. In another language, the 2n vertices of n-orthoplex are centres of
n − 1-dimensional faces of an n-cube. The n-orthoplex itself is the convex
hull of its vertices. The 4-orthoplex (often called 16-cell) has a particularly
nice structure: its vertices are elements of the group BD8 = Q8 !
This gives us an idea take a finite group G ⊆ U (H) and consider its
convex hull. The resulting 4-polytope is bound to have a high degree of
symmetry: left and right scrolls with respect to the elements of G are sym-
metries of the resulting polytope. Unfortunately, no other BD4n gives a
regular solid. It is instructive to realize why the hull of BD16 is not a 4-
cube: 2D-faces of a 4-cube are squares while two of the 2-sides of the hull
of BD16 are octagons.
The key is to find more finite subgroups. One gets them by lifting rota-
tional symmetries of other platonic solids. The first platonic solid is tetrahe-
dron. The group of its rotational symmetric has order 12. Its inverse image
in U (H) is called binary tetrahedral group:
√ 1 + I + J + K ((I−J+K)π/3√3)
BT =< e((I+J+K)π/3 3)
= ,e ,I >
2
∼
=< a, b, c | a3 = b3 = c2 = abc >
26
Wikipedia has numerous interlinked pages (http://en.wikipedia.org/wiki/Regular polytope)
with a wealth of information related to this lecture.
27 ∗
X = {v | ∀a ∈ X | < a, v >≤ 1}

115
Observe that BT has 24 elements but it is not isomorphic to S4 (again BT
has only 1 element of degree 2). In fact, BT is isomorphic to SL2 (Z3 ). It is
more essential for us that the convex hull of BT is a new self-dual regular
4D-polytope, called 24-cell. Self-duality manifests in the fact that it has 24
vertices and 24 3D-faces.
Cube and octahedron are dual of each other. They have the same ro-
tational symmetry group of order 24 (in fact it is isomorphic to S4 ). Its
inverse image in U (H) is called binary octahedral group:
1+I √ 1+I +J +K
BO =< e(Iπ/4) = √ , e((I+J+K)π/3 3) = ,I >
2 2
∼
=< a, b, c | a4 = b3 = c2 = abc >
This group of order 48 is unusual in many respects. In particular, it has no
other notable description as a group. Its convex hull fails to be a regular
solid but has some interesting properties.
We are left with icosahedron and dodecahedron, which are dual of each
other. Their rotational symmetry group has order 60 (in fact it is isomorphic
to S5 ). Its inverse image in U (H) is called binary icosahedral group:

BI =< e((I cos π/3+K sin π/3)π/5) , e((I cos π/5+K sin π/5)π/3) , I >
∼
=< a, b, c | a5 = b3 = c2 = abc >
This group of order 120 is isomorphic to SL2 (Z5 ). The convex hull of BT
is a new regular 4D-polytope, called 600-cell. Its 600 3D-faces are regular
tetrahedra.
The remaining regular 4D-polytope is the dual of 600-cell. It is called
120-cell. Its 120 3D-faces are dodecahedra.
27.6 Exercises
(i) Prove that if ab = ba for some a, b ∈ H then there exists an imagi-
nary unit x ∈ U (H) ∩ H0 such that a, b ∈ R(x) (cf. Exercise 26.6(i)).
(ii) Prove that if AB = BA for some A, B ∈ Mn (C) then28 eA+B =
eA eB .
(iii) Show that JxJ −1 = x−1 if x = eπI/n .
(iv) Prove Lemma 27.7.
(v) Using Lemma 27.7 prove that Sx Sy = φ(xy, yx).
(vi) Prove that the group of rotational symmetries of a tetrahedron is
isomorphic to A4 .
28
A general formula eA+B = F (eA , eB ) is called Baker-Campbell-Hausdorff formula.

116
27.7 Vista
There is no vista for you now: your limit is the sky now.
On a more serious note, if you were thinking of writing an essay29 on
quaternions you, can still do it. You can expand the last section where
no proofs were given. You can also discuss integer quaternions or 4-square
theorem. Another alternative is to describe quaternions by applying Cayley
process to complex numbers. You can apply Cayley process to quaternions
to obtain octonions or so called Cayley numbers. They are still a division
algebra, albeit nonassociative. Another direction is Clifford algebras and
spinors in dimension n.
Certainly, you can write an essay on Physics or Geometry as we have
not touched any of those issues.

29
Consult On quaternions and octonions by Conway and Smith or Regular Polytopes by
Coxeter.

117