Groups, Rings and Modules

Prof. C.J.B. Brookes

Lent 2005

A
L TEXed by Sebastian Pancratz
ii

These notes are based on a course of lectures given by Prof. C.J.B. Brookes in Part IB of
the Mathematical Tripos at the University of Cambridge in the academic year 20042005.

These notes have not been checked by Prof. C.J.B. Brookes and should not be regarded
as ocial notes for the course. In particular, the responsibility for any errors is mine 
please email Sebastian Pancratz (sfp25) with any comments or corrections.
 Contents

Introduction i

1 Groups 1
1.1 Basic concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Normal subgroups, quotient groups, homomorphism and isomorphisms . 2

1.3 Permutation groups, actions and representations . . . . . . . . . . . . . 6

1.4 Conjugacy classes, centralisers and normalisers . . . . . . . . . . . . . . 8

1.5 Finite p-groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.6 Finite abelian groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1.7 Sylow's theorems and applications to groups of small orders . . . . . . . 12

2 Rings 17
2.1 Denitions and examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.2 Homomorphisms, ideals, quotient rings and isomorphisms . . . . . . . . 19

2.3 Integral domains, elds of fractions, maximal ideals and prime ideals . . 23

2.4 Factorisation in integral domains  units, primes, irreducibles . . . . . . 26

2.5 Factorisations in polynomial rings, Gauss' lemma and Eisenstein's criterion 30

2.6 Gaussian integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.7 Z[α] with α an algebraic integer . . . . . . . . . . . . . . . . . . . . . . . 37

2.8 Hilbert's basis theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

3 Modules 41
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.2 Direct sums, free modules . . . . . . . . . . . . . . . . . . . . . . . . . . 44

3.3 Matrices over Euclidean domains, equivalence of matrices, Smith normal

form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

3.4 Modules over F[X] for a eld F  normal forms for matrices . . . . . . 51
 Introduction

Groups
This part of the course builds on the Part IA course Algebra & Geometry. The rst
punchline we aim for will be Sylow's theorems. We'll especially think p-groups, that is,
groups of order pa .

Rings
A ring is a set R together with two operations + and ×, where multiplication is distrib-
utive over addition. This includes elds, but also the integers Z and polynomial rings
C[X]. We will concentrate on commutative rings.

Number theory
The Gaussian integers are Z[i] ⊂ Q[i] ⊂ C. This is an example of adjoining roots of
integral polynomials to Q, or Z. These rings analogous to integers do not necessarily
have unique factorization. In fact, we do get unique factorization for Gaussian integers
since Euclid's algorithm works.

These ideas will be continued in the Part II course Number elds.

Algebraic geometry
A family J of polynomials in C[x1 , . . . , xn ] in n variables induces a subset of Cn which
consists of(λ1 , . . . , λn ) giving you zero for all polynomials in J , that is, a set of common
roots. Hilbert's basis theorem implies that all but nitely many of the polynomials in J
are redundant. The set of common zeros of J is equal to the set of common zeros of a
nite subset of J .

Modules
Modules are a generalisation of vector spaces, using scalars from a ring rather than a
eld. Our aim is a structure theorem for rings in which Euclid's algorithm works, for
example Z. We also consider the structure of algebraic groups, which is important in
algebraic topology, as well as C[X] and the Jordan normal form.

Lecture 1
ii Introduction

E-mail
The lecturer can be contacted at brookes@[Link].

Books
• There is a long list of recommended books in the schedules.

• J.B. Fraleigh, A First Course in Abstract Algebra

• B. Hartley, T.O. Hawkes, Rings, Modules and Linear Algebra
• P.J. Cameron, Introduction to Algebra
• M. Artin, Algebra

Lecture 1
 Chapter 1
Groups
1.1 Basic concepts
Denition (Group) . A set G is a group if there is a binary operation G × G → G,
satisfying

(i) If g1 , g 2 ∈ G then g1 g2 ∈ G (closure).

(ii) For all g1 , g2 , g3 ∈ G, (g1 g2 )g3 = g1 (g2 g3 ) (associativity).

(iii) There exists e∈G such that, for all g ∈ G, eg = ge = g (identity element).

(iv) For all g∈G there exists g −1 ∈ G, gg −1 = g −1 g = e (inverses).

Denition (Subgroup) . A subset H is a subgroup of G if it is a group under the

restriction of the operation. We write H ≤ G.

Example. (i) Additive groups, e.g. (Z, +), (R, +), (C, +).

(ii) Matrix groups, such as the general linear group GL(Rn ), i.e. the group of invertible
n×n n
matrices, or the special linear group SL(R ), i.e. matrices with determinant 1.

(iii) Permutation groups, i.e. the symmetric group Sn of permutations on {1, . . . , n}

and the alternating group An containing even permutations.

(iv) Cyclic groups of order n, i.e. {e, g, . . . , g n−1 }, and dihedral groups D2n of order
2n, i.e. the symmetries of the regular n-gon.

(v) Abelian groups, e.g. groups of order 4 including the Viergruppe V.

(vi) The quaternion group {±1, ±i, ±j, ±k}, ij = k, ji = −k of order 8.

Given a subgroup H ≤ G, we can dene an equivalence relation on G by

g1 ∼ g2 ⇐⇒ g1−1 g2 ∈ H.

This partitions G into equivalence classes, called left cosets of H in G,

gH = {gh : h ∈ H}.

All these cosets are of the same size |H|. Counting the elements of G we obtain

Lecture 1
2 Groups

Theorem 1.1 (Lagrange). Let G be a nite group with subgroup H. Then

|G| = |H||G : H|

where |G : H| is the number of left cosets of H in G, called the index of H in G.

Note 1. You can also do the same with the equivalence relation g1 ∼ g2 ⇐⇒ g2 g1−1 ∈
H, obtaining right cosets as equivalence classes. The number of right cosets is equal to
the number of left cosets.

Denition (Order). The order o(g) of g ∈ G is the least n ∈ N such that g n = e. Note
this implies that if g
m =e then o(g) | m.

Lemma 1.2. The order o(g) divides |G|.

Proof. The set {e, g, . . . , g o(g)−1 } is a subgroup of G. Now apply Theorem 1.1.

1.2 Normal subgroups, quotient groups, homomorphism

and isomorphisms
Example. The integers lie inside the real numbers, (Z, +) ≤ (R, +). We would like to
add cosets,
(Z + r1 ) + (Z + r2 ) = Z + (r1 + r2 ).

In general, we want to copy this procedure to dene an operation on the cosets of a

subgroup. If K ≤G we want to dene g1 Kg2 K = g1 g2 K . But this requires Kg2 K =
g2 K , after multiplying on left by g1−1 . In particular, we need Kg2 ⊂ g2 K for all g2 ∈ G,
and also Kg2−1 ⊂ g2−1 K which implies g2 K ⊂ Kg2 . So to make the denition work we
need that gK = Kg for all g ∈ G.

Denition (Normal subgroup). A subgroup K ≤ G is normal in G if gK = Kg for all

g ∈ G. Equivalently, gKg −1 = K for all g ∈ G. We write K / G.

Example. In abelian groups all subgroups are normal. In general, {e} and G are always
normal. In the dihedral group D2n , the subgroup of rotations is normal but {e, τ }, for
a reection τ, is not normal if n ≥ 3.

Proposition 1.3. K / G.
Let Then the set of cosets of K in G form a group under the
operation g1 K.g2 K = g1 g2 K .

Denition (Quotient group). This group is the quotient group G/K .

Proof. We need to check that the denition is well-dened, that is, if g1 K = g10 K and
g2 K = g20 K then g1 g2 K = g10 g20 K . (See Algebra & Geometry.)
The identity element is K, the inverse of gK is g −1 K . Associativity is inherited from
G.

Denition (Homomorphism) . A map θ: G → H where G and H are groups is a

homomorphism if
θ(g1 g2 ) = θ(g1 )θ(g2 ).

Lecture 2
 3

Lemma 1.4. (i) The kernel ker θ = {g ∈ G : θ(g) = e} / G is a normal subgroup.

(ii) The image Im θ = {θ(g) : g ∈ G} ≤ H is a subgroup.

Proof. (i) Let K = {g ∈ G : θ(g) = e}. Then the left coset gK is the set {g1 ∈ G :
θ(g1 ) = θ(g)} since

θ(g1 ) = θ(g)
⇐⇒ θ(g −1 g1 ) = e
⇐⇒ g −1 g1 ∈ K
⇐⇒ g1 ∈ gK.

Similarly, the right coset Kg is the set {g1 ∈ G : θ(g1 ) = θ(g)}. Thus Kg = gK .

(ii) We need to check that θ(g1 )θ(g2 )−1 ∈ Im θ for all g1 , g 2 . But this is θ(g1 g2−1 ).

Denition (Isomorphism). An isomorphism is a bijective homomorphism.

Observe that `isomorphic to' denes an equivalence relation on groups. The notation
used is ∼
=.

Theorem 1.5 (First isomorphism theorem). Let θ: G → H be a homomorphism. Then

ker θ is a normal subgroup and G/ ker θ ∼
= Im θ.

Proof. We dene a map Φ : G/K → Im θ, gK 7→ θ(g) where K = ker θ.

Check that Φ is well-dened,

g1 K = g2 K =⇒ g2−1 g1 ∈ K
=⇒ θ(g2−1 g1 ) = e
=⇒ θ(g2 )−1 θ(g1 ) = e
=⇒ θ(g1 ) = θ(g2 ).

Φ is a homomorphism,

Φ(g1 Kg2 K) = Φ(g1 g2 K)

= θ(g1 g2 )
= θ(g1 )θ(g2 )
= Φ(g1 K)Φ(g2 K).

Φ is injective. If θ(g1 ) = θ(g2 ) then θ(g1−1 g2 ) = e and so g1−1 g2 ∈ K . Hence g1 K = g2 K .

Clearly, Φ is surjective.

Example. Consider the map

θ : (R, +) → (C× , ×), r 7→ e2πir .

This is a group homomorphism, θ(r1 +r2 ) = θ(r1 )θ(r2 ). Its kernel is (Z, +). Theorem1.5
implies R/Z ∼
= Im θ = {e2πir : r ∈ R}. The map Φ as above is Φ(Z + r) = e2πir .

Lecture 2
4 Groups

Theorem 1.6 (Second isomorphism theorem). H ≤ G and K / G. Then HK ≤ G

Let
and H ∩K /G where HK = {hk : h ∈ H, k ∈ K} and (HK)/K ∼
= H/(H ∩ K).

Proof. First show that HK is a subgroup. Take hk ∈ HK . Then

(hk)(h1 k1 )−1 = h kk1−1 h−1

| {z } 1
∈K

But Kh−11 = h−1

1 K since K / G and so (kk1−1 )h−1
1 = h−1
1 k3 for some k3 ∈ K . So
(hk)(h1 k1 ) = hh−1
−1
1 k3 ∈ HK .
Dene
θ : H → G/K, h 7→ hK.
This is a group homomorphism,

θ(h1 h2 ) = h1 h2 K
= h1 Kh2 K
= θ(h1 )θ(h2 ).
So Theorem 1.5 implies that H/ ker θ ∼
= Im θ. Now

ker θ = {h ∈ H : hK = K} = H ∩ K
Im θ = {hK : h ∈ H} = (HK)/K,
cosets of K in HK . Thus H/(H ∩ K) ∼
= (HK)/K , as required.

In fact, there is a bijection between subgroups of G/K and the subgroups of G containing
K. This is given by the maps

{subgroups of G/K} {subgroups of G containing K}

X −→ {g ∈ G : gK ∈ X}
L/K ←− L

These maps are inverses of each other.

Take a normal subgroup K/G and form the quotient group G/K , with elements the
cosets gK and multiplication dened by g1 Kg2 K = g1 g2 K . The correspondence

{subgroups of G/K} ←→ {subgroups of G containing K}

restricts to

{normal subgroups of G/K} ←→ {normal subgroups of G containing K}

For a subgroup L/K ≤ G/K we have that

(gK)(L/K)(gK)−1 = (gLg −1 )/K

so

(gK)(L/K)(gK)−1 = L/K ⇐⇒ gLg −1 = L,

that is,

L/K / G/K ⇐⇒ L / G.

Lecture 3
 5

Theorem 1.7 (Third isomorphism theorem) . If K and L are normal subgroups of G

with K ≤L/G then (G/K)/(L/K) ∼
= G/L.

Proof. Apply Theorem 1.5 to the well chosen map

θ : G/K → G/L, gK 7→ gL.

We need to check this is well-dened,

g1 K = g2 K
=⇒ g1−1 g2 ∈ K ≤ L
=⇒ g1 L = g2 L.

θ is a homomorphism,

θ(g1 Kg2 K) = θ(g1 g2 K)

= g1 g2 L
= g1 Lg2 L
= θ(g1 K)θ(g2 K).

Clearly, Im θ = G/L. Also ker θ = {gK : gL = L} = L/K and so Theorem 1.5 implies

(G/K)/(L/K) ∼
= G/L.

Denition (Simplicity) . G is simple if the only normal subgroups in it are {e} and G
itself.

Example. The only simple abelian groups are Cp for p prime.

Proof. `Normal' for abelian groups is easy to understand as all subgroups are normal
because of commutativity. Take a non-trivial element g 6= e in a simple abelian group
G.
Then either g is of innite order and G contains the subgroup {. . . , g −1 , e, g, g 2 , . . . }.
Simplicity implies that G = {. . . , g −1 , e, g, g 2 , . . . }. But then there is also the subgroup
{. . . , g −2 , e, g 2 , g 4 , . . . }, a contradiction.

Or g is of nite order and G has the subgroup {e, g, g 2 , . . . , g o(g)−1 }. Simplicity implies
that G = {e, g, g 2 , . . . , g o(g)−1 }. But if o(g) = rs, say, then we also have the subgroup
{e, g r , g 2r , . . . , g r(s−1) }. Now simplicity implies that there are no non-trivial factorisa-
tions of o(g) so G = {e, g, . . . , g
p−1 } for some prime p.

We shall see shortly that the alternating group A5 on {1, . . . , 5} is simple. In fact, An
is simple for n ≥ 5, but we don't need to know the proof. Note that |A3 | = 3, so
A3 is isomorphic to C3 and simple. |A4 | = 12 and A4 contains the normal subgroup
V = {e, (12)(34), (13)(24), (14)(23)} / A4 , so A4 is not simple.
There are nitely many families of nite simple groups, and 26 groups that do not t
into those general families, called the sporadic ones.

Note|A5 | = 60 and |A6 | = 360. There is another nite simple group of order 168, GL3 (Z
mod 2), the group of invertible 3 × 3 matrices entries {0, 1} modulo 2. (This is left as
an exercise.)

Lecture 3
6 Groups

Theorem 1.8. Let G be a nite group. Then there are subgroups H1 , . . . , H s such that

G = H1 ≥ H2 ≥ · · · ≥ Hs = {e}

with Hi+1 / Hi and Hi /Hi+1 simple.

Warning. Not all Hi normal in G.

Proof. Let H2 be normal in G with |H2 | |G|. Then the

of largest order not equal to
correspondence between normal subgroups ofG containing H2 and the normal subgroups
of G/H2 implies that G/H2 simple. Repeat this to nd H3 / H2 with H2 /H3 simple.
The process stops when we reach Hs = {e}.

Remark. The simple groups Hi /Hi+1 are essentially unique, although there may be
lots of ways of picking the subgroups Hi , called composition factors.

Denition. G is soluble if all composition factors can be chosen to be cyclic of prime

order.

Remark. The terminology comes from Galois theory, see the relevant Part II course.
You all know a formula for solving quadratic equations. You can do something similar
involving roots for cubics and quartics, the process is called `solution by radicals'. But
you cannot necessarily do it for a quintic, e.g. t5 − 6t + 3 = 0. This can be shown by
dening a Galois group associated with the polynomial. If you can solve the polynomial
by radicals then the Galois group is soluble.

1.3 Permutation groups, actions and representations

The symmetry group Sn consists of permutations of {1, . . . , n}. Useful notation is the
disjoint cycle form, e.g. (1 2 3)(4 5)(6) where the single cycle is often suppressed.
It contains a subgroup An ≤ S n , the alternating group of even permutations.

We recall that every permutation can be expressed as a product of transpositions. An

element g is even if you need an even number of transpositions in such a product.
Equivalently in disjoint cycle form, g is even if the number of even length cycles is even,
e.g. (1 2 3) = (1 2)(2 3) is even but (1 2 3 4) = (1 2)(2 3)(3 4) is odd.

We note that the orders of these two groups are

|Sn | = n!,
n!
|An | = .
2

We can quickly see that the alternating group An / Sn is a normal subgroup. There are
two left cosets of An , Sn \ An = gAn for all permutations g and two right cosets of An ,
S n \ An = An g . Thus the left cosets are equal to the right cosets.

More generally, for any set Ω you can dene the symmetric group Sym G of all bijective
maps Ω → Ω.

Denition. A group G is a permutation group of degree n if G ≤ Sym Ω with |Ω| = n.

Lecture 4
 7

Example. Sn , An , D2n ≤ Sym Ω where Ω is the set of vertices of the regular n-gon.
Denition. An action of a group G on a set Ω is a map G × Ω → Ω, (g, α) 7→ g ∗ α,
often also written as g(α), such that

(i) g∗α∈Ω for each α ∈ Ω,

(ii) g1 ∗ (g2 ∗ α) = g1 g2 ∗ α,

(iii) e ∗ α = α.
Lemma 1.9. Let G act on Ω. Then for g∈G the map Φg : α 7→ g ∗ α is a permutation
of Ω and the map
Φ : G → Sym Ω, g 7→ Φg
is a homomorphism, a permutation representation of G.
The image Φ(G) Ω
is a subgroup of Sym Ω, denoted G . The kernel of Φ, ker Φ = {g ∈
G : g ∗ α = α for all α ∈ Ω}, also denoted G(Ω) , is a normal subgroup and G/GΩ ∼ = GΩ .

Proof. Φg−1 is the inverse of Φg and so Φg is a bijection. The map Φ : G → Sym Ω is a

homomorphism, from the second property of the denition of action. The rest is from
Theorem 1.5.

Example. (i) Let G be the symmetry group of the cube, |G| = 48, and Ω the set of
diagonals, |Ω| = 4. G acts on Ω,

G Ω = S4 ,
G(Ω) = {e, symmetry sending each vertex to the opposite one}.

(ii) The left regular action of G on Ω = G, given by

g ∗ g1 = gg1 .

Φg is left multiplication by g. The kernel is {e}, so Lemma 1.9 gives Cayley's

Theorem 1.10.

(iii) For a subgroup H of G, let Ω be the set of left cosets of H. Consider the action
of G on Ω given by
g ∗ g1 H = gg1 H.
T −1
The kernel is g1 ∈G g1 Hg1 since, for all g1 ∈ G,

gg1 H = g1 H
⇐⇒ g1−1 gg1 H = H
⇐⇒ g1−1 gg1 ∈ H
⇐⇒ g ∈ g1 Hg1−1 .

The kernel is the largest normal subgroup of G contained in H . If K / G, K ≤ H

−1
then g1 Kg1 =K since K is normal. But g1 Kg1−1 ≤ g1 Hg1−1 so K ≤ g1 Hg1−1 for
all g1 ∈ G.
Theorem 1.10 (Cayley). Any group G is isomorphic to a subgroup of Sym G.

Lecture 4
8 Groups

Theorem 1.11. Let G be a nite group and H a proper subgroup of G of index n.

Then there is a normal subgroup K of G contained in H such that G/K is isomorphic
to a subgroup of Sn . In particular |G : K| divides n! and is at least n.
Moreover, if G is non-abelian simple then G is isomorphic to a subgroup of An and
n ≥ 5.

Proof. Let K be the kernel of the action of G on Ω, where Ω is the set of left cosets
of H. Lemma 1.9 implies that G/K ∼
= GΩ , a subgroup of Sn . Lagrange's theorem 1.1
implies subgroups of Sn have order dividing n!. Since K ≤ H and H is of index n in G
we have that |G/K| ≥ n.
Now assume G is non-abelian simple. Then K = {e}, and so G∼
= GΩ subgroup of Sn .
But An / Sn and so GΩ ∩ An / GΩ . Simplicity implies that either

GΩ ∩ An = {e} or G Ω ∩ An = G Ω .

In the rst case, the second isomorphism theorem implies

GΩ /(GΩ ∩ An ) ∼
= (GΩ An )/An ≤ Sn /An

and hence |GΩ | ≤ 2 since GΩ ∩ An = {e}, a contradiction. Finally, we have n>4 since
A4 has no non-abelian simple subgroups.

Let us recall some denitions from the Part IA course Algebra & Geometry. Let G act
on Ω.
Denition. The orbit on Ω containing α is G(α) = {g ∗ α : g ∈ G}. The stabiliser of
α is Gα = {g ∈ G : g ∗ α = α}.
Theorem 1.12 (Orbit-stabiliser theorem). Let G act on Ω. Then |G : Gα | = |G(α)|.

Proof. The map

{cosets of Gα in G} → G(α), gGα 7→ g ∗ α
is a bijection.

1.4 Conjugacy classes, centralisers and normalisers

We consider the conjugation action of G on Ω = G, dened by

g ∗ x = gxg −1 .

for x ∈ Ω = G. In this case the permutation Φg : x 7→ gxg −1 as well as being bijective

is a homomorphism G → G. Thus they are automorphisms. The automorphism group
Aut G of isomorphisms G → G is a group under the composition of maps. We have
Φg ∈ Aut G.
Orbits are called conjugacy classes, cclG (x) = {gxg −1 : g ∈ G}. Stabilisers are called
centralisers of x, CG (x) = {g ∈ G : gxg −1 = x} = {g ∈ G : gx = xg}.
The orbit-stabiliser theorem implies |G : CG (x)| = |cclG (x)|. The kernel of the action

G(Ω) = {g ∈ G : ∀x ∈ Ω gxg −1 = x}

Lecture 5
 9

= {g ∈ G : ∀x ∈ Ω gx = xg}
\
= CG (x)
x∈G

is called the centre of G, Z(G).

Observe that |cclG (x)| divides |G|, which is part of Lagrange's Theorem 1.1.

Lemma 1.13.
P
Let G be a nite group. Then 1= 1/|CG (x)|, summing over distinct
conjugacy classes.

Proof.
P
We count the elements of G. |G| = |cclG (x)|, so

X |G|
|G| =
|CG (x)|
since |cclG (x)| = |G|/|CG (x)|. Now divide by |G|.

Example (Conjugacy classes of Sn and An ). Recall from Algebra & Geometry that two
permutations are conjugate in Sn precisely when they have the same cycle type when
expressed in disjoint cycle form, e.g. 22 1 denotes the cycle type of double transpositions
in S5 , e.g. (1 2)(3 4)(5).
In S5 , we have the following cycle types:

cycle type 15 22 1 3 12 5 2 13 32 41
# elements 1 15 20 24 10 20 30
even permutations odd permutations

For x ∈ An we have that cclAn (x) ⊂ cclSn (x). We also have

|cclSn (x)| = |Sn : CSn (x)|,

|cclAn (x)| = |An : CAn (x)|
by the orbit-stabiliser theorem. But CAn (x) = An ∩ CSn , and thus this set has index 1
or 2 in CSn (x). To see this, use the second isomorphism theorem, Theorem 1.6,

CSn (x)/(An ∩ CSn (x)) ∼

= (An CSn (x))/An ≤ Sn /An .

If all permutations that commute with x are even, i.e. CSn (x) = CAn (x),
|Sn | 2|An |
|cclSn (x)| = = = 2|cclAn (x)|.
|CSn (x)| CAn (x)

If there is an odd permutation commuting with x, i.e. |CSn (x) : CAn (x)| = 2,
|Sn | |An |
|cclSn (x)| = = = |cclAn (x)|.
|CSn (x)| CAn (x)
Example. Consider A5 , take x ∈ A5 . (1 2)(3 4)(5) commutes with (1 2),
S5 and
(1 2 3)(4)(5) commutes with (4 5). But conjugacy classes of 5 cycles do not split into 2
conjugacy classes of size 12. Consider (1 2 3 4 5) and its conjugates.

g(1 2 3 4 5)g −1 = (g(1) · · · g(5))

CSn (1 2 3 4 5) = {e, (1 2 3 4 5), (1 3 5 2 4), (1 4 2 5 3), (1 5 4 3 2)}
CSn (1 2 3 4 5) is of order 5, all elements are even.

Lecture 5
10 Groups

Proposition 1.14. A5 is simple.

Proof. Consider a normal subgroup K of A5 . Thus gkg −1 ∈ K for all k ∈ K , g ∈ A5 .

Thus K must be a union of conjugacy classes in A5 . But by Lagrange's Theorem 1.1,
|K| divides 60 = |A5 |. There is no way of taking a union of the conjugacy classes we've
worked out to give |K| dividing 60, unless K = {e} or K = A5 .

1.5 Finite p-groups

Denition (p-group). A nite group G is a p-group if |G| = pn for prime p.
Theorem 1.15. Let G be a nite p-group. Then Z(G) 6= {e}.

Proof. We count the elements of G,

X
|G| = |cclG (x)|.

But the size of conjugacy classes divides |G| = pn . So |cclG (x)| = pa for some 0 ≤ a ≤ n.

|G| = (no. of conjugacy classes of size 1) · 1

+ (no. of conjugacy classes of size p) · p
+ ···

But p divides |G|. So p divides the number of conjugacy classes of size 1. But {x} is a
conjugacy class if and only if

∀g ∈ G gxg −1 = x
⇐⇒ x ∈ Z(G)

So p | |Z|. Thus Z(G) 6= {e}.

Lemma 1.16. For any group G, if G/Z(G) is cyclic then G is abelian.

Proof. Suppose gZ generates G/Z(G) and thus each coset in Z is of the form (gZ)r =
gr Z for some r. Thus any
r
element x ∈ G is of the form g z for some z ∈ Z, r ∈ N.

x1 x2 = (g r1 z1 )(g r2 z2 )
= g r 1 g r 2 z 1 z2
= (g r2 z2 )(g r1 z1 )
= x2 x1

using that z1 ∈ Z . Thus G is abelian and so G = Z(G).

Proposition 1.17. If |G| = p2 for p prime then G is abelian.

Proof. |Z(G)| = p or p2 since by Lagrange's theorem |Z(G)| is 1, p

By Theorem 1.15,
2
or p . By Lemma 1.16, |Z(G)| =6 p, for if |Z(G)| = p then |G/Z(G)| = p and so G/Z(G)
2
cyclic. So |Z(G)| = p . Hence G is abelian.

Remark. Groups of order p3 are considered on the example sheet.

Lecture 5
 11

Recall the denitions of the direct product.

Denition (External direct product). G and H , G × H has a natural
Given two groups
group structure (g1 , h1 )(g2 , h2 ) = (g1 g2 , h1 h2 ). G × H has subgroups G1 = {(g, eH ) :
g ∈ G} ∼
= G and H1 = {(eG , h) : h ∈ H} ∼ = H . Note that G1 ∩ H1 is the trivial subgroup
of G × H and G1 H1 = G × H .

Denition (Internal direct product) . Given a group L and subgroups G1 , H1 with

G1 H1 = L, G1 ∩ H1 = {e}, all elements g ∈ G1 commute with all elements h ∈ H1 .
Note that L ∼
= G1 × H1 .
Theorem 1.18. Let G be a p-group of order pa . Then there is a subgroup of order pb
for any 0 ≤ b ≤ a.

Proof. By induction on a. The result is clearly true for a = 1. Suppose a > 1 and
pick an element x ∈ Z(G), x 6= e, noting that by Theorem 1.15 the centre is non-
trivial. By taking a suitable power of it, we may assume that x is of order p. Hence
K = {e, x, . . . , xp−1 } is a subgroup of
G, and it is normal in G since x is central.
Consider the quotient G/K .pa−1 and we can apply the inductive hypothesis
Its order is
to get L/K ≤ G/K of order p
b−1 . The correspondence between subgroups of G/K and

subgroups of G containing K gives a subgroup L of G containing K of order p .

b

1.6 Finite abelian groups

For example, by Proposition 1.17 groups of order p2 are abelian, and hence isomorphic
to Cp2 or Cp × Cp .
Theorem 1.19. Let G be a nite abelian group. Then

G∼
= Cd1 × Cd2 × · · · × Cdk
with di+1 | di and d1 · · · dk = |G|.

Proof. Postponed until Section 3.

Remark (Remark on proof ). One can be less sophisticated. As the rst step, consider
an abelian p-group G hxi generated by x ∈ G of maximal order. If
and the subgroup
hxi = G we are done. Otherwise, if hxi 6= G, the claim is that there exists a subgroup
H ≤ G with hxiH = G, hxi ∩ H = {e}, i.e. G ∼ = hxi × H .
Example. The abelian groups of order 8 are C8 , C4 × C2 and C2 × C2 × C2 . The abelian
groups of order 16 are C16 , C8 × C2 , C4 × C4 , C4 × C2 × C2 and C2 × C2 × C2 × C2 .
Lemma 1.20. If m and n are coprime then Cm × Cn ∼
= Cmn .

Proof. Take elements g of order m in Cm and h or order n in Cn .

Then (g, h) is of order mn in Cm × Cn since (g, h)r = (g r , hr ) and so the order of (g, h)
is the least common multiple of m and n which is mn since m and n are coprime.

But |Cm × Cn | = mn and so (g, h) is a generator for the group.

Example. C2 × C3 ∼
= C6 . Abelian groups of order 24 = 3 × 8 are C24 , C12 × C2 , and
C6 × C2 × C2 .

Lecture 6
12 Groups

1.7 Sylow's theorems and applications to groups of small

orders
Theorem 1.21 (Sylow) . Let G be a nite group of order pa m where p is prime and
p - m.

(i) There exists a subgroup P of order pa , called the Sylow p-subgroup.

(ii) All Sylow p-subgroups are conjugate.

(iii) The number np of Sylow p-subgroups satises np ≡ 1 (mod p) and divides |G|,
and hence np | m.
Remark. G acts on the set Ω of Sylow p-subgroups via conjugation, g ∗ P = gP g −1 . By
Sylow's second theorem, there is exactly one orbit. The stabiliser, called the normaliser
of P, is NG (P ) = {g ∈ G : gP g −1 = P }. By the orbit-stabiliser theorem, the size of the
orbit is np = |G : NG (P )|. Thus np | |G|.
Lemma 1.22. If np = 1 then the unique Sylow p-subgroup is normal in G.

Proof. Take the unique Sylow p-subgroup P . Then gP g −1 is also a Sylow p-subgroup.
Thus gP g
−1 =P for all g ∈ G.

Remark (continued). P /NG (P ). So applying Sylow's second theorem to

By denition,
NG (P ) we see that all Sylow p-subgroups of NG (P ) are conjugate to P . But P / NG (P )
and so all its conjugates are P itself. Thus P is the unique Sylow p-subgroup of NG (P ).

Corollary 1.23. Let G be a non-abelian simple group. Then |G| divides np !/2, and
np ≥ 5 once we have shown that all non-abelian subgroups of S4 are not simple.

Proof. G is acting on the set Ω of Sylow p-subgroups where |Ω| = np . IfG is non-abelian
simple then by Theorem 1.1 G is isomorphic to a subgroup of Anp , and np ≥ 5 subject
to proviso.

Example. Let |G| = 1000 = 23 53 . Then G is not simple.

Proof. n5 ≡ 1 (mod 5) and n 5 | 8. So n5 = 1 and so there exists a unique Sylow

5-subgroup which is normal. Thus G is not simple.

Example. Let |G| = 300 = 22 · 3 · 52 . Then G is not simple.

Proof. n5 ≡ 1 (mod 5) andn5 | 12. Assume G is simple and so n5 6= 1. Then n5 = 6.

But 300 does not divide 6!/2. Contradiction! Thus |G| = 300 implies that n5 = 1. (G
is not abelian simple since it is not of prime order.)

Example. Let |G| = 132 = 22 · 3 · 11. Then G is not simple.

Proof. n11 ≡ 1 (mod 11) and n11 | 12. Assume G is simple and so n11 6= 1. Then n11 =
12. n3 ≡ 1 (mod 3) and n3 | 44. By simplicity, n3 6= 1 and n3 6= 4 by Corollary 1.23, so
n3 = 22.
But since n11 = 12 and n3 = 22 there are 12 · (11 − 1) elements of order 11 and 22 · (3 − 1)
elements of order 3. But this gives too many elements, a contradiction!

Lecture 7
 13

This is a typical argument. We use Sylow's theorem to estimate the number of subgroups
and then count elements.

Lemma 1.24. Suppose |G| = 2p for an odd prime p. Then G∼

= C2p or D2p .

Proof. np ≡ 1 (mod p) and np | 2, so np = 1. Thus there is a normal subgroup K of

orderp. Sylow's theorem also implies there is a subgroup H of order 2. Let H = {e, g},
K = {e, x, . . . , xp−1 }. Note H ∩ K = {e} and G is a semidirect product of K by H .
Observe that |HK| = 2p.]

Now consider the action of H

K by conjugation. In particular, consider conjugation
on
by g  it permutes the elements of K . (In fact, conjugation by g gives an automorphism
of K .) This is uniquely determined by the image of x, that is, gxg
−1 = xr for some r .

Then also
2
x = g 2 xg −2 = gxr g −1 = (gxg −1 )r = xr .
So r2 ≡ 1 (mod p). This gives two choices. Suppose r ≡ 1 (mod p), then g and x
commute so gx is of order 2p and hence G ∼
= C2p . Otherwise suppose r ≡ −1 (mod p),
then gxg
−1 = x−1 = xp−1 and G ∼ D . In the latter case, x is a rotation and g is a
= 2p
reection.

There is a problem on the examples sheet for the case |G| = 15.

Proof of Sylow's theorems

Let G be a nite group of order |G| = pa m, where p is prime and p - m.

(i) There exists subgroups of order pa , called Sylow p-subgroups.

Proof. Consider the set Ω of subsets of G of size pa and let {g1 , . . . , gpa } ∈ Ω. G acts on
Ω via left multiplication,

g ∗ {g1 , . . . , gpa } = {gg1 , . . . , ggpa }.

Consider an orbit Σ ≤ Ω. Take g ∈ G. If {g1 , . . . , gpa } ∈ Σ then

gg1−1 ∗ {g1 , . . . , gpa } = {g, gg1−1 g2 , . . . , gg1−1 gpa } ∈ Σ.

So g appears as an entry in one of the pa -sets in orbit the Σ. Counting,

|G| pa m
|Σ| ≥ = = m.
pa pa

So we have two types of orbits Σ, (a) |Σ| = m, (b) |Σ| > m. By the orbit-stabiliser
theorem, |Σ| divides |G| = pa m. So for orbits Σ of type (b), where |Σ| > m, we deduce
that p | |Σ|.
The next step is to show that |Ω| is not divisible by p,
 a 
p m pa m pa m − 1 pa m − pa + 1
|Ω| = = · · · .
pa p a pa − 1 1

Lecture 7
14 Groups

Consider (pa m − k)/(pa − k) for 0 ≤ k ≤ pa − 1. If k=0 then

pa m − k pa m
= = m.
pa − k pa

If k>0 then set k = pb q , where p - q, so that

pa m − k pa m − pb q
=
pa − k pa − pb q
pa−b m − q
= a−b .
p −q
Note that b < a, so p does not divide the numerator after this cancellation process,
observing that b < a =⇒ k < pa . So after the cancellation for the product, p does not
divide the numerator. Thus p - |Ω|, as required.

Now count the elements of Ω,

|Ω| = sum of sizes of orbits

= sum of sizes of orbits of type (a)

+ sum of sizes of orbits of type (b).

But p - |Ω| and p divides the size of orbits of type (b), thus we do have orbits of type
(a).

Consider the orbit-stabiliser theorem for an orbit Σ of type (a),

|G| pa m
m = |Σ| = =
|P | |P |
where P is the stabiliser of an element chosen in Σ. So |P | = pa and thus there is a
a
subgroup of order p .

(ii) All Sylow p-subgroups are conjugate.

In fact, we will prove the following. If P is a Sylow p-subgroup and Q ≤ G with |Q| = pb
for 1 ≤ b ≤ a, then there exists g1 ∈ G such that Q ≤ g1 P g1−1 . In particular, if Q is of
order pa then Q = g1 P g1−1 .

Proof. This time consider the action of Q on the set of left cosets of P via left multipli-
cation,
g ∗ g1 P = gg1 P
for g ∈ Q, g1 ∈ G. We consider the orbits. By the orbit-stabiliser theorem, the size of
an orbit divides |Q|, thus is either 1 or a power of p.
The number of cosets is equal to the index of P in G,
|G| pa m
= a = m.
|P | p
In particular, m is equal to the number of orbits of size 1 plus pk times the number of
orbits of size k.
Counting cosets, we see that we must have an orbit of size 1. Suppose {g1 P } is such an
orbit. Thus gg1 P = g1 P for all g ∈ Q and hence g1−1 gg1 P = P . So g1−1 gg1 ∈ P and
g ∈ g1 P g1−1 . −1
Thus Q ≤ g1 P g1 , as required.

Lecture 7
 15

(iii) The number np of Sylow p-subgroups satises np ≡ 1 (mod p) and np | |G|.

Last time we observed that np is the index of NG (P ), the normaliser of a Sylow p-

subgroup P, in G and so np | |G|.

Proof. Consider the action of P on the set of Sylow p-subgroups of G via conjugation.
By the orbit-stabiliser theorem, orbits are of size 1 or of size divisible by p. We need to
show that there is exactly one orbit of size 1.
Visibly {P } is an orbit of size 1. Are there any others?

Suppose {Q} is an orbit of size 1. Thus gQg −1 = Q for all g∈P and so P ≤ NG (Q).
But we observed last time that the normaliser of a Sylow subgroup has a unique Sylow
subgroup. (This is a consequence of Sylow's second theorem applied to the normaliser.)
So P = Q, since both are Sylow subgroups of NG (Q).
Thus {P } is the only orbit of size 1.
Counting Sylow p-subgroups in G,

np = 1 + sum of sizes of other orbits.

But p divides the size of every other orbit. Therefore, np ≡ 1 (mod p).

Recall the denition of a soluble group, a group for which the composition factors can
be chosen to be abelian simple.

Theorem (1904, Burnside). If |G| = pa q b , where p and q are primes, then G is soluble.

Theorem (1937, Hall) . G is soluble if and only if whenever |G| factorises as |G| = mn
with m and n coprime there is a subgroup of order m.

Theorem (1963, Feit, Thompson, Odd Order Theorem). If |G| is odd then G is soluble.

Lecture 8
Lecture 8
 Chapter 2
Rings
2.1 Denitions and examples
Denition (Ring). A set R with two operations, + and ·, forms a ring if

(i) (R, +) is an abelian group,

(ii) multiplication is associative and there is a multiplicative identity 1R ∈ R such that

∀r ∈ R 1R · r = r · 1R = r,

(iii) multiplication is distributive over addition, that is

r1 (r2 + r3 ) = r1 r2 + r1 r3 ,
(r1 + r2 )r3 = r1 r3 + r2 r3 .

In this course all rings are assumed to be commutative, i.e. multiplication is commutative.

Denition (Subring) . A subset S of R is a subring if it is a ring under the restriction

of the two operations. In particular, 1R ∈ S . The notation is S ≤ R.
Example. We have the four nested rings Z√≤ Q ≤ R ≤ C√. In the introduction we also
met the Gaussian integers Z[i] ≤ C. Also Q[ 2] = {a + b 2 : a, b ∈ Q} ≤ R.
Remark. All elds are rings, e.g. Z modulo n, the set {0, . . . , n − 1} with addition and
multiplication modulo n.
Denition (Unit). A unit of R is an element with a multiplicative inverse in R, e.g. 2
is a unit in Q, but not a unit in Z.
Warning. There are various conventions.

(i) Not everyone insists on a multiplicative identity 1R .

(ii) Some people confusingly talk about 1R as the unit of R.
Example. The zero ring {0}. In this case the multiplicative identity is the same as the
additive one. But in all other rings 0 6= 1.
(0 + 0)r = 0r
(0 + 0)r = 0r + 0r
so 0r = 0 for all r ∈ R. Hence 0 is not the multiplicative identity if R has more than
one element.

Lecture 8
18 Rings

Example. Let R be a ring. A polynomial f over R is of the form

f (X) = an X n + · · · + a1 X + a0

with ai ∈ R. The degree of f is the largest n with an 6= 0. f is monic if an = 1, where

n is the degree of f . R[X] is the polynomial ring over R with ring operations

X
(f + g)(X) = (ai + bi )X i ,
i
i
X X 
(f g)(X) = aj bi−j X i
i j=0

ai X i , g(X) = bi X i .
P P
where f (X) =
ai X i
P
R is a subring of R[X] by identifying R with the constant polynomials f (X) =
with ai = 0 for all i ≥ 1.

R[[X]] is the ring of formal power series

f (X) = a0 + a1 X + a2 X 2 + · · ·

with the same addition and multiplication as for R[X] but note that there is no restriction
demanding that all but nitely many coecients be zero.

Laurent polynomials,
X
f (X) = ai X i ,
i∈Z
X X 
(f g)(X) = aj bi−j X i
i j

with the condition that all but nitely many ai are zero, e.g. X −1 + X .
Laurent series, X
ai X i
i∈Z

with the condition that all but nitely many of the ai for i≤0 are zero. Addition and
multiplication are dened in the same way as for Laurent polynomials.

Another example is given by the ring of all R-valued functions on a set A, f : A → R .

The ring operations can be dened by pointwise addition and multiplication,

(f + g)(a) = f (a) + g(a),

(f g)(a) = f (a)g(a).

For example, the set of continuous functions f : R → R forms a subring of the ring of
all functions R → R. R → R contains the subring
The set of continuous functions of
polynomial functions, f : R → R, a 7→ f (a), where f (X) ∈ R[X].
We can similarly consider Laurent polynomial functions, power series functions C→C
and Laurent series functions C → C.

Lecture 9
 19

2.2 Homomorphisms, ideals, quotient rings and

isomorphisms
Denition (Ring homomorphism). A map θ: R → S is a ring homomorphism if

(i) θ(r1 + r2 ) = θ(r1 ) + θ(r2 ),

(ii) θ(r1 r2 ) = θ(r1 )θ(r2 ),

(iii) θ(1R ) = 1S .

A bijective homomorphism is called an isomorphism. The kernel of θ is ker θ = {r ∈ R :

θ(r) = 0}.

Lemma 2.1. θ is injective if and only if ker θ = {0}.

Proof. θ is a homomorphism of the additive groups.

Denition (Ideal). A subset I of R is an ideal, written I / R, if

(i) I is a subgroup of R under addition,

(ii) whenever a∈I and r∈R then ar ∈ I .

The second condition is called the strong closure property.

Warning. An ideal in general is not a subring. If 1∈I then r = 1r ∈ I for all r∈R
and so I=R if 1 ∈ I. So the only ideal which is a subring of R is R itself.

Lemma 2.2. The kernel of a ring homomorphism θ: R → S is an ideal.

Proof. θ is a homomorphism of the additive groups (R, +) → (S, +) and so ker θ is an

additive subgroup of R.
If a ∈ ker θ then θ(a) = 0. Consider

θ(ar) = θ(a)θ(r) = 0.θ(r) = 0.

Thus ar ∈ ker θ, so we have the strong closure property.

Example. (i) In a eld F, the only ideals are {0} and F. (Multiply any non-trivial
element in the ideal by its inverse to see that 1 is in the ideal.)

(ii) In Z the ideals are of the form

nZ = {. . . , −2n, −n, 0, n, 2n, . . . }.

Proof. Certainly each set nZ is an ideal of Z.

Suppose I is a non-zero ideal. We show that all the non-zero additive subgroups
of Z nZ. Let n be the least positive element of I and use Euclid's
are of the form
algorithm. b ∈ I then b = nq + r with 0 ≤ r < n and some q . Note that
If
r = b − nq ∈ I . Minimality of n implies that r = 0 and hence b = nq . Thus
I = nZ.

Lecture 9
20 Rings

Denition. Let R be a ring and a ∈ R. Then the ideal generated by a is aR = {ar :

r ∈ R}. We often use the notation (a) for aR.

Note that this is the smallest ideal of R containing a. Such an ideal is called a principal
ideal.
More generally, the ideal generated by a1 , . . . , ak is

(a1 , . . . , ak ) = a1 R + . . . ak R
X k 
= ai ri : ri ∈ R .
i=1

Even more generally, the ideal generated by a subset A of R is

X 
(A) = ara : only nitely many ra ∈ R are non-zero .
a∈A

Examples of principal ideals include nZ in Z and (X) in C[X], where (X) is the ideal
of polynomials with zero constant term.

Proposition 2.3. Let I / R be an ideal. Then the quotient ring R/I has elements
consisting of the cosets r + I with the operations

(r1 + I) + (r2 + I) = (r1 + r2 ) + I,

(r1 + I).(r2 + I) = r1 r2 + I.

This denes a ring.

Proof. From Proposition 1.3 we have already checked that R/I forms a group under
addition as dened. So we need to check that multiplication is well-dened. If

r1 + I = r10 + I, r2 + I = r20 + I

then
r10 = r1 + a1 , r20 = r2 + a2
for some a1 , a2 ∈ I , and therefore

r10 r20 = (r1 + a1 )(r2 + a2 )

= r1 r2 + (a1 r2 + r1 a2 + a1 a2 )

where the second term is contained in I by the strong multiplicative closure property.
Thus r10 r20 + I = r1 r2 + I .
The multiplicative identity is 1+I since (1 + I)(r + I) = r + I .
Associativity, closure under multiplication and distributivity follow from the respective
properties in the ring R. (Check.)

Example. (i) nZ is an ideal of Z. Elements m + nZ of Z/nZ may be expressed as

one of
0 + nZ, 1 + nZ, . . . , (n − 1) + nZ.
Addition and multiplication correspond to arithmetic modulo n.

Lecture 9
 21

I = (X) / C[X]. Elements f (X) + I of the quotient ring C[X]/I

(ii) Consider the ideal
may be expressed in the form a + I where a ∈ C is the constant term of f (X).
Addition and multiplication correspond to that in C,

(a + I) + (b + I) = (a + b) + I,
(a + I)(b + I) = ab + I.
Therefore, C[X]/I ∼
= C.
Proposition 2.4 (Euclid's algorithm for polynomials) . Let F be a eld and consider
polynomials f (X), g(X) ∈ F[X]. Then we can write

f (X) = g(X)q(X) + r(X)

with deg r < deg g .

Proof. Write
n
X m
X
i
f (X) = ai X , g(X) = bi X i
i=0 i=0
with deg f (X) = n and deg g(X) = m.
If n < m then set q(X) = 0 and r(X) = f (X). Now assume n ≥ m, and argue by
induction on n. Set f1 (X) = f (X) − an b−1
m X
n−m g(X) so deg f (X) < deg f (X).
1 If
m=n then deg f1 (X) < m and we can apply our rst case to f1 (X),
f (X) = an b−1
m g(X) + f1 (X).

If n>m then we have by induction

f1 (X) = g(X)q1 (X) + r1 (X)

for suitable q1 (X) and r1 (X), so

f (X) = g(X)(q1 (X) + an b−1

m X
n−m
) + r1 (X).
Note 2. We require F to be a eld as we need b−1
m .

Example. Consider the ideal

I = (X 2 + 1) / R[X].
For any f (X) ∈ R[X], we can write f (X) = (X 2 + 1)q(X) + r(X) with deg r(X) ≤ 1.
Thus f (X) + I = r(X) + I . The elements of R[X]/I are of the form a + bX + I .

Addition takes the form

(a1 + b1 X + I) + (a2 + b2 X + I) = (a1 + a2 ) + (b1 + b2 )X + I,

and similarly multiplication is given by

(a1 + b1 X + I)(a2 + b2 X + I) = a1 a2 + (b1 a2 + a1 b2 )X + b1 b2 X 2 + I

= (a1 a2 − b1 b2 ) + (b1 a2 + a1 b2 )X + I.
This corresponds to addition and multiplication in C. In fact,

R[X]/(X 2 + 1) ∼
= C,
a + bX + I 7→ a + bi.

Lecture 10
22 Rings

Theorem 2.5 (First isomorphism theorem) . Let θ: R → S be a ring homomorphism.

Then ker θ / R and R/ ker θ ∼
= Im θ ≤ S .
Example. The map
X X
θ : R[X] → C, aj X j 7→ aj ij

is a ring homomorphism with ker θ = (X 2 + 1) and Im θ = C.

Proof. Lemma 2.2 states that ker θ is an ideal. Im θ is a subring of S,

(i) θ is a homomorphism of additive groups and so Lemma 1.4 implies that Im θ is an

additive subgroup of S.

(ii) θ(r1 )θ(r2 ) = θ(r1 r2 ) ∈ Im θ, so we have closure under multiplication. Associativity

is inherited from S . Moreover, θ(1R ) = 1S .

Let Φ : R/I → Im θ, r + I 7→ θ(r) and note I = ker θ. We know that Φ is well-dened

and bijective and an isomorphism of additive groups from proof of Lemma 1.4. It is left
to check that Φ is a ring homomorphism.

Φ((r1 + I)(r2 + I)) = Φ(r1 r2 + I)

= θ(r1 r2 )
= θ(r1 )θ(r2 )
= Φ(r1 + I)Φ(r2 + I),
Φ(1R + I) = θ(1R ) = 1S .

Thus Φ is a ring homomorphism.

Theorem 2.6 (Second isomorphism theorem) . Let R be a subring of S and J / S an

ideal. R ∩ J / S is an ideal
Then and {r + J : r ∈ R} = (R + J)/J ≤ S/J and
∼
R/(R ∩ J) = (R + J)/J ≤ S/J .

Proof. Let
θ : R → S/J, r 7→ r + J.
This is a ring homomorphism. (Check.) We have that

ker θ = {r ∈ R : r + J = J} = R ∩ J / R,
Im θ = {r + J : r ∈ R} = (R + J)/J ≤ S/J.

Theorem 2.5 implies that R/ ker θ ∼

= Im θ, that is, R/(R ∩ J) ∼
= (R + J)/J ≤ S/J .

There is a correspondence between additive groups,

   
additive subgroups of R additive subgroups
←→
containing ideal I of R/I

as in Chapter 1. This correspondence induces

   
subrings of R subrings
←→ ,
containing I of R/I

Lecture 10
 23
   
ideals ofR ideals
←→ .
containing I of R/I

(This needs to be checked.)

Theorem 2.7 (Third isomorphism theorem) . I and J

Let be ideals of R, with I ⊂ J.
Then (R/I)/(J/I) ∼
= R/J where J/I = {r + I : r ∈ J}.

Proof. Let θ be the map

θ : R/I → R/J, r + I 7→ r + J.

This is well-dened as in Theorem 1.7. One can check it is a ring homomorphism and
θ(1 + I) = 1 + J .
The rst isomorphism theorem implies that R/ ker θ ∼
= Im θ ≤ R/J . Noting that

ker θ = {r + I : r ∈ J} = J/I / R/I,

Im θ = R/J,

we get the required result.

Example. Given a ring R, there is a unique ring homomorphism

φ: Z → R
1 7→ 1R
m 7→ 1| + ·{z
· · + 1}
m times
The rst isomorphism theorem implies Z/ ker φ ∼
= Im φ ≤ R. Im φ is the prime subring
of R. ker φ / Z is an ideal and so is of the form nZ for some n.

This n is the characteristic of R. If R is one of Z, Q, R, or C then the characteristic is

0. If R = Z/(pZ) then the chacteristic is p.

2.3 Integral domains, elds of fractions, maximal ideals

and prime ideals
Denition (Integral domain) . A ring is an integral domain if ab = 0 implies a=0 or
b = 0 for all a, b ∈ R. A zero divisor a is non-zero and there exists b 6= 0 such that
ab = 0. (In an integral domain there are no zero divisors.)

Example. Z is an integral domain. All elds are integral domains. Subrings of integral
domains are integral domains, e.g. Z[i] ≤ C.
Denition (Principal ideal domain). A principal ideal domain is an integral domain in
which all ideals are principal ideals, i.e. of the form (a) = aR for some a ∈ R.
Example. Z is a principal ideal domain. We will see that any integral domain where
Euclid's algorithm applies is a principal ideal domain.

Lecture 11
24 Rings

Lemma 2.8. Let R be a nite integral domain. Then R is a eld.

Proof. Consider the map R → R, r 7→ ar, that is, multiplication by a 6= 0.

Since R is an integral domain this map is injective. (If ar1 = ar2 then a(r1 − r2 ) = 0
and so r1 − r2 = 0, i.e. r1 = r2 . Note that cancellation is valid in integral domains.)
Since R is nite the map is also surjective.
Thus there exists r ∈R with ar = 1. Thus a has a multiplicative inverse. So R is a
eld.

Lemma 2.9. If R is an integral domain then R[X] is an integral domain.

Proof. If
X X
f (X) = ai X i , g(X) = bi X i
with deg f = m and deg g = n then f (X)g(X) has degree m+n since am bn 6= 0 as R is
an integral domain. Therefore,

f (X) 6= 0, g(X) 6= 0 =⇒ f (X)g(X) 6= 0.

Thus R[X] is an integral domain.

Remark. By induction, R[X1 , . . . , Xn ] polynomial ring in indeterminates X1 , . . . , Xn is

an integral domain if R is. This is because R[X1 , X2 ] may be regarded as (R[X1 ])[X2 ].
Theorem 2.10. Let R be an integral domain. Then there is a eld of fractions F with
the properties

(i) R≤F is a subring,

(ii) F is a eld,

(iii) every element of F has the form ab−1 where a∈R and b−1 is the multiplicative
inverse in F of b ∈ R.
Example. With the notation as above, R = Z, F = Q.

Proof. Consider pairs (a, b) with a ∈ R, b ∈ R and b 6= 0. Dene an equivalence relation

(a, b) ∼ (c, d) ⇐⇒ ad = bc.

(For transitivity, note that (a, b) ∼ (c, d) ⇐⇒ ad = bc, (c, d) ∼ (e, f ) ⇐⇒ cf = de so

adf = bcf = bde as cancellation is valid in R. So af = be and (a, b) ∼ (e, f ).)
Write a/b for the equivalence class of (a, b). We dene addition by

a c ad + bc
+ =
b d bd
and multiplication by

a c ac
· = .
b d bd
One can easily check that these operations are well-dened.

Lecture 11
 25

If we let F = {a/b : a ∈ R, b ∈ R, b 6= 0} then F is a ring under these operations.

R may be identied with the subring of elements of the form r/1. The multiplicative
identity is 1/1. a/b has multiplicative inverse b/a if a 6= 0,

ab ab 1
= = .
ba ab 1

Every element a/b is of the form (a/1)(b/1)−1 since (b/1)−1 = 1/b.

Lemma 2.11. A non-zero ring R is a eld if and only if its only ideals are {0} and R.

Proof. Suppose that R is a eld and take an ideal I / R. If I 6= {0} then pick a 6= 0 ∈ I .
Then 1 = aa−1 ∈ I . Hence I = R.
Conversely, assume that {0} and R are the only ideals. Take a ∈ R with a 6= 0. Then
(a) 6= {0} and so (a) = R. Hence 1 ∈ (a) and there exists r such that ar = 1.

Denition (Maximal ideal) . An ideal I in R is maximal in R if I 6= R but whenever

I ⊂J /R then I=J or J = R.

Lemma 2.12. For an ideal I / R, the quotient ring R/I is a eld if and only if I is
maximal in R.

Proof. R/I is a eld if and only if I/I and R/I are the only ideals in R/I . By the usual
correspondence, this is equivalent to that the only ideals in R containing I are I and R,
which is the statement that I is maximal in R.

Denition (Prime ideal). An ideal I of R is a prime ideal in R if whenever ab ∈ I then

a∈I or b ∈ I.

Lemma 2.13. For an ideal I / R, the quotient ring R/I is an integral domain if and
only if I is a prime ideal in R.

Proof. Suppose R/I is an integral domain and ab ∈ I . Then I = ab + I = (a + I)(b + I)

and so either a + I = I or b + I = I . So a ∈ I or b ∈ I .
Conversely, suppose I is prime and (a + I)(b + I) = I . Then ab + I = I and so ab ∈ I .
Primeness implies a ∈ I or b ∈ I and so a + I = I or b + I = I .

Example. In Z, the maximal ideals are pZ where p is a prime number. Prime ideals
are pZ for p prime and {0}.

Lemma 2.14. The characteristic of an integral domain is 0 or a prime number p.

Proof. Recall that we have a unique ring homomorphism φ : Z → R. The prime subring
Im φ ∼
= Z/nZ, where n is the characteristic of R. If n properly factorises as n = st then
working modulo n we have st = 0 (mod n) and so we have zero divisors.

So if R is an integral domain, its prime subring is an integral domain and so n is prime

or 0.

Lecture 12
26 Rings

2.4 Factorisation in integral domains  units, primes,

irreducibles
Throughout this section R is assumed to be an integral domain.

Denition. a ∈ R is a unit if it has a multiplicative inverse. Equivalently,

An element
(a) = R. a divides b, written a | b, if there is c ∈ R such that b = ac.
We say
Equivalently, (b) ⊂ (a). Elements a and b are associates in R if a = bc for some unit
c ∈ R. Equivalently, (a) = (b). r ∈ R is irreducible in R if it is non-zero, not a unit and
whenever r = ab with a, b ∈ R then a or b is a unit. r ∈ R is prime in R if it is non-zero,
not a unit and if r | ab then r | a or r | b.

Remark. These denitions do depend on the ring R, e.g. 2 is prime and irreducible in
Z, but not in Q. 2X is irreducible in Q[X] but not in Z[X], as 2X = 2 · X and 2, X are
not units in Z[X].

Lemma 2.15. (r) is a prime ideal of R if and only if r is prime or r = 0.

Proof. r 6= 0 and r | ab then ab ∈ (r)

If and so if (r) is a prime ideal then a ∈ (r) or
b ∈ (r). Thus r | a or r | b.
(0) is a prime ideal in an integral domain. If r is prime and ab ∈ (r) then r | ab and
hence r|a or r | b. Thus a ∈ (r) or b ∈ (r).

Lemma 2.16. If r is prime in R then r is irreducible in R.

Proof. Suppose that r is prime in R and r = ab with a, b ∈ R. Then r | ab and so r | a

or r | b. Without loss of generality suppose r | a. So a = qr for some q ∈ R. So r = qrb.
Cancellation in integral domains gives 1 = qb, so b is a unit.

Example.
√ To show that the converse of Lemma 2.16 does not hold in general, consider
R = Z[ −5] ≤ C. R is an integral domain since it is a subring of a eld. Dene a norm,

√
N (a + b −5) = a2 + 5b2 ∈ Z≥0 .

Thus N (z) = z z̄ . N N (z1 z2 ) = N (z1 )N (z2 ). The units of R

is multiplicative, that is,
are precisely the elements of norm 1, namely ±1. Suppose z is a unit and so there exists
z1 with zz1 = 1 so N (z)N (z1 ) = N (zz1 ) = N (1) = 1 hence N (z) is a unit in Z and also
N (z) ≥ 0. So N (z) = 1 is the only possibility.
There are no elements in R of norm 2 or 3. (We cannot solve a2 + 5b2 = 2 or 3.)
√ √
Consider the identity 6 = 2 · 3 = (1 + −5)(1 − −5) in R.
To see that 2 is irreducible, express 2 = z1 z 2 and consider norms, 4 = N (z1 )N (z2 ).
Since there are no elements of norm 2
N (zj ) = 1 so zj is a
one of the unit.
√ √
But 2 is not prime in R since 2 | (1 + −5)(1 − −5) but 2 does not divide either
√ √
1 ± −5. (Consider norms, N (2) = 4, N (1 ± −5) = 6 and 4 - 6.)
√ √
Similarly, 3, 1 + −5, 1 − −5 are irreducible.

Denition (Euclidean domain) . An integral domain R is a Euclidean domain (ED) if

there is a function φ : R − {0} → Z≥0 called Euclidean function such that

Lecture 12
 27

(i) φ(ab) ≥ φ(a) for all a, b ∈ R \ {0},

(ii) if a, b ∈ R with b 6= 0 then there exist q, r ∈ R with a = qb + r with either r=0

or φ(r) < φ(b). (Euclid's algorithm.)
Example. (i) Z is a Euclidean domain with φ(n) = |n|.

(ii) F[X] with F a eld is a Euclidean domain with φ(f (X)) = deg f . (Section 2.4
gives Euclid's algorithm here.)
√
(iii) R = Z[i] = {a + bi : a, b ∈ Z} ≤ C is an integral domain. As with Z[ −5] we can
2 2
dene a norm via N (z) = z z̄ for z ∈ R. Thus N (a + bi) = a + b . In this case
the norm is a Euclidean function.

N is multiplicative and so property (i) holds,

N (z1 z2 ) = N (z1 )N (z2 ) ≥ N (z1 ),

since N (z2 ) ≥ 1. To verify property (ii) take z1 , z 2 ∈ R with z2 6= 0. Consider

z1 /z2 ∈ C. Then in the complex plane it is distance less than 1 from the nearest
element of R.

z1 /z2 = q + z3 with q ∈ R, z3 ∈ C and |z3 | < 1. So z1 = qz2 + z2 z3 . Set r = z2 z3 .

Thus z1 = qz2 + r .

N (r) = |z2 z3 |2 = |z2 |2 |z3 |2 < |z2 |2 = N (z2 ).

Remark. Similar arguments apply for other subrings of C and one can sometimes show
that the norm is a Euclidean function but one needs that any point in the complex plane
√
is distance less than 1 from a lattice point of R. Note for Z[ −5] this is not true.

Proposition 2.17. If an integral domain R is an ED then R is a PID.

Proof. Let R Φ : R \ {0} → Z≥0 = {n ∈ Z : n ≥ 0}.

be an ED with Euclidean function
Let I / R be an ideal and suppose that I is non-zero. Pick b ∈ I \ {0} with Φ(b) minimal.
Then take a ∈ I and use Euclid's algorithm: a = bq + r with r = 0 or Φ(r) < Φ(b).
Note that r = a − bq ∈ R, and hence minimality of Φ(b) implies r = 0. Thus a = bq , so
I = (b).

Example. Z, F[X] for a eld F and Z[i] are all PIDs. But
Z[X] is not a PID.
Xi
P

To see this, consider the ideal I = (2, X) = ai : a0 is even / Z[X]. Suppose that
I = (f (X)) for some f (X) ∈ Z[X], in particular 2 = f (X)g(X) for some g(X) ∈ Z[X].
We deduce that f (X) has to be a constant polynomial, that is, of degree 0, and must
be ±1 or ±2.

Also X = f (X)h(X) for some h(X) ∈ Z[X]. So f (X) cannot be ±2. So f (X) = ±1.
Hence (f (X)) = Z[X]. This is a contradiction, since I 6= Z[X], e.g. 1 6∈ I .
Example (Minimal polynomials of matrices with entries in a eld F). Let ∆ ∈ Mn×n (F).
Consider I = {f (X) ∈ F[X] : f (∆) = 0}.
We can check that I / F[X] is an ideal. But F[X] is a PID, so I = (m(X)), an by
multiplyingm(X) by a unit in F we may assume m(X) is monic. This is the minimal
polynomial of ∆. (See the course Linear Algebra where Euclid's algorithm is explicitly
used.)

Lecture 13
28 Rings

Denition (Unique factorisation domain). An integral domain is a unique factorisation

domain (UFD) if

(i) every non-zero element that is not a unit may be expressed as a product of nitely
many irreducibles,

(ii) whenever p1 · · · pm = q1 · · · qn for products of irreducibles then m = n and we can

reorder so that pi is an associate of qi . (So factorisation is unique up to ordering
and associates).

Our goal is the following proposition.

Proposition 2.18. If an integral domain R is a PID then R is a UFD.

Corollary 2.19. Z, F[X] for a eld F and Z[i] are all UFDs.
√
Example. Z[ −5] is not a UFD. Note that
√ √
6 = 2 · 3 = (1 + −5)(1 − −5)
√ √ √
where 2, 3, 1 + −5 and 1 − −5 are irreducibles. Units are ±1 in Z[ −5], so these
irreducibles are not associates of each other.

We need two lemmas about PIDs to prove Proposition 2.18.

Lemma 2.20. Let R be a PID. If p is an irreducible then it is prime. (This is the

converse of Lemma 2.16 for PIDs.)

Proof. Let p be an irreducible and suppose p | ab. Suppose p - a. The ideal (p, a) must
be principle in a PID, (p, a) = (d), say.

Thus p = q1 d and a = q2 d for some q1 , q2 ∈ R. p is irreducible so either q1 or d is a unit.

But if q1 is a unit, then d = q1−1 p and a = q2 d = q2 q1−1 p, so p | a, a contradiction. Hence
d is a unit. Thus (d) = R and there exist r, s ∈ R such that

1 = rp + sa

and so

b = rpb + sab.

Therefore, p | b.

Remark. In fact, d is a highest common factor of p and a. Highest common factors are
unique up to units, see Example Sheet 3.

Lemma 2.21 (Ascending chain condition, ACC) . Let R be a PID and Ij / R with
I1 ⊂ I2 ⊂ I3 ⊂ · · · . Then, for some n ∈ N, In = In+i for all i ≥ 0.

Proof.
S
I = j Ij / R is an ideal, check. (For example, if a ∈ Ij and b ∈ Ik ,
The union
j ≤ k then a ∈ Ik , B ∈ Ik , so a + b ∈ Ik ⊂ I .) R is a PID, so I = (a) for some a ∈ I .
We must have a ∈ In for some n ∈ N. For all i ≥ 0,

(a) ⊃ I ⊃ In+i ⊃ In ⊃ (a),

so we have equality throughout and in particular In+i = In .

Lecture 13
 29

Remark. Rings satisfying the ACC are called Noetherian.

Proof of (2.18). (i) Let a ∈ R, non-zero and not a unit. Assume it cannot be fac-
torised as a product of nitely many irreducibles. So in particular a itself is not
irreducible, hence a = a1 b1 with a1 , b1 non-zero and not units. We may assume
that a1 cannot be factorised as a product of nitely many irreducibles, so write
a1 = a2 b2 and continue. (If both a1 , b1 can be factorised into nitely many irre-
ducibles, then so can a.)

We obtain (a1 ) ( (a2 ) ( (a3 ) ( · · · with inequality at each stage, for if (ai ) =
(ai+1 ) then ai and ai+1 would be associates and bi+1 a unit.

(ii) Suppose we have p1 · · · pm = q1 · · · qn with products of irreducibles. Then p1 is

prime by Lemma 2.20 and p1 | q1 · · · qn , so p1 divides one of the qi , p1 | q1 , say,
and so q1 = p1 u. But q1 is irreducible and p1 is not a unit, hence u is a unit. This
is to say p1 and q1 are associates. We have

p1 · · · pm = up1 q2 · · · qn .

Cancellation in R gives

p2 · · · pm = (uq2 )q3 · · · qn .

Note that (uq2 ) is irreducible. Continuing in this way gives the result.

Remark. The proof of (ii) depends just on Lemma 2.20, that irreducibles are prime,
and (i) depends on the ACC.

There are several properties which follow quickly from the denition of a UFD.

(i) An element p∈R is irreducible if and only if it is prime.

We have already shown the `only if ' direction for integral domains.

Suppose p is an irreducible and p | ab. We can express a and b as

a = p1 · · · p m , b = q1 · · · qn

with pi , q i irreducible. Now write ab = pc with c = r 1 · · · rs where ri is irreducible.

Then
(p1 · · · pm )(q1 · · · qn ) = pr1 · · · rs
and uniqueness of factorisation implies that p is an associate of either some pi or
some qi . Hence p|a or p | b.

(ii) Highest common factors exist.

(iii) Lowest common multiples exist.

Denition (Highest common factor). d is a highest common factor of a1 , . . . , an , written

hcf(a1 , . . . , an ), if d | ai for each i and whenever d0 also divides each ai then d0 | d.
Note that one often refers to the highest common factor, although certainly multiplying
by a unit will give another highest common factor.

Lecture 14
30 Rings

Expressing each ai as a product of irreducibles, each ai is of the form

n
Y
ui pj ij
j

where pj is irreducible and pj is not associate to pk whenever j 6= k . ui is a unit in R

and nij ≥ 0.
Q mj
The claim is that j pj is a highest common factor of a1 , . . . , an where mj = mini {nij }.
0
It is clear that this is a factor of each ai , and if d | ai for each i then each irreducible
0
dividing d is also an irreducible dividing ai and so must be one of the pj . Thus
Y tj
d0 = u pj
j

where u is a unit. But the tj can be at most mini {nij } if d0 | ai for each i. Thus d0 | d.

2.5 Factorisations in polynomial rings, Gauss' lemma and

Eisenstein's criterion
IfF is a eld then F[X] is a ED, PID and UFD. Every ideal I / F[X] is principal,
I = (f (X)) for some f (X) ∈ F[X]. An element is irreducible if and only if it is prime.
F[X]/I is a eld, with elements of the form r(X) + I with r(X) = 0 or deg r < deg f , if
and only if I is maximal if and only if f (X) is irreducible in F[X].

Gauss' lemma helps to determine when polynomials are irreducible in F[X].

Denition (Content). Assume that the coecient ring R is a UFD. Let f (X) = a0 +
· · · + an X n with an 6= 0, deg f = n. The content c(f (X)) is the highest common factor
of a0 , . . . , an . f (X) is primitive if c(f (X)) is a unit, i.e. if the ai are coprime.

Our aim is to prove the following lemma.

Lemma 2.22 (Gauss' lemma) . Let R be a UFD with F its eld of fractions. Suppose
f (X) ∈ R[X] is primitive. Then f (X) is irreducible in R[X] if and only if f (X) is
irreducible in F[X].

In particular, when R = Z then f (X) is irreducible in Z[X] if and only if it is irreducible

in Q[X].
Example. X 3 + X + 1 is irreducible in Z[X], and is primitive in Z[X] and so Gauss'
lemma
3 3
implies X + X + 1 is irreducible in Q[X], and so Q[X]/(X + X + 1) is a eld.

Suppose X3 + X + 1 Z[X], so X 3 + X + 1 = g(X)h(X) for g(X), h(X) ∈

is reducible in
Z[X] not units. Primitivity implies that g(X) and h(X) are not constant polynomials,
so deg g, deg h ≥ 1. So one of g(X) and h(X) is of degree 1,

g(X) = b0 + b1 X,
h(X) = c0 + c1 X + c2 X 2 ,

say. Considering coecients of X0 and X 3,

b0 c0 = 1,
b1 c2 = 1.
So b0 = ±1, b1 = ±1. This is a contradiction since ±1 is not a root of X 3 + X + 1.

Lecture 14
 31

Lemma 2.23. If f (X) and g(X) are primitive in R[X] then so is f (X)g(X).

Proof. Let

f (X) = a0 + a1 X + · · · + am X m ,
g(X) = b0 + b1 X + · · · + bn X n

be primitive. Suppose the product f (X)g(X) is not primitive. So there is a prime p∈R
dividing c(f (X)g(X)) but p - c(f (X)) and p - c(f (X)).
Let k and l be such that

p | a0 , p | a1 , . . . , p | ak−1 , p - ak ,
p | b0 , p | b1 , . . . , p | bl−1 , p - bl .

The coecient ck+l of X k+l in f (X)g(X) is

· · · + ak+1 bl−1 + ak bl + ak−1 bl+1 + · · · .

p divides all of these terms apart from ak bl which it does not divide. So p - ck+l , contra-
dicting that p | c(f (X)g(X)). Thus f (X)g(X) is primitive.

Consider the polynomial ring R[X] and let F be the eld of fractions of R.

Corollary 2.24. For f (X), g(X) ∈ R[X], the content c(f (X)g(X)) is an associate
of c(f (X))c(g(X)). (Recall that highest common factors and hence contents are only
dened up to associates.)

Proof. As R is a UFD we may write

f (X) = c(f (X))f1 (X),

g(X) = c(g(X))g1 (X)

where f1 (X), g1 (X) are primitive. Then

f (X)g(X) = c(f (X))c(g(X))f1 (X)g1 (X)

where f1 (X)g1 (X) is primitive by Lemma 2.23. So a highest common factor of the
coecients of f (X)g(X) is c(f (X))c(g(X)).

Remark. An irreducible f (X) in R[X] is either in R (and irreducible in R) or it is

primitive in R[X].
To see this, assume that f (X) is irreducible and write f (X) = c(f (X))f1 (X) with f1 (X)
primitive. If f1 (X) 6∈ R then, since a polynomial of degree at least 1 cannot be a unit,
c(f (X)) must be a unit.

Proof of Lemma 2.22, Gauss' lemma. Take f (X) primitive in R[X]. Suppose it fac-
torises in R[X] as a product of non-units in R[X]. Since f (X) is primitive, neither of
these non-units is in R, the factors have degrees greater than 0. So f (X) factorises as a
product of non-units in F[X].

Lecture 15
32 Rings

Conversely, suppose that f (X) = g(X)h(X) with g(X), h(X) ∈ F[X] non-units and
hence not constant. Multiply by a ∈ R and b ∈ R repectively such that ag(X), bh(X) ∈
R[X]. Thus

abf (X) = (ag(X))(bh(X)).

Write

ag(X) = c(ag(X))g1 (X),

bh(X) = c(bh(X))h1 (X)

where g1 (X), h1 (X) are primitive in R[X]. Note that g1 (X), h1 (X) are not constant and
hence not units. Now Lemma 2.24 implies that ab is an associate of c(ag(X))c(bh(X))
by considering contents. So

abf (X) = uc(ag(X))c(bh(X))u−1 g1 (X)h1 (X)

where ab = uc(ag(X))c(bh(X)) and u a unit in R. Cancellation gives

f (X) = (u−1 g1 (X))h1 (X)

and thus f (X) factorises as a product of non-units in R[X].

Remark. A similar argument shows that if f (X) ∈ R[X], not necessarily primitive, and
f (X) = g1 (X)h(X) with g1 (X) primitive in R[X], h(X) ∈ F[X] then we can deduce
that in fact
f (X) = g1 (X)h0 (X)
with h0 (X) ∈ R[X].
We can see this as follows. There exists b ∈ R with bh(X) ∈ R[X]. Consider bf (X) =
g1 (X)(bh(X)). Write bh(X) = c(bh(X))h1 (X) with h1 (X) primitive. Consider contents,
b | c(g1 (X)bh(X)) and so b | c(bh(X)) since g1 (X) is primitive. Cancellation as in the
previous result gives f (X) = g1 (X)h0 (X) for some h0 (X) ∈ R[X].

Thus if

I = g1 (X)F[X] / F[X]

and

J = g1 (X)R[X] / R[X]

then I ∩ R[X] = J , with g1 (X) primitive in R[X].

Theorem 2.25. If R is a UFD then R[X] is a UFD.

Proof. Let f (X) ∈ R[X] be non-zero and a non-unit. Write f (X) = c(f (X))f1 (X) with
f1 (X) primitive. (The rst step is to show that we need only look at primitives.)

Observe that since R is a UFD, c(f (X)) is expressible as a product of irreducibles in R

in an essentially unique way. These irreducibles are irreducible in R[X].
If f (X) factorises as a product of irreducibles in R[X], then we can collect together
the irreducibles in R and the primitive irreducibles. The product of these primitive

Lecture 15
 33

irreducibles is primitive by Lemma 2.23 and so the content of f (X) is an associate of the
product of the irreducibles in R. The essentially unique factorisation of c(f (X)) means
that this product is essentially the same as the previous one.

Cancellation implies that the product of the primitives is an associate of f1 (X). Thus
we may assume that f (X) is primitive.

F[X] is a UFD. So we can factorise f (X) in F[X],

f (X) = p1 (X) · · · pk (X),

where pi (X) is irreducible in F[X].

There exists ai ∈ R with ai pi (X) ∈ R[X]. We have

ai pi (X) = ci qi (X)

with ci = c(ai pi (X)), qi (X) primitive in R[X] (and irreducible in R[X] by Gauss'
lemma). So

a1 · · · ak f (X) = c1 · · · ck q1 (X) · · · qk (X).

Considering contents and using the assumption that f (X) is primitive, we have

a1 · · · ak = uc1 · · · ck

for some unit u ∈ R since qi (X) is primitive by Lemma 2.23, i.e. q1 (X) · · · qk (X) is
primitive.

a1 · · · ak f (X) = (uc1 · · · ck )(u−1 q1 (X))q2 (X) · · · qk (X).

Cancellation gives

f (X) = (u−1 q1 (X))q2 (X) · · · qk (X),

a product of irreducibles in R[X]. This shows the existence part. For the uniqueness of
the factorisation assume that

f (X) = q1 (X) · · · qk (X)

= r1 (X) · · · rl (X)

with qi (X), ri (X) irreducible in R[X]. We are assuming that f (X) is primitive and so
each ri (X) must be primitive.
By Gauss' lemma, ri (X) is irreducible in F[X]. Uniqueness of factorisation in F[X]
implies that k = l, and after reordering we have qi (X) = ui ri (X) with ui a unit in F[X],
hence in F.
We can write ui = ai /bi for some ai , bi ∈ R with bi 6= 0. Then bi qi (X) = ai ri (X). But
qi (X) and ri (X) are primitive and hence bi and ai must be associates being the content
of bi qi (X) = ai ri (X). Cancelling gives that qi (X) and ri (X) are associates in R[X].
Thus factorisation is essentially unique in R[X].

Corollary 2.26. Let R be a UFD. Then R[X1 , . . . , Xn ] is a UFD.

Lecture 16
34 Rings

Proof. R is a UFD and repeated application of

We assume that Lemma 2.23 gives that
R[X1 ], R[X1 , X2 ] = (R[X1 ])[X2 ], . . . , R[X1 , . . . , Xn ] is a UFD.

Proposition 2.27 (Eisenstein's criterion). Let R be a UFD and let f (X) = a0 + a1 X +

· · · + an X n ∈ R[X], an 6= 0, be primitive. Assume that for some irreducible p we have
p - an , p | ai for 0 ≤ i < n and p2 - a0 . Then f (X) is irreducible in R[X], and hence in
F[X] by Gauss' lemma.

Proof. Suppose that f (X) = g(X)h(X) with

g(X) = rk X k + · · · + r0 ,
h(X) = sl X l + · · · + s0

where rk , sl 6= 0. Note that k + l = n. Since p | a0 = r0 s0 and p2 - a0 we may assume

that p | r0 and p - s0 . p - an = rk sl implies p - rk , p - sl .
Set j to be such that p | r0 , . . . , p | rj−1 , p - rj . Consider the term

aj = r0 sj + · · · + rj s0 .

Note p - aj . Thus j = n, and hence k = n, l = 0. As f (X) is primitive, the constant

polynomial h(X) must be a unit. Thus f (X) cannot be factorised as a product of
non-units.

Example. Consider R=Z and let f (X) = X n − p ∈ Z[X], where p is prime.

By Eisenstein's criterion, f (X) is irreducible in Z[X] and hence in Q[X]. So f (X) does
not have any roots in Q, as a root would yield a linear factor of f (X) in Q[X]. Thus p
has no nth roots in Q.

Example. Consider the cyclotomic polynomial

f (X) = X p−1 + X p−2 + · · · + 1

where p is prime. The claim is that this is irreducible in Z[X], and hence in Q[X].
Note that (X − 1)f (X) = X p − 1. We make the substitution X =1+Y and get

Y f (1 + Y ) = (1 + Y )p − 1,
   
p−1 p p−2 p
f (1 + Y ) = Y + Y + ··· +
1 p−1
p p



where
i
are binomial coecients. Eisenstein's criterion does now apply, we have p| i ,
p
p2 - p−1 = p. We deduce that f (1 + Y ) is irreducible and hence f (X) is irreducible in
Z[X], and hence in Q[X].

2.6 Gaussian integers

We consider the set of Gaussian integers as introduced before,

Z[i] = {a + ib : a, b ∈ Z}.

Lecture 16
 35

The norm

N (a + ib) = a2 + b2 = (a + ib)(a − ib) = z z̄,

where z = a + ib, is multiplicative, that is,

N (z1 z2 ) = N (z1 )N (z2 ).

The units have to be of norm 1, hence the only units are ±1, ±i.
Recall from an earlier example that Z[i] is a Euclidean domain and hence a PID and a
UFD. Thus the irreducibles and primes are the same, by Lemma 2.16 and Lemma 2.20.
What are they?

• 2 = (1 + i)(1 − i).

• 3, N (3) = 9. If it were to factorise as a product of two non-units then they would

have norm 3. But there are no elements of norm 3.

• 5 = (1 + 2i)(1 − 2i).

Lemma 2.28. A prime number p in Z is irreducible (and prime) in Z[i] if and only if
p is not of the form x2 + y 2 with x, y ∈ Z − {0}.

Proof. p = x2 + y 2 then p = (x + iy)(x − iy) and so p is not irreducible in Z[i].

If
2
Otherwise consider N (p) = p , p factorises into two non-units, necessarily of norm p,
2 2
only if there is x + iy with N (x + iy) = p. So x + y = p.

Proposition 2.29. The irreducibles (and primes) in Z[i] are up to associates

(i) p∈Z prime with p ≡ 3 (mod 4),

(ii) z with z z̄ = p, for p prime in Z with p=2 or p ≡ 1 (mod 4).

Proof. Let us rst show that these are actually irreducibles (and primes).

(i) If p ≡ 3 (mod 4) then it is not of the form x2 + y 2 since squares modulo 4 are 0
2 2
or 1 and so x + y ≡ 0, 1 or 2 (mod 4). By Lemma 2.28, p is irreducible (and
prime) in Z[i].

(ii) For example, p = 2 = (1 + i)(1 − i) is of the correct form, 1±i are irreducibles of
norm 2.
Consider the nite eld of p elements {0, 1, . . . , p − 1} = Fp ∼
= Z/pZ. Its multi-
plicative group Fp \ {0} is cyclic. If not then the structure theorem for abelian
groups implies we can nd a subgroup of form Cm × Cm for some m > 1.
Thus there are at least m2 elements satisfying am = 1. These are all roots of
Xm −1 in Fp [X] which is a UFD. X
m − 1 has a unique expression up to ordering
and associates as a product of at most m irreducibles in Fp [X].

In particular, there are at most m linear factors X −a and so at most m roots,

2
contradicting that there are at least m roots.

Lecture 16
36 Rings

Fp has a cyclic multiplicative group. A cyclic group hgi of order 4n has

In this case,
n
a subgroup of order 4, namely < g >, and a unique element of order 2, namely
2n
g . p − 1 is the element of order 2 in Fp − {0}, the multiplicative group.
An element of order 4 corresponds to a ∈ Z with a2 ≡ −1 (mod p). Thus p |
a2 + 1 = (a + i)(a − i). But p - a ± i and so p is not an irreducible (and prime) in
Z[i].
Thus p must factorise, p = z1 z2 with zi non-units in Z[i]. We have N (zi ) = p.
2 2
Write zi = x ± iy . We get p = x + y .

Now assume α is irreducible in Z[i], so ᾱ is irreducible. Take p | N (α). We use that

N (α) = αᾱ, a product of irreducibles.

If p ≡ 3 (mod 4) then p prime and unique factorisation implies that p is an associate of

either α or ᾱ. So p is an associate of α.

If p = 2, or p ≡ 1 (mod 4) then p = z z̄ | αᾱ and unique factorisation, so z is an associate

of α or ᾱ and so α is an associate of z or z̄ .
Thus our list is complete.

Corollary 2.30. Let n = pn1 1 · · · pnk k ∈ Z be the prime factorisation of n, with p1 , . . . , pk

distinct primes in Z. Then n is of the form x2 + y 2 if and only if whenever pi ≡ 3
(mod 4) then ni is even.

Proof. Suppose n = x2 + y 2 = (x + iy)(x − iy) = z z̄ with z = x + iy . Thus N (z) = n.

Express z as a product of irreducibles in Z[i],

z = α1 · · · αs ,

say. But we know what the irreducibles are from Proposition 2.29. Either N (αj ) = p2j
with
Q pj ≡ 3 (mod 4) or N (αj ) = pj with pj = 2 or pj ≡ 1 (mod 4). Thus n = N (z) =
N (αj ) is of the required form.
Conversely, if n = pn1 1 · · · pnk k with nj even for pj ≡ 3 (mod 4) then we can replace any
pj = 2 or pj ≡ 1 (mod 4) by pj = αj ᾱj and any other primes pj = αj = ᾱj and so
p2j = αj ᾱj for some irreducible αj . Thus n is of the form z z̄ for some z = x + iy and so
n = x2 + y 2 .

Example. Let n = 65 = 5 × 13 and note that 5 = (2 + i)(2 − i), 13 = (2 + 3i)(2 − 3i).

The unique factorisation up to reordering and associates is

n = (2 + i)(2 − i)(2 + 3i)(2 − 3i).

We use this to express n as x2 + y 2 = z z̄ with z = x + iy

n = [(2 + i)(2 + 3i)][(2 − i)(2 − 3i)] = (1 + 8i)(1 − 8i) = 12 + 82

or

n = [(2 + i)(2 − 3i)][(2 − i)(2 + 3i)] = (7 − 4i)(7 + 4i) = 42 + 72 .

Lecture 17
 37

2.7 Z[α] with α an algebraic integer

Denition. A complex number α∈C is an algebraic integer if there is a monic poly-
nomial f (X) ∈ Z[X] with f (α) = 0.

For example,

α∈C Minimal polynomial of α

√i f (X) = X 2 + 1
2
√ f (X) = X 2 − 2
1
2 (1 + −3) f (X) = X 2 − X + 1

Z[α] is the smallest subring of C containing Z and α. Z[α] ∼

= Z[X]/I where I is the
kernel of the map θ : Z[X] → Z[α], g(X) 7→ g(α).

Proposition 2.31. For an algebraic integer α this kernel is a principal ideal generated
by a monic irreducible polynomial in Z[X].

Denition. This generator is called the minimal polynomial of α, written fα (X).

Proof. By denition, f (α) = 0 for some monic f (X) ∈ Z[X]. So we can pick fα (X) ∈ I
of minimal degree ≥ 0, and we may assume fα (X) is primitive in Z[X]. The claim is
that this is the required polynomial.

We wish to show that I = (fα (X)).

Let h(X) ∈ I . Then, as Q[X] is a Euclidean domain, we can write

h(X) = q(X)fα (X) + r(X)

with r(X) = 0 or deg r(X) < deg fα (X). Clearing denominators, there exists a ∈ Z\{0}
such that
ah(X) = aq(X)fα (X) + ar(X)
with aq(X) ∈ Z[X] and ar(X) ∈ Z[X]. But ar(α) = 0 so ar(X) ∈ I . Minimality
of degree of fα (X) implies that ar(X) = 0 so r(X) = 0. So ah(X) = aq(X)fα (X).
Consider the contents of both sides,

a | c(ah(X)),
c(aq(X)fα (X)) = c(aq(X))

since fα (X) is primitive. So a divides all the coecients of aq(X). So q(X) ∈ Z[X].
Thus h(X) ∈ (fα (X)) / Z[X].
We claim that fα (X) is prime (and irreducible).

If fα (X) = f1 (X)f2 (X) then 0 = f1 (α)f2 (α) and so fi (X) ∈ I for some i. fα (X) | fi (X)
for some i.
Finally, we show that fα (X) is monic.

Since fα (X) | f (X) where f (X) as in the rst line.

Denition (Rational integers). The elements of Z are called rational integers.

Lecture 18
38 Rings

Lemma 2.32. If α is an algebraic integer and α∈Q then α ∈ Z.

Proof. fα (X) is irreducible in Z[X] and monic. By Gauss' lemma, Lemma 2.22, fα (X)
is irreducible in Q[X]. α is a root. So fα (X) = X − α. Thus α ∈ Z.

Example. Let p∈Z be prime and consider Z[X]/(p, fα (X)). We have Fp ∼

= Z/pZ. By
the isomorphism theorems,

Fp [X]/(f¯α (X)) ∼
= Z[X]/(p, fα (X)) ∼
= Z[α]/(p)

where f¯α (X) is the polynomial in Fp [X] obtained from fα (X) by taking coecients
modulo p.

For example, α = i, fα (X) = X 2 + 1. Then

Fp /(X 2 + 1) ∼
= Z[i]/(p).

For p=2 or p ≡ 1 (mod 4) this is not an integral domain, for p ≡ 3 (mod 4) it is an

integral domain.

Remark. There is quite a lot on algebraic integers in the Part II course

√ √ Number Fields.
For example, quadratic elds which are of the form
√ Q( d) = {a√+ b d : a, b ∈ Q} ≤ C.
In Q( d) the algebraic integers form a ring
√ R. However, (1 + −3)/2 is an algebraic
integer so R is not necessarily Z( d).
R is Euclidean if and only if d is −11, −7, −3, −2, −1, 2, 3, 5, 6, 7, 11, 13, 17, 19, 21,
29, 33, 37, 41, 57 or 73 (21 possibilities).
R is a UFD for d<0 −1, −2, −3, −7, −11, −19, −43, −67, −163 (9
if and only if d is
possibilities, cf. H.M. Stark An introduction to number theory ). For d > 0 there exist 38
possibilities with 2 ≤ d < 100 so that R is a UFD. This question is still open.

2.8 Hilbert's basis theorem

Recall that we showed that a PID satises the ascending chain condition (ACC) on
ideals.

Lemma 2.33. A ring R satises the ACC if and only if all ideals in R are nitely
generated.

Proof. Suppose that all ideals are nitely generated and consider a chain

I1 ⊂ I2 ⊂ · · ·
S S
with Ij / R. Then the union j Ij /R is an ideal too. So by supposition j Ij is nitely
generated. There exists N such that all these nitely many generators lie in IN .
S
If m≥N then Im ⊂ j Ij ⊂ IN . Also IN ⊂ Im and so Im = IN .
Conversely, assume the ACC and let J / R. Take a1 ∈ J with a1 6= 0. If J 6= (a1 ) pick
a2 ∈ J \ (a1 ). If J 6= (a1 , a2 ) pick a3 ∈ J \ (a1 , a2 ) etc. We are producing a chain

(a1 ) ( (a1 , a2 ) ( (a1 , a3 , a3 ) ( · · · .

By the ACC this process stops, so J = (a1 , . . . , aN ) for some N.

Lecture 18
 39

Denition. A ring with these properties is called Noetherian.

Theorem 2.34 (Hilbert's basis theorem). If R is Noetherian then R[X] is Noetherian.

Proof. Let J / R[X] be an ideal. We aim to show that it is nitely generated.

Consider
 j
X 
i
Ij = aj ∈ R : ∃f (X) ∈ J f (X) = ai X ∈ J ∪ {0},
i=0

the set of leading coecients of polynomials of degree j in J.

Pj Pj Pj
Ij / R is an ideal since if i=0 ai X i ∈ J , i=0 bi X i ∈ J then i=0 (ai + bi )X i ∈ J and
Pj i
if a ∈ R then i=0 aai X ∈ J .

Ij ⊂ Ij+1 since if ji=0 ai X i ∈ J then X

P Pj i ∈ J.


i=0 ai X
The ACC for R implies that there exists N with Im = IN for all m ≥ N and IN is
nitely generated by the leading coecients of f1 (X), . . . , fk (X), say.
Now take any f (X) ∈ J m ≥ N . The leading coecient of f (X) lies in
of degree
Im = IN r1 , . . . , rk ∈ R so that r1 f1 (X) + · · · + rk fk (X) has the
and so there exists
same leading coecient. So f (X) − (r1 f1 (X) + · · · + rk fk (X))X
m−N ∈ J and is of

degree less than m. Repeating this process yields q1 (X), . . . , qk (X) ∈ R[X] such that
f (X) − (q1 (X)f1 (X) + · · · + qk (X)fk (X)) ∈ J of degree less than N .
Now consider polynomials in J of degree less than N . For j < N there is a nite
S set Sj
of polynomials in J of degree j whose leading coecients generate Ij . Let S = j<N Sj ,
a nite set.

A similar argument to the one before shows that any polynomial in

P J of degree less than
N is of the form gi (X)hi (X) for gi (X) ∈ S , hi (X) ∈ R[X].
Thus J is generated by S ∪ {f1 (X), . . . , fk (X)}.

Corollary 2.35. If F is a eld then F[X1 , . . . , Xn ] is Noetherian and Z[X1 , . . . , Xn ] is

also Noetherian.

Corollary 2.36. Any ring image of Z[X1 , . . . , Xn ] is Noetherian.

Proof. Let θ : Z[X1 , . . . , Xn ] → S be a homomorphism. Then θ is surjective. If I /S is

an ideal then

J = {f (X1 , . . . , Xn ) ∈ Z[X1 , . . . , Xn ] : θ(f (X1 , . . . , Xn )) ∈ I}

is nitely generated. Then I is generated by the images under θ of a nite generating

set of J.

Lecture 18
Lecture 18
 Chapter 3
Modules
3.1 Introduction
Denition. Let R be a commutative ring. A set M is an R-module with operation +,
with (M, +) forming an abelian group, and also has a map

R × M → M, (r, m) 7→ m

satisfying

(r1 + r2 )m = r1 m + r2 m
r(m1 + m2 ) = rm1 + rm2
r1 (r2 m) = (r1 r2 )m
1m = m

for all r1 , r 2 , r ∈ R and m1 , m2 , m ∈ M .

Example. (i) Let R=F be a eld. Then all vector spaces over F are F-modules.
(ii) For any R, Rn forms an R-module,

r(r1 , . . . , rn ) = (rr1 , . . . , rrn ).

In particular, when n = 1, R itself is an R-module.

(iii) If I /R is an ideal then I is an R-module and R/I is an R-module.
(iv) Let R = Z. Z-modules are precisely the abelian groups. Given an abelian group
A with operation written as + then the module map is given by

| + ·{z
na = a · · + a} for n≥1
n times
0a = 0
(−n)a = −(an) for n≥1

(v) R = F[X], F a eld. If V is an F-vector space and α : V → V a linear map (vector

space endomorphism) then V may be regarded as a F[X]-module via

f (X).v = f (α)(v)

for v ∈V. Dierent maps α yield dierent F[X]-modules.

Lecture 19
42 Modules

(vi) If R≤S are rings then S can be regarded as an R-module via

rs = (rs)

forr ∈ R, s ∈ S . For example, let Z ≤ Z[α], α an algebraic integer. We can regard

Z[α] as a Z-module.

(vii) Let R be the prime subring of a nite eld F. Here R ∼= Fp for some prime p.
(Question 8 on Example Sheet 3 shows that F has cardinality pn for some n.) We
can view F as an Fp -vector space.

Remark. There exists exactly one eld of cardinality pn for every prime p and n ≥ 1,
up to isomorphism. For example,

Cardinality Field
4 F2 [X]/(X 2 + X + 1),
8 F2 [X]/(X 3 + X + 1)

noting that e.g. X2 + X + 1 is irreducible modulo 2. See also Question 9 on Example

Sheet 2.

Denition. A subset N of an R-module M is an R-submodule, written N ≤ M, if it is

an additive subgroup of M and rn ∈ N for all r ∈ R, n ∈ N .

Example. Ideals I /R are the R-submodules of the R-module R. In a F-vector space

the vector subspaces are the F-submodules.

Denition. If N ≤M then the quotient module M/N has elements m+N with the
operation + dened as for quotients of the additive groups,

r(m + N ) = rm + N

for r ∈ R, m ∈ M . Check that this turns M/N into an R-module.

Denition. A map θ: M → N is an R-module homomorphism if

θ(m1 + m2 ) = θ(m1 ) + θ(m2 ),

θ(rm) = rθ(m).

Example. If R = F, F a eld, then an R-module homomorphism is a linear map between

F-vector spaces.

Theorem 3.1 (First isomorphism theorem). If θ : M → N is an R-module homorphism

then ker θ is an R-submodule of M, the image of θ is a submodule of N and with

ker θ = {a ∈ M : θ(m) = 0},

Im θ = {θ(m) : m ∈ M }

we have that M/ ker θ ∼

= Im θ.

Proof. Left to the reader. The isomorphism theorem for additive subgroups shows that
M/ ker θ ∼
= Im θ for the additive groups. So we only need to check the multiplicative
properties.

Lecture 19
 43

As usual there is a 1−1 correspondence,

   
submodules of submodules of M
←→ .
M/N containing N

If N ≤L≤M are R-modules then

M/L ∼
= (M/N )/(L/N ).

Compare this with the third isomorphism theorem.

Example. In the special case of R = F a eld and vector spaces, we have quotient
spaces V /W where W ≤ V and there are isomorphism theorems for linear maps (F-
module homomorphisms).

Denition. Let m∈M then the annihilator of m is

Ann(m) = {r ∈ R : rm = 0}.

The annihilator of M is

Ann(M ) = {r ∈ R : ∀m ∈ M rm = 0}
\
= Ann(m).
m∈M

These annihilators are ideals in R since

r1 m = 0, r2 m = 0 =⇒ 0 = r1 m + r2 m = (r1 + r2 )m,
r1 m = 0 =⇒ (rr1 )m = 0.

Lemma 3.2. If M is an R-module and m∈M then

Rm ∼
= R/ Ann(m)

where Rm = {rm : r ∈ R}.

Proof. Apply Theorem 3.1 to the R-module homomorphism θ : R → M, r 7→ rm with

ker θ = Ann(m) and Im θ = Rm.

Modules of the form Rm are cyclic modules. More generally, if m1 , . . . , mk ∈ M and

M = Rm1 + · · · + Rmk = {r1 m1 + · · · + rk mk : r1 , . . . , rk ∈ R}

then M is generated by m1 , . . . , m k , and M is nitely generated.

Example. If R=Z then Z[α] is a Z-module where α is an algebraic integer. In fact, it

is a nitely generated Z-module. (Exercise.)

Lemma 3.3. Let N ≤M be R-modules. If M is nitely generated then M/N is nitely

generated.

Lecture 20
44 Modules

Proof. Suppose M = Rm1 + · · · + Rmk then M/N is generated by m1 + N, . . . , mk + N

since

m + N = r1 m1 + · · · + rk mk + N
= r1 (m1 + N ) + · · · + rk (mk + N ).

for some r1 , . . . , r k ∈ R .

Warning. N need not be nitely generated.

Example. The polynomial ring C[X1 , X2 , . . . ] in countably innitely many variables

X1 , X 2 , . . . . Consider the ideal
S I of polynomials with zero constant term. This is not
nitely generated since I= Ij where Ij = (X1 , . . . , Xj ) and hence I1 ( I2 ( I3 ( · · · .
Thus C[X1 , X2 , . . . ] is not Noetherian.

If R = C[X1 , X2 , . . . ] then the submodules are the ideals, and we have shown that there
is a submodule which is not nitely generated inside the cyclic module R.

3.2 Direct sums, free modules

Denition. If M1 , . . . , Mk are R-modules then the direct sum M1 ⊕· · ·⊕Mk has elements
(m1 , . . . , mk ) with mi ∈ Mi , addition

(m1 , . . . , mk ) + (m01 , . . . , m0k ) = (m1 + m01 , . . . , mk + m0k )

and scalar multiplication

r(m1 , . . . , mk ) = (rm1 , . . . , rmk ).

Denition. Let m1 , . . . , mk ∈ M . Then the set {m1 , . . . , mk } is independent if

r1 m1 + · · · + rk mk = 0 =⇒ r1 = · · · = rk = 0.

Denition. The subset S ⊂ M generates M freely if

(i) S generates M,

(ii) every map ψ : S → N, where N is an R-module, extends to an R-module homo-

morphism θ: M → N.
Remark. If such a θ exists then it is unique since if we have two such θ1 and θ2 then
S ⊂ ker(θ1 − θ2 ) and so M ≤ ker(θ1 − θ2 ) since S generates M. Thus θ1 = θ2 .
Denition. A module freely generated by some subset S is free and S is a basis.
Proposition 3.4. For S = {m1 , . . . , mk } ⊂ M the following are equivalent:

(i) S generates M freely.

(ii) S generates M and is an independent set.

(iii) Every element of M is uniquely expressible in the form r1 m1 + · · · + rk mk for some

ri ∈ R .

Lecture 20
 45

Proof. [(i) =⇒ (ii)] Suppose that S generates M freely. We need to show independence.
Let N= Rk and dene

ψ : S → N, mj 7→ (0, . . . , 1, . . . ),

i.e. a 1 in the j th place. So there is an R-module homomorphism θ: M → N. Then

θ(r1 m1 + · · · + rk mk ) = (r1 , . . . , rk ).

So if r1 m1 + · · · + rk mk = 0 then r1 = · · · = rk = 0.
The two implications (ii) =⇒ (iii) and (iii) =⇒ (i) are left as an exercise.

Example. For R = Z, M = Z, the set {2, 3} is a generating set, but contains no basis
for M. For example, {2} is independent, but is not a generating set.

Proposition 3.5. Let M be a R-module, generated by {m1 , . . . , mk }. Then there is a

free module Rk and a surjective homomorphism θ : Rk → M . Thus M ∼ = Rk / ker θ.

Proof. Dene an R-module homomorphism

θ : Rk → M, (r1 , . . . , rk ) 7→ r1 m1 + · · · + rk mk .

The kernel in Proposition 3.5 is the relation module.

Denition. R-module M is nitely presented if there exists a nite generating
An
set S = {m1 , . . . , mk } ⊂ M and the relation module is also nitely generated, by
{n1 , . . . , nl }. We say that M is generated by S subject to relations r1 m1 + · · · + rk mk = 0
for each ni = (ri1 , . . . , rik ).

Proposition 3.6. Let R be a non-zero ring, M be an R-module freely generated by

{u1 , . . . , um } and {v1 , . . . , vn }. Then m = n.

Proof. One can dene determinants for square matrices with coecients just as in Linear
Algebra by
X
det A = sgn(π)A1,π(1) · · · An,π(n) ,
π∈Sn
det AB = det A det B,
A adj(A) = (det A)I.

Thus A is invertible if and only if det A is a unit in R. Assume that n≥m with unique
expressions
X X
vj = Aij ui , uk = Bjk vj .
i j

So
XX
uk = Bjk Aij ui
j i
XX
= Aij Bjk ui
i j

Lecture 20
46 Modules

X
= (AB)ik ui .
i

As {u1 , . . . , um } is independent we have AB = I . If n > m then

 

B
A 0 = Im
0


so det A 0 6= 0. Contradiction. So m = n.

3.3 Matrices over Euclidean domains, equivalence of

matrices, Smith normal form
In this section we will assume that R is a Euclidean domain with Euclidean function
φ : R − {0} → Z≥0 .
If a, b ∈ R then there exists a hcf(a, b), dened up to associates, obtained by Euclid's
algorithm (cf. the Part IA course Numbers and Sets, Example Sheet Question 2). More-
over, hcf(a, b) = ax + by for some x, y ∈ R.
Denition. The elementary row operations on a m×n matrix A are:

(ER1) Add c times ith row to j th row. (Achieved by multiplying A on the left by I + C
where Ckl = 0 except for Cij = c.)
(ER2) Interchange rows i and j , i 6= j . (Multiplying by I+C where Ckl = 0, Cii =
Cjj = −1, Cij = Cji = 1.)
(ER3) Multiplying row i by a unit c. (Multiplying by C diagonal, all diagonal entries 1
apart from Cii = c.)

All of these are achieved by multiplication on the left by suitable matrices. These
operations are reversible as the corresponding matrices are invertible.

We could similarly consider elementary column operations (EC1), (EC2), (EC3), which
are achieved by multiplying on the right by suitable invertible matrices.

Denition. Two m×n matrices are equivalent if one can get from one to the other via
a sequence of elementary operations.

IfA, B are equivalent then there exists an invertible m × m matrix Q and an invertible
n × n matrix P with B = QAP −1 . We can see this by doing the bookkeeping on the
accumulation of row and column operations.

Theorem 3.7 (Smith normal form). Let A be an m×n matrix over a Euclidean domain
R. Then by a sequence of elementary operations, we can put it into the form
 
d1 0
 .. 
 . 
 

 dr 


 0 

 .. 
 . 
0 0

Lecture 21
 47

with d1 , . . . , d r non-zero and d1 | d2 | · · · | dr . These dk , the invariant factors, are unique

up to associates.

In fact, the product d1 · · · dk is a highest common factor of the k×k minors of A.

Recall the following denition. A k×k minor of A is the determinant of a k×k submatrix
of A. In particular, d1 is a highest common factor of the entries of A, d1 d2 is a highest
common factor of the 2 × 2 minors.

Lemma 3.8. The ideal in R generated by k×k minors of A is unchanged under

elementary operations.

Proof. Consider the 1×1 minors, i.e. the entries of A. Under an elementary operation,
Aij either stays the same or is replaced by Aij + cAik or Aij + cAkj or multiplied by a
unit or replaced by some other entry Akl , and so we deduce that the ideal generated by
the entries of the new matrix is contained in the ideal generated by the entries of the old
matrix. Now the other direction follows from the fact that the operations are revertible.
We thus obtain equality.

More generally, have a similar but messier argument.

Proof of Theorem 3.7. If A = 0 there is nothing to do. Suppose A 6= 0 and we may

assume after switching rows and columns that A11 6= 0. The idea is to use sequences of
elementary operations to reduce φ(A11 ). The process must stop since φ(A11 ) ∈ Z≥0 .
Case 1. A11
does not divide some entry A1j of the rst row then use Euclid's
If
algorithm, write A1j = qA11 + r with r 6= 0, so φ(r) < φ(A11 ). Subtract q times the
rst column from the j th column to give entry r in the (1 j) position. Switch columns
1 and j so that r is now in the top left hand corner.
Case 2. If A11 does not divide some entry of the rst column, use a similar process to
get a new matrix with r 6= 0 and φ(r) < φ(A11 ) in the top left hand corner.

Keep using these two cases until we cannot do so any longer, i.e. when Ak divides all
entries in rst row and column. Then subtract suitable multiples of the rst column
from the others and substract suitable multiples of the rst row from the others to get
matrix of the form  
d 0
0 C
with d 6= 0 for some (m − 1) × (n − 1) matrix C.
Case 3. If we have a matrix of the form
 
d 0
0 C
but d does not divide some entry of C then d - Aij , say. Use Euclid's algorithm so that
Aij = qd + r with r 6= 0 and φ(r) < φ(d). Add column 1 to column j . Subtract q times
row 1 from row i so that we have replaced Aij by r . Switch columns j and 1, rows i and
1, so r appears in top left hand corner.
We are now in a position to apply Case 1 and Case 2 again to reduce the φ-value of
the top left hand corner. Repeat the whole process (Cases 1, 2 and 3) until we can't
anymore, i.e. when the matrix is of the form
 
d 0
0 C

Lecture 21
48 Modules

with d dividing all entries of C.

Keep on going with elementary operations on C etc. The result follows.

Observe that for a matrix of the form

 
d1 0
 .. 
 . 
 

 dr 


 0 

 .. 
 . 
0 0

with d1 | d2 | · · · | dr and dk 6= 0 for all k the ideal generated by the k × k minors is

(d1 · · · dk ), thus by Lemma 3.8 the ideal generated by the k × k minors of the original
matrix is (d1 · · · dk ).

Example. We consider the following sequence of transformations,

       
2 −1 1 −1 1 0 1 0
A= .
1 2 3 2 3 5 0 5


1 0 .)
(i) Add column 2 to column 1. (Multiply on the right by 11


1 1 .)
(ii) Add column 1 to column 2. (Multiply on the right by 01

1 0


(iii) Subtract 3× row 1 from row 2. (Multiply on the left by −3 1 .)

Thus      
1 0 1 0 1 1
= A .
0 5 −3 1 1 2

Lemma 3.9. Let M be a free R-module of rank m, i.e. there is a basis of cardinality
m, and R be an Euclidean domain and suppose that N ≤M is a submodule. Then N
is nitely generated.

Proof. Pick a basis for M and so M∼

= Rm . Identify M with Rm . Consider the ideal

I = {r ∈ R : (r1 , r2 , . . . , rm ) ∈ N } / R.

Since R is an ED and hence a PID, I is generated by a ∈ R, say. Fix an element n ∈ N

of form (a, a2 , . . . , am ). Then for any (r1 , r2 , . . . , rm ) ∈ N we have r1 = ra for some
r ∈ R. Then

(r1 , r2 , . . . , rm ) − rn = (r1 , r2 , . . . , rm ) − r(a, a2 , . . . , am ) ∈ N

= (0, r2 − ra2 , . . . , rm − ram ).

Consider {(0, s2 , . . . , sm ) ∈ N } ≤ {(0, r2 , . . . , rm ) ∈ Rm } and apply induction to see

that the module on the LHS is of rank m − 1, generated by n2 , . . . , nm , say. Then
n, n2 , . . . , nm generate N .

Lecture 22
 49

Theorem 3.10. Let R be a Euclidean domain, N ≤ Rm . Then there is a ba-

sis{v1 , . . . , vm } of R
m such that N is generated by {d1 v1 , d2 v2 , . . . , dr vr } for some
1 ≤ r ≤ m and d1 | d2 | · · · | dr .

Proof. Use Lemma 3.9 to see that N is nitely generated by x1 , . . . , xn , say. So write
these as columns of a matrix A, an m × n matrix.
By Theorem 3.7 on Smith normal forms, we know that we can put A into the form

 
d1 0
 .. 
 . 
 
 dr 
 , di 6= 0, d1 | d 2 | · · · | d r

 0 

 .. 
 . 
0 0

by elementary row and columns operations.

Observe that performing an elementary row operation is associated with a change of

basis of Rm , e.g. adding c times row j to row i replaces the basis e1 , . . . , em of Rm by
e1 , . . . , ei , . . . , ej − cei , . . . , em (replacing the j th element ej by ej − cei ) since

a1 e1 + . . . + am em = a1 e1 + · · · + (ai + caj )ei · · · + aj (ej − cei ) + · · · + am em

Thus the ith coecient has been replaced by ai + caj .

Similary, column operations arise in connection with the change of the generating set of
N. Thus the elementary operations arise when changing basis for Rm and generating
set for N.
At the end of the process the matrix is in Smith normal form and the basis {v1 , . . . , vm }
of Rm is such that {d1 v1 , d2 v2 , . . . , dr vr , 0, . . . , 0} is a generating set for N, as required.
We can throw away the 0s.

Corollary 3.11. A submodule N of Rm when R is a ED is free of rank at most m.

Proof. The set {d1 v1 , . . . , dr vr } freely generates N since {v1 , . . . , vm } are free generators
of Rm .

Theorem 3.12. Let M be a nitely generated R-module, where R is an ED. Then

R R R
M∼
= ⊕ ⊕ ··· ⊕ ⊕ R··· ⊕ R
(d1 ) (d2 ) (dr )

for some dk 6= 0 with d1 | d2 | · · · | dr .

Remark. (i) We may assume that all the dk are non-units, for if dk is a unit then
R/(dk ) ∼
= {0} which may be thrown away.

(ii) A nitely generated R-module is a direct sum of cyclic R-modules. (Assuming

that R an ED.)

Lecture 22
50 Modules

Proof. Suppose the module M is nitely generated by {m1 , . . . , mm }. Then M∼

= Rm /N
by Proposition 3.5. This is because there is a R-module homomorphism

θ : Rm → M, (r1 , . . . , rm ) 7→ r1 m1 + · · · + rm mm

and Im θ = M {m1 , . . . , mm }
since generates M and N = ker θ ≤ Rm . By the isomor-
phism theorem, M∼
= Rm /N .
There exists a basis {v1 , . . . , vm } of Rm such that {d1 v1 , . . . , dr vr } is a generating set
for N. So
Rm /N ∼
= R/(d1 ) ⊕ · · · ⊕ R/(dr ) ⊕ R ⊕ · · · R,
as required.

Example. Take R = Z. The theorem tells us about nitely generated Z-modules, that
is, nitely generated abelian groups. Consider an abelian group A, written additively,
generated by a, b, c subject to relations

2a + 3b + c = 0,
a + 4b = 0,
5a + 6b + 7c = 0.

Then A is a Z-module and A ∼

= Z3 /N , where N is generated by (2, 3, 1), (1, 4, 0), (5, 6, 7).
Write these as columns of a matrix
 
2 1 5
3 4 6
1 0 7

and put this into Smith normal form,

 
1 0 0
0 1 0  .
0 0 21

(Check.) So A∼
= Z3 /N ∼
= Z/(1Z) ⊕ Z/(1Z) ⊕ Z/(21Z) ∼
= Z/(21Z), a cyclic abelian group
of order 21.

Theorem 3.13 (Structure theorem for nitely generated abelian groups) . A nitely
generated abelian group is isomorphic to

Cd1 × Cd2 × · · · × Cdr × C∞ × · · · × C∞ ,

where C∞ is the innite cyclic group.

Proof. Set R=Z in Theorem 3.12 and write the group operation multiplicatively.

Remark. For a nite group there are no copies of C∞ , see Theorem 1.19.

Proposition 3.14 (Primary decomposition) . Let R be a ED. Then

R/(d) ∼
= R/(pn1 1 ) ⊕ · · · ⊕ R/(pns s ),

where d = pn1 1 · · · pns s is the factorisation into irreducibles (and primes).

Lecture 23
 51

n
Proof (cf. (Lemma 1.20).


Split o one summand R/ pj j at a time using the following
lemma.

Lemma 3.15. If d = r1 r2 with hcf(r1 , r2 ) = 1 then M = R/(d) ∼

= R/(r1 ) ⊕ R/(r2 ).

Proof. Let m be a generator of M with Ann(m) = (d). If hcf(r1 , r2 ) = 1 then we can

write 1 = xr1 + yr2 for some x, y ∈ R by Euclid's algorithm. Then

m = 1.m = x(r1 m) + y(r2 m).

We have Ann(r1 m) = (r2 ), Ann(r2 m) = (r1 ) using that we have good factorisation in
R. Set

M1 ∼
= R/(r2 ), M2 ∼
= R/(r1 ).

M1 ∩ M2 = {0} since if sm ∈ M1 ∩ M2 then s is a multiple of both r1 and r2 , and hence

of r1 r2 since hcf(r1 , r2 ) = 1 and so sm = 0.

Consider the R-module homomorphism

M1 ⊕ M2 → M, (m1 , m2 ) 7→ m1 + m2 .

It is onto since M = M1 + M2 , M1 ∩ M2 = {0}. Hence M∼

= M1 ⊕ M2 .

Theorem 3.16. Let M be a nitely generated R-module, where R is a Euclidean

domain. Then

M∼
= N1 ⊕ N2 ⊕ · · · ⊕ Nt
n
where each Nj ∼
= R/(pj j ) and pj is prime (and irreducible) and nj ≥ 1, or Nj ∼
= R.

Proof. Use primary decomposition from Lemma 3.14 to split the components in Propo-
sition 3.12 as direct sums of modules of this form.
n
Note that the pj are not necessarily distinct. The pj j are the elementary divisors and
they are unique up to ordering. (Proof omitted.)

3.4 Modules over F[X] for a eld F  normal forms for

matrices
Let α : V → V be a linear map and V a nite dimensional vector space over a eld F.
We regard V as a F[X]-module via

g(X).v = g(α)(v)

for v ∈V, dependent on the choice of α.

Example. Cyclic modules over F[X], e.g. V = M ∼ = F[X]/(f (X)). Here f (X) is a
polynomial of least degree such that f (α) = 0. We may assume that f (X) is monic, it
is the minimal polynomial for α.

Lecture 23
52 Modules

(i) f (X) = X r . Take generator m of the F[X]-module M , Ann(m) = (f (X)). Then

m, Xm, X 2 m, . . . , X r−1 m is a vector space basis of M = V . Note that

(m, Xm, . . . , X r−1 m) = (m, α(m), . . . , αr−1 (m)).

α is represented by the matrix (with coecients in F)

 
0 0 0
1 0 
 
 .. 
0 1 . .
 
 .. 
 . 0 
0 1 0

(ii) f (X) = (X − λ)r . Then (α − λ)r = 0. Consider β = α − λ.ι then the minimal
r
polynomial of β is X . So there exists a vector space basis of M = V such that β
is represented by
 
0 0 0
1 0 
 
 .. 
0 1 . 
 
 .. 
 . 0 
0 1 0
and so α is represented by
 
λ 0 0
1 λ 
 
 .. 
0 1 . .
 
 .. 
 . λ 
0 1 λ

(iii) For a general f (X), for a generator m with Ann(m) = (f (X)) as in (i), we can
pick a vector space basis m, Xm, . . . , X r−1 m where

f (X) = a0 + a1 X + · · · + ar−1 X r−1 + X r

and

m, Xm, . . . , X r−1 m = m, α(m), . . . , αr−1 (m).

α is represented with respect to this basis by

 
0 0 −a0
1 0
 −a1 
.. .
.
 
0 1 . . .
 
 .. 
 . 0 
0 1 −ar−1

This is called the companion matrix of f (X), written C(f ).

Lecture 23
 53

Theorem 3.17 (Rational canonical form) . Let α : V → V be an endomorphism of

a nite dimensional F-vector space and F be a eld. Then, regarding V as an F[X]-
module M, M ∼ = M1 ⊕ · · · ⊕ Ms with each Mj cyclic and Mj ∼ = F[X]/(fj (X)) where
f1 (X) | f2 (X) | · · · | fs (X), and on choosing an vector space basis for each Mj as in
Example (iii), α is represented by a matrix (with coecients in F)

 
C(f1 ) 0
 C(f2 ) 
.
 
 ..
 . 
0 C(fs )

Proof. Theorem 3.12 splits M as a direct sum of cyclic modules of the right form, and
since M is nite dimensional as a F-vector space there are no components isomorphic
to F[X]. Now use Example (iii) to represent α.

Remark. The name is due to the special case where F = Q.

Remark. (i) The invariant factors fi (X) are unique up to associates. (This is not
quite proved here).

(ii) If A is a square n × n matrix with coecients in F, then A represents a linear map

α : V → V . So the theorem says we can pick a new basis with respect to which
α is represented in rational canonical form. Thus A is conjugate to a matrix in
rational canonical form.

(iii) Minimal polynomials: The minimal polynomial of α is a generator of the annihi-

lator Ann(M ), and this is equal to fs (X) after adjusting to make sure it is monic.

(iv) The characteristic polynomial of α is the product of the characteristic polynomials

of the C(fi ), that is, the product f1 (X) · · · fs (X).

Now we can use Proposition 3.14 on primary decomposition to split the Mi as direct
sums of cyclic modules with annihilators generated by powers of irreducibles.

Assume that F = C, so that the irreducibles are linear.

Theorem 3.18 (Jordan normal form for C). Let α : V → V be an endomorphism of a

nite dimensional C-vector space and regard V as a C[X]-module M . Then

M∼
= M 1 ⊕ · · · ⊕ Ms

where Mj ∼
= C[X]/((X − λj )aj ) for some λj ∈ C. Here, λ1 , . . . , λ s are not necessarily
distinct.

Taking a C-vector space basis for each Mj as in Example (ii), α is represented by a

Lecture 24
54 Modules

matrix of the form

 
λ1 0 0 0
 .. 
1 . 
 
 .. 
 . λ1 
 
0 1 λ1 0 
 

 0 λ2 .

 .. 
 1 . 
 
 .. 
 . λ2 
 

 0 1 λ2 

..
0 0 .

Remark. (i) A submatrix of the form

 
λ 0
 .. 
1 . 
 
 .. 
 . λ 
0 1 λ

is called a Jordan λ-block.

(ii) The Jordan blocks for α are unique up to reordering. (This is not proved here.)

(iii) Minimal polynomials of α: Observe that each Mi is in rational canonical form, so

the theorem yields a λ-block for each λ with X − λ dividing fi (X). Since

f1 (X) | f2 (X) | · · · | fs (X),

the powers of X −λ increase as we pass from f1 (X) to fs (X). Thus the size of
λ-blocks increases as we pass from M1 to Ms , so the largest λ-block arises from
Ms .
Recall that the minimal polynomial of α is
Y
fs (X) = (X − λ)aλ .
λ distinct
Then aλ is the size of the largest λ-block.

(iv) The characteristic polynomial of α factorises into irreducibles as follows,

Y
f1 (X) · · · fs (X) = (X − λ)bλ
λ distinct
where bλ is the sum of the sizes of λ-blocks. Observe that the λ are the eigenvalues
of α.

(v) The geometric multiplicity of λ is dened to be the dimension of the λ-eigenspace

and equal to the number of λ-blocks.

(vi) Given a square complex matrix A, it is conjugate to a matrix in Jordan normal

form.

Lecture 24
 55

Example (Solutions of linear dierence equations and dierential equations). Consider

the space V of complex sequences (zk ) ∈ C∞ that are solutions of

zi+k + ck−1 zi+(k−1) + · · · + c0 zi = 0

for i≥1 with c0 , . . . , ck−1 ∈ C.

Note that V is a nite dimensional C-vector space. Let α: V → V be the left-shift,

(z1 , z2 , . . . ) 7→ (z2 , z3 , . . . ).

The minimal polynomial of α is X k + ck−1 X k−1 + · · · + c0 = f (X), the auxiliary poly-

nomial. Factorise this is as
Y
f (X) = (X − λ)aλ .
λ distinct
Write down the C-vector space associated with the Jordan normal form, as sequences
for each λ,   
k k−a
λ , for 0 ≤ a ≤ aλ − 1
a
e.g. for a=0 we have the sequence (λ, λ2 , λ3 , . . . ) and for a=1 etc.

For dierential equations, the linear map is dierentiation.

Lecture 24