23.03.2013 Views

Without beauty, we are lost.

Without beauty, we are lost.

Without beauty, we are lost.

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

MAS276: Rings and Groups<br />

Lecturer’s version<br />

Dr E. Cheng<br />

J24 Hicks Building<br />

Semester 2, 2011–12<br />

http://cheng.staff.shef.ac.uk/mas276/<br />

<strong>Without</strong> <strong>beauty</strong>, <strong>we</strong> <strong>are</strong> <strong>lost</strong>.<br />

Contents<br />

Introduction 4<br />

I Rings 8<br />

1 Introduction to rings 8<br />

1.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8<br />

1.2 Subrings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15<br />

2 Division 19<br />

2.1 Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20<br />

2.2 Zero-divisors . . . . . . . . . . . . . . . . . . . . . . . . . . . 22<br />

2.3 Integral domains . . . . . . . . . . . . . . . . . . . . . . . . . 31<br />

3 Factorisation 37<br />

3.1 Unique factorisation . . . . . . . . . . . . . . . . . . . . . . . 38<br />

3.2 Euclidean domains . . . . . . . . . . . . . . . . . . . . . . . . 47


2 CONTENTS<br />

II Groups 55<br />

4 Revision 55<br />

4.1 Definitions and examples . . . . . . . . . . . . . . . . . . . . . 55<br />

4.2 Basic theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58<br />

4.3 Homomorphisms . . . . . . . . . . . . . . . . . . . . . . . . . 62<br />

5 Quotient groups 63<br />

5.1 Subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63<br />

5.2 Cosets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64<br />

5.3 Quotient groups . . . . . . . . . . . . . . . . . . . . . . . . . 70<br />

6 Conjugacy 72<br />

6.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73<br />

6.2 The class equation . . . . . . . . . . . . . . . . . . . . . . . . 79<br />

7 Homomorphisms 88<br />

7.1 Kernels and images . . . . . . . . . . . . . . . . . . . . . . . . 88<br />

7.2 Quotient groups revisited . . . . . . . . . . . . . . . . . . . . 96<br />

7.3 First isomorphism theorem . . . . . . . . . . . . . . . . . . . 102<br />

III Homework and Tutorial Questions 109<br />

8 Homework questions 110<br />

9 Tutorial questions 120


CONTENTS 3<br />

Information<br />

Please see the course <strong>we</strong>bpage for ans<strong>we</strong>rs to some Frequently Asked Ques-<br />

tions on matters such as what to do if you miss a lecture. If you email me<br />

with a question that is already ans<strong>we</strong>red in the FAQ, I will not reply.<br />

• Homework is due every <strong>we</strong>ek whether or not you have a tutorial. There<br />

is also tutorial work for every <strong>we</strong>ek whether or not you have a tutorial.<br />

These exercises and instructions <strong>are</strong> at the back of the booklet.<br />

• There will be an online test each <strong>we</strong>ek. These will count for<br />

The deadline for these is:<br />

...........................................<br />

...........................................<br />

• There will be “office hours” each <strong>we</strong>ek when you can come and ask for<br />

extra help with any lecture material or exercises. These <strong>are</strong> at<br />

...........................................<br />

• In <strong>we</strong>ek #n should should do tutorial sheet #n, which will help you<br />

with<br />

1. Online Test Week #n, due on Thursday of <strong>we</strong>ek #n, and<br />

2. Homework Sheet #n, due on Monday of <strong>we</strong>ek #(n + 1).<br />

• For this course you have two hours of lectures per <strong>we</strong>ek and one hour<br />

of tutorial per fortnight. We expect you to do at least three hours of<br />

private study per course per <strong>we</strong>ek as <strong>we</strong>ll. If you do not do this<br />

amount, you <strong>are</strong> likely to find that there is too much material<br />

to learn all at once before the exams.


4 Introduction<br />

Introduction<br />

You have already met groups.<br />

Groups <strong>are</strong> sets of things where those things interact with each other in<br />

nice ways.<br />

Some key examples <strong>are</strong>:<br />

• symmetry groups of various objects (rotations and reflections)<br />

• the integers under addition<br />

• similarly Q, R, C<br />

• n × n matrices under addition<br />

• polynomials under addition<br />

But in the cases apart from symmetries, saying something is a group simply<br />

doesn’t capture everything that’s going on. It’s a bit like saying<br />

A frying pan is a lump of metal.<br />

Beer is a liquid.<br />

In Numbers and Proofs you studied various properties of integers:<br />

• addition and subtraction<br />

• multiplication<br />

• divisibility—when can <strong>we</strong> divide?<br />

• division with remainder<br />

• highest common factor, Euclid’s algorithm<br />

• prime numbers<br />

• unique factorisation into primes


Introduction 5<br />

• modular arithmetic<br />

What <strong>we</strong>’re going to do now is find out what basic properties about Z<br />

enabled us to do all that.<br />

Question: Why do <strong>we</strong> want to know this?<br />

It’s like saying: Why might <strong>we</strong> want to know how a car/computer works?<br />

• sheer curiosity<br />

• so that if it goes wrong <strong>we</strong> can fix it<br />

• it might help us drive better, especially in bad <strong>we</strong>ather<br />

• it helps us be able to use a different one<br />

• <strong>we</strong> can show off to our friends<br />

Numbers <strong>are</strong> great. Numbers <strong>are</strong> everywhere. It would be nice to understand<br />

them better.<br />

But also, there <strong>are</strong> other number-like situations where it would be nice to be<br />

able to use all the techniques <strong>we</strong> know about integers—so <strong>we</strong> have to know<br />

whether that will work or not. Will everything go horribly wrong??<br />

Example: Fermat’s Last Theorem<br />

Example: Will there be some strange loopholes e.g. doctors vs nurses<br />

football?


can always divide<br />

e.g. Q, R, C<br />

6 Introduction<br />

fields <strong>are</strong> vacuously all of those<br />

Groups<br />

Rings<br />

Fields Integral domains<br />

Key examples<br />

Unique factorisation domains<br />

Euclidean domains<br />

these <strong>are</strong> about what<br />

happens when you can’t<br />

divide by some non-zero<br />

things<br />

• Z Q R C<br />

divide can do<br />

analysis<br />

can solve<br />

poly<br />

equs<br />

• R[i] throw some extra thing in and stir it up<br />

Z[i] = the “Gaussian integers”<br />

Z[ √ d ] where d is a squ<strong>are</strong>-free integer<br />

+, −,0<br />

also × and a distributive law<br />

no zero-divisors<br />

unique factorisations<br />

can do Euclid’s algorithm<br />

• Zn gives us different behaviour depending on whether n is prime or<br />

not<br />

• R[x] = polynomials with coefficients in R


Introduction 7<br />

• R[x1,... xn] = polynomials in n variables<br />

• Matn(R) = n × n matrices with coefficients in R


8 Section 1. Introduction to rings<br />

Part I<br />

Rings<br />

1 Introduction to rings<br />

1.1 Definitions<br />

Think about Z and what <strong>we</strong> can do with it:<br />

• add<br />

• subtract<br />

• multiply<br />

• <strong>we</strong> can’t necessarily divide<br />

• <strong>we</strong> have special numbers 0 and 1 — what is special about them?<br />

Definition 1.1.1. A ring is a set R equipped with two operations: addition<br />

+ and multiplication × (or just .) i.e.<br />

for all a,b ∈ R <strong>we</strong> have an element a + b ∈ R and a × b ∈ R<br />

satisfying the following axioms:<br />

1. associativity of + ∀a,b,c ∈ R (a + b) + c = a + (b + c)<br />

2. additive identity ∃0 ∈ R s.t. ∀a ∈ R a + 0 = a = 0 + a<br />

3. additive inverse ∀a ∈ R ∃(−a) ∈ R s.t. a + (−a) = 0<br />

4. commutativity of + ∀a,b ∈ R a + b = b + a<br />

these four say R is an Abelian group under +<br />

5. associativity of × ∀a,b,c ∈ R (ab)c = a(bc)<br />

6. multiplicative identity ∃1 ∈ R s.t. ∀a ∈ R a.1 = a = 1.a<br />

these two say R is a monoid under ×<br />

7. distributive law ∀a,b,c ∈ R a(b + c) = ab + ac<br />

∀a,b,c ∈ R (b + c)a = ba + ca


1.1 Definitions 9<br />

Examples 1.1.2. Some of these examples <strong>we</strong>’ll do in more detail later.<br />

1. Our most basic intuitive examples <strong>are</strong><br />

Z ⊂ Q ⊂ R ⊂ C<br />

—<strong>we</strong>’ll later see that these <strong>are</strong> subrings as shown.<br />

2. Note that the even numbers do not form a ring—why?<br />

there’s no unit<br />

They’re something else useful called an “ideal” which you’ll meet later<br />

if you do Rings and Modules.<br />

3. 2 × 2 real matrices form a ring.<br />

What is 1? What is 0?<br />

4. Polynomials with real coefficients form a ring.<br />

What is +? What is ×? What is 1? What is 0?<br />

5. Zn is a ring for each natural number n.<br />

Definition 1.1.3. A commutative ring is a ring R in which × is commu-<br />

tative, i.e.<br />

∀a,b ∈ R ab = ba<br />

Note that in this case <strong>we</strong> can drop some parts of the axioms—which?<br />

half of multiplicative identity, half of distributivity <br />

Remarks 1.1.4. Here <strong>are</strong> some remarks about the definition which <strong>are</strong><br />

boring, but it would be somehow wrong not to mention them.<br />

1. Some people don’t demand a 1 in their definition of ring.<br />

2. Some people don’t allow 0 = 1 in their definition of ring.<br />

3. Some people always demand commutativity of × in their definition of<br />

ring.<br />

Remarks 1.1.5. Here <strong>are</strong> some more interesting remarks.


10 Section 1. Introduction to rings<br />

1. We don’t demand multiplicative inverses. If <strong>we</strong> have them in a ring<br />

R, that ring is called a field.<br />

2. Many of the most obvious things <strong>we</strong> think <strong>are</strong> true <strong>are</strong> not actually<br />

axioms—<strong>we</strong> have to prove them from scratch.<br />

3. Identities and inverses <strong>are</strong> characterised by the axiom they obey, as<br />

<strong>we</strong> will now see.<br />

Lemma 1.1.6. “Identities <strong>are</strong> unique”<br />

Let R be a ring. Suppose <strong>we</strong> have u,u ′ ∈ R such that<br />

1. ∀x ∈ R u + x = x i.e. u is an additive identity, and<br />

2. ∀x ∈ R u ′ + x = x i.e. u ′ is an additive identity.<br />

Then u = u ′ .<br />

Proof.<br />

Putting x = u ′ in (1) gives u + u ′ = u ′ .<br />

Putting x = u in (2) gives u ′ + u = u.<br />

So <strong>we</strong> have<br />

u ′ = u + u ′<br />

= u ′ + u by commutativity of +<br />

= u<br />

as required. <br />

Since identities <strong>are</strong> unique <strong>we</strong> can “name this baby”. We name it 0, and<br />

then any element satisfying this property must be 0.<br />

Next <strong>we</strong> can do something very similar for additive inverses.<br />

Lemma 1.1.7. “Inverses <strong>are</strong> unique”<br />

Let R be a ring, and let x ∈ R.<br />

Suppose <strong>we</strong> have a,b ∈ R such that<br />

1. x + a = 0 i.e. a is inverse to x<br />

2. x + b = 0 i.e. b is inverse to x


1.1 Definitions 11<br />

Then a = b.<br />

Proof.<br />

But also<br />

x + b = 0 =⇒ a + (x + b) = a + 0<br />

a + (x + b) = (a + x) + b by<br />

= (x + a) + b by<br />

= 0 + b by<br />

= b by<br />

= a by definition of 0<br />

associativity of +<br />

.................................................<br />

commutativity of +<br />

.................................................<br />

assumption 1<br />

.................................................<br />

definition of 0<br />

.................................................<br />

Thus a = a + (x + b) = b as required. <br />

So, like with the identities, <strong>we</strong> can “name this baby”. We call it −x, and<br />

anything satisfying this property must be −x.<br />

Note 1.1.8.<br />

a − b is defined to be a + (−b).<br />

Lemma 1.1.9. Let R be a ring and x ∈ R.<br />

Then 0.x = 0.<br />

Proof.<br />

0x + 0x = (0 + 0)x by<br />

= 0x by<br />

Now, subtracting 0.x from both sides gives<br />

(0x + 0x) − 0x = 0x − 0x<br />

= 0 by<br />

distributive law<br />

.................................................<br />

definition of 0<br />

.................................................<br />

definition of −<br />

.................................................


12 Section 1. Introduction to rings<br />

But<br />

(0x + 0x) − 0x = 0x + (0x − 0x) by<br />

Thus<br />

= 0x + 0 by<br />

= 0x by<br />

0x = (0x + 0x) − 0x<br />

= 0<br />

associativity of +<br />

.................................................<br />

definition of −<br />

.................................................<br />

definition of 0<br />

.................................................<br />

as required. <br />

Lemma 1.1.10. Let R be a ring and x ∈ R. Then<br />

−(−x) = x<br />

Proof. We need to show that x is the additive inverse of −x, i.e.<br />

x + (−x) = 0.<br />

But this is true by definition of −x, so x = −(−x) as required. <br />

This is like Cinderella. Incidentally (and irrelevantly), did you know that<br />

in the original, her slipper wasn’t made of glass at all?<br />

Lemma 1.1.11. Let R be a ring and a,b ∈ R. Then<br />

Proof. We need to show that<br />

i.e. (−a)b + (ab) = 0<br />

(−a)b = −(ab).<br />

(−a)b is the additive inverse of (ab).<br />

i.e. .................................................................................................................


1.1 Definitions 13<br />

Now by the distributive law<br />

(−a)b + (ab) ((−a) + a).b<br />

LHS................................. = ................................<br />

0.b<br />

= .......................... by definition of additive inverse<br />

0<br />

= ................... by Lemma 1.1.9<br />

Thus (−a)b = −(ab) as required. <br />

Corollary 1.1.12. Let R be a ring and a ∈ R. Then<br />

Proof.<br />

Question: Can you tell the time?<br />

(−1)a = −a.<br />

(−1)a = −(1.a) by Lemma 1.1.11<br />

Definition 1.1.13. Integers mod n<br />

= −a by definition of 1. <br />

Let n ∈ N. Then the ring Zn or Z/nZ is defined as follows:<br />

• its set of elements is { 0, 1,... , n − 1 }<br />

• + is addition mod n<br />

• × is multiplication mod n <br />

Definition 1.1.13 by equivalence classes.<br />

Let x,y ∈ Z. We say x ≡ y (mod n) if n|x − y.<br />

This is an equivalence relation.<br />

Equivalence classes <strong>are</strong> represented by the elements {0,1,... n − 1}


14 Section 1. Introduction to rings<br />

The equivalence classes form a ring: <strong>we</strong> define operations<br />

and so on. This ring is Zn.<br />

[x] + [y] = [x + y]<br />

Definition 1.1.14. Polynomial rings<br />

Let R be a ring.<br />

Then <strong>we</strong> define R[x] to be the set of all polynomials with coefficients in R,<br />

i.e.<br />

for any n ∈ N0, where each ai ∈ R.<br />

This is a ring.<br />

Given<br />

<strong>we</strong> define<br />

So the first few terms of f.g <strong>are</strong>:<br />

a0 + a1x + a2x 2 + · · · + anx n<br />

f = a0 + a1x + a2x 2 + · · · + anx n<br />

g = b0 + b1x + b2x 2 + · · · + bmx m<br />

f + g = <br />

(ak + bk)x k<br />

k<br />

f.g = <br />

<br />

( aibj)x k<br />

k<br />

i+j=k<br />

a0b0 + (a0b1 + a1b0)x + (a0b2 + a1b1 + a2b0)x 2 + · · ·<br />

...........................................................................................................................<br />

We also need a 0 and a 1. These <strong>are</strong>:<br />

Definition 1.1.15. Matrix rings<br />

Let R be a ring, n ∈ N.<br />

0 is 0, 1 is 1<br />

................. <br />

We write Matn(R) for the set of all n × n matrices with coefficients in R.<br />

This is a ring with with the usual matrix addition and multiplication.<br />

0 and 1 <strong>are</strong>:<br />

⎛ ⎞ ⎛ ⎞<br />

0 0 1 0<br />

⎝ ⎠ and⎝<br />

⎠<br />

0 0 0 1


1.2 Subrings 15<br />

Note that this is a non-commutative ring. For example:<br />

⎛ ⎞ ⎛ ⎞<br />

1 1 0 0<br />

⎝ ⎠ and⎝<br />

⎠<br />

0 1 0 1<br />

do not commute: ⎛ ⎞ ⎛ ⎞ ⎛ ⎞<br />

1 1 0 0 0 1<br />

⎝ ⎠ ⎝ ⎠ = ⎝ ⎠<br />

0 1 0 1 0 1<br />

but ⎛ ⎞ ⎛ ⎞ ⎛ ⎞<br />

0 0 1 1 0 0<br />

⎝ ⎠ ⎝ ⎠ = ⎝ ⎠<br />

0 1 0 1 0 1<br />

1.2 Subrings<br />

Remember subgroups? Subrings <strong>are</strong> very similar.<br />

Definition 1.2.1. Subring<br />

Let R be a ring. A subring of R is a subset S ⊂ R such that S is a ring<br />

with the same +, ×,0,1 as R.<br />

Explicitly, this means for all a,b ∈ S:<br />

a + b ∈ S<br />

−a ∈ S<br />

0 ∈ S —automatic, but it’s good to remember it<br />

those say that S is a subgroup of R<br />

ab ∈ S<br />

1 ∈ S <br />

Examples 1.2.2. Some examples and non-examples of subrings:<br />

1. Z ⊂ Q ⊂ R ⊂ C<br />

2. Is Z2 a subring of Z?<br />

No: eg 1 + 1 = 0 ∈ Z2 but 1 + 1 = 2 ∈ Z.<br />

Comp<strong>are</strong> with: Zn is not a subgroup of Z. It is a quotient group.


16 Section 1. Introduction to rings<br />

3. Is Z2 a subring of Z4?<br />

No: same counterexample.<br />

4. Are the odd numbers a subring of Z?<br />

No: eg they <strong>are</strong> not closed under addition.<br />

5. Are the even numbers a subring of Z?<br />

No: eg they do not contain 1.<br />

The only subring of Z is Z itself.<br />

6. Define R = { a<br />

b<br />

This is a subring of Q.<br />

∈ Q | b is odd }.<br />

Given a c<br />

b , d ∈ R <strong>we</strong> can check:<br />

• a c ad + bc<br />

+ =<br />

b d bd<br />

which is in R because<br />

• − a<br />

b<br />

• a<br />

b .c<br />

ac<br />

=<br />

d bd<br />

is in R because<br />

which is in R because<br />

• 1 = 1<br />

1<br />

odd × odd = odd<br />

........................................<br />

b is odd<br />

...........................................<br />

which is in R because 1 is odd.<br />

7. Is the trivial ring 0 a subring of R?<br />

odd × odd = odd<br />

........................................<br />

No, unless R = 0 as <strong>we</strong>ll, because in 0, 0=1, which is only true if the<br />

only element is 0.<br />

8. Is Matn(R) a subring of R?<br />

No, as it isn’t even a subset of R.<br />

9. Is R[x] a subring of R? R is a subring of R[x]?<br />

1. No, as again it isn’t even a subset of R. 2. Yes<br />

10. Given a ring R and a natural number n, let S be the subset of R[x]<br />

consisting of polynomicals of degree at most n. Is this a subring of R?<br />

Only if n is 0, otherwise it won’t be closed under multiplication.


1.2 Subrings 17<br />

Definition 1.2.3. Z[ √ d] BRING YOGHURT<br />

Given d ∈ Z, the ring Z[ √ d] is defined to be all numbers of the form<br />

a + b √ d<br />

for any a,b ∈ Z. It is defined as a subring of C, so + and × <strong>are</strong> as in C.<br />

Note <strong>we</strong> need to check that Z[ √ d] is a ring: is it closed under the + and ×<br />

<strong>we</strong> have defined?<br />

√ √ √<br />

Given a1 + b1 d,a2 + b2 d ∈ Z[ d], <strong>we</strong> have<br />

√ √<br />

• a1 + b1 d + a2 + b2 d = (a1 + a2) + (b1 + b2) √ d<br />

√ √<br />

• (a1 + b1 d)(a2 + b2 d) = (a1a2 + b1b2d) + (b1a2 + a1b2) √ d<br />

√ √<br />

• −(a1 + b1 d) = −a − b d<br />

• 1 = 1 + 0. √ d<br />

So this really is a subring of C. <br />

We <strong>are</strong> taking the integers and “throwing something in”.<br />

So the interesting cases <strong>are</strong> when √ d is not an integer—the most interesting<br />

cases <strong>are</strong> when √ d is irrational, which is why <strong>we</strong> use the following condition.<br />

Definition 1.2.4. Squ<strong>are</strong>-free integers<br />

An integer d is called squ<strong>are</strong>-free if it has no repeated prime factors.<br />

Equivalently:<br />

n ∈ Z and n 2 |d =⇒ n = ±1.<br />

Note that 0 is not squ<strong>are</strong>-free. <br />

So, examples of squ<strong>are</strong>-free integers <strong>are</strong>: 2, 10, 15, .........................<br />

Examples of integers that <strong>are</strong> not squ<strong>are</strong>-free <strong>are</strong>: 8, 9, 12, 18, ...................<br />

Actually what <strong>we</strong> usually end up needing to use is:<br />

Lemma 1.2.5. Let d be squ<strong>are</strong>-free. Then for any a ∈ Z<br />

d|a 2 ⇒ d|a.


18 Section 1. Introduction to rings<br />

Lemma 1.2.6. Let d = 1 and squ<strong>are</strong>-free. Then √ d ∈ Q.<br />

Proof. Both of these <strong>are</strong> on Tutorial sheet 3. <br />

Proposition 1.2.7. Let d = 0,1 and squ<strong>are</strong>-free.<br />

Then every element of Z[ √ d] has a unique expression as a + b √ d.<br />

This should remind you of a basis for a vector space. It is very similar. It is<br />

very different from the situation with polynomials—with polynomials, none<br />

of the fruit gets stirred back in.<br />

Proof. By contradiction.<br />

√ √<br />

Suppose a1 + b1 d = a2 + b2 d with b1 = b2.<br />

Then<br />

⇒<br />

a1 − a2 = (b2 − b1) √ d<br />

√ a1 − a2<br />

d =<br />

b2 − b1<br />

since b1 = b2<br />

⇒ √ d is rational # contradicts Lemma 1.2.6<br />

So <strong>we</strong> must have b1 = b2. But this also gives a1 = a2. <br />

Definition 1.2.8. Gaussian integers<br />

When d = −1 <strong>we</strong> get Z[ √ d] = Z[i].<br />

This is called the Gaussian integers and consists of all complex numbers<br />

a + bi where a,b ∈ Z. <br />

Finally here’s a definition that <strong>we</strong> won’t use, but it’s here for completeness.<br />

Definition 1.2.9. Ring homomorphism Given rings R,S, a ring ho-<br />

momorphism is a function f : R −→ S such that<br />

1. ∀a,b ∈ R f(a + b) = f(a) + f(b) group homomorphism<br />

2. ∀a,b ∈ R f(ab) = f(a).f(b)<br />

3. f(1) = 1


Remarks 1.2.10.<br />

1. Just like for group homomorphisms, <strong>we</strong> also have f(0) = 0, but <strong>we</strong><br />

don’t have to include this as an axiom because it follows from (1),<br />

using additive inverses.<br />

f(a) + f(0) = f(a + 0) = f(a), so can add −f(a) to both sides.<br />

If <strong>we</strong> had multiplicative inverses, <strong>we</strong> wouldn’t have to demand f(1) =<br />

1.<br />

2. This definition ensures that if R is a subring of S, the inclusion R −→ S<br />

is a ring homomorphism.<br />

2 Division<br />

Division is hard when you teach it to small children. We didn’t demand it<br />

in a ring. Anyway, what is division?<br />

Comp<strong>are</strong> this with the question: What is subtraction?<br />

adding the additive inverse<br />

Ans<strong>we</strong>r: .................................................................<br />

So division is defined to be multiplication by the multiplicative inverse<br />

—if it exists!<br />

For example given x ∈ Z, 1<br />

x<br />

—in fact it’s only in Z if x = ±1.<br />

might not be in Z<br />

But given x = 0 ∈ Q <strong>we</strong> definitely have 1<br />

x ∈ Q<br />

and likewise for R, C.<br />

Now 1<br />

x<br />

then <strong>we</strong> can “divide by x” by multiplying by 1<br />

x .<br />

is “number such that when you multiply it by x you get 1” —and<br />

The elements <strong>we</strong> can divide by <strong>are</strong> called units.<br />

19


20 Section 2. Division<br />

2.1 Units<br />

Definition 2.1.1. Unit<br />

Let R be a ring and x ∈ R. Then x is called a unit if it has a multiplicative<br />

inverse, i.e.<br />

∃y ∈ R s.t. xy = 1 = yx.<br />

Note: We proved that if it a multiplicative inverse, then it is unique (this<br />

is the same proof as the additive case, Lemma 1.1.7). So <strong>we</strong> can “name this<br />

baby”, and <strong>we</strong> name it x −1 .<br />

Examples 2.1.2.<br />

1. In Z the only units <strong>are</strong> ±1. (They <strong>are</strong> very unit-like.)<br />

—in fact ±1 <strong>are</strong> always units.<br />

2. In Q every non-zero x is a unit: x −1 = 1<br />

x .<br />

Likewise in R and C.<br />

3. In Matn(R) the units <strong>are</strong> the<br />

invertible matrices.<br />

......................................<br />

4. On Tutorial sheet #1 you found the units in Zn for some n.<br />

On Tutorial sheet #2 you saw that the units <strong>are</strong> in Zn <strong>are</strong> easy to<br />

find.<br />

Lemma 2.1.3. If x and y <strong>are</strong> units then xy is a unit.<br />

Proof. We show that xy has an inverse given by: y −1 x −1<br />

(y −1 x −1 )(xy) = 1<br />

(xy)(y −1 x −1 ) = 1<br />

Theorem 2.1.4. Units in Zn.<br />

Let n > 1 be an integer and let a ∈ Zn. Then<br />

a is a unit in Zn ⇔ hcf(a,n) = 1


2.1 Units 21<br />

Proof. Recall from Numbers and Proofs:<br />

hcf(a,n) = 1 ⇐⇒ ∃r,s ∈ Z s.t. ar + ns = 1<br />

⇐⇒ ∃r,s ∈ Z s.t. ar − 1 = ns<br />

⇐⇒ ∃r ∈ Z s.t. n|ar − 1<br />

⇐⇒ ∃r ∈ Z s.t. ar ≡ 1 (mod n)<br />

⇐⇒ ∃y ∈ Zn s.t. ay = 1 ∈ Zn<br />

i.e. hcf(a,n) = 1 iff a is a unit in Zn. <br />

This theorem gives us a way of not only finding if something has an inverse<br />

in Zn, but of actually finding it.<br />

Example 2.1.5. Is 47 a unit in Z157? If so, find its inverse.<br />

1. We use Euclid’s algorithm to find hcf(157,47):<br />

157 = 3.47 + 16<br />

47 = 2.16 + 15<br />

16 = 1.15 + 1<br />

15 = 15.1 + 0<br />

So hcf(157,47) = 1<br />

......... and 47 is/is not* a unit in Z157.<br />

2. We feed the numbers back in to find r,s ∈ Z s.t. ar + ns = 1:<br />

Line 1: 16 = 157 − 3.47<br />

Line 2: 15 = 47 − 2.16<br />

= 47 − 2.(157 − 3.47)<br />

= 7.47 − 2.157<br />

Line 3: 1 = 16 − 1.15<br />

= (157 − 3.47) − (7.47 − 2.157)<br />

= 3.157 − 10.47<br />

So −10.47 ≡ 1 (mod 157).<br />

−10 ≡ 147<br />

and .......................... is the inverse of 47 in Z157.


22 Section 2. Division<br />

Proposition 2.1.6. The units of a ring R form a group U(R) under ×.<br />

Proof. We need to show<br />

1. x,y ∈ U(R) =⇒ x.y ∈ U(R)<br />

2. x ∈ U(R) =⇒ x −1 ∈ U(R).<br />

Then <strong>we</strong> automatically get 1 ∈ U(R). Now<br />

1.<br />

is Lemma 2.1.3<br />

.....................................................................<br />

2. We need to show that x−1 is a unit, i.e., that it has an inverse. We<br />

x<br />

show that its inverse is ............ :<br />

x −1 .x = 1 = x.x −1<br />

by definition of x −1 . <br />

Examples 2.1.7.<br />

1. In Z <strong>we</strong> have U(Z) = {1, −1} which is a group, isomorphic to C2, the<br />

cyclic group of order 2.<br />

2. In Q the non-zero elements form a multiplicative group.<br />

3. In Z8 the units <strong>are</strong> {1,3,5,7}. Every element has order 2.<br />

2.2 Zero-divisors<br />

How do <strong>we</strong> usually solve quadratics? E.g.<br />

x 2 + 3x = 2 = 0 as in HW # 1<br />

(x + 1)(x + 2) = 0<br />

⇒ x + 1 = 0 or x + 2 = 0<br />

⇒ x = −1, −2


2.2 Zero-divisors 23<br />

We crucially used the fact that<br />

ab = 0 ⇒ a = 0 or b=0.<br />

.........................................................................<br />

“You can’t multiply non-zero things and get zero.”<br />

Whereas in Z6<br />

ab = 0 ⇒ 1. a = 0<br />

or 2. b = 0<br />

or 3. {a,b} = {2,3}<br />

or 4. {a,b} = {3,4}<br />

So if <strong>we</strong> try to solve that quadratic <strong>we</strong> get<br />

(x + 1)(x + 2) = 0 ⇒ 1. x + 1 = 0 i.e. x = 5<br />

or 2. x + 2 = 0 i.e. x = 4<br />

or 3. {x + 1,x + 2} = {2,3} i.e. x = 1<br />

or 4. {x + 1,x + 2} = {3,4} i.e. x = 2<br />

So the solutions <strong>are</strong> x = 1,2,4,5 —there <strong>are</strong> four solutions!<br />

What about in Z9? Z12? Z30? this is long<br />

Z9 3,3,<br />

Z12 2,2,3<br />

Z30 2,3,5<br />

x + 1,x + 2 x<br />

x + 1 = 0 8<br />

x + 2 = 0 7<br />

x + 1,x + 2 x<br />

x + 1 = 0 11<br />

x + 2 = 0 10<br />

3,4 2<br />

8,9 7


24 Section 2. Division<br />

x + 1,x + 2 x<br />

x + 1 = 0 29<br />

x + 2 = 0 28<br />

5,6 4<br />

9,10 8<br />

14,15 13<br />

15,16 14<br />

20,12 19<br />

24,25 23<br />

The moral is that our methods for solving quadratics <strong>are</strong> rather more com-<br />

plicated if <strong>we</strong> have more options for ab = 0. In this example the problem<br />

was caused by the fact that 2.3 = 0 and also 3.4 = 0.<br />

In fact, it’s not just quadratics that go wrong, e.g.<br />

has two solutions:<br />

or more simply<br />

x ≡ 2,5<br />

.........................<br />

2x ≡ 4 (mod 6)<br />

2x ≡ 0 (mod 6)<br />

x ≡ 0,3<br />

has two solutions: .........................<br />

So even solving linear equations is a bit strange. We also know that<br />

does not necessarily mean a ≡ b e.g.<br />

3a ≡ 3b (mod 6)<br />

3.1 ≡ 3.5 , or 0,2; 2,4; 0,4; 3,5<br />

...................................................<br />

All this is because in Z6 <strong>we</strong> have the possibility of multiplying non-zero a<br />

and b getting zero.<br />

These <strong>are</strong> called zero-divisors.


2.2 Zero-divisors 25<br />

Definition 2.2.1. Let R be a ring and r ∈ R.<br />

Then r is called a left zero-divisor if<br />

r is called a right zero-divisor if<br />

∃s = 0 ∈ R s.t. rs = 0.<br />

∃s = 0 ∈ R s.t. sr = 0.<br />

If R is both a left and a right zero-divisor <strong>we</strong> simply say it is a zero-divisor;<br />

in particular if R is commutative then <strong>we</strong> don’t need to distinguish bet<strong>we</strong>en<br />

left and right.<br />

If R = 0 then 0 is certainly a zero-divisor. This is a “boring” case, so if<br />

r = 0 and is a zero-divisor <strong>we</strong> call it a proper zero-divisor. <br />

In this course almost all the rings <strong>we</strong> deal with will be commutative. The<br />

main exception is matrix rings.<br />

Examples 2.2.2.<br />

1. The zero-divisors in Z, Q, R, C <strong>are</strong>:<br />

Only 0, so there <strong>are</strong> no proper ones.<br />

2. The zero-divisors in Z4 <strong>are</strong>: 0,2<br />

The zero-divisors in Z6 <strong>are</strong>: 0,2,3,4<br />

The zero-divisors in Z8 <strong>are</strong>: 0,2,4,6<br />

Can you guess what the pattern is?<br />

⎛ ⎞<br />

1 1<br />

3. In Mat2(Z), ⎝ ⎠ is a left and right zero-divisor: e.g.<br />

2 2<br />

⎛ ⎞⎛<br />

⎞<br />

1 1 1<br />

⎝ ⎠⎝<br />

1<br />

⎠ = 0 and<br />

2 2 −1 −1<br />

⎛ ⎞⎛<br />

⎞<br />

−2 1 1 1<br />

⎝ ⎠⎝<br />

⎠ = 0<br />

−2 1 2 2<br />

In fact any matrix with determinant 0 is a zero-divisor.<br />

Theorem 2.2.3. Zero-divisors in Zn. (See also Tutorial sheet #2.)<br />

Let n > 1 be an integer and a ∈ Zn. Then


26 Section 2. Division<br />

1. a is a unit in Zn ⇐⇒ hcf(a,n) = 1, and<br />

2. a is a zero-divisor in Zn ⇐⇒ hcf(a,n) = 1.<br />

N.B. So every element of Zn is a unit or a zero-divisor and not both.<br />

We will see that there <strong>are</strong> some rings with elements that <strong>are</strong> neither a unit<br />

or a zero-divisor, but it is impossible to be both a unit and a zero-divisor.<br />

Before proving this theorem let’s look at some examples.<br />

Example 2.2.4. Consider Z12.<br />

Can <strong>we</strong> multiply 2 by something non-zero and get 0? 6<br />

Let’s try and do it a bit more systematically.<br />

3 4<br />

4 3<br />

6 2<br />

8 3<br />

9 4<br />

10 6<br />

Example 2.2.5. n = 12,a = 8<br />

Now hcf(8,12) = 4<br />

.......... so <strong>we</strong> expect 8 to be a zero-divisor in Z12.<br />

So can <strong>we</strong> find b such that 8b ≡ 0? We write<br />

12 = 3.4<br />

8 = 2.4<br />

so 2.4.3 ≡ 0 (mod 12)<br />

i.e. 8.3 ≡ 0 (mod 12)<br />

That example was easy enough to do just by staring, but for bigger numbers<br />

this technique is much more efficient than staring.


2.2 Zero-divisors 27<br />

Example 2.2.6. n = 201, a = 63<br />

First <strong>we</strong> find hcf(63,201) e.g. by finding their prime factorisations:<br />

201 = 3.67<br />

63 = 3.3.7<br />

So hcf(63,201) = 3<br />

................ so <strong>we</strong> expect 63 to be a zero-divisor in Z201.<br />

So can <strong>we</strong> find b such that 63b ≡ 0? We have<br />

3.3.7.67 ≡ 0 (mod 201)<br />

i.e. 63.67 ≡ 0 (mod 201)<br />

Before <strong>we</strong> prove the theorem, the following result is useful.<br />

Lemma 2.2.7. “You can’t be both a zero-divisor and a unit” boring<br />

lecture?<br />

Let R be a ring and a ∈ R. Then<br />

a is a unit =⇒ a is not a zero-divisor<br />

Equivalently (using the contrapositive):<br />

a is a zero-divisor =⇒ a is not a unit.<br />

Proof. Suppose a is a unit and ar = 0. Then<br />

r = a −1 .ar<br />

= a −1 .0<br />

= 0<br />

so a is not a zero-divisor. <br />

Proof of Theorem 2.2.3.<br />

1. Done as Theorem 2.1.4.<br />

2. Write hcf(a,n) = d > 1 and seek b = 0 such that ab ≡ 0 (mod n).<br />

Now<br />

n = n<br />

a<br />

.d, a =<br />

d d .d<br />

a n<br />

d , d ∈ Z


28 Section 2. Division<br />

Put b = n<br />

. Then ab = a.n<br />

d d<br />

a<br />

= .n ≡ 0 (mod n).<br />

d<br />

since a<br />

d ∈ Z. Finally <strong>we</strong> must check b = 0. Now<br />

0 < n<br />

d<br />

< n<br />

since d > 1, so<br />

n<br />

≡ 0<br />

d<br />

(mod n).<br />

For the converse <strong>we</strong> need to show that if a is a zero-divisor then<br />

hcf(a,n) = 1.<br />

Now, if a is a zero-divisor then by Lemma 2.2.7 it cannot be a unit,<br />

so by Theorem 2.1.4 <strong>we</strong> have hcf(a,n) = 1 as required. <br />

Example 2.2.8. In Z6[x], the element 1 + 3x is neither a unit nor a<br />

zero-divisor.<br />

For 1 + 3x to be a unit <strong>we</strong> need<br />

(1 + 3x)(a0 + a1x + a2x 2 + · · · anx n ) = 1 ∈ Z6[x]<br />

Comparing coefficients, this gives us<br />

comparing constants a0 ≡ 1 (mod 6)<br />

comparing coeffs of x i 3ai−1 + ai ≡ 0 (mod 6) ∀0 < i ≤ n<br />

comparing coeffs of x n+1 3an ≡ 0 (mod 6)<br />

Now 3a0 + a1 ≡ 0 =⇒ 3 + a1 ≡ 0 =⇒ a1 ≡ 3.<br />

Similarly 3a1 + a2 ≡ 0 =⇒ 3 + a2 ≡ 0 =⇒ a2 ≡ 3.<br />

Similarly ai ≡ 3 for all 0 < i ≤ n.<br />

But also 3an ≡ 0 #.<br />

Hence 1 + 3x is not a unit.<br />

To show that 1 + 3x is not a zero-divisor, consider<br />

(1 + 3x)(a0 + a1x + a2x 2 + · · · anx n ) = 0 ∈ Z6[x]


2.2 Zero-divisors 29<br />

Comparing coefficients, this gives us<br />

comparing constants a0 ≡ 0 (mod 6)<br />

comparing coeffs of x i 3ai−1 + ai ≡ 0 (mod 6) ∀0 < i ≤ n<br />

comparing coeffs of x n+1 3an ≡ 0 (mod 6)<br />

Now 3a0 + a1 ≡ 0 =⇒ a1 ≡ 0.<br />

Similarly 3a1 + a2 ≡ 0 =⇒ a2 ≡ 0.<br />

Similarly ai ≡ 0 for all 0 < i ≤ n.<br />

Hence 1 + 3x is not a zero-divisor. <br />

Remark 2.2.9. Note that to prove r is not a zero-divisor <strong>we</strong> prove<br />

rs = 0 =⇒ s = 0.<br />

So in general <strong>we</strong> have the following picture.<br />

In Zn it happens to be like this:<br />

R<br />

units zero-divisors<br />

R<br />

units zero-divisors


30 Section 2. Division<br />

We’ll see that in a field it’s like this:<br />

units<br />

and in an integral domain it’s like this:<br />

units<br />

We can still do something like divide by r even if r is not a unit—as long as<br />

it isn’t a zero-divisor.<br />

Example 2.2.10. In Z6 suppose <strong>we</strong> <strong>are</strong> given<br />

Can <strong>we</strong> deduce<br />

No: e.g. 3.1 ≡ 3.5 but 1 ≡ 5.<br />

Ho<strong>we</strong>ver <strong>we</strong> can do 3a − 3b ≡ 0<br />

so 3(a − b) ≡ 0<br />

R<br />

R<br />

3a ≡ 3b (mod 6)<br />

a ≡ b (mod 6) ?<br />

so if 3 is not a zero-divisor <strong>we</strong> can deduce a − b ≡ 0 i.e. a ≡ b.<br />

Lemma 2.2.11. Cancellation<br />

Let R be a ring. Suppose r = 0 ∈ R is not a zero-divisor, and ra = rb.<br />

Then a = b.<br />

0<br />

0


2.3 Integral domains 31<br />

Proof. ra = rb ⇒ ra − rb = 0<br />

⇒ r(a − b) = 0<br />

⇒ a − b = 0 since r is not a zero-divisor<br />

⇒ a = b<br />

Effectively <strong>we</strong> have “cancelled” r. <br />

Remember how in Numbers and Proofs you solved things like 2x ≡ 4 (mod 6) ?<br />

—You had to start by taking the hcf of 2 and 6. Effectively, that was to see<br />

if 2 was a zero-divisor or not (although they didn’t tell you at the time).<br />

We have seen that zero-divisors <strong>are</strong> a bit annoying. In the next section,<br />

<strong>we</strong> imagine <strong>we</strong>’re in a glorious world in which there <strong>are</strong> no zero-divisors.<br />

Such a world is called an integral domain.<br />

2.3 Integral domains<br />

We have been studying various things that go a bit pear-shaped if you have<br />

zero-divisors e.g. solving all kinds of equations. If <strong>we</strong> want to exclude that<br />

possibility <strong>we</strong> go into “integral domains”.<br />

Note that from this point on, all our rings <strong>are</strong> commutative.<br />

Definition 2.3.1. An integral domain (ID) is a commutative ring with<br />

no proper zero-divisors i.e. the only zero-divisor is 0. <br />

Examples 2.3.2.<br />

1. Z is an ID. So <strong>are</strong> Q, R, C.<br />

2. Z[ √ d] ⊂ C and <strong>we</strong> will see that this means it must be an ID.<br />

3. We will see: if R is an ID then R[x] is an ID (and the converse).<br />

N.B. In an ID <strong>we</strong> can cancel by anything non-zero, since by Lemma 2.2.11<br />

<strong>we</strong> can cancel by anything that isn’t a zero-divisor.<br />

In the case of Zn it is nice and easy to tell whether or not <strong>we</strong> have an ID.


32 Section 2. Division<br />

Theorem 2.3.3. Zn is an ID ⇐⇒ n is prime.<br />

Proof. Follows from Theorem 2.2.3:<br />

Zn is ID ⇔ ∀ 1 ≤ a ≤ n − 1, a is not a zero-divisor<br />

⇔ ∀ 1 ≤ a ≤ n − 1,hcf(a,n) = 1 by Theorem 2.2.3<br />

⇔ n is prime <br />

Examples 2.3.4. So for example<br />

• Z2, Z3, Z5, Z7, Z11,... <strong>are</strong> IDs<br />

• Z4, Z6, Z8, Z9,... <strong>are</strong> not IDs.<br />

Lemma 2.3.5. “A subring of ID is ID”<br />

Let S be a subring of R.<br />

Then R is ID =⇒ S is ID.<br />

Proof.<br />

Let a = 0,b = 0 ∈ S.<br />

Then a = 0,b = 0 ∈ R so ab = 0 ∈ R<br />

Then a = 0,b = 0 ∈ R so ab = 0 ∈ S<br />

since S has the same multiplication as R. <br />

Examples 2.3.6.<br />

1. Z[ √ d] is a subring of C so is always an ID.<br />

2. Any ring R is a subring of R[x].<br />

So if R[x] is an ID then R must be an ID.<br />

In practice <strong>we</strong> often use the contrapositive.<br />

The converse of that second part is more profound, and harder to prove:<br />

Theorem 2.3.7. Let R be a ring. Then R is ID ⇐⇒ R[x] is ID.<br />

Proof.<br />

“⇐=” is by Lemma 2.3.5 since R is a subring of R[x].


2.3 Integral domains 33<br />

“=⇒”<br />

Suppose R is ID.<br />

Let f,g = 0 ∈ R[x]. We need to show fg = 0.<br />

Let deg(f) = m, deg(g) = n.<br />

Recall deg(f) = largest m s.t. coeff of x m is not 0.<br />

So write f(x) = a0 + · · · + amx m<br />

g(x) = b0 + · · · + bnx n<br />

Now am, bn = 0 =⇒ ambn = 0 since R is ID.<br />

But ambn is the coefficient of x m+n in fg.<br />

So fg = 0. <br />

Now <strong>we</strong> turn our attention to finding out what the units <strong>are</strong> in various IDs.<br />

Theorem 2.3.8. Units in polynomial rings.<br />

Let R be an ID.<br />

Then the units in R[x] <strong>are</strong> just the units in R (so in particular they <strong>are</strong> all<br />

constants).<br />

Proof. —similar to Theorem 2.3.7.<br />

Consider f,g ∈ R[x] s.t. fg = 1.<br />

As before let deg(f) = m, deg(g) = n and write<br />

so in particular am, bn = 0.<br />

f(x) = a0 + · · · + amx m<br />

g(x) = b0 + · · · + bnx n<br />

Now <strong>we</strong> know that the coefficient of x m+n in fg is ambn<br />

and this is non-zero as R is an ID.<br />

But comparing coefficients in the equation fg = 1 <strong>we</strong> must have m + n = 0,<br />

since that is the only non-zero coefficient on the RHS.<br />

But m + n = 0 =⇒ m = 0 and n = 0.


34 Section 2. Division<br />

So <strong>we</strong> know that f = a0 and g = b0, constants.<br />

Moreover <strong>we</strong> know that a0b0 = 1 as constants, i.e. they <strong>are</strong> units in R. <br />

Next <strong>we</strong> investigate units in Z[ √ d]. These <strong>are</strong> not very obvious at first sight.<br />

To help with this <strong>we</strong> introduce something called a “norm”, which is a bit<br />

like a way of measuring how “big” elements of Z[ √ d] <strong>are</strong>. Eventually <strong>we</strong>’ll<br />

show that the units <strong>are</strong> precisely those elements of norm 1, but it will take<br />

us a few more lemmas to get there.<br />

Definition 2.3.9. Norm in Z[ √ d].<br />

Let d = 1 be a squ<strong>are</strong>-free integer, and let r = a + b √ d ∈ Z[ √ d].<br />

We define the norm of r as<br />

N(r) = |a 2 − db 2 | = |(a + b √ d)(a − b √ d)|.<br />

This is <strong>we</strong>ll-defined by uniqueness of a and b (Proposition 1.2.7).<br />

Note that this means N(r) = ±(a + b √ d)(a − b √ d). <br />

The following properties of the norm <strong>are</strong> extremely useful.<br />

Lemma 2.3.10. Let d = 1 be a squ<strong>are</strong>-free integer, and r,s ∈ Z[ √ d]. Then<br />

1. N(r) = 0 ⇐⇒ r = 0<br />

2. N(rs) = N(r)N(s).<br />

Proof.<br />

1. “⇐=” is by the definition of norm.<br />

Now for “=⇒”, put r = a + b √ d and suppose N(r) = 0<br />

i.e. |a 2 − b 2 d| = 0<br />

so a 2 = b 2 d. (1)<br />

If b = 0 <strong>we</strong> have √ d = ± a<br />

∈ Q #<br />

b<br />

since d is squ<strong>are</strong>-free (Lemma 1.2.6).<br />

So b = 0, and by (1) <strong>we</strong> then get a = 0.<br />

So r = 0 as required.


2.3 Integral domains 35<br />

√ √<br />

2. We simply calculate: Put r = a1 + b1 d, s = a2 + b2 d. Then<br />

N(r)N(s) = |a 2 1 − b2 1 d|.|a2 2 − b2 2 d|<br />

= |a 2 1 a2 2 + b2 1 b2 2 d2 − a 2 1 b2 2 d − a2 2 b2 1 d|<br />

N(rs) = N (a1a2 + b1b2d) + (a1b2 + b1a2) √ d <br />

= |(a1a2 + b1b2d) 2 − (a1b2 + b1a2) 2 d|<br />

= |a 2 1 a2 2 + 2a1a2b1b2d + b 2 1 b2 2 d2 − a 2 1 b2 2 d − 2a1a2b1b2d − b 2 1 a2 2 d|<br />

= |a 2 1 a2 2 + b2 1 b2 2 d2 − a 2 1 b2 2 d − b2 1 a2 2 d|<br />

= N(r)N(s) <br />

PS That was not fun to type. boring lecture?<br />

We can now prove that the norm really does tell us which elements <strong>are</strong> units.<br />

Theorem 2.3.11. Let d = 1 be a squ<strong>are</strong>-free integer and r ∈ Z[ √ d]. Then<br />

Proof.<br />

“⇐=”<br />

Consider r = a + b √ d with N(r) = 1.<br />

r is a unit in Z[ √ d] ⇐⇒ N(r) = 1.<br />

Then N(r) = |(a + b √ d)(a − b √ d)| = 1<br />

so r.(a − b √ d) = ±1<br />

so either (a − b √ d) or −(a − b √ d) is an inverse for r, i.e. r is a unit.<br />

“=⇒”<br />

Conversely suppose r is a unit in R. Then<br />

N(r)N(r −1 ) = N(rr −1 ) by Lemma 2.3.10<br />

= N(1)<br />

= 1<br />

Now N(r) and N(r −1 ) <strong>are</strong> both non-negative integers so must both be 1<br />

so in particular N(r) = 1 as required.


36 Section 2. Division<br />

Examples 2.3.12. We can use this to “test” whether a given element is a<br />

unit or not:<br />

1. Consider Z[ √ 2], r = 7 + 5 √ 2. Then N(r) = |49 − 25.2| = 1<br />

so 7 + 5 √ 2 isn’t a unit / is a unit* with inverse −(7 − 5√ 2)<br />

....................<br />

2. Consider Z[ √ 3], r = 4 + 7 √ 3. Then N(r) = |16 − 49.3| = 131 = 1<br />

so 4 + 7 √ 3 isn’t a unit / is a unit* with inverse<br />

3. Consider Z[ √ 3], r = 7 + 4 √ 3. Then N(r) = |49 − 16.3| = 1<br />

so 7 + 4 √ 3 isn’t a unit / is a unit* with inverse<br />

*delete as appropriate<br />

not a unit<br />

....................<br />

7 − 4 √ 3<br />

....................<br />

Note that this does give us a way of testing if a given element is a unit, but<br />

it doesn’t give us a way of actuallly finding all the units. In fact, finding all<br />

the units is hard in general—the following theorem tells us when it’s easy<br />

and when it’s hard.<br />

Theorem 2.3.13. Units in Z[ √ d]. Let d = 1 be a squ<strong>are</strong>-free integer.<br />

1. If d < −1 then U(Z[ √ d]) = {1, −1}.<br />

2. If d = −1 then U(Z[ √ d]) = {1, −1,i, −i}.<br />

3. If d > 1 then U(Z[ √ d]) is infinite.<br />

Proof.<br />

1. Put d = −m, with m > 1.<br />

Then N(a + b √ d) = |a 2 − b 2 d| = a 2 + b 2 m.<br />

But m > 1 so a 2 + b 2 m = 1 ⇒ b 2 = 0 and a 2 = 1<br />

⇒ b = 0 and a = ±1<br />

dummy text


2. (Gaussian integers)<br />

N(a + bi) = |a 2 − b 2 .(−1)| = a 2 + b 2 .<br />

Now a 2 + b 2 = 1 ⇒ {a 2 = 1 and b 2 = 0} or {a 2 = 0 and b 2 = 1}<br />

37<br />

⇒ {a = ±1 and b = 0} or {a = 0 and b = ±1}<br />

⇒ r = ±1 or ± i<br />

dummy text <br />

3. We will only prove this for the case d = 3.<br />

We know r = 7 + 4 √ 3 is a unit (see Examples 2.3.12).<br />

We also know that the units <strong>are</strong> closed under multiplication<br />

(Lemma 2.1.3), so r k is a unit for all k ∈ N.<br />

Now r > 1 so<br />

i.e. the numbers r k <strong>are</strong> distinct<br />

1 < r < r 2 < r 3 < · · ·<br />

so U(Z[ √ d]) is infinite. <br />

Note that to make the proof of part (3) work for any other value of d, <strong>we</strong><br />

just have to find one unit in Z[ √ d] other than ±1.<br />

Definition 2.3.14. A ring is called a field if every non-zero element is a<br />

unit.<br />

Example 2.3.15. Q, R, C <strong>are</strong> all fields.<br />

Lemma 2.3.16. Zn is a field if and only if n is prime.<br />

Proof.<br />

An element a = 0 ∈ Zn is a unit if and only if hcf(a,n) = 1 (Theorem 2.1.4).<br />

This is true for all a = 0 ∈ Zn if and only if hcf(a,n) = 1 for all 1 < a < n<br />

i.e. if and only if n is prime. <br />

3 Factorisation<br />

Now that <strong>we</strong> know a bit about dividing, <strong>we</strong> can think about factorisation.


38 Section 3. Factorisation<br />

3.1 Unique factorisation<br />

One of the most crucial things about the integers is unique prime factorisa-<br />

tion, that is:<br />

“Every integer has a unique factorisation into a product of prime<br />

numbers.”<br />

Question: What does “unique” mean? What about these different factori-<br />

sations of 30<br />

30 = 2 × 3 × 5 = (−2) × (−3) × 5 = 5 × 3 × 2?<br />

There could be some ±1 floating around, and <strong>we</strong> could change the order of<br />

things, but that doesn’t count as genuinely different.<br />

We’ll see that <strong>we</strong> have to be even more c<strong>are</strong>ful about this in rings with many<br />

units. But first <strong>we</strong> have to make sure <strong>we</strong> know what “prime” means in an<br />

arbitrary ring. The crucial property <strong>we</strong> want is that a prime number can’t<br />

be factorised any more. This is actually called “irreducibility”.<br />

Definition 3.1.1. “Irreducible”<br />

Let R be a commutative ring and r = 0 ∈ R. Then r is irreducible in R if<br />

1. r is not a unit, and<br />

2. r cannot be written as a product of non-units, i.e.<br />

r = st =⇒ s is a unit or t is a unit.<br />

Example 3.1.2. We can check that −3 is irreducible in Z:<br />

1. −3 is not a unit because the units in Z <strong>are</strong> just<br />

1,-1<br />

................


3.1 Unique factorisation 39<br />

2. If −3 = st then the possibilities for s,t <strong>are</strong>:<br />

1,-3; -3,1; -1, 3; 3, -1<br />

....................................................................<br />

and in each case s or t is a unit.<br />

Example 3.1.3. Suppose <strong>we</strong> remove the number 2 from existence.<br />

Then the following numbers become “irreducible”:<br />

4,6,8,10,14,...<br />

and <strong>we</strong> get some non-unique factorisations, e.g.<br />

Remarks 3.1.4.<br />

24 = 6.4 = 3.8<br />

1. The irreducibles in Z <strong>are</strong> ±p where p is prime.<br />

2. The condition saying r is not a unit is like the fact that 1 is not<br />

considered to be a prime number.<br />

3. Do the irreducibles form a group? Of course not—<strong>we</strong> can’t multiply<br />

them!<br />

4. Dire warning: in Z any prime p has the following property:<br />

p|ab =⇒ p|a or p|b.<br />

This property is called being prime. This is not the same as being<br />

irreducible in general, but in Z the notions happen to coincide. This<br />

is not true in all rings!<br />

Theorem 3.1.5. Criterion for irreducibility in Z[ √ d].<br />

Let d = 1 be a squ<strong>are</strong>-free integer, and r ∈ Z[ √ d]. Then<br />

Proof.<br />

N(r) is prime =⇒ r is irreducible in Z[ √ d].<br />

Let r ∈ Z[ √ d] with clN(r) not prime.<br />

So certainly N(r) = 1 so r is not a unit, by Theorem 2.3.11.


40 Section 3. Factorisation<br />

Now suppose r = st. We aim to deduce that s or t must be a unit.<br />

Then N(r) = N(s)N(t) by Lemma 2.3.10.<br />

But N(r) is prime, so <strong>we</strong> must have N(s) = 1 or N(t) = 1<br />

so by Theorem 2.3.11 s or t is a unit.<br />

Hence r is irreducible as required. <br />

Note that this condition is sufficient but not necessary, that is, the con-<br />

verse of the theorem is not true. i.e.<br />

There exists a squ<strong>are</strong>-free integer d and r ∈ Z[ √ d] such that r is irreducible<br />

in Z[ √ d] but N(r) is not prime.<br />

Examples 3.1.6.<br />

1. In Z[ √ −7], N(2 + √ −7) =<br />

prime<br />

so 2 + √ −7 is irreducible in Z[ √ −7].<br />

2. In Z[ √ −7] there is no element of norm 2:<br />

|4 + 7| = 11<br />

................................................. which is<br />

|a 2 + 7b 2 | = 2 =⇒ b = 0 otherwise a 2 + 7b 2 ≥ 7<br />

=⇒ a 2 = 2 # contradicts a ∈ Z<br />

This means that any element of norm 4 or 8 must be irreducible:<br />

if r has norm 4 it is certainly not a unit.<br />

Now suppose r = st, so N(r) = N(s)N(t).<br />

But there is no element of norm 2<br />

so <strong>we</strong> must have<br />

either N(s) = 1 and N(t) = 4, in which case s is a unit<br />

or N(s) = 4 and N(t) = 1, in which case t is a unit.<br />

A similar argument works for 8.<br />

3. On Tutorial sheet 5 you’ll show that in Z[ √ −11] there <strong>are</strong> no elements<br />

of norm 3 or 23, and hence an element is irreducible if it has norm<br />

3 2 , 23 2 , 3.23, 3.3.23, 3.23.23, 3 3 , 23 3


3.1 Unique factorisation 41<br />

For example N(5 + 2 √ −11) =<br />

4. In Z[ √ −7] <strong>we</strong> can factorise 11 as<br />

Now<br />

|25 + 44| = 69 = 3.23<br />

.....................................<br />

11 = (2 + √ −7)(2 − √ −7).<br />

N(2 + √ −7) = N(2 − √ −7) = 11,<br />

so neither of these factors is a unit<br />

so <strong>we</strong> have factorised 11 as a product of non-units<br />

i.e. 11 is not irreducible in Z[ √ −7].<br />

One moral of this is that a number can be irreducible in one ring but<br />

not another, so it is important to say what ring <strong>we</strong>’re working in when<br />

<strong>we</strong>’re talking about irreducibles.<br />

5. In general if N(r) = k then k is not irreducible, as in the above exam-<br />

ple.<br />

Now <strong>we</strong> need to think about what “uniqueness” means for unique factori-<br />

sations. We know that changing the order of the factors shouldn’t count<br />

as genuinely different, but here’s something else that might confuse us es-<br />

pecially if <strong>we</strong>’re in a ring with many units.<br />

We might factorise an element r as<br />

r = p.q<br />

but then given a unit u <strong>we</strong> could also put<br />

r = (pu).(u −1 q).<br />

This is cheating! This should not count as a “different” factorisation. For<br />

example in Z[i]<br />

this is (3 + 2i)(−i) × (i)(3 − 2i)<br />

13 = (3 + 2i)(3 − 2i) = (2 − 3i)(2 + 3i)<br />

We say that (3+2i) is an “associate” of (2 − 3i) because they only differ by<br />

a factor of a unit. That is, they’re not genuinely different. Likewise, (3-2i)<br />

is an “associate” of (2 + 3i).


42 Section 3. Factorisation<br />

Ho<strong>we</strong>ver the following example has genuinely different factors: check irr<br />

Definition 3.1.7. Associates<br />

6 = 2 × 3 = (1 + √ −5)(1 − √ −5)<br />

Elements s and t in a commutative ring R <strong>are</strong> said to be associates if there<br />

is a unit u such that t = us.<br />

(Note that this definition is symmetric, since t = us ⇐⇒ s = u −1 t.) <br />

Examples 3.1.8. It follows that associates <strong>are</strong> easiest to spot in rings with<br />

obvious units:<br />

1. −r is always an associate of r, as −1 is always a unit.<br />

2. In Z, the only units <strong>are</strong> ±1, so the only associates of r <strong>are</strong> ±r.<br />

This is also the case in Z[ √ d] when d < −1.<br />

3. In Z[i] the units <strong>are</strong> ±1, ±i, so the associates of a + bi <strong>are</strong><br />

−a − bi<br />

−b + ai<br />

b − ai<br />

4. In Z[ √ d] with d > 1 there <strong>are</strong> infinitely many units, so every element<br />

has infinitely many associates.<br />

5. In Z[ √ d], if s and t <strong>are</strong> associates then they have the same norm,<br />

because:<br />

t = us gives N(t) = N(u)N(s) = 1.N(s) = N(s)<br />

.....................................................................<br />

Is the converse true?<br />

No: it is possible to have the same norm but not be associates eg in<br />

Z[i] obviously 2 + 3i and 2 − 3i have the same norm, but can’t be<br />

associates as the only units <strong>are</strong> 1, −1,i, −i.


3.1 Unique factorisation 43<br />

Lemma 3.1.9. Let R be a commutative ring.<br />

If r is irreducible in R then all its associates <strong>are</strong> also irreducible in R.<br />

Proof. Let p be an associate of r, so p = ur where u is a unit.<br />

1. p cannot be a unit, since otherwise r = u −1 p would be a unit, but r is<br />

irreducible so is not a unit (by definition).<br />

2. We need to show: p = st ⇒ s or t is a unit.<br />

Now p = st means ur = st<br />

⇒ r = u −1 st = (u −1 s).t<br />

⇒ u −1 s is a unit or t is a unit (since r is irreducible)<br />

Now u −1 s is a unit ⇒ s = u.(u −1 s) is a unit.<br />

So <strong>we</strong> have s is a unit or t is a unit.<br />

Now, given any factorisation of an element s into irreducibles<br />

s = p1p2 · · · pk<br />

<strong>we</strong> could replace each pi by an associate of it, and that would be really<br />

boring. This does not count as different, just like in our example<br />

s = pq = (pu).(u −1 q).<br />

Example 3.1.10. In this example <strong>we</strong>’ll work in Z[ √ 3] and find two factori-<br />

sations of 20+7 √ 3 that look different, but <strong>are</strong> actually related by associates.<br />

That is, <strong>we</strong>’ll find irreducibles p1,p2,q1,q2 with<br />

p1p2 = q1q2 = 20 + 7 √ 3<br />

and a unit u such that q1 = p1u, q2 = u −1 p2.<br />

For the first factorisation, take p1p2 to be<br />

(1 + 2 √ 3)(2 + 3 √ 3) = 20 + 7 √ 3.<br />

We need to check that this is a factorisation into irreducibles:


44 Section 3. Factorisation<br />

• N(1 + 2 √ 3) = 1 − 4.3<br />

<br />

= 11<br />

N(1 + 2 √ 3) = ................................. ........................<br />

• N(2 + 3 √ 3) = 4 − 9.3<br />

<br />

= 23<br />

N(2 + 3 √ 3) = ................................. ........................<br />

so 1 + 2 √ 3 and 2 + 3 √ 3 <strong>are</strong> both irreducible<br />

11 and 23 <strong>are</strong> both prime (Theorem 3.1.5)<br />

because ..........................................................................................<br />

Now <strong>we</strong> pick a unit u = 7 + 4 √ 3. First <strong>we</strong> check it is a unit:<br />

N(7 + 4 √ 3) = .............1<br />

7 − 4 √ 3<br />

so 7 + 4 √ 3 is a unit, with inverse u −1 = ........................................<br />

Now consider<br />

31 + 18 √ 3<br />

q1 = p1u = (1 + 2 √ 3)(7 + 4 √ 3) = ...............................<br />

7 − 4 √ 3 −22 + 13 √ 3<br />

q2 = u −1 p2 = (...................)(2 + 3 √ 3) = ...............................<br />

20 + 7 √ 3<br />

Now check q1q2 = .............................................<br />

So <strong>we</strong> have two factorisations into irreducibles:<br />

20+7 √ 3 = (........................)(........................) = (........................)(........................)<br />

But these <strong>are</strong> not “really different” factorisations because<br />

they <strong>are</strong> related by associates<br />

......................................................................................................................<br />

We make this notion of “not really different factorisations” precise with the<br />

following definition of “equivalence” of factorisations.


3.1 Unique factorisation 45<br />

Definition 3.1.11. Equivalent factorisations<br />

Let R be a ring, r ∈ R.<br />

Let p1 · · · pn and q1 · · · qm be two factorisations of r into irreducibles.<br />

These <strong>are</strong> called equivalent if n = m and the pi’s and qi’s can be reordered<br />

so that for each i, pi and qi <strong>are</strong> associates. <br />

Definition 3.1.12. Unique factorisation domain<br />

An integral domain is called a unique factorisation domain (UFD) if<br />

1. every non-zero non-unit has a factorisation into irreducibles, and<br />

2. any two such factorisations <strong>are</strong> equivalent. <br />

Examples 3.1.13.<br />

1. Z is a UFD. This is the “fundamental theorem of arithmetic”, although<br />

personally I want to throw up every time I hear the words “fundamen-<br />

tal theorem of...”. Seriously, the fundamentality of a theorem should<br />

speak for itself, shouldn’t it?<br />

2. Q, R, C <strong>are</strong> UFD’s rather stupidly—there <strong>are</strong>n’t any non-zero non-<br />

units, so there <strong>are</strong> no elements that have to satisfy the condition.<br />

Therefore, they all do! We say the condition is “vacuously satisfied”.<br />

3. Z[ √ −5] and Z[ √ −11] <strong>are</strong> not UFDs. We’ll work through one of these<br />

examples a bit later.<br />

4. Z[i] is a UFD.<br />

5. Recall that if S ⊂ R and R is ID, then S is ID.<br />

Is the same true for UFDs? Why?<br />

No, because the irreducibles could be different—in a subring, it is likely<br />

that more things will be irreducible cf Example 3.1.3<br />

.....................................................................................................................<br />

Question: Is it going to be easier to show that something is or isn’t a UFD?<br />

To show something isn’t, <strong>we</strong> only have to exhibit one thing with non-unique<br />

factorisations. To show something is, <strong>we</strong> have to work much harder.<br />

Z[ √ −11] and Z[ √ −13] <strong>are</strong> on the homework and tutorial sheets.<br />

Now <strong>we</strong> will do Z[ √ −5].


46 Section 3. Factorisation<br />

Example 3.1.14. Z[ √ −5] is not a UFD since <strong>we</strong> have<br />

6 = 2.3 = (1 + √ −5)(1 − √ −5).<br />

We have to check that 2, 3, (1+ √ −5), (1− √ −5) <strong>are</strong> irreducible in Z[ √ −5],<br />

and <strong>are</strong> not associates. So <strong>we</strong> take norms:<br />

N(2) = 4<br />

N(3) = 9<br />

N(1 + √ −5) = N(1 − √ −5) = 1 + 5 = 6<br />

• First <strong>we</strong> show that in Z[ √ −5] there <strong>are</strong> no elements of norm 2 or 3:<br />

|a 2 + 5b 2 | = 2 =⇒ b = 0 otherwise a 2 + 5b 2 ≥ 5<br />

and similarly for norm 3.<br />

=⇒ a 2 = 2 # contradicts a ∈ Z<br />

• Next <strong>we</strong> deduce that any element of norm 4, 9 or 6 must be irreducible:<br />

If r has norm 4 it is certainly not a unit.<br />

Now suppose r = st, so N(r) = N(s)N(t).<br />

But there is no element of norm 2<br />

so <strong>we</strong> must have<br />

either N(s) = 1 and N(t) = 4, in which case s is a unit<br />

or N(s) = 4 and N(t) = 1, in which case t is a unit.<br />

A similar argument works for 9 and 6.<br />

• Next <strong>we</strong> show that these irreducibles <strong>are</strong> not associates:<br />

The units in Z[ √ −5] <strong>are</strong> just ±1, so 2 can’t be an associate of (1 ±<br />

√ −5).<br />

Also: associates have the same norm, since units have norm 1,<br />

so N(pu) = N(p).


3.2 Euclidean domains 47<br />

So <strong>we</strong> have found two non-equivalent factorisations of 6 into irreducibles in<br />

Z[ √ −5, showing that this ring is not a UFD.<br />

Example 3.1.15.<br />

Does the above example exhibit C as a non-UFD?<br />

No, because those elements <strong>are</strong> not irreducibles in C—they <strong>are</strong> units.<br />

Z[ √ −5] is a subring of C which is (vacuously) a UFD.<br />

...........................................................................<br />

In order to show that something is a UFD, it would help to know how<br />

to factorise it. In Z <strong>we</strong> have Euclid’s algorithm, and so <strong>we</strong> might ask<br />

ourselves whether <strong>we</strong> can do Euclid’s algorithm in other rings. The ans<strong>we</strong>r<br />

is: sometimes, but not always. When <strong>we</strong> can, the ring is called a Euclidean<br />

domain.<br />

3.2 Euclidean domains<br />

We usually do Euclid’s algorithm in Z≥0. How does it work?<br />

It depends on the crucial fact:<br />

∀a,b ∈ Z≥0 ∃ q,r ∈ Z≥0 s.t. a = qb + r and 0 ≤ r < b.<br />

This means that when <strong>we</strong> do the algorithm, the r’s will get smaller and<br />

smaller and will eventually have to become 0.<br />

In Numbers and Proofs you also did Euclid’s algorithm on polynomials,<br />

where in that case it was the degree of the polynomial remainder that got<br />

smaller and smaller.<br />

In general <strong>we</strong> can do Euclid’s algorithm in a ring R if there’s some way of<br />

measuring the “size” of elements, so that <strong>we</strong> can make remainders that get<br />

smaller and smaller.<br />

We’ve already seen one notion of “size” of elements of some rings:<br />

The norm in Z[ √ d].<br />

...................................................................................................


48 Section 3. Factorisation<br />

We’ll see that this does enable us to do Euclid’s algorithm on the Gaussian<br />

integers.<br />

Definition 3.2.1. Euclidean domain<br />

A Euclidean domain (ED) is an integral domain R such that there is a<br />

function<br />

such that for all a,b ∈ R \ {0}<br />

δ : R \ {0} −→ Z≥0<br />

1. ∃q,r ∈ R s.t. a = qb + r and r = 0 or δ(r) < δ(b)<br />

2. if b|a in R then δ(b) ≤ δ(a).<br />

Such a function is called a Euclidean function.<br />

recall b|a means ∃k ∈ R s.t. a = kb <br />

Examples 3.2.2.<br />

1. Z is ED with δ(a) = |a|.<br />

2. R[x] is ED with δ(f) = deg(f).<br />

3. Z[i] is ED with δ(r) = N(r) = |r| 2<br />

(remembering Z[i] ⊂ C).<br />

We won’t prove these, but <strong>we</strong>’ll look at some examples. The point is that in<br />

an ED <strong>we</strong> can use Euclid’s algorithm to find highest common factors, and<br />

do that ar + bs = h thing which is kind of fiddly and annoying but useful.<br />

Definition 3.2.3. Highest common factor<br />

Let R be an integral domain, a,b ∈ R.<br />

Then h is a highest common factor (hcf) of a and b if<br />

1. h|a and h|b ←− “it is a common factor”<br />

2. if d|a and d|b then d|h ←− “it is the highest one” <br />

Remarks 3.2.4.<br />

1. Hcf’s don’t necessarily exist if you’re not in a UFD.<br />

eg in Z[ √ −3: 4 = 2.2 = (1 + √ −3)(1 − √ −3) and (1 + √ −3).2


3.2 Euclidean domains 49<br />

2. Hcf’s <strong>are</strong>n’t unique: if h is one hcf and u is a unit, then uh is also an<br />

hcf. Ho<strong>we</strong>ver in an integral domain all hcf’s of a given pair of elements<br />

must be associates.<br />

Example 3.2.5. Hcf in R[x].<br />

We use Euclid’s algorithm to find the hcf of<br />

x 3 + 7x 2 + 14x + 8 and x 3 + 6x 2 + 11x + 6 ∈ R[x].<br />

x 3 + 7x 2 + 14x + 8 = 1.(x 3 + 6x 2 + 11x + 6) + (x 2 + 3x + 2)<br />

x 3 + 6x 2 + 11x + 6 = (x + 3)(x 2 + 3x + 2)<br />

So the hcf is x 2 + 3x + 2.<br />

Example 3.2.6. Hcf in Z[i].<br />

(This is extremely tedious and you should feel very glad that I’m not actually<br />

going to ask you to do it. This is probably the kind of thing it would be<br />

better to ask Maple to do, but I don’t believe in getting computers to do<br />

things for me—if I can’t do it myself I just do something more interesting<br />

instead! It’s possible I will get into trouble for saying that.)<br />

17+43i<br />

11+3i<br />

11+3i<br />

4+4i<br />

17+43i = 11+3i .11−3i<br />

11−3i<br />

11+3i = 4+4i .4−4i<br />

4−4i<br />

in C in Z[i] in Z[i]<br />

= 316<br />

130<br />

422 + 130i −→ 17 + 43i = (2 + 3i)(11 + 3i) + (4 + 4i)<br />

↑ ↑<br />

∼ 2.4 ∼ 3.2<br />

= 56<br />

32<br />

↑<br />

∼ 1.75<br />

So hcf is −1 − i or indeed 1 + i.<br />

Check that 1 + i really goes into both of those:<br />

and:<br />

17+43i<br />

1+i<br />

11+3i<br />

1+i<br />

17+43i = 1+i . 1+i 60<br />

1−i = 2<br />

11+3i = 1+i .1−i<br />

1−i = 7 + 4i.<br />

32 − 32i −→ 11 + 3i = (20 − i)(4 + 4i) + (−1 − i)<br />

26 + 2 i = 30 + 13i<br />

−→ 4 + 4i = −4(−1 − i)


50 Section 3. Factorisation<br />

Have you any idea how long it took to type that one example?<br />

two and a half hours.<br />

Ans<strong>we</strong>r: .................................<br />

One of the main points is to show that if <strong>we</strong> have Euclid’s algorithm then<br />

<strong>we</strong> have a UFD. In order to do this <strong>we</strong> need to know:<br />

1. In an ED, any a and b have an hcf.<br />

Moreover <strong>we</strong> have ar + bs = h for some r,s ∈ R.<br />

(The proof of this is exactly the same as in Z —<strong>we</strong> feed the numbers<br />

back into Euclid’s algorithm backwards.)<br />

2. In an ED, if p is irreducible then hcf(a,p) is 1 or p<br />

(or u or up where u is a unit).<br />

3. In an ED, if p is irreducible then<br />

i.e. “irreducible ⇒ prime”.<br />

—this follows from (1) and (2)<br />

p|ab ⇒ p|a or p|b<br />

4. In an ED, if r = ab for non-units a and b then<br />

δ(a) < δ(r)<br />

δ(b) < δ(r)<br />

strictly less—would only be equal if the other element is a unit<br />

We can then show that everything is a product of irreducibles, by contra-<br />

diction: suppose not, and take the “smallest” element that is not a product<br />

of irreducibles. In particular it is not irreducible—so it can be expressed as<br />

a product of “smaller” things. Contradiction. Boom. We then use (2) to<br />

show that the factorisation is unique.<br />

Theorem 3.2.7.<br />

Let R be a Euclidean Domain, p irreducible in R, a ∈ R.<br />

Then either 1 or p is an hcf of a and p.


3.2 Euclidean domains 51<br />

Proof.<br />

• Suppose p|a. Then p is an hcf of a and p<br />

since clearly p|a and p|p<br />

and {d|a and d|p} ⇒ d|p<br />

• Suppose p ∤ a. Then claim: 1 is an hcf of a and p.<br />

Clearly 1|a and 1|p.<br />

Now suppose d|a and d|p<br />

i.e. a = k1d and p = k2d for some k1,k2 ∈ R.<br />

But p irreducible =⇒ k2 is a unit or d is a unit.<br />

If k2 is a unit then d = k −1<br />

2 p<br />

so a = k1k −1<br />

2 p<br />

and thus p|a # so k2 is not a unit.<br />

So d must be a unit, in which case d|1<br />

hence 1 is an hcf as claimed. <br />

Theorem 3.2.8. “In an ED, irreducible implies prime”<br />

Let R be a Euclidean Domain, p irreducible in R, a,b ∈ R.<br />

Then p|ab =⇒ p|a or p|b<br />

Proof.<br />

Suppose p|ab and p ∤ a. We need to show p|b.<br />

Now, by Theorem 3.2.7 <strong>we</strong> know 1 must be an hcf of a and p.<br />

So <strong>we</strong> have ar + ps = 1 for some r,s ∈ R.<br />

Now multiplying by b gives bar + bps = b.<br />

But p|ab so p|abr + bps<br />

i.e. p|b. <br />

Definition 3.2.9. Prime<br />

Let R be a commutative ring. An element p ∈ R is called prime if<br />

p|ab =⇒ p|a or p|b.


52 Section 3. Factorisation<br />

Theorem 3.2.10.<br />

Let R be a Euclidean domain.<br />

If a = st for non-units s,t then<br />

δ(s) < δ(a)<br />

δ(t) < δ(a)<br />

Note that the definition gives us ≤ immediately, but <strong>we</strong> want strictly


3.2 Euclidean domains 53<br />

Now a cannot be irreducible, by hypothesis<br />

so <strong>we</strong> must have a = rs where r,s <strong>are</strong> non-units.<br />

But then by Theorem 3.2.10 <strong>we</strong> have<br />

δ(r) < δ(a)<br />

δ(s) < δ(a)<br />

By (1) this means r and s have factorisation into irreducibles<br />

r = r1r2 · · · rk<br />

s = s1s2 · · · sl<br />

But then a = r1r2 · · · rk.s1s2 · · · sl and is factorised. #.<br />

So every non-unit has at least one factorisation into a product of irreducibles.<br />

Next <strong>we</strong> show uniqueness of factorisations:<br />

Suppose there <strong>are</strong> non-equivalent factorisations of a.<br />

Choose the smallest n with<br />

a = p1p2 · · · pn = q1q2 · · · qm<br />

non-equivalent factorisations into irreducibles.<br />

Now pn|q1q2 · · · qm<br />

so by Theorem 3.2.8 pn must divide one of the qi.<br />

Wlog pn|qm.<br />

But qm is irreducible so <strong>we</strong> must have qm = upn for some unit u.<br />

Now <strong>we</strong> have<br />

p1p2 · · · pn−1pn = q1q2 · · · qm−1(u.pn)<br />

so p1p2 · · · pn−1 = q1q2 · · · qm−1u<br />

(We can cancel pn since R is a Euclidean domain, which means in particular<br />

it is an integral domain.)<br />

But these <strong>are</strong> both factorisations into irreducibles so must be equivalent,<br />

since n is the smallest n with non-equivalent factorisations into irreducibles.


54 Section 3. Factorisation<br />

But if these <strong>are</strong> equivalent then<br />

p1p2 · · · pn−1pn = q1q2 · · · qm−1(u.pn)<br />

must also be equivalent. #


Part II<br />

Groups<br />

4 Revision<br />

All of this section is supposedly “revision” from Groups and Symmetries. If<br />

you’re not sure about this material then you should revise it.<br />

4.1 Definitions and examples<br />

Definition 4.1.1. Group<br />

A group G is a set equipped with a binary operation . satisfying<br />

1. associativity ∀a,b,c ∈ G (ab)c = a(bc)<br />

2. identity ∃e ∈ G s.t. ∀g ∈ G g.e = g = e.g<br />

3. additive inverse ∀g ∈ G ∃g −1 ∈ G s.t. g.g −1 = g −1 .g = e<br />

Definition 4.1.2. Abelian group<br />

A group G is called Abelian if for all g,h ∈ G <strong>we</strong> have gh = hg. <br />

Definition 4.1.3. Order<br />

The order of a group G is: |G| = the number of elements of G<br />

The order of an element g ∈ G is the smallest natural number k > 0 such<br />

that g n = e.<br />

Examples 4.1.4. Permutation groups<br />

Sn is the group of permutations of n elements.<br />

55


56 Section 4. Revision<br />

|Sn| = n! HHH 1 : n, 2 : n!, 3 : n!<br />

2 , 4 : 2n, 5 : infinite<br />

This is also called the nth symmetric group.<br />

Recall that there <strong>are</strong> two different types of notation:<br />

• two-row notation ⎛<br />

⎞<br />

1 2 3 4 5 6<br />

⎝ ⎠<br />

2 4 5 1 3 6<br />

• disjoint cycle notation:<br />

(124)(35)<br />

This element has cycle type: 3,2<br />

Recall that every element can be written as a product of transpositions:<br />

(12)(24)(35)<br />

These <strong>are</strong> no longer necessarily disjoint.<br />

(12)(24)(35)(16)(16)<br />

Also, this isn’t unique e.g. ..............................................<br />

but all expressions for the same element will have the same parity.<br />

If it’s even, the element is called an even permutation.<br />

If it’s odd, the element is called an odd permutation.<br />

Do the even permutations form a subgroup of Sn? yes<br />

Do the odd permutations form a subgroup of Sn? no<br />

The alternating group An is the subgroup of Sn given by:<br />

the even permutations<br />

...........................................................................<br />

|An| = n!<br />

2 HHH<br />

Is Sn Abelian? No for n > 2. HHH<br />

(12)(23) = (123)<br />

(23)(12) = (132)


4.1 Definitions and examples 57<br />

Definition 4.1.5. Subgroup<br />

Let G be a group.<br />

A subgroup of G is a subset H ⊂ G such that H is a group with the same<br />

operation. <br />

Example 4.1.6. Cyclic groups<br />

The cyclic group of order n consists of elements<br />

with a n = 1.<br />

1,a,a 2 ,...a n−1<br />

Any group of order n is cyclic if it has an element of order n.<br />

Examples 4.1.7. Symmetry groups<br />

1. Dn is the group of symmetries of a regular n-gon.<br />

(n = 1,2 <strong>are</strong> slightly peculiar though.)<br />

|Dn| = 2n HHH<br />

It consists of ........... rotations and ............ reflections n and n HHH<br />

The group is generated by one rotation and one reflection.<br />

The rotations form a subgroup of Dn isomorphic to Cn.<br />

2. O2 is the group of symmetries of the circle.<br />

|O2| = ∞ HHH<br />

There <strong>are</strong> rotations rotα<br />

and reflections refα<br />

α<br />

This group can be generated from one reflection and all the rotations.<br />

α<br />

2


58 Section 4. Revision<br />

3. SO2 is the rotation group of the circle.<br />

SO2 is a subgroup of O2.<br />

4. Note that<br />

D1 ∼ = C2 ∼ = S2<br />

D3 ∼ = S3<br />

5. The smallest non-abelian group is D3 = S3.<br />

|D3| = 6 HHH<br />

Examples 4.1.8. Groups of matrices<br />

Do the n × n real matrices form a group under multiplication?<br />

No—<strong>we</strong> might not have inverses. HHH<br />

1. GLn(R) = the group of invertible real n × n matrices under multi-<br />

plication<br />

similarly <strong>we</strong> have GLn(F) for any field F.<br />

This is called the general linear group<br />

2. SLn(F) = matrices of determinant 1 with entries in a field F.<br />

This is a subgroup of GLn(F) and is called the special linear group.<br />

4.2 Basic theory<br />

Definition 4.2.1. Orders of elements<br />

Recall that the order of g ∈ G is the least n ∈ N such that g n = e if such<br />

an n exists; otherwise g has infinite order. <br />

Examples 4.2.2.<br />

1. What is the order of rot2π<br />

3<br />

What is the order of rot2π<br />

n<br />

in O2? 3 HHH<br />

in O2? n HHH<br />

What is the order of refα in O2? 2 HHH<br />

2. What is the order of a 3 in C12? 4 HHH<br />

What is the order of a 4 in C6? 3 HHH


4.2 Basic theory 59<br />

What is the order of a m in Cn?<br />

n<br />

hcf(m,n)<br />

explain!! HHH<br />

3. Every element g ∈ G generates a cyclic subgroup of G:<br />

〈g〉 = {1,g 2 ,... ,g n−1 }<br />

The order of this subgroup is the order of g.<br />

Definition 4.2.3. Cosets maltesers!!<br />

Let G be a group, a ∈ G and H a subgroup of G<br />

The left coset aH is given by<br />

aH = {ah | h ∈ H}<br />

and is a subset of elements of G (not a subgroup).<br />

The right coset Ha is given by<br />

Ha = {ha | h ∈ H}.<br />

Can <strong>we</strong> have aH = bH if a = b? yes HHH<br />

aH = bH ⇐⇒ a ∈ bH or equivalently b ∈ aH<br />

⇐⇒ ∃h ∈ H s.t. a = bh or equivalently b = ah<br />

⇐⇒ b −1 a ∈ H comp<strong>are</strong> b − a ∈ nZ for Zn<br />

For example if H = 〈a 2 〉 ⊂ C6 with C6 = {1,a,a 2 ,...a 5 } then aH is the<br />

same as<br />

a 3 H,a 5 H<br />

..........................................................................<br />

Do <strong>we</strong> have aH = Ha? In general no HHH If it’s equal for all a then it’s<br />

a normal subgroup.<br />

Example 4.2.4.<br />

Let G = S3, H = A3 ⊂ S3.<br />

How many left cosets <strong>are</strong> there? 2 HHH


60 Section 4. Revision<br />

Note:<br />

• |aH| = |H| for all a ∈ G.<br />

• Every element of G is in some coset since g ∈ gH.<br />

• a ∈ bH ⇐⇒ aH = bH.<br />

So if |G| is finite, the cosets form a partition of G into equal sized sets, which<br />

gives us Lagrange’s Theorem:<br />

Theorem 4.2.5. Lagrange<br />

For any group G and subgroup H<br />

Corollary 4.2.6.<br />

|H| | |G|.<br />

Given any g ∈ G, the order of g divides |G|<br />

since the order of g is the order of 〈g〉 ⊂ G.<br />

Corollary 4.2.7.<br />

If |G| is prime then G must be cyclic.<br />

Proof.<br />

Let |G| = p prime and let g be any non-identity element.<br />

Then by Corollary 4.2.6 <strong>we</strong> know that the order of g must divide p<br />

so must be 1 or p since p is prime.<br />

But g is not the identity, so the order of g cannot be 1<br />

so the order of g is p, showing that G is cyclic. <br />

Definition 4.2.8. index<br />

The index of H in G is |G|<br />

|H| . <br />

Definition 4.2.9. Group actions<br />

A group G acts on a (non-empty) set X if <strong>we</strong> have for each g ∈ G a function<br />

X −→ X<br />

x ↦→ g ∗ x


4.2 Basic theory 61<br />

interacting properly with the group structure, i.e.<br />

Example 4.2.10.<br />

1. e ∗ x = x for all x ∈ X<br />

2. g ∗ (h ∗ x) = gh ∗ x for all g,h ∈ G, x ∈ X <br />

D4 acts on the vertices of the squ<strong>are</strong>.<br />

D4 also acts on the edges of the squ<strong>are</strong>.<br />

Definition 4.2.11. Orbit and stabiliser<br />

Given an action of G on X as above, the orbit and stabiliser of x ∈ X <strong>are</strong><br />

given by:<br />

• orb(x) = { g ∗ x | g ∈ G } ⊂ X<br />

• stab(x) = { g ∈ G | g ∗ x = x } ⊂ G means subgroup<br />

Note that<br />

orb(x) = orb(y) ⇐⇒ x ∈ orb(y)<br />

so <strong>we</strong> get an equivalence relation ∼ on X<br />

x ∼ y ⇐⇒ they <strong>are</strong> in the same orbit.<br />

Theorem 4.2.12. Orbit-stabiliser<br />

Let G be a finite group acting on a set X. Then for any x ∈ X<br />

|orb(x)|.|stab(x)| = |G|<br />

Example 4.2.13. D4 acting on the squ<strong>are</strong><br />

Recall that D4 acts on the corners of the squ<strong>are</strong>:<br />

4<br />

Then orb(1) = {1,2,3,4} HHH<br />

stab(1) = {e,ref13} HHH<br />

So <strong>we</strong> can check the orbit-stabiliser theorem holds:<br />

3<br />

2<br />

1<br />

<br />

4.2 = 8 = |D4|<br />

......................


62 Section 4. Revision<br />

Note that there is a more delicate version of this theorem which also works<br />

for infinite groups. It involves cosets. Cosets <strong>are</strong> traditionally unpopular<br />

among students.<br />

orb(x)<br />

∼<br />

−→ cosets of stab(x)<br />

y ↦→ {g ∈ G | g ∗ x = y} “all ways of sending x to y”<br />

4.3 Homomorphisms<br />

Definition 4.3.1. Homomorphism<br />

A group homomorphism is a function<br />

such that for all x,y ∈ G<br />

θ : G −→ H<br />

θ(xy) = θ(x).θ(y)<br />

Note: it follows that θ(e) = e and θ(x −1 ) = (θ(x)) −1 . <br />

θ(x).θ(e) = θ(x) so θ(e) = e and θ(x).θ(x −1 ) = θ(e) = e<br />

Definition 4.3.2. Group isomorphism<br />

A group isomorphism θ : G −→ H is a homomorphism that is bijective.<br />

Equivalently: there exists a homomorphism φ : H −→ G such that<br />

Examples 4.3.3.<br />

φ ◦ θ = idH<br />

θ ◦ φ = idG. <br />

1. If H is a subgroup of G, the inclusion H −→ G is a homomorphism.<br />

2. Do <strong>we</strong> have a homomorphism yes HHH<br />

Z −→ Zn<br />

a ↦→ a (mod n) ?


3. Do <strong>we</strong> have a homomorphism no HHH<br />

Zn −→ Z<br />

a (mod n) ↦→ a ?<br />

No—just like in the ring case, because addition isn’t the same<br />

eg (n − 1) + 1 = 0<br />

4. How many group homomorphisms <strong>are</strong> there HHH 1, 2, 3, 4, 5:many<br />

Ans<strong>we</strong>r: one for each a ∈ Z<br />

Z<br />

θ<br />

−→ Z ?<br />

1 −→ a<br />

n ↦→ na = a + a + · · · + a or θ(n) = an<br />

There is always a group homomorphism G −→ 0 = {e}.<br />

Also, given any groups G,H there is always a trivial homomorphism<br />

5 Quotient groups<br />

G −→ H<br />

g ↦→ e<br />

We’re going to see that <strong>we</strong> can sort of multiply and divide groups under the<br />

right circumstances.<br />

5.1 Subgroups<br />

Definition 5.1.1. Subgroup<br />

H is a subgroup of a group G if it is a subset of G and a group under the<br />

same operation. Equivalently:<br />

1. for all a,b ∈ H <strong>we</strong> have ab ∈ H,<br />

2. e ∈ H, and<br />

3. for all a ∈ H <strong>we</strong> have a −1 ∈ H.<br />

We write H ⊆ G unless confusion is likely. <br />

63


64 Section 5. Quotient groups<br />

Examples 5.1.2.<br />

1. Z ⊂ Q ⊂ R ⊂ C under addition<br />

2. Is Zn a subgroup of Z? No, just like for rings.<br />

3. An ⊂ Sn<br />

Cn ⊆ Sn<br />

Dn ⊆ Sn<br />

Cn ⊆ Dn<br />

Dn ⊆ O2<br />

4. Is D3 ⊆ D4? No: 6 ∤ 8<br />

Is S3 ⊆ S4? Yes: always for n < m, Sn ⊂ Sm<br />

5. SLn(F) ⊆ GLn(F)<br />

6. The trivial group 0 = {e} is a subgroup of every group.<br />

5.2 Cosets<br />

Definition 5.2.1. Cosets<br />

Let H be a subgroup of G.<br />

Then the left coset aH is the set: { ah | h ∈ H }.<br />

The right coset Ha is the set: { ha | h ∈ H }. <br />

For example if H = {e,h1,h2,h3} then<br />

aH = {a,ah1,ah2,ah3}<br />

bH = {b,bh1,bh2,bh3}<br />

but now <strong>we</strong> actually multiply those elements out.


5.2 Cosets 65<br />

Example 5.2.2.<br />

Consider the cyclic group C12 with elements {e,a,a 2 ,a 3 ,... a 11 }<br />

so a 12 = a 0 = e.<br />

Let H = 〈a 4 〉<br />

so H has elements ............................................ e,a 4 ,a 8<br />

So <strong>we</strong> have the following cosets:<br />

eH = {e,a 4 ,a 8 } a 6 H = {a 6 ,a 10 ,a 2 }<br />

aH = {a,a 5 ,a 9 } a 7 H = {a 7 ,a 11 ,a 3 }<br />

a 2 H = {a 2 ,a 6 ,a 10 } a 8 H = {a 8 ,e,a 4 }<br />

a 3 H = {a 3 ,a 7 ,a 11 } a 9 H = {a 9 ,a,a 5 }<br />

a 4 H = {a 4 ,a 8 ,e } a 10 H = {a 10 ,a 2 ,a 6 }<br />

a 5 H = {a 5 ,a 9 ,a } a 11 H = {a 11 ,a 3 ,a 7 }<br />

|C12| = ...................12 |H| = ....................3<br />

So the number of distinct cosets = ................ |C12 |<br />

|H|<br />

The following cosets <strong>are</strong> the same:<br />

eH = a 4 H = a 8 H<br />

aH = a 5 H = a 9 H<br />

a 2 H = a 6 H = a 10 H<br />

a H = a 7 H = a 11 H<br />

= 4


66 Section 5. Quotient groups<br />

Example 5.2.3. Cosets in D3<br />

Recall that D3 is the symmetry group of the equilateral triangle:<br />

1<br />

a is rot anticlockwise, bi is ref through line through i (draw in dotted)<br />

We have elements: e,a,a 2 ,b1,b2,b3<br />

Note b2 = ab1<br />

b3 = a 2 b1.<br />

We have a subgroup H = {e,b1}.<br />

Now let’s find the left cosets:<br />

2<br />

eH = {e,b1}<br />

aH = {a,ab1} = {a,b2}<br />

a 2 H = {a 2 ,a 2 b1} = {a 2 ,b3}<br />

b1H = {b1,e}<br />

So the following cosets <strong>are</strong> equal:<br />

b2H = {b2,b2b1} = {b2,a}<br />

b3H = {b3,b3b1} = {b3,a 2 }<br />

3<br />

H = b1H<br />

aH = b2H<br />

a 2 H = b3H<br />

The number of distinct cosets is .............................. 3


5.2 Cosets 67<br />

Lemma 5.2.4.<br />

Let H be a subgroup of G and a ∈ G.<br />

Then the coset aH has the same number of elements as H.<br />

Proof.<br />

Certainly aH can’t have more elements that H.<br />

It can only have fe<strong>we</strong>r if ah = ah ′ for some h = h ′ ∈ H.<br />

But ah = ah ′ =⇒ h = h ′ . <br />

Lemma 5.2.5.<br />

Let H be a subgroup of G.<br />

If there is an element x ∈ G such that x ∈ aH and x ∈ bH<br />

then aH = bH.<br />

Proof.<br />

x ∈ aH so x = ah1, say.<br />

x ∈ bH so x = bh2, say.<br />

Then a = bh2h −1<br />

1 . (1)<br />

We show aH ⊆ bH: y ∈ aH =⇒ y = ah for some h ∈ H<br />

Similarly bH ⊆ aH<br />

=⇒ y = bh2h −1<br />

1 .h by (1)<br />

=⇒ y ∈ bH<br />

so aH = bH as required.


68 Section 5. Quotient groups<br />

Corollary 5.2.6. ABSOLUTELY KEY, MUST REMEMBER<br />

Let H be a subgroup of G and a,b ∈ G.<br />

Then aH = bH if and only if any one of the following equivalent conditions<br />

holds:<br />

• b ∈ aH • a ∈ bH<br />

• ∃ h ∈ H s.t. b = ah • ∃ h ∈ H s.t. a = bh<br />

• a −1 b ∈ H • b −1 a ∈ H<br />

We can then define an equivalence relation on the elements of G:<br />

a ∼ b ⇐⇒ aH = bH.<br />

Note that for right cosets it’s all the other way round:<br />

• b ∈ Ha • a ∈ Hb<br />

• ∃ h ∈ H s.t. b = ha • ∃ h ∈ H s.t. a = hb<br />

• ba −1 ∈ H • ab −1 ∈ H<br />

Example 5.2.7. Modular arithmetic bring sheet with nos<br />

Let G = Z.<br />

Put H = 4Z = { all multiples of 4 ∈ Z }<br />

= { k ∈ Z s.t. 4|k }<br />

= { 4n | n ∈ Z }<br />

= { ... , −8, −4,0,4,8,12,... }<br />

This is an additive group so the cosets <strong>are</strong> a + H.<br />

e.g. 1 + H = {... , −7, −3,1,5,9,13,...} = { k | k ≡ 1 (mod 4) }<br />

2 + H = {... , −6, −2,2,6,10,14,...} = { k | k ≡ 2 (mod 4) }<br />

3 + H = {... , −5, −1,3,7,11,14,...} = { k | k ≡ 3 (mod 4) }<br />

4 + H = H<br />

5 + H = 1 + H


5.2 Cosets 69<br />

So 1 + H = 5 + H = 9 + H = 13 + H = · · ·<br />

So a ∼ b ⇐⇒ a + H = b + H<br />

⇐⇒ b − a ∈ H<br />

⇐⇒ 4 | b − a<br />

⇐⇒ a ≡ b (mod 4)<br />

So <strong>we</strong> can achieve modular arithmetic by faffing around with cosets. Of<br />

course, that’s a bit over the top for modular arithmetic, but it shows us how<br />

<strong>we</strong> can generalise the idea of modular arithmetic to other (less obvious)<br />

groups. This is called quotient groups. In the above example <strong>we</strong> started<br />

with Z and “quotiented” (divided) by 4Z. In general <strong>we</strong> can quotient by<br />

any normal subgroup.<br />

Definition 5.2.8. Normal subgroup<br />

A subgroup H ⊆ G is a normal subgroup if<br />

equivalently: be ready to explain this<br />

∀ a ∈ G aH = Ha<br />

∀ a ∈ G, h ∈ H aha −1 ∈ H.<br />

Informally “H is stable under conjugation”.<br />

We write H G. <br />

Examples 5.2.9.<br />

1. If G is abelian then all subgroups <strong>are</strong> normal since<br />

aha −1 = aa −1 h = h ∈ H<br />

.....................................................................<br />

2. Any group G has normal subgroups 0 and G.<br />

3. An Sn (See Tutorial #7, q.1)<br />

4. Cn Dn but C2 is not. Recall from Example 5.2.3:<br />

H = {e,b1} ∼ = C2


70 Section 5. Quotient groups<br />

use b1a = b3,b1a 2 = b2<br />

eH = {e,b1} He = {e,b1}<br />

aH = {a,ab1} = {a,b2} Ha = {a,b1a} = {a,b3}<br />

a 2 H = {a 2 ,a 2 b1} = {a 2 ,b3} Ha 2 = {a 2 ,b1a 2 } = {a 2 ,b2}<br />

b1H = {b1,e} Hb1 = {b1,e}<br />

b2H = {b2,b2b1} = {b2,a} Hb2 = {b2,b1b2} = {b2,a 2 }<br />

b3H = {b3,b3b1} = {b3,a 2 } Hb3 = {b3,b1b3} = {b3,a}<br />

So the left cosets do not equal the right cosets, so H is not a normal<br />

subgroup.<br />

5. In fact if H is a subgroup of index 2 then it must be a normal sub-<br />

group. (See Tutorial #7, q.1)<br />

Later <strong>we</strong>’ll see that homomorphisms give rise to normal subgroups in an<br />

important way.<br />

Question: Which makes you feel more like a rat up a drainpipe—cosets or<br />

equivalence relations?<br />

5.3 Quotient groups<br />

Quotient groups <strong>are</strong> a really important piece of mathematics. I would say<br />

they <strong>are</strong> the most important and beautiful piece of theory in this course<br />

(though of course others might disagree).<br />

The point is that if <strong>we</strong> have a normal subgroup H G then the cosets form<br />

a group. Later <strong>we</strong>’ll see how to think of this as an equivalence relation.<br />

Definition 5.3.1. Quotient group<br />

Let H G. Then the cosets of H form a group called the quotient group<br />

G/H.


5.3 Quotient groups 71<br />

• The elements <strong>are</strong> the cosets aH.<br />

Remember g1H = g2H ⇐⇒ g −1<br />

2 g1 ∈ H.<br />

How many elements <strong>are</strong> there?<br />

• We define the group operation by<br />

• The identity element is eH = H.<br />

We have to check this makes sense i.e.<br />

|G/H| = |G|<br />

|H|<br />

(aH).(bH) = (ab)H<br />

if a1H = a2H and b1H = b2H (1)<br />

then <strong>we</strong> must have (a1b1)H = (a2b2)H. (2)<br />

Once <strong>we</strong> have checked this, the group axioms follow immediately.<br />

Now, by Corollary 5.2.6 <strong>we</strong> know that (1) means a −1<br />

2 a1 ∈ H and b −1<br />

2 b1 ∈ H.<br />

Also (2) means (a2b2) −1 (a1b1) ∈ H.<br />

Now (a2b2) −1 (a1b1) = b −1<br />

2 a−1<br />

2 .a1b1<br />

= b −1<br />

2 (a −1<br />

2 a1)<br />

<br />

∈H<br />

b2<br />

<br />

∈H<br />

. b −1<br />

2 b1<br />

<br />

∈H<br />

trick<br />

Note that this crucially depended on H being a normal subgroup. <br />

Examples 5.3.2.<br />

1. nZ = { nk | k ∈ Z }<br />

This is a normal subgroup of Z since Z is abelian.<br />

We formed the quotient group before: Z/nZ = Zn<br />

2. Sn/An ∼ = C2<br />

Dn/Cn ∼ = C2.<br />

3. R 2 /R gives us lines in R 2 draw pic


72 Section 6. Conjugacy<br />

4. R 3 /R gives us lines in R 3 draw pic<br />

5. We know that 0 G for any group. So put H = 0.<br />

{a}<br />

Then for any a ∈ G the coset aH is ................................<br />

So the quotient group G/0 is G<br />

.......................................<br />

6. G/G = 0<br />

.............<br />

Here’s something else <strong>we</strong> won’t really be using, but it’s included for com-<br />

pleteness. It’ll be very important in the future.<br />

Definition 5.3.3. Product Given groups A and B <strong>we</strong> can form a new<br />

group A × B as follows.<br />

• The elements <strong>are</strong> { (a,b) | a ∈ A,b ∈ B }<br />

• Multiplication is “pointwise”: (a1,b1).(a2,b2) = (a1a2,b1b2). <br />

If it’s an additive group <strong>we</strong> have<br />

(a1,b1) + (a2,b2) = (a1 + a2,b1 + b2)<br />

—does this remind you of anything? R × R = R 2<br />

6 Conjugacy<br />

Conjugacy is a relationship bet<strong>we</strong>en elements that means they behave in<br />

similar ways.<br />

E.g. getting past someone into a seat.


6.1 Definitions 73<br />

6.1 Definitions<br />

Definition 6.1.1. Conjugate<br />

Let G be a group and a,g ∈ G.<br />

The conjugate of a by g is gag −1 . <br />

Where have you seen conjugates before?<br />

Linear: conjugate matrices, diagonalisation etc<br />

..........................................................................<br />

Examples 6.1.2.<br />

1. In an Abelian group gag −1 = a<br />

so everything is only conjugate to itself.<br />

2. In a dihedral group and O2, conjugating a rotation through α by any<br />

reflection gives rotation through −α.<br />

actually this is the defining property of dihedral groups<br />

3. In Sn, conjugates <strong>are</strong> all those elements of the same cycle type.<br />

Lemma 6.1.3.<br />

Given a,g ∈ G, the order of gag −1 equals the order of a.<br />

Proof.<br />

Suppose the order of a is n, so n is the smallest natural number such that<br />

a n = e.<br />

First <strong>we</strong> check that (gag −1 ) n = e.<br />

Now<br />

(gag −1 ) n = gag −1 .gag −1 .... .gag −1<br />

<br />

n times<br />

= ga n g −1<br />

= geg −1<br />

= e


74 Section 6. Conjugacy<br />

Now <strong>we</strong> check that n is the smallest natural number such that<br />

(gag −1 ) n = e.<br />

Suppose there is a smaller one i.e. 0 < k < n such that (gag −1 ) k = e.<br />

Then<br />

(gag −1 ) k = ga k g −1<br />

= e<br />

so a k = g −1 g<br />

So such a k cannot exist.<br />

= e # contradicts the order of a being n<br />

So the order of gag −1 is n as required. <br />

Definition 6.1.4. Conjugacy class, centraliser<br />

Let G be a group and a ∈ G.<br />

The conjugacy class of a in G is<br />

The centraliser of a in G is<br />

conj G(a) = “all those elements conjugate to a”<br />

= { gag −1 | g ∈ G }<br />

centG(a) = “all those elements that commute with a<br />

= { g ∈ G | gag −1 = a } <br />

Does this remind you anything? group actions, orbit/stabiliser


6.1 Definitions 75<br />

Theorem 6.1.5. “Conjugation is a group action”<br />

Defining<br />

g ∗ a = gag −1 ∀g ∈ G<br />

gives a group action of G on its set of elements.<br />

Then for all a ∈ G: orb(a) = conj G(a)<br />

stab(a) = centG(a)<br />

So if G is finite, by the orbit-stabiliser theorem <strong>we</strong> have<br />

|conj G(a)| . |centG(a)| = |G|.<br />

.........................................................................<br />

In particular<br />

|conj G(a)| | |G|<br />

and <strong>we</strong> can partition G into conjugacy classes. not all equal sizes<br />

Also, it follows that centG(a) is a subgroup of G.<br />

Proof.<br />

eae −1<br />

1.∀a ∈ G : e ∗ a = ...............................<br />

= a tick<br />

g ∗ (hah −1 )<br />

2.∀g,h,a ∈ G : g ∗ (h ∗ a) = ............................... substitute definition of h∗<br />

g(hah −1 )g −1<br />

= ............................... substitute definition of g∗<br />

(gh).a.(gh) −1<br />

= ............................... re-associate, and<br />

use definition of (gh) −1<br />

= (gh) ∗ a


76 Section 6. Conjugacy<br />

Theorem 6.1.6. Conjugacy classes in Sn<br />

Given an element a ∈ Sn, the conjugacy class of a consists of all elements<br />

of the same cycle type as a.<br />

We won’t quite prove it, but <strong>we</strong> will show how to “do” it.<br />

Example 6.1.7.<br />

In S6 consider a = (123)(45).<br />

The conjugacy class is all elements of cycle type 3,2.<br />

(213)(36), (314)(25),...<br />

E.g. .....................................................................<br />

How many such elements <strong>are</strong> there?<br />

6.5.4<br />

3<br />

So <strong>we</strong> have |conjS6 (a)| = 120<br />

. 3 = 120<br />

6! 6!<br />

|centS6 (a)| = = = 3.2 = 6<br />

120 6.5.4<br />

In the next example <strong>we</strong>’ll actually exhibit the conjugacy.


6.1 Definitions 77<br />

Example 6.1.8. Conjugacy in S9<br />

Consider α = ( 1 9 6 3 )( 2 4 8 )( 5 7 ) ∈ S9<br />

β = ( 3 1 )( 9 2 4 7 )( 6 5 8 ) ∈ S9<br />

We claim that these <strong>are</strong> conjugate.<br />

To show this, <strong>we</strong> need to find θ ∈ S9 s.t. β = θαθ −1 .<br />

1. Line up the two permutations by re-ordering the disjoint cycles:<br />

α = 1 9 6 3 2 4 8 5 7 <br />

β =<br />

(9247)(658)(31)<br />

<br />

2. Re-order the vertical pairs of number to give a standard two-row no-<br />

tation:<br />

967534182<br />

θ =<br />

⎛<br />

⎜<br />

⎝<br />

3. Check your ans<strong>we</strong>r:<br />

i) Write θ in cycle notation:<br />

ii) Write θ −1 in cycle notation:<br />

1 2 3 4 5 6 7 8 9<br />

⎞<br />

⎟<br />

⎠<br />

(19264537)<br />

.........................................<br />

(73546291)<br />

.........................................<br />

iii) Calculate θαθ −1 (which should come out to be β):<br />

(19264537) . (1963)(248)(57) . (73546291)<br />

= (13)(2479)(586)<br />

= β<br />

So <strong>we</strong> have exhibited θ such that β = θαθ −1 ,<br />

showing that α and β really <strong>are</strong> conjugate.


78 Section 6. Conjugacy<br />

Example 6.1.9. Conjugacy classes in D4 bring plastic squ<strong>are</strong><br />

Recall that D4 is the group of symmetries of the squ<strong>are</strong>:<br />

In D4 <strong>we</strong> have elements e,a,a 2 ,a 3 ,b1,b2,b3,b4<br />

a is rot anticlockwise, bi is ref in the line i<br />

○4<br />

We know ref.rotα.ref −1 = rot−α<br />

• We know that e is conjugate to .................... only e<br />

○3<br />

○2<br />

○1<br />

a, since rotations commute<br />

• Conjugating a by a rotation gives ..................................<br />

a −1 = a 3 by the above<br />

Conjugating a by a reflection gives .................................<br />

{a,a 3 }<br />

So the conjugacy class of a is .......................................<br />

a 2 , since rotations commute<br />

• Conjugating a 2 by a rotation gives .................................<br />

(a 2 ) −1 = a 2 by the above<br />

Conjugating a 2 by a reflection gives ................................<br />

{a 2 }<br />

So the conjugacy class of a 2 is ......................................<br />

• For b1 <strong>we</strong> calculate (noting that b −1<br />

i = bi): just use plastic squ<strong>are</strong><br />

eb1e = b1 b1b1b1 = b1<br />

ab1a −1 = b3 b2b1b2 = b3<br />

a 2 b1a −2 = b1 b3b1b3 = b1<br />

a 3 b1a −3 = b3 b4b1b4 = b3<br />

{b1,b3}<br />

So the conjugacy class of b1 is ......................................


6.2 The class equation 79<br />

{b2,b4}<br />

• It follows that the remaining conjugacy class is .....................<br />

How do <strong>we</strong> know? What sort of elements <strong>are</strong> in conjugacy classes of<br />

their own? —only those that commute with everything, and <strong>we</strong> know<br />

that reflections don’t commute with rotations.<br />

6.2 The class equation<br />

We found that in D4 the conjugacy class sizes <strong>are</strong>:<br />

1, 1, 2, 2, 2 total elements = 8<br />

We can learn a surprising amount about a group just by looking at this<br />

app<strong>are</strong>ntly stupid equation<br />

Definition 6.2.1. Class equation<br />

1 + 1 + 2 + 2 + 2 = 8<br />

Let G be a finite group with conjugacy classes<br />

Then the class equation for G is<br />

C1,C2,... ,Ck.<br />

|C1| + |C2| + · · · |Ck| = |G|.<br />

Note that the class equation has the following features:<br />

• 1 must appear at least once (for e)<br />

• each number must divide |G|.<br />

Example 6.2.2. Class equation for S3<br />

We know that elements of S3 <strong>are</strong> conjugate precisely if they have the same<br />

cycle type, so all <strong>we</strong> have to do to find the conjugacy class sizes is<br />

1. write down every possible cycle type in S3, and<br />

2. count how many permutations there <strong>are</strong> of each type.


80 Section 6. Conjugacy<br />

So in S3 <strong>we</strong> have class equation<br />

1 + 2 + 3 = 6<br />

..........................................<br />

cycle type typical element no. of elements<br />

3 (123) 2<br />

2,1 (12) 3<br />

1,1,1 e 1<br />

Example 6.2.3. Class equation for S5 and A5<br />

See Tutorial #7 q.4<br />

Cycle types in S5:<br />

cycle type typical element # ODD EVEN<br />

5 (12345) = (12)(23)(34)(45) 5!<br />

5 24 X<br />

4,1 (1234) = (12)(23)(34) 5!<br />

5 30 X <br />

3,2 (123)(45) = (12)(23)(45) 5.4.3.2<br />

4 20 X <br />

3,1,1 (123) = (12)(23) 5.4.3<br />

3 .2.1<br />

2 20 X<br />

5.4<br />

2,2,1 (12)(34) 2 .3.2<br />

2 .1<br />

2 15 X<br />

5.4<br />

2,1,1,1 (12) 2 10 X <br />

1,1,1,1 e 1 X<br />

• The class equation for S5 is<br />

• The class equation for A5<br />

1 + 10 + 15 + 20 + 20 + 30 + 24 = 120<br />

60<br />

5! = 120 Total even = ............<br />

can’t be 1 + 15 + 20 + 24 = 60 because those numbers don’t all divide<br />

60<br />

.....................................................................


6.2 The class equation 81<br />

Example 6.2.4.<br />

Suppose |G| = 7.<br />

Then the class equation must be:<br />

1 + 1 + 1 + 1 + 1 + 1 + 1 = 7<br />

all the numbers must divide 7.<br />

since .....................................................................<br />

This holds for groups of order any prime p.<br />

Example 6.2.5.<br />

Suppose |G| = 8.<br />

Then the possibilities <strong>are</strong>:<br />

1 1 1 1 1 1 1 1<br />

1 1 1 1 1 1 2<br />

1 1 1 1 2 2<br />

1 1 1 1 4<br />

1 1 2 2 2 ←− this one is D4<br />

1 1 2 4<br />

Question: What do those 1’s tell us? That is, what kind of element is in<br />

a conjugacy class of its own?<br />

Ans<strong>we</strong>r: An element a such that<br />

“a commutes with everything”<br />

∀g ∈ G gag −1 = a<br />

i.e. ga = ag


82 Section 6. Conjugacy<br />

Definition 6.2.6. Centre<br />

Let G be a group and a ∈ G.<br />

We say a is central in G if: ∀g ∈ G ga = ag.<br />

The set of all central elements is called the centre of G:<br />

Z(G) = { a ∈ G | ∀g ∈ G ga = ag }<br />

Equivalent ways of saying a is central:<br />

∀g ∈ G, a ∈ centG(g)<br />

Examples 6.2.7.<br />

= { a ∈ G | ∀g ∈ G a ∈ centG(g) }<br />

= <br />

centG(g)<br />

g<br />

∀g ∈ G, ga = ag ∀g ∈ G, gag −1 = a<br />

centG(a) = G<br />

1. Z(D4) = { e, rotπ } —see Example 6.1.9.<br />

2. Z(D3) = { e } by studying multiplication table<br />

3. If G is abelian then Z(G) = G.<br />

4. Z(G) always contains e.<br />

Proposition 6.2.8.<br />

conj G(a) = {a}<br />

For any group G, Z(G) is a normal subgroup of G. Z(G) G<br />

Proof.<br />

• Z(G) is a subgroup of G because it is the intersection of subgroups<br />

centG(g). We can also check directly: given a,b ∈ Z(G) and g ∈ G,<br />

g.ab.g −1 = gag −1 .gbg −1 trick<br />

= ab since a,b ∈ Z(G)


6.2 The class equation 83<br />

• Now <strong>we</strong> show it is a normal subgroup.<br />

We need to show:<br />

for all a ∈ Z(G) and g ∈ G, gag −1 ∈ Z(G).<br />

.....................................................................<br />

a ∈ Z(G), since a ∈ Z(G)<br />

But gag −1 =........................................................<br />

So <strong>we</strong> have a normal subgroup as required. <br />

Note that looking at the class equation, <strong>we</strong> see Z(G) appearing in the form<br />

of all the 1’s.<br />

Examples 6.2.9. Centres and the class equation<br />

1. The class equation for D4 is<br />

1 + 1 + 2 + 2 + 2 = 8<br />

1<br />

<br />

+ 1<br />

<br />

+2 + 2 + 2 = 8<br />

centre<br />

These 1’s <strong>are</strong> the classes of e and rotπ.<br />

2. The class equation for S5 is:<br />

1 + 10 + 15 + 20 + 20 + 30 + 24 = 120<br />

trivial<br />

so the centre is .....................................................<br />

3. Let |G| = 8. We saw all the possible class equations in Example 6.2.5.<br />

We deduce that the centre must be non-trivial.


84 Section 6. Conjugacy<br />

Proposition 6.2.10.<br />

Let G be a group of order p k where p is prime.<br />

Then Z(G) must be non-trivial.<br />

Proof.<br />

The only possible numbers in the class equation <strong>are</strong>:<br />

1 or po<strong>we</strong>rs of p, since the numbers must divide p k .<br />

...........................................................................<br />

We know <strong>we</strong> must have at least one 1, because<br />

e is in a conjugacy class by itself<br />

...........................................................................<br />

Now if <strong>we</strong> have only one 1:<br />

• the LHS of the equation will be congruent to .....1 (mod p) , whereas<br />

• the RHS is congruent to .....0 (mod p) #.<br />

So <strong>we</strong> must have more than one 1 in the class equation.<br />

This shows that the centre is non-trivial because<br />

each 1 corresponds to an element in the centre<br />

...........................................................................<br />

We can also detect other normal subgroups.<br />

Theorem 6.2.11.<br />

Let G be a group and H a subgroup.<br />

Then H is normal if and only if it is a union of conjugacy classes.<br />

We’ll prove this after looking at some examples.


6.2 The class equation 85<br />

Example 6.2.12. Normal subgroups of D4<br />

For D4 <strong>we</strong> have<br />

1 + 1 + 2 + 2 + 2<br />

e a 2 {a,a 3 } {b1,b3} {b2,b4}<br />

We know that the possible orders of subgroups of D4 <strong>are</strong>:<br />

1,2,4,8<br />

...........................................................................<br />

by ...................................................Theorem Lagrange’s<br />

• We know <strong>we</strong> have normal subgroups 0 and G.<br />

—and clearly these <strong>are</strong> unions of conjugacy classes<br />

• For subgroups of order 2, there is only one possible union of conjugacy<br />

classes:<br />

{e} ∪ {a 2 }, since any subgroup must contain e<br />

.....................................................................<br />

This is indeed a subgroup of D4, so is a normal subgroup.<br />

• For subgroups of order 4, <strong>we</strong> have the following possible unions of<br />

conjugacy classes:<br />

{e} ∪ {a 2 } ∪ {a,a 3 }<br />

{e} ∪ {a 2 } ∪ {b1,b3}<br />

{e} ∪ {a 2 } ∪ {b2,b4}<br />

since e must be contained in any subgroup.<br />

We can check that these <strong>are</strong> all subgroups, so must be normal sub-<br />

groups.<br />

• So the following subgroups <strong>are</strong> not normal:<br />

{e,bi} for each i<br />

.....................................................................


86 Section 6. Conjugacy<br />

Example 6.2.13. Normal subgroups of S5<br />

In S5 <strong>we</strong> know that conjugacy classes <strong>are</strong> given by<br />

elements of the same cycle type<br />

...........................................................................<br />

• Is 〈 (123) 〉 a normal subgroup?<br />

No, because for example it doesn’t contain the conjugate cycle (345).<br />

Besides, its order is 3, whereas there <strong>are</strong> 5.4.3<br />

3<br />

cycle type, so some must be missing.<br />

= 20 elements of this<br />

.....................................................................<br />

• Is 〈 (12345) 〉 a normal subgroup?<br />

No, because its order is 5 whereas there <strong>are</strong> 5!<br />

5<br />

cycle type, so some must be missing.<br />

= 24 elements of this<br />

.....................................................................<br />

Proof of Theorem 6.2.11.<br />

• Suppose H is a subgroup of G and is a union of conjugacy classes.<br />

We aim to show that H is a normal subgroup.<br />

So, given h ∈ H, g ∈ G, <strong>we</strong> need to show ghg −1 ∈ H.<br />

Now <strong>we</strong> know that conj G(h) ⊆ H<br />

so ghg −1 ∈ conj G(h) ⊆ H.<br />

So H is a normal subgroup as required.<br />

• Conversely suppose H G.<br />

Given h ∈ H <strong>we</strong> want to show conj G(h) ⊆ H.<br />

So consider x ∈ conj G(h)<br />

i.e. x = ghg −1 for some g ∈ G.<br />

But ghg −1 ∈ H since H is normal, so conj G(h) ⊆ H as required.


6.2 The class equation 87<br />

Example 6.2.14. Conjugacy in A4.<br />

Consider the conjugacy class of α = (123) in S4.<br />

β θ s.t. θαθ −1 = β θ<br />

ODD EVEN<br />

(123) e X<br />

(132) (23) X <br />

(124) (34) X <br />

(142) (243) X<br />

(134) (234) X<br />

(143) (24) X <br />

(234) (1234) X <br />

(243) (124) X<br />

This conjugacy class splits into two classes in A4:<br />

• Elements conjugate to α in S4 via an even cycle:<br />

{(123), (142), (134), (243)}<br />

.....................................................................<br />

• Elements conjugate to α in S4 via an odd cycle:<br />

{(132), (124), (143), (234)}<br />

.....................................................................<br />

Theorem 6.2.15. Conjugacy in An<br />

If there is an odd permutation θ such that θα = αθ then<br />

conj An(α) = conj Sn (α).<br />

Otherwise the conjugacy class splits in two in An.


88 Section 7. Homomorphisms<br />

7 Homomorphisms<br />

Recall that a group homomorphism is a function θ : G −→ H such that:<br />

∀x,y ∈ G, θ(xy) = θ(x)θ(y).<br />

...........................................................................<br />

Here G and H <strong>are</strong> of course groups.<br />

Here’s a “schematic diagram” of the group homomorphism axiom:<br />

multiply in G<br />

G H<br />

do θ<br />

x y θ(x) θ(xy) θ(y)<br />

xy<br />

Remember it follows that<br />

• θ(e) = e<br />

• θ(a −1 ) = (θ(a)) −1<br />

do θ<br />

θ(xy)<br />

Homomorphisms produce some very interesting subgroups.<br />

7.1 Kernels and images<br />

multiply in H<br />

Homomorphisms bet<strong>we</strong>en groups <strong>are</strong> much more po<strong>we</strong>rful than functions<br />

bet<strong>we</strong>en sets. They draw their po<strong>we</strong>r from the group structure of the groups<br />

in question.<br />

Definition 7.1.1. Kernel and image<br />

Let G and H be groups, and θ : G −→ H be a homomorphism.


7.1 Kernels and images 89<br />

The kernel of θ is: Kerθ = “everything that is sent to e” ⊆ G<br />

= { g ∈ G | θ(g) = e }<br />

The image of θ is: Imθ = “everything that is hit by θ” ⊆ H<br />

Lemma 7.1.2.<br />

= { θ(g) | g ∈ G }<br />

Let θ : G −→ H be a homomorphism. Then<br />

1. Kerθ is a subgroup of G.<br />

2. Imθ is a subgroup of H.<br />

= { h ∈ H | h = θ(g) for some g ∈ G }<br />

also written θ(G) <br />

Note that to show that an element of G is in Kerθ <strong>we</strong><br />

apply θ to it and check that the ans<strong>we</strong>r is e.<br />

...........................................................................<br />

To show that an element of H is in Imθ <strong>we</strong><br />

find an element of G that get sent to it by θ<br />

...........................................................................<br />

Proof.<br />

1. • e ∈ Kerθ since .......................... θ(e) = e<br />

• Given a and b ∈ Kerθ <strong>we</strong> have ab ∈ Kerθ since<br />

θ(a)θ(b)<br />

θ(ab) = ................................... by definition of group homomorphism<br />

e.e<br />

= ........................ since a,b ∈ Kerθ<br />

e<br />

= ........................ by definition of e.


90 Section 7. Homomorphisms<br />

• Given a ∈ Kerθ <strong>we</strong> have a −1 ∈ Kerθ since<br />

(θ(a)) −1<br />

θ(a −1 ) = ................................... by standard result for group homomorphisms<br />

e −1<br />

= ........................ since a ∈ Kerθ<br />

e<br />

= ........................ by definition of e.<br />

2. • e ∈ Imθ since ............................ θ(e) = e<br />

• Given x and y ∈ Imθ <strong>we</strong> show xy ∈ Imθ:<br />

Put x = θ(a), y = θ(b), say.<br />

Then<br />

θ(a)θ(b)<br />

xy = ................................... (substitute)<br />

θ(ab)<br />

= ........................ by definition of group homomorphism<br />

So xy ∈ Imθ as required.<br />

• Given x ∈ Imθ <strong>we</strong> show x −1 ∈ Imθ:<br />

Put x = θ(a), say.<br />

Then<br />

(θ(a)) −1<br />

x −1 = ................................... (substitution)<br />

θ(a −1 )<br />

= ........................ by standard result for group homomorphisms<br />

Example 7.1.3.<br />

So x −1 ∈ Imθ as required. <br />

Let G = GLn(R), the group of invertible real n × n matrices.<br />

Let H = R \ {0} under multiplication.<br />

We define a group homomorphism θ : G −→ H by<br />

θ(A) = detA.


7.1 Kernels and images 91<br />

This is a homomorphism since<br />

det(AB) = detA.detB<br />

...........................................................................<br />

• Kerθ = { A ∈ GLn(R) | detA = 1 }<br />

• Imθ = H<br />

= SLn(R) by definition<br />

We can always find A with detA = x eg ( x 0<br />

0 1 )<br />

Example 7.1.4.<br />

D3 acts on the vertices of an equilateral triangle giving a homomorphism<br />

D3 −→ S3.<br />

• Kerθ = {e} which symmetries fix every vertex?<br />

• Imθ = S3 which perms of the vertices can be achieved by symmetries?<br />

We will see that this shows D3 ∼ = S3.<br />

In general Dn acts on the vertices of the regular n-gon, giving a homomor-<br />

phism<br />

• Kerθ = {e}<br />

Dn −→ Sn.<br />

• Imθ = not the whole of Sn unles n = 3<br />

eg can’t switch two adjacent vertices and nothing else, except on the<br />

triangle<br />

We will see that Imθ ∼ = Dn.<br />

More generally, if G acts on a set of n things, <strong>we</strong> get a homomorphism<br />

G −→ Sn.


92 Section 7. Homomorphisms<br />

Example 7.1.5.<br />

Consider the following polynomials in x1,x2,x3,x4:<br />

Then as S4 permutes x1,x2,x3,x4<br />

it also permutes p1,p2,p3<br />

p1 = x1x2 + x3x4<br />

p2 = x1x3 + x2x4<br />

p3 = x1x4 + x2x3<br />

E.g. (123) : x1x2 + x3x4 p1 also p2 p3<br />

p3<br />

x2x3 + x1x4 ......... p1 p2<br />

(14)(23) : x1x2 + x3x4 p1 p2 p3<br />

This gives rise to a homomorphism<br />

x4x3 + x2x1 p1 p2 p3<br />

............................ ......... ......... .........<br />

θ : S4 −→ S3<br />

and Kerθ is “all elements of S4 that fix p1, p2, and p3”<br />

(14)(23), (12)(34), (13)(24), e<br />

= ......................................................................... <br />

N.B. There <strong>are</strong> four other elements that fix p1:<br />

(12),(34),(1324),(1423)<br />

.........................................................................<br />

but they don’t fix p2 and p3, so they <strong>are</strong>n’t in the kernel.<br />

The kernel and image <strong>are</strong> intimately related to injectivity and surjectivity.<br />

In fact, they basically <strong>are</strong> injectivity and surjectivity.


7.1 Kernels and images 93<br />

Remember what “injective” and “surjective” mean?<br />

A function f : A −→ B is<br />

f(a1) = f(a2) =⇒ a1 = a2<br />

• injective if ........................................................<br />

∀b ∈ B ∃a ∈ A s.t. f(a) = b<br />

• surjective if .......................................................<br />

If you can’t remember this then write a 5-step plan for how you’re going to<br />

remember it:<br />

1. .....................................................................<br />

2. .....................................................................<br />

3. .....................................................................<br />

4. .....................................................................<br />

5. .....................................................................<br />

Do you think I’m joking? Yes/No (delete as appropriate)<br />

Lemma 7.1.6. Let θ : G −→ H be a homomorphism. Then<br />

Proof.<br />

“ =⇒ ”<br />

Suppose θ is injective.<br />

Let g ∈ Kerθ, so θ(g) = e.<br />

We need to show g = e:<br />

Now θ(g) = e = θ(e)<br />

θ is injective ⇐⇒ Kerθ = 0 = {e}<br />

but θ is injective so this implies g = e.


94 Section 7. Homomorphisms<br />

“⇐=”<br />

Conversely suppose Kerθ = 0.<br />

Let g1,g2 ∈ G with θ(g1) = θ(g2). (1)<br />

We need to show g1 = g2:<br />

Now <strong>we</strong> have<br />

θ(g1) −1 .θ(g2) = θ(g −1<br />

1 ).θ(g2) standard property of homs<br />

= θ(g −1<br />

1 .g2) by definition of hom<br />

But also θ(g1) −1 .θ(g2) = e by (1)<br />

So θ(g −1<br />

1 .g2) = e<br />

i.e. g −1<br />

1 .g2 ∈ Kerθ.<br />

Thus g −1<br />

1 .g2 = e since Kerθ = {e}.<br />

So g1 = g2. <br />

Lemma 7.1.7.<br />

Let θ : G −→ H be a homomorphism. Then<br />

θ is surjective ⇐⇒ Imθ = H<br />

I’m not sure that even deserves to be called a Lemma, as it’s just the defi-<br />

nition of Im. Oh <strong>we</strong>ll, it deserves emphasis.<br />

We <strong>are</strong> going to have an amazing relationship bet<strong>we</strong>en these things:<br />

This tells us many things e.g.<br />

G Kerθ ∼ = Imθ<br />

|Imθ| = |G|<br />

|Kerθ|<br />

Moreover, it means that to understand “multiple hits” <strong>we</strong> only have to<br />

understand what gets sent to e.<br />

We’d better recall quotient groups, and also check that Kerθ is a normal<br />

subgroup, so that <strong>we</strong> can quotient by it.


7.1 Kernels and images 95<br />

Proposition 7.1.8.<br />

Let θ : G −→ H be a homomorphism.<br />

Then Kerθ is a normal subgroup of G.<br />

Proof.<br />

We already know Kerθ is a subgroup of G (Lemma 7.1.2).<br />

To show it is normal, <strong>we</strong> need to show:<br />

∀a ∈ Kerθ and g ∈ G, gag −1 ∈ Kerθ<br />

...........................................................................<br />

Now<br />

θ(g).θ(a).θ(g −1 )<br />

θ(gag −1 ) = .......................................... since θ is a homomorphism<br />

θ(g).e.θ(g −1 )<br />

= .......................................... since a ∈ Kerθ<br />

θ(g).θ(g −1 )<br />

= .......................................... by definition of e<br />

θ(g.g −1 )<br />

= ........................ since θ is a homomorphism<br />

θ(e)<br />

= ................... by definition of inverse<br />

e<br />

= ............... since θ is a homomorphism<br />

So gag −1 ∈ Kerθ as required.


96 Section 7. Homomorphisms<br />

7.2 Quotient groups revisited<br />

We will treat quotient groups slightly differently this time.<br />

Let H G. We define the quotient group G H<br />

• The first important thing to remember about quotient groups is: what<br />

<strong>are</strong> the elements?<br />

The elements <strong>are</strong> the cosets aH<br />

.....................................................................<br />

• The next important thing to remember is: what is the group opera-<br />

tion?<br />

How can <strong>we</strong> multiply cosets?? If you had to multiply aH and bH,<br />

what would the ans<strong>we</strong>r be in your wildest dreams?<br />

(ab)H<br />

..............................<br />

In fact <strong>we</strong> can multiply any subsets of a group:<br />

if A and B <strong>are</strong> subsets of G then<br />

AB = { ab | a ∈ A, b ∈ B }<br />

Note that some of these elements may be the same as each other, as in the<br />

next example.<br />

Example 7.2.1. Multiplication of cosets<br />

• Consider the cyclic group C12 with elements<br />

Let A = {a,a 2 ,a 3 }<br />

B = {a 4 ,a 5 ,a 6 }<br />

{e,a,a 2 ,...,a 11 }.<br />

{a 5 ,a 6 ,a 7 ,a 8 ,a 9 }<br />

Then AB =.........................................................


7.2 Quotient groups revisited 97<br />

• Now let H be the subgroup {e,a 4 ,a 8 }, and put<br />

a,a 5 ,a 9<br />

A = aH = ......................<br />

a 2 ,a 6 ,a 10<br />

B = a 2 H = ......................<br />

{a 3 ,a 7 ,a 11 }<br />

Then AB =.........................................................<br />

a 3 H<br />

So (aH).(a 2 H) is the coset ...........................<br />

This works because<br />

H C12 since C12 is abelian.<br />

.....................................................................<br />

This makes your life easier so please remember<br />

easy<br />

it:<br />

If H is a normal subgroup of G then it is to<br />

multiply cosets. You do not actually have to go round multiplying every<br />

element by every other because:<br />

(aH)(bH) = (ab)H


98 Section 7. Homomorphisms<br />

Example 7.2.2.<br />

Consider the group D3 with multiplication table (with the usual notation):<br />

e a a 2 b1 b2 b3<br />

e e a a 2 b1 b2 b3<br />

a a a 2 e b2 b3 b1<br />

a 2 a 2 e a b3 b1 b2<br />

b1 b3 b4 b2 e a 2 a<br />

b2 b2 b1 b3 a e a 2<br />

b3 b3 b2 b1 a 2 a e<br />

1. Let H be the not normal subgroup {e,b1}.<br />

i) The coset aH = {ae,ab1} = { a,b2 }.<br />

ii) The set (eH).(aH) has elements<br />

iii) Is this a coset of H?<br />

2. Let V be the normal subgroup {e,a,a 2 }.<br />

a<br />

ea = ...........<br />

b2<br />

eb2 = ...........<br />

b4<br />

b1a = ...........<br />

a 2<br />

b1b2 = ...........<br />

No—wrong number of elements<br />

.............................................<br />

i) The coset b1V = {b1e,b1a,b1a 2 } = { b1,b3,b2 }<br />

ii) (eV ).(b1V ) has elements<br />

eb1 = b1 ......... ab1 = b2 ......... a2b1 = b3 .........<br />

eb2 = b2 ......... ab2 = b3 ......... a2b2 = b1 .........<br />

eb3 = b3 ......... ab3 = b1 ......... a2b3 = b2 .........<br />

iii) So (eV ).(b1V ) = { b1,b2,b3 } and is the coset ......................... b1V


7.2 Quotient groups revisited 99<br />

Definition 7.2.3. Quotient group<br />

Let H G.<br />

Then the cosets of H form a group called the quotient group denoted<br />

with<br />

G H<br />

• group operation given by multiplication of cosets, so<br />

• identity given by H = eH since<br />

(aH).(bH) = (ab)H<br />

H.(aH) = (aH).H = aH<br />

• inverses given by (aH) −1 = (a −1 )H since<br />

(aH)(a −1 H) = eH = H<br />

(a −1 H)(aH) = eH = H <br />

Another way of thinking about quotient groups is by equivalence rela-<br />

tions. We use the equivalence relation<br />

x ∼ y ⇐⇒ y −1 x ∈ H.<br />

Then the equivalence classes form a group! The operation is<br />

[x].[y] = [xy]<br />

which is <strong>we</strong>ll-defined because H is normal. same proof as for when <strong>we</strong> first<br />

defined quotient groups<br />

Actually these <strong>are</strong> just like cosets in disguise:<br />

[x] = [y] ⇐⇒ xH = yH<br />

cosets equivalence relation<br />

multiplication (aH)(bH) = (ab)H [a][b] = [ab]<br />

identity eH = H [e]<br />

inverse of aH (a −1 )H [a −1 ]


100 Section 7. Homomorphisms<br />

Here <strong>are</strong> a couple of applications of quotient groups.<br />

Theorem 7.2.4.<br />

Let G be a group with G Z(G) cyclic.<br />

Then G is abelian.<br />

Proof.<br />

Write Z = Z(G)<br />

and pick a generator gZ of G/Z. NB g is now fixed.<br />

This means every element of G/Z is of the form (gZ) n = g n Z.<br />

Now, given an element a ∈ G it is in some coset g n Z, say<br />

i.e. a = g n z for some n ∈ N0,z ∈ Z.<br />

Now <strong>we</strong> want to show ∀a,b ∈ G, ab = ba.<br />

Put a = g m z1<br />

b = g n z2<br />

Then ab = g m z1.g n z2<br />

= g m g n z2z1 since<br />

= g n g m z2z1 since<br />

z1 ∈ Z so commutes with everything<br />

...............................................<br />

g m g n = g m+n = g n g m<br />

...............................................<br />

= g n z2.g m z1 since moving z2 this time<br />

...............................................<br />

= ba


7.2 Quotient groups revisited 101<br />

Theorem 7.2.5.<br />

If p is prime then every group of order p 2 is abelian.<br />

Note <strong>we</strong> already know that every group of order p is<br />

Proof.<br />

We aim to show Z(G) = G.<br />

cyclic<br />

.....................<br />

We use Proposition 6.2.10 which says that in a group of order p k (with p<br />

prime), the centre must be non-trivial. from class equation<br />

So Z(G) is a non-trivial subgroup of G<br />

so by Lagrange’s Theorem its order must be<br />

Suppose |Z(G)| = p. aim for a contradiction<br />

Then |G/Z(G)| = p so G/Z(G) is cyclic.<br />

So by Theorem 7.2.4 G is abelian<br />

i.e. |Z(G)| = p2<br />

............................... #<br />

p or p 2<br />

.............................<br />

So <strong>we</strong> must have |Z(G)| = p 2 i.e. G is abelian as required.


102 Section 7. Homomorphisms<br />

7.3 First isomorphism theorem<br />

This is the really big theorem of the course.<br />

Theorem 7.3.1. The First Isomorphism Theorem for Groups<br />

Let θ : G −→ H be a group homomorphism. Then<br />

G Kerθ ∼ = Imθ.<br />

Before <strong>we</strong> prove this, let’s think about why it’s great.<br />

If <strong>we</strong> only had a function of sets, how could <strong>we</strong> ever tell anything about<br />

anything??<br />

stuff about even covering, e pulling things in<br />

Remarks 7.3.2.<br />

Note that <strong>we</strong> have some special cases:<br />

1. If θ is injective <strong>we</strong> have Kerθ = 0<br />

so the First Isomorphism Theorem says:<br />

G ∼ = Imθ.<br />

2. If θ is surjective <strong>we</strong> have Imθ = H<br />

so the First Isomorphism Theorem says:<br />

3. So if θ is an isomorphism<br />

G Kerθ ∼ = H.<br />

the First Isomorphism Theorem says:<br />

as expected.<br />

G ∼ = H<br />

4. If <strong>we</strong> have N G then <strong>we</strong> have a homomorphism<br />

G −→ G/N<br />

whose kernel is N and image is G/N.


7.3 First isomorphism theorem 103<br />

5. As a special case of the above, given a homomorphism θ : G −→ H <strong>we</strong><br />

know <strong>we</strong> have Kerθ G so <strong>we</strong> get a homomorphism<br />

G −→ G Kerθ ∼ = Imθ<br />

whose kernel is still Kerθ, but it is now surjective.<br />

<strong>we</strong> threw away everything that wasn’t hit<br />

Example 7.3.3.<br />

Fix n ∈ N.<br />

Dn acts on the vertices of a regular n-gon as in Example 7.1.4, giving a<br />

homomorphism<br />

We know that the kernel is 0............<br />

θ : Dn −→ Sn.<br />

so by the First Isomorphism Theorem<br />

Dn 0 ∼ = Imθ = Dn<br />

.........................................<br />

Example 7.3.4.<br />

Fix n ∈ N.<br />

Define θ : O2 −→ O2 by<br />

Then<br />

rotα ↦→ rotnα<br />

refα ↦→ refnα<br />

• Imθ = O2 since <strong>we</strong> can hit any α from α<br />

n<br />

.............................................................<br />

• Kerθ = 〈rot2π 〉<br />

n<br />

∼ = Cn<br />

............................................................<br />

So by the First Isomorphism Theorem: think harder about this<br />

We can picture the cosets as<br />

O2 Cn<br />

∼= O2<br />

.........................................<br />

•<br />

•<br />

•<br />

•<br />

•<br />

•<br />

•<br />

• • • • •


104 Section 7. Homomorphisms<br />

Example 7.3.5.<br />

Define θ : R −→ SO2 by<br />

Then<br />

α ↦→ rotα<br />

• Imθ = SO2<br />

.............................................................<br />

• Kerθ = {0, ±2π, ±4π,...} ∼ = 〈2π〉 ∼ = Z<br />

............................................................<br />

So by the First Isomorphism Theorem:<br />

Example 7.3.6.<br />

Define θ : Dn −→ Z2 by<br />

Then<br />

α ↦→<br />

R Z ∼ = SO2<br />

..................................................<br />

⎧<br />

⎨<br />

⎩<br />

0 if α is a rotation<br />

1 if α is a reflection<br />

• Imθ = Z2<br />

.............................................................<br />

• Kerθ =<br />

the rotations i.e. Cn<br />

............................................................<br />

So by the First Isomorphism Theorem:<br />

Dn Cn<br />

∼= Z2<br />

..................................................


7.3 First isomorphism theorem 105<br />

Example 7.3.7.<br />

D6 acts on the diagonals of the regular hexagon.<br />

So <strong>we</strong> get a homomorphism θ : D6 −→ S3.<br />

Then<br />

○3<br />

○2<br />

• Kerθ = {e,rotπ}<br />

............................................................<br />

• We can deduce what the image is from the First Isomorphism Theo-<br />

rem, which says:<br />

D6 Kerθ<br />

order = 12<br />

2<br />

So Imθ has order<br />

so must be<br />

○1<br />

∼= Imθ ⊆ S3<br />

↑ ↑<br />

= 6 order = 6<br />

6<br />

....... and is a subgroup of S3,<br />

the whole thing.<br />

.................................


106 Section 7. Homomorphisms<br />

Finally here’s the proof of the First Isomorphism Theorem.<br />

This is not examinable.<br />

Proof of First Isomorphism Theorem.<br />

We exhibit a group isomorphism<br />

f : G Kerθ<br />

∼<br />

−→ Imθ.<br />

For convenience <strong>we</strong> write K = Kerθ throughout.<br />

1. We define f(aK) = θ(a) ∈ Imθ.<br />

We check that this is <strong>we</strong>ll-defined<br />

i.e. aK = bK =⇒ θ(a) = θ(b).<br />

Now aK = bK ⇐⇒ a −1 b ∈ K<br />

=⇒ θ(a −1 b) = e<br />

=⇒ θ(a −1 ).θ(b) = e<br />

=⇒ θ(a) −1 .θ(b) = e<br />

=⇒ θ(a) = θ(b).<br />

2. We show that this is a group homomorphism.<br />

i.e. f((aK)(bK)) = f(aK).f(bK).<br />

Now (aK)(bK) = (ab)K<br />

so f((aK)(bK)) = f((ab)K)<br />

= θ(ab)<br />

= θ(a).θ(b) since θ is a homomorphism<br />

= f(aK).f(bK)


7.3 First isomorphism theorem 107<br />

3. We show that f is surjective.<br />

i.e. for all y ∈ Imθ, ∃x ∈ G Kerθ s.t. f(x) = y.<br />

Now y ∈ Imθ ⇐⇒ y = θ(a) for some a ∈ G.<br />

Then putting x = aK ∈ G Kerθ <strong>we</strong> have<br />

4. Finally <strong>we</strong> show that f is injective<br />

f(x) = f(aK)<br />

= θ(a)<br />

= y.<br />

i.e. Kerf = {e} (using Lemma 7.1.6).<br />

Let x ∈ Kerf<br />

so x = aK for some a ∈ G, and f(aK) = e.<br />

Now f(aK) = θ(a) and<br />

θ(a) = e =⇒ a ∈ Kerθ<br />

which is the identity in G Kerθ .<br />

=⇒ aK = K<br />

So f is a group isomorphism as required.


108 Section 7. Homomorphisms


Part III<br />

Homework and Tutorial Questions<br />

109<br />

• Homework #n is to be done in <strong>we</strong>ek n and handed in at the lecture<br />

on Monday of <strong>we</strong>ek #(n + 1).<br />

• Tutorial #n is to be done in <strong>we</strong>ek n regardless of whether or not you<br />

have a tutorial that <strong>we</strong>ek. In each tutorial you can get help with the<br />

previous two <strong>we</strong>eks’ tutorial work (and earlier if necessary).<br />

• In non-tutorial <strong>we</strong>eks you can come to my office hours to get help<br />

with that <strong>we</strong>ek’s tutorial work.<br />

• Please make sure you read any hints given in the questions. It’s<br />

extraordinary number of times tutors get asked questions which <strong>are</strong> in<br />

fact ans<strong>we</strong>red in the hints...<br />

• You should always justify all your ans<strong>we</strong>rs. If a question says “Is 14<br />

a zero-divisor?” don’t just ans<strong>we</strong>r yes or no – say how you decided<br />

that was the ans<strong>we</strong>r. When you write out your working, include some<br />

explanation about what you did. This is good practice for exams, but<br />

is also useful for the marker so they can understand what you <strong>we</strong>re<br />

doing (and therefore be in a better position to help you if it <strong>we</strong>nt<br />

wrong). It’s also useful for you, because when you’re revising you<br />

should be going back over your homework to make sure you understand<br />

it better than you did the first time round.


110 Section 8. Homework questions<br />

8 Homework questions<br />

Homework #1<br />

This set looks long, but it isn’t actually long. There’s a lot stuff here that isn’t so much<br />

question as some explanation about why these questions <strong>are</strong> useful things for you to<br />

think about.<br />

1. On the next page is a page of statements I have copied from some Semester One<br />

exams I have marked. Imagine this is work that a student has handed in. Mark it in<br />

the way that you would like to see your homework marked. (Pen or pencil? Marks<br />

out of 10 or a letter grade? Encouragement or criticism? Helpful comments?)<br />

This exercise is partly to help you think about how you should read over your own<br />

work to see if it is correct, partly so I can see what you consider to be useful<br />

feedback on your work, and partly to make sure you won’t ever make any of the<br />

same mistakes listed here.<br />

2. This question is to jog your memory about modular arithmetic. Let n ∈ N and<br />

a ∈ Z.<br />

• Recall that n|a means ∃k ∈ Z s.t. a = kn.<br />

• Recall that “x ≡ y (mod n)” means n|(x − y).<br />

Suppose x ≡ y (mod n). Show that:<br />

i) a + x ≡ a + y (mod n), and<br />

ii) ax ≡ ay (mod n).<br />

3. This question is about attempting to work in Z6. Taking squ<strong>are</strong> roots and solving<br />

quadratic equations don’t necessarily work the way <strong>we</strong>’re expecting...<br />

i) In a ring R, x is said to be a squ<strong>are</strong> root of y when x 2 = y. Find the elements<br />

of Z6 that have squ<strong>are</strong> roots i.e. find all the elements a ∈ Z6 such that there<br />

exists x ∈ Z6 s.t. x 2 with a (mod 6)<br />

You might find it useful to use the multiplication table from the tutorial sheet.<br />

You should find that not all elements have squ<strong>are</strong> roots: there <strong>are</strong> only four<br />

that do. Are you surprised?<br />

ii) Find all solutions of x 2 + 3x + 2 = 0 in Z6. You will probably need to do it<br />

by trial and error, because our usual methods for solving quadratic equations<br />

won’t work here. Why? Note how many solutions there <strong>are</strong>. Did it surprise<br />

you?


1.<br />

1 1 1<br />

+ =<br />

a b a + b<br />

2. (a + b) n = a n + b n<br />

3. (a − b) 3 = a 3 − 3a 2 b − 3ab 2 − b 3<br />

4. 3x = 2<br />

3<br />

=⇒ x = 2<br />

−1 3<br />

n + n<br />

5. 4 = 2(n + n<br />

2<br />

−1 ) 3<br />

1 1<br />

1− − 6. e 2 = e 2<br />

7. e 3<br />

<br />

1 2 2<br />

− + = e<br />

3 9 27<br />

3<br />

<br />

1 2<br />

−<br />

3 9<br />

8. (e x + e −x ) 3 = e 3x + e −3x<br />

9. 3e x + e −3x = 0<br />

10. tanx = 1 =⇒ x = 0.785 to 3 d.p.<br />

111


Homework #2<br />

112 Section 8. Homework questions<br />

1. Match the following informal statements with their formal counterparts. Note that<br />

these <strong>are</strong> all facts that could be true in a ring R, but <strong>are</strong>n’t necessarily true about<br />

all rings.<br />

i) The additive inverse of −1 is 1.<br />

ii) The multiplicative inverse of −1 is −1.<br />

iii) 0 has no multiplicative inverse in R.<br />

iv) If x has a multiplicative inverse then so does −x.<br />

v) If x and y <strong>are</strong> in R it isn’t necessarily true that xy is non-zero.<br />

vi) Every non-zero element of R has a multiplicative inverse.<br />

A) (−1)(−1) = 1<br />

B) ∃x,y ∈ R s.t. xy = 0<br />

C) ∀x = 0 ∈ R ∃y ∈ R s.t. xy = yx = 1<br />

D) −(−1) = 1<br />

E) ∀x ∈ R, 0x = 1<br />

F) ∃y ∈ R s.t. xy = yx = 1 ⇒ ∃z ∈ R s.t. (−x)z = z(−x) = 1<br />

2. Find all the units in Z10 and fill in a multiplication table for them.<br />

3. Let R = Z[ √ 7]. Let r = 8 − 3 √ 7 and s = 8 + 3 √ 7. Compute rs and deduce that<br />

r and s <strong>are</strong> units in R. Hence solve this equation in R:<br />

(8 + 3 √ 7)x = 3<br />

This should rather remind you of the process of “rationalising the denominator”.


Homework #3<br />

113<br />

1. Match the following statements with their negations. Note that I’m trying to catch<br />

you out.<br />

i) ∀x ∈ R x.0 = 0<br />

ii) ∃x = y ∈ R s.t. xy = 0<br />

iii) x ∈ R ⇒ 1.x = x<br />

iv) ∀x = y ∈ R, xy = 0<br />

v) ∃x ∈ R s.t. 1.x = x<br />

vi) ∃x ∈ R s.t. x.0 = 0<br />

A) ∀x ∈ R, x.0 = 0<br />

B) ∃x ∈ R s.t. x.0 = 0<br />

C) x ∈ R ⇒ 1.x = x<br />

D) ∀x = y ∈ R, xy = 0<br />

E) ∃x = y ∈ R s.t. xy = 0<br />

F) ∃x ∈ R s.t. 1.x = x<br />

2. For each of the following, determine whether or not a is a unit in Zn and, if it is<br />

a unit, use Euclid’s algorithm to find its inverse.<br />

i) n = 90, a = 25<br />

ii) n = 91, a = 26<br />

iii) n = 47, a = 25<br />

Hint: if you think you’ve found the inverse for a, multiply it back with a and check<br />

that you get something that’s congruent to 1 mod n. It’s so easy to check that<br />

something really is an inverse in Zn, you should never ever get the wrong ans<strong>we</strong>r<br />

– you might get stuck and not be able to find an ans<strong>we</strong>r (<strong>we</strong> all have bad days) but<br />

you shouldn’t ever get the wrong ans<strong>we</strong>r.<br />

3. For this question you can copy the method from tutorial sheet #3.<br />

i) Show, by finding its inverse, that 2 + 3x is a unit in Z9[x].<br />

ii) Show, by finding its inverse, that 3 + 4x is a unit in Z16[x].


Homework #4<br />

114 Section 8. Homework questions<br />

1. What does “hence” mean in a maths question? eg “i) Prove this. ii) Hence do<br />

that.” By contrast, what does “hence or otherwise” mean?<br />

You might think this is a silly question, but the reason I’m asking it is that loads<br />

of people always get it wrong in exams.<br />

2. For every element a in Z24 determine whether a is a unit or a zero-divisor. If a is<br />

a unit find its inverse, and if a is a zero-divisor find b = 0 such that ab = 0. Write<br />

your ans<strong>we</strong>rs out in a table, so that it’s easier to read.<br />

3. Is 523 is a unit or a zero-divisor in Z570? If it is a unit find its inverse, and if it is<br />

a zero-divisor find b = 0 such that 523b = 0.<br />

Homework #5<br />

1. Find a unit r ∈ Z[ √ 11] such that r > 1. Hence show that the group of units of<br />

Z[ √ 11] is infinite.<br />

2. This question is about Z[ √ −13].<br />

i) Show that Z[ √ −13] has no element of norm 2 or 11.<br />

ii) Hence show that any element of Z[ √ −13] with norm 4, 22 or 121 is irreducible<br />

in Z[ √ −13].<br />

iii) Calculate the norm of (3 + √ −13).<br />

iv) Hence express 22 as a product of two irreducible factors in Z[ √ −13] in two<br />

different ways, and deduce that Z[ √ −13] is not a unique factorisation domain.<br />

Hint: copy the procedure on Example 3.1.12. You must show that your two<br />

factorisations <strong>are</strong> non-equivalent, and that all your factors <strong>are</strong> irreducible.<br />

3. If R is a unique factorisation domain and S is a subring of R, does it follow that<br />

S is a unique factorisation domain?


Homework #6<br />

115<br />

This whole sheet is revision from Groups and Symmetries, so you may need to look things<br />

up from your old notes. I know it’s hard to remember things from previous modules, so<br />

this homework is to help you jog your memory in preparation for the groups part of our<br />

course.<br />

1. Write down the whole multiplication table for D3, the group of symmetries of the<br />

equilateral triangle. Then look it up on Wikipedia to check your ans<strong>we</strong>r.<br />

2. Consider the squ<strong>are</strong> with lines of symmetry labelled 1, 2, 3 and 4 as below.<br />

○4<br />

○3<br />

Recall that D4 is the group of symmetries of the squ<strong>are</strong>, and observe that it acts<br />

on the lines 1,2,3,4.<br />

i) For each element of D4, write down the corresponding permutation of 1,2,3,4<br />

in a table like the one below. In this table, e is the identity of the group, a is<br />

rotation through π<br />

2 anti-clockwise, and bi is reflection in the line i.<br />

e<br />

a<br />

a 2<br />

a 3<br />

ii) Find all the elements of the orbit of 1.<br />

b1<br />

b2<br />

b3<br />

b4<br />

○2<br />

○1<br />

1 2 3 4<br />

iii) Find all the elements of the stabiliser of 1.


116 Section 8. Homework questions<br />

iv) Verify that this satisfies the orbit-stabiliser theorem.<br />

v) Now consider the circle with similar looking lines on it as below:<br />

○4<br />

○3<br />

Are these lines <strong>are</strong> acted on by O2 (the group of symmetries of the circle) as<br />

for the squ<strong>are</strong> above?<br />

○2<br />

○1


Homework #7<br />

117<br />

1. Recall that D3 is the group of symmetries of the equilateral triangle. Consider the<br />

equilateral triangle with corners labelled as below.<br />

○2<br />

1<br />

3<br />

○3<br />

i) For each element of D3, write down the corresponding permutation of 1,2,3<br />

in a table like the one below. In this table, e is the identity of the group, a is<br />

rotation through 2π<br />

3 anti-clockwise, and bi is reflection in the line i.<br />

ii) This defines a function<br />

e<br />

a<br />

a 2<br />

b1<br />

b2<br />

b3<br />

2<br />

○1<br />

1 2 3<br />

θ : D3 −→ S3.<br />

Show that θ is a group isomorphism. This means it is a group homomorphism<br />

that is also a bijection.<br />

2. Go through your notes from the Rings part of the course and write a list of every<br />

definition you think you will need to learn for the exam.


Homework #8<br />

118 Section 8. Homework questions<br />

1. Write down all possible cycle types in S6. For each cycle type, write down a typical<br />

permutation of that type, and calculate the number of permutations of that type.<br />

Please try to lay out your ans<strong>we</strong>rs nicely so that they’re easy to read.<br />

2. Now consider the symmetric group S9 and let<br />

Find |conjS9 (α)| and |centS9 (α)|.<br />

⎛<br />

⎞<br />

1 2 3 4 5 6 7 8 9<br />

α = ⎝ ⎠.<br />

9 4 1 8 7 5 2 6 3


Homework #9<br />

119<br />

Note that Q.2 is just like the last question on Tutorial #9, so you can follow<br />

that model ans<strong>we</strong>r through.<br />

1. i) Write down all possible cycle types in S4, and the number of elements of S4<br />

of each type.<br />

ii) Using your ans<strong>we</strong>r to part (i), write down the class equation for S4.<br />

iii) Identify the conjugacy classes consisting of even permutations. These numbers<br />

cannot give the class equation for A4. Why?<br />

iv) Recall that A4 is a normal subgroup of S4. Use the class equation for S4 to<br />

show that if there is another non-trivial normal subgroup of S4 it must have<br />

order 4.<br />

Recall: to be a normal subgroup, its order must both divide 24 and be a sum<br />

of conjugacy class sizes including one class of size 1 for the identity.<br />

v) Write down the four elements that would have to be the four elements of this<br />

group, according to the conjugacy class sizes you found in part (i). Verify that<br />

these elements do in fact form a subgroup of S4.<br />

2. Let G be a group of order 39 and suppose that the centre of G is {e}.<br />

i) Determine the class equation for G, justifying your ans<strong>we</strong>r c<strong>are</strong>fully.<br />

ii) Find the number of elements of order 3 and the number of elements of order<br />

13 in G.<br />

iii) Let h ∈ G be an element of order 13. Let H = 〈h〉, the subgroup generated<br />

by h. Use the class equation to show that H is a normal subgroup of G.<br />

iv) Show that H is the only normal subgroup of G other than the trivial group<br />

{e} and G itself.


120 Section 9. Tutorial questions<br />

9 Tutorial questions<br />

Tutorial questions #1<br />

1. We write Zn for the ring of integers mod n. Fill in the following multiplication<br />

tables for Z3, Z4, Z5 and Z6:<br />

× 0 1 2<br />

0<br />

1<br />

2<br />

× 0 1 2 3<br />

0<br />

1<br />

2<br />

3<br />

× 0 1 2 3 4<br />

0<br />

1<br />

2<br />

3<br />

4<br />

× 0 1 2 3 4 5<br />

2. In a ring R, b is said to be a multiplicative inverse for a if ab = ba = 1. Use the<br />

tables you filled in above to find the multiplicative inverses for the elements of Z3,<br />

Z4, Z5 and Z6 if they exist.<br />

3. A field is a ring in which every non-zero element has a multiplicative inverse.<br />

Which of the above rings is a field? State precisely what it means for a ring not<br />

to be a field.<br />

4. An integral domain is a ring in which 1 = 0 and<br />

a = 0 and b = 0 ⇒ ab = 0.<br />

Which of the above rings is an integral domain? State precisely what it means for<br />

a ring not to be an integral domain.<br />

5. In the Countdown video <strong>we</strong> watched, the last stage of James Martin’s calculation<br />

was to divide 23,800 by 25. Carol Vorderman vaguely says out loud “Well to do<br />

that you multiply by 4...” which shows she’s using the trick that dividing by 25<br />

is the same as multiplying by 4 and dividing by 100 (both of which <strong>are</strong> probably<br />

easier to do in your head than dividing by 25 directly). Which ring axiom is she<br />

using here?<br />

6. We will define a binary operation ⊗ on Z by<br />

a ⊗ b = ab + 1.<br />

0<br />

1<br />

2<br />

3<br />

4<br />

5


121<br />

Show that this operation is not associative. Can you think of a binary operation<br />

on Z that is not commutative?<br />

7. In any ring R, <strong>we</strong> can prove using only the ring axioms that −(−1) = 1. Is the<br />

following a valid proof?<br />

∀x ∈ R, (−1)x = −x.<br />

Therefore − (−1) = (−1)(−1)<br />

but <strong>we</strong> know (−1)(−1) = 1.<br />

Observe that the result is true – but this doesn’t mean the proof is valid. Moreover<br />

observe that all the equalities valid – but this still doesn’t mean the proof is valid.<br />

What is wrong with it? Can you write a valid proof?<br />

If you found the earlier questions easy, see if you can prove that the following <strong>are</strong> true<br />

in every ring R. You should prove them directly from the ring axioms, without using<br />

any other “facts” that you know to be true, even if they seem very obvious!<br />

1. ∀x ∈ R, 0.x = 0.<br />

2. ∀x,y ∈ R, (−1)x = −x.<br />

3. ∀a,b,c ∈ R, a + b = 0 and a + c = 0 ⇒ b = c.


122 Section 9. Tutorial questions<br />

Tutorial questions #2<br />

1. The following is a proof that (x − y)(x + y) = x 2 − y 2 in any commutative ring.<br />

Ho<strong>we</strong>ver, the justification for each line of argument has not been included. Please<br />

fill them in. You may only use ring axioms and Lemmas from immediately after<br />

them in the notes.<br />

(x − y)(x + y) = x(x + y) + (−y)(x + y) by ...........................................(1)<br />

= (x 2 + xy) + (−y).x + (−y).y) by ...........................................(2)<br />

= x 2 + xy + ((−y).x + (−y).y <br />

= x 2 + (xy + (−y).x) + (−y).y <br />

= x 2 + (xy + (−(yx))) + (−y 2 ) <br />

= x 2 + (xy + (−(xy))) + (−y 2 ) <br />

by ...........................................(3)<br />

by ...........................................(4)<br />

by ...........................................(5)<br />

by ...........................................(6)<br />

= x 2 + (0 + (−y 2 )) by ...........................................(7)<br />

= x 2 + (−y 2 ) by ...........................................(8)<br />

= x 2 − y 2 by definition of subtraction<br />

Note that proofs should always include a justification of how each line of the argu-<br />

ment follo<strong>we</strong>d from the previous line.<br />

2. i) Write out a multiplication table for Z8.<br />

ii) Find the units in Z8. Recall that the units <strong>are</strong> those elements with a multi-<br />

plicative inverse.<br />

iii) Now write out a multiplication table just for the units in Z8. Use this to show<br />

that the units of Z8 form a multiplicative group.<br />

3. If x and y <strong>are</strong> units is x + y necessarily a unit?<br />

The previous question about the units in Z8 may help you with this.<br />

4. Let R be a commutative ring and x ∈ R. Prove that x has at most one multiplica-<br />

tive inverse, that is, if xa = 1 and xb = 1 then a = b. Just copy the proof of the<br />

result for additive inverses, but make it multiplicative instead.


123<br />

5. This question is to give you some practice with the rings Z[ √ d], which consists of<br />

all numbers of the form a + b √ d where a,b ∈ Z. We’ll look at the case d = 2.<br />

These might look a bit like polynomial rings but actually they’re a bit different.<br />

i) Let a,b ∈ Z. Calculate (a + bx)(a − bx) ∈ Z[x].<br />

ii) Let a,b ∈ Z. Calculate (a + b √ 2)(a − b √ 2) ∈ Z[ √ 2].<br />

iii) These two situations might not look very different so far. But now do it for<br />

a = 5,b = 3. What happens?<br />

iv) Can you think of some non-zero values of a and b that give (a+b √ 2)(a−b √ 2) =<br />

1?<br />

v) Can you think of some non-zero values of a and b that give (a+bx)(a−bx) = 1?<br />

vi) What’s the point I’m trying to make here??<br />

6. i) Let a,n,r,s be integers such that<br />

ar + ns = 1.<br />

Show that a and r <strong>are</strong> mutually inverse in Zn.<br />

ii) Deduce that a is a unit in Zn if and only if the highest common factor of a<br />

and n is 1.<br />

Can you remember a result from Numbers and Proofs that links the second<br />

part of this question to the first part, and thus enables you to deduce the result<br />

you’re asked for...??<br />

If you found the previous questions easy, try these:<br />

1. Let R be a ring. Show that if 1 = 0 ∈ R then x = 0 for all x ∈ R.<br />

2. Let R be a ring. Recall that R is called an integral domain if<br />

so R is not an integral domain if<br />

{ a = 0 and b = 0 } ⇒ ab = 0<br />

∃a = 0 and b = 0 s.t. ab = 0.<br />

Recall also that on the Tutorial Sheet 1 you found that Z3 and Z5 <strong>are</strong> integral<br />

domains, but Z4 and Z6 <strong>are</strong> not integral domains.<br />

Show that Zn is not an integral domain for n = 8,9,10. Can you guess what<br />

property of n determines whether or not Zn is an integral domain? Can you prove<br />

it?


124 Section 9. Tutorial questions<br />

Tutorial questions #3<br />

1. The following <strong>are</strong> some submitted “proofs” from homework #1. The question<br />

asked you to prove: if x ≡ y (mod n) then ax ≡ ay (mod n). Which of the<br />

following is a valid proof? What is wrong with the others?<br />

• “Proof” A:<br />

• “Proof” B<br />

ax ≡ ay (mod n) =⇒ n|ax − ay<br />

=⇒ n|a(x − y)<br />

=⇒ n|x − y<br />

=⇒ x ≡ y (mod n)<br />

x ≡ y (mod n) so multiplying both sides by a gives<br />

• “Proof” C<br />

• “Proof” D<br />

ax ≡ ay (mod n).<br />

x ≡ y (mod n) = n|x − y<br />

= n|a(x − y)<br />

= n|ax − ay<br />

x ≡ y (mod n) =⇒ n|x − y<br />

= ax ≡ ay (mod n)<br />

=⇒ x − y = kn for some k ∈ Z<br />

=⇒ ax − ay = akn<br />

=⇒ n|ax − ay<br />

=⇒ ax ≡ ay (mod n)<br />

2. i) Show that (x + 1)(x + 2)(x + 3) ≡ x 3 − x ∈ Z6[x].<br />

ii) What <strong>are</strong> the roots of (x + 1)(x + 2)(x + 3) in R?<br />

iii) What <strong>are</strong> the roots of x 3 − x in R?<br />

iv) What <strong>are</strong> the roots of (x + 1)(x + 2)(x + 3) in Z6?


v) Did this surprise you? Can you see why it’s true?<br />

125<br />

To find roots you need to find values of x for which the given formula comes out<br />

to 0.<br />

3. Show (1 + 3x) is a unit in Z9[x].<br />

Hint: you need to find integers a and b such that<br />

(1 + 3x)(a + bx) ≡ 1 ∈ Z9[x].<br />

4. Here is a method for solving the quadratic equation x 2 + 3x + 2 = 0 in the reals,<br />

producing two solutions. In the homework you should have seen that there <strong>are</strong><br />

four solutions in Z6, which means that this method doesn’t work in Z6. Which<br />

step goes wrong?<br />

x 2 + 3x + 2 = (x + 1)(x + 2)<br />

(x + 1)(x + 2) = 0 =⇒ x + 1 = 0 or x + 2 = 0<br />

=⇒ x = −1 or x = −2<br />

5. For any integer n > 0, define φ(n) to be the number of integers coprime to n in<br />

the set<br />

{1,2,3,... ,n}.<br />

Fill in the following table. In the second column you should write all the integers<br />

bet<strong>we</strong>en 1 and n and coprime to n that you will be counting in the third column.<br />

This function φ is called Euler’s Totient Function, and is quite a useful tool. In<br />

the next few questions <strong>we</strong>’ll see that there <strong>are</strong> clever ways of calculating φ(n) other<br />

than writing down all the relevant coprime integers and counting them.


126 Section 9. Tutorial questions<br />

n coprime integers φ(n)<br />

1 1 1<br />

2 1 1<br />

3 1,2 2<br />

4 1,3 2<br />

5 1,2,3,4 4<br />

6<br />

7<br />

8<br />

9<br />

10<br />

11<br />

12<br />

6. Show that if p is prime, φ(p) = p − 1.<br />

7. Show that if p and q <strong>are</strong> distinct prime numbers<br />

φ(pq) = φ(p)φ(q).<br />

(Why do <strong>we</strong> need p and q to be distinct?)<br />

8. Let d > 1 be a squ<strong>are</strong>-free integer, i.e. its prime factorisation has no repeated<br />

factors, or equivalently for all n ∈ Z<br />

n 2 |d ⇒ n 2 = 1.<br />

i) Show that if d|a 2 then d|a. Hint: think about the prime factors of d.<br />

ii) Hence prove that √ d is irrational.<br />

Hint: Copy the proof overleaf that √ 2 is irrational, but be c<strong>are</strong>ful about where<br />

you have to use part (i).


Proof that √ 2 is irrational: by contradiction<br />

Suppose √ 2 is rational.<br />

Put √ 2 = a<br />

b<br />

i.e. hcf(a,b) = 1.<br />

Then 2 = a2<br />

b 2<br />

=⇒ 2b 2 = a 2<br />

=⇒ 2|a 2<br />

where a,b ∈ Z and the fraction is in its lo<strong>we</strong>st terms<br />

=⇒ 2|a since 2 is prime<br />

=⇒ a = 2k for some k ∈ Z<br />

=⇒ 2b 2 = (2k) 2 = 4k 2<br />

=⇒ b 2 = 2k 2<br />

=⇒ 2|b 2<br />

=⇒ 2|b since 2 is prime<br />

So 2|a and 2|b # contradicts hcf(a,b) = 1.<br />

127<br />

Hence √ 2 is irrational as required.


128 Section 9. Tutorial questions<br />

Tutorial questions #4<br />

1. This question is about the group of units in Zn again, but this time <strong>we</strong>’ll go a step<br />

further and analyse what the group of units actually is. This is also to jog your<br />

memory a bit about some group stuff from MAS175. “Recall” that a cyclic group<br />

is one in which there is an element a such that every other element of the group is<br />

a po<strong>we</strong>r of a. Such an a is called a generator.<br />

i) On homework #2 you found the multiplication table for the unit group of Z10.<br />

Find an element of the unit group that has order 4. Remember, the order of<br />

an element a is the smallest integer k > 0 such that a k = 1.<br />

ii) Deduce that this group is cyclic.<br />

Oh OK I’ll do it for you: a group of order n is cyclic if and only if it has an<br />

element of order n. In fact any element of order n is a generator.<br />

iii) The above was the boringly efficient way of showing that the unit group in this<br />

case was cyclic. Something a bit more fun is to rewrite the multiplication table.<br />

Take your element of order 4 – let’s call it a. Then reorder your multiplication<br />

table like this:<br />

× 1 a a 2 a 3<br />

1<br />

a<br />

a 2<br />

a 3<br />

except that you’ll put the actual numbers in, instead of a,a 2 ,a 3 . Now fill in<br />

the table, and you should see the nice swirly pattern associated with a cyclic<br />

group.<br />

iv) Now, since I practically did that whole thing for you, do the whole thing for<br />

Z9. Start by writing down the units and their multiplication table, then show<br />

that the group is cyclic, and reorder the table to get the cyclic pattern.<br />

2. For every element a in Z15 determine whether a is a unit or a zero-divisor. If a is<br />

a unit find its inverse, and if a is a zero-divisor find b = 0 such that ab = 0.<br />

Hints:


129<br />

• You could of course do this by filling in a multiplication table, but I hope<br />

you’ll agree that that would be a bit of a long and boring way to do it.<br />

• We know <strong>we</strong> can find inverses using Euclid’s algorithm, but for numbers as<br />

small as this you may find it easier to do it by trial and error/staring. Don’t<br />

forget that if you can think of b such that ab ≡ −1 then you know that<br />

a.(−b) ≡ 1. This can sometimes help.<br />

• Note that you only really need to work all this out for half of the numbers,<br />

because once you’ve done a you can immediately write down the ans<strong>we</strong>r for<br />

−a in either case, using (−a)(−b) = ab.<br />

3. For each of the following values of a, determine whether a is a unit or a zero-divisor<br />

in Z570. Again, if a is a unit find its inverse, and if a is a zero-divisor find b = 0<br />

such that ab = 0.<br />

i) a = 46 ii) a = 299 iii) a = 105<br />

4. Let d be a squ<strong>are</strong>-free integer, and consider Z[ √ d]. Given an element r = a+b √ d ∈<br />

Z[ √ d] define<br />

N(r) = |a 2 − db 2 |.<br />

This is called the norm of an element, and <strong>we</strong>’ll be seeing more of this useful gadget<br />

soon.<br />

Calculate N(r) for the following elements.<br />

i) 3 + 2 √ 2 ∈ Z[ √ 2]<br />

ii) 8 + 3 √ 7 ∈ Z[ √ 7]<br />

iii) 17 − 12 √ 2 ∈ Z[ √ 2]<br />

5. i) Show that all the above elements <strong>are</strong> units in their respective rings.<br />

ii) What point do you think I’m trying to make here?<br />

6. This question is related to Homework #2, question 3.<br />

i) Let R be a ring, and a,b ∈ R. What does a<br />

mean ?<br />

b<br />

ii) On the homework, you solved the equation<br />

(8 + 3 √ 7)x = 3.<br />

Below is a solution of the app<strong>are</strong>ntly similar equation<br />

(10 + 3 √ 7)x = 3.


130 Section 9. Tutorial questions<br />

This method is the one that some people used in the homework. Ho<strong>we</strong>ver<br />

here it does not give a solution in Z[ √ 7], which means that something in the<br />

solution method isn’t valid in Z[ √ 7]. What is it? Hint: part (i) of this question<br />

is supposed to be relevant. The point of this question is to get you to think a<br />

bit more about what division “is”.<br />

(10 + 3 √ 7)x = 3<br />

x =<br />

=<br />

3<br />

10 + 3 √ 7<br />

3<br />

10 + 3 √ 7 . 10 − 3√7 10 − 3 √ 7<br />

= 30 − 9√ 7<br />

37<br />

iii) Do you think question 5 sheds any light on this matter?<br />

7. Do question 2 again but for Z21.


Tutorial questions #5<br />

131<br />

1. Recall that the norm of a + b √ d ∈ Z[ √ d] is |a 2 − b 2 d|. Find the norm of each the<br />

following elements:<br />

i) (1 + 2 √ 3), (1 − 2 √ 3), (−1 + 2 √ 3), (−1 − 2 √ 3) ∈ Z[ √ 3]<br />

ii) 3 ∈ Z[ √ 2], 3 ∈ Z[ √ 3], 3 ∈ Z[ √ −7], 3 ∈ Z[ √ −11]<br />

iii) (5 + 2 √ 6), (5 − 2 √ 6), (−5 + 2 √ 6), (−5 − 2 √ 6) ∈ Z[ √ 6]<br />

iv) 3 + 2 √ 2 ∈ Z[ √ 2]<br />

v) (5 + 2 √ −11), (2 + 5 √ −11) ∈ Z[ √ −11]<br />

2. Write down three more elements with the same norm as<br />

6174952171 + 6174955132 √ 410533<br />

Those <strong>are</strong> the phone and fax numbers of the Harvard maths department, in case<br />

you <strong>we</strong>re wondering.<br />

3. Which of the elements in Question 1 <strong>are</strong> units of their respective rings? Write<br />

down 4 more units in Z[ √ 2].<br />

Remember that r is a unit iff N(r) = 1.<br />

4. Write a list of all possible norms less than 20 of elements in Z[ √ −7].<br />

Norms have to be non-negative integers, but not all non-negative integers <strong>are</strong> valid<br />

norms in all rings. This gives us important information when <strong>we</strong>’re analysing<br />

those rings, sometimes, as <strong>we</strong>’ll see later.<br />

5. An element r ∈ R is called irreducible if r is not a unit, and<br />

r = st ⇒ s is a unit or t is a unit.<br />

i) Show that if N(st) is prime then either N(s) = 1 or N(t) = 1.<br />

Hint: use N(st) = N(s)N(t)<br />

ii) What does this tell us about r if N(r) is prime?<br />

Hint: what does N(s) = 1 tell us about s?<br />

iii) Which of the elements in Question 1 can <strong>we</strong> now deduce is definitely irre-<br />

ducible?<br />

6. This question is about the converse of the above question. We now know that<br />

N(r) prime ⇒ r irreducible.


132 Section 9. Tutorial questions<br />

But it turns out that r can be irreducible even if N(r) is not prime, and this<br />

depends on the fact that not all integers <strong>are</strong> valid values of N(r), as <strong>we</strong> saw in<br />

question 4.<br />

i) Show that there is no element in Z[ √ −11] whose norm is 3.<br />

Hint: to get a norm of 3, <strong>we</strong>’d have to find a,b ∈ Z such that a 2 + 11b 2 = 3.<br />

Is that possible?<br />

ii) Consider s,t ∈ Z[ √ −11]. Show that if N(st) = 9 then N(s) = 1 or N(t) = 1.<br />

iii) Deduce that 3 is irreducible in Z[ √ −11] even though N(3) is not prime.<br />

iv) Is 3 irreducible in Z[ √ −3]?<br />

7. This question pushes the previous question a bit further and finds some non-unique<br />

factorisation in Z[ √ −11].<br />

i) Show that there is no element of norm 23 in Z[ √ −11].<br />

ii) Show that if N(r) = 69 then r must be irreducible.<br />

Hint: 69 = 3 × 23.<br />

iii) Show that (5+2 √ −11), (5−2 √ −11), 3, and 23 <strong>are</strong> all irreducible in Z[ √ −11].<br />

iv) Write down two different factorisations of 69 into irreducibles in Z[ √ −11].<br />

That is, find rs = r ′ s ′ = 69 where r,s,r ′ ,s ′ <strong>are</strong> all irreducible, and r ′ ,s ′<br />

cannnot be obtained from r,s by multiplying by units. This shows that<br />

Z[ √ −11] is not a UFD.<br />

To do the last part you need to remember what the units in Z[ √ −11] <strong>are</strong>.<br />

8. This question is about expressing integers as the sum of two squ<strong>are</strong>s.<br />

Consider r = (1 + 2i)(3 + 4i) = −5 + 10i ∈ Z[i]. Now N(r) = 5 2 + 10 2 = 125 but<br />

also <strong>we</strong> know that (1 + 2i)(3 − 4i) must have the same norm as r. (Why??) And<br />

(1 + 2i)(3 − 4i) = 11 + 2i<br />

so taking the norm of that, <strong>we</strong> get 11 2 + 2 2 = 125. Expressing things as the sum<br />

of two squ<strong>are</strong>s is Interesting, and look! We’ve done it in two different ways.<br />

Now you do it – consider<br />

(2 + 3i)(4 − 5i) = 23 + 2i<br />

and copy the above procedure to express 533 as a the sum of two squ<strong>are</strong>s in two<br />

different ways.


Tutorial questions #6<br />

Note that this whole sheet is revision about groups.<br />

You should bring your Groups and Symmetries notes to the tutorial.<br />

133<br />

1. This question is to jog your memory about the symmetric groups Sn. We’ll do S9.<br />

i) Write down the following permutation in disjoint cycle notation:<br />

⎛<br />

⎞<br />

1 2 3 4 5 6 7 8 9<br />

⎝ ⎠ .<br />

9 4 1 8 7 5 2 6 3<br />

ii) Write down the following permutation in two-row notation:<br />

iii) Let<br />

and<br />

(1 5 3)(2 7).<br />

⎛<br />

⎞<br />

1 2 3 4 5 6 7 8 9<br />

α = ⎝ ⎠<br />

9 4 1 8 7 5 2 6 3<br />

⎛<br />

⎞<br />

1 2 3 4 5 6 7 8 9<br />

β = ⎝ ⎠.<br />

6 3 1 2 7 4 9 5 8<br />

Calculate βα. Remember this means do α first and then β afterwards.<br />

iv) Let α = (1 5 3) and β = (2 5 1). Calculate αβ and βα, and observe that they<br />

<strong>are</strong> not equal. Permutations do not commute, so Sn is not in general abelian.<br />

v) Write (1 5 4 7)(2 9) as a product of transpositions. Remember a transposition<br />

is a permutation that just swaps two numbers eg (1 6).<br />

2. Recall that a permutation is called even if, when it is written as a product of<br />

transpositions, it turns out to have an even number of transpositions.<br />

i) Show that the even permutations form a subgroup of Sn. This is called the<br />

alternating group An.<br />

ii) Do the odd permutations form a subgroup of Sn?<br />

3. Recall that D3 is the group of symmetries of the equilateral triangle.<br />

i) Write down all the elements of D3.


134 Section 9. Tutorial questions<br />

ii) Pick one reflection and one non-trivial rotation in D3, and express all the<br />

non-identity elements of D3 in terms of these two.<br />

This means that these two elements “generate” the group.<br />

iii) Find six subgroups of D3. (This is all the subgroups of D3).<br />

4. This question is about cosets. Recall that given a subgroup H of a group G, for<br />

each g ∈ G the left coset gH is the set<br />

{gh : h ∈ H}.<br />

Recall also that aH = bH ⇐⇒ a −1 b ∈ H.<br />

Now consider the following subgroup of S4:<br />

H = {e,(1 2)(3 4),(1 3)(2 4),(1 4)(2 3)}<br />

i) Let a ∈ S4 be the permutation (1 2). Write down all the elements of the coset<br />

aH.<br />

ii) Show that (1 4) and (2 3) <strong>are</strong> in the same left coset of H.<br />

iii) How many different left cosets of H <strong>are</strong> there?<br />

5. A subgroup H of G is called a normal subgroup if<br />

g ∈ G and h ∈ H ⇒ g −1 hg ∈ H.<br />

i) Show that if G is abelian, all its subgroups <strong>are</strong> normal.<br />

ii) Let H be the subgroup of D3 containing just the rotations. Convince yourself<br />

that H is a normal subgroup of D3.<br />

iii) Show that An is a normal subgroup of Sn. (See question 2 for definitions.)<br />

We will later show that if H is a normal subgroup, <strong>we</strong> can make the cosets of H<br />

into a group called the quotient group. This is one of the many reasons normal<br />

subgroups <strong>are</strong> important. “Normal” doesn’t mean “ordinary” here!<br />

6. Recall that the orthogonal group O2 is the group of symmetries of the circle. The<br />

elements of this group can be expressed as<br />

rotα = rotate anticlockwise through angle α<br />

refα = reflect in line through the centre at an angle of α<br />

to the horizontal<br />

2<br />

i) Here’s another way of expressing this group. Write α for rotα, and M = ref0<br />

ie reflection in the “x-axis”. In this new notation, what is refα?


135<br />

ii) Show that O2 has a subgroup isomorphic to D3. You can do this by algebra<br />

or by geometry.<br />

iii) Write down four other subgroups of O2 other than the trivial group and the<br />

whole group.


136 Section 9. Tutorial questions<br />

Tutorial questions #7<br />

1. This question is about the symmetric group S3, but the analogous result is true<br />

for Sn in general.<br />

i) Write down all the elements of S3, in cycle notation. (Check you have the<br />

right number of them.)<br />

ii) Write down all the elements of S3 as products of transpositions. Hence, find<br />

the even permutations i.e. the elements of A3.<br />

iii) Find all the left cosets of A3 in S3.<br />

iv) Check that, for all α ∈ S3, αA3 = A3α. This means that A3 is a normal<br />

subgroup of S3.<br />

v) When H is a normal subgroup of G then <strong>we</strong> can make the left cosets of H<br />

into a group called the “quotient group” G/H. What is the quotient group<br />

S3/A3? Hint: what is its order?<br />

2. i) Write 4Z for the subset of Z given by<br />

{k ∈ Z s.t. 4|k}.<br />

Show that 4Z is a subgroup of Z. Remember: Z is a group under addition.<br />

ii) What <strong>are</strong> all the left cosets of 4Z in Z? Note that since Z is a group under<br />

addition, a left coset of a subgroup H will be written a + H instead of aH.<br />

iii) We can define an operation ⊕ on the cosets of H = 4Z by<br />

(a + H) ⊕ (b + H) = (a + b) + H.<br />

This makes the cosets of H into a group. In what guise have you seen this<br />

group before?<br />

3. i) Consider D6, the symmetry group of the regular hexagon. What happens if<br />

you conjugate a rotation by a reflection? That is, if a is a rotation and b is a<br />

reflection, what is bab −1 ?<br />

ii) What happens in O2, the symmetry group of the circle?<br />

4. This question is about cycle types in S5. The “cycle type” of an element α of Sn<br />

is found as follows: write α in disjoint cycle form, with the cycles in descending<br />

order of length. Then list the lengths of the cycles (including the trivial ones of<br />

length 1 at the end). For example, the element (1 3)(5 2 6) ∈ S6 can be re-written<br />

(5 2 6)(1 3)(4)


137<br />

and has cycle type (3,2,1). The element (1 6)(2 4) has cycle type (2,2,1,1).<br />

i) Find all the possible cycle types of elements in S5.<br />

ii) Which ones <strong>are</strong> odd and which ones <strong>are</strong> even? Hint: to work this out, you<br />

need to rewrite the cycles as products of tranpositions.<br />

iii) How many elements of each cycle type <strong>are</strong> there?<br />

5. Conjugate the element (1 2 3)(4 5) by the following elements. What do you notice<br />

about the resulting cycle types?<br />

i) (3 4) ii) (1 4 5) iii) (1 2 3 4)


138 Section 9. Tutorial questions<br />

Tutorial questions #8<br />

1. Calculate θαθ−1 ⎛<br />

⎞<br />

1 2 3 4 5 6 7 8 9<br />

where θ = ⎝ ⎠ and α = (1 9 6 3)(2 4 8)(5 7).<br />

4 1 9 6 5 2 3 7 8<br />

2. For each of the following elements α of S9, find the number of elements of S9 of<br />

the same cycle type.<br />

i) (1 2)<br />

ii) (1 2 3)<br />

iii) (1 2 3 4)<br />

iv) (1 2 3 4 5)<br />

v) (1 2 3 4)(5 6 7)<br />

vi) (1 9 6 3)(2 4 8)(5 7)<br />

vii) (1 2)(3 4)<br />

3. For each of the elements α in question 2, find |conjS9 (α)| and |centS9 (α)|. Re-<br />

member:<br />

• In Sn, the elements conjugate to an element α <strong>are</strong> those with the same cycle<br />

type.<br />

• In any finite group G, |conj G(α)|.|centG(α)| = |G|.<br />

4. Let A,B be the following matrices with entries in Z7:<br />

⎛ ⎞ ⎛ ⎞<br />

1 2<br />

A = ⎝ ⎠ ,<br />

4 4<br />

B = ⎝ ⎠ .<br />

3 4 2 6<br />

i) Find the determinant of each of these matrices, to check that A ∈ GL2(Z7)<br />

and B ∈ GL2(Z7).<br />

Recall that GL2(Z7) is the group of invertible 2 × 2 matrices with entries in<br />

Z7.<br />

ii) Show that<br />

iii) Compute the conjugate ABA −1 .<br />

A −1 ⎛ ⎞<br />

5 1<br />

= ⎝ ⎠.<br />

5 3<br />

iv) Find the order of ABA −1 and hence find the order of B.<br />

5. In this question you should use the following facts:


139<br />

• The order of each conjugacy class must divide |G|, the order of the group G.<br />

• Every element is in precisely one conjugacy class, so orders of the conjugacy<br />

classes must add up to |G|. So for a group of order 12, for example, <strong>we</strong> could<br />

have 1,3,4,4 or 1,2,3,6 or 1,2,3,3,3 but not 1,3,3,3.<br />

i) What is the order of the conjugacy class of the identity e in any group? Hint:<br />

how many elements b,g can you find such that g = beb −1 ?<br />

ii) In a group of order 6, is it possible to have three conjugacy classes of order 2?<br />

Is it possible to have two conjugacy classes of order 3? What <strong>are</strong> the possible<br />

combinations of conjugacy class orders?<br />

iii) What <strong>are</strong> the possible combinations for a group of order 8?<br />

iv) What <strong>are</strong> the possible combinations for a group of order 10?


140 Section 9. Tutorial questions<br />

Tutorial questions #9<br />

1. Find the following conjugates in O2:<br />

i) rotφ rotθ(rotφ) −1<br />

ii) refφ rotθ(refφ) −1<br />

iii) rotφ refθ(rotφ) −1<br />

iv) refφ refθ(refφ) −1<br />

Hint: it may help you to remember refφ = rotφ ref0.<br />

2. In each of the following either find θ ∈ S7 such that θαθ −1 = β or explain why no<br />

such θ exists:<br />

i) α = (1 3 5 2 4)(6 7), β = (1 2)(3 4 5 6 7)<br />

ii) α = (1 2)(3 4)(5 6), β = (1 2)(6 7)<br />

3. Calculate θαθ−1 ⎛<br />

⎞<br />

1 2 3 4 5 6 7 8 9<br />

where θ = ⎝ ⎠ and α = (1 9 6 3 5)(2 4 8).<br />

4 1 9 6 5 2 3 7 8<br />

You should be able to write this straight down, without even working out what θ −1<br />

is!<br />

4. You <strong>are</strong> given that<br />

H = {e,(1 2)(3 4),(1 3)(2 4),(1 4)(2 3)}<br />

is a normal subgroup of S4, where e is the identity element of S4. Show that the<br />

following hold in S4/H:<br />

i) the inverse of (1 2)H is (3 4)H<br />

ii) (1 2 3)H(1 3 4)H = (1 3 4)H(1 2 3)H<br />

Hint: it will help you to use the fact that the identity element in the quotient group<br />

is eH which is the coset H.<br />

5. Let G be a group of order 21 and suppose that the centre of G is {e}.<br />

i) Determine the class equation for G.<br />

ii) Find the number of elements of order 3 and the number of elements of order<br />

7 in G.<br />

Hint: let a be an element of a conjugacy class of size 3. Now a is also an<br />

element of its own centraliser, which is a subgroup of G. We know what<br />

the order of this subgroup is, therefore <strong>we</strong> can work out the order of a. Do<br />

something similar for elements in conjugacy classes of size 7.


141<br />

iii) Let h ∈ G be an element of order 7. Let H = 〈h〉, the subgroup generated by<br />

h. Use the class equation to show that H is a normal subgroup of G.<br />

iv) Show that H is the only normal subgroup of G other than the trivial group<br />

{e} and G itself.<br />

v) Can you think of some other n (other than 21) where you could use this sort<br />

of argument on a group of order n?


142 Section 9. Tutorial questions<br />

Tutorial questions #10<br />

1. Consider the cyclic group C12 with elements<br />

2. Let<br />

i) Let<br />

Are A and B subgroups?<br />

ii) What is the set AB?<br />

{e,a,a 2 ,a 3 ,a 4 ,a 5 ,a 6 ,a 7 ,a 8 ,a 9 ,a 10 ,a 11 }<br />

A = {a 2 ,a 3 ,a 4 }<br />

B = {a 3 ,a 4 ,a 5 }.<br />

iii) Let H be the subgroup {e,a 4 ,a 8 }. Write down the elements of the cosets a 2 H<br />

and a 3 H.<br />

iv) What is the set (a 2 H)(a 3 H)? First work this out by multiplying every element<br />

of a 2 H by every element of a 3 H. Then check that it is the coset you <strong>we</strong>re<br />

expecting.<br />

v) What is the quotient group C12/H?<br />

p1 = (x1 + x2)(x3 + x4)<br />

p2 = (x1 + x3)(x2 + x4)<br />

p3 = (x1 + x4)(x2 + x3).<br />

The symmetric group S4 acts on these polynomials by permuting the variables,<br />

that is, for α ∈ S4<br />

α ∗ (xi + xj)(xk + xl) = (x α(i) + x α(j))(x α(k) + x α(l)).<br />

This definition of how the symmetric group acts on the polynomials is very formal<br />

and a bit impenetrable. So in this question <strong>we</strong>’re going to sit down and fiddle<br />

around with it to see what it actually does<br />

i) The element (2 3) acts on each polynomial by switching 2 and 3, so in effect<br />

x2 becomes x3 and x3 becomes x2. Write down the result of doing this to the<br />

polynomial<br />

(x1 + x2)(x3 + x4).


143<br />

ii) What you have just done is apply the element (2 3) to the polynomial p1.<br />

Which of p1,p2,p3 was the result?<br />

iii) Now apply (2 3) to p2 and to p3 and see which polynomial results in each case.<br />

iv) This means that the element (2 3) has permuted the polynomials {p1,p2,p3}.<br />

So <strong>we</strong> have produced a permutation in S3. Which one?<br />

v) Now try it for the element (1 2)(3 4). What happens if you apply this to each<br />

of p1,p2,p3?<br />

vi) So, what permutation of S3 have you produced from (1 2)(3 4)?<br />

vii) Find all the permutations in S4 that leave p1 fixed. This is the stabiliser of<br />

p1.<br />

viii) Find all the permutations that fix each of p1, p2, p3.<br />

ix) The above method gives a homomorphism f : S4 −→ S3. What is its kernel?<br />

x) What does the First Isomorphism Theorem tell us about this situation?

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!