HT2026 Groups & Group Actions Lecture Notes


Remaining TODOs: 4


1. Groups

Definition 1.1

Let 𝑆 be a set. A binary operation on 𝑆 is a function βˆ—:𝑆×𝑆→𝑆.

We define π‘Žβˆ—π‘β‰”βˆ—(π‘Ž,𝑏).

Remark 1.2

If βˆ— is understood implicitly, we often write π‘Žπ‘ for π‘Žβˆ—π‘.

Definition 1.3

βˆ— admits an identity element if βˆƒπ‘’βˆˆπ‘† s.t. π‘Žβˆ—π‘’=π‘Ž=π‘’βˆ—π‘Ž βˆ€π‘Žβˆˆπ‘†.

Definition 1.4

Suppose that 𝑒 is an identity element, then inverses exist for (𝑆,βˆ—) if βˆ€π‘Žβˆˆπ‘† βˆƒπ‘βˆˆπ‘† s.t. π‘Žβˆ—π‘=𝑒=π‘βˆ—π‘Ž.

Definition 1.5

A group is a pair (𝐺,βˆ—) where βˆ—:𝐺×𝐺→𝐺 is a binary operation which is associative, admits an identity, and for which inverses exist.

Definition 1.6

𝐺 is an abelian group if βˆ— is commutative.

Lemma 1.7

Let βˆ— be a binary operation on 𝑆.

  1. If 𝑒1,𝑒2 are identity elements, then 𝑒1=𝑒2.
  2. If π‘’βˆˆπ‘† is an identity element, π‘Žβˆˆπ‘†, βˆ— is associative, and 𝑏1,𝑏2 are inverses of π‘Ž under βˆ—, then 𝑏1=𝑏2.

Proof.

  1. 𝑒1=𝑒1βˆ—π‘’2=𝑒2
  2. 𝑏1=(id)𝑏1βˆ—π‘’=(inv)𝑏1βˆ—(π‘Žβˆ—π‘2)=(assoc)(𝑏1βˆ—π‘Ž)βˆ—π‘2=(inv)π‘’βˆ—π‘2=(id)𝑏2

⁠░

Definition 1.8

Let (𝑆,βˆ—) be a set with a binary operation, which admits inverses, then βˆ€π‘Žβˆˆπ‘†, we write π‘Žβˆ’1 to mean the inverse of π‘Ž.

Remark 1.9

βˆ’π‘Ž is also commonly used to denote the inverse of π‘Ž if (𝑆,+) is an abelian group.

Example 1.10 (symmetry group)

Definition 1.11

Let Sym(𝑋)≔{𝑓:𝑋→𝑋|𝑓bijective}.

Proposition 1.12

(Sym(𝑋),∘) is a group.

Proof. Suppose 𝑓,𝑔 are bijections. Then:

  • π‘“βˆ˜π‘” is injective:

    (π‘“βˆ˜π‘”)(π‘₯)=(π‘“βˆ˜π‘”)(𝑦)βŸΉπ‘“(𝑔(π‘₯))=𝑓(𝑔(𝑦))βŸΉπ‘”(π‘₯)=𝑔(𝑦)⟹π‘₯=𝑦
  • π‘“βˆ˜π‘” is surjective:

    βˆ€π‘₯βˆˆπ‘‹,βˆƒπ‘¦βˆˆπ‘‹s.t.𝑧=𝑓(𝑦)βˆƒπ‘₯βˆˆπ‘‹s.t.𝑦=𝑔(π‘₯)βŸΉπ‘§=𝑓(𝑔(π‘₯))=(π‘“βˆ˜π‘”)(π‘₯).

So π‘“βˆ˜π‘” is a bijection, and π‘“βˆ˜π‘” is a binary operation on Sym(π‘₯).

Composition is associative.

We have an identity, id𝑋:𝑋→𝑋;π‘₯↦π‘₯, which satisfies π‘“βˆ˜idπ‘₯=𝑓=idπ‘‹βˆ˜π‘“, βˆ€π‘“βˆˆSym(𝑓).

We have inverses: π‘“βˆ˜π‘“βˆ’1=π‘“βˆ’1βˆ˜π‘“=id𝑋.⁠░

Definition 1.13

If 𝑋={1,2,…,𝑛}, 𝑆𝑛≔Sym(𝑋) is the 𝑛th symmetric group.

Remark 1.14

Sym(𝑋) can also be interpreted as the permutations of 𝑋.

Example 1.15 (fields)

Let (𝐹,+,Γ—) be a field.

Then (𝐹,+) is an abelian group, with identity element 0𝐹, whereas (𝐹\{0𝐹},Γ—) (sometimes denoted πΉβˆ—) is also an abelian group with identity element 1𝐹.

For example, β„šβŠ‚β„βŠ‚β„‚ all form groups under addition, or under multiplication when zero is excluded.

Also β„€π‘βˆ—β‰” non-zero integers mod p, for prime 𝑝, is a group under multiplication.

Example 1.16 (matrices)

(𝑀𝑛(𝐹),+) is an abelian group (where 𝑀𝑛(𝐹) is the set of 𝑛×𝑛 matrices over 𝐹).

Definition 1.17

Define GL𝑛(𝐹)≔{π΄βˆˆπ‘€π‘›(𝐹)|det𝐴≠0}; (GL𝑛(𝐹),β‹…) is the 𝑛th general linear group over 𝐹.

Definition 1.18

Let SL𝑛(𝐹)≔{𝐴∈GL𝑛(𝐹)|det𝐴=1}; this is the 𝑛th standard linear group over 𝐹 under matrix multiplication.

Example 1.19

Definition 1.20

Let 𝑋 be a set, let π‘ŒβŠ†π‘‹.

Let Stab(π‘Œ)≔{π‘“βˆˆSym(𝑋):𝑓(π‘Œ)=π‘Œ}; that is, Stab(π‘Œ) is the set of permutations that map π‘Œ to itself.

Proposition 1.21

(Stab(π‘Œ),∘) is a group.

Proof. Necessary to prove: βˆ€π‘“,π‘”βˆˆStab(π‘Œ),π‘“βˆ˜π‘”βˆˆStab(π‘Œ): (π‘“βˆ˜π‘”)(π‘Œ)=𝑓(𝑔(π‘Œ))=𝑓(π‘Œ)=π‘Œ.

Associativity is inherited from the symmetric group.

id𝑋(π‘Œ)=π‘ŒβŸΉidπ‘‹βˆˆStab(π‘Œ).

Let π‘“βˆˆStab(π‘Œ), then 𝑓(π‘Œ)=π‘Œ. Hence π‘“βˆ’1(𝑓(π‘Œ))=π‘Œ, and π‘“βˆ’1(𝑓(π‘Œ))=π‘“βˆ’1(π‘Œ), so π‘“βˆ’1=π‘Œ, so π‘“βˆ’1∈Stab(π‘Œ).⁠░

Definition 1.22

(Stab(π‘Œ),∘) is the stabiliser subgroup of π‘Œ.

Definition 1.23

Let 𝐺 be a group, with binary operation βˆ—. A subset π»βŠ†πΊ is a subgroup if βˆ— restricts to a map βˆ—:𝐻×𝐻→𝐻 which then makes (𝐻,βˆ—/𝐻) into a group. We write 𝐻≀𝐺.

Example 1.24

(ℝ,+)≀(β„‚,+)

Proposition 1.25 (Subgroup Test)

Let 𝐺 be a group. A subset 𝐻 of 𝐺 is a subgroup iff :

  1. π‘’βˆˆπ»
  2. βˆ€π‘₯,π‘¦βˆˆπ», π‘₯π‘¦βˆ’1∈𝐻

Proof.

β€œβŸΉβ€:

Assume 𝐻 is a subgroup of 𝐺. Then 𝐻 is closed under multiplication and inversion. Therefore βˆ€π‘₯,π‘¦βˆˆπ», π‘₯π‘¦βˆ’1∈𝐻 and also π‘’βˆˆπ».

β€œβŸΈβ€: Associativity on 𝐻 is inherited from 𝐺.

(i) βŸΉπ‘’βˆˆπ».

Apply (ii) with π‘₯=𝑒, giving βˆ€π‘¦βˆˆπ»,π‘¦βˆ’1∈𝐻.

So (𝐻,βˆ—/𝐻) is a group.⁠░

Example 1.26

Let 𝐺=GL𝑛(ℝ).

Let O(𝑛)={π€βˆˆπΊ|π€βŠ€π€=𝐈}, the orthogonal matrices.

  1. 𝐈⊀𝐈=𝐈⟹𝐈∈O(𝑛).
  2. 𝐀,𝐁∈O(𝑛)⟹(π€πβˆ’1)⊀(π€πβˆ’1)=(πβˆ’1)βŠ€π€βŠ€π€πβˆ’1=(πβˆ’1)βŠ€πβˆ’1=(𝐁⊀)βˆ’1πβˆ’1=(𝐁𝐁⊀)βˆ’1=πˆβˆ’1(𝐁⊀𝐁=𝐈⟹𝐁𝐁⊀=𝐈for square𝐁)=𝐈

Hence by the subgroup test, O(𝑛)≀GL𝑛(ℝ).

Also SL𝑛(𝐹)≀GL𝑛(𝐹) can be shown by the subgroup test.

Lemma 1.27

If 𝐻≀𝐺 and 𝐾≀𝐺, then π»βˆ©πΎβ‰€πΊ.

Corollary 1.27.1

SO(𝑛)=SL𝑛(ℝ)∩O(𝑛)≀GL𝑛(ℝ).

Definition 1.28

Let (𝐺,βˆ—πΊ),(𝐻,βˆ—π») be groups.

The product group is (𝐺×𝐻,βˆ—πΊΓ—π»), where (𝑔1,β„Ž1)βˆ—πΊΓ—π»(𝑔2,β„Ž2)=(𝑔1βˆ—πΊπ‘”2, β„Ž1βˆ—π»β„Ž2).

Lemma 1.29

The product of two groups is a group.

Proof. Associativity is easy.

(𝑔,β„Ž) (𝑒,𝑒)=(𝑔𝑒,β„Žπ‘’)=(𝑔,β„Ž)=(𝑒,𝑒) (𝑔,β„Ž), so (𝑒,𝑒) is an identity.

The inverse of (𝑔,β„Ž) is (π‘”βˆ’1,β„Žβˆ’1).⁠░

Definition 1.30

Let 𝐺 be a group. 𝐺 is cyclic if βˆƒπ‘”βˆˆπΊ s.t. 𝐺={𝑔𝑛|π‘›βˆˆβ„€}, where

𝑔𝑛={π‘”π‘”π‘›βˆ’1if𝑛>0𝑒if𝑛=0(𝑔𝑛)βˆ’1if𝑛<0.

𝑔 is known as a generator of 𝐺.

Example 1.31

Definition 1.32

𝐺=(β„€,+) is a cyclic group with 𝑔=1.

This is the infinite cyclic group.

Definition 1.33

For π‘›βˆˆβ„€>0, 𝐢𝑛={𝑒,𝑔,𝑔2,…,π‘”π‘›βˆ’1} is a cyclic group with binary operation

π‘”π‘Žπ‘”π‘={π‘”π‘Ž+𝑏ifπ‘Ž+π‘β‰€π‘›βˆ’1π‘”π‘Ž+π‘βˆ’π‘›ifπ‘›β‰€π‘Ž+𝑏≀2π‘›βˆ’2
and inverse (π‘”π‘Ž)βˆ’1=π‘”π‘›βˆ’π‘Ž for 0β‰€π‘Ž<𝑛. 𝐢𝑛 is the cyclic group of order 𝑛.

Remark 1.34

A cyclic group can have more than one generator; for instance, (β„€,+) also has βˆ’1 as a generator.

Remark 1.35

𝐢5=βŸ¨π‘”2⟩, where, for β„Žβˆˆπ», we write βŸ¨β„ŽβŸ©β‰”{β„Žπ‘›|π‘›βˆˆβ„€}. This is because 2 and 5 are coprime.

Definition 1.36

An isometry 𝑇:ℝ𝑛→ℝ𝑛 is a distance-preserving function, i.e. |𝑇(𝒗)βˆ’π‘‡(π’˜)|=|π’—βˆ’π’˜| for all 𝒗,π’˜βˆˆβ„π‘›.

We write the set of isometries over ℝ𝑛 as Isom(ℝ𝑛).

Remark 1.37

By the subgroup test, Isom(ℝ𝑛)≀Sym(ℝ𝑛).

Remark 1.38

βˆ€π‘‡βˆˆIsom(ℝ𝑛),βˆƒ!π€βˆˆπ‘‚(𝑛),π’ƒβˆˆβ„π‘› s.t. 𝑇(𝒗)=𝐀𝒗+𝒃 for all π’—βˆˆβ„π‘›.

Definition 1.39

Let π‘›βˆˆβ„€>0. Let 𝑃𝑛 be the regular 𝑛-sided polygon centred at the origin 𝑂.

The dihedral group 𝐷2𝑛 is

𝐷2𝑛≔Isom(ℝ2)∩Stab(𝑃𝑛);

that is, it is the group of distance-preserving transformations that stabilise the regular 𝑛-sided polygon centred at the origin.

We denote the rotation by 2πœ‹/𝑛 anticlockwise by π‘Ÿ, and the reflection in an axis (which one depends on which way we choose to draw the polygons) by 𝑠. Then the elements of 𝐷2𝑛 are 𝑒,π‘Ÿ,π‘Ÿ2,…,π‘Ÿπ‘›βˆ’1,𝑠,π‘Ÿπ‘ ,π‘Ÿ2𝑠,…,π‘Ÿπ‘›βˆ’1𝑠. Note that this is 2𝑛 elements.

Proposition 1.40

𝑆2𝑛={π‘Ÿπ‘–π‘ π‘—|0≀𝑖<𝑛,π‘—βˆˆ{0,1}}.

Proof. Suppose π‘Ÿπ‘–π‘ π‘—=π‘Ÿπ‘Žπ‘ π‘ w.l.o.g. 𝑖β‰₯π‘Ž. Then π‘Ÿβˆ’π‘Žπ‘Ÿπ‘–π‘ π‘—=π‘Ÿβˆ’π‘Žπ‘Ÿπ‘Žπ‘ π‘=𝑠𝑏, so w.l.o.g. π‘Ž=0.

If 𝑗=𝑏, then π‘Ÿπ‘–=π‘Ÿ0=π‘’βŸΉπ‘Ÿπ‘–(𝑝0)=𝑒(𝑝0) for some point 𝑝0, so 𝑖=0.

If 𝑗≠𝑏, π‘Ÿπ‘–=𝑠. Take dets to get 1=βˆ’1, contradiction.

Now let π‘ƒβˆˆπ·2𝑛. Then 𝑃(𝑝0)=𝑝𝑖=π‘Ÿπ‘–(𝑝0) for some π‘–βˆˆ{0,…,π‘›βˆ’1}.

Hence π‘Ÿβˆ’π‘–π‘ƒ(𝑝0)=𝑝0.

Now |π‘Ÿβˆ’π‘–π‘ƒ(𝑝1)βˆ’π‘Ÿβˆ’π‘–π‘ƒ(𝑝0)|=|𝑝1βˆ’π‘0|, hence π‘Ÿβˆ’π‘–π‘ƒ(𝑝1)∈{𝑝1,π‘π‘›βˆ’1}.

TODO finish this off⁠░

Definition 1.41

The order of a group 𝐺 is |𝐺|, the number of elements in 𝐺.

Definition 1.42

If 𝐺={𝑔1,…,𝑔𝑛}, its Cayley table is an 𝑛×𝑛 array where with 𝑔𝑖𝑔𝑗 in the 𝑖th row and 𝑗th column.

Definition 1.43

Let 𝐺,𝐻 be groups. An isomorphism between 𝐺 and 𝐻 is a bijection πœ‘:𝐺→𝐻 s.t.

πœ‘(𝑔1𝑔2)=πœ‘(𝑔1) πœ‘(𝑔2).

If there is an isomorphism between 𝐺 and 𝐻, then 𝐺 and 𝐻 are isomorphic, written 𝐺≅𝐻.

Remark 1.44

Isomorphic finite groups will have the β€œsame” Cayley tables (up to row/column orders).

Definition 1.45

For a group 𝐺, the order of an element π‘”βˆˆπΊ is

π‘œ(𝑔)≔min{π‘Ÿβˆˆβ„€>0|π‘”π‘Ÿ=𝑒}.

If π‘”π‘Ÿβ‰ π‘’ for all π‘Ÿβˆˆβ„€>0, we say that 𝑔 has infinite order.

Lemma 1.46

If πœ‘:𝐺→𝐻 is an isomorphism, then βˆ€π‘”βˆˆπΊ,

π‘œ(πœ‘(𝑔))=π‘œ(𝑔).

Proof.

First we show that πœ‘(𝑒)=𝑒:

πœ‘(𝑒)=πœ‘(𝑒2)=πœ‘(𝑒)2βŸΉπ‘’=πœ‘(𝑒)βˆ’1πœ‘(𝑒)=πœ‘(𝑒)βˆ’1πœ‘(𝑒)2=πœ‘(𝑒).

Then

π‘”π‘Ÿ=π‘’βŸΊπœ‘(π‘”π‘Ÿ)=πœ‘(𝑒)=π‘’βŸΊπœ‘(𝑔)π‘Ÿ=𝑒.

Hence

π‘œ(𝑔)=min{π‘Ÿ>0|π‘”π‘Ÿ=𝑒}=min{π‘Ÿ>0|πœ‘(𝑔)π‘Ÿ=𝑒}=π‘œ(πœ‘(𝑔)).

2. Permutation groups

Remark 2.1

|𝑆𝑛|=𝑛!.

Remark 2.2

We write permutations on the right, i.e. if π‘–βˆˆ{1,…,𝑛},πœŽβˆˆπ‘†π‘›, then π‘–πœŽβ‰”πœŽ(𝑖).

This means that composition is backwards from the usual order.

Definition 2.3

A cycle is a permutation πœŽβˆˆπ‘†π‘› such that βˆƒ{π‘Ž1,…,π‘Žπ‘š}βŠ†{1,…,𝑛} s.t. π‘Žπ‘˜πœŽ=π‘Žπ‘˜+1 for π‘˜βˆˆ{1,…,π‘šβˆ’1} and π‘Žπ‘šπœŽ=π‘Ž1, and for all π‘₯∈{1,…,𝑛}\{π‘Ž1,…,π‘Žπ‘š}, π‘₯𝜎=π‘₯.

We denote this cycle as (π‘Ž1 π‘Ž2 … π‘Žπ‘š).

Remark 2.4

We could denote 𝑒=(1)(2)β‹―(𝑛).

Lemma 2.5

Disjoint cycles commute.

Proof. Let 𝜎=(π‘Ž1 … π‘Žπ‘š) and 𝜏=(𝑏1 … π‘π‘˜) be disjoint cycles, i.e. {π‘Ž1,…,π‘Žπ‘š}∩{𝑏1,…,π‘π‘˜}=βˆ….

Let π‘–βˆˆ{1,…,𝑛}. Then

π‘–πœŽπœ={π‘Žπ‘—+1𝜏=π‘Žπ‘—+1if𝑖=π‘Žπ‘—π‘–πœ=𝑏𝑗+1if𝑖=𝑏𝑗𝑖otherwise;
π‘–πœπœŽ={π‘–πœŽ=π‘Žπ‘—+1if𝑖=π‘Žπ‘—π‘π‘—+1𝜎=𝑏𝑗+1if𝑖=𝑏𝑗𝑖otherwise.

Theorem 2.6

Every πœŽβˆˆπ‘†π‘› can be written as a product of disjoint cycles.

Moreover, this factorisation is unique up to cycling of elements within cycles, and order of (commuting) factors.

Proof. Fix π‘Ž1∈{1,…,𝑛} and consider π‘Ž1,π‘Ž1𝜎, π‘Ž1𝜎2,….

There exists 0≀𝑖<𝑗 such that π‘Ž1πœŽπ‘–=π‘Ž1πœŽπ‘˜. We choose such a pair (𝑖,𝑗) with minimal 𝑖.

Then applying πœŽβˆ’π‘–, we get that π‘Ž1=π‘Ž1πœŽπ‘—βˆ’π‘–, and note that π‘œ<π‘—βˆ’π‘–. Since 𝑖 was minimal, 𝑖=0.

Let 𝑗>0 be the least such that π‘Ž1πœŽπ‘—=𝜎1, then π‘Ž1,π‘Ž1𝜎,π‘Ž1𝜎2,…,π‘Ž1πœŽπ‘—βˆ’1 must all be distinct.

The set {π‘Ž1,π‘Ž1𝜎,…,π‘Ž1πœŽπ‘—βˆ’1} is known as the orbit of π‘Ž1 under 𝜎, orbit(π‘Ž1).

Note that orbit(π‘Ž1)={π‘Ž1πœŽπ‘›|π‘›βˆˆβ„•}={π‘Ž1πœŽπ‘›|π‘›βˆˆβ„€}.

Now pick π‘Ž2∈{π‘Ž,…,𝑛}\orbit(π‘Ž1). orbit(π‘Ž2) must be disjoint from orbit(π‘Ž1), otherwise π‘Ž2∈orbit(π‘Ž1), since if π‘Ž1πœŽπ‘Ÿ=π‘Ž2πœŽπ‘ , then π‘Ž2=π‘Ž1πœŽπ‘Ÿβˆ’π‘ .

We continue until we have a partition

{1,…,𝑛}=orbit(π‘Ž1)βˆͺ…βˆͺorbit(π‘Žπ‘Ÿ).

Then 𝜎=(π‘Ž1 π‘Ž1𝜎 … π‘Ž1πœŽπ‘—1βˆ’1)(π‘Ž2 π‘Ž2𝜎 … π‘Ž2πœŽπ‘—2βˆ’1)…(π‘Žπ‘Ÿ π‘Žπ‘ŸπœŽ … π‘Žπ‘ŸπœŽπ‘—π‘Ÿβˆ’1), where 𝑗𝑖=|orbit(π‘Žπ‘–)| for π‘–βˆˆ{1,…,π‘Ÿ}.

If 𝜎=𝑐1⋅𝑐2β‹…β€¦β‹…π‘π‘Ÿ=𝑐1′⋅𝑐2′⋅….β‹…π‘π‘Ÿβ€², then 1 must appear in one of the 𝑐𝑖′s, supposing W.L.O.G. that 𝑖=1. Also 1 appears in 𝑐1. Then 𝑐1=(1 1𝜎 … 1πœŽπ‘—1βˆ’2), and 𝑐1β€²=(1 1𝜎 … 1πœŽπ‘—1β€²βˆ’1), hence 𝑗1=𝑗1β€² and 𝑐1=𝑐1β€². Repeat for all 𝑐𝑖′s, then the factorisation is unique up to commutation and cycling.⁠░

Proposition 2.7

Let 𝜎=𝜌1𝜌2β€¦πœŒπ‘˜βˆˆπ‘†π‘› where the πœŒπ‘–s are disjoint cycles. Then π‘œ(𝜎)=lcm(π‘œ(𝜌1),…,π‘œ(πœŒπ‘˜)) and π‘œ(πœŒπ‘–)=length ofπœŒπ‘–.

Proof. If 𝑐 is a cycle of length 𝑙, 𝑐=(π‘Ž1 π‘Ž2 … π‘Žπ‘™), then π‘Žπ‘–π‘π‘™=π‘Žπ‘–, whereas if π‘₯β‰ π‘Žπ‘– then π‘₯𝑐=π‘₯ so π‘₯𝑐𝑙=π‘₯. Therefore 𝑐𝑙=𝑒.

Conversely, if π‘π‘š=𝑒 for π‘š>0 then π‘Ž1π‘π‘š=π‘Ž1𝑒=π‘Ž1, so π‘šβ‰₯𝑙 therefore π‘œ(𝑐)=𝑙.

Then let 𝐿≔lcm(𝑙𝑖),𝑙𝑖=length(πœŒπ‘–)=π‘œ(πœŒπ‘–).

Then π‘™π‘–βˆ£πΏ βˆ€π‘–, so 𝜎𝐿=(𝜌1 𝜌2 … πœŒπ‘˜)𝐿=𝜌1π‘™πœŒ2π‘™β€¦πœŒπ‘˜πΏ (note that this is not a valid application of distributivity in the general case!).

Lemma 2.8

If π‘”βˆˆπΊ has finite order, then for π‘›βˆˆβ„€, 𝑔𝑛=𝑒 iff π‘œ(𝑔)βˆ£π‘›.

Proof. β€œβŸΈβ€:

If π‘œ(𝑔)βˆ£π‘›, then 𝑛=π‘œ(𝑔)π‘š, hence 𝑔𝑛=π‘”π‘œ(𝑔)π‘š=(π‘”π‘œ(𝑔))π‘š=π‘’π‘š=𝑒.

β€œβŸΉβ€:

We write 𝑛=π‘žπ‘œ(𝑔)+π‘Ÿ for π‘žβˆˆβ„€ and 0β‰€π‘Ÿ<π‘œ(𝑔). Then 𝑔𝑛=π‘’βŸΉπ‘”π‘žπ‘œ(𝑔)+π‘Ÿ=𝑒 hence π‘”π‘Ÿ=𝑒. But π‘Ÿ<π‘œ(𝑔), so π‘Ÿ=0 by the defn of π‘œ(𝑔).⁠░

We have that 𝜎𝐿=𝑒; by the lemma, π‘œ(𝜎)∣𝐿.

Now consider πœŽπ‘œ(𝜎)=𝑒. Then (𝜌1 … πœŒπ‘˜)π‘œ(𝜎)=𝑒 hence 𝜌1π‘œ(𝜎)β€¦πœŒπ‘˜π‘œ(𝜎)=𝑒.

If we take π‘¦βˆˆ{1,…,𝑛}, saying W.L.O.G. that 𝑦 appears in πœŒπ‘–. Then 𝑦 does not appear in any other πœŒπ‘—, hence π‘¦πœŒπ‘–π‘œ(𝜎)=π‘¦βŸΉπœŒπ‘–π‘œ(𝜎)=𝑒 for all 𝑖.

By the lemma, π‘œ(πœŒπ‘–)βˆ£π‘œ(𝜎), so 𝐿=lcm(π‘œ(πœŒπ‘–))βˆ£π‘œ(𝜎).⁠░

Definition 2.9

For πœŽβˆˆπ‘†π‘›, we define π‘ƒπœŽβˆˆGL𝑛(ℝ) to besuch that the 𝑖th row of π‘ƒπœŽ has a 1 in column π‘–πœŽ, and has all other entries zero. Then π‘ƒπœŽ is a permutation matrix.

That is,

(π‘ƒπœŽ)𝑖𝑗=π›Ώπ‘–πœŽ,𝑗

Proposition 2.10

βˆ€πœŽ,πœβˆˆπ‘†π‘›, π‘ƒπœŽπ‘ƒπœ=π‘ƒπœŽπœ.

Proof. Fix 1≀𝑖,𝑗≀𝑛. Then

(π‘ƒπœŽπ‘ƒπœ)𝑖𝑗=βˆ‘π‘˜=1𝑛(π‘ƒπœŽ)π‘–π‘˜(π‘ƒπœ)π‘˜π‘—=βˆ‘π‘˜=1π‘›π›Ώπ‘–πœŽ,π‘˜π›Ώπ‘˜πœ,𝑗=𝛿(π‘–πœŽ)𝜏,𝑗=𝛿𝑖(𝜎𝜏),𝑗=𝑃(𝜎𝜏)𝑖𝑗.

Definition 2.11

The sign of a permutation is the function sgn:𝑆𝑛→ℝ, πœŽβ†¦detπ‘ƒπœŽ.

Remark 2.12

sgn(π‘Ž1 … π‘Žπ‘˜)=βˆπ‘–=1π‘˜βˆ’1sgn(π‘Žπ‘–π‘Žπ‘–+1)=(βˆ’1)π‘˜βˆ’1.

Definition 2.13

A permutation 𝜎 is even if sgn(𝜎)=1 and odd if sgn(𝜎)=βˆ’1.

Remark 2.14

sgn(𝜎𝜏)=sgn(𝜎)sgn(𝜏).

Definition 2.15

πœŽβˆˆπ‘†π‘› is a transposition if it is a 2-cycle, i.e. 𝜎=(𝑖 𝑗) for some 1≀𝑖<𝑗≀𝑛.

Remark 2.16

sgn(𝜎)=(βˆ’1)π‘š where π‘š is the number of transpositions of which 𝜎 can be be written as a product.

Definition 2.17

The alternating group is

𝐴𝑛≔{πœŽβˆˆπ‘†π‘›|sgn(𝜎)=1}.

Remark 2.18

𝐴𝑛≀𝑆𝑛 by the Subgroup Test; also |𝐴𝑛|=12|𝑆𝑛|=12𝑛!.

3. Conjugacy

Definition 3.1

Let 𝐺 be a group, 𝑔,β„ŽβˆˆπΊ. 𝑔 is conjugate to β„Ž (in 𝐺) if βˆƒπ‘₯∈𝐺 s.t. π‘₯βˆ’1𝑔π‘₯=β„Ž.

We write this as π‘”βˆΌβ„Ž.

Lemma 3.2

Conjugacy is an equivalence relation.

Proof. Let π‘₯,𝑦,π‘§βˆˆπΊ.

Reflexivity: π‘₯=π‘’βˆ’1π‘₯π‘’βŸΉπ‘₯∼π‘₯

Symmetry:

π‘₯βˆΌπ‘¦βŸΉβˆƒπ‘”βˆˆπΊ. 𝑦=π‘”βˆ’1π‘₯π‘”βŸΉπ‘₯=(π‘”βˆ’1)βˆ’1π‘₯π‘”βˆ’1βŸΉπ‘¦βˆΌπ‘₯.

Transitivity: Suppose π‘₯βˆΌπ‘¦ and π‘¦βˆΌπ‘§, then 𝑧=π‘”βˆ’1𝑦𝑔=π‘”βˆ’1β„Žβˆ’1π‘₯β„Žπ‘”=(β„Žπ‘”)βˆ’1π‘₯(β„Žπ‘”), so π‘₯βˆΌπ‘§.⁠░

Definition 3.3

The conjugacy class of π‘₯∈𝐺 is the equivalence class π‘₯ under ∼. We denote this 𝑔𝐺. (This is not standard notation).

Remark 3.4

𝐺 is the disjoint union of its conjugacy classes,

𝐺=⨃{𝑔𝐺|π‘”βˆˆπΊ}.

Proposition 3.5

If 𝐺 is an abelian group, then π‘¦βˆΌπ‘₯ iff 𝑦=π‘₯.

Proof.

𝑦=π‘”βˆ’1π‘₯π‘”βŸΊπ‘¦=π‘₯π‘”βˆ’1𝑔=π‘₯.

Corollary 3.5.1

The conjugacy classes of an abelian group are singletons.

Lemma 3.6

Any conjugate of a π‘˜-cycle (π‘Ž1 … π‘Žπ‘˜)βˆˆπ‘†π‘› is itself a π‘˜-cycle.

Proof.

πœŒβˆ’1(π‘Ž1 … π‘Žπ‘˜)𝜌=(π‘Ž1𝜌 … π‘Žπ‘˜πœŒ).

⁠░

Theorem 3.7

For 𝜎,πœβˆˆπ‘†π‘›, 𝜎∼𝜏 iff 𝜎 and 𝜏 have the same cycle type (that, is they have the same number of cycles of each length when factored into cycles).

Proof. β€œβŸΉβ€: Suppose 𝜎∼𝜏, then it follows from unqieueness of cycle factorisation, Lemma 3.6 and distributivity that 𝜎 and 𝜏 have the same cycle type.

β€œβŸΈβ€:

Suppose 𝜎 and 𝜏 have the same cycle type, writing 𝜎=𝑐1β€¦π‘π‘˜ and 𝜏=𝑐1β€²β€¦π‘π‘˜β€². By Lemma 3.6, we can choose any πœŒβˆˆπ‘†π‘› such that each 𝑐𝑖=πœŒβˆ’1π‘π‘–β€²πœŒ, then by distributivity 𝜎=πœŒβˆ’1𝜏𝜌 so 𝜎∼𝜏.⁠░

4. Subgroups

Definition 4.1

Let 𝑆 be a subset of a group 𝐺.

The subgroup generated by 𝑆 is:

βŸ¨π‘†βŸ©β‰”β‹‚{𝐻≀𝐺|π»βŠ‡π‘†}.

Remark 4.2

βŸ¨π‘†βŸ© is the smallest subgroup of 𝐺 that contains 𝑆.

Lemma 4.3

Arbitrary intersections of families of subgroups are subgroups.

Definition 4.4

Let π‘†βŠ†πΊ. A word in 𝑆 is some π‘€βˆˆπΊ s.t. βˆƒπ‘ 1,𝑠2,…,π‘ π‘›βˆˆπ‘† s.t. 𝑀=𝑠1𝑠2…𝑠𝑛.

Remark 4.5

If 𝑛=0 in the above definition, then the word 𝑀=𝑒.

Proposition 4.6

Let π‘†βŠ†πΊ, 𝑀(𝑆)≔{words in𝑆}, π‘†βˆ’1≔{π‘ βˆ’1|π‘ βˆˆπ‘†}.

Then βŸ¨π‘†βŸ©=𝑀(𝑆βˆͺπ‘†βˆ’1).

Proof.

β€œβŠ‡β€: It suffices to prove (STP) that βˆ€π»β‰€πΊ s.t. π»βŠ‡π‘†, 𝑀(𝑆βˆͺπ‘†βˆ’1)βŠ†π».

So let π‘€βˆˆπ‘€(𝑆βˆͺπ‘†βˆ’1). Then 𝑀=𝑠1𝑠2…𝑠𝑛 for π‘ π‘–βˆˆπ‘†βˆͺπ‘†βˆ’1. Since π‘†βŠ†π», each π‘ π‘–βˆˆπ». Since 𝐻≀𝐺, π‘€βˆˆπ».

β€œβŠ†β€: STP that 𝑀(𝑆βˆͺπ‘†βˆ’1) is a subgroup of 𝐺 containing 𝑆, so then βŸ¨π‘†βŸ©βŠ†π‘€(𝑆βˆͺπ‘†βˆ’1).

π‘’βˆˆπ‘€(𝑆)βŠ†π‘€(𝑆βˆͺπ‘†βˆ’1).

If π‘₯,π‘¦βˆˆπ‘€(𝑆βˆͺπ‘†βˆ’1), then π‘₯=𝑠1…𝑠𝑛, 𝑦=𝑑1β€¦π‘‘π‘š for 𝑠𝑖,π‘‘π‘–βˆˆπ‘†βˆͺπ‘†βˆ’1. Then π‘₯π‘¦βˆ’1=𝑠1β€¦π‘ π‘›π‘‘π‘šβˆ’1…𝑑1βˆ’1βˆˆπ‘€(𝑆βˆͺπ‘†βˆ’1).

For all π‘ βˆˆπ‘†, 𝑠 is a word in 𝑆, so π‘†βŠ†π‘€(𝑆βˆͺπ‘†βˆ’1).⁠░

Remark 4.7

⟨{𝑔}⟩=βŸ¨π‘”βŸ©={𝑔𝑛|π‘›βˆˆβ„€}.

Corollary 4.7.1

If 𝑔 has finite order, then π‘œ(𝑔)=|βŸ¨π‘”βŸ©|.

Theorem 4.8

Let 𝐺=βŸ¨π‘”βŸ© be a cyclic group.

  1. If |𝐺|=𝑛 is finite, then 𝐺≅𝐢𝑛
  2. Otherwise, 𝐺≅(β„€,+).

Proof.

  1. 𝐺={𝑒,𝑔,𝑔2,…,π‘”π‘›βˆ’1}.

    Say 𝐢𝑛={𝑒,π‘₯,π‘₯2,π‘₯π‘›βˆ’1}.

    Then πœ‘:𝐺→𝐢𝑛 is defined by 𝑔𝑖↦π‘₯𝑖.

  2. Define πœ‘:℀→𝐺;𝑛↦𝑔𝑛.

    Then πœ‘(𝑛+π‘š)=𝑔𝑛+π‘š=π‘”π‘›π‘”π‘š=πœ‘(𝑛)πœ‘(π‘š).

    πœ‘ is clearly surjective as 𝐺={𝑔𝑛|π‘›βˆˆβ„€}. If πœ‘(𝑛)=πœ‘(π‘š), then 𝑔𝑛=π‘”π‘š, so π‘”π‘›βˆ’π‘š=𝑒; since 𝐺 has infinite order, π‘›βˆ’π‘š=0 so 𝑛=π‘š. Hence πœ‘ is also injective, so it is a bijection.

    ⁠░

Theorem 4.9

Any subgroup of a cyclic group is cyclic.

Proof. Say that 𝐺=βŸ¨π‘”βŸ©, and let 𝐻≀𝐺. If 𝐻={𝑒}, done.

Otherwise, βˆƒπ‘šβ‰ 0 such that π‘”π‘šβˆˆπ».

Since 𝐻 is closed under inverses, if π‘š<0 then also π‘”βˆ’π‘šβˆˆπ». So we can assume that π‘š>0.

Let 𝑛≔min{π‘š>0|π‘”π‘šβˆˆπ»}. Then π‘”π‘›βˆˆπ», and if 0<π‘š<𝑛 then π‘”π‘šβˆ‰π».

Claim: 𝐻=βŸ¨π‘”π‘›βŸ©.

β€œβŠ‡β€:

π‘”π‘›βˆˆπ»βŸΉ(𝑔𝑛)π‘Ÿβˆˆπ» βˆ€π‘Ÿβˆˆβ„€βŸΉβŸ¨π‘”π‘›βŸ©={π‘”π‘›π‘Ÿ|π‘Ÿβˆˆβ„€}βŠ†π»

β€œβŠ†β€: Take β„Žβˆˆπ». Since π»βŠ†πΊ=βŸ¨π‘”βŸ©, βˆƒπ‘˜βˆˆβ„€ s.t. β„Ž=π‘”π‘˜.

Divide π‘˜ by 𝑛 with remainder to get π‘˜=π‘žπ‘›+π‘Ÿ where π‘ž,π‘Ÿβˆˆβ„€ and 0β‰€π‘Ÿ<π‘›βˆ’1.

Then π‘Ÿ=π‘˜βˆ’π‘žπ‘›, so π‘”π‘Ÿ=π‘”π‘˜βˆ’π‘žπ‘›=π‘”π‘˜(𝑔𝑛)βˆ’π‘ž.

Since π‘”π‘˜,π‘”π‘›βˆˆπ», by the subgroup test π‘”π‘Ÿβˆˆπ», so by our choice of 𝑛 and π‘Ÿ<𝑛, it must be that π‘Ÿ=0. Hence π‘˜=π‘žπ‘›, so β„Ž=π‘”π‘˜=(𝑔𝑛)π‘žβˆˆβŸ¨π‘”π‘›βŸ©.⁠░

Corollary 4.9.1

The subgroups of β„€ are π‘šβ„€ for each π‘šβˆˆβ„€.

Proposition 4.10

Let π‘š,π‘›βˆˆβ„€. Then βˆƒπ‘’,π‘£βˆˆβ„€ s.t. π‘’π‘š+π‘£π‘›β‰•β„Ž generates βŸ¨π‘š,π‘›βŸ© (=π‘šβ„€+𝑛℀).

If β„Ž>0, then β„Ž=gcd(π‘š,𝑛).

Proof.

βŸ¨π‘š,π‘›βŸ© is a subgroup of (β„€,+), so, by Theorem 4.9, βŸ¨π‘š,π‘›βŸ©=βŸ¨β„ŽβŸ© for some β„Žβˆˆβ„€.

Then π‘šβˆˆβŸ¨β„ŽβŸ©, so π‘š=π‘Ÿβ„Ž for some π‘Ÿβˆˆβ„€. Similarly, 𝑛=π‘ β„Ž for some π‘ βˆˆβ„€.

Thus β„Žβˆ£π‘š and β„Žβˆ£π‘›.

To show that β„Ž=gcd(π‘š,𝑛), we will show that if π‘˜βˆ£π‘š and π‘˜βˆ£π‘›, then π‘˜βˆ£β„Ž

But βˆƒπ‘’,π‘£βˆˆβ„€ s.t. β„Ž=π‘’π‘š+𝑣𝑛 because β„ŽβˆˆβŸ¨π‘š,π‘›βŸ©, and now π‘˜βˆ£π‘š and π‘˜βˆ£π‘› βŸΉπ‘˜βˆ£π‘’π‘š+𝑣𝑛=β„Ž.⁠░

Lemma 4.11 (BΓ©zout's Lemma)

βˆ€π‘š,π‘›βˆˆβ„€, βˆƒπ‘’,π‘£βˆˆβ„€ s.t. gcd(π‘š,𝑛)=π‘’π‘š+𝑣𝑛.

Proof. Equivalent to Proposition 4.10.⁠░

Theorem 4.12 (Chinese Remainder Theorem)

Suppose π‘š,π‘›βˆˆβ„€+ are coprime. Then πΆπ‘šΓ—πΆπ‘› is cyclic, and is isomorphic to πΆπ‘šΓ—π‘›.

Proof. Say πΆπ‘š=βŸ¨π‘”βŸ© and 𝐢𝑛=βŸ¨β„ŽβŸ©.

STP: πΆπ‘šΓ—πΆπ‘›=⟨(𝑔,β„Ž)⟩.

Let π‘˜β‰”π‘œ((𝑔,β„Ž)). Then (π‘”π‘˜,β„Žπ‘˜)=(𝑔,β„Ž)π‘˜=(𝑒,𝑒). Therefore π‘šβˆ£π‘˜ and π‘›βˆ£π‘˜, hence π‘šπ‘›βˆ£π‘˜,= because π‘š,𝑛 coprime, by BΓ©zout’s Lemma.

Also, (𝑔,β„Ž)π‘šπ‘›=(π‘”π‘šπ‘›,β„Žπ‘šπ‘›)=(𝑒𝑛,π‘’π‘š)=(𝑒,𝑒). So π‘˜βˆ£π‘šπ‘›.

Therefore π‘˜=π‘šπ‘›.⁠░

5. Modular arithmetic

Definition 5.1

For a set 𝑋 and relation π‘…βˆˆπ‘‹2, we write

𝑋/𝑅≔{[π‘₯]𝑅|π‘₯βˆˆπ‘‹}
to be the set of equivalence classes in 𝑋 under 𝑅.

Lemma 5.2

≑𝑛 is an equivalence relation (where ≑𝑛 is congruence modulo 𝑛).

Proof. See e.g. Discrete Maths (MT2025).⁠░

Definition 5.3

We write ℀𝑛 to mean β„€/≑𝑛.

Definition 5.4

We write π‘₯̅≔[π‘₯]≑𝑛.

Example 5.5

β„€2={{…,βˆ’4,βˆ’2,0,2,4,…},{…,βˆ’3,βˆ’1,1,3βƒ—..}}={0Μ…,1Μ…}.

Lemma 5.6

Let π‘₯,π‘₯β€²,𝑦,π‘¦β€²βˆˆβ„€ such that π‘₯≑π‘₯β€²mod𝑛 and 𝑦≑𝑦′mod𝑛.

  1. π‘₯+𝑦≑π‘₯β€²+𝑦′mod𝑛
  2. π‘₯𝑦≑π‘₯′𝑦′mod𝑛

Proof.

  1. π‘›βˆ£π‘₯βˆ’π‘₯β€² and π‘›βˆ£π‘¦βˆ’π‘¦β€², so

    π‘›βˆ£(π‘₯βˆ’π‘₯β€²)+(π‘¦βˆ’π‘¦β€²)=(π‘₯+𝑦)βˆ’(π‘₯β€²+𝑦′)
    hence π‘₯+𝑦≑π‘₯β€²+𝑦′mod𝑛.
  2. π‘₯β€²π‘¦β€²βˆ’π‘₯𝑒=π‘₯β€²(π‘¦β€²βˆ’π‘¦)+(π‘₯β€²βˆ’π‘₯)𝑦, which is an integer multiple of 𝑛 if π‘›βˆ£π‘₯βˆ’π‘₯β€² and π‘›βˆ£π‘¦βˆ’π‘¦β€². Hence π‘₯′𝑦′=π‘₯𝑦mod𝑛.

    ⁠░

Proposition 5.7

There following are well-defined binary operations on ℀𝑛:

Proof. It suffices to prove (STP) that these definitions do not depend on the choices made for the two 𝛼,π›½βˆˆβ„€π‘›:

  • pick π‘Žβˆˆπ›Ό,π‘βˆˆπ›½
  • set 𝛼+𝛽=π‘Ž+𝑏̅.

If π‘Žβ€²βˆˆπ›Ό and π‘β€²βˆˆπ›½, then π‘Žβ€²+π‘β€²β‰‘π‘Ž+𝑏mod𝑛 by the Lemma (TODO ref), so π‘Žβ€²+𝑏′̅=π‘Ž+𝑏̅. Similarly π‘Žβ€²π‘β€²Μ…=π‘Žπ‘Μ….⁠░

Proposition 5.8

  1. (℀𝑛,+) is an abelian group, and moreover is cyclic.
  2. (℀𝑛,Γ—) satisifies associativity and commutativity, and admits an identity element
  3. π‘₯(𝑦+𝑧)=π‘₯𝑦+π‘₯𝑧 for all π‘₯,𝑦,π‘§βˆˆβ„€π‘›.

Proof.

  1. π‘₯Μ…+(𝑦̅+𝑧̅)=π‘₯Μ…+(𝑦+𝑧̅)=𝑧+(𝑦+𝑧)Μ…=(π‘₯+𝑦)+𝑧̅=π‘₯+𝑦̅+𝑧̅=(π‘₯Μ…+𝑦̅)+𝑧̅.

    0Μ…+π‘₯Μ…=0+π‘₯Μ…=π‘₯Μ…=π‘₯+0Μ…=π‘₯Μ…+0Μ…, so (℀𝑛,+) admits an identity.

    The inverse of π‘₯Μ… is βˆ’π‘₯Μ….

    This group is cyclic because π‘˜Μ…=1Μ…+…+1Μ…βŸπ‘˜times for all π‘˜βˆˆβ„€>0.

    Hence (℀𝑛,+) is generated by 1Μ…. 1Μ… has order π‘œ(1Μ…)=𝑛.

  2. Inherited from the integers in a similar manner to the above, with 1Μ… as the identity.
  3. Also inherited from the integers.

    ⁠░

Remark 5.9

℀𝑛 is a unital commutative ring.

Lemma 5.10

The function from {0,1,…,π‘›βˆ’1}→℀𝑛 defined by π‘Ÿβ†¦π‘ŸΜ… is a bijection.

Proof.

Suppose 0β‰€π‘Ÿ,π‘Ÿβ‰€π‘›βˆ’1 and π‘ŸΜ…=𝑠̅. Then π‘Ÿβ‰‘π‘ mod𝑛, so π‘›βˆ£π‘Ÿβˆ’π‘ , i.e. π‘Ÿβˆ’π‘ =π‘›π‘˜ for some π‘˜βˆˆβ„€. WLOG suppose 𝑠β‰₯π‘Ÿ, then if π‘˜β‰ 0 then |π‘ βˆ’π‘Ÿ|β‰₯𝑛. But |π‘ βˆ’π‘Ÿ|β‰€π‘›βˆ’1, so π‘˜=0 and π‘Ÿ=𝑠. Hence the function is injective.

Now take π›Όβˆˆβ„€π‘›, and say 𝛼=π‘ŽΜ… for some π‘Žβˆˆβ„€. Divide π‘Ž by 𝑛 with remainder to get π‘Ž=π‘žπ‘›+π‘Ÿ for some π‘ž,π‘Ÿβˆˆβ„€ with 0β‰€π‘Ÿβ‰€π‘›βˆ’1.

Then π‘›βˆ£π‘žπ‘›=π‘Žβˆ’π‘Ÿ, so π‘Žβ‰‘π‘Ÿmod𝑛. Note that 𝛼=π‘ŽΜ…=π‘ŸΜ…, so the function is surjective.

Hence the function is a bijection.⁠░

Proposition 5.11

  1. Let π‘₯Μ… in ℀𝑛. Then π‘₯Μ… admits a multiplicative inverse iff gcd(π‘₯,𝑛)=1.
  2. Define ℀𝑛×≔{π‘₯Μ…βˆˆβ„€π‘›|βˆƒπ‘¦βˆˆβ„€π‘›s.t.π‘₯̅𝑦̅=1Μ…}. Then ℀𝑛× is an abelian group.
  3. For prime 𝑝, ℀𝑝×=℀𝑝\{0Μ…}, so ℀𝑝 is a field.

Proof.

  1. exercise
  2. If π‘ŽΜ…,π‘Μ…βˆˆβ„€π‘›Γ—, then π‘ŽΜ…π‘Μ…=1Μ…=𝑏̅𝑑̅ for some 𝑐̅,π‘‘Μ…βˆˆβ„€π‘›. Hence (π‘ŽΜ…π‘Μ…)(𝑐̅𝑑̅)=(π‘ŽΜ…π‘Μ…)(𝑏̅𝑑̅)=1Μ…β‹…1Μ…=1Μ….
  3. Let 0Μ…β‰ π‘₯Μ…βˆˆβ„€π‘. Then π‘βˆ€π‘₯. Since 𝑝 is prime, gcd(𝑝,π‘₯)=1. Hence by (a), π‘₯Μ…βˆˆβ„€π‘›Γ—.⁠░

6. Cosets & Lagrange’s Theorem

Definition 6.1

Let 𝐻≀𝐺 not necessarily finite.

A left coset of 𝐻 is a subset of 𝐺 of the form 𝑔𝐻 for some π‘”βˆˆπΊ, where

𝑔𝐻≔{π‘”β„Ž|β„Žβˆˆπ»}.

A right coset of 𝐻 is a subset of 𝐺 of the form 𝐻𝑔 for some π‘”βˆˆπΊ, where

𝐻𝑔≔{β„Žπ‘”|β„Žβˆˆπ»}.

Definition 6.2

We write 𝐺/𝐻 (β€œπΊ mod 𝐻”) to mean the set of left cosets of 𝐻 in 𝐺; we write 𝐻\𝐺 to mean the set of right cosets of 𝐻 in 𝐺

Definition 6.3

The index of 𝐻 in 𝐺 is

[𝐺:𝐻]≔|𝐺/𝐻|.

Example 6.4

𝐺=(β„€,+); fix π‘›βˆˆβ„€>0.

Let 𝐻=βŸ¨π‘›βŸ©={π‘›π‘Ž|π‘Žβˆˆβ„€=𝑛℀}.

The left cosets of 𝐻 are

{0+𝑛℀,1+𝑛℀,2+𝑛℀,…,π‘›βˆ’1+𝑛℀}.

Hence β„€/𝑛℀=℀𝑛.

Remark 6.5

If 𝐺 is abelian, then its left cosets are the same as its right cosets.

Remark 6.6

In general |𝐺/𝐻|=|𝐻\𝐺|.

Lemma 6.7 (Coset Equality Lemma)

Let 𝐻≀𝐺 and 𝑔1,𝑔2∈𝐺.

Then 𝑔1𝐻=𝑔2𝐻 iff 𝑔2βˆ’1𝑔1∈𝐻.

Proof.

β€œβŸΉβ€:

π‘’βˆˆπ»βŸΉπ‘”1π‘’βˆˆπ‘”1π»βŸΉπ‘”1βˆˆπ‘”2π»βŸΉβˆƒβ„Žβˆˆπ»s.t.𝑔1=𝑔2β„Ž.

Hence 𝑔2βˆ’1𝑔1=β„Žβˆˆπ».

β€œβŸΈβ€ :

Suppose 𝑔2βˆ’1𝑔1∈𝐻, then 𝑔1𝐻=𝑔2𝑔2βˆ’1𝐻=𝑔2𝐻

because π‘₯∈𝐻, then π‘₯𝐻={π‘₯β„Ž|β„Žβˆˆπ»}=𝐻.⁠░

Remark 6.8

Left cosets have the property that if π‘₯∈𝐺 and π‘”π»βˆˆπΊ/𝐻, then π‘₯π‘”π»βˆˆπΊ/𝐻.

Theorem 6.9 (Lagrange's Theorem)

For 𝐺 a finite group, and 𝐻≀𝐺 a subgroup of 𝐺, then |𝐻|∣|𝐺|.

Proof. We will show that 𝐺/𝐻 forms a partition of 𝐺, and that βˆ€π‘”1𝐻,𝑔2𝐻∈𝐺/𝐻, |𝑔1𝐻|=|𝑔2𝐻|.

π‘”π»βˆ‹π‘”βŸΉπ‘”π»β‰ βˆ…. Moreover, if π‘”βˆˆπΊ, then

𝑔=π‘”π‘’βˆˆπ‘”π»βŸΉπΊ=β‹ƒπΆβˆˆπΊ/𝐻𝐢.

If 𝑔1𝐻 and 𝑔2𝐻 are not disjoint, then βˆƒπ‘₯βˆˆπ‘”1βˆ©π‘”2𝐻. We write π‘₯=𝑔1β„Ž1=𝑔2β„Ž2 for some β„Ž1,β„Ž2∈𝐻.

Then 𝑔2βˆ’1𝑔1=β„Ž2β„Ž1βˆ’1∈𝐻 as 𝐻≀𝐺.

Hence by the Coset Equality Lemma, 𝑔1𝐻=𝑔2𝐻. Therefore distinct left cosets are disjoint.

Say that 𝐺/𝐻={𝑔1𝐻,𝑔2𝐻,…,𝑔𝑛𝐻} where 𝑛=|𝐺/𝐻|.

The function πœ‘:𝐻→𝑔𝑖𝐻;β„Žβ†¦π‘”π‘–β„Ž is a bijection, with inverse π‘¦β†¦π‘”π‘–βˆ’1𝑦. Hence |𝑔𝑖𝐻|=|𝐻|=|𝑔𝑖𝐻|.

Then |𝐺|=βˆ‘π‘–=1𝑛|𝑔𝑖𝐻|=𝑛|𝐻|=[𝐺:𝐻]|𝐻|.

⁠░

Corollary 6.9.1

For 𝐺 a finite group and π‘”βˆˆπΊ, then π‘œ(𝑔)∣|𝐺|.

Proof. Recall that π‘œ(𝑔)=min{π‘›βˆˆβ„€>0|𝑔𝑛=𝑒}, but also π‘œ(𝑔)=|βŸ¨π‘”βŸ©|. By Lagrange’s Theorem, |βŸ¨π‘”βŸ©|∣|𝐺|, so π‘œ(𝑔)∣|𝐺|.⁠░

Corollary 6.9.2

Suppose 𝑝 is prime and 𝐺 is a group of order 𝑝. Then 𝐺≅𝐢𝑝, and all non-identity elements of 𝐺 have order 𝑝.

Proof. By Corollary 6.9.1, for π‘”βˆˆπΊ, π‘œ(𝑔)=1 or π‘œ(𝑔)=|𝐺|. π‘œ(𝑔)=1 iff 𝑔=𝑒, so for all π‘”βˆˆπΊ\{𝑒}, π‘œ(𝑔)=𝑝.

Now pick some π‘”βˆˆπΊ s.t. 𝑔≠𝑒, then βŸ¨π‘”βŸ©β‰…πΆπ‘œ(𝑔)=𝐢𝑝.⁠░

Corollary 6.9.3

If 𝐺 is a finite group, π‘”βˆˆπΊ, then 𝑔|𝐺|=𝑒.

Proof. From Corollary 6.9.1, |𝐺|=π‘Ÿβ‹…π‘œ(𝑔) for some π‘Ÿβˆˆβ„•. Hence

𝑔𝐺=π‘”π‘Ÿβ‹…π‘œ(𝑔)=(π‘”π‘œ(𝑔))π‘Ÿ=π‘’π‘Ÿ=𝑒.

Theorem 6.10 (Euler's Totient Theorem)

Let 𝑛β‰₯2, and let π‘Žβˆˆβ„€ be coprime to 𝑛. Then

π‘Žπœ™(𝑛)≑1mod𝑛,

where

πœ™(𝑛)≔|℀𝑛×|=|{0<π‘˜<𝑛|gcd(π‘˜,𝑛)=1}|.

Proof. Apply Corollary 6.9.3 to ℀𝑛×:

π‘Ž+π‘›β„€βˆˆβ„€π‘›Γ—,|℀𝑛|=πœ™(𝑛)⟹(π‘Ž+𝑛℀)πœ™(𝑛)=1+π‘›β„€βŸΉπ‘Žπœ™(𝑛)=1mod𝑛.

⁠░

Theorem 6.11 (Fermat's Little Theorem)

For prime 𝑝 and π‘Žβˆˆβ„€ coprime to 𝑝,

π‘Žπ‘βˆ’1≑1mod𝑝.

Proof. Apply Euler’s Totient Theorem with 𝑛=𝑝:

πœ™(𝑝)=|℀𝑝×|=|{1Μ…,2Μ…,…,π‘βˆ’1}|=π‘βˆ’1.

Theorem 6.12 (Wilson's Theorem)

For prime 𝑝, (π‘βˆ’1)!β‰‘βˆ’1mod𝑝.

Proof. Let 𝐺 be a finite group and define ∼, a relation on 𝐺 by

π‘₯βˆΌπ‘¦βŸΊπ‘¦=π‘₯βˆ¨π‘¦=π‘₯βˆ’1.

∼ is an equivalence relation; reflexivity is obvious, symmetry is by taking inverse of both sides, and transitivity is also simple.

Then

𝐺={π‘”βˆˆπΊ|π‘₯=π‘₯βˆ’1}βˆͺβ‹ƒπ‘¦βˆˆπΊ;π‘¦β‰ π‘¦βˆ’1{𝑦,π‘¦βˆ’1}.

Then if 𝐺 is finite, and π‘šβ‰”{π‘₯∈𝐺|π‘₯=π‘₯βˆ’1}, then |𝐺|β‰‘π‘šmod2.

We then apply these ideas to ℀𝑝×={π‘₯βˆˆβ„€π‘Γ—|π‘₯=π‘₯βˆ’1}βˆͺβ‹ƒπ‘¦βˆˆβ„€π‘Γ—{𝑦,π‘¦βˆ’1}.

Consider ∏π‘₯βˆˆβ„€π‘Γ—π‘₯. This is (π‘βˆ’1)!Μ…. Since the 𝑦s cancel with π‘¦βˆ’1,

(π‘βˆ’1)!Μ…=∏π‘₯Μ…βˆˆπ•π‘Γ—;π‘₯Μ…=π‘₯Μ…βˆ’1π‘₯Μ….

But π‘₯Μ…=π‘₯Μ…βˆ’1⟺π‘₯Μ…2=1Μ…βŸΊ(π‘₯Μ…βˆ’1Μ…)(π‘₯Μ…+1Μ…)=0.

Since ℀𝑝 is a field, π‘₯Μ…2=1Μ…βŸΉπ‘₯Μ…=1Μ… or π‘₯Μ…=βˆ’1Μ….

Hence (π‘βˆ’1)!Μ…=1Μ…β‹…βˆ’1Μ…=βˆ’1Μ….

⁠░

Corollary 6.12.1

Given a group 𝐺 with |𝐺| even, then βˆƒπ‘”βˆˆπΊ with π‘œ(𝑔)=2.

Proof. Consider {𝑔,π‘”βˆ’1} for each π‘”βˆˆπΊ. |{𝑔,π‘”βˆ’1}|=2 iff π‘”β‰ π‘”βˆ’1 iff 𝑔2≠𝑒.

That is, |{𝑔,π‘”βˆ’1}|=1 iff 𝑔2=𝑒 iff 𝑔=𝑒 or π‘œ(𝑔)=2.

Now 𝐺=β‹ƒπ‘”βˆˆπΊ{𝑔,π‘”βˆ’1} so |𝐺|≑1+|{π‘”βˆˆπΊ:π‘œ(𝑔)=2}|mod2. Since |𝐺| is even by assumption, there must exist at least one (in fact, an odd number of) 𝑔 with π‘œ(𝑔)=2.⁠░

Theorem 6.13

Let 𝑝β‰₯3 be prime, and let 𝐺 be a group with |𝐺|=2𝑝. Then either 𝐺≅𝐢2𝑝 or 𝐺≅𝐷2𝑝.

Proof. By Lagrange’s Theorem, βˆ€π‘”βˆˆπΊ, π‘œ(𝑔)=|βŸ¨π‘”βŸ©|∣|𝐺|.

Since 𝑝β‰₯3 is prime, this forces π‘œ(𝑔)∈{1,2,𝑝,2𝑝}.

If βˆƒπ‘”βˆˆπΊ with π‘œ(𝑔)=2𝑝, then 𝐺=βŸ¨π‘”βŸ©β‰…πΆ2𝑝.

Suppose then βˆ„π‘”βˆˆπΊ with π‘œ(𝑔)=2𝑝.

If βˆ€π‘”βˆˆπΊ,π‘œ(𝑔)=1 or 2, then 𝑔2=𝑒 for all π‘”βˆˆπΊ. In this case, 𝐺≅𝐢2𝑛 for some π‘›βˆˆβ„•. Then 2𝑝=|𝐺|=2𝑛 which is a contradiction.

Therefore, βˆƒπ‘₯∈𝐺 with π‘œ(π‘₯)=𝑝.

Moreover, by Corollary TODO, βˆƒπ‘¦βˆˆπΊ with π‘œ(𝑦)=2.

Now we have ⟨π‘₯⟩,⟨π‘₯βŸ©π‘¦βŠ†πΊ.

Note that these are disjoint: suppose that π‘₯𝑖=π‘₯𝑗𝑦, then 𝑦=π‘₯π‘–βˆ’π‘—βˆˆβŸ¨π‘₯⟩ so 2∣|⟨π‘₯⟩|=𝑝, which is a contradiction.

Hence 𝐺=⟨π‘₯⟩TODO⟨π‘₯βŸ©π‘¦={π‘₯𝑖𝑦𝑗|0β‰€π‘–β‰€π‘βˆ’1,π‘œβ‰€π‘—β‰€1}.

Now note that

𝑦π‘₯π‘¦βˆ’1∈(π‘¦βŸ¨π‘₯⟩)π‘¦βˆ’1=(πΊβˆ’βŸ¨π‘₯⟩)π‘¦βˆ’1=⟨π‘₯βŸ©π‘¦π‘¦βˆ’1=⟨π‘₯⟩.

Then 𝑦π‘₯π‘¦βˆ’1=π‘₯𝑖 for some 0β‰€π‘–β‰€π‘βˆ’1.

Now π‘₯=𝑦2π‘₯π‘¦βˆ’2=𝑦(𝑦π‘₯π‘¦βˆ’1)π‘¦βˆ’1=𝑦(π‘₯𝑖)π‘¦βˆ’1=(𝑦π‘₯π‘¦βˆ’1)𝑖=(π‘₯𝑖)𝑖=π‘₯𝑖2.

Hence 𝑖2≑1mod𝑝 as π‘œ(π‘₯)=𝑝.

Since 𝑝 is prime, this forces 𝑖=1 or π‘–β‰‘βˆ’1mod𝑝, so π‘–βˆ’1 or 𝑖=π‘βˆ’1.

If 𝑖=1, then 𝑦π‘₯π‘¦βˆ’1=π‘₯βŸΉπ‘¦π‘₯=π‘₯𝑦, so π‘œ(π‘₯𝑦)∣2𝑝. (π‘₯𝑦)2=π‘₯2𝑦2=π‘₯2β‰ π‘’βŸΉπ‘œ(π‘₯𝑦)β‰ 2. Similarly (π‘₯𝑦)𝑝=π‘₯𝑝𝑦𝑝=π‘¦β‰ π‘’β‡’π‘œ(π‘₯𝑦)≠𝑝. So 𝑖=π‘βˆ’1.

Therefore

𝐺={π‘₯𝑖𝑦𝑗|0β‰€π‘–β‰€π‘βˆ’1,0≀𝑗≀1}

and 𝑦π‘₯π‘¦βˆ’1=π‘₯𝑝1=π‘₯βˆ’1.

Hence 𝐺≅𝐷2𝑝.⁠░

7. Homomorphisms

Definition 7.1

Let 𝐺,𝐻 be groups. A function πœ‘:𝐺→𝐻 is a homomorphism if πœ‘(𝑔1βˆ—πΊπ‘”2)=πœ‘(𝑔1)Γ—π»πœ‘(𝑔2) for all 𝑔2,𝑔2∈𝐺.

Remark 7.2

An isomorphism is a bijective homomorphism.

Definition 7.3

A monomorphism is an injective homomorphism. We write 𝐺β†ͺ︎𝐻.

Definition 7.4

A epimorphism is a surjection homomorphism. We write 𝐺↠𝐻.

Example 7.5

Definition 7.6

The trivial homomorphism is the homomorphism πœ‘:𝐺→𝐺;𝑔↦𝑒𝐻.

Example 7.7

Linear maps between vector spaces are group homomorphisms (under addition).

Theorem 7.8

Hom(β„€,𝐺)≅𝐺.

Proof. Let πœ‘π‘”:℀→𝐺 be defined by 𝑛↦𝑔𝑛 for each π‘”βˆˆπΊ.

Then STP that the function 𝐺→Hom(β„€,𝐺);π‘”β†¦πœ‘π‘” is bijective.

If 𝑔,β„ŽβˆˆπΊ and πœ‘π‘”=πœ‘β„Ž, then 𝑔=πœ‘π‘”(1)=πœ‘β„Ž(1)=β„Ž, so π‘”β†¦πœ‘π‘” is injective.

Let πœ“:℀→𝐺 be a homomorphism, then if π‘”β‰”πœ“(1)∈𝐺, then πœ‘π‘”(𝑛)=𝑔𝑛=πœ“(1)𝑛=πœ“(𝑛) for all π‘›βˆˆβ„€. Then πœ“=πœ‘π‘”, so π‘”β†¦πœ‘π‘” is surjective.⁠░

Definition 7.9

An automorphism of 𝐺 is an isomorphism πœ“:𝐺→𝐺.

Remark 7.10

Automorphisms form a group Aut(𝐺) under composition.

Definition 7.11

Fix π‘”βˆˆπΊ. Define πœƒπ‘”:𝐺→𝐺 by π‘₯β†¦π‘”βˆ’1π‘₯𝑔; πœƒπ‘” is the conjugation by 𝑔.

Proposition 7.12

πœƒπ‘” is an automorphism.

Proof. πœƒπ‘”(π‘₯𝑦)=π‘”βˆ’1π‘₯𝑦𝑔=π‘”βˆ’1π‘₯π‘”π‘”βˆ’1𝑦𝑔=πœƒπ‘”(π‘₯) πœƒπ‘”(𝑦). Hence each πœƒπ‘” is a homomorphism.

Moreover, πœƒπ‘” is invertible : πœƒπ‘”βˆ’1=πœƒπ‘”βˆ’1.⁠░