13 Quotient Groups (cont.)

left/right cosets

For any NG (see Subgroup) and any gG let:
gN={gn|nN};Ng={ng|nN}
be called (respectively) a left coset and a right coset of N in G. Any element of a coset is called a representative for the coset.

We denote the set of left cosets by GH, as we see in Quotient Group.

Warning

In general gHHg for all gG. Essentially, once the left/right cosets agree with each other, then we have a Normal Subgroup specifically for that H.

Examples

Say we have G=S3 (see Symmetric Group & Permutations) and H=(1 2){1,(1 2)}. Then the left-cosets of H are;

gG gH
1 {1 1,1(1 2)}={1,(1 2)}=H
(1 2) {(1 2)1,(1 2)(1 2)}={(1 2),1}
(1 3) {(1 3),(1 2 3)}
(2 3) {(2 3),(1 3 2)}
(1 2 3) {(1 2 3),(1 3)}
(1 3 2) {(1 3 2),(2 3)}
Notice that these are partitioned. Namely name:
  • 1H,(1 2)H=C1
  • (1 3)H,(1 2 3)H=C2
  • (2 3)H,(1 3 2)H=C3
    creating a partition of 3 elements.
Note

For one right coset see the order matters:
H(1 3)={1(1 3),(1 2)(1 3)}={(1 3),(1 3 2)}(1 3)H

Definition

Let φ:GH be a Homomorphism (or Group Morphism) with Kernel K. The quotient group, GK (read G modulo K or simply G mod K) is the group whose elements are the Fiber (Group Theory) of φ with group operation defined below:

Namely if X is the Fiber (Group Theory) above a and Y is the fiber above b, then the product of X with Y is defined to be the fiber above the product ab.

The notation implies the idea that the Kernel K is a single element in the group GK, and we shall see in Properties of Quotient Groups that, as in the case of ZnZ as seen here that the other elements of GK are just the "translates" of the kernel K. We can think of GK as "dividing out" by K, hence the name (and why we refer to it by mod K, as the K is dividing it into distinct parts).

Definition using Cosets

Definition

We can define the set GH where HG as the set of left cosets of H in G. We have a canonical map:
π:GGH;ggH

Example

Consider G=S3={1,(1 2),(1 3),(2 3),(1 2 3),(1 3 2)}, and let H=(1 3)={1,(1 3)}.

We can do the same for the right cosets:

  • H1 is the first one
  • H(1 2)={1(1 2),(1 3)(1 2)}={(1 2),(1 2 3)}. But notice then H(1 2)(1 2)H so then this is not a Normal Subgroup.
  • By the remaining number of elements then we have H(2 3)={(2 3),(1 3 2)}.

Consider instead KG with K=(1 2 3)={1,(1 2 3),(1 3 2)}. Then:

  • 1K={1,(1 2 3),(1 3 2)}=K1 as expected.
  • We haven't seen (1 2)K={(1 2),(1 3),(2 3)}=K(1 2) since the size of our coset is the same for each, so we have the other half of the elements.
  • Thus K is a Normal Subgroup.
Theorem

Suppose G is a finite Group and HG (see Subgroup). Then |H| divides |G| (see Order (Groups)).

Proof

The left cosets of H partition G, so they all have size |H| and are disjoint. Thus:
$
|G| = (# \text{ cosets of H in G})|H|
$
Thus |H| divides |G|.

An alternative method of this is saying:

Theorem

The number of left cosets of H in G equals |G||H|

Proof

See Index (Groups).

index

For HG (see Subgroup) the index of H in G is the number of cosets of H in G.

This is often denoted [GH] or |G:H|. Notice that this just |GH|=[GH] by definition. If G is finite (|G|<) then [GH]=|G||H| via Lagrange's Divisibility Theorem of Order of Subgroups.


dg-publish: true

Theorem

If HG (see Subgroup) and [GH]=2 (see Index (Groups)) then HG.

Proof

Use 11 Basic Axioms and Examples (Groups)#25 and apply it to H, using Equivalences of Group Actions.

So all of this means that for any gG then |g| must divide |G|:

Proof

ordG(g)=|g|, but using Lagrange's Divisibility Theorem of Order of Subgroups then it must divide |G|.

But when does the set GH have a group structure? Namely one such that π:GGH to be a Homomorphism (or Group Morphism). The answer is that:

Claim

If π:GGH is a Homomorphism (or Group Morphism), then:

π(g1g2)=π(g1)π(g2)

for all g1,g2G. But all of this is only iff:

(g1g2)H=(g1H)(g2H)

So we must define the Coset Product by:

(g1H)(g2H):=(g1g2)H

We have to consider that this well defined.

coset product

We define the coset product for HG as:
(g1H)(g2H):=(g1g2)H
for all g1,g2G. Then this operation is well defined iff HG.

We want to prove that this is well defined; namely if g1H=g1~H and g2H=g2~H do we have:
(g1g2)H=(g1~g2~)H

Example

To highlight that it is not always well-defined, consider the example H={1,(1 3)} for G=S3. We had:

Notice (1 2)H(2 3)H=(1 2)(2 3)H=(1 2 3)H. Namely if the green coset is C2 and the red one is C3 then does it always equal C3 (as it did in our calculation).

What we see is that (1 3 2)H(2 3)H=(1 3 2)(2 3)H=(1 3)H is in C1instead, which is different. Thus we needed the other condition of being a Normal Subgroup to be actually well defined.

The Proof

Let's do the proof using the importance of Normal Subgroup:

Proof

(): We want to show that HG. Take any hH and gG. We got to show that ghg1H.

First notice that ghgH. That means that since ggH then (gh)H=gH are the same cosets. But then because g1H=g2Hg21g1Hg11g2H, then we can compare (gh)H and gH. Namely:
(gh)Hg1H=(ghg1)H
and:
gHg1H=(gg1)H=H
Thus (ghg1)H=H so then ghg1H, as we wanted.

(): TBD

As a consequence, for any KG, the set GK is a Group with the operation defined in the Coset Product. The projection map π:GGK:ggK is a Homomorphism (or Group Morphism). The identity of GK is the coset 1K since:

(gK)(1K)=(g1)K=gK=(1g)K=(1K)(gK)

Further notice that (gK)1=g1K.

Also:

Ker(π)={gG|gK=1K}={gG|gK}=K

so as a consequence, every Normal Subgroup is a Kernel of a Homomorphism (or Group Morphism).

That means that whenever you have some KG then there is some corresponding π map where Ker(π)=K.

So for any φ:GH Homomorphism (or Group Morphism) then Ker(φ)G.

But how does GKer(φ) fit into the picture?