16 - The Four Horsemen Theorems

Last time we talked about thee first isomorphism theorem:

Theorem

Suppose φ:G→H is a Homomorphism (or Group Morphism). Then:

  1. Ker(φ)⊴G
  2. im(φ)≤H
  3. The map GK→im(φ) where gK↦φ(g) is an Isomorphism, where K=Ker(φ). Thus GKer(φ)≅im(φ)

An Example

This will be the most common way to understand a Quotient Group. To highlight this, consider the example G=D8. We claim that D8K where in this case K=⟨r2⟩ then has:
D8K≅Z2×Z2
The idea is to find some surjective morphism φ:D8→Z2×Z2 in which Ker(φ)=K.

Know that D8 is generated by r,s, so φ is determined by φ(r) and φ(s). We also know the key relations are that r4=1=s2,rs=sr3. Thus if we want such a morphism φ then:
φ(r4)=4φ(r)=0=(0―,0―)

When we look at the codomain of φ its in terms of adding $\mathbb

Let's use a different example to look at the other 3 theorems we'll cover.

Foundational Example

Consider G=Z24Z. Recall that because it's Cyclic Group via Cyclic Groups of the Same Order are Isomorphic, then the subgroups of Z24Z is 1:1 with the divisors of 24. Let's look at the lattice.

Notice how the lattice makes a literal diamond between subgroups. The next theorem covers when these cases happen.

Theorem

Suppose A,B≤G and A≤NG(B) (see Normalizer, so every element of A normalizes B, so aBa−1=B for all a∈A). Then:

  1. AB=⟨A∪B⟩ is the smallest subgroup containing A and B. It is the "tip" of the diamond in the lattice.
  2. A∩B⊴A and B⊴AB
  3. ABB≅AA∩B

This may be the most useless theorem ever, but the last finding is really important. Namely, for when you want to "cancel" B from AB then you have to consider the factor of A∩B (ie: you get some elements of B that "bleed over")

Note

If B⊴G (see Normal Subgroup) then NG(B)=G so A≤NG(B).

Example

Here A=⟨6⟩ and B=⟨4⟩. Then AB=⟨2⟩ and A∩B=⟨12⟩.

If you take AB and divide it into the cosets, you can make pairings between cosets of AB versus that of just A w.r.t A∩B.

Proof

Proof

(Sketch of (3)): Define a map φ:A→ABB such that a↦aB (note that aB∈ABB still, our notation still makes sense). It's a Homomorphism (or Group Morphism) since a1B⋅a2B=a1a2B so φ(a1a2)=φ(a1)φ(a2).

Note that φ is surjective since if you take any (ab)B for some ab∈AB, then check that:
(ab)B=aB
since a−1(ab)=b∈B via Left (and Right) Cosets form a Partition in G. Then φ(a)=aB=(ab)B so then φ is surjective.

Notice also Ker(φ)={a∈A|aB=1B}. But aB=1B⟺a∈B, so then:
Ker(φ)={a∈A|a∈B}=A∩B
showing injectivity.

Using the The First Isomorphism Theorem of Groups, The Fundamental Theorem for Group Morphisms, then:
AKer(φ)=AA∩B≅ABB
☐

Theorem

Suppose H,K⊴G (see Normal Subgroup) and H≤K. Then:

  1. Since KH={kH|k∈K} (see Quotient Group) then using πH:G→GH such that g↦gH has πH(K)=KH and then KH⊴GH.
  2. (GH)KH≅GK

Example

Like for the other theorems consider Z24Z.

Here H=⟨8⟩ and K=⟨4⟩. Here:
((Z24Z)HKH)={(0+H)+KH,(1+H)+KH,…,(3+H)+KH}
Which equals:
(Z24Z)K= the same thing

Proof

Proof

The strategy is to find a surjection φ:GH→GK with Ker(φ)=KH defined by gH↦gK. If you're worried about well-definedness, it turns out this will just work.

☐

Finally for the last one:

Theorem

Suppose K⊴G (see Normal Subgroup) with π:G→GK be the cannonical projection. Then:

  1. A≤B in G iff A―≤B― in GK=G―
  2. A⊴G implies A―⊴G―. This becomes an iff when we have subgroups H≤G that contain K.
    Then assuming H≤G that contain K:
  3. A≤B then |B:A|=|B―:A―| (see Index (Groups))
  4. A∩B―=A―∩B―
  5. ⟨A∪B⟩―=⟨A―∪B―⟩

Example

Consider D8:

If we alternatively look at K=⟨r2⟩={1,r2} then the left cosets of K in D8 are:

Notice we don't care which representative we use for each subgroup of cosets, so then we can use the bar notation to indicate a coset:

  1. 1―=r2―
  2. r―=r3―
  3. s―=sr2―
  4. sr―=sr3―
    which all are elements of D8K={1―,r―,s―,sr―}. But look! We can make some map:
    {1,s,sr2,r2}→{1―,s―,sr2―,r2―}={1―,s―}
    namely, use π:D8→D8K such that g↦gK=g―. Then for each normal subgroup in the lattice (see the orange boxes in the picture of the lattice):
  5. H1={1,s,sr2,r2} has π(H1)=H1―
  6. H2={1,r,r2,r3}→{1―,r―,r2―,r3―}={1―,r2―}=H2―.
  7. ... we can repeat with each normal subgroup, and we'll get π(Hi)=Hi―.
  8. H3={1,sr3,r2,sr}→{1―,sr3―,r2―,sr―}={1―,sr―}=H3―
  9. But some of them became the other normal subgroups. Namely π({1,s})=H1―
  10. π{1,sr2}=H4―
  11. ...

We can make a lattice for what becomes what, and what stays the same:

Which is the same lattice picture we had before, but only with the normal subgroups!

s, hence multiplying by 4.

Further using the other relations:
φ(s2)=2φ(s)=0,φ(r)+φ(s)=φ(s)−φ(r)
Lastly, we want φ to be surjective and Ker(φ)={1,r2}.
Z2×Z2={(0,0),(0,1),(1,0),(1,1)}
The idea here is to treat the numbers as binary numbers. Namely r↦(0,1) and s↦(1,0) as they are flags for if we have an extra r,s from the Left and Right Cosets of K.

Notice here φ is surjective as we wanted. We can make a table to check:

1 r r2 r3 s sr sr2 sr3
(0,0) (1,0) (0,0) (1,0) (0,1) (1,1) (0,1) (1,1)
Notice:
  • The 1K coset had φ(1K)=(0,0).
  • The other outcomes got partitioned by their cosets.
  • Thus we got Ker(φ)=K as desired.

But notice that in this case we really could've used a whole bunch of other maps since we only have mod2 (some arithmetic just works out).

Proof

Proof

The first two will be quick to check, so let's look at (3). Define ψ:GKer(φ)→im(φ) by ψ(gKer(φ))=φ(g).

Let's see if ψ is well defined. Suppose g1Ker(φ)=g2Ker(φ). Then using Left (and Right) Cosets form a Subgroup, then g2−1g1∈Ker(φ). Thus φ(g2−1g1)=1. Then φ(g2)−1φ(g1)=1H being a morphism. Thus φ(g1)=φ(g2) showing it's well defined.

We should check ψ is a morphism:
ψ((g1Ker(φ))(g2Ker(φ)))=ψ(g1g2Ker(φ))=φ(g1g2)=φ(g1)φ(g2)=ψ(g1Ker(φ))ψ(g2Ker(φ))
Now is ψ surjective? Let h∈im(φ), then there is some g∈G where φ(g)=h. But then ψ(gKer(φ))=φ(g)=h.

Now is ψ injective? Notice:
Ker(ψ)={gKer(φ)∈GKer(φ)|φ(g)=1H}={gKer(φ)|g∈Ker(φ)}={gKer(φ)|gKer(φ)=1Ker(φ)}={1Ker(φ)}={1GKer(φ)}
so ψ is injective. Thus ψ is an isomorphism between the two groups!

☐

Let's use a different example to look at the other 3 theorems we'll cover.

Foundational Example

Consider G=Z24Z. Recall that because it's Cyclic Group via Cyclic Groups of the Same Order are Isomorphic, then the subgroups of Z24Z is 1:1 with the divisors of 24. Let's look at the lattice.

!16 - The Four Horsemen Theorems 2025-02-13 07.15.14.excalidraw

Notice how the lattice makes a literal diamond between subgroups. The next theorem covers when these cases happen.

!The Second Isomorphism Theorem (The Diamond Iso. Theorem)

!The Third Isomorphism Theorem (The Double Fraction Theorem)

Finally for the last one:

!The Fourth Isomorphism Theorem (The Lattice Isomorphism Theorem)