Notes:Coset stuff

From Maths
Jump to: navigation, search
See /Quotient group for this applied to the quotient group (which is after all the point)

Stuff

Let (G,×) be a group and let HG be a subgroup. Proper or not. Then

  • Any set of the form gH is called a left coset, where gH:={g×h | hH}
  • Any set of the form Hg is called a right coset, where Hg:={h×g | hH}

H itself is a coset as eH=H clearly (for e the identity of G)

Claims

  1. For x,yG we can define an equivalence relation on G: xyx1yH[Note 1]. Done - Alec (talk) 21:23, 23 October 2016 (UTC)
    • By symmetry this is/must be the same as y1xH - this is true as H is a subgroup.
  2. x,gG[x[g]xgH]. Done - Alec (talk) 21:23, 23 October 2016 (UTC)
  3. For x,yG we can define another equivalence relation on G: xyxy1H. Done - Alec (talk) 21:23, 23 October 2016 (UTC)
    • By symmetry this is/must be the same as yx1H - this is true as H is a subgroup.
  4. x,gG[x[g]xHg]. Done - Alec (talk) 21:23, 23 October 2016 (UTC)

Going forward

I have shown we get two equivalence relations. and Thus we get two partitions of G, and:

  • π:GG given by π:g[g] and π:GG given by π:g[g]

Can we factor anything through G or G?

We can factor a map, f:GX (for some other thing X) if:

  • g,hG[π(g)=π(h)f(g)=f(h)]

Actually lets try factoring the group operation through this!

  • (g,h),(g,h)G[π(g,h)=π(g,h)×(g,h)=×(g,h)][Note 2]
    • Then ([g],[h])=([g],[h]) so [g]=[g] and [h]=[h]
    • So gg and hh
      • Thus h1,h2H such that g1g=h1 and h1h=h2. We want to show gh=gh
        • Well h1h2=g1gh1h

Here we get stuck. If we had gH=Hg then we could go further and factor the × through, then we can proceed to:

  • gh4h=gh for some h4H (I did h4:=h1h3 on paper but I don't know if I used the same h1 here. h3 comes from turning either gH into Hg or hH into Hh with h1 or h2 what it was "before")

The result is, if h4 is known to be e then we can factor. So if H is the trivial group, we can factor. We must have ×=¯×π so this is another way of saying a group "over" the trivial group is isomorphic to the group.

Really stupid and pointless way of saying it.

We can get multiplication through:

If we want π to be a group homomorphism we require this in fact.

Proof of claims

[Expand]

Proofs here

Notes

  1. Jump up It must be this way as we will require xx, then we get x1x=eH as H is a subgroup.
  2. Jump up We're informal about how we use π here, we really mean:
    • π1:G×GG×G given by π:(g,h)([g],[h])=(π(g),π(h))