Notes:Quotient
Terminology
Let X be a set and let ∼ be an equivalence relation on the elements of X.
- Then X∼ denotes the "equivalence classes" of
This is best thought of as a map:
- π:X→X∼ by π:x↦[x] where recall:
- [a]={x∈X|x∼a}, the notation [a] makes sense, as by the reflexive property of ∼ we have a∈[a]
Quotient structure
Suppose that ⊙:X×X→X is any map, and writing x⊙y:=⊙(x,y) when does ⊙ induce an 'equivalent' mapping on X∼?
- This is a mapping: ⊙:X∼×X∼→X∼ where [x]⊙[y]=[x⊙y]
- should such an operation be 'well defined' (which means it doesn't matter what representatives we pick of [x] and [y] in the computation)
Alternatively
We have no concept of ⊙ on X∼, but we do on X. The idea is that:
- Given a [x] and a [y] we go back
- To an x and a y representing those classes.
- Compute x⊙y
- Then go forward again to [x⊙y]
In functional terms we may say:
- ⊙:X∼×X∼→X∼ given by:
- ⊙([x],[y])↦π(π−1([x])⊙π−1([y])⏟if ⊙ makes sense)=[π−1([x])⊙π−1([y])]
Here \pi^{-1}([x]) is a subset of X containing exactly those things which are equivalent to x (as these things all map to [x]).
- We can say A\odot B (for A\subseteq X and B\subseteq X) if a\odot b\sim a'\odot b'
As then
- We can define \pi(A) (for A\subseteq X ) properly if
This all seems very contrived
As a diagram
I seem to be asking when a map (dashed line) is induced such that the following diagram commutes:
\begin{xy}\xymatrix{X\times X \ar@{->}[r]^-\odot \ar@{->}[d]_-{\pi\times\pi} \ar@{-->}[dr]^{\pi\circ\odot} & X \ar@{->}[d]^-{\pi} \\ \frac{X}{\sim}\times\frac{X}{\sim} \ar@{.>}[r]_-{\odot} & \frac{X}{\sim} }\end{xy} |
Diagram |
---|
It is quite simple really:
- The dashed arrow exists by function composition.
- Using the Factor (function) idea, if we have (for (v,v')\in V\times V and (u,u')\in V\times V - from wanting the diagram to commute):
- [(\pi\times\pi)(v,v')=(\pi\times\pi)(u,u')]\implies[\pi(+(v,v'))=\pi(+(u,u'))] then
- there exists a unique function, \odot:\frac{X}{\sim}\times\frac{X}{\sim}\rightarrow\frac{X}{\sim} given by: \odot:=(\pi\circ+)\circ(\pi\times\pi)^{-1}
- [(\pi\times\pi)(v,v')=(\pi\times\pi)(u,u')]\implies[\pi(+(v,v'))=\pi(+(u,u'))] then
Note that while technically these are not functions (as they'd have to be bijective in this case) as FOR ANY x=(\pi\times\pi)^{-1}(a,b) we have \odot(x) being the same, it doesn't matter what element of (\pi\times\pi)^{-1} we take.