Processing math: 100%

Thursday, April 17, 2014

Equivalence relations, partitions, and surjections

[This is a draft!]
In this post, we fix attention on a fixed set XSet.
We consider, compare, and count three structures X may possess:
  1. Equivalence relations EX on X;
  2. Partitions πP(X) of X;
  3. Surjections (i.e., onto functions) f:XY out of X.
There are two fundamental relationships between those structures:
  1. There is canonical bijection between equivalence relations on X and partitions of X.
  2. There is a coreflexive (pronounced co-reflexive) adjoint equivalence between the category of equivalence relations on X and the category of surjections out of X.
The topics to be covered in this post are:
  1. Equivalence relations
  2. Partitions
  3. Quotient sets, equivalence classes, and quotient maps
  4. Kernels (sometimes known as “kernel pairs”)
  5. Adjoint equivalences

[Everything below the horizontal line is just experimenting with formatting.]
Equivalence relations
A brief review of equivalence relations. The standard definition of equivalence relations is that it is a relation which is reflexive, transitive, and symmetric.
Our definition of an equivalence relation is logically equivalent: that it is a thin groupoid, called by some a setoid.
A groupoid is just a category in which each arrow is invertible. Being a category gives the reflexive and transitive parts of the traditional definition, being “thin” (meaning that there is at most one arrow between any two objects) makes it a relation, and being each arrow being invertible makes the relation symmetric.
Partitions
A brief review of partitions. A partition π of a set X is a family of non-empty subsets of X which are (pair-wise) disjoint and such that their union is all of X. Thus each element xX is a member of one and only one of the non-empty subsets which comprise π.
So we have a canonical surjection Xπ : xX the member of π which contains x. It is a surjection because each member of π is non-empty.
Quotient sets, equivalence classes, and quotient maps
A brief review of quotient sets, equivalence classes, and quotient maps. Given an equivalence relation E on X, we define the equivalence class [x] of xX to be its orbit or connected component under the groupoid (!). That is equivalent to the customary definition.

For a simple example, we consider the set X={1,2,3}. {1,2,3} has five partitions, each of which is the quotient object in a short exact sequence. The exact sequences corresponding to three of the five partitions are: E as function X×X{0,1}EX×XXX/EPartition typecomment(111111111){1,2,3}×{1,2,3}{1,2,3}{{1,2,3}}3the coarsest, chaotic, case(110110001){(1,1),(1,2),(2,1),(2,2),(3,3)}{1,2,3}{{1,2},{3}}2+1(100010001){(1,1),(2,2),(3,3)}{1,2,3}{{1},{2},{3}}1+1+1the finest, discrete, case The other two partitions are also of type 2+1; the difference is that the block of the partition consisting of only one element contains respectively 1 and 2.

Consider the following diagram ("ker" is kernel): ker(f)Xclass mapX/ker(f)ker(f)XfY Here the surjection f is at lower right;
the partition is X/ker(f) at top right;
the equivalence relation is ker(f) on the left.
An important adjoint equivalence (in combinatorics and elsewhere; XSet) is Equivrelns(X)X/kernelXSurj
Equivrelns(X)X/Part(X)kernelform the class mapXSurj
Equivrelns(X)X/Part(X)form the class mapXSurj=Equivrelns(X)kernelXSurj

No comments:

Post a Comment

MathJax 2.7.9