Cosheaf

From testwiki
Jump to navigation Jump to search

In topology, a branch of mathematics, a cosheaf is a dual notion to that of a sheaf that is useful in studying Borel-Moore homology.Template:Further explanation needed

Definition

We associate to a topological space X its category of open sets Op(X), whose objects are the open sets of X, with a (unique) morphism from U to V whenever UV. Fix a category 𝒞. Then a precosheaf (with values in 𝒞) is a covariant functor F:OpX𝒞, i.e., F consists of

  • for each open set U of X, an object F(U) in 𝒞, and
  • for each inclusion of open sets UV, a morphism ιU,V:F(U)F(V) in 𝒞 such that
    • ιU,U=idF(U) for all U and
    • ιU,VιV,W=ιU,W whenever UVW.

Suppose now that 𝒞 is an abelian category that admits small colimits. Then a cosheaf is a precosheaf F for which the sequence

(α,β)F(Uα,β)(α,β)(ιUα,β,UαιUα,β,Uβ)αF(Uα)αιUα,UF(U)0

is exact for every collection {Uα}α of open sets, where U:=αUα and Uα,β:=UαUβ. (Notice that this is dual to the sheaf condition.) Approximately, exactness at F(U) means that every element over U can be represented as a finite sum of elements that live over the smaller opens Uα, while exactness at αF(Uα) means that, when we compare two such representations of the same element, their difference must be captured by a finite collection of elements living over the intersections Uα,β.

Equivalently, F is a cosheaf if

  • for all open sets U and V, F(UV) is the pushout of F(UV)F(U) and F(UV)F(V), and
  • for any upward-directed family {Uα}α of open sets, the canonical morphism limF(Uα)F(αUα) is an isomorphism. One can show that this definition agrees with the previous one.[1] This one, however, has the benefit of making sense even when 𝒞 is not an abelian category.

Examples

A motivating example of a precosheaf of abelian groups is the singular precosheaf, sending an open set U to Ck(U;), the free abelian group of singular k-chains on U. In particular, there is a natural inclusion ιU,V:Ck(U;)Ck(V;) whenever UV. However, this fails to be a cosheaf because a singular simplex cannot be broken up into smaller pieces. To fix this, we let s:Ck(U;)Ck(U;) be the barycentric subdivision homomorphism and define Ck(U;) to be the colimit of the diagram

Ck(U;)sCk(U;)sCk(U;)s.

In the colimit, a simplex is identified with all of its barycentric subdivisions. One can show using the Lebesgue number lemma that the precosheaf sending U to Ck(U;) is in fact a cosheaf.

Fix a continuous map f:YX of topological spaces. Then the precosheaf (on X) of topological spaces sending U to f1(U) is a cosheaf.[2]

Notes

Template:Reflist

References

Template:Ref begin

Template:Ref end


Template:Topology-stub