context | ⟨X,TX⟩ … topological space |
definiendum | F in it |
inclusion | F in SetOp(X)op |
for all | U∈TX |
for all | s,t∈FU |
for all | CU … open cover(U) |
postulate | (∀(V∈CU). s|V=t|V)⟹s=t |
Here we use the notation discussed in presheaf. I.e. in the last line, “s|V=t|V” is notation for “F(i)(s)=F(i)(t)”, where i:V→U.
A presheaf in topology assigns a set FU to each open set U of a topological space ⟨X,T⟩. A sheaf is a presheaf fulfilling two axioms ('locality' and 'gluing') and it's intended to capture the case where FU are sets of sections over a topological space (e.g. the section of a fibre bundle p:E→X, e.g. vector fields, e.g. all 1-forms) and the arrows F(i) are restrictions of those section to smaller open sets. The definition of a seperated presheaf is the first step towards the definition of a sheaf.
Locality axiom: To understand the postulate, recall that for sections s,t:U→Y (or function in general) we trivially have that
s=t⟹∀(x∈U).s(x)=t(x) and as corollary we have s=t⟹∀(V⊆U).s|V=t|V .
Sections also fulfill function extensionality, which goes in the reverse direction
(∀(x∈U).s(x)=t(x))⟹s=t and consequently, if CU is a covering of U, then (∀(V∈CU).s|V=t|V)⟹s=t .
Now if the FU's of a sheaf are to be a sets of such sections over a topological space, then it should be that the maps F(i) between them can be seen as restriction of the function domains. To this end, if i:V→U is the inclusion of a small open set V in another U, require that if s,t∈FU agree on all restricted domains which make up the cover of U, i.e. F(i)(s)=F(i)(t), then there can be no other way in which they can differ, so s=t.
Wikipedia: Sheaf