nLab
open map

Context

Topology

Topos Theory

Contents

Definition

For maps between topological spaces

A function f:XY between topological spaces is called open if the image of every open set in X is also open in Y.

Recall that f is a continuous map if the preimage of every open set in Y is open in X. For defining open maps typically one restricts attention to open continuous maps, although it also makes sense to speak of open functions that are not continuous.

Examples

  • For any two topological spaces X, Y, the projection map π:X×YY is open.

  • If G is a topological group and H is a subgroup, then the projection to the coset space p:GG/H, where G/H is provided with the quotient topology (making p a quotient map), is open. This follows easily from the observation that if U is open in G, then so is

    p 1(p(U))=UH= hHUhp^{-1}(p(U)) = U H = \bigcup_{h \in H} U h
  • If p:AB and q:CD are open maps, then their product p×q:A×CB×D is also an open map.

For morphisms between locales

A continuous map f:XY of topological spaces defines a homomorphism f *:Op(Y)Op(X) between the frames of open sets of X and Y. If f is open, then this frame homomorphism is also a complete Heyting algebra homomorphism; the converse holds for sober spaces (maybe as long as Y is T 0?). Accordingly, we define a map f:XY of locales to be open if it is, as a frame homomorphism f *:Op(Y)Op(X), a complete Heyting algebra homomorphism, i.e. it preserves arbitrary meets and the Heyting implication.

This is equivalent to saying that f *:Op(Y)Op(X) has a left adjoint f ! (by the adjoint functor theorem for posets) which satisfies the Frobenius reciprocity condition that f !(Uf *V)=f !(U)V.

For geometric morphisms of toposes

Categorifying, a geometric morphism f:XY of toposes is an open geometric morphism if its inverse image functor f *:YX is a Heyting functor.

For morphisms in a topos

A class RMor() of morphisms in a topos is called a class of open maps if it satisfies the following axioms.

  1. Every isomorphism belongs to R;

  2. The pullback of a morphism in R belongs to R.

  3. If the pullback of a morphism f along an epimorphism lands in R, then f is also in R.

  4. For every set S the canonical morphism ( sS*)* from the S-fold coproduct of the terminal object to the terminal object is in R.

  5. For {X if iY i} iIR then also the coproduct iX i iY i is in R.

  6. If in a diagram of the form

    Y p X g f B\array{ Y &&\stackrel{p}{\to}&& X \\ & {}_{\mathllap{g}}\searrow && \swarrow_{\mathrlap{f}} \\ && B }

    we have that p is an epimorphism and g is in R, then f is in R.

The class R is called a class of étale maps if in addition to the axioms 1-5 above it satisfies

  1. for f:XY in R also the diagonal YY× XY is in R.

  2. If in

    Y p X g f B\array{ Y &&\stackrel{p}{\to}&& X \\ & {}_{\mathllap{g}}\searrow && \swarrow_{\mathrlap{f}} \\ && B }

    we have that p is an epimorphism, and p,gR, then fR.

For instance (JoyalMoerdijk, section 1).

References