0
$\begingroup$

For any topological space $X$ one can consider the so called frame of all open subsets of $X$ to be denoted by $\mathcal{O}(X)$. If $f:X \to Y$ is continuous taking the inverse image we get the morphism between frames $\mathcal{O}(f): \mathcal{O}(Y) \to \mathcal{O}(X)$. In this manner we obtain a contravariant functor. On the other side, for a frame $F$ one can define the space $Pt(F)$ of points of $F$ to be $Hom(F,\{0,1\})$ declaring all the sets of the form $\Phi(U):=\{p \in Pt(F): p(U)=1\}$ as opens. If $g:F \to G$ is a frame morphism we get a continuous map in the opposite direction $Pt(g):Pt(G) \to Pt(F)$ by $Pt(g)(q)=q \circ g$.

A topological space $X$ is called sober if any irreducible subset $A$ of $X$ is a closure of the unique point. This is equivalent to the condition that the map $\delta:X \to Pt(\mathcal{O}(X))$ is a homeomorphism where $\delta_x(U):=1$ if $x \in U$ and $0$ otherwise.

Take $X$ to be any topological space (not necessarilly sober). Then even the map $\delta$ need not to be homeomorphism, still $\mathcal{O}(\delta):\mathcal{O}(Pt(\mathcal{O}(X))) \to \mathcal{O}(X)$ is an isomorphism of frames. This isomorphism turns out to be equal to $\Phi^{-1}$. In particular $\mathcal{O}(X)$ is always spatial (see below).

Now we would like to repeat the argument but for the other direction: we call a frame $F$ to be spatial if the following condition holds: if it is not true that $a \leq b$ in $F$ then there exists $p \in Pt(F)$ such that $p(a)=1$ and $p(b)=0$. Now it is not true that for any frame $F$ the map $Pt(\Phi)$ is always a homeomorphism: it is $\iff$ $Pt(\Phi)$ is surjective $\iff \Phi$ is injective $\iff F$ is spatial. In this case $Pt(\Phi)=\delta^{-1}$.

Question 1: Is the above reasoning concerning spatiality of $F$ correct?

In particular it is not true that always (i.e. for any frame $F$-not necessarily spatial) $Pt(F)$ is isomorphic to $Pt(\mathcal{O}(Pt(F)))$: it happens precisely when $F$ is spatial which holds iff $Pt(F)$ is sober. I'm pretty sure that I am missing something here: so let me ask

Question 2: Is the space $Pt(F)$ always sober? Is it true that always $Pt(F) \cong Pt(\mathcal{O}(Pt(F)))$ or it is true precisely when $F$ is spatial?

I would be grateful if someone can clarify these issues.

$\endgroup$
1
  • $\begingroup$ Also, in case you're not aware, I think many people's go-to reference for this stuff is Johnstone's Stone Spaces. $\endgroup$
    – Tim Campion
    Feb 12, 2021 at 17:57

1 Answer 1

4
$\begingroup$

Yes, it is always true that $Pt(F) \cong Pt(\mathcal{O}(Pt(F)))$ (i.e. $Pt(F)$ is sober), whether or not $F$ is spatial. Indeed, this follows formally from the fact that $\mathcal{O}(Pt(\mathcal{O}(X))) \cong \mathcal{O}(X)$ for all spaces $X$, using the theory of idempotent adjunctions.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.