3
$\begingroup$

Let $(X,\tau)$ be a topological space. We topologize $\tau$ itself in the following way. For $K\subseteq X$ compact, we set $${\cal V}_K=\{U\in \tau: U \supseteq K\}.$$ The compact-open topology on $\tau$ is the topology generated by $\{{\cal V}_K: K\subseteq X\text{ is compact}\}.$ (The reason I call this topology compact-open is given below.)

Question: If ${\cal K}$ is a compact subset of $\tau$ with the compact-open topology, is $\bigcap {\cal K}$ necessarily open?


Let $S$ be the Sierpinski space. Then $\tau$ can be identified with the set $\text{Hom}(X,S)$ of continuous functions from $X$ to $S$. We can endow $\text{Hom}(X,S)$ with the compact-open topology and "transfer" it to $\tau$ - and that's what we get above.

$\endgroup$
4
$\begingroup$

The answer is yes if $X$ is locally compact. Indeed, consider the topological space $X \times \mathcal{K}$ and let $V = \{(x,U) \in X \times \mathcal{K} | x \in U\}$. We claim that $V$ is open in $X \times \mathcal{K}$. To see this, take $(x,U) \in V$ and let $W$ be an open neighborhood of $x$ such that $\overline{W}$ is compact and contained in $U$. Then the set $W \times (\mathcal{K} \cap \mathcal{V}_{\overline{W}}) \subseteq X \times \mathcal{K}$ is an open neighborhood of $(x,U)$ which is contained in $V$, and so $V$ is open. Let us now show that $\cap_{U \in \mathcal{K}} U \subseteq X$ is open. First observe that $y \in \cap_{U \in \mathcal{K}} U$ if and only if $y \times \mathcal{K}$ is contained in $V$. Since $V$ is open and $\mathcal{K}$ is compact, if $y \times \mathcal{K} \subseteq V$ then there exists an open neighborhood $U_y$ of $y$ such that $U_y \times \mathcal{K} \subseteq V$ (this is a standard compactness argument). We may finally conclude that $U_y \subseteq \cap_{U \in \mathcal{K}} U$ and so $\cap_{U \in \mathcal{K}} U$ is open.

If $X$ is not locally compact then I'm not sure what the answer should be, but I would guess that the claim is false in general. This is partly because, as you said, this is a particular case of the compact-open topology on function spaces, which is known to be well-behaved for locally compact spaces (and some more general spaces), but not for all spaces.

Edit: Here is a template counter-example which shows that the claim is not true for general topological spaces. Let $X$ be a $T_1$-space and $A \subseteq X$ a subset satisfying the following two properties:

1) The intersection of $A$ with any compact subset of $X$ is finite.

2) $A$ is not closed.

Let us set $U_a = X \backslash \{a\}$. Then the collection $\mathcal{K} = \{U_a\}_{a \in A}$ is compact in the compact-open topology because by (1) every open set in $\mathcal{K}$ is cofinite. On the other hand, $\cap_{U \in \mathcal{K}}U = X \backslash A$ is not open by (2). To show that such spaces exist choose a non-principal ultrafilter $\mathcal{F}$ on the set $\mathbb{N}$ of natural numbers and set $X = \mathbb{N} \cup \{\infty\}$ where the open neighborhoods of $\infty$ are given by $\infty \cup U$ for $U \in \mathcal{F}$. Let $A = \mathbb{N} \subseteq X$. Then it is not hard to verify that $A$ satisfies properties (1) and (2) above.

$\endgroup$

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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