0

If $X$ is a locally compact, second-countable topological space, then is its Alexandroff extension $X^*$ also second-countable?

Our definition of locally compact is that for every $x$ in $X$, we have an open subset $U$ of $X$ such that $\operatorname{cl}_X U$ is compact and $x \in U$. I was able to prove that if $\mathcal{B}$ is a base for $X$, then the subset

$$ \bigl\{U \in \mathcal{B} : \text{$\operatorname{cl}_X U$ is compact}\bigr\} $$

is also a base for $X$. So we can assume without loss of generality that our (countably many) basic open sets have compact closure.

Since the set of finite subsets of a countable set is also countable, we have that the set $\mathcal{B}_\mathrm{f}$ of finite unions of sets in $\mathcal{B}$ is also countable. Define

$$ \mathcal{C} := \mathcal{B} \cup \bigl\{X^* \setminus \operatorname{cl}_X U : U \in \mathcal{B}_\mathrm{f}\bigr\} $$

which is a countable collection of open subsets of $X^*$.

Is the following proof that every open subset of $X^*$ is a union of sets in $\mathcal{C}$ correct?

Let $x \in U$ which is an open subset of $X^*$. If $x$ is a point in $X$, then $U \cap X$ is open in $X$, and so there is a $V$ in $\mathcal{B} \subset \mathcal{C}$ such that $x \in V$ and $V \subset U \cap X \subset U$. Otherwise $x$ is the point at infinity, so $U = X^* \setminus K$ where $K$ is a compact, closed subset of $X$. We can cover $K$ by finitely many sets $U_n$ in $\mathcal{B}$, so that $K \subset \bigcup_n U_n \in \mathcal{B}_\mathrm{f}$. It follows that $x \in X^*\setminus \operatorname{cl}_X \bigl(\bigcup_n U_n\bigr)$ which is contained in $X^* \setminus K = U$.

In both cases, we have found a set in $\mathcal{C}$ which is a neighborhood of the point $x$ and is contained in $U$. Hence, $\mathcal{C}$ is a base for $X^*$.

shoteyes
  • 1,425
  • Are you sure "It follows that $x \in X^\setminus \operatorname{cl}_X \bigl(\bigcup_n U_n\bigr)$"? I can only see that $x \in X^\setminus \bigl(\bigcup_n U_n\bigr).$ – Anne Bauval Sep 13 '22 at 11:06
  • @AnneBauval I suppose that since $K \subset \bigcup_n U_n \subset \operatorname{cl}_X\bigl(\bigcup_n U_n\bigr)$, then the reverse inequality holds for their complements in $X^*$. As for the possible duplicate, I think these are different questions since I don't assume the space is Hausdorff. – shoteyes Sep 13 '22 at 11:13
  • And actually the closure in $X$ is a subset of $X$, and therefore doesn't contain the point at infinity (which I was assuming $x$ was in that context). – shoteyes Sep 13 '22 at 11:19
  • The definition of "locally compact" is unusual. I would expect taht each point $x$ has a compact neigbhborhood, but not necessarily a closed compact one. – Paul Frost Sep 13 '22 at 11:24
  • I agree that "the reverse inequality holds for their complements" but this didn't prove the claim, knowing only that $x$ belongs to $X^*\setminus K.$ On the other hand, for $x=\infty$, the same claim is obvious hence why such an intricate argument? – Anne Bauval Sep 13 '22 at 11:27
  • @PaulFrost I think you make a good point. I have seen the property I used be referred to as strongly locally compact before. – shoteyes Sep 13 '22 at 11:28
  • 1
    @AnneBauval I needed to find a set in $\mathcal{C}$ which is contained in $X^* \setminus K$ and is a neighborhood of $x$ (which in the context of my proof there was the point at infinity). We both agree that $X^* \setminus \operatorname{cl}_X\bigl(\bigcup_n U_n\bigr)$ achieves the first part by reversing the inequality, and it achieves the second part of being a neighborhood of $x = \infty$ which we both also agree on. – shoteyes Sep 13 '22 at 11:34

1 Answers1

0

Your definition of "locally compact" is unusual. I know two variants:

  1. Each point has a compact neighborhood.

  2. Each point has a neighborhood base consisting of compact sets.

They are equivalent for Hausdorff spaces, but you do not require spaces to be Hausdorff. As you say, you use

  1. Each point has a compact closed neighborhood.

This property also seems to be denoted as strongly locally compact.

In the context of the Alexandroff extension this variant is in fact extremely useful because the open neigborhoods of $\infty$ in $X^*$ are precisely the complements $X^* \setminus K$, where $K \subset X$ is compact and closed in $X$.

Your proof is correct. Its essence is that $X$ has a countable base $\mathcal B$ with the following properties:

  • For each $U \in \mathcal B$ the closure $\operatorname{cl}_X U$ is compact.
  • Each compact $K \subset X$ is contained is some $U \in \mathcal B$.

You start with a countable base $\mathcal B_0$ of $X$ and observe that $\mathcal B_1 = \{ U \in \mathcal B \mid \operatorname{cl}_X U \in \mathcal B \}$ is also a base for $X$ (which is clearly countable). Then trivially the set $\mathcal B$ of all finite unions of elements of $\mathcal B_1$ is again a basis. Clearly $\mathcal B$ is countable and for each $U \in \mathcal B$ the closure $\operatorname{cl}_X U$ is compact. This holds because if $U = \bigcup_{i=1}^n U_i$ with $U_i \in \mathcal B_1$, then $U \subset C := \bigcup_{i=1}^n \operatorname{cl}_X U_i$, thus $\operatorname{cl}_X U \subset C$. Because $C$ is compact and $\operatorname{cl}_X U$ is closed in $C$, it is compact. Moreover each compact $K \subset X$ can be covered by finitely many $U_i \in \mathcal B_0$, thus it is contained in some $U \in \mathcal B$.

Therefore $\mathcal{C} := \mathcal{B} \cup \bigl\{X^* \setminus \operatorname{cl}_X U : U \in \mathcal{B}\bigr\} $ is a countable base of $X^*$.

Paul Frost
  • 76,394
  • 12
  • 43
  • 125
  • You consolidated the properties of a desirable countable base pretty well with those two bullet points. I'll admit while coming up with my solution I was adding more sets to my base ad hoc at times when I realized I didn't have enough basic sets to prove the next fact. So I appreciate that you stripped the construction to its essentials with those two properties. Working backwards from that is cleaner, but of course, I didn't have that foresight when I was sketching up the proof – shoteyes Sep 13 '22 at 13:35