3

For a topological space $X$ that is Hausdorff, and $\{B_k\}_{k\in I}$ is an arbitrary collection of compact sets, I need to show that $\bigcap\limits_{k\in I} B_k$ is also compact.

Here's the rub. I do not have closed sets to work with. The proof that compact sets in a Hausdorff space are closed is proven in the next section of the book, so I am forbidden from using it in my proof. Unfortunately, I cannot find a proof online anywhere that doesn't rely on closedness. And I've been struggling with this proof for three weeks. Any pointers or full fledged proofs that avoid closedness would be appreciated. I thought I could prove this on my own but have only failed. Im out of time and even when cheating I couldn't find a single useful bit of advice.

Edit A few people have offered some advice, which I appreciate. But to be clear, my instructor isn't going to expect much more beyond the open set cover definition of compactness. We really have little else to work with, sorry to say. I don't know what duals are, I cannot infer that compact sets are closed, and I'm sure there are lots of tricks up people's sleeves that I am just not allowed to use at this juncture.

  • What definition of compactness do you have? The open cover one? Because then you discussed open sets but not closed? – Marc May 29 '18 at 04:04
  • Open covers is how compactness is defined, yes. Every open cover has a finite subcover. I dont have a closed set definition for compactness, if that is what youre asking. – CogitoErgoCogitoSum May 29 '18 at 04:07
  • Its funny too because I dont know how else to use Hausdorff... aside from concluding closedness of compact sets in some other secondary proof what else good does it do me? I dont know why else its here in the question if Im not supposed to infer closedness. But Im not supposed to and no proof seems to exist that doesnt. – CogitoErgoCogitoSum May 29 '18 at 04:56
  • Just a quick thought (and it's a very very long time since have looked at this stuff!) - for each of your sets B_k (on the move so apols for words and no mathjax) an open cover A of the intersection can form part of an open cover, where every point in the set B_k but not in the starting open cover A (of the intersection) would be contained in an open set disjoint from A. Of this resultant open cover for B_k you can find a finite subcover. Would the intersection of all these finite subcovers do it? Again apologies for a) thinking out loud and b) badly expressed and not typeset rudimentary idea! – Mehness May 29 '18 at 05:10
  • I dont know, tbh. Id have to think about it and its too late for me at the moment, will have to return. I appreciate the suggestion though. All you need to do is enclose your latex in dollar signs and it will render. – CogitoErgoCogitoSum May 29 '18 at 05:15
  • Sure, (late here too hence forgot the in line $) key thing wanted to incorporate directly in some way was the hausdorff nature, hope something direct surfaces! – Mehness May 29 '18 at 05:20

1 Answers1

3

The first part is taken from Brian M. Scott in How to prove that a compact set in a Hausdorff topological space is closed?.

Lemma: Let $B$ and $C$ be two compact sets in $X$ and fix $x\in B\setminus C$. Then there exists an open set around $x$ that is disjoint from $C$.

Proof: For each $y\in C$ we can, by Hausdorff, construct two disjoint open sets $U_y$ and $V_y$ such that $x\in U_y$ and $y\in V_y$. Now $\left\{V_y\mid y\in C\right\}$ is an open cover of $C$, so by compactness there exists a subcover $\left\{V_y\mid y\in \tilde C\right\}$, where $\tilde C \subset C$ is finite. Now $$ \bigcap_{y\in \tilde C} U_y $$ is an open set that contains $x$ and is disjoint from $C$.


Let $Y = \bigcap_{k\in I} B_k$ and let $\mathcal{O}$ be an open cover of $Y$. Now, for each $x \in B_1\setminus Y$ there exists a $k\in I$ such that $x\notin B_k$. Therefore, by the lemma, there exists an open set $O_x(k)$ such that $x\in O_x(k)$ and $O_x(k) \cap Y \subseteq O_x(k)\cap B_k = \emptyset$. We then have an open cover of $B_1$ given by $$ \mathcal{O} \cup \left\{ O_x(k)\mid x\in B_1\setminus Y\right\}, $$ which by the compactness of $B_1$ has a finite subcover. Since any $y\in Y$ is not covered by any element in $\left\{ O_x(k)\mid x\in B_1\setminus Y\right\}$ we conclude that there must be a finite subset of $\mathcal{O}$ that covers $Y$.

Marc
  • 6,861
  • 1
    Hi in your last sentence did you mean $\left{ O_x(k)\mid x\in B_1\setminus Y\right}$? - I think this proof was along the lines of what I was loosely thinking earlier, thanks! – Mehness May 29 '18 at 11:04
  • 1
    Yes, definitely! Thank you for the careful reading:) – Marc May 30 '18 at 14:01