# Rudin proof: closed subsets of compact sets are compact

I'm reading (self-study) Rudin's mathematical analysis book and I'm going through the proof of the following theorem (theorem 2.35 p38):

Closed subsets of compact sets are compact

I will now write out my proof, so can someone verify whether I filled in the details correctly? (so in general, is every step I made valid, are the explanations correct etc. Any questions to test my understanding will be appreciated as well.

Proof (edit): Assume $F \subset K \subset X$ with $F$ closed and $K$ compact. Let $\{V_a\}$ be an open cover of $F$. Then $F \subset \bigcup_a V_a$. We prove that there is a finite subcover of $F$, from which compactness will follow.

Consider the union $\{V_a\}\cup \{F^c\} := V$. We prove that this is an open cover of $K$. Any set in the collection $\{V_a\}$ is open, by hypothesis that this is an open cover of $F$, and $F^c$ is open, since $F$ is closed. So every set in this collection is open. Since $X = F \cup F^c \subset V \cup F^c$, it follows that $K \subset V \cup F^c$ (as $K \subset X$) and this proves that the union written above is an open cover of $K$.

Because $K$ is compact, it follows that there is a finite collection $\Omega \subset V$ which is a subcover of $K$, so $K \subset \bigcup_{A \in \Omega} A$, and because $F \subset K$, it also follows that $F \subset \bigcup_{A \in \Omega} A$.

Now, if $F^c \notin \Omega$, then $\Omega = \{V_{a_1}, \dots, V_{a_n}\}$ for open sets $V_{a_1}, \dots, V_{a_n}$, which is a finite subcover of $F$, since $F \subset \bigcup_{A \in \Omega} A$. Otherwise, if $F^c \in \Omega$, then $\Omega = \{F^c, V_{a_1}, \dots, V_{a_m}\}$, but $F \cap F^c = \emptyset$, so $F \subset V_{a_1}\cup \dots \cup V_{a_n}$, such that $\Omega - \{F^c\}$ is a finite subcover of $F$.

QED

• You can shorten your proof that $C=\{V_a\}_a \cup \{F^c\}$ is an open cover of $X$ by noting that $C$ is (obviously) an open family, and that $\cup C=(\cup_aV_a)\cup F^c \supset F\cup F^c=X.$ – DanielWainfleet Jul 21 '17 at 3:07
• Isn't this what I prove? – user370967 Jul 21 '17 at 8:28
• Yes you did. With more words. But there's nothing wrong with your work – DanielWainfleet Jul 21 '17 at 17:43

There's a mistake:

so we obtain an open cover $K^c \cup V$ of $K$, and because $K$ is compact

You don't know that $K$ is compact. You are trying to prove that $K$ is compact. You only know that $K$ is closed and $F$ is compact. In fact, the mere fact that you don't use $F$ anywhere in your proof should be cause for alarm.

Also, you are confusing covers (which are collections of sets) with their coverage (i.e., union of their component sets).

So, when you write that you take a finite subcover $\Omega$ of $K$, you then say that $K\subset \Omega$, which is not true. What is true is that $$K\subset\bigcup_{A\in\Omega} A$$ which is different!

I advise you to restart the proof again, and here's a couple points to start you off:

• You need to prove that $K$ is compact
• That means you need to prove every open cover of $K$ has a finite subcover
• That means you take some cover, $\mathcal V$, such that $K$ is covered by $\mathcal V$ (you were on point until here).
• Now you need to prove there exists a finite subcover of $\mathcal V$.

And a little hint:

• You know $F$ is compact, so any open cover of $F$ has a finite subcover.
• Can you add one more set to $\mathcal V$ such that the resulting cover will be a cover for $F$?

You still have that misstake of cover vs coverage. So, instead of saying that $K^c\cup V$ is an open cover of $F$, you should say that $\{K^c\}\cup \{V_a\}$ is an open cover of $F$.

I am even more concerned that I think you are confused a bit about what an open cover is. When you say

Now, consider the union $K^c \cup V$. Since $K \subset V$, it follows that $X =K^c \cup K \subset K^c \cup V$, meaning that $X = K^c \cup V$, so $F \subset K^c \cup V$ and $X$ is an open subset of itself

I'm thinking "yes, all he said is true, but it's irrelevant".

An open cover is a collection of open sets, not a collection of sets whose union is open. So, you don't need to prove that $$K^c\cup V$$ is open (even though it is), you need to prove that every element of $$\{K^c\}\cup\{V_a\}$$ is open (which is not hard, it's just that if you don't do that, the proof is incorrect).

• I corrected the proof. Is it correct now? (don't look at my misunderstanding of open cover/open coverage, thank you for pointing that out though!) – user370967 Jul 20 '17 at 15:15
• @Math_QED I added an explanation into my answer. – 5xum Jul 20 '17 at 15:29
• Your answer and edits made me realise a lot of mistakes. Thank you for that. – user370967 Jul 20 '17 at 15:31
• @Math_QED That's why we're here :) – 5xum Jul 20 '17 at 15:32
• @Math_QED It looks much much better. Well done! – 5xum Jul 21 '17 at 6:12