1
$\begingroup$

I'm reading proofs about a closed subspace of a compact space being compact, but in my mind I think i can prove it for far more than only closed subspaces. For example, $X$ being compact, then for every open cover of $X$ I can find a finite cover for $X$. Since the elements of this finite cover are opens in $X$, let's name them $U_{\alpha}$, when I take the intersection $U_{\alpha}\cap Y$, I have a finite cover for $Y$ made of open sets... Hmmm, so, what if I just do like this:

Suppose $A$ an open cover for $Y$, then for each $A_{\alpha}$, take $U_\alpha$ such that $A_\alpha\cap U_\alpha$ is open in $X$. I now must construct an open cover for $X$ with these things.. Hmmm, I think that in the case where $Y$ is closed, I can just make the union with the complementar, but in the other cases I'm not sure if the union of those things is gona give the entire $X$. Is this the reason why I must suppose a closed $Y$?

$\endgroup$
  • $\begingroup$ If a (Hausdorff) set is not closed then it can't be compact (take a sequence not converging in the set, and draw countably many small opens balls around its elements) $\endgroup$ – reuns Nov 16 '16 at 18:25
  • $\begingroup$ @user1952009: That’s true in Hausdorff spaces but not in general. $\endgroup$ – Brian M. Scott Nov 16 '16 at 18:26
  • 1
    $\begingroup$ The answer to your last question is yes. If $Y$ is not closed, you may not be able to find an open cover of $X$ whose trace on $Y$ is the one with which you started. As an example, let $X=[0,1]$ and $Y=(0,1)$. Let $U_n=\left(\frac1n,1\right)$ for $n\in\Bbb Z^+$. Then $\{U_n:n\in\Bbb Z^+\}$ is an open cover of $Y$, but there is no way to find open $V_n$ in $[0,1]$ such that $V_n\cap Y=U_n$ for each $n\in\Bbb Z^+$ and $\{V_n:n\in\Bbb Z^+\}$ covers $X$. $\endgroup$ – Brian M. Scott Nov 16 '16 at 18:28
1
$\begingroup$

In a Hausdorff space (for instance, a metric space), a compact subspace is closed. So your attempt at a proof is wrong.

A counterexample is easy: $(-1,1)$ is not compact, because the open cover by the intervals $(-1 + 1/n,1 - 1/n)$ (for integer $n>0$) has no finite subcover; however $[-1,1]$ is compact.

Where is your attempt wrong? You have to start with an open cover for $Y$, not for $X$. Not every open cover for $Y$ is necessarily obtained by considering $U_\alpha\cap Y$, where the sets $U_\alpha$ form an open cover for $X$.

On the other hand, if $Y$ is closed in $X$, an open cover for $Y$ is always of that type, because you can consider $X\setminus Y$, which is open in $X$.

$\endgroup$
  • $\begingroup$ @Masacroso Sorry, typo. $\endgroup$ – egreg Nov 16 '16 at 20:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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