3
$\begingroup$

Is it possible to have a topological space having all compact subsets closed, but the space itselt is not Hausdorff?

I couldn't find any counterexample. I tried $\mathbb R$ with cofinite, point inclusion and point exclusion topologies, but these didn't work.

Any suggestion in this context will be helpful for me.

$\endgroup$
  • 2
    $\begingroup$ The cocountable topology on $\mathbb{R}$ will work $\endgroup$ – tfw cant into math Oct 18 '13 at 5:53
2
$\begingroup$

A space in which all compact sets are closed is known as a $KC$ space; $T_B$ space is another term that has been used.

A standard example of a $KC$ space that is not Hausdorff is the one-point compactification $Q^*$ of the space of rational numbers with the usual Euclidean topology. $\Bbb Q$ is not locally compact, so $\Bbb Q^*$ is not Hausdorff. Let $p$ be the point at infinity in $\Bbb Q^*$, and suppose that $K\subseteq\Bbb Q^*$ is compact. If $p\notin K$, then $K$ is a compact subset of $\Bbb Q$ with the usual topology, so $K$ is closed and compact in $\Bbb Q$, $\Bbb Q^*\setminus K$ is open in $\Bbb Q^*$, and $K$ is therefore closed in $\Bbb Q^*$.

Suppose, then, that $p\in K$. If $K$ is not closed, there is some $x\in(\operatorname{cl}_{\Bbb Q^*}K)\setminus K$. Clearly $x\in\Bbb Q$, and $\Bbb Q$ is open in $\Bbb Q^*$, so $x\in\big(\operatorname{cl}_{\Bbb Q}(K\cap\Bbb Q)\big)\setminus K$, and $K\setminus\{p\}$ is therefore not closed in $\Bbb Q$. There is therefore a sequence $\langle x_n:n\in\Bbb N\rangle$ of distinct points of $K\setminus\{p\}$ converging in $\Bbb Q$ to some $x\in\Bbb Q\setminus K$. Let $C=\{x\}\cup\{x_n:n\in\Bbb N\}$; then $C$ is a closed, compact subset of $\Bbb Q$, so $\Bbb Q^*\setminus C$ is an open nbhd of $p$ in $\Bbb Q^*$. Let $D=\{x_n:n\in\Bbb N\}$; $D$ is a discrete set in $\Bbb Q$, so we can find open sets $U_n$ in $\Bbb Q$ such that $U_n\cap D=\{x_n\}$ for each $n\in\Bbb N$. The sets $U_n$ are also open in $\Bbb Q^*$, so $\{U_n:n\in\Bbb N\}\cup\{\Bbb Q^*\setminus C\}$ is an open cover of $K$ with no finite subcover, contradicting the hypothesis that $K$ is compact. Thus, $K$ must be closed, and $\Bbb Q^*$ is $KC$.

$\endgroup$
2
$\begingroup$

Another example of a $KC$ space is the real line with the cocountable topology, where the open sets are just the complements of countable sets. An infinite set $A$ contains a countable subset $a_1, a_2, ... $. The cover $U_1, U_2, ...$ where $U_n=\Bbb R-\{a_n,a_{n+1},...\}$, is then an open cover without a finite subcover. This means that $A$ is not compact. In other words, the compact sets in this topology are just the finite ones, so they are closed.

Note that a $KC$ space has unique sequential limits, i.e. a sequence that converges has a unique limit. In a first countable space it can then be shown that the space must be Hausdorff.

$\endgroup$

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.