In proving that a locally compact Hausdorff space $X$ is regular, we can consider the one-point compactification $X_\infty$ (this is not necessary, see the answer here, but bear with me). Since $X$ is locally compact Hausdorff, $X_\infty$ is compact Hausdorff. As a result, $X_\infty$ is normal.

Imitating the idea in my proof in the link above and taking into consideration the correction pointed out in the answer, let $A,B\subseteq X$ be two disjoint, closed sets in $X$, and consider $X_\infty$. Since $X_\infty$ is normal, there are disjoint open sets $U,V \subseteq X_\infty$ such that $A\subseteq U$ and $B \subseteq V$. Then considering $X$ as a subset of $X_\infty$, we can take the open sets $U \cap X$ and $V \cap X$ as disjoint, open subsets of $X$ that contain $A$ and $B$, respectively...or so I thought. I see from the answers to this question that this does not succeed in proving that a locally compact Hausdorff space is normal, since this is not true.

So my question is simply: why does the above proof fail?



$A$ and $B$ are closed in $X$. They need not be closed in the compactification $X_\infty$. You could try to fix this by replacing them with their closures in $X_\infty$, but then these need not be disjoint.

  • 3
    $\begingroup$ A specific example: The Tikhonov plank is the one-point compactification of the deleted Tikhonov plank $X$, which is not normal: the sets $H$ and $K$ of this proof of non-normality are closed and disjoint in $X$, but the extra point of $X^*$ is in the closure in $X^*$ of each of them. $\endgroup$ – Brian M. Scott Jan 29 '13 at 21:59
  • 1
    $\begingroup$ I have made that kind of mistake before. Maybe after making it this time I'll remember not to make it the next time! Hopefully. Thanks for the answer. $\endgroup$ – Alex Petzke Jan 30 '13 at 2:00

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.