5
$\begingroup$

This is a follow up to my previous question (unfortunately closed as a duplicate). There the problem was to turn the Moore plane into a normal space by adding a single point. Brian M. Scott gave an answer to this specific problem years ago.

This got me thinking about a more general question: Can you add one point to any completely regular not normal space $X$ to obtain a normal space $Y$ of which $X$ is a subspace?

Note that for me a completely regular spaces and normal spaces are Hausdorff. The restriction to completely regular spaces is natural since every subspace of a completely regular space is itself completely regular, and normal spaces are completely regular. Since $Y$ is supposed to be normal, the subspace $X$ must be completely regular.

I know that if $X$ is locally compact you can take $Y$ to be the one-point compactification. However if $X$ is not locally compact the one-point compactification fails to be Hausdorff.

$\endgroup$
  • $\begingroup$ Maybe add a point $p$ such that for all closed sets $A$ in $X$, $p$ is in its closure in $Y =X \cup \{p\}$. This would "kill" all non-normality showing pairs (which then cannot be closed and disjoint any more). ? Just an idea that might or might not work. $\endgroup$ – Henno Brandsma May 5 '17 at 18:42
  • $\begingroup$ @HennoBrandsma Wouldn't that mean that all singletons from $X$ are no longer closed in $Y$? $\endgroup$ – Alice Munro May 5 '17 at 21:22
  • $\begingroup$ No, only add points to all counterexamples to normality, and singletons are never part of those, by complete regularity of $X$ $\endgroup$ – Henno Brandsma May 5 '17 at 21:55
  • $\begingroup$ Brian M Scott did a similar thing: all counterexamples to normality came from the $x$-axis, so he made the new point a 1-point that would be a limit point to all those sets. You had a similar idea expressed by Ravski, as a comment to your new duplicate question, IIRC. $\endgroup$ – Henno Brandsma May 5 '17 at 21:57
5
$\begingroup$

In general it is not possible to add a point to a Tychonoff space to get a normal space. Let $X= {\mathbb N}^\kappa$ where $\kappa$ is uncountable and where the topology is the product topology, that is, a base consists of products $\prod U_\lambda$ where finitely many sets $U_\lambda$ are single points and all others are all of $\mathbb N$. It is known that $X$ is not normal. Each basic open set is closed in $X$ and homeomorphic to $X$ (and therefore not normal). Now let $Y = X \cup \{p\}$ be a Hausdorff space, where $X$ has the subspace topology. To show that $Y$ is not normal, it is enough to find a closed subset $K$ which is not normal. Choose any $x \in X$ and let $U$ and $V$ be disjoint neighborhoods of $x$ and $p$, respectively. Let $K = Y \setminus V$. Then $U,$ and, hence, $K$, contains a basic open set, which is closed and non-normal. Therefore, $K$ is not normal.

$\endgroup$
  • $\begingroup$ Nicely done. I would add that the basic open set that is a subset of $U$ is closed in $Y$ because it is closed in $X$ and is disjoint from the nbhd $V$ of $p.$ $\endgroup$ – DanielWainfleet May 6 '17 at 18:37
  • $\begingroup$ Thanks a lot, this is great! $\endgroup$ – Alice Munro May 26 '17 at 13:32

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.