A Hausdorff topological space $X$ is called strongly zero-dimensional whenever for every closed subset $A$ of $X$ and every open subset $U$ of $X$ such that $A \subseteq U$, there exists a clopen subset $V$ of $X$ such that $A \subseteq V \subseteq U$.

Now, let $X'$ be a Hausdorff topological space such that for every open subset $O$ of $X'$ and every closed subset $C$ of $X'$ such that $O \subseteq C$, there exists a clopen subset $V$ of $X'$ such that $O \subseteq V \subseteq C$. In this case, we say $X'$ is a $*$-space.

It is clear that a $*$-space is strongly zero-dimensional.

I am trying to answer the following questions:

1) Is it true that a strongly zero-dimensional is $*$-space? If not, is there an additional assumption $A $ on $X'$ such that $A $ + strongly zero-dimensional implies $X'$ is $*$-space?

2) Are $*$-spaces famous? ( For example, are $*$-spaces well-known cases of topological spaces that I do not know)

  • 1
    $\begingroup$ Are you the one who asked this question earlier? Because what is called "property $*$" there is what is called "strongly $0$-dimensional" here, and what is called a $*$-space here implies what is called "extremally disconnected" there, which is insanely confusing. Where did you get this terminology? At any rate, you should check the examples from Steen & Seebach which I mentioned in the comments there, maybe one of them will help in some way, or at least, clarify the question. $\endgroup$
    – Gro-Tsen
    Mar 26 '17 at 17:27

(Having posted this, I saw that all of it is in the comment by Gro-Tsen)

Taking, in the definition of *-space, $C=\text{closure of $O$}$ shows that closure of an open set must be clopen. This is clearly also sufficient. So *-spaces are exactly extremally disconnected Hausdorff spaces - those with closures of open sets clopen. For an example of a strongly zero-dimensional space which is not extremally disconnected, take the Stone space of any incomplete Boolean algebra, e. g. one-point compactification $S\cup\{\infty\}$ of an infinite discrete space $S$. In there, take $O\subset S$ any infinite subset with infinite complement, and $C$ its closure $O\cup\{\infty\}$. Then $C$ is not clopen (the only clopens of $S\cup\{\infty\}$ are finite subsets of $S$ and their complements), so there is obviously no clopen $V$ with $O\subseteq V\subseteq C$.

  • $\begingroup$ It's not entirely in my comment, because I hadn't noticed that "extremally disconnected" was also sufficient for what OP calls being a $*$-space. $\endgroup$
    – Gro-Tsen
    Mar 26 '17 at 17:40

What you call a strongly zero-dimensional space is commonly known as an ultranormal space. An ultranormal space is a Hausdorff space where for every pair of disjoint closed sets $C,D$ there is a clopen set $Z$ with $C⊆Z⊆D^{c}⊆Z⊆D^c$. A strongly zero-dimensional space is a space whose Stone-Cech compactification is zero-dimensional. The ultranormal spaces are precisely the strongly zero-dimensional normal spaces. See my expository for more details on these discussions. So it is true that the $*$-spaces are called extremally disconnected spaces. Finally, there are extremally disconnected spaces which are Hausdorff but not normal (nor even regular) such as this one. In fact, the absolute of a normal space is always completely regular and extremally disconnected but not necessarily normal.

  • $\begingroup$ The π-Base link does not work with me - says "Loading..." indefinitely. $\endgroup$ Mar 26 '17 at 18:50
  • $\begingroup$ The pi-base link works for me on chrome. I wonder what may be going on with the link for others? $\endgroup$ Mar 28 '17 at 4:14
  • $\begingroup$ Weird. I am on chrome too; when going to the front page, there is an announcement There is a major data cleanup and standardization effort underway on Github. Data will be in-flux while that effort is in progress. If I click any link there, it says the same Loading... without any further results $\endgroup$ Mar 28 '17 at 6:09
  • $\begingroup$ I don't understand the containment $Z \subseteq D^c \subseteq Z$ … do you mean $Z = D^c$, or is one of those $Z$'s supposed to be something else? Reading the introduction to your paper suggests that maybe you just meant $C \subseteq Z \subseteq D^c$. $\endgroup$
    – LSpice
    Apr 12 '18 at 19:14

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.