A space $X$ is said to be starcompact if for every open cover $\mathcal U$ of $X$ there exists a finite subset $\mathcal V\subseteq\mathcal U$ such that $\operatorname{St}(\bigcup\mathcal V,\mathcal U)=X$.

I think the Isbell–Mrówka space $\Psi(\mathcal A)$ with $\lvert\mathcal A\rvert=\omega_1$ is not starcompact. But I can't prove it.


1 Answer 1


Enumerate $\mathcal{A}$ as $\langle A_\alpha:\alpha\in\omega_1\rangle$. Enumerate $[\omega_1]^{<\omega}\times\omega$ as $\bigl<\langle F_\alpha,n_\alpha\rangle:\alpha\in\omega_1\bigr>$, with $F_\alpha\subseteq\alpha$ always. Define recursively a sequence of neighourhoods: $U_\alpha$ of $A_\alpha$ such that $U_\alpha\cap(\bigcup\{A_\beta:\beta\in F_\alpha\}\cup n_\alpha)=\emptyset$. Then $\mathcal{U}=\{U_\alpha:\alpha\in\omega_1\}\cup\bigl\{\{n\}:n\in\omega\bigr\}$ is an open cover. If $\mathcal{V}$ is a finite subfamily of $\mathcal{U}$ then there is an $\alpha$ such that $\mathcal{V}\subseteq\{U_\beta:\beta\in F_\alpha\}\cup\bigl\{\{i\}:i\in n_\alpha\bigr\}$. Then $U_\alpha\cap\bigcup\mathcal{V}=\emptyset$, so the point $A_\alpha$ is not in $\operatorname{St}(\bigcup\mathcal{V},\mathcal{U})$.

Addendum (2022-01-30): this proof works for every (M)AD family of any cardinality; the conclusion is that $\Psi(\mathcal{A})$ is never starcompact.

  • $\begingroup$ Does $St(\bigcup\mathcal V,\mathcal U)$ not contain uncountable number of elements of $\Psi(\mathcal A)$? If not, then can we construct $\mathcal U$ in such a way that it will happen? $\endgroup$
    – Nur Alam
    Jan 26 at 10:09
  • $\begingroup$ What are your quantifiers? Do you want it to happen for all $\mathcal{V}$? Or just one finite subfamily of $\mathcal{U}$? $\endgroup$
    – KP Hart
    Jan 26 at 10:22
  • $\begingroup$ Note: this proof does not depend on $\omega_1$; it will work for every (M)AD family. So $\Psi(\mathcal{A})$ is never starcompact. $\endgroup$
    – KP Hart
    Jan 26 at 10:24
  • $\begingroup$ We want it to happen for all $\mathcal V$. $\endgroup$
    – Nur Alam
    Jan 26 at 10:39
  • 1
    $\begingroup$ Not necessarily; it depends on the AD family: if, in our notation, $\{A_n:n\in\omega\}$ is pairwise disjoint and also $A_n\cap A_\alpha=\emptyset$ whenever $n<\omega\le\alpha$ then there will always be $\mathcal{V}$ with a countable star. $\endgroup$
    – KP Hart
    Jan 26 at 11:10

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.