My M.Sc. student has the following question, that I assume has an answer in the literature, and we are looking for references.

The generalized Cantor space is the space $2^\kappa$, with basic open sets $$ [\sigma] := \{f\in 2^\kappa : \sigma\subseteq f\}, $$ for $\sigma\in 2^{<\kappa}$.

A space is $\kappa$-compact if every open cover has a subcover of cardinality (strictly) smaller than $\kappa$.

For which cardinals $\kappa$ is the generalized Cantor space $2^\kappa$ $\kappa$-compact? We are interested in strong limit cardinals $\kappa$, mainly strongly inaccessible $\kappa$.

We would especially appreciate references.

Update: A related question.

  • $\begingroup$ If my memory serves me right this is strongly compact cardinals (I'd say weakly compact, but I am going to hedge my bets here, and claim my memory says in the weakly compact case it is only true for covers of cardinality $\kappa$). But I've got a mild case of the flu, so there's no reason to trust my memory. $\endgroup$
    – Asaf Karagila
    Mar 7, 2016 at 16:59
  • $\begingroup$ @Asaf: what about this argument by contradiction; for each $\alpha<\kappa$ the basic opens $[\sigma]$ with $\sigma \in 2^{\alpha}$ is an open covering of cardinality less than $\kappa$, so for at least one $\sigma$ the open $[\sigma]$ is not contained in the original open covering. Then by weak compactness there is a cofinal branch made of these $\sigma$'s which cannot be covered by the original open cover. $\endgroup$
    – godelian
    Mar 7, 2016 at 17:14
  • $\begingroup$ @AsafKaragila: Be healthy, we need you here. :) Note that every open cover of $2^\kappa$ is refined by one with basic open sets. We assume $2^{<\kappa}=\kappa$, so we may restrict attention to open covers of cardinality $\kappa$. $\endgroup$ Mar 7, 2016 at 17:36
  • $\begingroup$ @godelian: You mean you are applying the tree property to the tree induced by all these $\sigma$-s? How do you prove the branch is not covered? In any case, I think I see why the tree property suffices, but with a slightly more involved argument. And what about the converse implication, why is it necessary? $\endgroup$ Mar 7, 2016 at 17:44
  • $\begingroup$ Boaz: Looks like something I should have noticed. So I will unhedge my bet, and go with weakly compact. @godelian: I'm afraid I don't have the mental capacity right now to think about this sort of mathematics. $\endgroup$
    – Asaf Karagila
    Mar 7, 2016 at 17:45

1 Answer 1


A cardinal $\kappa$ is weakly compact if and only if $2^{\kappa}$ is $\kappa$-compact and $\kappa$ is strongly compact if and only if $2^{I}$ is $\kappa$-compact for all sets $I$ where $2^{I}$ is given the topology with basis of open sets of the form $[\sigma]$ where $\sigma:J\rightarrow 2$ and $|J|<\kappa$.

To prove the tree property from the Tychonoff theorem characterization of weak compactness, one uses the following argument. Suppose that $T$ is a $\kappa$-tree. Then for each $\alpha<\kappa$, let $T_{\alpha}$ be the $\alpha$-th level of this tree. Then by a compactness argument, one can show that the inverse limit $\varprojlim_{\alpha<\kappa}T_{\alpha}\subseteq\prod_{\alpha<\kappa}T_{\alpha}$ is a non-empty closed subset (the proof of this fact is the same in the ordinary compactness case). However, $\varprojlim_{\alpha<\kappa}T_{\alpha}$ is the set of all $\kappa$-branches of $T$. Therefore $\kappa$ satisfies the tree property.

To prove that every weakly compact cardinal satisfies the $\kappa$-Tychonoff theorem, one could use the tree property and inaccessibility to show that every $\kappa$-filter on a $\kappa$-algebra of sets can be extended to a $\kappa$-ultrafilter and from this fact one could show that weakly compact cardinals satisfy a $\kappa$-version of Alexander's subbase theorem. From Alexander's subbase theorem, one then establishes that weakly compact cardinals satisfy the $\kappa$-Tychonoff theorem.

This result was originally proven in the paper Additions to some Results of Erdos and Tarski by Donald Monk and Dana Scott.

Furthermore, a long list of characterizations of weak compactness including several topological characterizations is given in the book The Theory of Ultrafilters (1974).

  • $\begingroup$ Much appreciated! $\endgroup$ Mar 8, 2016 at 19:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.