We say that a $T_2$-space $(X,\tau)$ has homeomorphic open sets if every non-empty open set $U\subseteq X$ endowed with the subspace topology is homeomorphic to $(X,\tau)$.

The rationals with the Euclidean topology are an example of a space with homemorphic open sets, as well as $\{0,1\}^\lambda$, where $\lambda$ is an infinite cardinal and $\{0,1\}$ carries the discrete topology.

Is there for every infinite cardinal $\kappa$ a space of cardinality $\kappa$ with homeomorphic open sets?

Edit. Apologies for falsely claiming that $\{0,1\}^\lambda$ is a space with homeomorphic open sets - Joel David Hamkins' comment below gives an argument refuting my claim.

  • 1
    $\begingroup$ If you remove a point from $2^\omega$, it is no longer compact, and so not homeomorphic to the whole space $2^\omega$. Doesn't this refute your claim about $\{0,1\}^\lambda$? $\endgroup$ May 15, 2018 at 10:45
  • $\begingroup$ You are right - thanks Joel! I'll add this as a note in the question $\endgroup$ May 15, 2018 at 11:27
  • $\begingroup$ Do you have any uncountable examples of this phenomenon? $\endgroup$ May 15, 2018 at 11:44
  • $\begingroup$ I thought I had :-) but now it looks like $\mathbb{Q}$ is the only example I have. It would also be interesting to see another countable example, not homeomorphic to $\mathbb{Q}$. $\endgroup$ May 15, 2018 at 11:47
  • 1
    $\begingroup$ I wonder whether $2^{<\kappa}$ for uncountable ordinals $\kappa$ would be an uncountable example analogous to $\mathbb{Q}$, but I worry that cofinality differences in the order could show up in the homeomorphism types of different open sets. $\endgroup$ May 15, 2018 at 12:04

1 Answer 1


The Baire space $\omega^\omega$ has homeomorphic open sets: it is an immediate consequence of the fact that it is the unique nonempty zero-dimensional Polish space, up to homeomorphism, every compact subset of whose has empty interior (theorem 7.7 in Kechris). In fact we actually have better:

Proposition. For every infinite set $X$ with the discrete topology, $X^\omega$ has homeomorphic open sets.

Proof. Let $U \subseteq X$ be nonempty open. I claim that $U$ is a union of exactly $\kappa$ disjoint cones, where $\kappa = |X|$. To see this, it is enough to show that $U$ is a union of $\leqslant \kappa$ disjoint cones, as a cone is itself a union of $\kappa$ disjoint cones. But if $T \subseteq X^{< \omega}$ is a nonempty tree such that $[T] = U^c$, then $U$ is the disjoint union of all cones $N_s$ where $s \notin T$ but $s_{\restriction |s| - 1} \in T$.

Now, writing $U$ as a union of $\kappa$ disjoint cones, we can get a homeomorphism between $U$ and $X$ by gluing together homeomorphism between these cones and cones of the form $N_{(x)}$, $x \in X$.

So the answer to Dominic's question is positive at least for cardinals of the form $\kappa^\omega$.

Edit: Will Brian remarked, in the comments, that the same argument actually enables to show that the subspace of $X^\omega$ whose elements are eventually constant also had homeomorphic open sets. So for every infinite cardinal, there exists a space with homeomorphic open sets.

  • 1
    $\begingroup$ Every nonempty compact subset of $\omega^\omega$ has nonempty interior? What about singletons? $\endgroup$
    – Nik Weaver
    May 15, 2018 at 12:59
  • 3
    $\begingroup$ @NikWeaver: I think it should say empty interior. $\endgroup$
    – Will Brian
    May 15, 2018 at 13:01
  • 2
    $\begingroup$ However @DominicvanderZypen, maybe you should not accept the answer yet, the question is still interesting as we have no example for a lot of cardinals. It would be nice if other people could look at this question. $\endgroup$ May 15, 2018 at 13:46
  • 5
    $\begingroup$ Doesn't your argument apply equally well to the subspace of $X^\omega$ consisting of eventually constant sequences? It's still true that the closed sets can be expressed as trees, and it's still true that every open set is a union of disjoint cones. But this subspace has cardinality $|X|$, so this version of the argument would work for any infinite cardinal. $\endgroup$
    – Will Brian
    May 15, 2018 at 14:14
  • 2
    $\begingroup$ Of course @WillBrian you are right! That's nice, maybe you should post it as an answer? Or if you prefer, I can add it on my answer. $\endgroup$ May 16, 2018 at 9:57

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.