# Homeomorphic open sets and homogeneity

If $(X,\tau)$ is a $T_2$-space such that all non-empty open sets are homeomorphic (with the subspace topology) to $X$, is $(X,\tau)$ necessarily homogeneous?

• Don't forget to link preceding relevant questions of yours like mathoverflow.net/questions/300253/… – Pietro Majer May 24 '18 at 7:44
• Related: In the Cantor set all clopen sets are homeomorphic (I believe) and it is not homogeneous. – HenrikRüping May 24 '18 at 8:17
• @HenrikRüping Why do say that the Cantor set is not homogeneous? It is, after all, the product of countably many $2$-point discrete spaces. Isn't a product of homogeneous spaces homogeneous? – bof May 24 '18 at 8:28
• @bof: Thanks of course it is homogeneous.There are points which are endpoints of some removed intervals and points which are not. I was mistakenly thinking that a homeomorphism can't map one of the first kind to one of the second kind. However that idea at least shows that there are no order preserving (thinking of the Cantor set as a subset of the real line) homeomorphisms that map a point of the first kind to one of the second kind. – HenrikRüping May 24 '18 at 8:34
• @RamirodelaVega And if $X$ is compact and has exactly two non-homeomorphic "types", $X$ is homogeneous. – Henno Brandsma May 25 '18 at 9:34

But it's not hard to prove that a zero-dimensional, first countable space of diversity one is homogeneous (it's also in that paper, but the proof idea is classical); examples of such spaces are $\mathbb{Q}$ and $\mathbb{P}$ (the irrationals). Any compact space of diversity two is homogeneous (reference in the linked paper) and a compact metric space of diversity two (diversity one is impossible for a compact space) is homeomorphic to the Cantor set. (the two types being clopen = Cantor set, and open, non-clopen = Cantor set minus a point).