This question was about spaces in which all non-empty open sets "look alike".

Now I am interested in the opposite: Is there a $T_2$-space $(X,\tau)$ with $|X|>1$ such that whenever $U\neq V$ are open subsets of $X$, they are not homeomorphic when endowed with the subspace topology?

  • $\begingroup$ Re: close, is there an easy example or an easy argument for a negative answer that I missed? $\endgroup$ – Dominic van der Zypen Nov 1 '18 at 10:27
  • $\begingroup$ I seem to recall that this is a problem which appeared at the Vojtěch Jarník International Mathematical Competition held in Ostrava something like 10+ years ago. $\endgroup$ – Tomasz Kania Nov 1 '18 at 10:51
  • 1
    $\begingroup$ I wonder if there is a subspace of $\mathbb R$ with that property. $\endgroup$ – bof Nov 1 '18 at 10:59

There is an example in this paper, A method for constructing ordered continua, of an ordered continuum in which no two intervals are homeomorphic.

Because an open set is a union of a disjoint family of intervals a homeomorphism between two open sets it has to map constituents intervals to constituent intervals; as that can only be done by the identity we see that homeomorphic open sets are equal.

And to answer @bof's question: yes there is also a subset of $\mathbb{R}$ with this property. To see that enumerate the set of triples $\langle f,A,B\rangle$, where $A$ and $B$ are disjoint uncountable $G_\delta$-sets and $f:A\to B$ is a homeomorphism as $\bigl<\langle f_\alpha,A_\alpha,B_\alpha\rangle:\alpha<\mathfrak{c}\bigr>$. Recursively, using that the sets $A_\alpha$ and $B_\alpha$ have cardinality $\mathfrak{c}$, choose $x_\alpha\in A_\alpha$ such that $x_\alpha$ and $f_\alpha(x_\alpha)$ are both not in the union $\{x_\beta:\beta<\alpha\}\cup\{f_\beta(x_\beta):\beta<\alpha\}$. In the end let $X=\{x_\alpha<\alpha<\mathfrak{c}\}$. Claim: $X$ is as required. Assume $f:U\to V$ is a homeomorphism between open subsets and assume $f(x)\neq x$ for some $x$. Pick an open interval $I$ around $x$ such that $f[I\cap X]\cap I=\emptyset$. Apply Lavrentieff's theorem to find $G_\delta$-sets $A$ around $I\cap X$ and $B$ around $f[I\cap X]$ and a homeomorphism $\tilde f:A\to B$ that extends $f$. We can assume $A\subseteq I$, so $A\cap B=\emptyset$. Then $\langle f,A,B\rangle = \langle f_\alpha,A_\alpha,B_\alpha\rangle$ for some $\alpha$ and we find that $x_\alpha\in X$ but $f(x_\alpha)=f_\alpha(x_\alpha)\notin X$. This contradiction shows that $f$ must be the identity and hence that $U=V$.

| cite | improve this answer | |

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.