My question is the following:

Do there exist two non-homeomorphic topological spaces $X$ and $Y$ such that there are embeddings $f : X \hookrightarrow Y$, $g : Y \hookrightarrow X$, with both $f(X)$ dense in $Y$ and $g(Y)$ dense in $X$?

It is easy to see that the answer is positive if one drops the requirement that the image be dense for one of the embeddings. For example, let $X = [0, 1[$, $Y = [0, 1]$, $f : X \rightarrow Y$ the inclusion, $g : Y \rightarrow X$ defined by $g(y) = \frac{1}{2} y$ for any $y \in Y$. It is clear that $f$ and $g$ are embeddings, and $f(X)$ is dense in $Y$, but $g(Y)$ is not dense in $X$.

However, I can't figure out an example of the stronger situation above, and I wouldn't know how to prove that in such a case the two spaces must in fact be homeomorphic.

Thanks in advance.


Here is an example with a natural topology: $X$ is an open unit disk in the plane, and $Y$ is the same disk punctured at the origin. Here $Y$ is obviously embedded in $X$. If one removes one of the radii of $Y$ (i.e. interval connecting the origin to the boundary circle), one obtains an open set homeomorphic to the disk $X$.

| cite | improve this answer | |

Update: Actually not an answer, just related idea.

Note that by composing $f$ and $g$ your question is stronger that being able to find non-trivial embedding of $X$ to itself with dense image or being able to find a nontrivial dense subspace homeomorphic to whole space. To this there exists at least this trivial example: Inclusion of (infinite) subset of the same cardinality as the whole space, with indiscrete topology so every non-empty subspace is dense.

Another example: Any non-trivial subspace (e.g. remove a point) of rational numbers without isolated points is homeomorphic to rational numbers.

Update: So spaces that contain non-trivial dense copy of itself include: any indiscrete space, $\mathbb{Q}$, $\mathbb{R^n}$ (by @user72694) and any topological sum of those.

Actually, $\mathbb{Q}$ is an anti-example to your question in the following sense: For $X = \mathbb{Q}$ there is no $Y, f, g$ such that your condition holds since $\mathbb{Q}$ is not densely embeddable to any its subspace with isolated points and any its subspace without isolated points is homeomorphic to it.

| cite | improve this answer | |
  • $\begingroup$ Of course any answer to my question provides an example for your statement, but I can't see why the converse should hold. If I find such an embedding of X, what should I choose as Y? $\endgroup$ – Luca Bressan Aug 11 '13 at 17:42
  • $\begingroup$ Just take $Y = X$ and the same embedding. $\endgroup$ – user87690 Aug 11 '13 at 17:52
  • $\begingroup$ But then they would be homeomorphic, contrary to the assumption that they are not. $\endgroup$ – Luca Bressan Aug 11 '13 at 17:55
  • $\begingroup$ @Luca: Oh, you're right. My bad. Just take my “answer” as related idea. $\endgroup$ – user87690 Aug 11 '13 at 18:05

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.