4
$\begingroup$

I'm interested in the dual question to: continuous images of open intervals, about surjections onto open intervals.

Namely, if $X$ is a topological space, when can we guarantee that there exists a topological embedding of $X$ into some Euclidean space?

$\endgroup$
7
  • 5
    $\begingroup$ There is ambiguity between the title question and body question. (“Embedding” usually means that $f$ is a homeomorphism $X\to f(X)$ with subspace topology on the latter.) $\endgroup$ Apr 14, 2020 at 13:09
  • $\begingroup$ Another thing which should be clarified is whether $d$ represents finite number or whether products of infinitely many copies are allowed, too. $\endgroup$ Apr 14, 2020 at 13:10
  • 1
    $\begingroup$ @MartinSleziak should be finite. $\endgroup$
    – ABIM
    Apr 14, 2020 at 13:14
  • $\begingroup$ It's fine I'll accept the answer but would be open to other posts (purely for scientific interest) $\endgroup$
    – ABIM
    Apr 14, 2020 at 13:52
  • 4
    $\begingroup$ @New_Topologist_On_The_Block Then I suggest changing the question & title question back to “embedding” so that it matches the answer you accepted. $\endgroup$ Apr 14, 2020 at 14:21

1 Answer 1

10
$\begingroup$

There's an old theorem of Deák that gives an interesting characterization. Given a topological space $X$, define a relation on subsets of $X$ as follows: we write $U \sqsubseteq V$ if and only if $\overline{U} \subseteq V$.

Theorem (Deák): A separable metrizable space is homeomorphic to a subset of $\mathbb R^n$ if and only if its topology has a subbasis generated by $\leq n+1$ collections of open sets, each totally ordered by $\sqsubseteq$.

For example, to generate the topology of $\mathbb R^2$ with $3$ collections of this kind, think of how $3$ families of open half-planes can be used to form a small open triangle around every point of the plane. This theorem appears in

J. Deák, "A new characterization of the class of subspaces of a Euclidean space," Studia Sci. Math. Hungar. 11 (1980), pp. 253-258.

I don't have a link to this paper, but the result is discussed in section 2 of this paper, which contains a few other related things as well. My favorite part of this theorem is that by changing it a little, one arrives at an interesting (to me) conjecture:

Conjecture: A separable metrizable space is homeomorphic to a subset of $\mathbb R^n$ if and only if its topology has a subbasis generated by $\leq n+1$ collections of open sets, each totally ordered by $\subseteq$.

The conjecture is true for $n=1$, but I don't think it's known for larger $n$. (Does the Klein bottle have a subbasis generated by $4$ nested collections of open sets? Even this special case does not seem trivial to me.)

$\endgroup$
1
  • $\begingroup$ Note that the question was originally about the existence of a continuous injection, which is a priori weaker (for non-separable metrizable spaces it's weaker, just considering an uncountable discrete space of at most continuum cardinal). (But the question might be amended retroactively...) $\endgroup$
    – YCor
    Apr 14, 2020 at 14:28

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.