14
$\begingroup$

Is there an interval of the form $[\sigma,\tau]$ in the lattice of topologies on some set $X$ such that $|[\sigma,\tau]| = \aleph_0$? In other words, do there exist two topologies $\sigma$ and $\tau$ on $X$ such that there are a countably infinite number of topologies on $X$ that are finer than $\sigma$ and coarser than $\tau$?

I can find intervals like this of size $n$ for every finite $n > 0$, and intervals of size $\mathfrak{c}$. But I don't see how to obtain closed intervals of any intermediate size. I have a proof (well, an idea for a proof) that if $|[\sigma,\tau]| = \aleph_0$, then $\sigma$ cannot be a Hausdorff topology. Other than this, though, I can't seem to say much of anything about this question.

$\endgroup$
2
  • $\begingroup$ Doesn't it work if we take $\mathbb{N}$ with the (highly non-Hausdorff) topologies $\sigma = \{[2n,+\infty) : n\in\mathbb{N}\}$ and $\tau = \{[n,+\infty) : n\in\mathbb{N}\}$, or am I missing something ? $\endgroup$ Commented Jan 13, 2023 at 8:31
  • 1
    $\begingroup$ Ah, of course not, I misunderstood the use of "interval" in this context, assuming that it should be a totally ordered set. $\endgroup$ Commented Jan 13, 2023 at 8:54

1 Answer 1

7
$\begingroup$

The special case where $\sigma=\{\emptyset,X\}$ is the trivial topology is easy to resolve. In this case, if $\tau$ is finite, then the interval $[\sigma,\tau]$ is finite. If $\tau$ is infinite, then there must exist an $\omega$-chain $T_0\subsetneq T_1\subsetneq T_2\subsetneq\cdots$ of nonempty, proper open sets in $\tau$ or a dual $\omega$-chain of nonempty, proper open sets in $\tau$. Assuming the former, for any nonempty subset $I\subseteq \omega$ let ${\mathcal B}_I=\{T_i\;|\;i\in I\}$ be a basis for a subtopology of $\tau$. Since $I\neq J$ implies that ${\mathcal B}_I$ and ${\mathcal B}_J$ generate different subtopologies of $\tau$, we get at least $\mathfrak c$ topologies in the interval $[\sigma,\tau]$.

$\endgroup$
4
  • $\begingroup$ Does this really work? If $I = \{1\}$ and $J = \{0,1\}$, don't ${\mathcal{B}}_I$ and ${\mathcal{B}}_J$ both generate $T_1$? $\endgroup$
    – Anonymous
    Commented Jan 13, 2023 at 14:04
  • 1
    $\begingroup$ @Anonymous: The topology generated by your ${\mathcal B}_I$ is $\{\emptyset, T_1, X\}$. The topology generated by your ${\mathcal B}_J$ is $\{\emptyset, T_0, T_1, X\}$. These are different. $\endgroup$ Commented Jan 13, 2023 at 14:15
  • $\begingroup$ Oh, right. I misunderstood what you were saying. Thanks. $\endgroup$
    – Anonymous
    Commented Jan 13, 2023 at 14:27
  • $\begingroup$ Interesting -- thanks! I'm starting to suspect that the answer to my question might be negative. $\endgroup$
    – Will Brian
    Commented Jan 13, 2023 at 17:41

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .