22
$\begingroup$

I'm quite new to topology, and a homework question (which I solved without knowing the answer to this question) got me thinking:

If $X$ is a countable set, and $\tau$ is a topology on it, does it necessarily have a countable basis?

Since $\tau \subset 2^{X}$, it might have an uncountable number of sets in it, but a base can be made of a very small subset of $\tau$, so pure intuition says that the answer is yes, but I couldn't prove it, and now I'm not sure that it's true at all.

Thanks!

$\endgroup$
0
20
$\begingroup$

http://en.wikipedia.org/wiki/Arens%E2%80%93Fort_space

this can be found in "counterexamples in topology" as referenced on the wikipedia page (a book i sometimes wish i had)

$\endgroup$
9
$\begingroup$

Good question! I suspect the answer is no. Here is my suggestion for a counterexample. Let $U$ be a nonprincipal ultrafilter on $\mathbb N$. Then I claim $U\cup\{\emptyset\}$ is a topology on $\mathbb N$, which follows from the axioms of an ultrafilter. Now I suspect that this topology does not have a countable basis, mainly because there are $2^{2^{|\mathbb N|}}$ many ultrafilters, implying you need to make uncountably many choices to specify it. If you could get away with specifying a countable basis, that would only be countably many choices.

Edit: Yoyo just posted a definitive answer.

$\endgroup$
1
  • 1
    $\begingroup$ Indeed, for any free ultrafilter this space has uncountable base. This is well-known. $\endgroup$ – Henno Brandsma May 25 '11 at 6:00
6
$\begingroup$

The Hewitt-Marczewski-Pondiczery theorem implies that $[0,1]^{\mathbb{R}}$ in the product topology has a countable dense subset $D$. This $D$, as a space in its own right (in the subspace topology) has a local base of minimal size $\mathfrak{c}$ at every point, i.e. its character at every point is $\mathfrak{c}$. I show this and more in this answer, where you can find other countable examples based on ultrafilters as well.

$\endgroup$
2
  • $\begingroup$ Your first link is not working for me $\endgroup$ – Alessandro Codenotti Feb 20 '18 at 18:01
  • $\begingroup$ @AlessandroCodenotti search the name of the theorem on Wikipedia instead? $\endgroup$ – Henno Brandsma Feb 20 '18 at 18:18

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.