All of the famous examples for sequentially indistinguishable topologies on a set $X$ are provided on an uncountable set $X$ (an uncountable set $X$ with discrete and cocountable topology or the $l^1(\mathbb{N})$-space with norm and weak topology, see this thread).

A natural question arises: Does there exist on a countable set $X$ two topologies that have the same convergent sequences? (Note that the countability of $X$ does not has a large influence on the sequentiality of the topology, e.g. the Arens-Fort space is countable and not sequential.) If this is so, can such topologies be explicitely given?

  • 3
    $\begingroup$ Can you have a topology on $\mathbb Z$, other than the discrete topology, where the only convergent sequences are eventually constant? $\endgroup$ Jan 11 '16 at 15:14
  • $\begingroup$ I see, so you mean that we can take the Arens-Fort space in which all subsets are sequentially open or equivalently all convergent sequences are eventually constant. In other words, the family of sequentially open sets which forms a topology is the strictly finer discrete topology. $\endgroup$
    – yada
    Jan 11 '16 at 16:00
  • $\begingroup$ See also this post at math.SE: Example of different topologies with same convergent sequences $\endgroup$ Dec 6 '16 at 3:30

There are plenty of topologies on a countable set for which all convergent sequences are eventually constant.

The most constructive example I know is the Arens-Fort space, given as example 26 in Steen and Seebach's Counterexamples in Topology.

This topology is constructed on the set $\mathbb{N}^2 \cup \{\infty\}$. Every point of $\mathbb{N}^2$ is isolated, and open neighborhoods of $\infty$ are defined to contain all but finitely many points from all but finitely many columns of $\mathbb{N} \times \mathbb{N}$ (it helps to draw a picture).

Obviously no nontrivial sequence converges to a point of $\mathbb{N}^2$. Suppose $(x_n)$ is a sequence in $\mathbb{N}^2$: we will show it does not converge to $\infty$. Either (a) it has points in only finitely many columns, so some neighborhood of $\infty$ misses the whole sequence, or (b) it has points in infinitely many columns, and we can find a infinite subsequence with one point in each column, which shows that some neighborhood of $\infty$ misses infinitely many of the $x_n$.

A little less constructively, any countable subset of the Stone-Cech compactification of the integers will have the property that every convergent sequence is eventually constant. Using this fact, you obtain $2^{2^{\aleph_0}}$ non-homeomorphic countable spaces, none of which have any nontrivial convergent sequences.

  • $\begingroup$ Thanks. I didn't know this property of the Stone-Cech compactification. $\endgroup$
    – yada
    Jan 11 '16 at 16:04
  • $\begingroup$ For a good explanation of why it's true, see Joseph van Name's answer to this question: mathoverflow.net/questions/138161/… $\endgroup$
    – Will Brian
    Jan 11 '16 at 16:06

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.