5

Let $\omega_{cf}$ be the countable space $\omega=\{0,1,2,3,\dots\}$ endowed with the cofinite topology $$\tau_{cf}=\{\emptyset\}\cup\{U\subseteq\omega:\omega\setminus U\mbox{ is finite}\}.$$ It is easy to see that the space $\omega_{cf}$ is second-countable, compact and connected, and so is its countable power $\omega_{cf}^\omega$.

Being motivated by the 5th problem of Kihara, I became interested in studying properties of metrizable subspaces of $\omega_{cf}^\omega$. Since $\omega_{cf}$ contains a discrete doubleton $\{0,1\}$, the countable power $\omega_{cf}^\omega$ contains the Cantor cube $\{0,1\}^\omega$ and consequently, contains a topological copy of any zero-dimensional metrizable separable space. What about higher-dimensional metrizable subspaces of $\omega_{cf}^\omega$?

Problem. Is each metrizable (or better Hausdorff) subspace of $\omega_{cf}^\omega$

  1. hereditarily disconnected?
  2. totally disconnected?
  3. zero-dimensional?

Let us recall that a topological space $X$ is

$\bullet$ zero-dimensional if $X$ has a base of the topology consisting of clopen subsets;

$\bullet$ totally disconnected if for any distinct points $x,y\in X$ there exists a clopen set $U\subseteq X$ such that $x\in U$ and $y\notin U$;

$\bullet$ hereditarily disconnected if each nonempty connected subspace of $X$ is a singleton;

$\bullet$ punctiform if each nonempty compact connected subspace of $X$ is a singleton.

Remark. By known Sierpinski Theorem of 1918, a compact connected Hausdorff space cannot be written as the countable union of pairwise disjoint nonempty closed sets. This implies that each compact metrizable subspace of $\omega_{cf}^\omega$ is a singleton, and each Hausdorff subspace of $\omega_{cf}^\omega$ is punctiform. In his comment to this MO-post, @Arno claims that he knows how to embed the Erdős space into $\omega_{cf}^\omega$, which would imply that the answer to the last subquestion (on zero-dimensionality of subspaces of $\omega_{cf}^\omega$) has negative answer.

Taras Banakh
  • 40,791
  • @bof The answer to your first comment is yes. To answer my first question, it suffices to prove that each nonempty connected metrizable subspace of $\omega_{cf}^\omega$ is a singleton. – Taras Banakh Jul 25 '20 at 07:23
  • @bof Concerning your second comment, unfortunately, I am not an expert in false memory syndromes. Yet, I follow the standard terminology from the Engelking's book in which Engelkings writes (in Historical Notes to the section "Various kinds of diconnectedness") that hereditarily disconnected spaces were introduced by Hausdorff in 1914 who called them "totally diconnected spaces", but after Sierpinski (1921) whose introduced totally disconnected spaces (in their modern meaning) the terminology has changed. But this happened long-long ago (by the way, what is your age)? – Taras Banakh Jul 25 '20 at 07:25
  • I too learned that totally disconnected meant that the connected components are singletons, while totally separated meant points can be separated by clopen sets. – shane.orourke Jul 25 '20 at 08:09
  • 1
    @shane.orourke This simply means that in different areas of mathematics, the same mathematical objects have different names (this is contrary to a definition of mathematics as a science that gives the same names to different concepts). In this situation, it is better to follow the standards from the main discipline to which a concept is attributed. In our case this is General Topology where Engelking's "General Topology" is a standard reference. – Taras Banakh Jul 25 '20 at 09:07
  • @bof Wikipedia is a good reference (especially if it is written by a specialist). Unfortunately, in this case we have two rather standard references (Engelking and Willard) using the same terminology in two different senses. And this is bad. – Taras Banakh Jul 25 '20 at 11:30
  • @bof I looked trough internet resources and Willard's meaning of a totally disconnected space is used in https://mathworld.wolfram.com/TotallyDisconnectedSpace.html, https://encyclopediaofmath.org/wiki/Totally-disconnected_space, https://ncatlab.org/nlab/show/totally+disconnected+space, https://encyclopediaofmath.org/wiki/Totally-disconnected_space, https://proofwiki.org/wiki/Definition:Totally_Disconnected_Space On the other hand, Engelking's meaning can be found only in one resource: https://www.emathzone.com/tutorials/general-topology/totally-disconnected-space.html – Taras Banakh Jul 25 '20 at 11:36
  • @bof It is also interesting how (often) topologists use these notions: hereditarily disconnected vs totally disconnected and totally disconnected vs totally separated in their papers. But this is just a terminological question (which is resolved by writing the precise definitions). What about the answer to the ordiginal question: does $\omega_{cf}^\omega$ contain a non-trivial connected metrizable subspace? – Taras Banakh Jul 25 '20 at 11:42
  • @TarasBanakh It's fairly easy to show that $\omega_{\mathrm{cf}}^\omega$ does not contain any connected Polish space. Even though there are examples of connected Polish spaces which decompose into countably many disjoint non-empty closed sets. $\mathfrak E_{\mathrm{c}}\cup {\infty}$ has such a decomposition. – D.S. Lipham Jul 27 '20 at 16:46
  • @D.S.Lipham Then please write this as a (partial) solution. I will vote up :) By the way, it seems (but I did not write the proof to be 100% sure) that any topological embedding of a metrizable space $X$ into $\omega_{cf}^\omega$ extends to a topological embedding of an absolute $F_{\sigma\delta}$-set containing $X$. So, it remains to make one step more (from Polish to $F_{\sigma\delta}$-sets) and the problem will be resolved completely! – Taras Banakh Jul 27 '20 at 17:00
  • @TarasBanakh It seems that I overlooked a small detail, but I'll keep working on it. I would be very surprised if $\omega_{\mathrm{cf}}^\omega$ contains a metrizable connected space, seeing as it's difficult to even construct one that's Hausdorff. – D.S. Lipham Jul 27 '20 at 22:27
  • then again, maybe Arno knows how to embed $\mathfrak E_{\mathrm{c}}\cup {\infty}$... I would love to hear more about how to embed Erdos space – D.S. Lipham Jul 28 '20 at 00:11
  • Please correct Arno to @Arno . Otherwise he can overlook your comment. – Taras Banakh Jul 28 '20 at 04:43

0 Answers0