Let $‎\langle‎ ‎‎X_i,\mathcal{T}_i \rangle_{i\in I}$ be a family of topological spaces. Consider $X=\prod_{i\in I} X_i$ with product topology.

Question. Is there a topological property that holds in $X$ when $|I|=\aleph_{\beta}$, but does not hold whenever $|I|=\aleph_{\alpha}$, for some $0<\alpha<\beta$ ?

Any reference or helpful comment will be appreciated.

  • $\begingroup$ Well, if $X$ is any separable space with at least 2 points, then $\prod_IX$ is separable iff $I$ is countable - does this count? $\endgroup$ Nov 11, 2015 at 1:56
  • 5
    $\begingroup$ @Noah, the product of $\mathfrak{c}$ many separable spaces is separable. That's Hewitt–Marczewski–Pondiczery theorem. $\endgroup$ Nov 11, 2015 at 2:03
  • 1
    $\begingroup$ If $X_i$ are all metrizable and have at least two points, then the property, not metrizable, holds for $\beta =\aleph_1$, but not for any cardinal strictly less. I doubt this is what you had in mind. $\endgroup$ Nov 11, 2015 at 3:26
  • 1
    $\begingroup$ I think that the concept of cardinal functions in topology might provide examples of what you are looking for. In the eponymous book by Juhász (readily available online) there is a whole chapter devoted to the topic of their behaviour with respect to products. $\endgroup$
    – dalry
    Nov 11, 2015 at 6:19
  • 3
    $\begingroup$ Addendum to the comment by @KarolSzumiło : The product of $\mathfrak c^+$ copies of a two-point discrete space is not separable. So separability is an example of what the OP asked about, with the "break" occurring at the step from $\mathfrak c$ to $\mathfrak c^+$. $\endgroup$ Nov 11, 2015 at 11:49

2 Answers 2


The splitting number $\mathfrak{s}$ is the minimal cardinal $\kappa$ such that if $\mathcal{A} \subseteq \mathcal{P}(\omega)$ has size $< \kappa$, there exists some $X \subseteq \omega$ such that for all $A \in \mathcal{A}$, $X \setminus A$ is finite or $X \cap A$ is finite.

Then if all $X_i$ are sequentially compact Tychonoff spaces, and $|I| <\mathfrak{s}$ then $\prod_{i \in I} X_i$ is also sequentially compact. And $\{0,1\}^{\mathfrak{s}}$ is not sequentially compact. What $\aleph_\alpha$ equals $\mathfrak{s}$ cannot be said in ZFC. It could be $\alpha = 1$ (under CH, e.g.) or much bigger, depending on the size of the continuum. See van Douwen's paper in the Handbook of Set Theoretic Topology on cardinal invariants of the continuum.


The following is Example 11.8 of Hodel´s article in the Handbook of Set-Theoretic Topology, which provides several examples in the vein of Karol and Andreas's comment:

For the Cantor cube $X=2^\kappa$, we have:

  • $\phi(X)=\kappa$ where $\phi$ is any of the cardinal functions: spread, hereditary Lindelof degree, hereditary density, net weight, weight, $\pi$-weight, hereditary $\pi$-weight, point separating weight, diagonal degree, tightness, $\pi$-character, hereditary $\pi$-character, pseudo-character, character.
  • $d(X)=\log \kappa$ (this is Karol-Andreas's example).

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.