Let $X$ be a Lindelöf topological space. Does this imply that every $\omega$-cover has a countable subcover which is also an $\omega$-cover? if not, is there an example of a topological Lindelöf space with an $\omega$-cover for which there isn't a countable sub $\omega$-cover?

We say that an open cover $\mathcal{U}$ of $X$ is an $\omega$-cover, if $X \notin \mathcal{U}$ and for any finite set $A \subset X$ there is a $U \in \mathcal{U}$ such that $A \subseteq U$.

Thank you!


In general, the answer is negative. I found an abstract of an article “A new class of spaces with all finite powers Lindelöf” by Natasha May, Santi Spadaro, and Paul Szeptycki. (Topology Appl. 170, 104-118 (2014)):

An $\omega$-cover of a set $X$ is a cover with the following property: every finite subset of $X$ is contained in an element of the cover. An $\iota$-cover is an $\omega$-cover with the additional property that disjoint finite sets are separated by members of the cover (meaning that if $F, G \in [X]^{<\omega}$ are disjoint and $\mathcal{U}$ is an $\iota$-cover then there is $U \in \mathcal{U}$ such that $F \subseteq U$ and $G \cap U = \emptyset$).

A topological space $X$ is called an $\epsilon$-space if every finite power of $X$ is Lindelöf; equivalently, $X$ is an $\epsilon$-space if every open $\omega$-cover has a countable $\omega$-subcover. The authors remark that every space with a countable network is an $\epsilon$-space, and they show that every Hausdorff space with a countable network has the property that every open $\iota$-cover has a countable refinement that is also an $\iota$-cover. Motivated by this, they call a space with the latter property an $\iota$-space.

In the paper under review, the authors investigate a number of issues related to $\epsilon$-spaces and $\iota$-spaces. A ZFC example of a regular $\iota$-space with no countable network is provided. The authors show that the property of being an $\epsilon$-space and the property of being an $\iota$-space are equivalent for regular spaces with a $G_\delta$ diagonal. A consistent example of a hereditarily $\epsilon$-space whose square is not hereditarily Lindelöf is constructed, as well as an absolute (meaning, ZFC) example of a non $D$-space that has a countable open $\iota$-cover. The paper ends with the following questions: is every $\iota$-space a $D$-space? Is every hereditarily $\iota$-space a $D$-space?

So, the answer is positive iff every finite power of the space $X$ is Lindelöf. In particlular, the answer is positive when $X$ has a countable network or is $\sigma$-compact. But there is a classical example of a Lindelöf space which square is not Lindelöf. It is so-called Sorgenfrey arrow $\Bbb L$, that is the real line endowed with the Sorgenfrey topology generated by the base consisting of half-intervals $[a,b)$, $a<b$. In is well known that the space $\Bbb L$ is Lindelöf, but its square contains a closed discrete space $\{(x,-x):x\in\Bbb R\}$ of cardinality $\frak c$.


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.