Let $X$ be a Lindelöf space, and $E \subset X$.

I am interested in the following property.

(1) $E$ is countable $\Leftrightarrow$ $\forall x \in X, \exists U_x$ open set such that $x \in U_x$ and $U_x \cap E$ is countable

Which in case $E$ is closed leads to :

(2) $E$ is countable $\Leftrightarrow$ $\forall x \in E, \exists U_x$ open set such that $x \in U_x$ and $U_x \cap E$ is countable

(by taking $U_x = X \backslash E$ for $x \in X \backslash E$)

For example, (2) can be used to prove that a discrete and closed subspace of a Lindelöf space (like $\mathbb{R}$ or any $\sigma$-compact space) is countable.

Here is the proof for ($\forall x \in X, \exists U_x$ open set such that $x \in U_x$ and $U_x \cap E$ is countable $\Rightarrow E$ is countable) :

We have $X = \cup _{x \in X} U_x$. Because $X$ is Lindelof, we have a countable family $\{x_i\}_{i \in \mathbb{N}}$ such that $X = \cup _{i \in \mathbb{N}} U_{x_i}$. Then $E = X \cap E = \cup _{i \in \mathbb{N}} (U_{x_i} \cap E )$ is a countable union of countable subsets, so it is countable.

My questions are :

  1. Does this property have a name ?

  2. If not, is it linked to some property of Lindelöf spaces (like a countability axiom)

  3. If not, is it an ingredient to prove some known thing related to Lindelöf spaces ?

The property $(1)$ is equivalent to the requirement that if $A\subseteq X$ is uncountable, then there is an $x\in X$ such that $U\cap A$ is uncountable for each open nbhd $U$ of $x$. In other words, each uncountable subset of $X$ has a condensation point in $X$. As you observe, every Lindelöf space has this property; the converse, however, is not true.

Example. Let $X$ be $\omega_2$ with the order topology. For $\alpha\in\omega_2$ let


and let $\mathscr{U}=\{U_\alpha:\alpha\in\omega_2\}$; $\mathscr{U}$ is an open cover of $X$ with no countable subcover. (Indeed, $\mathscr{U}$ has no subcover of cardinality $\omega_1$, either.) Now suppose that $A\subseteq X$ is uncountable; there is a least $\alpha\in\omega_2$ such that $U_\alpha\cap A$ is uncountable. If $\alpha$ were not a condensation point of $A$, there would be a $\beta<\alpha$ such that $(\beta,\alpha]\cap A$ was countable, but then $U_\beta\cap A$ would be uncountable, contradicting the choice of $\alpha$. Thus, $\alpha$ is a condensation point of $A$.

The fact that $\mathscr{U}$ in the example has no subcover of cardinality $\omega_1$ is not an accident.

Theorem. If $X$ has property $(1)$, and every open cover of $X$ has a subcover of cardinality at most $\omega_1$, then $X$ is Lindelöf.

Proof. If $X$ is not Lindelöf, it has an open cover $\mathscr{U}$ with no countable subcover, and we may assume that $|\mathscr{U}|=\omega_1$ and enumerate $\mathscr{U}=\{U_\xi:\xi<\omega_1\}$. For each $\eta<\omega_1$ let $$F_\eta=X\setminus\bigcup_{\xi<\eta}U_\xi\;;$$ each $F_\eta$ is a non-empty closed set, and $\bigcap_{\eta<\omega_1}F_\eta=\varnothing$. For $\eta<\omega_1$ let $x_\eta\in F_\eta$, and let $A=\{x_\eta:\eta<\omega_1\}$. The points $x_\eta$ may not all be distinct, but I claim that $A$ is uncountable.

Suppose not; then there is an $x\in A$ such that $E=\{\eta<\omega_1:x_\eta=x\}$ is uncountable. But for each $\eta\in E$ we have $x=x_\eta\notin\bigcup_{\xi<\eta}U_\xi$, so $$x\notin\bigcup_{\eta\in E}\bigcup_{\xi<\eta}U_\xi=\bigcup\mathscr{U}=X\;,$$ which is impossible.

Now let $x\in X$; clearly $x\in U_\eta$ for some $\eta<\omega_1$. Let $V=X\setminus F_{\eta+1}=\bigcup_{\xi\le\eta}U_\xi$; $V$ is an open nbhd of $x$, and $V\cap A\subseteq\{x_\xi:\xi\le\eta\}$, so $V\cap A$ is countable, and $x$ is not a condensation point of $A$. Thus, $A$ has no condensation point, contradicting the hypothesis that $X$ has property $(1)$. It follows that $X$ must be Lindelöf after all. $\dashv$

I’ve not seen property $(1)$ used much, but a strengthening of it is equivalent to $X$ being hereditarily Lindelöf.

Theorem. A space $X$ is hereditarily Lindelöf if and only if each uncountable subset of $X$ contains a condensation point of itself.

This is not hard to prove, and you can also find a proof in Dan Ma’s Topology Blog.


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.