4
$\begingroup$

I have a sequential, hereditarily Lindelöf topological space $\mathcal{X}$, and some subset $A \subseteq \mathcal{X}$. I am interested in the following properties:

  1. There is some compact set $B$ with $A \subseteq B \subseteq \mathcal{X}$.
  2. Every sequence $(a_n)_{n \in \mathbb{N}}$ in $A$ has a subsequence which converges in $\mathcal{X}$.
  3. For every open cover $\bigcup_{n \in \mathbb{N}} U_n \supseteq \mathcal{X}$ there is some $N \in \mathbb{N}$ with $A \subseteq \bigcup_{n \leq N} U_n$.

I know that each of these implies the next, but do the implications reverse?

(Trying to Google sources is made difficult by mentions of relatively compact dominating; and since I don't wish to assume Hausdorffness, that is not the notion I am looking for.)

$\endgroup$
1
  • $\begingroup$ That's a nice question. It feels like (3) should be weaker than (1), but also that it should quite possibly be equivalent to (1). $\endgroup$
    – Asaf Karagila
    Commented Jun 14, 2023 at 12:35

1 Answer 1

2
$\begingroup$

It seems that Gutik's hedgehog is a required counterexample.

I recall that Gutik's hedgehog is the set $$X=\{(0,0)\}\cup\{(\tfrac1n,0):n\in\mathbb N\}\cup\{(\tfrac1n,\tfrac1{nm}):n,m\in\mathbb N\}$$ endowed with the topology $\tau$ consisting of subsets $U\subseteq X$ satisfying the following two conditions:

(1) if $(\frac1n,0)\in U$ for some $n\in\mathbb N$, then there exists $m\in\mathbb N$ such that $(\frac1n,\frac1{nk})\in U$ for all $k\ge m$;

(2) if $(0,0)\in U$, then there exists $m\in\mathbb N$ such that $(\frac1n,\frac1{nk})\in U$ for all $n\ge m$ and all $k\in\mathbb N$

It is easy to see that Gutik's hedgehog is countable, second-countable, and Hausdorff but not regular.

In Gutik's hedgehog consider the subset $A=\{(0,0)\}\cup\{(\frac 1n,\frac1{nm}):n,m\in\mathbb N\}$.

It is easy to see that every nontrivial sequence in $A$ has a subsequence that converges in $X$ (to a point in the set $\{(0,0)\}\cup\{(\frac1n,0):n\in\mathbb N\}$). On the other hand, no subspace of $X$ containing $A$ is compact. So, the property (2) is satisfied but (1) does not.

The conditions(2) and (3) are equivalent for sequential spaces. This equivalence follows from two lemmas.

Lemma 1. Let $A$ be a subset of a topological space $X$ such that every open cover of $X$ contains a finite subfamily covering $A$. Then every sequence in $A$ has an accumulation point in $X$.

Proof. Assuming that $A$ contains a sequence $(a_n)_{n\in\omega}$ without accumulation points in $X$, for every $x\in X$, we can find a neighborhood $U_x\subseteq X$ of $X$ such that the set $\{n\in\omega:a_n\in U_x\}$ is finite. Then for every finite subfamily $\mathcal F$ of the open cover $\{U_x:x\in X\}$ of $X$ the set $\{n\in\omega:a_n\in \bigcup\mathcal F\}$ is finite and hence $\{a_n:n\in\omega\}\not\subseteq\bigcup\mathcal F$ and $A\not\subseteq \bigcup\mathcal F$. $\quad \square$

Lemma 2. If a sequence $(x_n)_{n\in\omega}$ in a sequential space $X$ has an accumulation point in $X$, then $(x_n)_{n\in\omega}$ has a subsequence that converges to some point of $X$.

Proof. Let $a$ be an accumulation point of the sequence $(x_n)_{n\in\omega}$ in $X$.

For every point $x\in X$ let $I_x$ be the intersection of all neighborhoods of $x$ in $X$. It is easy to see that for every $x\in X$ and $y\in I_x$ we have $I_y\subseteq I_x$.

For a subset $A\subseteq X$, consider the set $\dot A:=\{x\in X:I_x\cap A\ne\emptyset\}$ and observe that $A\subseteq \dot A$.

If for some $x\in X$ the set $\Omega_x:=\{n\in\omega:x_n\in I_x\}$ is infinite, then $(x_n)_{n\in\Omega_x}$ is a required convergent (to $x$) subsequence of the sequence $(x_n)_{n\in\omega}$ and we are done. So, assume that for every $x\in X$ the set $\Omega_x$ is finite. In this case, the set $A=\{x_n:n\in\omega\setminus\Omega_a\}$ is not closed in $X$ as $a\in\overline{A}\setminus A$.

Moreover, the set $\dot A$ is not closed as $A\subseteq \dot A\subseteq\bar{A}$ and $a\notin \dot A$. By the sequentiality of $X$, the set $\dot A$ contains a sequence $(a_k)_{k\in\omega}$ that converges to some point $b\in X\setminus \dot A$. If $a_k\in I_b$ for some $k\in\omega$, then $I_{a_k}\subseteq I_b$ and $a_k\in \dot A$ imply $I_b\cap A\supseteq I_{a_k}\cap A\ne\emptyset$ and hence $b\in \dot A$, which contradicts the choice of $b$. This contradiction implies that $I_b\cap\{a_k:k\in\omega\}=\emptyset$ and hence the set $\{a_k:k\in\omega\}$ is infinite. Since $\{a_k:k\in\omega\}\subseteq \dot A$, for every $k\in\omega$ there exists a number $i_k\in\omega$ such that $x_{i_k}\in I_{a_k}$. We claim that $b$ is the limit of the sequence $\{x_{i_k}:k\in\omega\}$. Indeed, for every open neighborhood $O_b$ of $b$ in $X$, there exists $m\in\omega$ such that $\{a_k:k\ge m\}\subseteq O_b$. Then for every $k\ge m$ we have $x_{i_k}\in I_{a_k}\subseteq O_b$, which means that the sequence $(x_{i_k})_{k\in\omega}$ converges to $b$.

Assuming that the set $F=\{x_{i_k}:k\in\omega\}$ is finite and taking into account that $b\notin \dot A$, we can find a neighborhood $O_b$ such that $O_b\cap F=\emptyset$, which contradicts the convergence of the seqeunce $(x_{i_k})_{k\in\omega}$ to $b$. This contradiction shows that the set $F$ is infinite. Then we can choose an increasing number sequence $(n_k)_{k\in\omega}$ such that $F=\{x_{n_k}\}_{k\in\omega}$ and $x_{n_i}\ne x_{n_j}$ for $i\ne j$. For every $k\in\omega$ find $p_k\in\omega$ such that $n_k=i_{p_k}$. The convergence of the sequence $(x_{i_p})_{p\in\omega}$ to $b$ implies the convergence of the sequence $(x_{n_k})_{k\in\omega}=(x_{i_{p_k}})_{k\in\omega}$ to $b$. Therefore, $(x_{n_k})_{k\in\omega}$ is a required convergent subsequence of the sequence $(x_n)_{n\in\omega}$. $\quad\square$

$\endgroup$
4
  • 1
    $\begingroup$ I imagine you take this hedgehog as your $\mathcal X$, what would your $A$ be? $\endgroup$
    – Wojowu
    Commented Jun 14, 2023 at 21:09
  • $\begingroup$ @Wojowu Thank you for the comment. I added more details to my answer. Is it Ok now? $\endgroup$ Commented Jun 14, 2023 at 21:20
  • 1
    $\begingroup$ If I'm not missing something, the example already separates Properties (1) and (2) - any sequence in $A$ should have a converging subsequence, too. (And the space is a great example in many other ways, thanks!) $\endgroup$
    – Arno
    Commented Jun 15, 2023 at 5:37
  • $\begingroup$ @Arno You are right this example separates (1) and (2). So now the problem is to separate (2) and (3), or prove that they are equivalent (at least under some separation axioms). $\endgroup$ Commented Jun 15, 2023 at 6:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .