Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $(X,\mathscr T)$ be a topological space. Given $A\subseteq X$, we say that $x$ is a complete limit point of $A$ if for every nbhd $N$ of $x$, $|N\cap A|=|A|$. I want to prove

Suppose $(X,\mathscr T)$ is compact. Then every infinite subset $A$ of $X$ has a complete limit point in $X$.

PROOF Assume the contrary. Then there exists an infinite subset $A$ of $X$ such that for any $x\in X$, there is a nbhd $N_x$ such that $|N_x\cap A|<|A|$. The collection $\{N_x:x\in X\}$ covers $X$. By compactness, there exists a finite subcover $\{N_{x_1},\ldots, N_{x_n}\}$. Then $A=\bigcup_{i=1}^n A\cap N_{x_i}$. But this is impossible, since $A$ is infinite, and $\bigcup_{i=1}^n A\cap N_{x_i}$ is a finite union of sets strictly smaller than $A$.

Is this correct? The problem is in "Introductory Real Analysis" by Kolmogorov.

Also, is a proof of the converse too complicated? Kolmogorov gives a reference as "P.S. Alexandroff op.sit. pp. 250-251; J.L. Kelley op.sit. pp. 163-164" but I have no idea what this means. I am honestly clueless on how to approach it. Could you provide hints? I wouldn't mind a complete answer, if you think it is requires something too advanced.

share|cite|improve this question
Kelly offers it as an exercise, with an extended hint: "If $X$ is not compact choose an open cover $A$ with no finite subcover such that the cardinal number $c$ of $A$ is as small as possible. Let $C$ be a well-ordered set of cardinal $c$ such that the set of predecessors of each member has a cardinal less than $c$ (It is shown in the appendix that $c$ is such a set.) Let $f$ be a one-to-one map of $C$ onto $A$. Then for each member $b$ of $C$ the union $\cup\{ f(a):a<b\}$ does not cover $X$ and, in fact, the complement of this union must have cardinal number at least as great as $c$..." – David Mitra Jul 4 '13 at 1:45
"...It is therefore possible to choose $x_b$ from the complement such that $x_a\ne x_b$ for $a<b$. Consider the set of all $x_b$." – David Mitra Jul 4 '13 at 1:46
@DavidMitra Thank you. Why not put it as an answer? Now someone posted the same! =) – Pedro Tamaroff Jul 4 '13 at 3:48
I'm confused as to why $|N_{x_i}| < |A|$. Can you explain this? – Ink Jul 4 '13 at 5:25
@Ink Yes, it seems the last line is wrong. However, it is easy to repair. Note that the union of the sets $N_{x_1}\cap A,\dots,N_{x_n}\cap A$ is $A$; thus, by the pidgeonhole principle, $|N_{x_i}\cap A|=|A|$ for some $i$. – Karl Kronenfeld Jul 4 '13 at 13:51
up vote 7 down vote accepted

Your proof is correct. For the converse, suppose $X$ is not compact. Let $\mathcal{O}$ be an open cover of $X$ of minimal cardinality [such that no finite subcover exists], and let $\{O_\alpha\}_{\alpha < |\mathcal{O}|}$ be an enumeration of $\mathcal{O}$. Construct a set $A$ as follows:

  • let $x_\alpha \in X$ be any element not covered by any of the $O_\beta$ for $\beta < \alpha$. Such $x_\alpha$ exists or else $\{O_\beta\}_{\beta < \alpha}$ would be an open cover, contradicting the minimality of $|\mathcal{O}|$.
  • let $A = \{x_\alpha : \alpha < |\mathcal{O}|\}$

We'll show that $A$ has no complete limit point. Indeed, for any $x \in X$ there is some $\alpha$ such that $O_\alpha$ is a neighbourhood of $x$, but $O_\alpha$ avoids all the $x_\beta$ for $\beta > \alpha$, and so:

$$\left |A \cap O_\alpha\right | \leq \left|\{x_\gamma : \gamma \leq \alpha\}\right| = |\alpha| < |\mathcal{O}| = |A|$$

share|cite|improve this answer
Thanks, fixed... – Amit Kumar Gupta Jul 4 '13 at 5:15
$\{X\}$ is an open cover of minimal cardinality... Maybe you want to require that it doesn't have a finite subcover? – Asaf Karagila Jul 4 '13 at 19:15
@AsafKaragila Yes, he means the same as David. =) – Pedro Tamaroff Jul 4 '13 at 19:23
How do you know, that $A$ is an infinite set? – el_tenedor Jan 4 at 16:34

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.