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

I just need a bit of help clarifying the definition of a compact set.
Let's start with the textbook definition:

A set $S$ is called compact if, whenever it is covered by a collection of open sets $\{G\}$, $S$ is also covered by a finite sub-collection $\{H\}$ of $\{G\}$.

Question: Does $\{H\}$ need to be a proper subset of $\{G\}$? If, for instance, $\{G\}$ is already a finite collection, does that mean $S$ is automatically covered by a finite sub-collection of $\{G\}$? Also, is there any need for the open sets in $\{H\}$ to be bounded sets?

share|cite|improve this question
up vote 16 down vote accepted

As with many statements involving nested quantifiers, it may help to think of this in terms of a game. Suppose you are trying to prove that a certain space $G$ is compact. $G$ is compact if, for every open covering $C$ of $G$, there is a finite subcovering. So the game goes like this:

  1. You say “$G$ is compact.”

  2. Your adversary says “It is not. Here is an open covering $C$.” (The adversary gives you a family of open sets whose union contains $G$.)

  3. You reply “Here is a finite subcovering of $C$.” (You reply with a finite subset of $C$ whose union still contains $G$.)

If you succeed in step 3, you win. If you fail, you lose. (If you're trying to prove that $G$ is not compact, you and the adversary exchange roles.)

If the adversary presents a finite open covering $C$ in step 2, you have an easy countermove in step 3: just hand back $C$ itself, and you win!

But to prove that $G$ is compact you also have to be able to find a countermove for any infinite covering $C$ that the adversary gives you.

Must your finite subcovering be a proper subset of $C$? No. If this were required, the adversary would always be able to win in step 2 by handing you a covering $C$ with only a single element, $C=\{ G \}$. Then the only proper subset you could hand back would be $\lbrace\mathstrut\rbrace$, which is not a covering of $G$, and therefore the would be no nonempty compact sets. That would be silly, so you have to be allowed to hand back $C$ unchanged in step 3.

share|cite|improve this answer
+1: Never seen a game approach to quantifying statements. I like it! – Cameron Buie Jun 20 '12 at 2:10
I didn't invent this. Wikipedia has an article (although it isn't very good). And the idea is crucial to understanding why the strategies of so many games are PSPACE-complete. This is because the central PSPACE-complete problem, analogous to SAT for NP-complete problems, is the satisfiability problem for quantified boolean formulas, and asserting that P1 wins some game can be understood as claiming that there exists a P1 move such that for all P2 moves there exists a P1 move… etc. – MJD Jun 20 '12 at 2:27
Great explanation, Mark! Thank you very much. – Sakif Khan Jun 20 '12 at 23:18
@CameronBuie I forgot to ping you on the reply above. – MJD Jun 21 '12 at 13:48
It's okay. Since it was posted right after mine, it pinged me by default. – Cameron Buie Jun 21 '12 at 15:36

Does $\{H\}$ need to be a proper sub-collection? No. If $\{G\}$ is finite to start with, then $\{G\}$ is a perfectly fine sub-collection. As an example, cover $[0,1]$ by $\{G\}=\{(-1,3/4),(1/2,2)\}$. If $\{G\}$ is infinite, then the sub-collection $\{H\}$ must be a proper subset due to its finiteness.

The open sets in $\{H\}$ do not need to be bounded. For instance, we could cover the interval $[0,\infty)$ by $$ \{G\} = \{ (a,\infty) : a \in \mathbb{R} \} $$

An obvious finite sub-cover is given by $$ \{H\} = \{ (-1,\infty) \} $$

However, $[0,\infty)$ is not compact since there is an open cover which has no finite sub-cover:

$$ \{G\} = \{ (a,a+2) : a \in \mathbb{R} \} $$

share|cite|improve this answer
That clears up quite a few things for me. Thanks! – Sakif Khan Jun 20 '12 at 23:19

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.