Reviewed second countability. A metric space is separable iff it is second countable.

Example 166.1.

Bounded sets need not be totally bounded. In , let be the sequence defined by . for all . Now, the set . Any open ball of radius can contain at most one . Thus, is not totally bounded.

c45e7a

Compactness

Compact totally bounded.
Proof is trivial.

Any closed and bounded set in is sequentially compact (Cf Heine Borel Theorem)

Closed subsets of compact sets are compact.

Compact sequentially compact (Cf Theorem 176.5).
Vasanth’s proof: Suppose is compact and not sequentially compact, that is, there exists a sequence such that does not have any convergent subsequence. We claim that is closed; this should be easy to see. Since is compact, this implies that is compact too. Since does not have any converging subsequence, there exist such that for each . This yields an open cover of which does not have a finite subcover, contradicting our earlier conclusion that must be compact.

If is sequentially compact, then for any open cover there exists such that for any , there exists such that .
This is the Lebesgue covering lemma with ‘compact’ replaced with ‘sequentially compact’ in the hypothesis (note that we haven’t shown these to be equal yet).
Vasanth’s proof: Suppose there does not exist such a . Then for each , there exists such that for all . WLOG, assume . must be contained in for some and . Choose large enough such that and . Then, we have , a contradiction.

Compact complete.
Let be compact. We have shown that this implies is sequentially compact. If is a Cauchy sequence in , then it has a convergent subsequence, and hence must itself converge.

TFAE:

  1. is compact;
  2. is complete and totally bounded.
  3. is sequentially compact;

We have shown: , ,

: Let . If has finitely many distinct elements, then we are done. Otherwise, is the union of finitely many -balls, and at least one such ball must contain infinitely many elements of . Take the closure of this ball. It is totally bounded (since is totally bounded), so we can cover it with finitely many balls of radius , once of which must again contain infinitely many points of . Keep going to obtain a contracting sequence of closed sets, whose intersection must be a singleton since is complete. It is now easy to obtain a subsequence of converging to .

: Let be an open cover of . Since is sequentially compact, there exists such that for every , there is such that . We now claim that there exist such that . Indeed, if this were not true, we would be able to construct a sequence such that for any , contradicting sequential compactness.

Here, we have shown the implications in the opposite directions.