For a metric space , the following are equivalent:

  1. is Compact
  2. is Sequentially compact
  3. is Totally bounded and complete

Proof(s)

Suppose is compact. Suppose it is not sequentially compact. Find a sequence which doesn’t have a convergent subsequence. Suppose that there is some s.t. every nbd of has infinitely many points of the sequence. Now there must be a subsequence of converging to - contradiction. Hence, for every find open nbd of which contains at most finitely many points in the sequence. Then are an open cover of . But is compact, so find finite s.t. . Hence, this union contains at most finitely many points of the sequence, but there are infinitely many of them in - contradiction.

Suppose is sequentially compact. Take any Cauchy sequence. It has a convergent subsequence. But that convergent subsequence will bind our Cauchy sequence, so the Cauchy sequence converges, hence is complete. Now suppose is not totally bounded. Find an s.t. every -net is infinite. Pick any point and keep picking (always possible because finitely many points can’t cover ). Now find a convergent subsequence of this but the sequence is not Cauchy! And a convergent sequence is always Cauchy! Contradiction.

Suppose is totally bounded and complete. Pick a sequence in . Let . For any let and find a finite -net. One of the balls in it will contain infinitely many terms of the sequence , so let be this infinite set contained in this ball. Hence define . Then the sequence is Cauchy (for any , the sequence is contained in a ball of radius ) hence convergent.

Suppose is sequentially compact. Because of we have already proved that is totally bounded and complete. Let be an open cover of . Use Lebesgue number lemma to find a Find a finite -net. By definition of , find for each in the -net. Then cover . Hence is compact.