Compactness, completeness and total boundedness

Try

On the real line, closed and bounded sets are compact. In a general metric space, those words are insufficient: an infinite discrete space is bounded and closed in itself but not compact. The correct replacement is completeness plus total boundedness. The first supplies missing limits; the second supplies finite approximation and Cauchy subsequences. Each property addresses a different obstruction.

Predict

Predict the fate of a sequence in a complete totally bounded space. Total boundedness extracts a Cauchy subsequence, and completeness gives its limit inside the space. This proves sequential compactness. To recover the open-cover definition of compactness, more work is needed: a cover may contain infinitely many differently sized neighborhoods, and one must show finitely many suffice.

Experiment

Organize the equivalence.

Starting property	Intermediate conclusion	Tool
Compact	Totally bounded	Cover by all equal-radius balls
Complete and totally bounded	Sequentially compact	Cauchy extraction and completeness
Sequentially compact	Compact	Lebesgue number and finite net

The table is a proof dependency map. Do not use an unproved arrow in reverse merely because the final properties are equivalent. Each implication has its own mechanism and hypotheses.

Understand

A compact metric space is totally bounded because the balls of any fixed positive radius form an open cover with a finite subcover. It is sequentially compact as well. If a sequence had no cluster point, each point would have a neighborhood containing only finitely many indexed terms; a finite subcover would then contain only finitely many indices in total, impossible. A cluster point supplies a subsequence by choosing increasing indices in successively smaller balls. A Cauchy sequence with a convergent subsequence converges to the same limit, so compactness also implies completeness.

Now assume sequential compactness. It implies total boundedness because failure would produce an ε-separated sequence with no convergent subsequence. For a given open cover, prove a Lebesgue number exists: otherwise choose x_n whose radius-1/n ball is contained in no cover member. Extract x_(n_k)→x and choose a cover member containing B(x,r). Eventually both d(x_(n_k),x) and 1/n_k are below r/2, so the supposedly offending ball lies in B(x,r), a contradiction.

Apply

Choose a finite net of radius less than that Lebesgue number. Each net ball lies in a cover member, and those finitely many members cover the space. Thus sequential compactness implies open-cover compactness. Combining the implications proves the announced equivalence.

Repair “complete means compact.” R is complete but lacks total boundedness. Repair the opposite shortcut “totally bounded means compact.” The interval (0,1) has finite nets at every scale but is incomplete. These two examples isolate the independent requirements rather than merely presenting unrelated failures.

A continuous image of a compact space is compact by pulling an open cover back to the domain and then pushing the finite selection forward. This argument does not require the image map to be injective.

Continue

Optional depth derives uniform continuity of continuous maps from compact metric domains using the paired-sequence proof from Analysis I. The same reasoning works because sequential compactness supplies a limit and continuity controls both image sequences there.

Save a complete chain from one chosen compactness formulation to another. State whether balls are open and whether centers lie in the space. Next, connectedness will control a different global feature: absence of a separation into two open pieces, rather than the existence of finite covers or convergent subsequences.