compactness

References

Summary

Definitions

Ultrafilters: Compactness

A space is compact if and only if every ultrafilter on it converges to at least one point.

Intuition

Really, every ultrafilter will always converge to something — the thing is, it just might not be in your space; it might be a kind of infinite point. The ultrafilter definition of compactness then says that no limits escape to infinity

Ultrafilter Lemma

The problem is, how do we know a space even has an ultrafilter on it? The ultrafilter lemma tells you that every space does, but it's equivalent to the axiom of choice.

Theorems

Continuous Maps Preserve Compactness

Let f : A → B be continuous. Then if K ⊆ A is compact, so is f(K).

Take an open cover f(K) ⊆ ⋃αUα. Then pulling everything back, we have K ⊆ f−1f(K) ⊆ f−1(⋃αUα) = ⋃αf−1(Uα). We get a finite subcover, since K is compact, and then push that forward: f(K) ⊆ f(⋃if−1(Ui)) = ff−1(⋃iUi) ⊆ ⋃iUi, getting a finite subcover for f(K).

Tychonoff's Theorem

Let {Cα} be an arbitrary collection of compact spaces. Then ∏αCα is compact

Set Theory

Tychonoff's theorem is equivalent to the axiom of choice

Inside compact: Closed ⊆ Compact

T2: Compact ⊆ Closed

Counterexamples

  1. Zariski Topology

    Consider the Zariski topology on ℝ: the only closed sets are finite points and the whole space.

    Every subset is compact

    You can then just pick one set for each of these missing points, and that gives you a finite subcover.

    Now the set (0,1) is compact but not closed.

Maps: Compact → T2 are Closed

Let f : A → B map from a compact space to a Hausdorff one. Then f is closed

Let C ⊆ A be closed. Then it is also compact, and so it gets mapped to a compact subset, which is then closed, since B is T2.

Continuous Bijections: Compact → T2 are Homeos

Maps: Compact → T2 are Closed tells us that our morphism is closed. We can use this to show that f−1 is continuous:

Let C ⊆ A be closed. Then (f−1)−1(C) = f(C) is closed.