14
$\begingroup$

"A paracompact space is a topological space in which every open cover admits a locally finite open refinement" is the definition of paracompactness on Wikipedia.

Comparing with the definition of compactness, "a topological space is called compact if each of its open covers has a finite subcover".

In a first understanding, the difference you notice is that, to be compact, the space must have a finite subcover for EACH of its open covers, so every compact space would be paracompact.

I would like to know what is the motivation on that definition, in what way that concept helps in "extending" the notion of compactness and some examples in which paracompactness is an important property; if they are examples from differential topology or functional analysis, even better. Thanks in advance.

$\endgroup$
  • $\begingroup$ If we restrict to a Hausdorff (connected) space $X$, then the exhaustion of $X$ by compact sets is equivalent to $X$ being paracompact and locally compact. Notice that all connected topological manifolds have this property. So $X$ is then the direct limit of compact sets, so each point of $X$ is "eventually" in a compact set. This is what I think of in terms of extending the notion of compactness, but that's just me. $\endgroup$ – Moya Mar 15 '16 at 4:06
  • 10
    $\begingroup$ You need paracompactness to get partitions of unity. You need partitions of unity to do anything. $\endgroup$ – user98602 Mar 15 '16 at 4:16
3
$\begingroup$

As Mike Miller said, paracompactness gives you partitions of unity, and anything you might want to do uses partitions of unity. Let me just give a short list of desirable properties a paracompact smooth manifold enjoys.

One can always define a Riemannian metric, and hence an ordinary distance function (non-paracompact spaces are not metrizable!)

Every real vector bundle is isomorphic to its dual (this is essentially the point above).

Definition of integration of compactly supported forms on an oriented manifold requires paracompactness.

Vector bundles over a manifold can be classified: A vector bundle over the manifold corresponds to the pullback of the tautological bundle over some grassmannian by some map. This does not hold for non-paracompact manifolds.

$\endgroup$
  • 2
    $\begingroup$ Any motivation from "nondifferential" analysis or general topology? Because the ony time i heard this term is during differential geometry courses and have only seen examples that You given. Would be interesting to see more general setting. $\endgroup$ – wroobell Mar 15 '16 at 17:54
  • 1
    $\begingroup$ The vector bundle works for general paracompact spaces. It is a requirement for metrizability as well. $\endgroup$ – Thomas Rot Mar 15 '16 at 19:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.