While looking at proofs here and there on the existence of a Riemannian metric, I always miss something.

Given a manifold $M$ I would like to give myself an atlas $\{(U_\alpha,\phi_\alpha)\}$ with a countable number of maps. Then on each of the open sets of the atlas I can define my metric $g_i$. Then I know there exists a partition of unity subordinate to $U_\alpha$. Then I define my Riemannian metric as $g = \sum_ig_i\theta_i$.

But How do I get my locally finite countable cover?

  • 1
    $\begingroup$ "Riemannian metric" (not "riemannien metric"). $\endgroup$ – Dietrich Burde Nov 21 '17 at 9:42
  • $\begingroup$ The existence of the smooth partition of unity is guaranteed for a smooth manifold. The property of locally finite is in fact one of the requirement of partition of unity. So when we claimed that we have a partition of unity subordinate so a open cover, it automatically understood that the family of supports of our function $\psi_{\alpha}$ is locally finite. $\endgroup$ – Sou Nov 21 '17 at 9:42
  • $\begingroup$ What I understand is every manifold is supposed paracompact so this gives me the locally finite cover. But how do I get the countability? because the partition of unity is a countable but nothing guarantees that I can say that I have a countable atlas. $\endgroup$ – tomak Nov 21 '17 at 10:07
  • 1
    $\begingroup$ Look math.stackexchange.com/questions/2440003/… $\endgroup$ – Sou Nov 21 '17 at 11:53
  • 1
    $\begingroup$ What is your definition of manifold? Paracompactness alone is not enough to get a countable atlas - consider an uncountable disjoint union of copies of $\mathbb R$. However, I don't think this is a problem for partitions of unity - an uncountable sum can still make sense thanks to the local finiteness. $\endgroup$ – Anthony Carapetis Nov 21 '17 at 12:47

Generally, we suppose that a manifold is paracompact t ensure the existence of partitions of unity.


  • $\begingroup$ I get that we have our partition of unity but how do I get a countable cover so that the $g_i$ are countable? (i.e. there is a countable atlas) $\endgroup$ – tomak Nov 21 '17 at 10:09
  • $\begingroup$ paracompact implies that $\endgroup$ – Tsemo Aristide Nov 21 '17 at 10:11
  • $\begingroup$ I don't see it. How does the fact that for every cover we can extract a locally finite cover imply that we can extract a countable cover? $\endgroup$ – tomak Nov 21 '17 at 10:39

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.