3
$\begingroup$

I am working on the following hw problem: If we have that $X$ is a compact Hausdorff space, with $\{U_\alpha\}_{\alpha\in A}$, then we can find a finite number of continous functions $f_1,...,f_k$, with $f_i:X\mapsto [0,1]$ such that $f_1(x)+...+f_k(x)=1$ for all $x$, and for each $i$ there exists a $\alpha_i$ such that $\overline{f_i^{-1}((0,1])}\subset U_{\alpha_i}$.

So this are my thoughts at the moment. Since $X$ is Hausdorff and compact it is normal, so we know that Urysohn's lemma applies. Meaning, if I have any two disjoint sets I can find a continous function that is $0$ is one of the sets and $1$ in the other one.

My intuition is telling me to use compactness and find $U_{\alpha_1},..., U_{\alpha_k}$ that cover $X$, and then by the last requirement of the problem we would need $f_i(x)=0$ if $x\notin U_{\alpha_i}$.

The problem is very clear if I can write my space $X$ as the disjoint union of finite closed sets cause then I could just apply Urysohns and get my functions very easily, but I do not see how to apply this lemma using the open cover given. Any hints would be greatly appreciated.

$\endgroup$
  • $\begingroup$ Hi Daniel, I've given a direct answer to the question, which may be helpful for you. $\endgroup$ – Paul May 20 '13 at 7:25
2
$\begingroup$

Such a family of functions is called a partition of unity (where unity = 1, because of the summing to 1 property), and these exists for all paracompact Hausdorff spaces.

For a complete proof, see this, e.g. A compact (Hausdorff) space is paracompact Hausdorff (we can work with finite covers everywhere, but that makes the proof only slightly easier..).

Chapter 2 of the linked document handles the finite case, without going to paracompactness.

$\endgroup$
  • $\begingroup$ Which linked document? $\endgroup$ – Olivier Bégassat May 20 '13 at 6:52
  • $\begingroup$ under "this" in the second paragraph $\endgroup$ – Henno Brandsma May 20 '13 at 6:53
  • $\begingroup$ Indeed, I missed it! $\endgroup$ – Olivier Bégassat May 20 '13 at 6:54
  • $\begingroup$ Excellent. That shrinking lemma was exactly what I needed ! $\endgroup$ – Daniel Montealegre May 20 '13 at 7:07
1
$\begingroup$

Let us recall some important definitions fristly:

A family $\{f_s\}_{s\in S}$ of continuous functions from a space $X$ to the closed unit interval $I$ is called a partition of unity on the space $X$ if $\Sigma_{s\in S}f_s(x)=1$ for every $x\in X$.

A partition of unity $\{f_s\}_{s\in S}$ on a spaceis subordinated to a cover $\mathscr U$ of $X$ if the cover $\{f_s^{-1}((0,1])\}_{s\in S}$ of the space $X$ is a refinement of $\mathscr U$.

Theroem 1: For every $T_1$-space $X$ the following conditions are equivalent:

1) The space $X$ is paracompact;

2) Every open cover of the space $X$ has a partition of unity subordinated to it.

The proof of it can be seen the General Topology by Engelking Page 302.

Note that every compact Hausdorff space is paracompact.

$\endgroup$

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.