0
$\begingroup$

I'm trying to prove the next:

If $C$ is a closed subset of a complete Riemannian manifold $M,$ then for each $p\in M,$ there is a point $p^{'}\in C$ that is as close to $p$ as possible.

So I'd like to prove that there is $p^{'}\in C$ such that $d(p,p^{'})=d(p,C):=\displaystyle\inf_{q\in C}d(p,q).$

I'm stuck proving this. If $C$ were bounded, the Hopf-Rinow theorem would imply the compactness of $C$ and we can get such point utilizing the sequence criteria of compactness.

However, $C$ only is closed. Then, since $(M,d)$ is metric space, then the distance from $p$ to $C$ is a continuous function. If we consider the boundary of $C,$ $d(p,C)$ attains a minimum in such set, but, what about the rest of $C?$

Is there another easier way to proof this?

Any kind of help is thanked in advanced

$\endgroup$
  • $\begingroup$ Pick $q$ in $C$ and consider only the points $p'$ in $C$ with $d(p,p') \leq d(p,q)$. $\endgroup$ – Catalin Zara May 30 '18 at 2:06
  • $\begingroup$ Thanks to answer @CatalinZara. Let $A:=\{p^{'}\in C:d(p,p^{'})\leq d(p,q)\}$ where $q\in C$ and $p\in M$ fixed point. Such set is closed because of continuity of Riemannian metric, and is bounded because $d(p,q)$ is fixed positive number with $q$ fixed, right? Then $A$ is compact for Hopf-Rinow and we have the case mentioned above. But why this is enogh to finish the proof? $\endgroup$ – Squird37 May 30 '18 at 2:19
  • $\begingroup$ Or I don't understand your idea @CatalinZara? $\endgroup$ – Squird37 May 30 '18 at 2:26
0
$\begingroup$

As suggested in the comment, one can consider the

$$ C_R = C\cap B_R(p).$$

For $R$ large enough, $C_R$ is nonempty. Since $M$ is complete, the Hopf-Rinow theorem implies that $B_R(p)$ (and thus $C_R$) is compact. Thus there is $p' \in C_R$ so that $$ d(p,p') = \inf_{s\in C_R} d(p,s).$$

Since $d(p, p')\le R$, we also have

$$ d(p,p') = \inf_{s\in C} d(p,s).$$

$\endgroup$
  • $\begingroup$ Thanks @JohnMa. Now everything is clear. $\endgroup$ – Squird37 May 30 '18 at 2:57

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.