2
$\begingroup$

Let $X$ be an Alexandrov space with curvature bounded from below (if necessary, $X$ might be assumed to be finite dimensional or even compact).

Question 1. Is it true that every point of $X$ has a neighborhood $U$ such that any two points from $U$ can be connected by at most one shortest path (which does not have to be contained in $U$)?

Question 2. In the previous question, can one choose $U$ to be geodesically convex, i.e. for any two points from $U$ any shortest path between them (if it is not unique) is contained in $U$?

A reference would be helpful.

Remark. Of course, if $X$ is a smooth Riemannian manifold then the answers to both questions are 'yes'.

$\endgroup$
6
  • 4
    $\begingroup$ I am afraid, the vertex $p$ of the cone $X$ over the circle of length less than $2\pi$ has no such neighborhood $U$. $\endgroup$
    – valeri
    Apr 1, 2015 at 9:50
  • $\begingroup$ @valeri why don't you make this an answer ? $\endgroup$ Apr 1, 2015 at 11:30
  • $\begingroup$ @Thomas Richard May be Semyon Alesker means that the curvature is bounded from above? Then the answer might be different. $\endgroup$
    – valeri
    Apr 1, 2015 at 11:37
  • $\begingroup$ @valeri Thank you! I think this is the final answer. $\endgroup$
    – asv
    Apr 1, 2015 at 12:07
  • 1
    $\begingroup$ @valeri is right that an upper bound on the curvature implies a positive answer to question 1, because one can construct triangles of "arbitrary fatness" from non-unique shortest paths between two points. $\endgroup$ Apr 1, 2015 at 18:03

1 Answer 1

6
$\begingroup$

Counterexample: the vertex p of the cone X over the circle of length less than 2π has no such neighborhood U.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.