2
$\begingroup$

Let $X$ be an $n$-dimensional Alexandrov space with curvature at least -1. Assume that at every point it has an $(n,\delta)$-strainer of length $\mu$, where $\delta$ and $\mu$ are independent of a point.

Does there exist $\sigma=\sigma(\delta, \mu)$ such that for any geodesic triangle of diameter less than $\sigma$ if one of its angles is at least $\pi-1/1000$ then the other two are less than $1/10$?

For smooth Riemannian manifolds with the injectivity radius at least $\mu$ the answer is positive as it was explained in the answer to this post: A property of geodesic triangles in manifolds with lower bounds on curvature and injectivity radius However I do not see how to generalize directly the argument.

$\endgroup$

1 Answer 1

1
$\begingroup$

First note that for any direction $\xi$ at $x$ there is a geodesic $[xz]$ of length $\tfrac{\mu}{10^n}$ which runs in the $\varepsilon$-close direction to $\xi$.

(This part follows from volume comparison. The space is locally-bi-Lipschitz-Euclidean, therefore voulme of small balls are almost as in the Euclidean space. On the other hand if there are no long $\tfrac{\mu}{10^n}$-long geodesics in the directions $\varepsilon$-close to $\xi$ then the volume of $2\cdot\tfrac{\mu}{10^n}$ is too small.)

Next note that if $\measuredangle [x^y_z]=\alpha$ and $|x-y|=|x-z|=r<\tfrac{\mu}{10^n}$ then $$|y-z|>\tfrac{r}{10}\cdot(\alpha-\varepsilon),$$ where $\varepsilon$ is a small constant depending on $n$ and $\delta$.

(It says roughly that in distance coordinates the geodesics from $x$ are nearly straight, say it can not turn by big angle.)

These two observations plus the standard comparison lead to the solution.

$\endgroup$
2
  • $\begingroup$ Do you mean that the answer to my question is positive, may be after one modifies the constants? $\endgroup$
    – asv
    Sep 24, 2015 at 10:07
  • 1
    $\begingroup$ @sva yes, but maybe with worse constants. $\endgroup$ Sep 24, 2015 at 10:21

Your Answer

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

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