A subset $S\in M$ is called strongly convex, or geodesically convex, if for any $p,q\in S$ there is a unique normal minimal geodesic $\gamma$ joining $p$ to $q$, and $\gamma$ is contained in $S$.

For example, on $(\mathbb{S}^{2},g_{\mathbb{S}^{2}})$, any geodesic ball of radius $r<\frac{\pi}{2}$ is strongly convex, (since, as I see, I take the unit sphere $\mathbb{S}^{1}$ on $\mathbb{R}^{2}$ his parametrization is $(\cos(\theta),\sin(\theta))$ then for any $r<\frac{\pi}{2}$ I will be still in the sphere, and as the sphere in $\mathbb{R}^{3}$ is the "same" the sphere $\mathbb{S}^{1}$ we're done).

And the book's claim, each strongly convex set is contractible, and the intersection of a family of strongly convex sets in $M$ is again strongly convex if it is not empty, I don't think is so trivial see this. And finally, can you give me an example of geodesically convex that does not belong to space, Thanks!!

  • $\begingroup$ Out of curiosity, what book is this from? $\endgroup$ – littleO May 3 '16 at 22:16
  • 1
    $\begingroup$ The last paragraph is from the book Riemannian Geometry By Takashi Sakai. $\endgroup$ – User166523 May 3 '16 at 22:48
  • $\begingroup$ I think he (or you) is missing further assumption. Without this assumption it is unclear why the unique geodesic varies continuously. For instance, if you assume that $S$ is compact this will be the case. Or you need no conjugate points assumption. $\endgroup$ – Moishe Kohan May 5 '16 at 15:23
  • $\begingroup$ Related: math.stackexchange.com/questions/479022/…. From this answer you can conclude that your $S$ is contractible if, say, $(M,g)$ is complete and $S$ is closed. $\endgroup$ – Moishe Kohan May 5 '16 at 15:30

The intersection rule is trivial. For any two points $p$ and $q$ in the intersection, they are both in each set, and therefore the unique geodesic $\gamma$ between them must have been in each set. But if the geodesic is in each set it is in the intersection. Therefore the definition is satisfied.

Contractibility seems harder. Let $p$ be a point in $M$. For $q$ in $M$ and $t$ in $[0, 1]$ define $f(t, q)$ to be the point a portion $t$ of the way along the unique geodesic $\gamma$ from $q$ to $p$. This is well-defined. Also $f(0, q) = q$ and $f(1, q) = p$. That has to be the contraction. The question is whether this map is continuous.

The challenge is this. If you have a triangle between two "nearby" points and a distant one, need the edges remain close? Well, in a general manifold, no. The "unique geodesic" condition has to come in somehow. But I can't see how to do it.

As for the last part of your question, the hyperbolic plane does not fit in space and is strongly convex.


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.