9
$\begingroup$

Suppose $M$ is a 1-connected closed manifold with sectional curvature $\ge 1$. So the diameter $D$ of $M$ satisfies $$ D \le \pi $$ When equality holds $M$ is isometric to round sphere. In fact this rigidity holds under the assumption of $Ric \ge n-1$. Hence it is natural to ask what happens to almost extreme case. i.e. when $$D \ge \pi -\epsilon $$Under the Ricci assumption only, the manifold is not necessary sphere by a counter example of M. Andersen. However with extra assumption: sectional curvature is bounded from below, Perelman proved it is homeomorphic to a twist sphere. (Is Perelman the first one prove this?)

By Grove-Shiohama's Diameter Sphere Thereom, under the assumption sectional curvature $\ge 1$ and $D\ge \pi-\epsilon$, $M$ is a twist sphere. (It also follows from Perelman's theorem above)

My question is: Is $M$ diffeomorphic to the standard sphere?

Either Perelman's proof or Grove-Shiohama's Diameter sphere theorem uses the 'soft' approach, i.e. It is not by convergence argument, nor a Lipschitz distance between $M$ and $S^n$ is derived.

Actually, I suspect that it is not Gromov-Hausdorff close to the round sphere, as one might round off two tips of $S^2/\mathbb Z_p$, where $\mathbb Z_p$ acts on $S^2$ by rotation along the $z$-axis.

What if one assume there is also an upper bound $K$ on sectional curvatures, i.e. $$1\le sec(M) \le K$$

$\endgroup$
2
  • $\begingroup$ @J Ge: This is an open problem, if memory doesn't fail me. I actually remember talking about it with Grove in one of our group meetings. Also, I believe that the recent work of Curtis Pro might shed some light on that, although, if I remember well, he assumes almost maximal volume. Perhaps you should contact those two people to figure out the state of the art regarding this question. $\endgroup$ May 2, 2013 at 2:14
  • $\begingroup$ @J Ge: Your question is essentialy Problem 1.2 in the "Diameter pinching" section of the problem list from the "Manifolds with non-negative sectional curvature" workshop at AIM (aimpl.org/nnsectcurvature/1). The problem is still open, to the best of my knowledge. $\endgroup$ May 17, 2013 at 12:29

1 Answer 1

3
$\begingroup$

I will answer the last question. Namely, let me show that if $1\le \mathrm{sec}\,M \le K$ and $\mathrm{diam}\, M>\pi-\varepsilon$ for sufficiently small $\varepsilon>0$ then it has to be diffeomorphic to $\mathbb S^m$.

Asssume contrary, then there is a sequence of $m$-dimensional manifolds $M_n$ such that $1\le \mathrm{sec}\,M_n \le K$ and $\mathrm{diam}\, M_n\to\pi$ as $n\to \infty$.

Let's pass to a converging sequence $M_n\to A$. Note that $A$ is $m$-dimensional and $1\le \mathrm{sec}\, A \le K$ in the sense of Alexandrov and $\mathrm{diam}\, A=\pi$. It follows that $A$ is isometric to $\mathbb S^m$

Finally note that $M_n$ is diffeomorphic to $A$ for all large $n$; the later was essentally proved by Cheeger in his thesis.

$\endgroup$

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.