MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top


We know that the space of riemannian metrics on a compact manifold is an open cone in the space of symmetric 2-tensors. Is it reasonable to think that metrics with positive sectional curvature (even positive at a specific point $x \in M$) also form a convex cone?

This is a question about the local behaviour of metrics, so I am not imposing the condition that the sectional curvature be positive everywhere.

Also, due to certain corollary of this statement for a class of metrics, I am quite certain that this cannot hold for metrics of negative curvature.

share|cite|improve this question
Do you want to impose to have a "convex" cone? otherwise this seems obvious. – Reimundo Heluani Sep 29 '11 at 0:30
Yes Reimundo, the question is whether it is convex or not. – S.A.A Sep 29 '11 at 0:34
I added the word convex to the question. Thanks. – S.A.A Sep 29 '11 at 0:35
The following reference does not answer your question, but seems relevant: There Fernando Coda Marques proves that the space of positive scalar curvature metrics on a given orientable 3-dimensional manifold is path-connected. It makes essential use of the Ricci flow and the work of Perelman. – Benoît Kloeckner Sep 29 '11 at 8:26
up vote 16 down vote accepted

No, the formula for curvature is nonlinear with respect to metric tensor in a very essential way.

In particular, a convex combination of two positively curved metrics can have negative curvature. In fact, arbitrary large negative sectional curvature.

For example, the induced metric on any embedding $\mathbb{S}^2\hookrightarrow\mathbb{S}^2\times \mathbb{S}^2$, such that both projections are diffeomorphisms is a convex combination of metrics with constant curvature 1/2. But it is not hard to make such metric arbitrary bad.

share|cite|improve this answer
What's the easiest counterexample for this? – Deane Yang Sep 29 '11 at 1:27
@Deane, I include a source for such examples. – Anton Petrunin Sep 29 '11 at 1:45
Thanks! I know impressively little about these things. – Deane Yang Sep 29 '11 at 1:58

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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