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

I am dealing with the following concrete situation that could be familiar to Riemannian geometers more experienced than myself.

Suppose that $M$ is a smooth compact manifold of dimension $m$ and $g$ is a smooth Riemann metric on $M$. $\newcommand{\ve}{{\varepsilon}}$ $\newcommand{\pa}{\partial}$ Suppose that $(g^\ve)_{\ve>0}$ is a family of smooth Riemann metrics satisfying the following properties.

A. We know that for any $p\in M$ there exists an open neighborhood $U\ni p$ and local coordinates $x^1,\dotsc, x^m$ on $U$ such that

$$ g^\ve_{ij} \to g_{ij} $$

uniformly on the compacts of $U$, where

$$ g^\ve =\sum_{i,j} g^\ve_{ij}dx^idx^j,\;\;g=\sum_{i,j}g_{ij}dx^idx^j. $$

B. (Edited following Deane Yang's inquiry.) The note by $Gr_2(TM)$ the bundle of Grassmanians of $2$-planes in the tangent bundle. The sectional curvature $K^\ve$ can then be viewed as a function $K^\ve: Gr_2(TM)\to\mathbb{R}$. We know that there exists a smooth function $K^0: Gr_2(TM)\to \mathbb{R}$ such that $K^\ve\to K^0$ uniformly.

Question. Can we conclude that the function $K^0$ in B is the sectional curvature of $g$?

share|cite|improve this question
I just want to confirm that the sectional curvature converges only for co-ordinate 2-planes and not for other 2-planes? Others can answer your question better than me, but my experience has been that controlling the sectional curvature only along co-ordinate 2-planes (for a fixed set of co-ordinates) is not enough. – Deane Yang May 15 '12 at 20:46
It's a bit better than the uniform convergence mentioned in my question. let me edit the question. – Liviu Nicolaescu May 15 '12 at 22:14
Have you looked at the paper by Stefan Peters, "Convergence of riemannian manifolds", Compositio Math 62 (1987) 3-16? You can't conclude that the limiting sectional curvature is continuous but it is bounded almost everywhere. – Deane Yang May 15 '12 at 22:41
I have looked at the paper you mentioned. That paper uses only the assumption that $K^\varepsilon$ is uniformly bounded. In my case $K^\varepsilon$ is convergent to a smooth limit. The question is weather this limit is indeed the curvature of the limiting metric $g$ which we know is smooth. – Liviu Nicolaescu May 16 '12 at 0:10
You're absolutely right, and Anton Petrunin has a much better answer to your question. – Deane Yang May 16 '12 at 8:05

Yes, it is true. Here is a sketch of proof, but I am sure that there is a simper way to see it.

Note that by Gauss–Bonnet, this is true in dimension 2.

Assume we know it in dimension $n-1$.

There is a construction of smooth convex hypersurfaces $H^\varepsilon$ in $(M,g^\varepsilon)$ which converge nicely to a smooth hypersurfaces $H$ in $(M,g^\varepsilon)$. "Nicely" means that one can apply the induction hypothesis for $H^\varepsilon\to H$. Taking many hypersurfaces like that plus linear algebra finishes the proof.

The construction. To construct $H^\varepsilon$ near $p\in M$, choose n points $a_1,a_2,\dots a_n$ such that in all the metrics the angles $\angle a_ipa_j\approx\pi/2$ for all $i\ne j$. Then define $$H^\varepsilon=\{\\, x\in M\mid \sum\phi(|a_i-x|_{g^\varepsilon}-|a_i-p|_{g^\varepsilon})= 0 \\,\}$$ where $\phi(0)=0$, $\phi'(0)=1$ and $\phi''\ll -2$.

share|cite|improve this answer
@ Anton Thank you for your answer. I need to process the strategy that you outlined above. I have to admit that I do not see why $H^\varepsilon$ is smooth and what is the role of the angle condition that you mentioned. – Liviu Nicolaescu May 16 '12 at 9:35
@Liviu, sorry, it is very cryptic. Look at 4.3 in our "Extremal subsets in ..." and then look at Lemma B.5 in… . Hope it helps. – Anton Petrunin May 16 '12 at 10:26
@ Anton: Thanks for the ref. I'll check it out. – Liviu Nicolaescu May 16 '12 at 12:45
@Anton: Perhaps it is obvious, but why does it follow from Gauss-Bonnet in dimension 2? – Dan Lee May 16 '12 at 18:31
@Dan. Fix 3 very close points, let $\triangle^\varepsilon$ be geodesic triangles for $g^\varepsilon$ with the vertices at these points. Note that the angles and area of $\triangle^\varepsilon$ converge to the angles and area of $\triangle^0$. Apply Gauss–Bonnet. – Anton Petrunin May 17 '12 at 2:17

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.