1

I am reading certain proof ( John Lee's Introduction to riemannian manifold, Theorem 11.14 ; refer to question Q.3-2) in my questions Understanding the Gunther's Volume comparison theorem ( John Lee's Introductino to Riemannian manifold ) ) and some question arises.

Let $(M, g)$ be a connected riemannian manifold of dimension $n$ and $\psi:=\operatorname{exp}_p|_V^{U} : V:=B_{\delta}(0) \subseteq T_p M \to U:=\operatorname{exp}_p(B_{\delta}(0))$ the exponential map to a geodesic ball.

Let $\varphi : U \xrightarrow{\cong} W\subseteq \mathbb{R}^{n}$ be the assoicated normal coordinates, where the $W$ is the open ball in $\mathbb{R}^{n}$ of radius $\delta$ centered at the origin (C.f. Simple question on normal coordinates on geodesic ball ( image of normal coordinate on geodesic ball can be also ball ? ) ) Consider pull back metric $(W, (\varphi^{-1})^{*}g|_U)$. (Well-defined?)

Q. Then by question is, each $(U,g|_U)$ or $(W, (\varphi^{-1})^{*}g|_U)$, as regarded an open submanifolds, has constant sectional curvatures $0$, or more strongly(?), has constant Riemannian curvature $0$ ?

If the exponential map $\psi : \operatorname{exp}_p$ is local isometry, then from the local isometric invariance of Riemann curvature tensor and the surjectivity of $\psi$, we maybe show the zero constant Riemannian curvature (of $(U,g|_U)$). True? But if not ( C.f. Exponential map as a local radial isometry ), how can we show this? My queston is true?

And more general, how about when the $U$ is just arbitrary normal neighborhood ?

Can anyone help?

Plantation
  • 2,417
  • 2
    No, $U$ need not have constant curvature. We only know the (sectional in this case) curvature is bounded above (as per given condition on $M\supseteq U$), and we pull it back to $V\subset T_p(M)$ via the exponential map. – user10354138 Oct 02 '23 at 09:35
  • Yes. Although the ( open ball in ) tangent space maybe of zero curvature - depends on metric over it- , its exponential image (geodesic ball) may not be zero curvature, since the exponential map may bend the tangent space ( so that its curvature is not necessarly zero.) For example, let's think the exponential map of Sphere endowed with the round metric. ( Do you also think/agree that this really can be counter example ? ) – Plantation Oct 04 '23 at 07:33
  • 1
    You cannot pick an arbitrary metric! You need to pull back the metric on $M$ back to this open ball of $T_pM$. – user10354138 Oct 04 '23 at 08:49
  • @user10354138 O.K. I will think about it more. Anyway thanks. :) – Plantation Oct 04 '23 at 09:00

0 Answers0