Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. 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

In Frankel's paper: "Manifolds with positive curvature", he proved the following theorem:

If $M^n$ is complete Riemannian manifold with positive sectional curvature, and $V^r$, $W^s$ are two compact totally geodesic submanifolds. If $r+s\ge n$ then $V$ and $W$ have a non-empty intersection.

The proof is by contradiction. If there is no intersection, then chose $\gamma$ connecting $V$ to $W$ and realized the distance between $V$ and $W$. Due to the dimension reason, he find a parallel vector fileld $X$ along $\gamma$, then apply the second variational formula.

For the boundary term $g(\nabla_X X, \dot{\gamma})$. He claimed this is zero due to the totally geodesic property. My question is: In order to take covariant derivative $\nabla_v Y$, the vector field $Y$ has to be defined at least along one curve $\sigma$ with $\dot{\sigma}=v$, right? But for $\nabla_X X$ in the proof, the vector filed $X$ is only defined along $\gamma$ not along the tangent direction $X$. (My guess is it does not depend on the extension of $X$, but I can't see why, or it's too trivial so Frankel didn't write it down?)

edit: I found that probably, we can just extend $X$ at $\gamma(0)$ and $\gamma(\ell)$ as the tengent vector of geodesic along direction $X$, and extend $X$ at $\gamma(t)$ arbitrary. So this will give the desired boundary condition. Is my claim correct?

share|cite|improve this question

My thought :

At $p$, $\nabla_XY =X^l \frac{\partial Y^k }{\partial x^l} \frac{\partial }{\partial x^k} + X^l Y^m \Gamma_{lm}^k \frac{\partial }{\partial x^k}$ so that $Y$ must be defined on a neighborhood of $p$.

Note that $X(t)$ is defined on $\gamma(t)$ bacause of parallel transport.

So for each $t$, on a neighborhood $U$ of $\gamma(t)$, we can define a vector field $W$ such that $W(\gamma(t))=X(t)$ and $\frac{d}{ds}_{s=a} \exp_{\gamma(t)} (sX(t)) = W( \exp_{\gamma(t)} (aX(t))$ for $0 < a < \epsilon$ where $\epsilon > 0$ is small.

Furthermore since $V$ is totally geodesic, we can define a vector field $W$, corresponded to $X(0)$, which is tangent to $V$.

share|cite|improve this answer

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.