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

Let $M$ be a $n$-dimensional smooth riemannian manifold and $\varphi\colon(-\varepsilon,\varepsilon)\rightarrow M$ an embedding. $\varphi$ will denote the image of $\varphi$, too.

Consider the bundle of orthonormal $(n-k)$-frames in $M$ over $\varphi$, where $k>2$, which are orthogonal to $\varphi$. The fiber of this bundle is the orthonormal stiefel manifold $V_{n-k}(\mathbb{R}^{n-1})$, which is connected, since $n-k<n-1$.

My first idea to show, that this bundle is smoothly trivial was, to transport one fixed fiber via parallel transport along $\varphi$ of the Levi-Civita-connection of $M$. This works, because parallel transport along $\varphi$ gives me isometries, which leave vectors tangent to $\varphi$ fixed. (Just to be sure, this works, right?)

Now take a global section $\eta$ of this bundle and consider the bundle of orthonormal $(k-1)$-frames in $M$ over $\varphi$, which are orthogonal to $\varphi$ and to $\eta$. This gives me a bundle with fiber $V_{k-1}(\mathbb{R}^{n-1-(n-k)})=V_{k-1}(\mathbb{R}^{k+1})$.

I am searching for an argument to show, that this second bundle is also smoothly trivial. The proof considered before doesn't work, because parallel transport along $\varphi$ doesn't preserve beeing orthogonal to $\eta$ in general. Can someone help me out by giving me a hint how to prove that?

I don't want to use any classification results or any sophisticated bundle-theory. I think there must be an more-or-less elementary argument like the one of my first idea.

share|cite|improve this question

The first idea only works when $\phi$ is a reparametrized geodesic. Write $\dot \phi= \tfrac{d}{dt}\phi$ and denote parallel transport along $\phi$ by $P_{\phi}$. Then $P_{\phi}(\dot\phi(0))$ remains tangent to $\phi$ iff there is an $\alpha:(-\varepsilon,\varepsilon)\to \mathbb{R}$ s.t. $P_{\phi}(\dot \phi(0))(t)=\alpha(t)\dot \phi(t)$ and then $\nabla_{t}(\alpha\dot \phi)=0$, i.e. $\nabla_t(\tfrac{d}{dt}\phi(\int_{0}^{t}\alpha(\tau)d\tau))=0$.

If $\phi$ is not a reparametrized geodesic then assume it is normalised, choose an orthonormal (n-k)-frame $\{e_{i}\}$ at $T_{\phi(0)}M$ which is orthogonal to $\dot\phi(0)$ and solve the system of equations: $$\nabla_{t}E_{i}=\langle E_{i},\nabla_{t}\dot \phi\rangle \dot \phi,\quad E_{i}(0)=e_{i}$$ The solutions satisfy $\langle \nabla_{t}E_{i},\dot \phi\rangle=\langle E_{i},\nabla_{t}\dot \phi\rangle$ and therefore $\tfrac{d}{dt}\langle E_{i},\dot \phi\rangle=0$ and the orthogonality at $t=0$ is preserved for all $t$. Then also $\langle \nabla_{t}E_{i},E_{j}\rangle=0$ and therefore orthonormality of the $\{e_{i}\}$ is preserved too.

Something similar will work for you second question, but I think you will need to involve $\eta$ in the system of equations.

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.