2
$\begingroup$

Suppose $M$ is a complete Finsler manifold of finite dimension. Is there always a smooth proper real-valued function $f$ on $M$ so that the norms of the first and second derivative are bounded ? The existence of a smooth function with bounded first derivative follows from MR3083289 applied to the function that gives distance from the base-point.

$\endgroup$
  • $\begingroup$ do you know how to do it for Riemannian manifolds? $\endgroup$ – Vitali Kapovitch Sep 10 '15 at 21:43
  • $\begingroup$ What are the title and authors of MR3083289? I don't have access to the AMS MR database. $\endgroup$ – Alex M. Jun 18 '18 at 11:41
3
$\begingroup$

For complete Riemannian manifolds of bounded sectional curvature the answer is YES by L.F. Tam. Exhaustion functions on complete manifolds MR2648946. Then the answer for the Hilbert metric on a properly convex manifold is also YES since such a manifold admits a Riemannian metric of bounded curvature that is bi-Lipshitz to the Hilbert metric

| cite | improve this answer | |
$\endgroup$
1
$\begingroup$

I have not worked out the details but I suspect the following surface has no such function.

The example is a smooth surface $S$ in Euclidean space with the induced Riemannian metric.

Start with the infinite circular cylinder $C$ in Euclidean space $x^2+y^2=1$. Construct S by attaching small handles near points $(x_n,y_n,z_n)=(1,0,z_n)$ on C so that $z_{n+1}=z_n + (1/n)$

The derivative of a function $f$ on $S$ must vanish at some point $p_n$ in each handle (like a Morse function). The distance between successive vanishing points is of order $1/n$. I think properness of $f$ forces the second derivative of $f$ to be unbounded.

I am really interested in properly convex manifolds with the Hilbert metric, for which I do not know the answer.

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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