Let $(M,g)$ be a three dimensional smooth Riemannian manifold and suppose that $\Gamma$ is an embedded minimal surface in $M$. Define the Fermi or semigeodesic coordinates around this surface through the local diffeomorhism $Z:\Gamma \times \mathbb{R} \to M$ $ Z(y,z) = Exp_y (z N(y)) $ where $N$ is the unit vector along $\Sigma$

Define $\Gamma_{t} = \left\{ {x_3=t} \right\} $. and let $H_{x_3}$ denote the mean curvature of $\Gamma_{x_3}$

Does there exist a conformal factor $c$ such that if $ \hat{g} = c g$ then $\hat{H_t}$ is constant for all sufficiently small t?


  • $\begingroup$ there is a related result to this question which shows that it is always possible to build a conformal factor such that the geodesic spheres around a point will be of constant mean curvature locally... $\endgroup$ – Ali Nov 28 '14 at 23:32

Yes, this can always be done as long as $\Gamma$ has a positive injectivity radius.

In fact, calculation yields the following formula: Let $V$ be the unit vector field (defined in a neighborhood of $\Gamma$) that is $g$-perpendicular to the level sets of $t$ (i.e., $V$ is perpendicular to each $\Gamma_t$). This $V$ is well-defined on a $\Gamma$-neighborhood $U\subset M$. Then $$ \mathrm{d}c(V) = 2 H\,c - 2\hat H\,c^{3/2}, $$ where $H$ is the function on $U$ that gives the mean curvature of the level sets $\Gamma_t$ in the metric $g$ and $\hat H$ is the function on $U$ that gives the mean curvature of the level sets $\Gamma_t$ in the metric $\hat g = c g$.

Thus, one can specify any desired function $\hat H$ (in particular, a function constant on the each level set $\Gamma_t$) and choose any desired initial function $c_0>0$ on $\Gamma=\Gamma_0$, and there will be a unique solution $c>0$ to the above equation (which is a first order PDE) that satisfies $c = c_0$ along $\Gamma_0$. This solution $c$ will be defined on an open neighborhood $\hat U\subset U$.

  • $\begingroup$ Thanks Robert for your nice answer. May I ask you where the assumption about the positive injectivity radius was used? $\endgroup$ – Ali Nov 29 '14 at 21:03
  • $\begingroup$ Well, that probably was a bad use of terminology. All I meant to assume was that there is an $\epsilon>0$ such that the map $Z$ that you defined in your question is an injective diffeomorphism on $\Gamma\times (-\epsilon,\epsilon)\subset \Gamma\times\mathbb{R}$. That way, you can be sure that for $t$ sufficiently small, the level sets $\Gamma_t$ are disjoint. I didn't really want to assume anything about the injectivity radius of the induced metric on $\Gamma$. $\endgroup$ – Robert Bryant Nov 29 '14 at 21:21

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.