Suppose $ f: \mathbb{R}^3 \to \mathbb{R}$ is a smooth function and $V = \nabla f$ is a smooth vector field. Let $\theta : \mathbb{R} \times \mathbb{R}^3 \to \mathbb{R}^3$ be the integral curves of $V$. It would be very good for my problem if I could construct a chart $\psi = (t,\theta,\varphi)$ with the property that $(f\circ \psi)^{-1}(a) = \{p\} \times [0,2\pi]\times [0,\pi]$ is some vertical two dimensional plane in the domain of $\psi$ for each $a$ in the image of $f$. I would very much like to determine the Laplacian in these coordinates. I have taken an introductory course to smooth manifolds during my master studies, so I have a basic understanding of smooth manifolds. I feel this should be possible for the particular $f$ that I am using, but I am not sure how to do it. (Please let me know if the question is not well-posed, I am a beginner, so that would be helpful.) If you know how it is possible to solve my problem, it would be great if you can provide some details, such that I can read up on it and see if it works.


You can do it away from the zero of $V$ (i.e. the critical point of $f$).

Let $q \in \mathbb R^3$ be a point where $V(q)\neq 0$. By the Straightening theorem for vector fields, there is an open set $U$ with coordinates $(t, x, y)$ and a local coordinates $\varphi : U \to \mathbb R^3$ so that $V = \frac{\partial }{\partial t}$ in $\varphi(U)$. By shrinking $U$, we assume $U = I \times V$ for some interval $I$. Under this coordinates, the integral curve $\theta$ is given by

$$ \theta (t_0, (t, x, y )) = (t_0+ t, x, y)$$

and after a scaling in the $(x, y)$ direction, this coordinates has the properties you need.

  • 1
    $\begingroup$ Great news Arctic Char! Do you know offhand how to get a formula for the Laplacian of these coordinates, or must I go through the proof of the straightening theorem to see if I can get an expression for $\varphi$? $\endgroup$ – Mikkel Rev Nov 29 '19 at 23:30
  • $\begingroup$ I have gone through the proof now. It seems difficult get an explicit formula for the coordinates because I must invert a mean-looking solution of a PDE. Is the easiest way to get the laplacian to use the Voss-Weyl formula? If so, do you know how I can get the metric in such coordinates? Thank you! $\endgroup$ – Mikkel Rev Nov 30 '19 at 0:38
  • $\begingroup$ Well you can represent the Laplacian as the sum of Laplacian (of the level set $f=constant$) and the mean curvature of the level set (together with one more term). Is that what you want? @MikkelRev $\endgroup$ – Arctic Char Nov 30 '19 at 1:18

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.