I am reading the Colding and Miniczzi book "A course in Minimal Surfaces". In chapter 1, section 8 there is a derivation of the second variation formula, and the stability operator $L$, but just for the case of normal variations.

A normal vector fields $J$ on a minimal surface $\Sigma$ is said to be a Jacobi field if $LJ = 0$.

Then in the book they consider the case when $\Sigma \subset \mathbb{R}^3$ is the graph of a function satisfying the minimal surface equation. Then they consider the variation $F: \Sigma \times \mathbb{R} \to \mathbb{R}^3$ defined as: $$ F \colon (x_1, x_2, x_3, t) \mapsto (x_1, x_2, x_3 + t). $$ The variation field on the surface is $(0, 0, 1)$ and clearly $$ t \mapsto Vol\big(F(\Sigma, t) \big) $$ is constant.

Then in the book they conclude that the normal component of the variation field of this variation is a Jacobi Field.

I don't understand why this should be true. I mean, how can we assert that the normal component of the variation $F$ does not change the volume? In principle this could be false.


If you're only interested in geometry (and not parametrization), then the normal component of a variation field is the only part that matters - any field can be written as the sum of a tangential part and a normal part, and variations of a map tangential to its image do not change its image.

Thus, the variation generated by the normal component of $(0,0,1)$ will have image coinciding with that of $F$, which clearly has constant volume.

  • $\begingroup$ Yes thanks! I actually just realized that it was a silly question. Because the first variation formula is: $$ \frac{d}{dt} Vol(F(\Sigma, t)) = - \int_\Sigma (F_t, H) $$ where F is a variation (not necessarily normal) with compact support. And since the mean curvature vector $H$ is normal, it's clear that the tangent component of the variation does not contribute. Thank you very much!! $\endgroup$ – Onil90 Jan 19 '17 at 10:38
  • $\begingroup$ There is one technical detail @Onil90, the variation $F$ (the translation) is not of compact support, so for your first variation formula, there would be a boundary term. $\endgroup$ – user99914 Jan 19 '17 at 19:31
  • $\begingroup$ @JohnMa you are right. How can I solve this problem? $\endgroup$ – Onil90 Jan 19 '17 at 20:26
  • $\begingroup$ In fact you can just calculate $L f$, where $f = \vec n \cdot (0,0,1)$ is the orthogonal part of $(0,0,1)$ and check that to be zero. Of course it is not so geometric. @Onil90 $\endgroup$ – user99914 Jan 19 '17 at 20:32
  • $\begingroup$ @JohnMa Yes, right, but it would be nice to avoid the computation.. :D mmmh $\endgroup$ – Onil90 Jan 19 '17 at 20:45

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.