2
$\begingroup$

Let $M \subset \Bbb R^3$ be a regular surface with a parametrization where the induced metric is $$ds^2=\frac{du^2+dv^2}{(u^2+v^2+c)^2}$$

Where $c>0$ is constant. Prove that the Gaussian curvature of $M$ is constant.


I succeeded to solve it by computing the Riemannian curvature by the definition, but it was very long and not elegant. Is there a simpler way to calculate it, by using the direct definition of Gaussian curvature (i.e using a parametrization)?

$\endgroup$
1
$\begingroup$

The curvature of $$\frac{du^2+dv^2}{h(u,v)^2}$$ is given by $$K = h\,\triangle h - \|\nabla h\|^2.$$Here $h(u,v)=u^2+v^2+c$ gives that $$K = 4(u^2+v^2+c) -(4u^2+4v^2) = 4c.$$The proof of this formula is a direct application of Gauss' formula $$K = \frac{-1}{\sqrt{EG}}\left(\left(\frac{(\sqrt{G})_u}{\sqrt{E}}\right)_u+\left(\frac{(\sqrt{E})_v}{\sqrt{G}}\right)_v\right)$$for orthogonal parametrizations.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Thank you, but this is actually how I originally solved it. I didn't learn the above formula, so I had to calculate $K$ by definition. Is there another way to do it, maybe by finding a parametrization? $\endgroup$ – user401516 Feb 23 at 19:32
  • $\begingroup$ When you write the metric as a combination of $du$ and $dv$, a parametrization is already implied. The given metric expression is already local. We are already working in coordinates, and the fact that $M$ is embedded in $\Bbb R^3$ is irrelevant. Also, unless you want to try using big guns, it can't get much simpler than using the formula for the curvature of a conformal metric to the flat one. $\endgroup$ – Ivo Terek Feb 23 at 19:35
1
$\begingroup$

I will give an answer using solely differential forms which I believe gives a more elegant solution as it rewrites the above formulae using less variables; hopefully you are familiar with them. In my opinion (not necessarily right or more instructive), it is easier to remember this flow of ideas than the formula-based approach.

The induced metric (first fundamental form) can be written as $$I=\theta_1^2+\theta_2^2,$$ where in your case you have $\theta_1=\dfrac{du}{u^2+v^2+c}$ and $\theta_2=\dfrac{dv}{u^2+v^2+c}$.

Also, we know that the Gaussian curvature appears in the Gauss equation $$d\omega_{12} = K\theta_1\wedge\theta_2,\tag{1}\label{Gauss}$$ where $\omega_{12}$ is the $\textit{connection form}$ satisfying the structure equations $$d\theta_1 + \omega_{12}\wedge\theta_2=0$$ $$d\theta_2 - \omega_{12}\wedge\theta_1=0$$

Computing the exterior derivatives you get $d\theta_1=\dfrac{2v\;du\wedge dv}{(u^2+v^2+c)^2}$ and $d\theta_2=-\dfrac{2u\;du\wedge dv}{(u^2+v^2+c)^2}$. Using these along with the two equations above you get $$\omega_{12}=\dfrac{-2v\;du+2u\;dv}{u^2+v^2+c},$$ which replaced in (1) gives you $\dfrac{4c\;du\wedge dv}{(u^2+v^2+c)^2}=K\theta_1\wedge\theta_2$ from which you can conclude $K=4c$.

Hope it helps and shows a less computational way of finding out the Gauss curvature.

| 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.