This is a problem from the my last exam in Differential Geometry II and I didn't solve it. I'm studying again, but without success. So I need help.

Does there exist a surface $S \subset \mathbb{R}^3$ which is homeomorphic to the torus $\mathbb{T}^2$ and has Gaussian curvature $K \geq 0$?

What I have to work with: Differential forms, Gauss-Bonnet Theorem, Stokes Theorem, Euler characteristic, etc.

Can someone help me?

  • 4
    $\begingroup$ What does Gauss Bonnet tell you? What is $\chi(\Bbb T^2)$? $\endgroup$ – Ted Shifrin Jan 27 '14 at 18:04
  • $\begingroup$ Must be $S\subset{\Bbb R}^3$. $\endgroup$ – Martín-Blas Pérez Pinilla Jan 27 '14 at 18:13

Assume such a surface $S$ exists. A torus is a surface of genus $g=1$, therefore the Euler characteristic is $\chi(S)=\chi(\mathbb{T}^2)=2-2g=0$. By the Gauss-Bonnet theorem it follows that

$$\int_{S} K dA=0$$

where $K$ is the Gaussian curvature of $S$ and $dA$ the area element on $S$. Assume $K\ge 0$. Since $K$ cannot be identically zero (*), the left hand side is positive. Contradiction.

Therefore such a surface cannot exist.


(*) Because the only surfaces in $\mathbb{R}^3$ with constant Gaussian curvature $0$ are planes and cylinders.

  • $\begingroup$ The Gaussian curvature is invariant by homeomorphisms? $\endgroup$ – Felipe Jan 27 '14 at 18:17
  • 1
    $\begingroup$ @Felipe: no (deform a little bit the tip of a sphere for example), but the Euler characteristic is. Therein lies the power of Gauss-Bonnet! It combines topological data (Euler characteristic) with geometrical data (global curvature). $\endgroup$ – Bruno Stonek Jan 27 '14 at 18:20
  • 1
    $\begingroup$ $K$ not being identically 0 isn't something to take for granted. This is where you need to use the fact that you are in $\mathbb{R}^3$, and not in some other ambient space. $\endgroup$ – Nate Jan 27 '14 at 18:22
  • $\begingroup$ I am satisfied with the part that if $ K> 0$ the left side of the integral is positive and this results in a contradiction but I could not quite understand what happens when $ K = $ 0 and where the absurdity of it. $\endgroup$ – Felipe Jan 27 '14 at 18:23
  • 3
    $\begingroup$ @Felipe: any compact surface in $\mathbb{R}^3$ has a point with positive curvature. Indeed, it is bounded, thus it lies inside of a big enough sphere. Now reduce the sphere's radio until it first touches the surface in a point. That point of the surface must therefore have positive curvature. (You should fill in the details for this argument). $\endgroup$ – Bruno Stonek Jan 27 '14 at 18:27

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.