1

Show, using the Gauss-Bonnet theorem, that $\mathbb{S}^2$ and $\mathbb{T}^2$ are not diffeomorphic.

I find this exercise a bit strange, since by $\chi(\mathbb{S}^2)\neq \chi(\mathbb{T}^2)$ we obviously have that $\mathbb{S}^2$ and $\mathbb{T}^2$ are not homeomorphic and especially since any diffeomorphism would be a homeomorphism, not diffeomorphic. However, as the exercise suggests, I end up with the result, that the Gaussian curvature of $\mathbb{S}^2$ and $\mathbb{T}^2$ are different. But how can I conclude that they are not diffeomorphic? I mean if I use the Theorema Egregium, we have that there cannot be an isometry, however not every diffeomorphism is an isometry.

TheGeekGreek
  • 7,869

1 Answers1

1

If $F : S^2\rightarrow (T^2,g)$ is diffeomorphism, then we have $F^\ast g$ is a metric on $S^2$

Note that $\int_{T^2 } K(g)\ dVol_g =2\pi \chi (T^2)$ Here first term is invariant under isometry So $$ \int_{T^2 } K(g)\ dVol_g = \int_{S^2 } K(F^\ast g) \ dVol_{F^\ast g} =2\pi \chi (S^2)$$ Hence we have a contradiction

HK Lee
  • 19,964
  • Sorry to ask, but $F^*$ does denote the pullback? Brilliant answer though, I knew that I had to end up with an isometry, but wasn't sure how to achieve this. Thanks! +1 – TheGeekGreek Dec 17 '16 at 18:21