5

Let $M$ be a be a closed orientable surface, and suppose that its tangent bundle $TM$ splits into a direct sum of line bundles.

How to prove that $M$'s Euler characteristic is zero?


Unfortunately, I am entirely unfamiliar with Characteristic Classes, but I would accept a solution based on properties of these objects;

if there is a reference for a book where I could find this claim or details of a proof, that would be great.

Asaf Shachar
  • 25,111
  • Are you allowed to use the classification of surfaces? – hunter Nov 09 '20 at 17:11
  • 1
    Well, this is not homework so I am allowed to use anything I want:) If there is a more direct proof which does not rely on heavy tools like the classification theorem I would prefer that, but that is not necessary- so feel free to use anything you want to. – Asaf Shachar Nov 09 '20 at 17:14
  • More generally, if the tangent bundle of a closed manifold admits a line subbundle, then it has Euler characteristic zero. See this answer for example. – Michael Albanese Aug 17 '21 at 20:32

2 Answers2

7

To complement Connor's answer:

Classifing spaces argument: Because $M$ is orientable, its tangent bundle is classified by a map $\phi:M\rightarrow BSO(2)=\mathbb{C}P^\infty$. The Euler class of the bundle (which gives $\pm \chi(M)$ when paired with a fundamental class) is by definition the element $\phi^\ast(z)$, where $H^\ast(\mathbb{C}P^\infty;\mathbb{Z})\cong \mathbb{Z}[z]$. Identifying $H^2(M;\mathbb{Z})$ with $\mathbb{Z}$, $\phi^\ast(z)$ can directly be interpreted as $\pm\chi(M)$.

So, we need to show that the condition that the tangent bundle splits implies that $\phi^\ast(z) = 0$.

Now a splitting of the tangent bundle implies a reduction of the structure group to $B:=BS(O(1)\times O(1))\subseteq BSO(2)$. That is, $\phi$ factors through the inclusion $B\rightarrow BSO(2)$. Since $B\simeq \mathbb{R}P^\infty$, and $H^2(\mathbb{R}P^\infty;\mathbb{Z})\cong \mathbb{Z}/2\mathbb{Z}$, we see that on degree $2$ cohomology, $\phi^\ast$ factors as a composition $\mathbb{Z}\rightarrow \mathbb{Z}/2\mathbb{Z}\rightarrow \mathbb{Z}$. Since any homomorphism $\mathbb{Z}/2\mathbb{Z}\rightarrow \mathbb{Z}$ is trivial, we deduce $\phi^\ast(z) = 0$ as desired.

By the way, the assumption that $M$ is orientable is not needed. For, if $M$ is non-orientable, and $\pi:\widetilde{M}\rightarrow M$ is the orientation double covering, then the fact that $T\widetilde{M}\cong \pi^\ast TM$ shows that splitting of $TM$ gives a splitting of $\widetilde{M}$. By the orientable case, $\chi(\widetilde{M}) = 0$, from which it follows that $\chi(M) = 0$ as well.

5

Characteristic class argument: The Euler class of a vector bundle vanishes if there is a nonvanishing section. The Euler class of the tangent bundle evaluated on the fundamental class is the Euler characteristic. If we have a trivial subbundle, we automatically have a section. We know that we have a 1-dimensional subbundle. Since we are orientable, this implies this subbundle is either trivial or both it and its complement are nonorientable. In the first case, this gives the Euler characteristic 0 by before.

In the second case, see Jason's answer.

Singularity argument: A vector field on a surface has the sum of the indices of the singularities equal to its Euler characteristic. Since a nonvanishing section of the tangent bundle is a vector field with no singularities, we deduce that if either of the bundles is trivial, the Euler characteristic of the surface vanishes.

In case both are nontrivial, we can pullback to a cover as in Jason's answer.

Connor Malin
  • 11,661
  • 1
    I'm not sure we know that the line bundles are trivial. For example, the tangent bundle to $T^2$ can be decomposed as a sum of Mobius bundles. – Jason DeVito - on hiatus Nov 09 '20 at 17:37
  • Oh sorry I read the question wrong, will fix. – Connor Malin Nov 09 '20 at 17:38
  • 2
    I see a fix: if $TM$ splits as a sum of non-trivial line bundles, this implies $H^1(M;\mathbb{Z}/2\mathbb{Z})$ is non-trivial, which implies $\pi_1(M)$ has an index two subgroup. Pulling back $TM$ to the cover $\widetilde{M}$ of $M$ which corresponds to this index two subgroup, you get a splitting of $T\widetilde{M}$ into orientable (so trivial) line bundles. Now use either of your arguments on $\widetilde{M}$ and then conclude about $M$. – Jason DeVito - on hiatus Nov 09 '20 at 17:48
  • @JasonDeVito That's very nice. I was trying to explicitly come up with a section and was unable to. Maybe one thing: you need orientability to conclude that the pullback has both bundles trivial (because they must combine to give an orientable bundle). But this is not necessary since you only need 1 bundle trivial to begin with. – Connor Malin Nov 09 '20 at 17:52
  • Agreed. Alternatively, if you pull back and one of the bundles is still non-orientable, then you can pull back again. By the way, I'm happy to delete all my comments, if you'd like me to. I'm also happy to keep them up. Whichever you prefer. – Jason DeVito - on hiatus Nov 09 '20 at 18:03
  • @JasonDeVito I'm of the opinion that unless an answer has been substantially changed or there is pointless fighting, there isn't a reason to delete comments. – Connor Malin Nov 09 '20 at 18:04
  • Maybe I missed this in all the comments ... But if one subbundle is orientable, the other is orientable as well if and only if $TM$ is orientable. – Ted Shifrin Nov 09 '20 at 19:19