In Tu's "An Introduction to Manifolds", exercise 16.8, we are supposed to prove that $M$ has a trivial tangent bundle $\Leftrightarrow$ M admits a smooth frame $\{X_1, \dots, X_n\}$ (i.e., smooth vector fields $X_1, \dots, X_n$ such that $\{X_{1_p}, \dots, X_{n_p}\}$ is a basis for $T_pM$ for every $p\in M$).

When I read this, I concluded that $M$ is orientable $\Rightarrow$ M admits a smooth frame $\{X_1, \dots, X_n\}$ $\Rightarrow M$ has trivial tangent bundle. However, I've learned that $\mathbb{S}^2$ is an orientable manifold which has no trivial tangent bundle.

What am I missing here?

  • $\begingroup$ If $M$ has trivial tangent bundle, then $M$ is orientable. The converse is not true as your example demonstrates. $\endgroup$ – Michael Albanese Nov 29 '16 at 22:19
  • $\begingroup$ How do you read that orientable implies parallelizable(existence of global frame? $\endgroup$ – melomm Nov 29 '16 at 22:29
  • $\begingroup$ @melomm if $M$ is orientable, then there are smooth vector fields which form an oriented base for the tangent spaces, no? Wouldn't that be a smooth frame? $\endgroup$ – rmdmc89 Nov 29 '16 at 22:31
  • 1
    $\begingroup$ What definition of orientabiliy you are using? $\endgroup$ – melomm Nov 29 '16 at 22:35
  • 1
    $\begingroup$ The @Andreas Cap's answer is the definition of orientability in terms of local vector fields. But I think so you are confuse because for an hypersurface $S$ on $\mathbb{R}^{n}$ the orientability is equivalent to esistence a global nornmal vector field, in particular the normal bundle is trivial. (The same is true if you change $\mathbb{R}^{n}$ by an orientable manifold ) $\endgroup$ – melomm Nov 30 '16 at 13:13

I am referring to your discussion with @melomm in the comments to the question. The right definition of orientability involving vector fields is the following. You have to specify an orientation of the tangent space $T_xM$ for all $x\in M$, which are compatible in the sense that for each $x\in M$, there is an open neighborhood $U$ of $x\in M$ and there are local vector fields $X_1,\dots X_n\in\mathfrak X(U)$ such that for each $y\in U$ the values $X_1(y),\dots,X_n(y)$ define a positively oriented basis for $T_yM$. Of course, this implies that $TM$ is trivial over $U$, but $TM$ is locally trivial anyway.

To see that this is equivalent to the definition via oriented atlasses roughly goes as follows. It easily follows from the defintions that for a connected local chart $(U_\alpha,\phi_\alpha)$ the choice of an orientation of $T_xM$ for one $x\in U_\alpha$ gives rise to a compatible choice for all $y\in U_\alpha$. If you have an oriented atlas than you can extend this orientation to all points in $M$. Conversely, if you have compatible orientations the charts inducing this given orientation form an oriented atlas.

  • $\begingroup$ that's a much better explanation then the ones I've found in textbooks, thanks! $\endgroup$ – rmdmc89 Nov 30 '16 at 14:24

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.