4
$\begingroup$

Let $M$ be a smooth manifold of dimension $n.$ It is well known that it is not allways possible to find such global vector fields $X_1,\ldots, X_n$ which would be a basis for the tangent space at each point of $M.$ Indeed, that can only happen for parallelizable manifolds $M.$

But how about if we allow a larger number of fields, so that despite a few of them having to be linearly dependent at any given point $p\in M,$ they will still span $T_pM$? I feel like such a set of vector fields should always exist, even under the added requirement that it should be finite, but I don't know how to prove either ...

$\endgroup$

1 Answer 1

4
$\begingroup$

Let $f:M\to\mathbb R^N$ be an embedding (such a thing exists for $N$ sufficiently large). Let $X_i$ be the normal projection of the constant "coordinate" field $e_i$ on $\mathbb R^N$ to the tangent subspaces to $M$. This gives (by pullback) $N$ vector fields on $M$ which span the tangent subspaces at each point.

For example, $S^2$ embeds in $\mathbb R^3$ and this gives three tangent vector fields on $S^2$ which span the tangent space at each point. You should write them explicitly, for fun.

$\endgroup$
2
  • $\begingroup$ We know there is an embedding for $N=2\dim M$. In fact, an immersion is enough to do this and that can be gotten in dimensin $2\dim M-1$. A non trivial invariant of the manifold is the minimum number of fields needed. $\endgroup$ Jun 23, 2015 at 0:17
  • $\begingroup$ Thank you. Just appealing to Whitney and reaping the rewards; that is very pleasently simple. $\endgroup$ Jun 23, 2015 at 0:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .