5
$\begingroup$

I'm having trouble to prove the following statement:

If the $n$-sphere $S^n\subset\mathbb{R}^{n+1}$ admits a (continuous) nonvanishing tangent vector field, then $n$ is odd.

The idea of the proof is pretty clear on my mind: we use the fact that the antipodal map has degree $(-1)^{n+1}$ and that the Brouwer degree is invariant under homotopies.

We have a (continuous) nonvanishing tangent vector field $v\colon S^n\to\mathbb{R}^{n+1}$ that can be supposed unitary (otherwise, just consider $u=v/\|v\|$). Paraphrasing Hirsch: "a homotopy of $S^n$ from the identity map to the antipodal map is obtained moving each $p\in S^n$ to $-p$ along the great semicircle in the direction $v(p)$".

My question is: is there an explicit formula for this homotopy?

$\endgroup$

2 Answers 2

6
$\begingroup$

The application $H\colon S^n\times[0,1]\to S^n$ given by $$H(x,t) = \cos(\pi t)x + \sin(\pi t) v(x)$$ is what I was looking for.

$\endgroup$
2
  • $\begingroup$ That will not stay in $S^n$. $\endgroup$ Jun 3, 2017 at 20:25
  • 1
    $\begingroup$ @PVAL-inactive Well, for all $t\in[0,1]$ and all $x\in S^n$, we have that $\|\cos(\pi t)x + \sin(\pi t)v(x)\|^2 = \langle \cos(\pi t)x + \sin(\pi t)v(x), \cos(\pi t)x + \sin(\pi t)v(x) \rangle = \cos^2(\pi t)\|x\|^2 + \cos(\pi t)\sin(\pi t)\langle x,v(x)\rangle + \sin^2(\pi t)\|v(x)\|^2 = \cos^2(\pi t) + \sin^2(\pi t) = 1$, since $\|x\| = \|v(x)\| = 1$ and $\langle x,v(x) \rangle = 0$. $\endgroup$ Jun 4, 2017 at 4:54
5
$\begingroup$

How about this? Define $\Phi:S^n\times[0,1]\to S^n$, $n=2m-1$ by $\Phi(v,t)=A(t)v$ where $A(t)$ is the matrix $$\pmatrix{ \cos \pi t&\sin\pi t&&&&&&\\ -\sin \pi t&\cos\pi t&&&&&&\\ &&\cos \pi t&\sin\pi t&&&&\\ &&-\sin \pi t&\cos\pi t&&&&\\ &&&&\ddots&&&\\ &&&&&\cos \pi t&\sin\pi t\\ &&&&&-\sin\pi t&\cos\pi t\\ }.$$

$\endgroup$
2
  • $\begingroup$ Nice trick! But I have to prove that $n$ is odd, haven't I? $\endgroup$ May 31, 2017 at 18:30
  • $\begingroup$ @rldias Lord Shark is thinking of $S^n$ as a subset of $\Bbb R^{n+1}$. There is a reason this construction only makes sense when $n$ is odd (i.e. when $n+1$ is even). $\endgroup$ Jun 3, 2017 at 20:27

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.