2

Prove that invertible linear map in $T:\Bbb{R}^n\to \Bbb{R}^n$,that $\det{T}>0$ is isotopic to the identity map.That is find some smooth map $\Bbb{R}^n\times I \to \Bbb{R}^n$ that is diffeomorphism (invertible) at each time $t\in I$.

yi li
  • 4,786
  • 1
    Well, of course there is no hope of such an isotopy passing through only linear maps, because as soon as $\det T<0$ you are guaranteed to send any basis to a lineraly dependent set at some point. Also, the claim fails for $n=1$ and $T=-id$ because $H(0,t)-H(1,t)$ is $1$ at $t=0$ and $-1$ at $t=1$, therefore it will be $0$ for some $0<t'<1$. But $H(0,t')-H(1,t')$ means that $f_{t'}(1)=f_{t'}(0)$. –  Oct 04 '21 at 14:51
  • Oh sorry,I didn't realized $T$ needs to preserve orientation,that is $\det{T}>0$ yesterday.I found this question in Milnor's book,but forget one condition. – yi li Oct 05 '21 at 01:56
  • 1
    $\det(T) > 0$ does not imply $\lambda(T) > 0$. For example $T = \operatorname{diag}{-1/4, -1, 1}$ fails for $t = 1/2$ in your method. Or am I missing something? – obareey Oct 05 '21 at 11:08
  • Oh sorry,I realized this point before,I don't know why I write it on the post.....@obareey – yi li Oct 05 '21 at 11:13
  • See https://math.stackexchange.com/questions/100978/how-many-connected-components-does-mathrmgl-n-mathbb-r-have . – Thorgott Oct 05 '21 at 11:22
  • thank you,I didn't realize this is equivalent to set of matrix $GL(\Bbb{R}^n)^+ = {\det{T}>0}$ are path connected – yi li Oct 05 '21 at 11:27
  • there is another nice note talking about it,see http://virtualmath1.stanford.edu/~conrad/diffgeomPage/handouts/gramconnd.pdf – yi li Oct 05 '21 at 11:30

0 Answers0