15
$\begingroup$

Suppose I have a smooth Riemannian manifold $X$ with induced distance function $d$, and a bi-Lipschitz (with respect to $d$) homeomorphism $$\phi: X \to X.$$

Under what circumstances could $\phi$ be smoothable to a diffeomorphism? By "smoothable" in this case I mean "homotopic to a diffeomorphism through bi-Lipschitz homeomorphisms" (this might not be standard, I suppose). Are there any clear obstructions?

$\endgroup$
3
  • 1
    $\begingroup$ What do you mean by "smoothable"? $\endgroup$ Jul 19, 2020 at 20:38
  • $\begingroup$ @Igor Belegradek: I guess the OP means "approximated by $C^1$ diffeomorphisms", with uniform convergence? $\endgroup$ Jul 19, 2020 at 21:08
  • $\begingroup$ Sorry, just edited the question to make it more clear what I mean! $\endgroup$ Jul 19, 2020 at 22:58

2 Answers 2

13
$\begingroup$

Any self-homeomorphism of a manifold of dimension $\neq 4$ is topologically isotopic to a bi-Lipschitz homeomorphism, see lemma 2.4 in Lipschitz and quasiconformal approximation of homeomorphism pairs by Jouni Luukkainen.

There are exotic spheres (e.g. in dimension $7$) that admit a self-homeomorphism that is not homotopic to a diffeomorphism, see here. This gives a bi-Lipschitz homeomorphism that is not homotopic to a diffeomorphism.

$\endgroup$
5
  • $\begingroup$ This is interesting, thanks. Is there a counterexample in dimension 4? Perhaps one can appeal to Donaldson-Sullivans work on Lipschitz/quasi-conformal 4-manifolds... $\endgroup$ Jul 20, 2020 at 16:13
  • $\begingroup$ @RohilPrasad: I do not know what happens in dimension 4. You now have all the references that I have, and should be able to explore it further. $\endgroup$ Jul 20, 2020 at 16:27
  • $\begingroup$ I am not sure if the answer is correct. The link shows a self-homeomorphis, that is not isotopic to a diffeomorphism. This is because the manifold with reversed orientation are not diffeomorphic. I think the homeomorphism cannot be bi-Lipschitz due to uniqueness of the Lipschitz structure in dimensions $\neq 4$. $\endgroup$ Sep 23, 2020 at 1:39
  • $\begingroup$ @PiotrHajlasz: I do not claim to understand Luukkainen's paper, but I think what I say is a formal consequence of his Lemma 2.4 for $A=\emptyset$ and $B=M$. $\endgroup$ Sep 23, 2020 at 1:59
  • $\begingroup$ I am sorry, I was stupid. Exotic spheres are bi-Lipschitz homeomorphic exactly for a reason I mentioned: uniqueness of a Lipschitz structure. $\endgroup$ Sep 23, 2020 at 3:03
3
$\begingroup$

There is some interest in a related question in non-linear elasticity, specifically people there would consider a function "smoothable" if there is a close-by (in some norm applying both to function and its inverse) diffeomorphism.

In 2D there are some results with regards to this, see e.g. Smooth approximation of bi-Lipschitz orientation-preserving homeomorphisms by Danieri & Pratelli, who prove that there is a close diffeomorphism in some bi-Sobolev norm for domains in the plane (which if I am not mistaken should imply the same result at least for compact manifolds). But the proof uses a bi-Lipschitz extension theorem, so I am not sure if one can construct homotopies from that result easily and there seems to be no way to extend this to higher dimensions.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.