Let $f_k:B_1\rightarrow \mathbb R^n$ be a sequence of injective differentiable volume-preserving maps (i.e. $\mu(f_k(A))=\mu(A)$ for any measurable $A\subset B_1$) that converges uniformly to $f:B_1\rightarrow\mathbb R^n$, where $B_1\subset \mathbb R^n$ is the open unit ball, $\mu$ is the Lebesgue measure. Then can we show that $f$ is also volume-preserving? I can prove that $$\mu(f(A))\ge \mu(A)$$ for any measurable set $A$, but the reverse direction is difficult for me.
I ask this question because in p60 of the book "Symplectic Invariants and Hamiltonian Dynamics (2nd edition)" by Hofer and Zehnder, this statement seems to be implicitly used (although the theorem and idea of argument used in that book are correct), and I suspect it, to be honest.
Current progress: if we are given that $f$ is injective, then using a result in algebraic topology (see Moving of sphere embedding and its interior defined by Jordan-Brouwer separation theorem), we can show that $f_k^{-1}$ converges locally uniformly to $f^{-1}$ on $f(B)$. Then the reverse direction also follows.
$\begingroup$
$\endgroup$
9
-
2$\begingroup$ Not sure about the details, but one way to get in trouble in the plane might be if your $f$ ends up mapping the unit circle to a Jordan curve of positive measure, but all your $f_n$ are differentiable (so the image of the unit circle has measure zero). $\endgroup$– Martin M. W.Commented Oct 25, 2023 at 20:28
-
1$\begingroup$ I am not sure, but I would expect a counterexample. See however Lemma 1.2 in arxiv.org/pdf/1510.05575.pdf. I must day, I like your questions. $\endgroup$– Piotr HajlaszCommented Oct 25, 2023 at 23:53
-
1$\begingroup$ the dimension $n$ is not the same $n$ of the sequence’s index, is it? $\endgroup$– Pietro MajerCommented Oct 26, 2023 at 11:57
-
1$\begingroup$ @PietroMajer Thanks for your comment, I have modified my question. $\endgroup$– Tian LANCommented Oct 26, 2023 at 12:01
-
1$\begingroup$ @MartinM.W. I agree that this may lead to a counterexample (although the approximation might be very difficult to construct) $\endgroup$– Tian LANCommented Oct 26, 2023 at 12:17
|
Show 4 more comments