Let $(X,\omega)$ be a connected symplectic manifold, possibly with boundary. Let $g_1, g_2: B(1) \to X$ be two balls in $X$. Is it true that if $\delta$ is sufficiently small, then there is an ambient isotopy from $g_1|_{B(\delta)}$ to $g_2|_{B(\delta)}$? To be clear, by this I mean a smooth family of symplectic maps $\Psi_t : X \to X$ such that $\Psi_0 = id$ and $\Psi_1(g_1) = g_0$.

If this is true, does anyone have a proof or a reference?


  • $\begingroup$ This is relevant: mathoverflow.net/questions/61994/… . $\endgroup$ – Marco Golla Jul 22 '14 at 6:18
  • $\begingroup$ According to "Introduction to Symplectic Topology" by McDuff and Salamon, it is not known. See Section "Blowing up and down". I don't know what was done after the book was written. $\endgroup$ – Hwang Jul 22 '14 at 9:16
  • $\begingroup$ @Hwang Thanks for your response. I don't have a copy of Introduction to Symplectic Topology, and I couldn't find a .pdf online, but are you sure that it is stated there that this question is not known? Note that the question does not ask for a $\delta > 0$ such that any two $B(\delta)$ are isotopic, which is known to be not known; rather, the question asks about subballs of two fixed balls, which I think should be easier. $\endgroup$ – bentspoon Jul 22 '14 at 16:08
  • $\begingroup$ Now I have a problem in understanding your question. You pick two arbitrary balls of size 1 and take subballs of them. Is it different from picking arbitrary two balls of size $\delta$? $\endgroup$ – Hwang Jul 23 '14 at 0:00
  • $\begingroup$ @Hwang I think it is different, yes (it's formally easier; however, if you can convince me that the problems are equivalent that would be great) $\endgroup$ – bentspoon Jul 23 '14 at 2:25

The answer is yes, and the construction of the isotopy follows from rather elementary techniques:

  • First, Hamiltonian isotopies act transitively on points in a connected symplectic manifold. In other words, we may assume that $g_1(0)=g_2(0)=p$.

  • Second, after another Hamiltonian isotopy, we may even suppose that $D_0g_1=D_0g_2$ (consider e.g. paths of linear symplectomorphisms in a Darboux chart appropriately cut off by using Hamiltonians).

  • Third, and finally, one can perform the "Alexander trick": a linear interpolation $g_i(tx)/t$, $t \in (0,1]$ where the (outer) scalar multiplication is defined using some Darboux chart around $p \in X$.

It can be checked that these two paths of symplectomorphisms are well-defined in some sufficiently small ball $B(\delta)$ and that both paths moreover extend to an isotopy from $g_i|_{B(\delta)}$ to the restriction of the linear symplectomorphism representing $D_0g_1=D_0g_2$ in that Darboux chart.


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.