Let $X$ be the Hopf vector field on the three-sphere. Is there a smooth nowhere zero function $f$ so that the modified vector field $fX$ is not the Reeb vector field of any contact form on the three-sphere?

This question is a follow-up on aglearner's interesting question on how to construct volume-preserving flows on the three-sphere that are not Reeb vector fields.

I'm also looking for evidence to support the following guess: If $X$ is a vector field on the three-sphere, there exists a nowhere zero smooth function $f$ so that the vector field $fX$ is not a Reeb vector field for any contact form on the three-sphere.

  • $\begingroup$ If you can make sure that $fX$ has at most one closed orbit, you violate Cristofaro-Gardiner--Hutching's refinement of Taubes' theorems (a.k.a. the Weinstein conjecture). $\endgroup$ – Marco Golla Jan 17 '17 at 0:59
  • $\begingroup$ @MarcoGolla, I think Reeb flows are really special and I'm looking for easier arguments. $\endgroup$ – alvarezpaiva Jan 17 '17 at 16:59

The answer is yes.

Proposition. Let $\alpha$ be the standard contact form on the three-sphere (for which the Reeb vector field is the Hopf vector field $X$). If $f$ is a strictly positive function on $S^3$ such there exist two Hopf circles $\gamma_1$ and $\gamma_2$ for which $$ \int_{\gamma_1} f^{-1} \alpha \neq \int_{\gamma_2} f^{-1} \alpha \ , $$ then the vector field $fX$ is not the Reeb vector field of any contact form on $S^3$.

The key idea is that all solutions of $\dot{x}(t) = f(x(t))X(x(t))$ are periodic and that if $fX$ were a Reeb vector field, all these solutions would have the same period.

Indeed, assume $fX$ is the Reeb vector field of a contact form $\beta$. If $\gamma_1$ and $\gamma_2$ are integral curves of $fX$ with periods $T_1$ and $T_2$, then $$ T_2 - T_1 = \int_{\gamma_2} \beta - \int_{\gamma_1} \beta = \int_\Sigma d\beta \ , $$ where $\Sigma$ is a cylinder with boundary $\gamma_2 - \gamma_1$ and foliated by Hopf circles (I'm being a bit sloppy with the word foliated). It follows that this last integral is zero and so $T_1 = T_2$.

Now we just have to notice that the period of $\gamma_i$ ($i = 1, 2$) equals the integral of $f^{-1}\alpha$ over $\gamma_i$: $$ \int_{\gamma_i} f^{-1} \alpha = \int_0^{T_i} f^{-1}(\gamma_i(t))\alpha(\dot{\gamma_i}(t)) \, dt = \int_0^{T_i} f^{-1}(\gamma_i(t))\alpha(f(\gamma_i(t)) X(\gamma_i(t))) \, dt = T_i $$


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.