2
$\begingroup$

Is there a standard or "simple" contact structure on the $3$-dimensional torus $T^3$, like there are for example for the Eucliden space and the $3$-sphere?

My first thought was to consider a structure descending to $T^3$ from the standard structure defined by the form $$ \omega_{std} = dz + x\wedge dy$$ on $\mathbb{R^3}$, but this doesn't work because $\omega_{std}$ is not invariant under the action of $\mathbb{Z}^3$. Also, if such a structure were to descend to $T^3$, the Reeb flowlines would consist of periodic curves parallel to the $\partial_z$ direction, and in particular it would be transversal to a $2$-dimensional torus $T^2$. That, on the other hand, can not happen, because the contact condition would imply the existence of an exact volume form on $T^2$, contradicting Stokes theorem.

I know there are contact structures on $T^3$ because of a theorem of Thurston's that states that every closed smooth $3$ manifold supports a contact structure, but I couldn't find any explicit examples for the torus. I am particularly interested in explicit descriptions of the associated Reeb field and its flow.

$\endgroup$
1
  • 1
    $\begingroup$ Yes, look up Luttinger surgery and Fintushel-Stern's Lagrangian tori invariants and related papers of Eliashberg. It's a crucial reason we have all these exotic 4-manifolds. There are very explicit descriptions of the "standard" contact structure which correspond to a geodesic flow on the 2-torus factor (i.e. Reeb orbits in $T^2\times S^1$ are lifts of geodesics in $T^2$). $\endgroup$ Aug 30, 2022 at 18:34

1 Answer 1

3
$\begingroup$

Consider the $1$-form on $\mathbb{R}^3$ given by $$ \alpha = \cos(2\pi z) dx + \sin(2\pi z) dy, $$

where $(x,y,z)$ are the standard coordinates on $\mathbb{R}^3$. This is a contact form on $\mathbb{R}^3$ with Reeb vector field given by $$R_{\alpha} = \cos(2\pi z) \partial_x + \sin(2\pi z) \partial_y.$$ This vector field can be integrated explicitly and its integral curve passing through $(x_0,y_0,z_0)$ at time $t=0$ is given by $$\phi(t) = (x_0 + \cos(2\pi z_0)t, y_0 + \sin(2\pi z_0)t, z_0). $$

Consider the standard action of $\mathbb{Z}^3$ on $\mathbb{R}^3$ by translations. The $1$-form $\alpha$ is invariant under this action and, therefore, descends to a $1$-form on the quotient $\mathbb{R}^3/\mathbb{Z}^3 \simeq T^3$. This gives an example of a contact structure on $T^3$.

More generally, given any $k \in \mathbb{Z} \smallsetminus \{0\}$, one can define a contact form on $\mathbb{R}^3$ by $$ \cos(2\pi k z) dx + \sin(2\pi k z) dy. $$ This contact form is invariant under the above action of $\mathbb{Z}^3$ and, therefore, descends to a contact form on $T^3$.

$\endgroup$
1
  • $\begingroup$ Dou you know if these contact forms on $\mathbb{R}^3$ and $\mathbb{T}^3$ are overtwisted? $\endgroup$ Jan 23, 2023 at 15:12

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.