Bott-integrability of overtwisted contact structures

We show that an overtwisted contact structure on a closed, oriented 3-manifold can be defined by a contact form having a Bott-integrable Reeb flow if and only if the Poincaré dual of its Euler class is represented by a graph link.

💡 Research Summary

The paper establishes a complete classification of overtwisted contact structures on closed, oriented three‑dimensional manifolds that admit a Bott‑integrable Reeb flow. A contact form α is called Bott‑integrable if there exists a Morse–Bott function f that is invariant under the Reeb vector field R (i.e. df(R)=0). The main result, Theorem 1.2, states that an overtwisted contact structure ξ is Bott‑integrable if and only if the Poincaré dual of its Euler class e(ξ) can be represented by a graph link.

The authors first develop the local theory of Bott‑integrable Reeb flows. They prove neighbourhood theorems for critical tori (Theorem 2.2) and critical Klein bottles (Theorem 2.3), showing that near such a critical surface the contact form can be put into a standard Lutz‑type model. In these coordinates the Bott integral has the simple quadratic form f(r)=c±r², and the Reeb flow is tangent to the torus fibres.

Next, they show how to eliminate critical surfaces altogether. By a C^∞‑small perturbation of the Bott integral they replace each critical torus or Klein bottle by a pair of isolated periodic Reeb orbits—one elliptic and one hyperbolic—while preserving Bott‑integrability. Consequently the critical set of a generic Bott‑integrable Reeb flow consists solely of isolated periodic orbits, whose union is a link L_f.

Section 3 relates this critical link to the Euler class. Using the orientations of the elliptic and hyperbolic components, the authors prove that L_f represents the Poincaré dual of e(ξ). Hence any Bott‑integrable contact structure forces its critical link to be a graph link. This yields the “only‑if’’ direction of Theorem 1.2.

For the converse, the paper relies on Yano’s work on graph links in graph manifolds. Yano proved that in a graph manifold every homology class in H₁(M) can be represented by a graph link. The authors adapt this result to the contact setting: given a graph link L that represents PD(e(ξ)), they construct a Bott‑integrable contact form whose critical link contains L. The construction proceeds in several steps. First, on a Seifert‑fibered piece they produce a Bott‑integrable form whose critical link includes a prescribed sublink of L. Then they use a fiber‑connected sum operation for Bott‑integrable contact forms to glue together the local models across the JSJ tori. Throughout this process they repeatedly apply the neighbourhood theorems and the surface‑removal perturbations to keep the critical set purely periodic. The result is a global Bott‑integrable contact form on the whole manifold whose Euler class dual is exactly L, establishing the “if’’ direction.

Section 4 focuses on Seifert manifolds. The authors prove that every overtwisted contact structure on a closed Seifert manifold is Bott‑integrable (Corollary 4.4). This follows from the fact that any homology class on a Seifert manifold can be realized by a sublink of a critical link, combined with Eliashberg’s classification of overtwisted structures.

Section 5 supplies the necessary background on the JSJ decomposition, graph manifolds, and Yano’s homology‑completeness theorem, making the paper self‑contained for readers not specialized in 3‑manifold topology.

Finally, Section 6 presents a detailed example: the mapping torus of Arnold’s cat map. By analysing its JSJ decomposition and applying the previous constructions, the authors completely classify the Bott‑integrable contact structures on this graph manifold, including the tight ones. They also exhibit overtwisted structures that are not Bott‑integrable, thereby illustrating the sharpness of their main theorem.

Beyond the classification, the paper discusses dynamical consequences. By a theorem of Paternain, any Bott‑integrable Reeb flow has zero topological entropy. Consequently, Theorem 1.2 provides a criterion for an overtwisted contact structure to not force entropy, answering a question of Katok and of Alves–Colin–Honda in the negative for Seifert manifolds. Conversely, the authors exhibit overtwisted structures on certain graph manifolds that force any associated Reeb flow to have positive entropy, showing that the dichotomy “entropy forced vs. Bott‑integrable’’ is sharp.

In summary, the work bridges contact dynamics, Morse–Bott theory, and 3‑manifold topology, delivering a precise homotopy‑theoretic classification of Bott‑integrable overtwisted contact structures via the combinatorial notion of graph links. This advances our understanding of the interplay between contact geometry and low‑dimensional topology, and opens new avenues for studying entropy and integrability in Reeb dynamics.