Weak and strong fillability of higher dimensional contact manifolds (Q2377347): Difference between revisions

The study of 3-dimensional contact geometry has been governed by notions such as tight and overtwisted, weak and strong symplectic fillability, and Giroux torsion. This groundbreaking paper attempts to give the right generalizations of these notions for higher-dimensional contact geometry. It is hard to describe its content in a short review as it contains a large number of definitions, constructions and results. In dimension 3 one distinguishes between strong and weak symplectic fillings \((W,\omega)\) of a contact manifold \((M=\partial W,\alpha=ker(\xi))\), where the former means that \(\omega=d\alpha\) in a neighborhood of \(\partial W\) while the latter only demands \(\omega\mid_\xi=d\alpha\mid_\xi\). McDuff has shown that in dimensions \(2n+1\geq 5\) these two conditions are equivalent, and the authors therefore consider a weaker notion of weak fillability which boils down to the inequality \(\alpha\wedge(w\mid_\xi+\tau d\alpha)^n>0\) for all \(\tau\geq0\) to hold. With this weaker definition, they prove that a symplectic manifold is a weak filling if and only if it admits an almost complex structure \(J\) tamed by \(\omega\) that makes \((M,\xi)\) the pseudoconvex boundary of \((W,\omega)\). A higher-dimensional generalization of overtwisted disks had been introduced under the name ``plastikstufe'' by [\textit{K.\ Niederkrüger}, Algebr. Geom. Topol. 6, 2473-2508 (2006; Zbl 1129.53056)]. In this paper the authors introduce a topologically much less restrictive notion of ``PS-overtwistedness'', namely the existence of a ``bordered Lagrangian open book'' (bLob), that is an \((n+1)\)-dimensional submanifold with Legendrian boundary which is a relative open book with Legendrian fibers. They prove that a PS-overtwisted contact manifold can not have a weak symplectic filling \((W,\omega)\) such that the restriction of \(\omega\) to the bLob is exact. In dimension 3, the emblematic examples of tight but not fillable contact structures are the contact structures \(\xi_k=ker(\cos ks d\theta+\sin ks dt)\) on the \(3\)-torus which are fillable only for \(k=1\). This led to the definition of Giroux torsion of \((M,\xi)\) as the maximal \(n\), such that \((T^2\times I,\xi_{2n})\) can be embedded into \((M,\xi)\). In the paper under review the authors introduce a higher-dimensional generalization of this set of examples under the name ``Giroux domains''. Giroux domains are constructed as follows. One starts with an oriented \((2n-1)\)-manifold that admits a ``Liouville pair'' \(\alpha_\pm\), that is a pair of contact forms such that \(\pm\alpha_\pm\wedge d\alpha_\pm^{n-1}>0\) and such that the \(1\)-form \(\beta:=e^{-s}\alpha_-+e^s\alpha_+\) on \({\mathbb R}\times M\) satisfies \((d\beta)^n>0\). The associated Giroux domain is then the manifold \(\left[0,\pi\right]\times S^1\times M\) with the contact form \(\lambda_{GT}=\frac{1+\cos s}{2}\alpha_++\frac{1-\cos s}{2}\alpha_-+\sin s dt\). The authors describe a number of explicit constructions of Liouville pairs and thus Giroux domains. The obstruction to fillability is then given by the theorem that a contact manifold containing a submanifold with nonempty boundary obtained by gluing together two Giroux domains can not be strongly fillable. (Under a certain homological assumption, for example if \(H_1(M)=0\), the manifold is also not weakly fillable.) The authors also define what they call ``Lutz-Mori twists'' to homotope such a contact structure \(\xi\) into a family of contact structures \(\xi_k\). As a result the authors obtain the following theorem. For any closed manifold \(M\) with a Liouville pair \((\alpha_+,\alpha_-)\) there is a sequence \(\left\{\xi_k\right\}_{k>0}\) of contact structures on \(T^2\times M\) such that \(\xi_1\) is exactly fillable (that is strongly fillable and the contact form extends globally over \(W\)), \(\xi_k\) is not strongly fillable for any \(k\geq 2\), all \(\xi_k\) and \(\xi_l\) are homotopic through a family of almost contact structures, and if \((\alpha_+,\alpha_-)\) is hypertight, then all \(\xi_k\) are hypertight (allow Reeb vector fields without contractible closed orbits) and no two of them are isotopic (and if moreover \(\pi_1M\) has trivial center then no two of them are contactomorphic). Moreover they show that there are certain \(3\)-manifolds with Liouville pairs such that the resulting \(5\)-dimensional contact manifolds \((T^2\times M,\xi_k)\) are all weakly fillable, thus producing sequences of \(5\)-manifolds that are weakly but not strongly fillable.