On the symplectic fillings of standard real projective spaces (Q2139552)

A fundamental problem in contact and symplectic topology is that of understanding (strong) symplectic fillings of a given contact manifold. As there is a simple operation, namely symplectic blow-up, that gives new symplectic fillings (with more complicated topology) starting from any given one, in studying this problem one must focus on the ``simplest'' symplectic fillings. In high dimensions, the word ``simplest'' means usually \textit{symplectically aspherical}, i.e., without symplectic (embedded) spheres; indeed, in this setting the study of moduli spaces of pseudo-holomorphic spheres can be done more effectively as there is more control on nodal degenerations. Two natural questions can then be studied in this direction: are there any symplectically aspherical filling of a given contact manifold? And if so, ``how many'' (e.g./ up to symplectomorphism, diffeomorphism, homotopy, homology, etc)? The paper under review is concerned with both of these questions. The first and main result by the authors is the non-existence of symplectically aspherical fillings of the (standard) contact projective space \(\mathbb{R}P^{2n-1}\) for \(n\geq 3\) and odd. The proof passes through the study of a moduli space of pseudo-holomorphic spheres in a closed symplectic manifold coming from a hypothetical filling \(W\) of \(\mathbb{R}P^{2n-1}\). More precisely, by performing a symplectic reduction at the boundary \(\mathbb{R}P^{2n-1}\) of \(W\), one gets a closed symplectic manifold \(X^{2n}\) together with a distinguished symplectic divisor \(\mathbb{C}P^{n-1}\). Given a point and a complex hyperplane in the divisor, the authors then consider the moduli space of pseudo-holomorphic spheres passing through them. By analysing carefully the possible degenerations of families in this moduli space, the authors then deduce that the homology class of a complex line in the divisor must be twice a homology class in \(X\) up to torsion. As the Chern class of the complex line is \(n+2\), this gives a contradiction under the assumption of \(n\) odd. (It should be noted that, after a preliminary draft by the authors proving the above result for the case of \(\mathbb{R}P^5\), Zhou proved, using instead subtle properties and computations of symplectic cohomology, that the statement in fact holds more generally for \(\mathbb{R}P^{2n-1}\) with \(n\neq 2^k\); see [\textit{Z. Zhou}, Geom. Topol. 25, No. 6, 3013--3052 (2021; Zbl 1486.53090)].) The second result in the paper is that any semipositive symplectic filling \(W\) of \(\mathbb{R}P^{2n-1}\) is simply connected. Semipositivity is a technical condition concerning symplectic spherical homology classes, which is typically used to guarantee that the undesired nodal degenerations of families of pseudo-holomorphic spheres happen only in codimension at least \(2\); moreover, a symplectically aspherical filling is in particular semipositive. As the authors point out, their result is likely to also hold (at least on a homological level) for more general symplectic fillings using abstract perturbation techniques. The proof of this second result has the same setup as that of the first one, i.e., one looks at the closed symplectic manifold \(X\) obtained by symplectic reduction at the boundary \(\mathbb{R}P^{2n-1}\) of \(W\). Then, the authors argue, using the fact that the codimension of the bad degenerations is at least \(2\), that any loop in \(X\setminus \mathbb{C}P^{n-1}=W\setminus \mathbb{R}P^{2n-1}\) can be pulled back under the evaluation map \(ev\) the (smooth part of the) marked moduli space of spheres considered in the previous proof. As these spheres have a unique intersection point with the hyperplane in the divisor of \(X\), the pulled back loop can be homotoped in the marked moduli space to the preimage of the neighborhood of the divisor via \(ev\). Taking the image via \(ev\) of this homotopy then gives the homotopy from the original loop to another one near the divisor in \(X\); in other words, \(\mathbb{R}P^{2n-1}\) surjects at the \(\pi_1\)-level onto \(W\), thus concluding. It is worth pointing out that the strategy of the proofs of the two above results allow the authors to give yet another proof of Eliashberg-Floer-McDuff's result (see Theorem 1.5 of [\textit{D. McDuff} and \textit{D. Salamon}, \(J\)-holomorphic curves and symplectic topology. 2nd ed. Providence, RI: American Mathematical Society (AMS) (2012; Zbl 1272.53002)]) on the uniqueness of the diffeomorphism type of symplectically aspherical symplectic fillings of the standard \((2n-1)\)-sphere.