Compactness for punctured holomorphic curves (Q862194)

A proof of the compactness theorem for moduli spaces of punctures holomorphic curves arising in symplectic field theory \textit{F. Bourgeois, Y. Eliashberg, H. Hofer, K. Wysocki} and \textit{E. Zehnder} [Geom. Topol. 7, 799--888 (2003; Zbl 1131.53312)] hereafter refered to as [1]) is presented. Roughly speaking, the compactness theorem asserts the existence of a subsequence which converges to a stable broken holomorphic curve of the sequence \(f_k:\Sigma_k\to X\), where \(X\) is a closed symplectic manifold with a tamed almost complex structure, the \(\Sigma_k\)'s are holomorphic curves of the same genus such that the areas \(\int_{\Sigma_k}f^*_k \omega\) are uniformly bounded. The exact statement of the compactness theorem (Theorem 1 and Theorem 2.9) and the meaning of convergence of the sequence \(f_k:\Sigma_k\to X\) are stated in introduction and \S2. The authors say the proof in [1] and the proof in this paper are orthogonal. The proof of the compacness theorem in [1] focuses on the domains, ignoring the level structure in the image (and reconstructing it a posteriori), while the proof in this paper focuses on the level structure in the image, ignoring the domains (and reconstruct them a posteriori). Technically, the proof in [1] is based on repeated bubbling off analysis at points where the gradient explodes, while the proof in this paper follows the strategy of \textit{M. Gromov} [Invent. Math. 82, 307--347 (1985; Zbl 0592.53025)], hereafter refered to as [2]). In \S 2.8--\S 2.10, changes of action and topology of sub-level sets are analyzed. They reduce the proof to two separate cases: long cylinders of small area and regions with compact image. Uniform convergence of holomorphic cylinders of small area is proved in \S 4 (Theorem 4.18), by using the fact that the \(M\)-components of long holomorphic cylinders approach closed Reeb orbit (Prop. 4.5. cf. Prop. 5.7 of [1] and \textit{H. Hofer, K. Wysocki} and \textit{E. Zehnder} [Ergodic Theory Dyn. Syst. 22, No. 5, 1451--1486 (2002; Zbl 1041.57009)]. The second case requires to generalize Gromov compactness ([2]) to holomorphic curves with free boundary. This is done in \S 3. The result is stated as Theorem 3.2, which says that a sequence of holomorphic curves \(f_k:\Sigma_k\to X\), \(X\) not necessarily closed, converges to a nodal holomorphic curve provided the boundary circles have disjoint collars whose modulus is uniformly bounded below and the differentials of \(f_k\) on these collars are uniformly bounded below and above and the areas of \(\Sigma_k\) are uniformly bounded and their images are contained in a compact set. After these preparations, the compactness theorem (Theorem 2.9) is proved in \S 5. As a useful case of the compactness theorem, it is shown that if the genus is zero and all the \(f_k\) represent a homology class \(A\) which cannot be written as \(A=B+C\), \(\omega(B) \cdot\omega(C)>0\), in \(H_2(X;\mathbb{Z})\), then the limit of the subsequence is a broken holomorphic curve without node (Corollary 2.10). It is remarked that if \(X\) is not closed but already a symplectic cobordism, replacing the uniform area bound condition by a uniform bound condition of the Hofer energy \[ E(f)=\int_\Sigma f^*\omega+\sup_{\varphi\in S} \int_\sigma f^*\bigl(\varphi'(r) dr\wedge\lambda \bigr), \] where \(\lambda\) is the associated 1-form of an \(\omega\)-tamed translation invariant almost complex structure on \(\mathbb{R}\times M\), \(M\) a compact odd-dimensional manifold with a stable Hamiltonian structure \(\omega\) (cf. \S 4.1), everything in this paper remains valid (Remark 2.15). Hence the proof of compactness theorem can be adopted for a general symplectic cobordism ([1], cf. \S 2.4).