Compactness for holomorphic curves with switching Lagrangian boundary conditions (Q711543)

Let \((X,\omega)\) be a symplectic manifold with an \(\omega\)-compatible almost complex structure \(J\), \(L\) be an immersed real analytic Lagrangian submanifold of \(X\) with clean self-intersection along a compact submanifold \(K\) such that \(J\) is integrable near \(K\). Let \((S, j)\) be a connected Riemann surface with boundary \(\partial S\). A holomorphic curve \(f: (S,\partial S, j)\to(X,L,J)\) is a continuous map \(f: S\to X\) which maps \(\partial S\) to \(L\) and is \((j,J)\)-holomorphic in the interior. \((X,L,\omega,J)\) may be noncompact with cylindrical ends and \(S\) may have punctures in the interior as well as on the boundary. Intersections of \(f|_{\partial S}\) with \(K\) are called switches. They are not treated as boundary punctures. Assuming \(X=\overline X\cup(\mathbb{R}_+\times M)\), \(\partial\overline X= M\), the energy \(E(f)\) of \(f\) is a sum \(E_\omega(f)+ E_\lambda(f)\), where \[ \begin{aligned} E_\omega(f)&= \int_{f^{-1}(X)} f^*\omega+ \int_{f^{-1}(\mathbb{R}_+\times M)} f^*_M\omega,\\ E_\lambda(f)&= \sup_{\phi\in{\mathcal C}^+} \int_{f^{-1}(\mathbb{R}_+\times M)} (\phi\circ f_{\mathbb{R}})\,df_{\mathbb{R}}\wedge f^*_M\lambda. \end{aligned} \] Here, \({\mathcal C}^+\) means the set of nonnegative functions \(\phi\) on \(\mathbb{R}_+\) such that \(\int^\infty_0 \phi(s)\,ds= 1\). Then, it is shown: Theorem 1.1. Under suitable hypotheses on \((X,L,\omega, J)\), each sequence of holomorphic curves \(f_n\) of finite signature and uniformly bounded energy has a subsequence converging to a stable holomorphic curve. The authors remark that, if the number of switches where the map changes branches is a priori bounded by \(M\), this theorem has already been proved by \textit{S. Ivashkovich} and \textit{V. Shevchishin} [Commun. Contemp. Math. 4, No.~1, 65--106 (2002; Zbl 1025.32024)] (Th. 1.2). But in Theorem 1.1, boundedness of the numbers of switches is not assumed. As a consequence of this theorem, the following finiteness result for the number of switches, according to the authors, the motivation of this work, are proved. Theorem 1.2. Assume \((X,L,\omega= d\lambda)\) satisfies the hypotheses of Theorem 1.1 and is exact with convex end. Then, for each \(s\in\mathbb{N}\) and \(C> 0\), there exists a constant \(\kappa(s,C)\in\mathbb{N}\) such that every holomorphic disc \(f: ((\dot D),\partial\dot D,j)\to (K,L.J)\) with at most a boundary punctures and energy \(\leq C\) has at most \(\kappa(s,C)\) switches. Several hypotheses on \((X,L,\omega,J)\) (indexed \((X)\), \((Y)\), \((L)\), \((K)\) and \((R)\)) are described in \S3. \((X)\), \((L)\) and \((K)\) are refinements of the above statements on \(X\), \(L\) and \(K\). \((Y)\) asserts the existence of a compact subset \(\overline Y\) of \(\overline X\) such that every \(J\)-holomorphic maps \(f: S\to X\) from a compact Riemann surface with boundary satisfying \(f(\partial S)\subset Y=\overline Y\cup(\mathbb{R}_+ \times (\overline Y\cap M)\) is entirely contained in \(Y\). \((R)\) asserts that no Reeb orbit meets \(\Lambda\) and that all closed Reeb orbits and Reeb chords are nondegerate. Here, \(\mathbb{R}_+\times\Lambda= L\cap (\mathbb{R}_+\times M)\) and a Reeb chord is a Reeb orbit \(\gamma: [0,T]\to M\) with \(\gamma(0)\), \(\gamma(T)\in\Lambda\). As for symplectic forms, \(\lambda|_\Lambda= \omega|_\Lambda=0\) is assumed in \((L)\). \((K)\) asserts existence of local biholomorphism which maps a self intersection to the coordinate axes. An example of such an \((X,L,\omega,J)\) is \(X= T^*Q\), \(Q\) a Riemannian manifold with the Liouville one-form \(\lambda=pdq\), \(\omega= d\lambda\), \(J\) the almost complex structure on \(T^*Q\) induced by the Riemannian metric, and \(L= Q\cup NK\), \(K\) is a compact submanifold of \(Q\) and \(NK\) is its conormal bundle. \((K)\) is the condition that \(L\) is real analytic near \(K\) with \(J\)-orthogonal self-intersection along \(K\). But if \(L\) is two-sided near \(K\), \((K)\) can be arranged by a deformation of the compatible almost complex structure near \(K\) (Proposition 3.1. To avoid later confusion, it is better to change the numbering of this proposition from 3.1 to 3.0). The definition and properties of the energy of \(f\) are also given in \S3, showing the monotonicity of the energy (Lemma 3.4). By this Lemma, a continuous map \(f:D\setminus\{0\}\to X\) of finite energy which is holomorphic in the interior is shown to be extendable to a continuous map \(D\to X\) (Lemma 3.5). Then, asymptotic properties of a holomorphic curve \(f\) with finite energy and unbounded image (Proposition 3.1. This is another Proposition 3.1, and the later references to Proposition 3.1 in this paper mean this Proposition.) and statements on Reeb orbits and chords of the holomorphic cylinder and strip [Proposition 3.2. cf. \textit{F. Bourgeois}, \textit{Y. Eliashberg}, \textit{H. Hofer}, \textit{K. Wysocki} and \textit{E. Zehnder}, Geom. Topol. 7, 799--888 (2003; Zbl 1131.53312), hereafter referred to as [1]; \textit{H. Hofer}, \textit{K. Wysocki} and \textit{E. Zehnder}, Ergodic Theory Dyn. Syst. 22, No.~5, 1451--1486 (2002; Zbl 1041.57009)] are given. Before preparing this global theory, the properties of functions \(f: S\to\mathbb{C}\), where \((S, j)\) is a connected Riemann surface with boundary \(\partial S\), \(f\) is continuous, holomorphic on \(S\setminus\partial S\) and mapping \(\partial S\) to \(\mathbb{R}\cup i\mathbb{R}\) are treated as the local theory in \S2. The main results are sequential compactness of \(C^0\)-bounded set of these functions (Lemma 2.4) and a variant of Vitalli's theorem, which gives a vanishing condition of the limit of \(C^0\)-bounded series of these functions (Lemma 2.5). After these preparations, Theorem 1.1 is proved in \S4, first, under additional assumptions on the Reeb chords (and the closed Reeb orbits), and estimates of divergences of \(f_n\) by using Lemma 2.4. Then, adopting the proof of compactness for holomorphic curves in symplectic field theory ([1]), additional assumptions are shown to be satisfied under the assumptions of Theorem 1.1. Theorem 1.2 is proved in \S5 by using Theorem 1.1 and Lemma 2.5. In Appendix (\S6), moduli spaces \({\mathcal M}(f)\) of holomorphic curves in \(X\) with smooth boundary on \(L\), interior punctures asymptotic to closed Reeb orbits, boundary punctures asymptotic to Reeb cords and Lagrangian intersetion punctures are defined and their formal dimensions are computed (Theorem A.1) applying an index theorem on capping operators \(\overline\partial^{\pm}_\alpha\) [cf. \textit{F. Bourgeois} and \textit{K. Mohnke}, Math. Z. 248, No.~1, 123--146 (2004; Zbl 1060.53080)].