Middle dimensional symplectic rigidity and its effect on Hamiltonian PDEs (Q2301920)

The author proves several inequalities involving Viterbo's symplectic capacities and their relation with symplectic reduction. The inequalities are used to prove a middle dimensional coisotropic camel theorem that applies to Hamiltonians generated by sub-quadratic Hamiltonians. The second part of the article extends the results to infinite-dimensional Hilbert spaces via finite-dimensional reduction arguments. Gromov's non-squeezing theorem says that a symplectomorphism of \((\mathbb{R}^{2n},\omega_0)\) cannot send a ball of radius \(r\) into a symplectic cylinder of radius \(R\) if \(R < r\). Eliashberg and Gromov showed that this is equivalent to \[ \mbox{area}(\Pi_1\phi(B_r^{2n})) \geq \pi r^2 \quad \mbox{for every symplectmorphism } \phi, \] where \(\Pi_1:\mathbb{R}^{2n} \rightarrow \mathbb{R}^2\) denotes projection onto the first 2 coordinates of \(\mathbb{R}^{2n}\) and \(B_r^{2n}\) denotes the ball of radius \(r\). A ``middle dimensional'' generalization of this statement might be \[ \mbox{Vol}(\Pi_k\phi(B_r^{2n})) \geq \mbox{Vol}(\Pi_k B_r^{2n}) = \mbox{Vol}(B_r^{2k}) \quad \mbox{for every symplectmorphism } \phi, \] where \(\Pi_k:\mathbb{R}^{2n} \rightarrow \mathbb{R}^{2k}\) denotes projection onto the first \(2k\) coordinates of \(\mathbb{R}^{2n}\). \textit{A. Abbondandolo} and \textit{R. Matveyev} proved that this inequality is false in general [J. Topol. Anal. 5, No. 1, 87--119 (2013; Zbl 1268.37074)], which shows that middle dimensional volume symplectic rigidity does not hold in general. In this paper, the author proves a similar inequality involving the capacities \(c\) and \(\gamma\) defined by \textit{C. Viterbo} [Math. Ann. 292, No. 4, 685--710 (1992; Zbl 0735.58019)]. The main theorem in the first half of the paper concerns symplectomorphisms generated by sub-quadratic Hamiltonians, i.e., functions \(H:\mathbb{C} \times \mathbb{R} \rightarrow \mathbb{R}\) that satisfy \[ \lim_{|z| \rightarrow +\infty} \frac{\nabla H_t(z)}{|z|} = 0 \quad \mbox{uniformly in } t. \] Theorem 1.1 (Coisotropic camel theorem). Let \(X \subset \mathbb{C}^k\) be a compact set. Consider \(X \times \mathbb{R}^{n-k} \subseteq \mathbb{C}^k \times \mathbb{C}^{n-k}\) and let \(\psi = \psi_1^H\) be a Hamiltonian diffeomorphism of \(\mathbb{C}^n\) generated by a sub-quadratic Hamiltonian \(H\). Then \[ c(X) \leq \gamma(\Pi_k(\psi(X \times \mathbb{R}^{n-k}) \cap \mathbb{C}^k \times i \mathbb{R}^{n-k})). \] If we let \(W = \mathbb{C}^k \times i \mathbb{R}^{n-k}\), then this inequality can be written as \[ c(X) \leq \gamma(\mbox{Red}_W(\psi(X \times \mathbb{R}^{n-k}))), \] where \(\mbox{Red}_W\) denotes the symplectic reduction by \(W\). Moreover, the above inequality and the monotonicity of \(\gamma\) imply that \[ c(X) \leq \gamma(\Pi_k \psi(X \times \mathbb{R}^{n-k})). \] In Remark 1.3, the author points out that Theorem 1.1 does not hold for general symplectomorphisms, with a counterexample for the previous weaker inequality given by rearranging the coordinates. In the second part, the author studies middle dimensional symplectic rigidity in infinite-dimensional Hilbert spaces. Theorem 1.1 is extended to Hilbert spaces for semilinear PDEs of the type described in [\textit{S. B. Kuksin}, Commun. Math. Phys. 167, No. 3, 531--552 (1995; Zbl 0827.35121)]. Examples of such equations include a nonlinear string equation on \(\mathbb{T}\), a quadratic nonlinear wave equation on \(\mathbb{T}^2\), a nonlinear membrane equation on \(\mathbb{T}^2\), and the Schrödinger equation with a convolution nonlinearity on \(\mathbb{T}^n\). With the right compactness assumptions on the nonlinearity, the flow maps can be approximated on bounded sets by finite dimensional symplectomorphisms, and Theorem 1.1 can be applied. For concreteness, the author limits the discussion to the periodic nonlinear string equation. He defines a product of Hilbert spaces \(E = E_+ \times E_- = H^{\frac{1}{2}}(\mathbb{T}) \times H^{\frac{1}{2}}(\mathbb{T})\) and proves the following theorem: Theorem 1.4. Denote by \(\Phi^t:E \rightarrow E\) the flow of the nonlinear string equation satisfying the previous hypothesis. For every \(k \in \mathbb{N}\), every compact subset \(X\) of \(E_k\) and every \(t \in \mathbb{R}\), we have \[ c(X) \leq \gamma(\Pi_k \Phi^t(X \times E^k_+)). \] The paper contains two appendices containing results used to show that A) Theorem 1.1 does not hold for general symplectomorphisms and B) the set of diffeomorphisms for which Theorem 1.1 does hold is strictly bigger than the group of compactly supported Hamiltonian diffeomorphisms. Proposition A.1. Consider the coisotropic subspace \(\mathbb{C}^k \times \mathbb{R}^{n-k} \subseteq \mathbb{C}^k \times \mathbb{C}^{n-k}\) with \(0 \leq k < n\). We have \[ c(\mathbb{C}^k \times \mathbb{R}^{n-k}) = 0 = \gamma(\mathbb{C}^k \times \mathbb{R}^{n-k}). \] Proposition B.1. Denote by \(\mbox{Ham}^{\mbox{dL}}(\mathbb{C}^n)\) the set of Hamiltonian diffeomorphisms \(\varphi^H_t\) such that \(H_t\), \(\varphi^H_t\) and \((\varphi^H_t)^{-1}\) are all Lipschitz in space over compact time intervals. Then \(\mbox{Ham}^{\mbox{dL}}(\mathbb{C}^n)\) is a subgroup of \(\mbox{Sympl}(\mathbb{C}^n)\). Moreover \(\mbox{Ham}^{\mbox{dL}}(\mathbb{C}^n)\) is strictly bigger than the group of compactly supported Hamiltonian diffeomorphisms.