Higher \(P\)-symmetric Ekeland-Hofer capacities (Q2669238)

Symplectic capacities were first introduced by \textit{I. Ekeland} and \textit{H. Hofer} [Math. Z. 200, No. 3, 355--378 (1989; Zbl 0641.53035); Math. Z. 203, No. 4, 553--567 (1990; Zbl 0729.53039)]. Since then, many new capacities have been defined and they are important in the quantitative study of symplectic geometry. The paper under review is concerned with the definition and computations of higher Ekeland-Hofer capacities of partially symmetric (\(P\)-symmetric) subsets in the standard symplectic space \((\mathbb{R}^{2n},\omega_0)\). It is built upon and motivated by the article [\textit{Y. Dong} and \textit{Y. Long}, J. Differ. Equations 196, No. 1, 226--248 (2004; Zbl 1109.37048)] which studies closed characteristics on \(P\)-symmetric compact convex hypersurfaces in \((\mathbb{R}^{2n}, \omega_0)\). Letting \(P := \operatorname{diag}(-I_{n-\kappa},I_\kappa,-I_{n-\kappa},I_\kappa)\) for some \(\alpha \in [0,n)\) we say that a subset \(A \subset \mathbb{R}^{2n}\) is \(P\)-symmetric if \(PA = A\). Furthermore let \[ \mathcal B(\mathbb{R}^{2n}) := \left\{B \subset \mathbb{R}^{2n} \mid PB=B, \; B \cap \operatorname{Fix}(P) = \varnothing\right\}\, . \] A closed characteristic \(z \colon \mathbb{R}/T\mathbb{Z} \longrightarrow \mathcal S\) of a \(P\)-symmetric hypersurface \(\mathcal S \subset \mathbb{R}^{2n}\) is called \(P\)-symmetric if \(z(t+ \frac T2) = Pz(t)\). The action of such a closed characteristic is defined by \(\mathcal A(z) = \frac 12 \int_0^T \langle -J_0 \dot z, z\rangle dt\). Consider the action spectra \begin{align*} \varSigma_{\mathcal S} &:= \left\{ k \mathcal A(x) > 0 \mid x \text{ is a closed characteristic of } \mathcal S, \; k \in \mathbb{N}\right\} \\ \varSigma_{\mathcal S}^P &:= \left\{ k \mathcal A(x) > 0 \mid x \text{ is a \(P\)-symmetric closed characteristic of } \mathcal S, \; k \in \mathbb{N}\right\}\, . \end{align*} The main construction of the paper is that of (higher) \(P\)-symmetric Ekeland-Hofer capacities \[ c^j_P \colon \mathcal B(\mathbb{R}^{2n}) \longrightarrow [0,\infty], \quad j = 1,2,\ldots\, , \] which implicitly depends on the integer \(\kappa\in [0,n)\) in the definition of \(P\) above. The maps \(c^j_P\) satisfies the requirements of being symplectic capacities by the work of \textit{Y. Dong} and \textit{Y. Long} [J. Differ. Equations 196, No. 1, 226--248 (2004; Zbl 1109.37048)]. The authors prove the following \begin{itemize} \item[1.] For a \(P\)-symmetric bounded convex domain \(D \subset \mathbb{R}^{2n}\) with boundary \(\mathcal S = \partial D\) of class \(C^{1,1}\). If \(D\) contains a fixed point of \(P\), then \[ c^1_P(D) = c^1_P(\mathcal S) = \min \varSigma^P_{\mathcal S}\, . \] \item[2.] \(c^j_P(\mathcal S) = c^j_P(B_{\mathcal S}) \in \varSigma^P_{\mathcal S}\) where \(B_{\mathcal S}\) is the bounded component of \(\mathbb{R}^{2n} \setminus \mathcal S\) for \(j = 1,2,\ldots\). \end{itemize} Furthermore when \(j = 1\), the authors prove that if \(D \in \mathcal B(\mathbb{R}^{2n})\) is a bounded convex domain, then \[ \begin{cases} c^1_P(D) = c^1_{\text{EH}}(D), & \kappa = 0\\ c^1_P(D) = 2c^1_{\text{EHZ}}(D), & 1\leq \kappa < n \end{cases} \] where \(c^1_{\text{EH}}\) is the Ekeland-Hofer capacity and \(c^1_{\text{EHZ}}\) is the generalized Ekeland-Hofer-Zehnder capacity defined in [\textit{R. Jin} and \textit{G. Lu}, ``Generalizations of Ekeland-Hofer and Hofer-Zehnder symplectic capacities and applications'', Preprint, arXiv:1903.01116]. The authors compute the \(P\)-symmetric Ekeland-Hofer capacities for ellipsoids and polydisks. Furthermore they compute \(c^2_P(D^2 \times_L D^2) = 8\) and give an estimate for the value of \(c^1_P(D^2 \times_L D^2)\), where \[ D^2 \times_L D^2 := \left\{(x_1,x_2,y_1,y_2) \mid x_1^2+x_2^2 < 1, \; y_1^2+y_2^2 < 1\right\} \subset \mathbb{R}^4\, . \]