Non-trapping estimates near normally hyperbolic trapping (Q2345432)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Non-trapping estimates near normally hyperbolic trapping
scientific article

    Statements

    Non-trapping estimates near normally hyperbolic trapping (English)
    0 references
    0 references
    0 references
    22 May 2015
    0 references
    In two pseudodifferential frameworks (semiclassical and b-calculus), the authors derive resolvent estimates for pseudodifferential operators \(P\) under assumptions of normally hyperbolic trapping. These assumptions essentially say that the associated Hamiltonian dynamics has an invariant set \(\Gamma\) which is normally hyperbolic, and that the incoming/outgoing tails \(\Gamma_\pm\) are of codimension \(1\) and intersect symplectically at \(\Gamma\). Normally hyperbolic trapping holds for Kerr-de Sitter spaces; see \textit{J. Wunsch} and \textit{M. Zworski} [Ann. Henri Poincaré 12, No. 7, 1349--1385 (2011; Zbl 1228.81170)]. In the semiclassical case, \(P=P_h(z)\in\Psi_h(X)\) is a family of operators which is formally selfadjoint for real parameters \(z\). The main result, stated in Theorem 1.1 of the paper, is an estimate \[ \|u\|_{\mathcal{H}_{h,\Gamma}} \leq Ch^{-1} \|P_h(z) u\|_{\mathcal{H}_{h,\Gamma}^*}. \tag{*} \] If the norms were \(L^2\)-norms \(\|\cdot\|\), such an \(O(h^{-1})\) estimate would be called lossless because it would translate to regularity estimates with no loss of Sobolev regularity. It is known that in the presence of trapping lossless \(L^2\) estimates do not hold. The Hilbert spaces \(\mathcal{H}_{h,\Gamma}\) and their (starred) duals, which occur in \((*)\), are equal to \(L^2\) microlocally outside \(\Gamma\). Near \(\Gamma\) the squared \(\mathcal{H}_{h,\Gamma}\)-norm is \(\|Q_+u\|^2+\|Q_-u\|^2\), where \(Q_\pm\in\Psi_h(X)\) have principal symbols which are defining functions of \(\Gamma_\pm\). Therefore (*) improves the \(O(h^{-1})\) estimate of a cutoff resolvent obtained in [\textit{K. Datchev} and \textit{A. Vasy}, Ann. Inst. Fourier 62, No. 6, 2347--2377 (2012; Zbl 1271.58014); addendum ibid. 62, No. 6, 2379--2384 (2012)]. In Section 3 of the paper, \(P\) is a b-pseudodifferential operator on a manifold \(M\) with boundary \(X\). The characteristic set \(\Sigma\) is a subset of b-cosphere bundle \({}^{\roman b}S^*M\). Normal hyperbolic trapping, \(\Gamma=\Gamma_+\cap\Gamma_-\), \(\Gamma_\pm\subset\Sigma\), is assumed to occur above \(X\). The main result, Theorem 3.2 of the paper, is an estimate in terms of (weighted) b-Sobolev norms which degenerate at \(\Gamma\) in fashion similar to the semiclassical case. The author's motivation for the result is its application to the analysis of linear and non-linear waves in general relativity. The proofs involve commutator estimates. The semiclassical case is viewed as a dilation invariant special case of the boundary case.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    semiclassical resolvent estimates
    0 references
    trapping
    0 references
    normal hyperbolicity
    0 references
    propagation of singularities
    0 references
    b-calculus
    0 references
    0 references