Fractal Weyl law for open quantum chaotic maps (Q2436024)

The purpose of the article under review is the study of the semiclassical quantizations of open hyperbolic maps arising in chaotic scattering problems. More precisely, the general aim of the article is to study the spectral properties of \(\hbar\)-Fourier integral operators quantizing open hyperbolic (symplectic) maps. Such operators (or their projection on a finite dimensional space) are called by the authors hyperbolic open quantum maps, and holomorphic families of such maps are called hyperbolic quantum monodromy operators. The main result of the article (Theorem 4) gives a fractal Weyl upper bound on the number of ``resonances'' for hyperbolic quantum monodromy operators in terms of the Minkowski dimension of the trapped set. As an application of their general study of these hyperbolic open quantum maps, the authors derive a fractal Weyl upper bound in the case of scattering by several convex obstacles. Precisely, consider \(\mathcal{O}=\bigcup_{j=1}^J\mathcal{O}_j\) the union of open, bounded strictly convex subsets of \(\mathbb{R}^n\) with smooth boundaries and satisfying Ikawa condition \[ \overline{\mathcal{O}}_k\cap\text{convex hull}(\overline{\mathcal{O}}_l\cup \overline{\mathcal{O}}_j)=\emptyset,\;\;j\neq k\neq l. \] The classical Hamiltonian flow associated to this geometric situation is defined by free motion away from obstacles and normal reflections on the obstacles. The information that will be relevant in the study of scattering resonances will be contained in the geometry of the trapped set \(K\) which consists of points that do not escape at infinity when \(t\rightarrow\pm\infty\). In this geometric context, the open quantum maps will be associated to the billiard map on \(B^*\partial\mathcal{O}\) constructed from the classical Hamiltonian flow. One is then interested in counting the number of scattering resonances, i.e., counting with their multiplicity the poles of the meromorphic continuation of \[ R(\lambda):=(-\Delta-\lambda^2)^{-1}:L^2_{\mathrm{comp}}(\mathbb{R}^n\backslash\mathcal{O})\longrightarrow L^2_{\mathrm{loc}}(\mathbb{R}^n\backslash\mathcal{O}). \] Under the above geometric assumptions, the authors deduce from their main result the following fractal Weyl upper bound for any fixed \(\alpha>0\): \[ \sum_{-\alpha<\text{Im} \lambda, r\leq|\lambda|\leq r+1}m_R(\lambda)=\mathcal{O}(r^{\mu+0}),\;\text{as}\;r\rightarrow+\infty, \] where \(2\mu+1\) is the box dimension of the trapped set, and \(m_R(\lambda)\) is the multiplicity of a (nonzero) resonance, i.e., \[ m_R(\lambda)=\text{rank}\oint_{\gamma}R(\zeta)d\zeta,\;\gamma:t\mapsto \lambda+\epsilon e^{2i\pi t},\;0<\epsilon\ll 1. \] This kind of upper bound was predicted by the second author in [Duke Math. J. 60, No. 1, 1--57 (1990; Zbl 0702.35188)] where similar bounds were obtained for families of semiclassical Schrödinger operators satisfying some analyticity assumptions.