Kleinian Schottky groups, Patterson-Sullivan measures, and Fourier decay (Q2037844)

Let \(\mu\) be a Borel probability measure in \(\mathbb{R}^d\) and let \(\widehat{\mu}(\xi)\) be its Fourier transform defined for any \(\xi \in \mathbb{R}^d\). Let \(K\) be a nonempty compact subset of \(\mathbb{R}^d\). Its Hausdorff dimension can be expressed as \[\dim_H(K)= \sup\left\{s\in[0,d]:\int_{\mathbb{R}^d} |\widehat{\mu}(\xi)|^2|\xi|^{s-d} d\xi < \infty \text{ for some } \mu \in \mathscr{P}(K)\right\},\] where \(\mathscr{P}(K)\) is the set of Borel probability measures on \(K\). The Fourier dimension of \(K\) is defined by \[ \dim_F (K)=\sup\{s \in [0,d]: \sup_{\xi} |\widehat{\mu}(\xi)|^2|\xi|^{s} < \infty \text{ for some } \mu \in \mathscr{P}(K)\}. \] It is an interesting problem to build families of deterministic sets with positive Fourier dimension, that is, compact sets \(K\) with fractal Hausdorff dimension such that one can find a Borel probability measure \(\mu\) on \(K\) whose Fourier transform has polynomial decay: \(\widehat{\mu}(\xi)=O(|\xi|^{-\epsilon})\). \textit{J. Bourgain} and \textit{S. Dyatlov} obtained in [Geom. Funct. Anal. 27, No. 4, 744--771 (2017; Zbl 1421.11071)] a generalized Fourier decay bound for the Patterson-Sullivan measure for convex co-compact hyperbolic surfaces. In the paper under review, the authors study the case of Schottky groups and point out that it is enough to completely solve the problem of the positiveness of the Fourier dimension of limit sets for all Zariski-dense subgroups of \(\text{PSL}_2(\mathbb{C})\). The main result is Theorem 1.1. Assume that \(\Gamma\) is a Zariski-dense Schottky group in \(\text{PSL}_2(\mathbb{C})\), and let \(\mu\) be any Patterson-Sullivan measure on the limit set \(\Lambda_{\Gamma}\). Fix any neighborhood \(\mathscr{U}\) of \(\Lambda_{\Gamma}\). Let \(g\) be in \(C^1(\mathscr{U},\mathbb{C})\) and let \(\phi\) be in \(C^2(\mathscr{U},\mathbb{R})\) with \( M:= \inf_{z\in \mathscr{U}} |\nabla_z \phi| > 0\) on \(\Lambda_{\Gamma}\). Assume that \(\|g\|_{C^1}+ \|\phi\|_{C^2} \leq M'\). Then there exist constants \(C>0\) and \(\epsilon >0\), with \(C\) depending on \(M, M' \text{ and } \Gamma\), and \(\epsilon\) depending only on \(\mu\), such that for all \(t \in \mathbb{R}\) with \(|t| \geq 1\), \[ \left|\int_{\Lambda_{\Gamma}} e^{it\phi(z)} g(z) d\mu(z) \right| \leq C |t|^{-\epsilon}. \] The authors also give a variation of this (Theorem 1.1'). The statement is very similar to Theorem 1.1 but now \(\inf_{z\in \mathscr{U}} |\nabla_z \phi| >|t|^{-\alpha}\) (where \(\alpha >0\), depends only on \(\mu\)), so that the constant \(C\) depends only on \(M'\). The proof of Theorem 1.1 is given in Section 3. A compact set \(K \subset \mathbb{C}\) is said to have \(C^2\)-stable positive Fourier dimension if and only if for all \(C^2\)-diffeomorphism \(\phi: \mathscr{U} \rightarrow \phi(\mathscr{U}) \subset \mathbb{C}\), defined on a neighborhood \(\mathscr{U}\) of \(K\), the set \(\phi(K)\) has positive Fourier dimension. As a consequence of Theorem 1.1, Corollary 1.4 is obtained: Let \(\Gamma\) be a non-elementary subgroup of \(\text{PSL}_2(\mathbb{C})\) such that \(\Lambda_{\Gamma}\) does not equal \(\mathbb{P}^1(\mathbb{C})\). Then \(\Lambda_{\Gamma}\) has \(C^2\)-stable positive Fourier dimension if and only if \(\Gamma\) is Zariski-dense in \(\text{PSL}_2(\mathbb{C})\). In an appendix by the first author, a random walk is constructed on a convex co-compact subgroup of \(\text{SO}_0(1,n)\), which has exponential moment and such that the associated Patterson-Sullivan measure of the convex co-compact group can be realized as a stationary measure.