Rates of convergence in the CLT for two-dimensional dispersive billiards. (Q5961020)

The author considers billiards in the two-dimensional torus with convex obstacles. In the first section, a brief overview of the model is given in which all relevant notions are introduced in much detail. Although various central limit theorems have been established in the past for regular functions for the billiard transformation [\textit{L. A. Bunimovich} and \textit{Y. G. Sinai}, Commun. Math. Phys. 78, 479--497 (1981; Zbl 0459.60099); \textit{L. A. Bunimovich, N. I. Chernov} and \textit{Y. G. Sinai}, Russ. Math. Surv. 46, No. 4, 47--106 (1991; Zbl 0780.58029); \textit{L.-S. Young}, Ann. Math. (2) 147, 585--650 (1998; Zbl 0945.37009)] , the problem of the rate of convergence has not been answered in full detail yet. The current paper provides a profound discussion on this topic. By carefully rewriting of the proof of the exponential rate of decorrelation given in [L.-S. Young (loc. cit.)], the author is able to extend the previous results to a more general class of functions, the so-called space of \(\eta \)-Hölder functions in \(m\) future coordinates. By adapting the proof of \textit{C. Jan} [Vitesse de convergence dans le TCL pour des processus associés à des systèmes dynamiques et aux produits de matrices aléatoires (Ph. D. thesis, University of Rennes 1, France) (2001); C. R. Acad. Sci., Paris, Sér. I, Math. 331, 395--398 (2000; Zbl 0963.60018); Rates of convergence for some processes under mixing conditions and application to random matrix products, Preprint (2001)], the rate of convergence for the billiard transformation is obtained. A fully detailed technical proof is given in the appendices at the end of the paper which keeps the first part of the paper also readable for nontechnically-inclined readers. In the next section, a rate of convergence for the billiard flow is obtained, which however is not claimed to be optimal. Again, all technical details of the proof can be found in the appendices. Some of the proofs may seem rather lengthy but on the other hand this allows the interested reader to verify all crucial steps of the proof in much detail.