An explicit Berry-Esséen bound for uniformly expanding maps on the interval (Q1758954)

The author proves a Berry-Esséen bound for uniformly expanding Markov transformations on an interval and for Lipschitz observables. The Berry-Esséen inequality has been known from many sources (all of them are listed in the paper) with a non-explicit constant \(C\). To formulate the main result, some notations are needed. Let \(T\) be a uniformly expending map on \([0,1]\) which satisfies the Markov condition. Let \(\operatorname{P}\) be the probability distribution given by the Gibbs measure associated to \(T\). The observable \(f\) satisfies \(\operatorname{E} f= 0\) and is supposed to be Lipschitz. Denote \[ \gamma= \inf|T'|> 1,\quad S_nf= \sum^{n-1}_{k=0} f\circ T^k,\quad \sigma= \lim_{n\to\infty} (1/n) \operatorname{E}(S_n f)^2. \] The constant \(G\) below is the supremum over each inverse branch \(\sigma_j\) of \(T\) of the Lipschitz constant of \(\log|T'\circ\sigma_j|\). Let \(d\) be a metric compatible with the topology on \([0,1]\) and \[ |f|_l= \sup_{x\neq y} {|f(x)- f(y)|\over d(x,y)}. \] The main result reads as follows. Theorem. Assume that \(\sigma>0\). Then we have for all \(x\in \mathbb{R}\), all \(n\geq 1\), \[ \Biggl| \operatorname{P}\Biggl({S_nf\over \sigma\sqrt{n}}< x\Biggr)-{1\over \sqrt{2\pi}} \int^x_{-\infty} e^{-t^2/2} dt\Biggr|\leq C{\cosh^6(D_{\mathbb{R}}/4)\| f\|_\infty(\| f\|_\infty+\| f\|_l)^2\over \sigma^3\sqrt{n}}, \] where \(C\) is an absolute constant, \(C\leq 11460\), and \[ D_{\mathbb{R}}= {2(\gamma^2 G+1)\over \gamma(\gamma-1)}+ 2\log{2\gamma^2 G+\gamma+1\over \gamma-1}. \] The novelty is that the author indicates an upper bound for the constant \(C\).