Rates in almost sure invariance principle for slowly mixing dynamical systems

TL;DR

This paper establishes near-optimal rates for the almost sure invariance principle in slowly mixing dynamical systems, improving significantly over previous results by employing a Young-tower-like Markov chain representation.

Contribution

It introduces a novel approach using Young-tower-like Markov chains to achieve optimal rates in the almost sure invariance principle for slowly mixing systems.

Findings

Achieved rates of $o(n^eta)$ with slowly varying functions.

Improved previous rate bounds from $O(n^{1/4})$ to near optimal.

Applied advanced approximation methods for dependent processes.

Abstract

We prove the one-dimensional almost sure invariance principle with essentially optimal rates for slowly (polynomially) mixing deterministic dynamical systems, such as Pomeau-Manneville intermittent maps, with H\"older continuous observables. Our rates have form $o(n^\gamma L(n))$, where $L(n)$ is a slowly varying function and $\gamma$ is determined by the speed of mixing. We strongly improve previous results where the best available rates did not exceed $O(n^{1/4})$. To break the $O(n^{1/4})$ barrier, we represent the dynamics as a Young-tower-like Markov chain and adapt the methods of Berkes-Liu-Wu and Cuny-Dedecker-Merlev\`ede on the Koml\'os-Major-Tusn\'ady approximation for dependent processes.