Thin annuli property and exponential distribution of return times for weakly Markov systems

Abstract

We deal with the problem of asymptotic distribution of first return times to shrinking balls under iteration generated by a large general class of dynamical systems called weakly Markov. Our ultimate main result is that these distributions converge to the exponential law when the balls shrink to points. We apply this result to many classes of smooth dynamical systems that include conformal iterated function systems, rational functions on the Riemann sphere $\widehat{\mathbb C}$, and transcendental meromorphic functions on the complex plane $\mathbb{C}$. We also apply them to expanding repellers and holomorphic endomorphisms of complex projective spaces. One of the key ingredients in our approach is to solve the well known, in this field of mathematics, problem of appropriately estimating the measures of, suitably defined, large class of geometric annuli. We successfully do it. This problem is, in the existing literature, differently referred to by different authors; we call it the Thick Thin Annuli Property. Having this property established, we prove that for non--conformal systems the aforementioned distributions converge to the exponential one along sets of radii whose relative Lebesgue measure converges fast to one. But this is not all. In the context of conformal iterated function systems, we establish the Full Thin Annuli Property, which gives the same estimates for all radii. ln this way, we solve a long standing problem. As a result, we prove that the convergence to the exponential law holds along all radii for essentially all conformal iterated function systems and, with the help of the techniques of first return maps, for all aforementioned conformal dynamical systems.