Correlation functions in linear chaotic maps

Abstract

The simplest examples of chaotic maps are linear, area-preserving maps on the circle, torus, or product of tori; respectively known as the Bernoulli map, the cat map, and the recently introduced "spatiotemporal" cat map. We study correlation functions in these maps. For the Bernoulli map, we compute the correlation functions in a variety of ways: by direct computation of the integral, through Fourier series, through symbolic dynamics, and through periodic orbits. In relation to the more standard treatment in terms of eigenfunctions of the Perron-Frobenius operator, some of these methods are simpler and also extend to multipoint correlation functions. For the cat map, we compute correlation functions through a Fourier expansion, review and expand on a prior treatment of two-point functions by Crawford and Cary, and discuss the limitations of shadowing. Finally, for the spatiotemporal cat map -- intended to be a model of many-body chaos -- we show that connected correlation functions of local operators vanish.