Highly accurate and fine-scale estimation of equilibrium measures

Abstract

Equilibrium measures are special invariant measures of chaotic dynamical systems and iterated function systems, commonly studied as salient examples of fractal measures. While useful analytic expressions are rare, computational exploration of these measures can yield useful insight, in particular in studying their Fourier decay. In this note we present simple, efficient computational methods to obtain weak estimates of equilibrium and related measures (i.e. as integrals against smooth functions) at high spatial resolution. These methods proceed via Chebyshev-Lagrange approximation of the transfer operator. One method, which estimates measures directly from spectral data, gives exponentially accurate estimates at spatial scales larger than the approximation's resolution. Another, method, which generates random point samples, has a Central Limit Theorem-style accuracy down to an exponentially small spatial resolution. This means that these measures and their Fourier decay can be studied very accurately, and at very high Fourier frequencies.