Spectral Analysis of Current Fluctuations in Periodically Driven Stochastic Systems

Abstract

Current fluctuations play an important role in non-equilibrium statistical mechanics, and are a key object of interest in both theoretical studies and in practical applications. So far, most of the studies were devoted to the fluctuations in the time-averaged current -- the zero frequency Fourier component of the time dependent current. However, in many practical applications the fluctuations at other frequencies are of equal importance. Here we study the full frequency dependence of current statistics in periodically driven stochastic systems. First, we show a general method to calculate the current statistics, valid even when the current's frequency is incommensurate with the driving frequency, breaking the time periodicity of the system. Somewhat surprisingly, we find that the cumulant generating function (CGF), that encodes all the statistics of the current, is composed of a continuous background at any frequency accompanied by either positive or negative peaks at current's frequencies commensurate with the driving frequency. We show that cumulants of increasing orders display peaks at an increasing number of locations but with decreasing amplitudes that depend on the commensurate ratio of frequencies. All these peaks are then transcribed in the behaviour of the CGF. As the measurement time increases, these peaks become sharper but keep the same amplitude and eventually lead to discontinuities of the CGF at all the frequencies that are commensurate with the driving frequency in the limit of infinitely long measurement. We demonstrate our formalism and its consequences on three types of models: an underdamped Brownian particle in a periodically driven harmonic potential; a periodically driven run-and-tumble particle; and a two-state system.