Representation of intermediate time-scale motions in stochastic modeling: Analysis on stochastic description of classical Hamiltonian dynamics in relation with measurement imperfection

Abstract

It is a well established result that, in classical dynamical systems with sufficient time-scale separation, the fast chaotic degrees of freedom are well modeled by (Gaussian) white noise. In this paper, we present the stochastic dynamical description for intermediate time-scale motions with insufficient time-scale separation from the slow dynamical system. First, we analyze how the fast deterministic dynamics can be viewed as stochastic dynamics under experimental observation by intrinsic errors of measurement. Then, we present how the stochastic dynamical description should be modified if intermediate time-scale motions exist: the time correlation of the noise ξis modified to <ξ(t)ξ(t')> = C(x,p)δ(t-t'), where C(x,p) is a smooth function of the slow coordinate (x,p), and generally the cumulants of ξexcept its average vary as a smooth function of the slow coordinates (x,p). The analysis given in this work actually shows that, regardless of the sufficiency of time-scale separation, any complex (chaotic and ergodic) dynamical system can be well described using Markov process, if we perfectly construct the deterministic part of (extended) stochastic dynamics.