Path-averaged Kinetic Equation for Stochastic Systems

Abstract

For a stochastic system, its evolution from one state to another can have a large number of possible paths. Non-uniformity in the field of system variables leads the local dynamics in state transition varies considerably from path to path and thus the distribution of the paths affects statistical characteristics of the system. Such a characteristic can be referred to as path-dependence of a system, and long-time correlation is an intrinsic feature of path-dependence systems. We employed a local path density operator to describe the distribution of state transition paths, and based on which we derived a new kinetic equation for path-dependent systems. The kinetic equation is similar in form to the Kramers-Moyal expansion, but with its expansion coefficients determined by the cumulants with respect to state transition paths, instead of transition moments. This characteristic makes it capable of accounting for the non-local feature of systems, which is essential in studies of large scale systems where the path-dependence is prominent. Short-time correlation approximation is also discussed. It shows that the cumulants of state transition paths are equivalent to jump moments when correlation time scales are infinitesimal, as makes the kinetic equation derived in this paper has the same physical consideration of the Fokker-Planck equation for Markov processes.