Estimating time-dependent entropy production from non-equilibrium trajectories

Abstract The rate of entropy production provides a useful quantitative measure non-equilibrium system and estimating it directly from time-series data experiments is highly desirable. Several approaches have been considered for stationary dynamics, some which are based on variational characterization the rate. However, issue obtaining in case non-stationary dynamics remains largely unexplored. Here, we solve this open problem by demonstrating that can be generalized to give exact value even dynamics. On basis result, develop an efficient algorithm estimates continuously time using machine learning techniques validate our numerical analytically tractable Langevin models experimentally relevant parameter regimes. Our method only requires interest without any prior knowledge system’s parameters.