Dynamical thermostatting and statistical ensembles

摘要

Dynamical thermostatting constitutes a procedure for computing thermodynamical mean values of classical dynamical systems that is of interest both from the practical and from the conceptual points of view. Here we extend and unify previous partial results, showing that the dynamical thermostatting approach can be implemented in order to simulate a wide family of statistical ensembles of general dynamical systems with a vanishing divergence and admitting an integral of motion. As a particular illustration, the thermostatting procedure is applied to power law-like maximum entropy ensembles.