A gentle stochastic thermostat for molecular dynamics (Q2391046)

A dynamical technique for sampling the canonical measure in molecular dynamics is presented. The method generalizes the scheme by Samoletov, Chaplain and Dettmann, which aims to combine the advantages of the Langevin-thermostat with those of the Nosé-Hoover method. It can be viewed as a Nosé-Hoover method in which the thermostat variable is a Brownian particle. In contrast to Langevin dynamics, where noise is added directly to each physical degree of freedom, the new scheme relies on an indirect coupling to a single Brownian particle. For a model with harmonic potentials, we show under a mild non-resonance assumption that we can recover the canonical distribution. In spite of its stochastic nature, experiments suggest that it introduces a relatively weak perturbative effect on the physical dynamics, as measured by perturbation of temporal autocorrelation functions. The examples of a harmonic oscillator and of three particle connected by springs, interacting with each other through a Lennard-Jones potential are investigated. The kinetic energy is well controlled even in the early stages of a simulation. The primary distinction between the method presented here and others in the literature is an analysis of ergodicity, making use of the concept of hypoellipticity with respect to the operator defining the right hand side of the Fokker-Planck equations. The new dynamics has an invariant probability measure, which is proportional to the Boltzmann-Gibbs distribution. It is proved analytically that under a non-resonance assumption, an open connected set U with full measure can be constructed, such that the probability measure is ergodic on U.