MONTE-CARLO ENTROPIC SAMPLING FOR THE STUDY OF METASTABLE STATES AND RELAXATION PATHS

We present a continuous extension of the recent Monte Carlo entropic m ethod for sampling a density of states restricted in dimensionless mac roscopic parameters. The method performs a random walk through a two-d imensional macrostate space and provides complete information in the f orm of continuous functions of the system's coupling constants. For th e example of an Ising system, we project relaxation paths from Monte C arlo Metropolis dynamic over the two-dimensional state space and compa re them with a ''most probable path'' associated with the equilibrium distribution, derived from the density of states. We observe a close a greement between them in the stochastic regime, i.e., before the syste m escapes from the metastable state. We establish a Markovian macrosco pic dynamic over the two macroscopic parameters and we discuss it with respect to the Metropolis microscopic dynamic.