Multiensemble sampling: An alternative efficient Monte Carlo technique

We present a general recipe for the successful non-Boltzmann sampling simulation from which data including the free energy of a system over a moderately wide range of thermodynamic states can be obtained. A universal form of the non-Boltzmann weighting function W is derived by a functional minimization so that a most favorable condition for the simultaneous investigation of several states with uniform accuracy is satisfied. The realization of the method is simple and systemic. The accuracy of the method is demonstrated by comparison with reliable earlier results for a Lennard-Jones fluid.