Alexander P. Lyubartsev and Aatto Laaksonen
Division of Physical Chemistry, Arrhenius Laboratory, Stockholm University, S-109 61, Stockholm, Sweden
A new approach is presented to solve the reverse problem of statistical mechanics: reconstruction of interaction potentials from radial distribution functions. The method consists in iterative adjustment of the interaction potential to known radial distribution functions using Monte Carlo simulation technique and statistical-mechanical relations to connect deviations of canonical averages with Hamiltonian parameters. The method is applied to calculate the effective interaction potentials between the ions in aqueous NaCl solutions at two different concentrations. The reference ion-ion radial distribution functions, calculated in separate Molecular Dynamics simulations with water molecules, are reproduced in Monte Carlo simulations, using the effective interaction potentials for the hydrated ions. Application of the present method should provide an effective and economic way to simulate equilibrium properties for very large molecular systems (e.g. polyelectrolytes) in the presence of hydrated ions as well as to offer an approach to reduce a complexity in studies of various associated and aggregated systems in solution.
PACS number(s): 05.20.Gg,02.70.-c,61.20.Ja,61.20.Qg
Paper (PostScript file)
Later this algorithm have been referred as "Inverse Monte Carlo" or "Inverse Newton".