An ensemble reweighting method for combining the information of experiments and simulations

Abstract The accuracy of molecular dynamics (MD) simulation can be assured by the reproductivity of experimental results. We propose a new post-processing method for optimizing weighting factors of MD simulation snapshots to improve the agreement with the experimental data without an undesirable change of the distribution generated by the simulation. This method minimizes the Kullback–Leibler divergence between the optimized and original probability distribution, provided that the error between experimentally observed and calculated values is smaller than some given value. In this method, the reweighting method was applied to ethane in a vacuum and lysozyme in a solvent.