Molecular Dynamics (MD) is a powerful simulation technique for capturing the dynamics and equilibrium properties of molecular and macromolecular systems at atomic resolution. However, MD faces significant practical challenges due to the limited time and spatial scales it can reach. To address these challenges, various coarse-grained (CG) and multiscale methods have been developed. In particular, Multiscale Factorization (MF) is a promising multiscale framework that provides a self-consistent and efficient way of coevolving the atomistic and CG states without requiring calibration of the CG model. MF achieves this coevolution by backmapping the CG state to an ensemble of all-atom microstates consistent with the latter. In this study, we introduce a force field-driven backmapping method that yields improved accuracy and numerical stability over existing methods, enabling the use of larger CG timesteps in the course of a multiscale simulation.
Guo et al. (Mon,) studied this question.