Building a Coarse-Grained Model Based on the Mori-Zwanzig Formalism

摘要

We present a novel method to build a coarse-grained (CG) model based on the Mori-Zwanzig (MZ) formalism that reproduces kinetics. Our approach leads to the computation of a generalized Langevin equation (GLE), which includes the memory kernel and the fluctuation that are consistent with brute force molecular dynamics (MD) simulations. Our CG model based on the MZ formalism successfully reproduces kinetics, i.e. the distribution of first passage times (FPT) and velocity autocorrelation functions (VACF), for alanine dipeptide. In addition, we show that the memory part of the CG model of GLE is essential to reproduce kinetics. In other words, the Markovian model fails to reproduce brute force MD results, whereas the GLE model succeeds.