762b A New Multiscale Coarse-Grain Methodology: The Self-Consistent Force-Matching Method

Jhih-Wei Chu, Chemical Engineering, University of California, Berkeley, 101A Gilman, UC Berkeley, Berkeley, CA 94720 and Hyung Min Cho, Chemical Engineering, UC Berkeley, 6 Gilman, UC Berkeley, Berkeley, CA 94720.

Force matching may be considered as a multiscale method that can be used to develop coarse-grained models directly from all-atom (or fine-grained) simulations. Recently, it has been shown that the force-matching equation is closely related to the Yvon-Born-Green (YBG) equation that describes the hierarchal connection between the structural correlation functions of simple liquids. In this work, we show that it is possible to extend the force-matching method to solve the YBG equation self-consistently. The resulting methodology may thus be employed to compute the underlying pair interactions that are required to give rise to a specified pair distribution. The implication of such capability in molecular design will be addressed. The SCFM is applied to develop the coarse-grained force fields of liquid water and liquid hexane from all-atom molecular dynamics simulations. By reproducing pair distribution functions, we found that the agreement of three-body correlation functions between all-atom and coarse-grained simulations is also improved. The SCFM method is contrasted with the commonly inverse Monte Carlo strategy in developing coarse-grain potential. The issues of transferability of coarse-grain force fields, the thermodynamic and dynamic consistency of coarse-grained simulations, and the integration of coarse-grained and all-atom models will also be discussed.