(432a) A Conservative Algorithm For An On-The-Fly Change Of Resolution In Mixed Atomistic / Coarse-Grained Multiscale Simulations

Heyden, A., University of South Carolina
Truhlar, D. G., University of Minnesota

Coarse-graining techniques are commonly used to increase the computational efficiency of molecular simulations of complex soft matter systems. A difficulty in applying these methods for describing complex systems is the need to obtain accurate coarse-grained potential energy parameters for strongly interacting systems where atoms become polarized or react. As a result, coarse-graining techniques often cannot be applied to these systems, and less efficient but generally more accurate all-atom force fields have to be employed. In order to combine the efficiency of coarse-grained force fields with the accuracy of atomistic methods, mixed-resolution force fields have been developed that use a coarse-grained description for the part of the system distant from an active site and an atomistic description for the active site and its direct environment. Since atoms or groups of atoms may diffuse during a molecular simulation, the algorithm needs to permit an on-the-fly reclassification (from atomistic to coarse-grained or vice versa) of atoms or groups of atoms as they transition between the high- and low-resolution regimes. Previous algorithms for adaptive-resolution molecular dynamics simulations are equivalent to using non-conservative force fields, which result in a lack of conservation of both total system energy and total angular momentum.(1-3)

Recently, we introduced an adaptive partitioning method for treating the potential energy function for systems with an active zone modeled at a high-level of theory and a surrounding environmental zone modeled at a low-level of theory.(4) This algorithm permits an on-the-fly change in level of theory of the potential energy function of an atomistic simulation as atoms or groups of atoms transition between the active and environmental zones. By extending this adaptive partitioning method for potential energy functions to adaptive partition Lagrangian functions, we were able to derive a Hamiltonian for mixed-resolution systems that allows for a change in resolution of selected groups during a simulation. An NVE simulation protocol is developed in which total energy, angular momentum, and linear momentum are conserved. As an application, we present simulation results for the structure and diffusion of a water molecule in hexane.

(1) Praprotnik, M.; Delle Site, L.; Kremer, K. J. Chem. Phys. 2005, 123, 224106. (2) Praprotnik, M.; Delle Site, L.; Kremer, K. Physical Review E 2006, 73, 066701. (3) Praprotnik, M.; Delle Site, L.; Kremer, K. J. Chem. Phys. 2007, 126, 134902. (4) Heyden, A.; Lin, H.; Truhlar, D. G. J. Phys. Chem. B 2007, 111, 2231.