MD + EFP --> ??? : A Tale of Molecular Dynamics and the Effective Fragment Potential Method

Heather Netzloff, Iowa State University

Dynamic simulations of a chemical system allow investigation of time-dependent phenomena and provide a means to study characteristics of a chemical system which become computationally intensive as the size of a system grows. The Effective Fragment Potential (EFP) method for solvation in the GAMESS program is currently being coupled with a Molecular Dynamics (MD) algorithm with these goals in mind. The EFP method reduces computational expenses by treating the solute with an ab initio (AI) method of choice and adds the interaction effect of the solvent/“fragment” molecules as one-electron terms. Terms include Coulomb/electrostatic and polarization (dipole - induced dipole) interactions, and exchange repulsion/charge transfer. The first MD version simulates EFP - EFP interactions in both NVE (constant volume and energy) and NVT (constant volume and temperature) ensembles on a variety of water clusters (up to 256 waters). EFP - EFP MD will allow the study of small water clusters, larger clusters, and finally bulk behavior. AIMD has also been implemented. While this type of MD is much more computationally demanding, it allows an accurate study of small systems and gives a benchmark for the EFP method. Extensions to solute/AI - solvent/EFP MD are currently underway. Since most biological and experimental processes occur in solution, this will be a very important way to study a chemical system. As the size of the system and/or number of solvent molecules increases, static optimization methods in the search for energy minima become unreasonable. The use of MD and/or MD with Simulated Annealing (SA) will provide a viable means to overcome this problem.

Abstract Author(s): Heather M. Netzloff and Mark S. Gordon