Electrostatics play a crucial role in biophysical and biochemical processes, such as protein-ligand binding, ionic interactions, and PH-dependent processes etc. The quantitative modellings of the electrostatics with the classical force fields pose great challenges due to the lack of capturing the polarization effects. Hence great efforts have been casted into developing the polarizable force fields. However, there are multiple issues with the use of classical point models of polarization in existing approaches. Usually these simple point polarizable models deviate far from reality, leading to many ad hoc attempts to patch its artifacts and numerical issues. This results in lower degree of self-consistency and transferability, both are important for robust modeling of molecular systems based on classical physics. Here in this work, we report our effort exploring a more realistic polarizable Gaussian Multipole (pGM) model for more satisfactory representation of the atomic charge distributions in molecular systems. In the pGM model, all charges and multipoles are represented by Gaussian densities instead of classical points. The more realistic representation of molecular charge distributions allows for higher accuracy, self-consistency, and transferability. To implement the pGM model, we first proposed a new local natural frame, which overcomes the artifact induced by the rigid-body assumption of molecules in classical point polarizable force fields where the local Cartesian frame is often used. In addition, we derived all the force and torque interaction terms which carefully considered the continuous distribution of all multipoles. To accelerate the calculation, we also explored PME framework for pGM model. Finally, we analyzed the potential implementation of pGM model onto GPUs to seek more efficiency.

Implementing polarizable Gaussian multipole model for molecular dynamics simulations

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.