Abstract: I will describe several research projects that applied automatic force field parameterization methods to build approximate force fields for water and the lessons learned therein. The force fields discussed span a wide range of functional forms, including: (1) fixed-charge three and four point models (TIP3P-FB and TIP4P-FB), (2) a six-point fluctuating charge model (FT-6FQ), (3) a direct polarization multipole model (iAMOEBA), (4) a single-site fully polarizable multipole model (uAMOEBA), and (5) an all-atom fully polarizable multipole model (AMOEBA14). The model optimization was facilitated by ForceBalance, an open-source software package that organizes and automates the calculations of observables to achieve the goal of reproducibility for this class of calculation. Over the course of these projects, we learned several interesting lessons, such as: (a) Reparameterized direct polarization models are able to fully capture the effect of mutual induction. (b) The short-range repulsive part of the intermolecular interaction plays a principal role in determining the shape of the first peak of the radial distribution function. (c) When fitting parameters to data points from a reference potential energy surface, it is essential to include points from the thermodynamic ensembles of the model we are fitting and the reference potential surface. I will summarize the lessons of the past by presenting promising future directions for model development.

Lessons learned in building polarizable and fixed-charge water models

252nd National Meeting (2016)

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