Machine Learning Directed Optimization of Classical Molecular Modeling Force Fields

TL;DR

This paper introduces a machine learning-driven multiobjective optimization workflow for developing more accurate classical molecular force fields, significantly reducing the number of costly simulations needed.

Contribution

It presents a novel workflow that efficiently searches for optimal force field parameters using machine learning, applicable to diverse molecular systems.

Findings

Successfully identified low-error force field parameters for HFC vapor-liquid equilibrium.

Achieved accurate force fields for ammonium perchlorate crystal phase.

Demonstrated the workflow's generality across different molecular systems.

Abstract

Accurate force fields are necessary for predictive molecular simulations. However, developing force fields that accurately reproduce experimental properties is challenging. Here, we present a machine learning directed, multiobjective optimization workflow for force field parameterization that evaluates millions of prospective force field parameter sets while requiring only a small fraction of them to be tested with molecular simulations. We demonstrate the generality of the approach and identify multiple low-error parameter sets for two distinct test cases: simulations of hydrofluorocarbon (HFC) vapor-liquid equilibrium (VLE) and an ammonium perchlorate (AP) crystal phase. We discuss the challenges and implications of our force field optimization workflow.