Molecular modeling with machine-learned universal potential functions

TL;DR

This paper demonstrates that neural networks can be trained to create universal, smooth, and predictive energy potential functions for molecular modeling, enhancing drug discovery processes.

Contribution

It introduces a neural network-based universal potential function trained automatically on large-scale crystal structures, showing improved versatility over traditional methods.

Findings

Neural networks can effectively model molecular energy potentials.

The proposed model outperforms traditional force fields in tests.

The approach is scalable and adaptable to large datasets.

Abstract

Molecular modeling is an important topic in drug discovery. Decades of research have led to the development of high quality scalable molecular force fields. In this paper, we show that neural networks can be used to train a universal approximator for energy potential functions. By incorporating a fully automated training process we have been able to train smooth, differentiable, and predictive potential functions on large-scale crystal structures. A variety of tests have also been performed to show the superiority and versatility of the machine-learned model.