AI and the World's Hard Problems · Chemistry & materials · arXiv · 2026-10-26
Machine Learning Accelerated Simulation of Charge Regulation Effects
Simulating materials that gain charge from surface groups requires hybrid Monte Carlo and molecular dynamics, and the Monte Carlo step is often as costly as the dynamics. The authors introduce CR-ML, a graph attention network that predicts each surface site's ionization probability from the instantaneous configuration. It reproduces the equilibrium results of the original method for a polyanion and for two nanoparticles linked by a polyelectrolyte, while updating charges about ten times faster.