Towards predictive design of electrolyte solutions by accelerating ab initio simulation with neural networks

Ab initio molecular dynamics can be massively accelerated using equivariant neural networks applicable to predict the properties of electrolyte solutions for predictive design in materials applications.