Gibbs–Helmholtz graph neural network: capturing the temperature dependency of activity coefficients at infinite dilution

Abstract

A hybrid model that combines the Gibbs–Helmholtz equation with Graph Neural Networks for predicting limiting activity coefficients.