Learning the exchange-correlation functional from nature with fully differentiable density functional theory
This work offers an incremental improvement in quantum chemistry modeling for materials scientists by enhancing the accuracy of molecular property predictions with limited experimental data.
This paper addresses the challenge of improving molecular property prediction in ab initio simulations by training a neural network to replace the exchange-correlation functional within a differentiable Kohn-Sham DFT framework. Using only eight experimental data points, their method improved the prediction accuracy of atomization energies across 104 molecules, including those with novel bonds and atoms not in the training set.
Improving the predictive capability of molecular properties in ab initio simulations is essential for advanced material discovery. Despite recent progress making use of machine learning, utilizing deep neural networks to improve quantum chemistry modelling remains severely limited by the scarcity and heterogeneity of appropriate experimental data. Here we show how training a neural network to replace the exchange-correlation functional within a fully-differentiable three-dimensional Kohn-Sham density functional theory (DFT) framework can greatly improve simulation accuracy. Using only eight experimental data points on diatomic molecules, our trained exchange-correlation networks enable improved prediction accuracy of atomization energies across a collection of 104 molecules containing new bonds and atoms that are not present in the training dataset.