NeuralNEB—neural networks can find reaction paths fast

Abstract

Quantum mechanical methods like density functional theory (DFT) are used with great success alongside efficient search algorithms for studying kinetics of reactive systems. However, DFT is prohibitively expensive for large scale exploration. Machine learning (ML) models have turned out to be excellent emulators of small molecule DFT calculations and could possibly replace DFT in such tasks. For kinetics, success relies primarily on the models’ capability to accurately predict the potential energy surface around transition-states and minimal energy paths. Previously this has not been possible due to scarcity of relevant data in the literature. In this paper we train equivariant graph neural network-based models on data from 10 000 elementary reactions from the recently published Transition1x dataset. We apply the models as potentials for the nudged elastic band algorithm and achieve a mean average error of 0.23 eV and root mean squared error of 0.52 eV on barrier energies on unseen reactions. We compare the results against equivalent models trained on QM9x and ANI1x. We also compare with and outperform Density Functional based Tight Binding on both accuracy and required computational resources. The implication is that ML models are now at a level where they can be applied to studying chemical reaction kinetics given a sufficient amount of data relevant to this task.