Equivariant message passing for the prediction of tensorial properties and molecular spectra

TL;DR

This paper introduces PaiNN, a rotationally equivariant message passing neural network that improves data efficiency and accuracy in predicting molecular properties and spectra, significantly accelerating simulations.

Contribution

The paper extends message passing neural networks to include rotationally equivariant representations, enabling tensorial property prediction and faster molecular spectra simulations.

Findings

PaiNN outperforms previous models on molecule benchmarks.

PaiNN reduces model size and inference time.

Achieves 4-5 orders of magnitude speedup in molecular spectra simulation.

Abstract

Message passing neural networks have become a method of choice for learning on graphs, in particular the prediction of chemical properties and the acceleration of molecular dynamics studies. While they readily scale to large training data sets, previous approaches have proven to be less data efficient than kernel methods. We identify limitations of invariant representations as a major reason and extend the message passing formulation to rotationally equivariant representations. On this basis, we propose the polarizable atom interaction neural network (PaiNN) and improve on common molecule benchmarks over previous networks, while reducing model size and inference time. We leverage the equivariant atomwise representations obtained by PaiNN for the prediction of tensorial properties. Finally, we apply this to the simulation of molecular spectra, achieving speedups of 4-5 orders of magnitude compared to the electronic structure reference.