We refine the OrbNet model to accurately predict energy, forces, and other response properties for molecules using a graph neural-network architecture based on features from low-cost approximated quantum operators in the symmetry-adapted atomic orbital basis. The model is end-to-end differentiable due to the derivation of analytic gradients for all electronic structure terms, and is shown to be transferable across chemical space due to the use of domain-specific features. The learning efficiency is improved by incorporating physically motivated constraints on the electronic structure through multi-task learning. The model outperforms existing methods on energy prediction tasks for the QM9 dataset and for molecular geometry optimizations on conformer datasets, at a computational cost that is thousand-fold or more reduced compared to conventional quantum-chemistry calculations (such as density functional theory) that offer similar accuracy.
@article{arxiv.2011.02680,
title = {Multi-task learning for electronic structure to predict and explore molecular potential energy surfaces},
author = {Zhuoran Qiao and Feizhi Ding and Matthew Welborn and Peter J. Bygrave and Daniel G. A. Smith and Animashree Anandkumar and Frederick R. Manby and Thomas F. Miller},
journal= {arXiv preprint arXiv:2011.02680},
year = {2020}
}
Comments
Accepted for presentation at the Machine Learning for Molecules workshop at NeurIPS 2020