Researchers at DeepMind have open-sourced DM21, a neural network model for mapping electron density to chemical interaction energy, a critical component of quantum mechanical modelling. DM21 outperforms standard models on various benchmarks, and it’s accessible as a PySCF simulation framework addition.
DM21 uses a neural network to approximate the energy function component of Density Functional Theory (DFT) which describes the quantum mechanical behaviour of molecules.
“By expressing the functional as a neural network and incorporating these exact properties into the training data, we learn functionals free from important systematic errors — resulting in a better description of a broad class of chemical reactions,” said DeepMind researchers.
The researchers used supervised learning to train an MLP neural network. The training dataset consisted of 1161 examples; the inputs contained Kohn-Sham (KS) orbital features sampled on a spatial grid, while the output values were “high-accuracy reaction energies.”
Researchers tested DM21 on three benchmarks, including GMTKN55 and QM9, which contain data for chemical tasks “very distinct” from the training data. DM21 set new state-of-the-art performance on these benchmarks, outperforming four previous methods.
“DM21 is better than the best hybrid functional and approaches the performance of the much more expensive double-hybrid functionals,” said researchers.
DM21 was able to achieve greater accuracy in modelling the quantum mechanical behaviour of electrons.
