Accurate Many-Body Repulsive Potentials for Density-Functional Tight-Binding from Deep Tensor Neural Networks

TL;DR

This paper introduces a hybrid approach combining density-functional tight-binding with deep tensor neural networks to accurately predict molecular properties, outperforming standard methods and enabling large-scale electronic structure calculations.

Contribution

The paper presents a novel DFTB-NN$_{ ext{rep}}$ model that learns many-body interatomic repulsive energies using deep neural networks, improving accuracy over traditional pairwise models.

Findings

Accurate predictions of energies, geometries, and vibrational frequencies.

Outperforms standard DFTB and DFT-PBE0 in tests.

Potential for large-scale, chemically accurate simulations.

Abstract

We combine density-functional tight-binding (DFTB) with deep tensor neural networks (DTNN) to maximize the strengths of both approaches in predicting structural, energetic, and vibrational molecular properties. The DTNN is used to learn a non-linear model for the localized many-body interatomic repulsive energy, which so far has been treated in an atom-pairwise manner in DFTB. Substantially improving upon standard DFTB and DTNN, the resulting DFTB-NN$_{\sf{rep}}$ model yields accurate predictions of atomization and isomerization energies, equilibrium geometries, vibrational frequencies and dihedral rotation profiles for a large variety of organic molecules compared to the hybrid DFT-PBE0 functional. Our results highlight the high potential of combining semi-empirical electronic-structure methods with physically-motivated machine learning approaches for predicting localized many-body interactions. We conclude by discussing future advancements of the DFTB-NN$_{\sf{rep}}$ approach that could enable chemically accurate electronic-structure calculations for systems with tens of thousands of atoms.