DeePKS: a comprehensive data-driven approach towards chemically accurate density functional theory

TL;DR

DeePKS introduces a machine learning framework that creates accurate, adaptable density functionals for quantum chemistry, capable of improving with more data and providing reliable predictions across diverse molecules.

Contribution

This work presents a novel data-driven method for constructing density functionals that are both accurate and broadly applicable within Kohn-Sham DFT, advancing the integration of machine learning in quantum chemistry.

Findings

Achieves chemically accurate energy and property predictions

Demonstrates adaptability with increasing data

Provides consistent results across diverse molecular systems

Abstract

We propose a general machine learning-based framework for building an accurate and widely-applicable energy functional within the framework of generalized Kohn-Sham density functional theory. To this end, we develop a way of training self-consistent models that are capable of taking large datasets from different systems and different kinds of labels. We demonstrate that the functional that results from this training procedure gives chemically accurate predictions on energy, force, dipole, and electron density for a large class of molecules. It can be continuously improved when more and more data are available.