Pushing the frontiers of density functionals by solving the fractional electron problem

Density functional theory describes matter at the quantum level, but all popular approximations suffer from systematic errors that arise from the violation of mathematical properties of the exact functional. We overcame this fundamental limitation by training a neural network on molecular data and on fictitious systems with fractional charge and spin. The resulting functional, DM21 (DeepMind 21), correctly describes typical examples of artificial charge delocalization and strong correlation and performs better than traditional functionals on thorough benchmarks for main-group atoms and molecules. DM21 accurately models complex systems such as hydrogen chains, charged DNA base pairs, and diradical transition states. More crucially for the field, because our methodology relies on data and constraints, which are continually improving, it represents a viable pathway toward the exact universal functional.

Automated summary

Density functional theory has long been plagued by systematic errors in approximations, but a new neural network-based functional, DM21, shows promise in accurately describing complex systems and outperforming traditional functionals in benchmarks. By relying on data and constraints, DM21 represents a viable pathway toward the exact universal functional.