Artificial intelligence sees split electrons

Article Multidisciplinary Sciences

Interpretations of ground-state symmetry breaking and strong correlation in wavefunction and density functional theories

John P. Perdew et al.

Summary: Strong correlations within a symmetry-unbroken ground-state wavefunction may manifest in approximate density functional theory as symmetry-broken spin densities or total densities, arising from soft modes of fluctuations such as spin-density or charge-density waves. An approximate density functional that breaks symmetry can be more revealing than an exact functional that does not, with examples including the stretched H-2 molecule, antiferromagnetic solids, and the static charge-density wave/Wigner crystal phase of a low-density jellium. Time-dependent density functional theory quantitatively shows that the static charge-density wave is a soft plasmon, with the frequency of a related density fluctuation dropping to zero.

PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA (2021)

Add to Collection

Article Chemistry, Physical

What Types of Chemical Problems Benefit from Density-Corrected DFT? A Probe Using an Extensive and Chemically Diverse Test Suite

Golokesh Santra et al.

Summary: The performance of HF-DFT relative to self-consistent DFT under different XC functionals has been studied. HF-DFT is highly beneficial for subsets dominated by dynamical correlation, particularly at low HF exchange percentages. However, in subsets with significant nondynamical correlation, HF-DFT may do more harm than good.

JOURNAL OF CHEMICAL THEORY AND COMPUTATION (2021)

Add to Collection

Article Multidisciplinary Sciences

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

James Kirkpatrick et al.

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.

SCIENCE (2021)

Add to Collection

Article Multidisciplinary Sciences