Data for: Neural network self-consistent fields for density functional theory

It contains the data generated and used in the study of NeuralSCF, a machine-learning framework that models the Kohn-Sham density map and emulates self-consistent field iterations to predict ground-state electron densities and derived electronic-structure properties. The study evaluates NeuralSCF on molecular datasets including QM9, MD17/MD22 trajectories, BFDb-SSI non-covalent systems, bond-rotation examples, and larger QMugs molecules, using PySCF calculations with the cc-pVTZ basis, PBE exchange-correlation functional, and cc-pVTZ-AutoAux density representation.

Source data for the figures and related computational outputs are included for reproducibility.