A Kohn-Sham scheme based neural network for nuclear systems

A Kohn-Sham scheme based multi-task neural network is elaborated for the supervised learning of nuclear shell evolution. The training set composed single-particle wave functions and occupation probabilities 320 nuclei, calculated by Skyrme density functional theory. It found that deduced distributions, momentum charge radii are in good agreements with benchmarking results untrained nuclei. In particular, accomplishing evolution leads to a remarkable improvement extrapolation density. After further charge-radius-based calibration, evolves stronger predictive capability. This opens possibility infer correlations among observables combining experimental data complex systems.