Predicting electronic structures at any length scale with machine learning

The properties of electrons in matter are fundamental importance. They give rise to virtually all molecular and material determine the physics at play objects ranging from semiconductor devices interior giant gas planets. Modeling simulation such diverse applications rely primarily on density functional theory (DFT), which has become principal method for predicting electronic structure matter. While DFT calculations have proven be very useful point being recognized with a Nobel prize 1998, their computational scaling limits them small systems. We developed machine learning framework any length scale. It shows up three orders magnitude speedup systems where is tractable and, more importantly, enables predictions scales infeasible. Our work demonstrates how circumvents long-standing bottleneck advances science frontiers intractable current solutions. This unprecedented modeling capability opens an inexhaustible range astrophysics, novel materials discovery, energy solutions sustainable future.