Some recent progresses in density-functional theory: efficiency, accuracy, and applicability

Abstract

Three new developments of density functional theory are reported here. First, we have developed a highly efficient first-principles method for excited electronic states, whose computational time scales linearly with the system size. Secondly, Neural Networks is introduced to construct the exchange-correlation functional of density functional theory; and lastly, a rigorous first-principles formalism has been established for open electronic systems.