Aktivität: Vorträge › Vortrag › Science to Science
Beschreibung
We develop a machine learning (ML) approach that maps the random-phase approximation (RPA) to a pure density functional. The ingredients for the ML-RPA energy density are only averages of the electronic density and its gradient in some real-space environment. That is, our ML-RPA functionals can be considered as non-local extensions to the usual gradient approximations. RPA exchange-correlation potentials obtained via the optimized effective potential method serve as derivative information for the ML fit. This greatly enhances the data set size contrast to common approaches using only energies for fitting. We apply the ML-RPA method to create an RPA substitute for elements C, H, and O. Evaluation on molecules, surfaces and bulk materials shows that our ML-RPA functional can capture covalent and van der Waals interactions as well as hydrogen bonding.