Speaker
Michele Ceriotti
(EPFL Lausanne)
Description
Machine-learning models have greatly extended the scope of application of atomistic simulations, that can evaluate, with predictive accuracy, the stability and properties of materials and molecules. In this lecture, I will briefly overview traditional modeling techniques that can be used to estimate vibrational (IR, Raman, ...) and electronic spectroscopies (optical excitations in particular). I will then discuss how these methods can be combined with data-driven approaches, based on machine-learned approximations of the interatomic potentials and of atomic-scale dielectric and electronic response properties.
