Pär Håkansson

My current research interests concerns the interpretation of NMR relaxation experiments involving the questions: 1) what microscopic information about molecular order and dynamics can we deduce via computational modeling? 2) How can we extend these modeling capabilities with machine learning to build property surfaces of the most costly modeling components? These questions are typically addressed with atomistic or Brownian Dynamics simulations to mimic the molecular dynamics and interaction tensors for the spin-system computed with quantum chemistry methods. Probability methods such as Gaussian processes and Polynomial chaos expansions are of interest in the construction of property surfaces.