Gaussian Processes Optimization of atomic structure for collisional problems calculation

A. M. P. MENDEZ; J.I. DI FILIPPO; S.D. LOPEZ; D.M. MITNIK

Deauville

Conferencia; The 20th International Symposium on Correlation, Polarization and Ionization in Atomic and Molecular Collisions (COPIAMC); 2019

ICPEAC

Synopsis The accurate description of the target is required when describing collisional processes. Generally, theatomic structure is obtained through the optimization of electronic configurations and parameters to minimizethe differences between the numerical and experimental values of the energies and the oscillator strengths. Byimplementing a machine learning approach through a Bayesian Optimization with Gaussian Processes, we canreproduce the experimental atomic structure with high accuracy. Results for neutral beryllium are presented.