Atomistic kinetic Monte Carlo simulation of precipitation and segregation in metals for nuclear applications, using a novel methodology based on artificial neural networks

In this thesis, a novel methodology for modeling the effects of neutron irradiation in steels is proposed. The atomic level is retained, but the point-defects transition rates are calculated with little approximation using an interatomic potential, taking all effects of static relaxation and long-ranged chemical interactions into account. Artificial Neural Networks are used to speed-up the simulation. This novel approach is first successfully applied to the simulation of thermal annealing experiments, highlighting its qualities. Next, convincing first steps are taken towards the application to the simulation of irradiation damage.