Quark Mass Models and Reinforcement Learning

A bstract In this paper, we apply reinforcement learning to the problem of constructing models in particle physics. As an example environment, use space Froggatt-Nielsen type for quark masses. Using a basic policy-based algorithm show that neural networks can be successfully trained construct which are consistent with observed masses and mixing. The policy lead from random phenomenologically acceptable over 90% episodes after average episode length about 20 steps. We also capable finding proposed literature when starting at nearby configurations.