Machine Learining of Few-Body Scattering Problem

One of a basic problem in few body scattering physice is solving the scattering length while we know details of scattering potential. We propose a new method to solve scattering length based on neural network. Using a two hidden layer fully-connected neural network, we can see it works very well to solve this problem. After training we open the neural network and see how it works while training potential is santisfied perturbation theory. We see it does first and second order Born-Approximation while potential strength is very shallow.