Machine Learning Approach to Study of Low Energy Alpha-Deuteron Elastic Scattering using Phase Function Method

Abstract

Central idea: To obtain the interaction potential using the inverse scattering method, we have employed the Physics-Informed Machine Learning (PIML) approach. In this framework, the machine learning algorithm is guided by the underlying physical laws, enabling the accurate extraction of the inverse scattering potential from the elastic scattering data. Methodology: As a reference potential, a combination of three smoothly joined Morse functions has been utilized, characterized by ten model parameters. These parameters are optimized in an iterative fashion using a Genetic Algorithm to ensure the best fit to the phase shifts extracted from the experimental scattering data. The process of optimization is guided by the computed scattering phase shifts by solving the phase equation using 5th order RK-method for the reference potential in each iteration Results: Our approach yields inverse potentials for both single and multi channel scattering. Using the Scattering Phase Shifts obtained from these inverse potentials, we calculate the partial cross-section to determine the resonance energies and decay width. The obtain values of resonance energies and decay width for 3D1, 3D2 and 3D3 states of alpha-deuteron are in correspondence with the experimental results. Conclusion: It can be concluded that our machine learning-based approach for constructing the inverse potential offers a novel and complementary technique to existing direct methods.