Dynamics of Ion Cyclotron Wave under Perturbed Environment using Physics Informed Neural Networks

We apply Physics Informed Neural Networks (PINN) to study the dynamics of ion cyclotron waves in plasma in a perturbed environment. We investigate the propagation of bright one-soliton solutions under three different physical conditions: linear damping in collisionless plasma and collisional plasma, and thermal fluctuations of the magnetic field. We investigated evolution of the soliton by analyzing its dynamical behavior in terms of amplitude, energy, and momentum. For linear damping, both in collisional and collisionless plasma environments, the soliton dynamics were found to be irreversible, characterized by exponential decay of both energy and momentum. In contrast, under thermal fluctuations of the magnetic field, the soliton remains stable as the total energy and amplitude are conserved on average. We also demonstrate that PINN achieves an accuracy comparable to the inverse scattering transform method, while being significantly simpler to implement. Source codes are available at https://github.com/gautamksaharia/pFLE