February 12, 2025:
Dr. Petros Steanou
Affiliation: Universidad de Alicante, Spain
Title: 3D neutron star magnetospheres with physics-informed neural networks
Abstract: Physics-Informed Neural Networks (PINNs) have emerged recently as a powerful new method for solving PDEs. They leverage state-of-the-art machine learning techniques to provide surrogate solutions that satisfy the equations that govern a physical system. Recent advancements in the field have established PINNs as competitive alternatives to traditional numerical methods in terms of efficiency and accuracy, even surpassing them for high dimensional problems. Furthermore, as flexible and mesh-less solvers, they can be easily adapted to different physical setups. We employ our PINN solver to find solutions of the force-free field in the magnetosphere of neutron stars in various 3D configurations. Our results pave the road for future extensions of the problem that can shed light to prominent questions in the field.
More