An Adjoint Formulation of Energetic Particle Confinement

2025/11/15 by McDevitt, Christopher J., Arnaud, Jonathan S.
#FOS: Physical sciences #Plasma Physics (physics.plasm-ph)

paper · doi:10.48550/arxiv.2511.11968

Abstract

An adjoint formulation of energetic particle confinement in axisymmetric geometry is derived and evaluated using a Physics-Informed Neural Network (PINN). The PINN estimates the escape time of energetic ions by solving an inhomogeneous adjoint of the drift kinetic equation with a Lorentz collision operator, yielding predictions of the escape time of fast ions in tokamak geometry due to direct ion orbit loss and collisional transport. This is the first time a PINN has been used to solve the drift kinetic equation in tokamak geometry, a challenging problem due to the large time scale separation present between the rapid transit time of energetic ions, and their slow collision time scale. It is shown that a careful and intentional design of a PINN is able to learn the escape time for the majority of the geometry considered, suggesting a path toward constructing a rapid surrogate for use in a broader optimization framework.

Citations

• Generative AI Models for Learning Flow Maps of Stochastic Dynamical Systems in Bounded Domains
• A Runaway Electron Avalanche Surrogate for Partially Ionized Plasmas
• Gradient Alignment in Physics-informed Neural Networks: A Second-Order Optimization Perspective
• Optimizing the Optimizer for Physics-Informed Neural Networks and Kolmogorov-Arnold Networks
• A Physics-Constrained Deep Learning Treatment of Runaway Electron Dynamics
• From PINNs to PIKANs: Recent Advances in Physics-Informed Machine Learning
• SOAP: Improving and Stabilizing Shampoo using Adam
• A physics-constrained deep learning surrogate model of the runaway electron avalanche growth rate
• The Impact of Collisionality on the Runaway Electron Avalanche during a Tokamak Disruption
• Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's next
• Physics-informed neural networks (PINNs) for fluid mechanics: A review
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data
• Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Adam: A Method for Stochastic Optimization
• On the limited memory BFGS method for large scale optimization

Related

• An efficient numerical method for charged particle transport based on hybrid collision model and machine learning
• A Physics-Constrained Deep Learning Treatment of Runaway Electron Dynamics
• Application of Neural Ordinary Differential Equations for Tokamak Plasma Dynamics Analysis
• Development of grid-based and PINN solvers for electron kinetics in collisional non-thermal plasmas
• Encoder-decoder neural network for solving the nonlinear Fokker-Planck-Landau collision operator in XGC