Integration of Machine Learning-Based Plasma Acceleration Simulations into Geant4: A Case Study with the PALLAS Experiment

2025/03/15 by Sytov, A., Cassou, K., Kubytskyi, V. +2
#Accelerator Physics (physics.acc-ph) #FOS: Physical sciences #Plasma Physics (physics.plasm-ph)

paper · doi:10.48550/arxiv.2503.12154

Abstract

We present the development and integration of a Machine Learning (ML)-based surrogate model, trained on Particle-In-Cell (PIC) simulations of laser-driven plasma wakefield acceleration source of electrons, into Geant4 simulation toolkit. Our model enables the generation and tracking of plasma-accelerated beams within complete experimental setups, unifying plasma acceleration and Monte Carlo-based simulations, which significantly reduces their complexity and computational cost. Our implementation focuses on the PALLAS laser-plasma accelerator test facility, integrating its full experimental setup into Geant4. We describe the ML model, its integration into Geant4, and key simulation results, demonstrating the feasibility of start-to-end simulations of plasma acceleration facilities and applications within a unified framework.

Related

• Accelerating kinetic plasma simulations with machine learning generated initial conditions
• Surrogate Models studies for laser-plasma accelerator electron source design through numerical optimisation
• The Artificial Scientist -- in-transit Machine Learning of Plasma Simulations
• From Compact Plasma Particle Sources to Advanced Accelerators with Modeling at Exascale
• Invertible Surrogate Models: Joint surrogate modelling and reconstruction of Laser-Wakefield Acceleration by invertible neural networks