Speaker
Ningdong Wang
(Cornell University)
Description
Space-charge beam dynamics simulations are computationally expensive, especially for high-intensity beamlines where accurate 3D field solves dominate runtime. We present a deep surrogate model that predicts full 3D electric fields from particle charge distributions with orders-of-magnitude speedup over traditional Poisson solvers. The surrogate uses a U-Net architecture trained on self-consistent particle-in-cell space charge solver. We describe the data generation strategy, model design, and training methodology, and physics-informed augmentation.
Funding Agency
This work was supported by DOE Tigner Traineeship in Accelerator Science and DOE grant DE-SC0025351.
| In which format do you inted to submit your paper? | LaTeX |
|---|
Author
Ningdong Wang
(Cornell University)
Co-author
Georg Hoffstaetter
(Cornell University (CLASSE))