Optimizing the IFMIF-DONES Particle Accelerator with Differentiable Deep Learning Surrogate Models
Abstract
In this work, Deep Learning Surrogate Models are employed to optimize the quadrupole values in the initial section of the High Energy Beam Transport Line of the IFMIF-DONES accelerator. Two Fourier Neural Operator models were trained: one for predicting two-dimensional beam profiles and another for forecasting one-dimensional beam statistics along the accelerator's longitudinal axis. These models offer up to 3 orders of magnitude speedup compared to traditional simulations, with a trade-off of maintaining accuracy within percentage errors below 6$\%$. Moreover, their differentiability allows seamless integration with optimization algorithms, enabling efficient tuning of quadrupole values to achieve specific beam objectives. This approach offers a robust solution for enhancing the performance of IFMIF-DONES accelerator and other scientific experiments.
Cite
Text
Romero et al. "Optimizing the IFMIF-DONES Particle Accelerator with Differentiable Deep Learning Surrogate Models." NeurIPS 2024 Workshops: D3S3, 2024.Markdown
[Romero et al. "Optimizing the IFMIF-DONES Particle Accelerator with Differentiable Deep Learning Surrogate Models." NeurIPS 2024 Workshops: D3S3, 2024.](https://mlanthology.org/neuripsw/2024/romero2024neuripsw-optimizing/)BibTeX
@inproceedings{romero2024neuripsw-optimizing,
title = {{Optimizing the IFMIF-DONES Particle Accelerator with Differentiable Deep Learning Surrogate Models}},
author = {Romero, Galo Gallardo and Rodríguez-Llorente, Guillermo and Rodríguez, Lucas Magariños and Navascués, Rodrigo Morant and Petrovsky, Nikita Khvatkin and Martín, Roberto Gómez-Espinosa},
booktitle = {NeurIPS 2024 Workshops: D3S3},
year = {2024},
url = {https://mlanthology.org/neuripsw/2024/romero2024neuripsw-optimizing/}
}