RF photoinjector modeling and machine learning optimization in NSRRC

Chun-Yan Lin^1*, An-Ping Lee¹, Wai-Keung Lau¹

¹Light Source Division, NSRRC, Hsinchu, Taiwan

* Presenter:Chun-Yan Lin, email:yuweifzr@hotmail.com

Photoinjector are the source of low-emittance electron beams used in many advanced accelerator facilities. In National Synchrotron Radiation Research Center (NSRRC), an S-band photoinjector has been built for light source development. In order to improve the quality of output electron beam, we use the ASTRA program developed by DESY to model beam dynamics in the system. Physical simulations are powerful tools for charged particle accelerators system design and modeling. But in order to optimize system performance efficiently, we need to introduce sufficiently accurate machine learning model for this purpose.

Keywords: Photoinjector, emittance, machine learning, Genetic algorithm, neural network