Speaker
Description
Control of 6-dimensional beam phase space is one of the most important steps for high-brightness X-ray Free Electron Lasers (XFELs). Such beam manipulation can be properly performed through accurate beam diagnostics. Recently, generative phase space reconstruction (GPSR) based on neural networks and differentiable simulations has been developed for robust phase space diagnostics, including coupled information. We incorporated GPSR method into conventional beam diagnostics techniques, such as quadrupole and RF cavity phase scans, to enable more efficient and complete 6-dimensional phase space reconstruction using standard accelerator elements. Both simulations and experimental demonstrations at the Pohang Accelerator Laboratory Free Electron Laser facility show that the reconstructed phase space accurately predicts not only the training and test datasets but also other phase spaces further downstream in the beamline that are not included in the GPSR training. These observations reveal that the reconstructed phase space closely resembles the ground-truth physical distribution and opens a path for robust beam diagnostics using elements readily available in many accelerator facilities.
| I have read and accept the Privacy Policy Statement | Yes |
|---|