Speaker
Description
Numerical beam dynamics simulation codes are essential for designing and studying particle accelerators, but their computational cost can make them unsuitable for online use and predictions during operations. The use of machine learning-based surrogate models can significantly reduce the required computational time whilst still providing an accurate prediction of the beam properties. In this paper, we present the first results on the training of surrogate models for the prediction of the longitudinal phase space (LPS) at the European XFEL. Finally, we discuss the potential application of such models in the development of a virtual diagnostic tool for use in the European XFEL control room as well as a fast estimator for the final LPS based on the user-provided compression parameters.
| Paper status | Resubmitted proceeding files received and assigned to an editor. Accepted. |
|---|