Machine learning tools applied to linacs and FELs

My research is mainly involved in the development of machine learning methods to further the understanding of electron beams in modern accelerators. I work with the sychrotron facility MAX IV, where electrons are accelerated to very near the speed of light and utilized to extract X-rays for a varied range of scientific work. Since the electron beams moves at extremely high speeds, it is difficult to extract information of the distribution of the beam in time and energy. A new diagnostic tool has recently been installed in the facility for this exact purpose, but the measurment is destructive and the beam can not be reused after being measured in this way. This is where machine learning can prove useful. In my current project I develop methods for extracting the same information as the new diagnostic tool without disturbing the beam. I collect data from the entire accelerator and train artificial neural networks to map this non-destructive information to the destructive measurments.