Lasers suffer from long-term pointing instabilities due to environmental and other influences. In addition, UV femtosecond pulse beams, as the front-end laser for SwissFEL, damage relatively fast multiple optics in the beam path. This beam pointing and beam quality degradation is a problem in many lasers and laser applications in science and industry. We studied feasibility of Bayesian optimization method from Machine Learning to optimize the laser beam in those multi-dimensional aspects in a fast and reliable way.

Project Partners:
Prof. Dr. Bojan Resan (FHNW, laser system development)
Dr. Alexandre Trisorio (PSI, laser and accelerator physics)
Dr. Alisa Rupenyan (ETHZ/Inspire, Bayesian optimization in Machine Learning)
Romain Carreto (TLD Photonics AG, laser control development)