Machine-learning supported compression of laser pulses in a control loop

Klaus Jünemann, HAW Hamburg

Tim Laarmann, DESY

Laser pulse compression in a neural network based self-learning loop.

Copyright: HAW Hamburg

Owing to its power of extracting essential information from large amounts of data, machine learning is bringing a revolutionary reform to research in the physical sciences. The central goal of the present project application is the development of a neural network based self-learning loop, which connects advanced laser technology, an optimization algorithm, and an experimental feedback signal. Laser pulse manipulation and control in an optimization feedback loop supported by neural networks is a hot topic in modern laser science. It may open-up the door to teach laser beams how to control complex function in matter, materials and in the life sciences on the quantum level of electrons and atomic constituents. Similar approaches have been successfully used for optimization feedback loops at particle accelerator facilities and will certainly enter in the design, operation and application of next generation of high-repetition-rate x-ray FELs. The goal must be to provide accurate knowledge and ultimate control of complex x-ray pulses at MHz repetition rate. These challenges can only be met by closely linking DESY's research activities in these fields on the Campus in the Science City Bahrenfeld with methodologically oriented applied research in data science and applied mathematics at HAW Hamburg.