Speaker
Description
We use Machine Learning with the event-generator (Sar$t$re) for
the for exclusive diffraction processes in electron-nucleus scattering. The second moment of the amplitude includes averaging of hundreds of initial nucleon configurations of the heavy nucleus, a procedure which is extremely CPU intense. We show that we can make the production of the cross-sections used for event generation at least 90% more efficient using machine learning. Sartre is using lookup tables for the first and second amplitude of the interaction. With our new technique, the production time of these lookup tables is reduced from several months to a few days. This will enable us to generate exclusive diffraction events for all heavy nuclear species, with any exclusive final state.
