Machine Learning in Fast Beam Diagnositcs

In accelerator physics, detailed studies of beam often require disruptive measurement and take a long time to perform. This includes the measurement of the emittance where a quadruple scan is needed. In this report, a method using machine learning of fast emittance estimation is introduced. Using the image of the beam and the machine settings of the experiment, one can achieve a single % error estimation of emittance across a large range. The study is conducted using simulation data from ASTRA.