Simulating Experimental Beamline Conditions using Pepper Pot Emission Monitoring and Bayesian Optimization

A realistic simulation of a beamline is challenging to create as the exact initial conditions of the particle source (at injection) must be measured experimentally. An alternative to this, used in this experiment, is the employment of a pepper-pot emittance monitor (PPEM) to obtain some of the quantities crucial to the development of such a simulation such as emittance, particle density and momentum. A new pepper-pot screen and monochromatic camera were affixed to an PPEM attachment for the Nanogan beamline at RCNP, U. Osaka. As the PPEM is modular, it was then placed at various points along the Nanogan track and data was collected at each point. Analysis was then performed on the data collected using the PPEM and the aforementioned quantities were obtained. A program making use of Bayesian optimization to optimize simulation parameters such as quadrupole or solenoid strength was then written and used to replicate experimentally-obtained values for emittance in the simulation. In particular, X and Y emittance have been replicated closely using the Bayesian optimization software. This program works well with regards to replicating numerical quantities like emittance at specific points in the simulation, but replication of more complex characteristics (phase space and momentum distributions, particle density) will require the additional development of a machine learning program to assign numerical value to images displaying these distributions before passing this value on to the Bayesian program.