Machine Learning Applications for the SBS Hadron Calorimeter

The new HCAL-J is a segmented hadron calorimeter constructed to measure the energy of several GeV protons and neutrons. HCAL-J will be used this fall in the initial Super BitBite (SBS) collaboration series of experiments in Jefferson Lab's Hall A, beginning with the G_Mⁿ experiment which will precisely measure the magnetic form factor of the neutron. HCAL-J is composed of 288 individual calorimeter modules measuring 15cmx15cmx1m. These modules consist of 40 layers of iron, which cause the hadrons to shower, alternating with 40 layers of scintillator, which sample the energy. HCAL-J has a time resolution of 0.5 ns, a position resolution as good as 3-4 cm, and detects protons and neutrons with near identical efficiency. This talk will address efforts to implement neural network based cluster finding algorithms and particle identification for HCAL-J, with the goal of reducing background signals during these experiments. Plans to implement these neural networks on FPGAs to be used as a real time experimental trigger for HCAL-J will also be discussed. If successful a neural network based trigger could select physics events over background more efficiently than traditional methods, collecting cleaner data samples and saving beam time.