Speaker
Description
Nuclear targets can be polarized through dynamic nuclear polarization using high magnetic fields, low temperatures, and RF. The polarization of the material can be maximized through adjusting the microwave frequency. Human operators typically make the adjustments, but machine learning techniques may be able to automate this system, as well as optimize the polarization. The targets can also be partially repaired through annealing, and machine learning can potentially be used to predict the optimal time to anneal to achieve the maximum overall polarization.
Previous work used Gaussian process-based and multilayer perceptron models to predict polarization of the target and train a reinforcement learning agent for positively polarized targets using CLAS12 data from Run Group C experiments using the APOLLO cryogenic target system. This work expands that to include negative polarization to make progress towards the goal of using surrogate models and reinforcement learning to tune the microwave frequency during experiments. This work also looks at using g2p annealing data to train a model to predict when to anneal.