Maintaining optimal beam brightness and luminosity using machine learning

While human operators are very good at finding the optimal conditions that maximize the luminosity in RHIC, maintaining those conditions can be a demanding task. There is also no metric that can measurably determine if those optimal conditions are truly optimal, given the degrees of freedom are very high and there are multiple and competing objectives. In this talk we will describe how we use machine learning and improved physics models to build systems for optimizing the beam brightness during injection at the BNL Booster and AGS synchrotrons and efforts to maintain maximum luminosity in RHIC.