BROOD: Bilevel and Robust Optimization and Outlier Detection for Efficient Tuning of High-Energy Physics Event Generators

The parameters in Monte Carlo (MC) event generators are tuned on experimental measurements by evaluating the goodness of fit between data and MC predictions. relative importance each measurement is adjusted manually an often time-consuming, iterative process to meet different needs. In this work, we introduce several optimization formulations algorithms with new decision criteria for streamlining automating process. These designed two formulations: bilevel robust optimization. Both applied datasets used ATLAS A14 tune dedicated hadronization generated SHERPA generator, respectively. corresponding generator compared using three metrics. We compare quality our automatic tunes published tune. Moreover, analyze impact a pre-processing step that excludes cannot be described physics models generators.