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

TL;DR

This paper introduces new bilevel and robust optimization algorithms to automate and improve the tuning of Monte Carlo event generators in high-energy physics, reducing manual effort and enhancing tuning quality.

Contribution

It presents novel optimization formulations and algorithms for automating MC generator tuning, applied to ATLAS and sherpa datasets, with comparative analysis of results.

Findings

Automatic tuning quality comparable to manual ATLAS A14 tune

Pre-processing data improves model fit and tuning accuracy

Algorithms effectively handle outliers and complex parameter spaces

Abstract

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