GEANT4 Parameter Tuning Using Professor

TL;DR

This paper explores tuning Geant4 simulation parameters using the Professor framework to improve agreement with experimental data, aiming to better quantify uncertainties in high energy physics measurements.

Contribution

It demonstrates the impact of parameter variations on Geant4 models and introduces a framework for propagating model uncertainties to physics measurements.

Findings

Parameter tuning improves dataset agreement

More degrees of freedom needed for full accuracy

Framework sets foundation for uncertainty propagation

Abstract

The Geant4 toolkit is used extensively in high energy physics to simulate the passage of particles through matter and to predict effects such as detector efficiencies and smearing. Geant4 uses many underlying models to predict particle interaction kinematics, and uncertainty in these models leads to uncertainty in high energy physics measurements. The Geant4 collaboration recently made free parameters in some models accessible through partnership with Geant4 developers. We present a study of the impact of varying parameters in three Geant4 hadronic physics models on agreement with thin target datasets and describe fits to these datasets using the Professor model tuning framework. We find that varying parameters produces substantially better agreement with some datasets, but that more degrees of freedom are required for full agreement. This work is a first step towards a common framework for propagating uncertainties in Geant4 models to high energy physics measurements, and we outline future work required to complete that goal.