Simulation validation epistemics in a Geant4 case study

Abstract

This paper summarizes the results of a three year long project concerning the validation of Geant4-based simulation of electron backscattering. Accurate simulation of backscattering is of great importance in many experimental scenarios from calorimetry to medical dosimetry, as it affects the pattern of energy deposition in detectors or other kinds of sensitive volumes. All Geant4 multiple and single scattering models, as well as Geant4 recommended settings for PhysicsLists, have been involved in a validation test against several thousand experimental measurements documented in the literature. Comparison with experiment is carried out through rigorous statistical methods. The effect of dependencies of physics modeling on other parts of the software has been investigated. The discussion of this validation test, its methodology and its statistical analysis tools offer the opportunity to review the foundations of simulation validation epistemology and provide concrete guidelines to the experimental community to assess the reliability of simulation applications in experiments.