Speaker
Description
Monte-Carlo (MC) simulations play a key role in high energy physics. MC generators and their interfaces to the experiment-specific software framework evolve continuously. Thus, a periodic validation is indispensable for obtaining reliable and reproducible physics simulations. For that purpose, ATLAS has developed a central semi-automated validation system: PMG Architecture for Validating Evgen with Rivet (PAVER). It evaluates each new revision of a MC generator as well as each software update of the full MC production chain. Distributions of observables are generated using publicly available Rivet routines, enabling direct comparisons to MC reference distributions and to measured data.
PAVER is operated via a web interface that enables a straightforward configuration and highlights distributions that changed compared to earlier generator or software versions. This way, potential issues can be detected in small-scale test samples before generating large-scale samples for the collaboration, which is crucial for a sustainable and low-cost MC production procedure at ATLAS.