I describe the work of the CEDAR collaboration in developing tools for tuning and validating Monte Carlo event generator programs. The core CEDAR task is to interface the Durham HepData database of experimental measurements to event generator validation tools such as the UCL JetWeb system - this has necessitated the migration of HepData to a new relational database system and a Java-based interaction model. The "number crunching" part of JetWeb is also being upgraded, from the Fortran HZTool library to the new C++ Rivet system and a generator interfacing layer named RivetGun. Finally, I describe how Rivet is already being used as a central part of a new generator tuning system, and summarise two other CEDAR activities, HepML and HepForge.
The parameters tuning of event generators is a research topic characterized by complex choices: the generator response to parameter variations is difficult to obtain on a theoretical basis, and numerical methods are hardly tractable due to the long computational times required by generators. Event generator tuning has been tackled by parametrisation-based techniques, with the most successful one being a polynomial parametrisation. In this work, an implementation of tuning procedures based on artificial neural networks is proposed. The implementation was tested with closure testing and experime…
Apprentice is a tool developed for event generator tuning. It contains a range of conceptual improvements and extensions over the tuning tool Professor. Its core functionality remains the construction of a multivariate analytic surrogate model to computationally expensive Monte-Carlo event generator predictions. The surrogate model is used for numerical optimization in chi-square minimization and likelihood evaluation. Apprentice also introduces algorithms to automate the selection of observable weights to minimize the effect of mis-modeling in the event generators. We illustrate our improveme…
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 d…
