Improved tuning methods for Monte Carlo generators

TL;DR

This paper introduces enhanced tuning methods for Monte Carlo event generators in high energy physics, utilizing Bayesian optimization and adaptive interpolation to improve parameter fitting stability and accuracy.

Contribution

The paper develops and tests extensions to the Professor MC tuning system, incorporating Bayesian methods and adaptive algorithms for better optimization of generator parameters.

Findings

Improved stability of tuning results with new extensions

Enhanced optimization accuracy in Pythia8 tuning

Demonstrated performance gains over standard methods

Abstract

The Monte Carlo event generators (MC) are used for the simulation of different processes in high energy physics. To achieve the best description of the data, the parameters of simulations are adjusted (tuned) with different methods. In this thesis extensions of the Professor MC tuning system were developed and tested. The extensions improve the optimization algorithm for the search of the MC parameters that provide the best description of data. The first extension enables a Bayesian approach in the optimization procedure and the second implements a new adaptive interpolation algorithm for a search of the optimum in the MC parameter space. The performance of the developed extensions was studied with a tuning of the Pythia8 MC event generator and visible improvements in the stability of the results were found in comparison to the results delivered by the standard approach.