Humboldt-Universität zu Berlin - Mathematisch-Naturwissen­schaft­liche Fakultät - Experimentelle Elementarteilchenphysik


Professor represents a systematic approach to the subject of Monte Carlo event generator tuning. The idea is to interpolate the generator response under parameter changes and to fit this approxmiation to high precision data. It is of universal character, i.e applicable to any generator that has an interface to the standard HepMC event record. So far, fortran Pythia has been the focus of our work. Our strategy was to consecutively tune distinct generator aspects one after another. First, we tuned the flavour parameters to LEP-data (multiplicities of identified hadrons + some OPAL spectra). The second stage was to tune fragmentation parameters to general event shape observables (Thrust, etc.) measured by LEP. Recently, we successfully tuned Pythias modeling of the 'Underlying event' (multiple interactions, initial state radiation, beam-beam remnants, colour reconnection) to TEVATRON data (CDF and D0). In general, the Professor tunings yield impressive results compared to both, the quality of and the time spent to achieve the default tunings. The Professor tunings of fortran Pythia have found there way into the release of version 6.4.20, the last release ever.

The Professor project is hosted on HepForge. link


The following plots compare the tuning obtained with Professor to standard tunings.

Tune A was done by Rick Field and has been the standard tuning for Underlying Event

studies (most important background) for several years. The ATLAS tuning was done by

A. Moraeaes and is considered to be one of the standard tunings used in the ATLAS experiment.


CDF: Z-Boson p_T distribution


Z-Boson p_T distribution, measured by CDF

here we are very sensitive to gluon emissions prior to the hard 2->2 scattering since
initial state radiation of a gluon is more or less the only chance for the Z-Boson to gain
a p_T different from zero.
It is quite obvious that this observable was not looked at while doing tune A.


CDF: Average Lepton Pair p_T vs. Charged Multiplicity


 average p_T of the lepton pair vs. charged multiplicity in Drell-Yann events


This is a measurement from TEVATRON Run 2 @ 1960 GeV where one is interested in the average p_T of the lepton pair

from reconstructed Drell-Yan events (qqbar --> Z0 --> llbar) as a function of the overall charged multiplicity.

The ATLAS tuning dramatically overshoots the data, tune A produces too few soft events, while the systematic tuning (Professor) is dead on.