MC-TESTER: a universal tool for comparisons of Monte Carlo predictions for particle decays in high energy physics

TL;DR

MC-TESTER provides a ROOT-based, two-stage framework to compare Monte Carlo decays across generators by automatically extracting decay channels from event records and producing a downloadable booklet of invariant-mass histograms and ratios. It introduces three Shape Difference Parameter ($SDP$) algorithms to quantify distribution differences, with options for user-defined SDP via SETUP.C and wrappers to integrate with FORTRAN77 and C++ generators. The tool outputs a data file mc-tester.root and a LaTeX booklet, enabling rapid, automated validation of decay predictions (demonstrated with TAUOLA and PYTHIA for tau decays) and easy extension to other generators and future features. While it provides a practical, extensible testing framework, it omits certain physics aspects (e.g., intermediate states, polarization, Levi-Civita tensor effects) and highlights memory and scaling considerations for high-multiplicity decays, guiding users toward cautious, mid-scale analyses with planned enhancements.

Abstract

Theoretical predictions in high energy physics are routinely provided in the form of Monte Carlo generators. Comparisons of predictions from different programs and/or different initialization set-ups are often necessary. MC-TESTER can be used for such tests of decays of intermediate states (particles or resonances) in a semi-automated way. Our test consists of two steps. Different Monte Carlo programs are run; events with decays of a chosen particle are searched, decay trees are analysed and appropriate information is stored. Then, at the analysis step, a list of all found decay modes is defined and branching ratios are calculated for both runs. Histograms of all scalar Lorentz-invariant masses constructed from the decay products are plotted and compared for each decay mode found in both runs. For each plot a measure of the difference of the distributions is calculated and its maximal value over all histograms for each decay channel is printed in a summary table. As an example of MC-TESTER application, we include a test with the tau lepton decay Monte Carlo generators, TAUOLA and PYTHIA. The HEPEVT (or LUJETS) common block is used as exclusive source of information on the generated events.

Figures (8)

• Figure 1: Example of booklet's informational pages produced at analysis step. At the bottom of the table \ref{['fig-booklet1']}(b), the $T_1$, $T_2$ coefficients quantifying the difference in all decay channels combined are printed (see chapter \ref{['subsec:recent_updates']} for details).
• Figure 2: Examples of plots produced at analysis step.
• Figure 3: Comparison of two histograms with SDP$=0.177~$ (in the limit of infinite samples SDP would equal $0.20$).
• Figure 4: Comparison as in the previous figure, but samples $1000$ times smaller. SDP$=0.00139~$ now.
• Figure 5: Two completely disjoint histograms. SDP$=1.97$.