Reweighting Method for Lund String Breaks in PYTHIA

Abstract

Event generators are useful for simulating collision experiments in high-energy particle physics. In the event generator PYTHIA 8, parameters may be varied to compare competing models against experimental data. It is then beneficial to employ reweighting techniques to explore the results for multiple parameter values with only one simulation.

This study aims to develop and implement a reweighting method for meson production in electron-positron collisions, within the Lund string model for hadronization. We expect the number of ss¯, uu¯ and dd¯ string breaks to follow a multinomial distribution and the produced mesons to distribute correspondingly. Thus, a ratio between two multinomial mass functions using the string break probabilities for two different sets of parameters is developed and used as statistical weight for each event. By varying the s ¯ s suppression parameter as well as η and η′ rejection parameters the weight for each event is calculated and applied to the distribution of final s ¯ s breaks and mesons from a set of test simulations. The reweighted test distributions are compared to target distributions for comparison.

The results show a high accuracy of the method when applied to the number of string breaks but much lower for the number of final mesons, pointing towards a discrepancy between the predicted correlation of string breaks and mesons and the true correlation.

The reweighting technique introduced can be naturally extended to reweighting around the mixing angles of the pseudoscalar and vector mesons as well as the vector-topseudoscalar suppression factor also present. It can also be further generalised to include baryons and account for hadron decays. In the end, one hopes to use this method for comparison with experimental data.

Popular Abstract

The event generator is a powerful tool in particle physics which simulates particle collisions. The event generators generate much simulated data. To reduce the generated data, techniques can be implemented that enable us to take shortcuts in the computing process. One of these shortcuts is reweighting, which makes a theoretical prediction of how varying input parameters would change the results. If this is possible, one only has to run the simulation once and then simply reweight the results to explore different sets of outcomes.

Say you want to simulate a collision between two high-energy particles just as it is done at CERN’s large accelerators. You then need to know how the particles behave before they reach the detectors. This is the job of event generators, which use physics models implemented as computer algorithms to predict exactly how a particle collision would behave in real life. Then, you want to run more simulations to obtain more data. You may want to do one hundred, or why not a million, or even a hundred million simulations. This is such that the amount of data from the simulations is comparable to the huge amount from the real experiments. Now consider, you have two different values for an input parameter you would like to compare. You would then have to produce two hundred million simulations. What if you want to vary fifty parameters, each over a thousand values? It is easy to see how this will become incredibly time, energy and money-consuming. By narrowing down on one part of the simulation known as hadronization, we can simplify the problem.

Hadronization is the process of constituent particles known as quarks turning into composite particles called hadrons. Picture two quarks connected by a string. If the string has enough energy, it can break in two, producing a quark in each new string end. The breaking can continue until the energy of the string is depleted. The quarks in each string piece then combine to form particles known as mesons. This is the Lund String model of hadronization.

The quarks produced in the string breaks are of three types: up, down and strange, each with an associated probability of being produced. This can be directly linked to the probability of producing a certain number of hadrons in a simulation. Modifying the individual quark probabilities by altering the input parameters, we can predict how the number of hadrons will change and adjust the result accordingly. Instead of running a new simulation for each parameter value, the original simulation can instead be used to obtain all the results. This is the basis of the reweighting method used in the thesis.

• author: Engman, Hugo
• supervisor: Christian Bierlich
• organization: Department of Physics; Particle and nuclear physics
• course: FYSK04 20241
• year: 2024
• type: M2 - Bachelor Degree
• subject: Physics and Astronomy
• language: English
• id: 9164963
• date added to LUP: 2024-06-24 08:29:38
• date last changed: 2024-06-24 08:29:38