Implementation of Novel NLO Matching Schemes for p p > Z Z

Abstract

The task of matching fixed-order event generation with the parton shower is physically non-trivial and closely related to an efficiency bottleneck of event generators. In this thesis, we examine and implement two novel matching schemes in KrkNLO and MAcNLOPS for the LHC process p p > Z Z. As both schemes generate positive-definite event samples, they can be interesting alternatives to MC@NLO. The implementation of KrkNLO will be modified compared to original formulation, allowing MSbar PDFs to be used. Whereas for MAcNLOPS, this will be its first implementation. Both implementations rely on MadGraph5_aMC@NLO (MG5_aMC) and PYTHIA8 for hard event generation and parton shower respectively, using also elements of MG5_aMC for the matching procedure. The results are obtained for various distributions sensitive to matching schemes or valuable for practical purposes. In both cases, the results agree well with MC@NLO up to the higher order effects. For KrkNLO, it was additionally observed to be insensitive to factorisation scheme choice.

Popular Abstract

Throughout the history of physics, the dominant philosophy had generally been inclined towards reductionism. This belief in the possibility of understanding complex phenomena in terms of basic constituents motivated the ultimate goal of finding truly fundamental laws of nature. An incredible advancement towards this goal took place in the past century, where physicists managed to identify the most basic constituents of matter, in the form of fundamental particles, and proposed a theory describing their interactions. This theoretical framework with the ambitious name of the Standard Model is the most successful theory of known elementary particles and their non-gravitational interactions.

However, despite being powerful and conceptually elegant, using the Standard Model entails extremely complicated calculations, posing a huge challenge for connecting theory with experiment. In fact, the data obtained from detectors at collider experiments are cannot be directly interpreted without sophisticated reconstruction and simulation tools. To translate them to physically meaningful results, computer simulations must be used. These simulations start from theory and give the expected results at an experiment, hence can be used backwards to comprehend the actual data. They are known as event generators and are now indispensable tools in particle physics.

As event generators are so important, their efficiency is a huge concern, especially considering the pending upgrades to collider experiments. Take the High-Luminosity Large Hadron Collider as an example. When its construction is complete in 2030, the data it is expected to collect per year is 1.5 times the total data collected in the entire three years of run 2. This is a great challenge on the computation front, both in terms of CPU usage and storage capacity. Unfortunately, bottlenecks on the efficiency of event generators exist due to physical constraints. One of the more prominent bottlenecks is associated with weights of the events generated. If the weights are non-uniform, we will get greater imprecision. Since the requirement is typically put on the precision of results, a greater sample size is needed, leading to more CPU time and storage.

Up to the common target of producing the accurate result, these weights can be manipulated by the usage of different matching schemes, with the standard scheme giving an event sample uniform in magnitude but with a fraction being negatively weighted. Recently, two schemes taking the names KrkNLO and MAcNLOPS were proposed. They are closely related to the standard scheme, but have the advantage of fully removing the negative weights with negligible downside. This is a solid step in the right direction of having event samples with uniform and positive weights.

My work is centred on a new take on the implementation of the two schemes, providing fresh insights into their performance and future developments. The results obtained are very promising, generally presenting them as plausible alternatives to the standard scheme. If further tests continue to deliver similar results, a discussion about replacing the standard scheme with one of them might take place in the future. Considering the fact that MAcNLOPS is already capable of producing positive and uniform event samples, its wide implementation may represent a final resolution to this computational bottleneck.

• author: Che, Yuxiao
• supervisor: Rikkert Frederix
• organization: Department of Physics; Particle and nuclear physics
• course: FYSM64 20261
• year: 2026
• type: H2 - Master's Degree (Two Years)
• subject: Physics and Astronomy
• keywords: QCD Phenomenology, Event Generators, NLO Matching
• language: English
• additional info: Part of this work is written up as a paper for journal submission. The preprint is on arXiv with number 2605.22108 and will soon be submitted to EPJ C.
• id: 9229458
• date added to LUP: 2026-06-01 14:12:18
• date last changed: 2026-06-01 14:12:18