Parton showers are immensely flexible tools, which simulate the strong-interaction physics that occurs between the TeV energy scales that the LHC was designed to probe, and the GeV mass scale of the protons and other hadrons that the LHC collides and detects. They are used in almost every LHC experimental analysis.
The goal of the project was to transform the way in which parton showers are conceived, by introducing innovative methods that establish the relation with another field of research called resummation. The figure of merit is the so-called "logarithmic" accuracy, with existing parton showers being leading logarithmic (LL), while we hoped to achieve next-to-leading and next-to-next-to-leading logarithmic accuracies (NLL and NNLL), corresponding to higher precision.
The first major milestone (Phys. Rev. Lett. 125, 052002 (2020)) involved tasks 1 and 2 of the project. It brought the demonstration (a) that we could formulate a sound definition of what NLL accuracy should achieve; (b) that we could rigorously determine the logarithmic accuracy of a range of existing parton showers, and (c) that we could design parton showers that simultaneously provided key complementary aspects of NLL accuracy. This was done for electron-positron (e+e-) collisions. Each of these represented the first time such achievements were made in the field.
To elaborate on this work at NLL, we then took two directions.
The first was to achieve showers that had complete NLL accuracy for e+e- collisions. This involved the important question of so-called subleading colour corrections (JHEP03(2021)041), which need to be correct, at the very least, at LL accuracy (and generally were not, in particular for the wide class of dipole and antenna showers). It also involved fully solving a problem known as spin correlations, where we designed new observables for testing spin correlations and addressed a long-standing issue in a well-established algorithm, connected with spin correlations for low energy emissions.
The second direction at NLL was the extension to hadron-hadron collisions (JHEP11(2022)019), where we provided complete solutions notably for colour singlet production. Two of the early-career researchers involved in the project then independently extended this to deep inelastic scattering and vector-boson fusion. This completed task 2 of the project.
The next major component was task 3, the development of NNLL-accurate parton showers. This was the highest-risk, highest-gain element of the project. At the time of the proposal, NNLL resummations didn't even exist for several of the major classes of observables that are relevant for testing. Where resummations did exist, it turned out they were not always completely correct, as we discovered when trying to reproduce the results using independent methods. For yet another class of observables (multiplicities), we carried out the first resummations ourselves. A crucial element that we required, so-called next-to-leading-order matching, existed; but it became clear that it didn't exist in a form that could be straightforwardly combined with our logarithmically accurate parton showers.
We worked through these problems and then started addressing NNLL accuracy for final-state showers through successive classes of observables. In 2023 (Phys.Rev.Lett. 131 (2023) 16, 161906) we demonstrated that a shower could achieve this accuracy (strictly, α^2 L^(2n-2) or α^n L^(n-1)) for two out of four major classes of observables: multiplicities and energy flow in a restricted angular region. About a year later,
accepted for publication in Phys.Rev.Lett.(opens in new window) showed how to obtain a parton shower with this accuracy for what is arguably the largest class of observables, event shapes. In our view, these two references represent major milestones.
In addition to core parton shower development, we have carried out a range of phenomenological work, including a method for rigorously determining whether a jet of partons has quark or gluon flavour, and new observables sensitive to the phenomenon of multi-parton interactions.
We have also documented and publicly released the code developed as part of the project and it is available at
https://gitlab.com/panscales/panscales-0.X(opens in new window).
