Sudakov evolution without unitarity

J. Altmann, H. T. Li, L. Scyboz, P. Skands · 2025

Read Sudakov evolution without unitarity by J. Altmann, H. T. Li, L. Scyboz, P. Skands (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-00111.txt.

Introduction

Altmann and co-authors present a sampling method for particle physics calculations in a 2025 paper. The catalog record shortens the author names to initials, while the text byline spells them out in full. Their tool handles sharp, spiky functions defined over spaces of many particles.

Three ideas drive the method. A Sudakov step sweeps the space in order from hard to soft events. Spaces for more particles nest inside simpler ones in sequence. And near the edges, complex amplitudes split into simpler factors. The team built the method in code, named Sunshine, and ran it on decays of the Z boson. Tests checked a sector shower program's accuracy at a set order of precision. The work may sharpen event generators used across collider physics.

Catalog record

CREATORJ. Altmann, H. T. Li, L. Scyboz, P. Skands
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE67 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-00111

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Sudakov evolution without unitarity J. Altmann, H. T. Li, L. Scyboz, P. Skands Abstract We present a method for sampling singular functions defined on (nested) multi-particle phase spaces, based on a generalisation of parton-shower phase-space generation techniques. At the heart of the method are three key ingredients: 1) the Sudakov sampling by which shower-style calculations sweep across phase space in an ordered manner, from hard to soft; 2) the sequential nesting of multiparticle phase spaces; and 3) the factorisations obeyed by singular multiparton amplitudes on the edges of these phase spaces. We demonstrate a C++ implementation of the proposed algorithm, dubbed Sunshine , 1 1 1 Loosely derived from Sudakov Nesting of Hard Integrals. for hadronic Z decays, and use it to test the tree-level accuracy of the Vincia sector shower through 𝒪 ⁡ ( α s 2 ) {\cal O}(\alpha_{s}^{2}) . 1 Introduction This work originated in the study of tree-level expansions of parton showers, which are among the main diagnostics tools to assess the accuracy of parton showers at higher orders [ 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 ] . So far, fixed-order expansions of showers have typically been examined not by directly running the actual shower algorithm, but instead by constructing explicit products of (sums over) their radiation kernels and phase-space mappings. This should of course be equivalent but does require writing additional code, which can be complex, and also introduces the risk that the actual shower differs in some unanticipated way from the (separately constructed) direct expansion. In this paper, we propose a simple and general technique for obtaining fixed-order expansions straight from shower generators. The crucial observation is that the Sudakov factors (or, really, no-emission probabilities) produced by parton-shower algorithms express the principle of detailed balance: for each (n+1) -parton configuration created by a shower, a corresponding n -parton event is “annihilated”. This ensures that the shower operator is unitary : the action of the shower does not change the total (inclusive) cross section. Thus, in general, the “weight” of an (n+m) -parton configuration produced by a shower that starts from an n -parton configuration is, schematically: w_{n+m}~=~w_{n}\prod_{i=n}^{n+m-1}{\cal P}_{i\mapsto i+1}\,\Delta(t_{i},t_{i+1})\,, (1) where w_{n}\propto|M_{n}|^{2} is the weight of the starting ( n -parton) configuration, {\cal P} represents (fixed-order) branching kernels of the shower at hand (e.g., DGLAP splitting kernels [ 9 , 10 , 11 ] or dipole/antenna functions), and the Sudakov factors \Delta\sim\exp(-\int{\cal P}) represent no-emission probabilities between the sequential (nested) branching scales, t_{i+1}<t_{i} , with the shower evolution variable t a measure of quantum mechanical resolution scale analogous to a jet-resolution scale. (If a shower algorithm has multiple ways to produce the given (n+m) -parton configuration, one must further sum over all such paths.) In the context of studying tree-level expansions of parton-shower algorithms, one would ideally like to “remove” the Sudakov factors, and study just the tree-level expansion, w_{n+m}^{(0)}~=~w_{n}^{(0)}\prod_{i=n}^{n+m-1}{\cal P}^{(0)}_{i\mapsto i+1}\,, (2) where the superscript (0) indicates expansion to the first non-trivial order. This is what has traditionally been done by direct constructions of the product \prod{\cal P}_{i} in eq. ( 2 ). In this work, we point out that this can also be accomplished by running the shower algorithm in a dedicated mode in which the principle of detailed balance is not imposed. We present an explicit implementation of this proposal in the framework of Pythia 8.3 [ 12 ] , with detailed validations, and use it to study tree-level expansions of the Vincia sector shower [ 13 ] , for hadronic Z decays with and without tree-level matrix-element corrections up to second order. We note that most of our discussion here will focus on QCD but the general principle also applies to QED showers, or indeed to any Sudakov-based formalism that implements detailed balance. In sec. 2 , we summarise the standard method of Sudakov-based sampling of nested phase spaces in shower-style algorithms, and establish some notation that we shall use throughout. In sec. 3 , we discuss the problem of obtaining fixed-order expansions from such algorithms in detail, present the Sunshine algorithm, and give a mathematical proof that it produces the desired tree-level-expanded weights. We also discuss subtleties associated with ensuring full phase-space coverage, shower algorithms with general n\mapsto n+m branchings (for m\geq 2 ), the infrared cutoff in shower algorithms, and finally present a set of validation checks which demonstrate that our implementation of the proposed algorithm is self-consistent. Sec. 4 illustrates the application of the algorithm to tree-level tests of the Vincia sector shower at 1 st and 2 nd order, comparing it against both analytic results [ 14 , 15 ] and to results obtained with Event2 [ 16 , 17 ] . Finally, in sec. 5 , we conclude and give an outlook. The algorithm as we implemented it in Pythia is given in appendix A . 2 Standard Sudakov sampling of nested phase spaces Mathematically, parton-shower algorithms are formulated as Markov chains which operate on partonic states. These states consist of a set of partons specified by their momenta and information about their flavours and other internal quantum numbers. The latter typically includes colour flow in a leading-colour (LC) limit. 2 2 2 Conventionally represented by so-called Les Houches colour tags [ 18 , 19 ] . Optionally, helicity information may also be present and may then be used to generate helicity-dependent showers as in refs. [ 20 , 21 ] . (More sophisticated treatments of spin-correlation and polarisation effects are beyond the scope of our discussion here.) For each input state, one seeks to generate radiation patterns that have eikonal factors as their soft limits and DGLAP kernels as their collinear limits (which we refer to collectively as infrared limits), \mathrm{ant}_{n\to n+m}\sim\frac{|M_{n+m}|^{2}}{|M_{n}|^{2}}\,, (3) where \mathrm{ant}_{n\to n+m} expresses an approximation which, at the very least, should reproduce the leading pole structure(s) of the matrix elements. For definiteness, here we focus on the sector-antenna approach [ 22 , 23 , 24 , 25 , 26 , 13 ] . Starting from a Born state which we denote \Phi_{0} , the veto algorithm (see, e.g. refs. [ 27 , 28 , 12 ] ) is used to generate (unweighted) branchings distributed according to \frac{\mathrm{d}{\cal P}}{\mathrm{d}\Phi_{1}}=\mathrm{ant}_{0\mapsto 1}\,\Delta_{0}(t_{0},t_{1})\,|M_{0}|^{2}\,, (4) where M_{0}(\Phi_{0}) is the Born-level matrix element, t represents the evolution variable of the shower with t_{0}(\Phi_{0}) the starting scale and t_{1}(\Phi_{1})<t_{0} the evolution scale associated with the 0\mapsto 1 branching, and \Delta_{0}(t_{0},t_{1}) is the Sudakov factor, that expresses the probability that the shower does not generate any branchings in the interval t\in[t_{0},t_{1}] . In a shower Markov chain this sampling is iterated, to build up higher multiplicities with a compound probability distribution proportional to \frac{\mathrm{d}{\cal P}}{\mathrm{d}\Phi_{n}}=|M_{0}|^{2}\,\prod_{i=0}^{n-1}\mathrm{ant}_{i\to i+1}\,\Delta_{i}(t_{i},t_{i+1})\,. (5) The product of Sudakov factors explicitly represents the resummation carried out by the shower between each of the successive scales t_{i} . In this work, we are interested in modifying this sampling to eliminate the Sudakov factors, thus making the resulting sampling proportional to the tree-level expansion of the shower instead. We note that a variant of the above that will be especially interesting to us is one in which iterated matrix-element corrections (MECs) [ 3 ] are applied during the shower

Read the complete text