Event generation with Sherpa 2.2

Abstract

SHERPA is a general-purpose Monte Carlo event generator for the simulation of particle collisions in high-energy collider experiments. We summarise essential features and improvements of the SHERPA 2.2 release series, which is heavily used for event generation in the analysis and interpretation of LHC Run 1 and Run 2 data. We highlight a decade of developments towards ever higher precision in the simulation of particle-collision events. Figure 1: Overview of the SHERPA 2.2 event generator framework. Interfaces/OutputsAMEGIC [6] and COMIX [7,8]. They are used for the simulation of parton-level events within the Standard Model and beyond, and for the decay of heavy resonances such as W , Z, or Higgs bosons or top quarks. Both include automated methods for efficient phase-space integration and algorithms for the subtraction of infrared divergences in calculations at next-to-leading order (NLO) in QCD [9, 10, 11] and the electroweak theory [12]. For the evaluation of virtual corrections at NLO accuracy SHERPA relies on interfaces to dedicated one-loop providers, e.g. BLACKHAT [13], OPENLOOPS [14] and RECOLA [15,16]. The default parton-showering algorithm of the SHERPA 2.2 series is the CSSHOWER [17], based on Catani-Seymour dipole factorisation [9,10,18]. As of version 2.2.0 SHERPA also features an independent second shower implementation, DIRE [19,20,21]. For the matching of NLO QCD matrix elements with parton showers SHERPA implements the MC@NLO method [22,23]. For NNLO QCD calculations the UN 2 LOPS method [24, 25] is used. The merging of multi-jet production processes at leading order [26,27,28] and next-to-leading order [29,30] is based on truncated parton showers. Multiple parton interactions are implemented via the Sjöstrand-van-Zijl model [31]. The hadronisation of partons into hadrons is modelled by a cluster fragmentation model [32]. Alternatively, in particular for uncertainty estimations, an interface to the Lund fragmentation model [33] of PYTHIA [34] is available. SHERPA provides a large library for the simulation of τ -lepton and hadron decays, including many form-factor models. Furthermore, a module for the simulation of QED final-state radiation in particle decays [35], which is accurate to first order in α for many channels is built-in. To account for spin correlations in production and subsequent decay processes the algorithm described in [36] is implemented. Events generated with SHERPA can be cast into various output formats for further processing, with the HEPMC [37] format being the most commonly used. In the specific case of parton-level events, at the leading and next-to-leading order in QCD, additional output formats are supported. They include Les Houches Event Files [38], NTUPLE files for NLO QCD events [39] and cross-section interpolation grids produced via MCGRID [40,41] in the APPLGRID [42] and FASTNLO [43,44] formats. To analyse events on-the-fly a runtime interface to the RIVET package [45] can be used conveniently.

Event generation with normalizing flows

“…This section presents a numerical study of the novel phase-space integrator. All computations are performed using the event generation framework Sherpa [40][41][42], with matrix elements provided by Comix [30]. This use case differs slightly from Ref.…”

“…In this section we compare the performance of our integrator based on the normalizing flow technique to the best alternative method available in the public event generation program Sherpa [40][41][42]. The basic computational setup is analogous to Ref.…”

Resummed predictions for jet-resolution scales in multijet production in e+e− annihilation

“…We validate all non-trivial colour correlators in Eq. (27) by comparing the eikonal approximation to exact (n + 1) tree-level matrix elements (see also Ref. [28]), in form of the ratio…”

“…To achieve NLL resummation for jet resolutions, we rely on an independent implementation of the CAESAR formalism [24,25] in the SHERPA [26,27] event-generator framework, presented in [28]. A brief introduction to the CAESAR approach and details on the implementation of automated NLL resummation in the SHERPA framework are discussed in Sec.…”

Exploring phase space with Neural Importance Sampling

“…This includes total and differential production rates in the framework of the Standard Model or hypothetical New Physics scenarios. To allow for a direct comparison with experimental data, multi-purpose event generators such as PYTHIA [1], HERWIG [2] or SHERPA [3,4] proved to be vital tools. Starting from the evaluation of partonic hard-scattering cross sections they accomplish a fully differential and exclusive simulation of individual scattering events by invoking parton-shower simulations, particle decays, models for the parton-to-hadron transition and, in case of composite colliding entities such as protons, multiple interactions per collision.…”