DELPHES

DELPHES
Name	DELPHES
Developer	CERN collaborators, CNRS groups, Université Paris-Saclay
Initial release	2012
Latest release	2020s
Programming language	C++
Operating system	Linux, macOS
License	LGPL-compatible (open-source)

DELPHES is a modular, fast-simulation framework for high-energy physics collider experiments. It provides a parametrized, detector-level simulation and reconstruction chain that approximates full simulation outputs from experiments such as ATLAS and CMS for use by phenomenologists and analysis-level studies by groups like LHCP, IHEP, and university collaborations. DELPHES bridges event generators and analysis tools, enabling rapid prototyping of searches and measurements similar to workflows at CERN experiments and theory groups including Fermilab, DESY, and SLAC.

Overview

DELPHES was created to translate generator-level events from tools such as PYTHIA, HERWIG, and MADGRAPH into reconstructed objects resembling those produced by detectors like ATLAS and CMS. It accepts formats produced by generators and matrix-element programs like MG5_aMC@NLO and Sherpa and outputs ROOT trees compatible with analysis frameworks used by collaborations such as ROOT-based workflows, Rivet, and experiment-specific toolchains. The framework is widely used by phenomenologists at institutions like Institut de Physique Théorique, University of Oxford, and University of California, Berkeley to evaluate sensitivity studies for searches inspired by results from LHC Run 1, LHC Run 2, and future projects like the FCC and ILC.

Architecture and Simulation Chain

DELPHES implements a chain that starts with generator-level particles and proceeds through fast detector effects: tracking efficiency, calorimeter response, particle-flow algorithms, and object reconstruction. The architecture is modular: detector cards and modules define behavior used in pipelines developed by contributors from CEA Saclay, INFN, and KEK. Typical chains include modules that emulate magnetic-field bending like in CMS, calorimeter granularity resembling ATLAS, and b-tagging parametrizations used in analyses by CDF and DØ legacy studies. The modularity allows integration with pile-up modeling techniques adopted by Pileup Per Particle Identification (PUPPI), reconstruction steps from FastJet for jet clustering, and isolation algorithms tested in collaborations such as Belle II.

Detector Modules and Parameterization

DELPHES ships with detector configuration cards parameterized to mimic experiments and proposals like ATLAS, CMS, ILC Detector Concept (ILD), and generic detectors for future colliders like CEPC. Modules include tracker resolution and efficiency maps inspired by measurements from ATLAS Inner Detector, calorimeter energy smearing tuned to CMS Electromagnetic Calorimeter and Hadronic Calorimeter performance, and muon system models reflecting designs from Muon Spectrometer subsystems. b-tagging and c-tagging efficiencies/ mistag rates can be set to match working points used in analyses from ATLAS Collaboration and CMS Collaboration. Jet reconstruction uses algorithms such as anti-kt from FastJet and supports grooming techniques referenced by studies at Snowmass workshops and by groups in IHEP Beijing.

Input/Output and Interfaces

DELPHES reads common event formats: HepMC produced by PYTHIA and HERWIG, Les Houches Event (LHE) formats from MadGraph5_aMC@NLO and CalcHEP, and ROOT TTree files produced by custom generator chains in groups like MPI Munich. Outputs are ROOT trees with branches for objects (electrons, muons, photons, jets, MET) that integrate with analysis stacks used by Rivet, MadAnalysis 5, and collaboration-level notebooks common at CERN Summer Student Programme. Interface utilities allow conversion between formats, and users often pipeline DELPHES with sample management systems used in projects at Fermilab and Brookhaven National Laboratory.

Validation and Performance

Validation of DELPHES configurations is performed by comparing distributions to full simulation and public results from ATLAS and CMS notes, conference presentations at ICHEP, and phenomenology papers by groups at SLAC and LPTHE. Performance benchmarks focus on CPU time and memory usage versus GEANT4-based full simulation used by ATLAS Simulation Group and CMS Simulation Group; DELPHES provides orders-of-magnitude speed-ups enabling large-sample scans for studies by teams at Perimeter Institute and Max Planck Institute for Physics. Systematic uncertainties can be propagated by varying detector card parameters to reproduce shifts reported in public analyses by ATLAS Collaboration and CMS Collaboration.

Applications and Usage Examples

DELPHES is used to evaluate search strategies for signatures studied by collaborations such as exotic resonances similar to analyses presented at Moriond, precision Higgs coupling fits inspired by Higgs Combination Group results, and dark-matter simplified model studies common in workshops at KITP. It supports reinterpretation efforts where groups at IPPP Durham, CERN Theory Department, and University of Chicago rederive limits using public LHC data and simplified models. Educational uses include tutorials at summer schools run by CERN and computational exercises in courses at Princeton University and EPFL.

Development and Licensing

DELPHES development is community-driven with contributions from researchers at CERN, CNRS, INFN, and international collaborators from IHEP, KEK, and universities worldwide. The codebase is written in C++ and distributed under an open-source license compatible with academic reuse; contributors follow practices similar to those in projects like ROOT and Geant4 collaboration. Releases often coincide with major conference cycles and are accompanied by detector cards reflecting evolving designs from ATLAS Upgrade and CMS Phase-2 Upgrade studies.

Category:Particle physics software