MadGraph

MadGraph
Name	MadGraph
Title	MadGraph
Developer	CERN; SLAC National Accelerator Laboratory; LAPTH; MIT
Released	2002
Latest release	(see Development and Versions)
Programming language	Python; Fortran; C++
Operating system	Linux; macOS; Windows (via WSL)
License	GNU General Public License; other open-source components

MadGraph is a widely used particle physics event generator and matrix-element calculator for simulating high-energy collisions and decays. It provides automated generation of tree-level and loop-amplitude processes, event samples, and parton-level predictions that interface with showering and detector-simulation tools. MadGraph underpins phenomenology studies across collider experiments, theoretical investigations, and pedagogy in particle physics.

Overview

MadGraph automates the construction of scattering amplitudes and cross sections for processes relevant to experiments such as Large Hadron Collider, Tevatron, LEP, and planned facilities like International Linear Collider and Future Circular Collider. It interoperates with parton-shower programs and detector frameworks including PYTHIA, HERWIG, DELPHES, and GEANT4 to produce realistic simulated data for collaborations like ATLAS, CMS, LHCb, and ALICE. MadGraph supports model implementations from repositories and frameworks such as FeynRules, UFO (format), and HEPMDB, enabling studies inspired by theories associated with Standard Model, Supersymmetry, Extra Dimensions, and Effective Field Theory. The software is used by research groups at institutions including CERN, Fermilab, DESY, Brookhaven National Laboratory, and University of Oxford.

Development and Versions

The MadGraph project originated in the early 2000s with contributions from teams at CERN, Caltech, and University of California, Berkeley, evolving through major releases like MadGraph 4 and MadGraph5. Key collaborations and developers include researchers associated with SLAC National Accelerator Laboratory, LAPTH, IPPP Durham, and groups at MIT and NYU. Successive versions introduced features such as the Universal FeynRules Output via FeynRules integration, next-to-leading order (NLO) automation via the MadLoop and aMC@NLO frameworks, and substantial Python-based reengineering. Community governance, workshops, and summer schools at institutions like CERN and DESY have shaped feature priorities and documentation practices. Recent development focus includes performance improvements, GPU acceleration initiatives tied to projects at NERSC and Argonne National Laboratory, and integration with analysis ecosystems developed at Institute for Quantum Computing and national laboratories.

Theoretical Framework and Capabilities

MadGraph computes amplitudes using Feynman-diagrammatic techniques for perturbative quantum field theories relevant to models implemented in the UFO format. It automates tree-level matrix elements and, via MadLoop, one-loop amplitudes using algorithms connected to tools such as CutTools, OpenLoops, and integrand-reduction methods championed by researchers at Institut de Physique Théorique (IPhT). For NLO calculations MadGraph interfaces with subtraction schemes and matching procedures like MC@NLO and POWHEG, enabling combination with parton showers from PYTHIA and HERWIG. MadGraph supports multi-leg processes, heavy-flavor treatments developed in collaborations involving CERN and Fermilab, and EFT operators often studied in contexts associated with High Energy Physics phenomenology groups. It implements color and spin-correlated matrix elements and uses phase-space integration strategies refined by teams at SLAC and IPPP Durham.

Usage and Workflow

Typical workflows begin by selecting or importing a model via FeynRules-generated UFO files or community model repositories like HEPMDB and then specifying processes using MadGraph's process syntax. Users generate matrix elements, produce parton-level event files in formats compatible with Les Houches Accord conventions, and pass those events to showering programs such as PYTHIA or HERWIG for hadronization. Subsequent detector-level simulation is done with packages like DELPHES or full simulation stacks in collaborations including ATLAS and CMS. MadGraph offers scripting and batch execution suitable for high-throughput computing centers such as CERN IT, NERSC, and Fermilab Grid sites. Documentation, tutorials, and example cards are distributed via community workshops at places like CERN and DESY, and training occurs in summer schools organized by SUSY group networks and phenomenology consortia.

Validation and Benchmarking

Validation of MadGraph involves comparisons against analytic results, alternative generators including COMIX, SHERPA, and fixed-order tools developed at IN2P3 and KIT, and experimental data from ATLAS, CMS, CDF, and D0. Benchmark studies often target key processes such as diboson and top-quark production, Higgs boson associated production probed by teams at CERN and Brookhaven National Laboratory, and beyond-Standard-Model signatures explored by groups at IPPP Durham and DESY. Performance benchmarking assesses integration efficiency, scaling on clusters managed by HTCondor or SLURM, and CPU/GPU utilization in collaborations involving Argonne National Laboratory and NERSC. Continuous validation is supported by regression test suites maintained by developer consortia and by comparisons with precision computations from institutions like KITP and Perimeter Institute.

Applications in Research and Education

MadGraph is central to phenomenology papers from authors at CERN, Fermilab, Brookhaven National Laboratory, University of Cambridge, and Princeton University that explore Higgs properties, new-physics searches, and precision electroweak studies. It is used in analysis workflows within experimental collaborations such as ATLAS and CMS for signal modeling, background estimation, and systematic studies. In education, MadGraph features in graduate courses and summer schools at CERN and MIT, enabling students to reproduce calculations from landmark papers by researchers like those affiliated with SLAC and Caltech. Outreach and training materials leverage examples tied to landmark results from Large Hadron Collider runs and historical measurements from LEP and Tevatron.

Category:Computational physics software