Efficient amplitude computing : Optimizing matrix elements in MadGraph5_aMC@NLO

Details

Supervisors

Maltoni, Fabio ; Mattelaer, Olivier

Faculty

Faculté des sciences

Degree label

Master [120] en sciences physiques, à finalité approfondie

Abstract

enThe computation of squared matrix elements lies at the heart of event simulation at the LHC. As such, the MadGraph5_aMC@NLO framework plays an essential role in bridging local quantum field theories and experimentally measurable observables through Monte-Carlo integration. This master’s thesis explores and proposes concrete optimizations to MadGraph5_aMC@NLO’s matrix element evaluation routines, focusing particularly on tree-level processes. Motivated by physical symmetries such as charge conjugation and flavour indistinguishability at high energies, we propose two kernel-reducing strategies: a flavour optimization, applicable to processes differing only by lepton or light quark flavour, and a C-symmetry inspired optimization, applicable to particle-antiparticle crossings in QED and QCD. We prove a result allowing to recast coloured ordered amplitudes when inversing a fermionic flow. Both optimizations are implemented directly in MadGraph’s standalone Fortran output.