MadAgents

TL;DR

MadAgents introduces an agent-based framework that tightly couples with MadGraph to enable rapid, communicative, and autonomous LHC simulations. It combines automated installation, tailored training for novices, advanced support for experienced users, and autonomous campaign generation driven by PDFs or papers. The work demonstrates end-to-end capabilities—from building MadGraph ecosystems (including ROOT, Pythia, and Delphes) to designing six tutorials, performing precision top-quark studies with NLO+PS and multi-jet merging, and reproducing HEPTAPOD-like leptoquark analyses. Collectively, these contributions offer a path toward AI-assisted, reproducible, and scalable LHC research workflows that can accelerate both theory and experiment analyses while preserving human guidance and accountability.

Abstract

We uncover an effective and communicative set of agents working with MadGraph. Agentic installation, learning-by-doing training, and user support provide easy access to state-of-the-art simulations and accelerate LHC research. We show in detail how MadAgents interact with inexperienced and advanced users, support a range of simulation tasks, and analyze results. In a second step, we illustrate how MadAgents automatize event generation and run an autonomous simulation campaign, starting from a pdf file of a paper.

Figures (3)

• Figure 1: MadAgents structure.
• Figure 2: Comparison of the $m_{t\bar{t}}$ and $p_{T}^{t\bar{t}}$ distributions for precision simulations of the production process, figure generated autonomously by MadAgents.
• Figure 3: Reconstructed minimum leptoquark mass for $m_{S1} \in [1, 1.5, 2]$ TeV, figure generated autonomously by MadAgents.