AnalysisG Documentation
AnalysisG is a Graph Neural Network Analysis Framework for High Energy Physics. It provides a complete pipeline for translating ROOT n-tuples into graph-structured data, training and evaluating Graph Neural Networks, and running cut-based selections — all from a Python interface backed by high-performance C++ and CUDA.
Overview
User Guide
Python Interface
- Analysis (Python)
- ParticleTemplate (Python)
- EventTemplate (Python)
- GraphTemplate (Python)
- SelectionTemplate (Python)
- MetricTemplate (Python)
- ModelTemplate (Python)
- IO (Python)
- Meta / AMI (Python)
- Tools (Python)
- Notification (Python)
- Plotting (Python)
- ROC (Python)
- OptimizerConfig (Python)
- Structs (Python)
Built-In Events
Built-In Graphs
Built-In Models
Built-In Metrics
Built-In Selections
- Example MET Selection (
selections.example.met) - Analysis Regions Selection (
selections.analysis.regions) - MC16 Children Kinematics (
selections.mc16.childrenkinematics) - MC16 Decay Modes (
selections.mc16.decaymodes) - MC16 Missing ET / Neutrino Reconstruction (
selections.mc16.met) - MC16 Parton Energy Fractions (
selections.mc16.parton) - MC16 Top + Reconstructed-Jet Matching (
selections.mc16.topjets) - MC16 Top Kinematic Distributions (
selections.mc16.topkinematics) - MC16 Top-Quark Matching (
selections.mc16.topmatching) - MC16 Top + Truth-Jet Matching (
selections.mc16.toptruthjets) - MC16 Z′ Selection (
selections.mc16.zprime) - MC20 Truth Matching (
selections.mc20.matching) - MC20 Top Kinematic Distributions (
selections.mc20.topkinematics) - MC20 Top-Quark Matching (
selections.mc20.topmatching) - MC20 Z′ Selection (
selections.mc20.zprime) - Combinatorial Neutrino Reconstruction (
selections.neutrino.combinatorial) - Neutrino Reconstruction Validation (
selections.neutrino.validation) - Top Efficiency (
selections.performance.topefficiency)
C++ User Templates
C++ Framework
Neutrino Solutions (NuSol)
pyc — CUDA Extensions
About