hep-viz v0.1.5

hep-viz v0.1.5 hep-viz is a specialised visualisation tool developed to address the specific challenges of inspecting High Energy Physics (HEP) data in the era of machine learning. It connects detector energy signatures from truth particles and allows for the inspection of reconstruction algorithm outputs. Key Features Zero-Setup Environment: Inspect collision events and verify truth-level clustering immediately. Modern Tech Stack: Decoupled client-server model using a FastAPI backend and Three.js frontend. Lazy Loading: Efficiently explores large-scale datasets (e.g., ColliderML) via PyArrow predicate pushdown. ML Integration: visualises decision boundaries for Graph Neural Networks (GNNs) and attention mechanisms for Transformers. Physics Clustering: Integrated FastJet (Anti-kt, R=0.4) for dynamic comparison of Truth vs. Reco jets. Installation pip install hep-viz