Introduction

LARDON (Liquid Argon Reconstruction Done in pythON) is a command-line reconstruction framework for liquid argon time projection chamber (LArTPC) detectors. It reads raw HDF5 readout files produced by DUNE and ProtoDUNE detectors and produces fully reconstructed physics objects — hits, 2D and 3D tracks, and photon detection signals — stored in HDF5 output files. LARDON is used in the DUNE and ProtoDUNE experimental programs at CERN and Fermilab. It supports both the charge readout (TPC) and light readout (PDS) systems, and can process data from a range of detector geometries with a common interface.

​

What LARDON does

The reconstruction pipeline covers the full chain from raw waveform data to physics objects:

• Signal processing — pedestal estimation, noise filtering, and coherent noise removal on TPC wire waveforms
• Hit finding — identifies signal regions of interest on each wire view
• 2D track reconstruction — Hough-transform-based track seeding with Kalman filter refinement, per detector module
• 3D track reconstruction — combines 2D tracks from multiple wire views into 3D space points
• PDS reconstruction — processes photon detection system waveforms, finds peaks, and clusters light signals
• TPC–PDS matching — correlates reconstructed charge tracks with light clusters for full event reconstruction

​

Supported detectors

LARDON supports the following detector configurations, specified with the -det flag:

​

Key capabilities

​

Dependency management with Pixi

LARDON uses Pixi to manage its Python dependencies and environments. Pixi resolves and installs all required packages — including numpy, scipy, numba, h5py, scikit-learn, and xrootd — in isolated, reproducible environments. You do not need an existing Python installation or conda environment to use LARDON. Pixi handles everything.