John Marshall

My major research interest is the reconstruction of subatomic particle interactions in fine-granularity detectors. Such detectors provide detailed images of complex event topologies, in which individual particles must be resolved. Whilst the human eye/brain can excel at identifying features in the recorded events, it is a significant challenge to develop an automated solution.

I am author of the Pandora software framework for developing and running pattern-recognition algorithms. Pandora promotes a multi-algorithm approach to pattern recognition, in which many algorithms gradually build up a picture of structures in images. Pandora algorithms frequently use machine-learning approaches to drive pattern-recognition decisions.

Pandora algorithms provide an automated reconstruction of neutrino interactions in the Liquid- Argon Time-Projection Chambers (LArTPCs) being operated and developed for the short- and long-baseline neutrino physics programmes. The algorithms are used by MicroBooNE, SBND and DUNE.

Pandora algorithms also provide the particle flow calorimetry used by the International Linear Collider and Compact Linear Collider experiments. The algorithms provide the baseline for past studies of detector optimisation and physics sensitivity for a future e+e- linear collider.

Some selected publications:

• The Pandora multi-algorithm approach to automated pattern recognition of cosmic-ray muon and neutrino events in the MicroBooNE detectorLink opens in a new window
• The Pandora software development kit for pattern recognitionLink opens in a new window
• Performance of particle flow calorimetry at CLICLink opens in a new window