Abstract

This seminar is a candid review of the Pandora software project, which originated 20 years ago in the context of the International Linear Collider (ILC). Vital lessons were learned about what it really takes to deliver a Particle Flow Algorithm (PFA), via the implementation of the highly-successful package PandoraPFA. These lessons then fed into a two-year programme of software engineering, in which the Pandora Software Development Kit (SDK) was created, and PandoraPFA completely reimplemented. The Pandora SDK supports a multi-algorithm approach to pattern recognition, in which large numbers of algorithms gradually build up a picture of complex events. The new PandoraPFA successfully delivered a programme of physics-sensitivity and detector-optimisation studies for both the ILC and Compact Linear Collider (CLIC). The modular design of the Pandora SDK then opened up an all-new opportunity, in the realm of neutrino physics: the reconstruction of events in Liquid-Argon Time-Projection Chamber (LArTPC) detectors. LArTPC detectors can provide “photograph-quality” images of the charged particles emerging from neutrino interactions, but the images can be extremely complex and difficult to interpret. The Pandora LArTPC reconstruction has seen extensive use at the MicroBooNE experiment, and is now used by SBND and ICARUS too. Today, it is being developed for use at the Deep Underground Neutrino Experiment (DUNE), in order to unlock the physics potential of DUNE ’s imaging detectors. This seminar will review some of the key Pandora milestones, as well as the pattern-recognition algorithms deployed for use at e+e- linear colliders and for LArTPC detectors. Some current development highlights at DUNE will be introduced, alongside a few reflections about this now long-running project.