The ICARUS collaboration, as part of the Short-Baseline Neutrino (SBN) Program at Fermilab, has completed its first oscillation analysis with the Booster Neutrino Beam (BNB): a single-detector search for muon-neutrino disappearance in the context of a 3+1 sterile neutrino model, using data collected in 2022–2023. Two independent reconstruction frameworks were used to select charged-current events with a single muon, at least one proton, and no pions in the final state, and no statistically significant disappearance is observed at the 600 m ICARUS baseline. One of these frameworks is SPINE, an end-to-end machine-learning-based reconstruction that goes from raw liquid argon time projection chamber (LArTPC) images to complete particle-level descriptions of each neutrino interaction, enabling a high-efficiency, high-purity selection. In this talk, I will describe how SPINE works and how it was integrated into the analysis, including its performance relative to traditional approaches. Upcoming joint SBND–ICARUS analyses will build on this result, constraining flux and interaction model uncertainties with a near detector and further leveraging ML-based reconstruction. I will close by discussing how these reconstruction advances are positioned to extend the physics reach of LArTPC experiments, including SBN and DUNE.
Justin Mueller, Research Associate, Fermilab
Host: Susan Dittmer
Joan West
(847) 491-3645
Email