SearcharxivSearch

arXiv subjects

Katharina Lachner

Publications and source records attributed to Katharina Lachner.

2 recordsLinked to original sources

Interaction-model dependence in calorimetric energy reconstruction methods due to non-linear material effects in modern neutrino detectors

Neutrino oscillation experiments rely on high precision neutrino energy reconstruction. A common reconstruction technique in LAr-TPCs and scintillators is via the calorimetric sum of visible particles created in the interaction. However, non-linearities in the detector response, such as Birks quenching for scintillators and recombination effects for TPCs, lead to ambiguities in the reconstruction of visible hadronic energies. Interaction-model-dependent assumptions are required to resolve these ambiguities, which introduces a bias in reconstruction of neutrino energy. This introduces a systematic uncertainty separate from the well-studied bias due to interaction-model dependent modelling of missing energy caused by, for example, the production of final state neutrons. In this work, we evaluate the interaction-model dependence of the bias caused by these material effects across multiple tunes of the GENIE, NEUT, NuWro, and GiBUU neutrino interaction event generators for cases representative of calorimetric energy reconstruction at the T2K (ND280), NO$\nu$A, MINER$\nu$A, $\mu$BooNE, and DUNE experiments. Using pure calorimetric reconstruction, our results show significant differences in the mean neutrino-energy reconstruction bias between models, at the level of $\sim$7-9\,MeV for scintillator detectors and $\sim$11-18\,MeV for argon-based detectors in the relevant energy range. The latter is shown to be reduced (down to $\sim$3.5\,MeV) when using an idealised hybrid energy reconstruction based on tracking and calorimetry. Overall, we conclude that neutrino-energy reconstruction bias due to material effects may imply non-negligible systematic uncertainties for neutrino oscillation and cross-section measurements, and discuss alternative analysis strategies to mitigate the issue.

hep-ex

Deep-learning-based low-energy trigger algorithms for the Hyper-Kamiokande experiment

Modern machine learning techniques have become increasingly important in particle physics because of their powerful pattern-recognition capabilities, including in real-time data acquisition where stringent runtime constraints apply. This paper details the performance of deep-learning-based trigger algorithms for a large water Cherenkov detector such as Hyper-Kamiokande aimed at low-energy neutrino events (below 7 MeV). The performance of custom neural-network supervised classifiers is shown alongside two anomaly-detection approaches trained solely on detector noise: a pure autoencoder and an energy-based model based on Manifold Projection--Diffusion Recovery (MPDR). The supervised model shows signal identification efficiencies of 76.7% for single electrons of 3 MeV kinetic energy, significantly exceeding signal efficiencies obtained from a traditional hit-count-based trigger of 26.4%, as does the MPDR approach with 31.8%. Runtime evaluations on GPU yield per-window inference latencies well below the millisecond scale, indicating that real-time operation is feasible.

physics.ins-det