Search for an anomalous excess of charged-current quasi-elastic νe interactions with the MicroBooNE experiment using Deep-Learning-based reconstruction

We present a measurement of the νe-interaction rate in the MicroBooNE detector that addresses the observed MiniBooNE anomalous low-energy excess (LEE). The approach taken isolates neutrino interactions consistent with the kinematics of charged-current quasi-elastic (CCQE) events. The topology of such signal events has a final state with 1 electron, 1 proton, and 0 mesons (1e1p). Multiple novel techniques are employed to identify a 1e1p final state, including particle identification that use two methods of deep-learning-based image identification, and event isolation using a boosted decision-tree ensemble trained to recognize two-body scattering kinematics. This analysis selects νe -candidate events in the reconstructed neutrino energy range of 200–1200 MeV, while 29.0 ± 1.9(sys) ± 5.4(stat) are predicted with νμ CCQE interactions used as a constraint. We use a simplified model to translate the MiniBooNE LEE observation into a prediction for a νe signal in MicroBooNE. A ∆χ2 test statistic, based on the combined Neyman–Pearson χ2 formalism, is used to define frequentist confidence intervals for the LEE signal strength. Using this technique, we exclude 0.25 (0.38) times the median MiniBooNE LEE signal strength at the 90% (2σ) confidence level. The expected upper limit under the assumption that there is no LEE signal is 0.75 (0.98) at the 90% (2σ) confidence level.


A link to the paper submitted to Phys. Rev. D is posted on the archive at arXiv:2110.14080.

Additional Material

A link to the accompanying supplementary materials is available here.


Back to the electron-neutrino analysis overview.