Penn, Gabriel
(2023)
Selection of muon-antineutrino charged-current single-pion events with boosted decision tree particle identification in the ND280 detector.
PhD thesis, University of Liverpool.
|
Text
201283313_Feb2023.pdf - Author Accepted Manuscript Download (13MB) | Preview |
Abstract
Tokai to Kamioka (T2K) is a long-baseline neutrino oscillation experiment situated in Japan, studying electron-neutrino appearance in a muon-neutrino beam and thus probing multiple parameters of the Pontecorvo-Maki-Nakagawa-Sakata matrix. A beam of muon-(anti)neutrinos is produced at the J-PARC accelerator complex and sampled by near and far detectors to compare the beam content before and after oscillation. The off-axis near detector, ND280, serves both to characterise signals and backgrounds observed at the far detector, and to measure a variety of neutrino interaction cross-sections on carbon and oxygen nuclear targets. These measurements contribute to our understanding of both neutrino oscillations and the interactions between neutrinos and atomic nuclei. Selecting specific interaction types from the ND280 data requires accurate identification of charged particle tracks, but the conventional cut-based methods perform poorly in many areas. To make full use of the wide array of information available from all of the ND280 subdetectors and thus develop a high-performing particle identification (PID) algorithm, a multivariate analysis approach is necessary, but such a tool has yet to be developed and deployed at T2K. This thesis presents the development of a general-purpose PID tool for charged particle tracks in ND280 using boosted decision trees, and demonstrates its superior classification power compared to the existing PID algorithms currently in use. The tool has been designed to be broadly applicable and outperforms purpose-built conventional PID at identifying each particle type in almost all kinematic regions tested. Furthermore, it has been integrated into a selection algorithm for muon-antineutrino charged-current single-pion events, and found to substantially improve the selection performance. These results demonstrate the power of multivariate PID methods for charged particle tracks in ND280, and strongly motivate their use in future T2K analyses.
| Item Type: | Thesis (PhD) |
|---|---|
| Divisions: | Faculty of Science and Engineering > School of Physical Sciences |
| Depositing User: | Symplectic Admin |
| Date Deposited: | 16 Feb 2023 10:52 |
| Last Modified: | 16 Feb 2023 10:52 |
| DOI: | 10.17638/03168409 |
| Supervisors: |
|
| URI: | https://livrepository.liverpool.ac.uk/id/eprint/3168409 |
Dimensions
Dimensions
