Long Baseline Neutrino Physics with Theia 70m 18m Mike Wilking - PowerPoint PPT Presentation

Long Baseline Neutrino Physics with Theia 70m 18m Mike Wilking Stony Brook University T HEIA25 MOOD Workshop November 12th, 2019 20m

Challenges in DUNE LBL Physics E rec for selected E true bins fully reconstructed μ Ar 3.5 kt-yr 40 ν π + × -like Events / 40 30 n Missing Energy n energy 20 p measurement π + π 0 10 e ν Selected 0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 E (GeV) Rec. • Subtle mistakes in our modeling of ν -Ar interactions on argon can produce large biases on neutrino oscillation parameters • Missing energy (e.g. neutrons) cause feed-down in E rec vs E true • Mismodeling the shape of this feed-down can cause biases • Understanding the detector uncertainties is also critical (e.g. in hadronic energy measurements)

DUNE-PRISM to the Rescue? • DUNE-PRISM can provide strong constraints on ν -Ar interaction modeling • By measuring a continuously varying set of neutrino energy spectra, the E true -> E rec relationship can be constrained • However, challenges still remain • Differences in detector efficiency and resolution between the ND (ArgonCube + Muon spectrometer) and FD still must be precisely understood • Since the FD is on-axis, DUNE-PRISM cannot sample higher energies to constrain high-E ν feed-down (other strategies, such as changing the horn current are under investigation) • Uncertainties in the neutrino flux prediction must be well constrained 
 (beamline geometry, wrong-sign backgrounds, etc.)

Advantages of Adding a WbLS Detector • A WbLS detector provides several complementary features to the DUNE LBL program • A different (simpler?) target nucleus • Different detector systematic uncertainties (and coupling of detector modeling to cross section modeling) • Improved neutron detection • Good energy resolution • Fast timing • In the era of systematics limitations (i.e. when a 4th detector would come online), providing extra constraints on systematic uncertainties will be a high priority

Theia LBL Sensitivity Studies • The initial LBL studies have focused on a pure water (Cherenkov-only) phase • The additional benefits of WbLS have not yet been included (hadronic energy measurements, neutron tagging, etc.) • Previous studies of a Water Cherenkov detector in the LBNF beam occurred in LBNE • These studies used older (“SK1”) reconstruction tools and analysis techniques • An updated set of studies has been conducted for Theia based on the latest Water Cherenkov analysis tools

Reminder of LBNE Studies 300 • LBNF beam with a water Cherenkov Neutrino&running& detector at Homestake NUEQE=562 250 ANUEQE=4 NUENQE=684 200 • Prior studies (LBNE) made the following Number of Events ANUENQE=7 NC=1390 assumptions: BEAM NUE=289 150 NUMU CC=75 100 1. Only single-ring events are selected (~20% ν e -CCnQE efficiency) 50 • Largest interaction mode at DUNE 0 2 4 6 8 energies of ~2-3 GeV is resonance Reconstructed Energy (GeV) (CC π ) events / GeV) / GeV) 1.4 1.4 2. Neutral current background rejection 1.2 1.2 2 2 cm cm -38 -38 1 1 is based on older reconstruction tools 
 (10 (10 TOTAL 0.8 0.8 � � cross section / E cross section / E (pre-FiTQun and even pre-POLFit) QE 0.6 0.6 DIS 0.4 0.4 • Both of these assumptions have been RES 0.2 0.2 revisited with updated reconstruction � � 0 0 -1 -1 2 2 10 10 10 10 1 1 10 10 tools E E (GeV) (GeV) � � � � � � � �

Advances in Cherenkov Reconstruction Fraction of μ misIDed as e • Since the LBNE WC studies, the FiTQun 5 Miss-ID rate [%] 4.5 FiTQun - event reconstruction package has been µ 4 3.5 implemented in T2K & SK apfit 3 2.5 2 • A likelihood-based fitter that generates 1.5 1 charge and time PDFs for all PMTs for 0.5 0 any proposed set of final state particles 0 100 200 300 400 500 600 700 800 900 1000 Visible energy [MeV] • Substantial improvements are seen in e/ μ separation and NC ( π 0 ) rejection Improved low-E γ Detection Rejection Efficiency 1 • FiTQun is now exclusively used for all T2K 0.8 oscillation analyses, and in the latest SK MC π 0 events atmospheric analysis 0.6 0 to 500 MeV/c 0.4 0 π • FiTQun can naturally incorporate fiTQun 0.2 POLFit scintillation light, but this has not yet 0 0 10 20 30 40 50 60 70 80 90 100 been implemented True Energy of the Less Energetic (MeV) γ

FiTQun π 0 Rejection γ Photon Conversions Vertex • Goal : identify a low-E photon in the 
 π 0 γ presence of a high-E photon • To reject π 0 : Compare best fit likelihoods of π 0 fit & single-e fit (as a function of reconstructed π 0 mass) • Large improvement in finding low energy 2nd ring • ~70% reduction in π 0 background relative to POLFit 
 (but not even POLFit was used in the LBNE studies) 400 1 Rejection Efficiency ) e /L 350 0 350 π ln(L ν Signal 0.8 300 e fiTQun 300 250 0 Background π 250 POLFit 0.6 MC π 0 events 200 200 0 to 500 MeV/c T2K ν e 0.4 150 0 150 π Background fiTQun 100 100 0.2 POLFit 50 50 0 0 0 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 100 120 140 160 180 200 0 50 100 150 200 250 0 2 True Energy of the Less Energetic γ (MeV) Mass (MeV/c) 0 π π Mass (MeV/c )

Multi-Ring Events Atmospheric MC Event Hit Charge Distribution • FiTQun can currently reconstruct up to 6 rings in a staged approach • Each step sequentially adds a “track- like” ( π + ) or “shower-like” (e) ring • The chain terminates when adding a ring does not sufficiently improve the fit • Ring counting & PID are significantly Atmospheric MC Event Reconstructed charge Predicted charge improved 10 Reconstructed “Mean” Charge Sample Fit Sequence Fit Fit e No π Fit Improves Improves Improvement Improves ee e π π e ππ eee ee π π ee π e π ππ e πππ ee π e ee ππ π e π e π e ππ ππ ee ππ e π πππ e ππππ π e ππ e π e πππ Reconstructed charge Predicted charge 10

“1-Ring” ν e -CC π + in T2K e CC1 selection: reconstructed energy fiTQun APfit • The newest T2K ν e sample is ν e -CC π + Reconstructed energy where the π + is below Cherenkov Number of events 20 (7.48 × 10 POT) distribution of Osc. CC ν e threshold Osc. ν CC e CC1 -enriched 1.5 / CC ν ν µ µ Beam ν / ν CC e e e sample. NC 2 MC w/ sin =0.0217 θ 13 • Still a 1-ring event, but with a Michel FiTQun selection 1 yields a much electron smaller numu CC 0.5 background, • Previously, these events were e CC1 selection: reconstructed energy which has a large 0 0 500 1000 systematic contaminated with ν μ -CC background Reconstructed ν energy (MeV) uncertainty. fiTQun APfit • Improved e/ μ separation now allows Reconstructed energy for a high purity 1-ring, 1-Michel ν e Number of events 20 (7.48 × 10 POT) Xiaoyue Li fiTQun event selection February 10, 2017 19 / 69 distribution of Osc. ν CC e selection Osc. ν CC e CC1 -enriched 1.5 / CC ν ν µ µ Beam / CC ν ν e e e sample. NC 2 MC w/ sin θ =0.0217 • Eventually, Theia may have a better tag 13 FiTQun selection 1 yields a much of below Cherenkov pions via smaller numu CC 0.5 scintillation (if separable from protons, background, etc.), but this is not yet included which has a large 0 systematic 0 500 1000 Reconstructed energy (MeV) ν uncertainty. Xiaoyue Li fiTQun event selection February 10, 2017 19 / 69

Theia ν e Samples (0 vs 0+1 Decay-e) FiTQun FiTQun nue&QE& nuebar&QE& nue&NQE& nuebar&NQE& NC& numu&CC& ≤ 1 decay 0 decay Electrons Electrons • By adding in the new “1-ring” CC π + sample, we see a very large gain in ν e CC non-QE efficiency x 25% • More than 50% increase in the 2-3 GeV region • These events have the largest cross section at the DUNE oscillation maximum

Boosted Decision Tree π 0 Cut NC ν e THEIA collaboration meeting • The best-fit likelihoods and reconstructed kinematics of the multi-ring fits were combined into a boosted decision tree • The primary goal of this cut is to remove neutral current ( π 0 ) background (as in the LBNE analysis)

1-Ring Event Samples 300 Neutrino&running& NUEQE=562 250 ANUEQE=4 NUENQE=684 200 Number of Events ANUENQE=7 1-ring, NC=1390 BEAM NUE=289 150 NUMU CC=75 0 decay e 100 50 1-ring, 0 1 decay e 2 4 6 8 Reconstructed Energy (GeV) • The 1-ring, 0-decay-e sample has a substantially reduced NC background • The new 1-ring, 1-decay-e sample increases the statistics by ~30% • The purity of this sample is also higher

Multi-ring Samples • Additional 2- and 3-ring samples also have controllable backgrounds • Selections have not yet been optimized for CP sensitivity