Decomposition Methods in the NOvA Near Detector forνe appearance
Nitish Nayak (UC-Irvine)
for the NOvA collaboration
New Perspectives 2017
Neutrino Oscillations
I (νe , νµ, ντ ) superpositions of propagating mass eigenstates (ν1, ν2, ν3) →many parameters still unknown!
I NOvA looks at νµ → νµ and νµ → νe channels to constrain theseparameters
I mass hierarchy (NH, IH)
I mixing angle, θ23
I CP-violating phase, δCP
New Perspectives 2017 1 / 14 Nitish Nayak (UC-Irvine)
NOvA
I ND useful for handle on neutrinospectrum
I Ensures we don’t get overwhelmedby systematics at FD
New Perspectives 2017 2 / 14 Nitish Nayak (UC-Irvine)
The νe appearance analysis
Muon
Proton
Michel e-
Electron
Proton
π0 (→γγ)
νμ + n → μ + p
νe + n → e + p
ν + X → ν + X'Proton
1m
1m
νμ Charged Current
νe Charged Current
Neutral Current
I Look for νeCC signal at FD
I Use ND to predict both signal and bkg (beam νeCC, NC, νµCC) at FD
I Fit in oscillation parameter space
New Perspectives 2017 3 / 14 Nitish Nayak (UC-Irvine)
Event classification
FEATURE MAPS
::
I CVN : Raw pixel maps →convolutional filters → NN layers
I CVN improvement in sensitivity ≡30% more exposure
New Perspectives 2017 4 / 14 Nitish Nayak (UC-Irvine)
ND selection
I ND data/MC needed for beambackground decomposition
I Different backgrounds → differentoscillation rates
I Apply FD optimized PID selectionto ND
I Cut out events in surrounding rock
I Ensure good match withkinematics at FD
New Perspectives 2017 5 / 14 Nitish Nayak (UC-Irvine)
Prediction Strategy
I Further bin FD energy into 3 PID bins → same for ND
Decomposition :
I Crude option : split data/MCproportionally based on fractionsin MC
I Better option : find data-drivenconstraints
New Perspectives 2017 6 / 14 Nitish Nayak (UC-Irvine)
BEN Decomposition for beam νe
I beam νe shares ancestry with high-statistics νµCC
I By constraining flux from π, K through νµCC → constrain beam νe
True Neutrino Energy (GeV)0 5 10 15 20
Eve
nts
/ Bin
/ 6E
20 P
OT
0
5
10
15
20
310×
All AncestorsOther Ancestor
Ancestor0LK Ancestor+K Ancestor+π
beam νe CC
True Neutrino Energy (GeV)0 5 10 15 20
Eve
nts
/ Bin
/ 6E
20 P
OT
0
1
2
3
4
610×
All AncestorsOther Ancestor
Ancestor0LK Ancestor+K Ancestor+π
νµCC
New Perspectives 2017 7 / 14 Nitish Nayak (UC-Irvine)
BEN Decomposition
I beam νe shares π phasespace with contained νµ
I Shape correction →important for systematics
Reconstructed neutrino energy (GeV)0 2 4 6 8 10
PO
T20
10×E
vent
s / 3
.72
0
20
40
60
80310×
in NDµνContained
CC from Pionsµν
CC from Kaonsµν
Bkgd. + other
NOvA Preliminary
(GeV/c)πz
tp0 10 20 30 40 50 60
(G
eV/c
)π T
tp
0
0.2
0.4
0.6
0.8
1
0.7
0.8
0.9
1
1.1
1.2
1.3Pion weights
New Perspectives 2017 8 / 14 Nitish Nayak (UC-Irvine)
BEN Decomposition
I Normalization correction fromoverall scales incontained/uncontained events
I π scale goes down by 3%, K goesup by 17%
I Smaller fraction of beam νes fromK → only normalizationconsidered for K
Reconstructed neutrino energy (GeV)0 2 4 6 8 10 12
PO
T20
10×E
vent
s / 3
.72
0
50
100
150
310×
in NDµνUncontained
CC from Pionsµν
CC from Kaonsµν
Bkgd. + other
NOvA Preliminary
New Perspectives 2017 9 / 14 Nitish Nayak (UC-Irvine)
Michel Decomposition for νµ, NC
I On average, 1 more Michel e− in νµCC
I µ decay lifetime ∼ 2.2µs � our resolution → find at µ track-end
I Michels from beam νe and NC come from hadronic portion (π → µ → e)
Number of Michels0 1 2+
PO
T20
10×E
vent
s / 3
.72
0
5000
10000
15000
NOvA Preliminary
Data
CCµνMC
MC NC
eνMC Beam I Input BEN result → NC, νe
degeneracy broken
I Fit with Poisson Likelihood
New Perspectives 2017 10 / 14 Nitish Nayak (UC-Irvine)
Combo
I BEN + Michel prefers higherνµCC and NC (17.7% and 10.4%)
I Small increase in beam νes (1%)
I Takes into account uncertainties influx, cross-section and νe selectionefficiency
Reconstructed neutrino energy (GeV)
Rat
io
0
0.5
1
1.5
2
2.5
CVN
0.75 < CVN < 0.87 0.87 < CVN < 0.95 0.95 < CVN < 1
0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 5
NOvA Preliminary
/ MCeνBeam
CC / MCµν
NC / MC
Reconstructed neutrino energy (GeV)
PO
T20
10×E
vent
s / 3
.72
0
1000
2000
3000
NOvA Preliminary
0.75 < CVN < 0.87 0.87 < CVN < 0.95 0.95 < CVN < 1
0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 5
Data
eνMC Beam
CCµνMC
MC NC
→
Reconstructed neutrino energy (GeV)
PO
T20
10×E
vent
s / 3
.72
0
1000
2000
3000
NOvA Preliminary
0.75 < CVN < 0.87 0.87 < CVN < 0.95 0.95 < CVN < 1
0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 5
Data
eνBeam
CCµν
NC
Uncorrected MC
New Perspectives 2017 11 / 14 Nitish Nayak (UC-Irvine)
Extrapolation for νµ → νe
I Signal : Correct νµ MC by Reco/True Mapping, map back to Reco in νeafter extrapolation
I F/N helps flux/cross-section uncertainties, but care still needed!
I MC embedded in procedure, Reco/True map doesn’t help reco-levelsystematics
I Background : PID misidentifies at both ND, FD → no need for Reco/Truemap, extrapolate directly
New Perspectives 2017 12 / 14 Nitish Nayak (UC-Irvine)
νe selection efficiency
I Expect good EM showersimulation but hadronic?
I νµCC events → remove µ-tracks,simulate EM shower (MRE)
I ∼ 1% data/MC disagreement
I Similar strategy for EM showers →FD cosmic muons with brems(MRBrem)
I Very good agreement
New Perspectives 2017 13 / 14 Nitish Nayak (UC-Irvine)
Summary
I Setting expectations of what we should be seeing at FD is tricky but wehave..
I Number of data-driven techniques using the ND to make precisepredictions
I Cross-checks at ND to make sure we are on the right track
I Good systematics treatment taking advantage of near-identical detectors
I Look out for Shiqi’s talk on the νe appearance results and many othersfrom NOvA
I Stay tuned for further exciting results!
New Perspectives 2017 14 / 14 Nitish Nayak (UC-Irvine)
Backup
New Perspectives 2017 15 / 14 Nitish Nayak (UC-Irvine)
νe ND Data/MC
New Perspectives 2017 16 / 14 Nitish Nayak (UC-Irvine)
νµ ND Data/MC
Reconstructed muon energy (GeV)0 1 2 3 4 5
Eve
nts
310
0
20
40
60
80
100 Simulated selected eventsSimulated backgroundData
syst. rangeσFull 1- POT20 10×ND POT norm., 3.72
NOvA Preliminary
Hadronic energy (GeV)0 0.5 1 1.5 2 2.5 3
Eve
nts
310
0
100
200
300Simulated selected eventsSimulated backgroundData
syst. rangeσFull 1- POT20 10×ND POT norm., 3.72
NOvA Preliminary
Muon ID0 0.2 0.4 0.6 0.8 1
Eve
nts
310
410
510
610
710 Simulated selected eventsSimulated backgroundData
syst. rangeσFull 1- POT20 10×ND POT norm., 3.72
NOvA Preliminary
Length of primary track (m)0 5 10 15
Eve
nts
310
0
20
40
60
80Simulated selected eventsSimulated backgroundData
syst. rangeσShape-only 1- POT20 10×ND area norm., 3.72
NOvA Preliminary
New Perspectives 2017 17 / 14 Nitish Nayak (UC-Irvine)
Systematics Budgeting
New Perspectives 2017 18 / 14 Nitish Nayak (UC-Irvine)
Simulated Flux at NOvA
E (GeV)0 5 10 15
CC
/ 6E
20 P
OT
/ kT
ON
/ 50
MeV
ν 610
0.5
1
1.5
2
2.5
Total
µν
µν
eν + eν
A SimulationνNO
FLUKA11
E (GeV)0 5 10 15
CC
/ 6E
20 P
OT
/ kT
ON
/ 50
MeV
ν 3−10
2−10
1−10
1
10
Total
µν
µν
eν + eν
A SimulationνNO
FLUKA11
ND FD
New Perspectives 2017 19 / 14 Nitish Nayak (UC-Irvine)
νe Selected Event
NOvA - FNAL E929
Run: 19741 / 3Event: 67330 / --
UTC Sun Jun 7, 201517:09:42.290953856 sec)µt (
218 220 222 224 226 228
hits
110
210
q (ADC)10 210 310
hits
110
210
400 600 800 1000 1200 1400x
(cm
)
200
300
400
z (cm)400 600 800 1000 1200 1400
y (c
m)
200−
100−
0
New Perspectives 2017 20 / 14 Nitish Nayak (UC-Irvine)
Muon Removed Electron Events (MRE)
New Perspectives 2017 21 / 14 Nitish Nayak (UC-Irvine)
Muon Removed Bremstrahlung Events (MRBrem)
New Perspectives 2017 22 / 14 Nitish Nayak (UC-Irvine)