FIRST OSCILLATION
RESULTS FROM NOνA
Yale NPA Seminar
Dec 3rd 2015
Louise Suter, Argonne National Laboratory
for the NOνA collaboration
L. Suter
220 collaborators
from
40 institutions and
6 countries
Yale NPA Seminar
2
L. Suter
Yale NPA Seminar
3
NOvA physics goals
•  Designed to measure 4 channels in
NuMI beam
•  νe appearance (ν
_µ
_
•  νe appearance (νµ
ν_e)
νe)
•  νµ disappearance (νµ
ν_µ)
_
_
•  νµ disappearance (νµ
νµ)
_
_
•  Probability of νe(νe) appearance in νµ(νµ)
beam depends on θ13, θ23, δCP, Δm231,
Δm232
•  Results in sensitivity to the neutrino
mass hierarchy, CP violation and the
θ23 octant
•  Broad non-oscillation program
•  Neutrino cross-sections
•  Sterile neutrino searches
•  Supernova searches
•  Searches for monopoles and other exotics
Inverted hierarchy
Normal hierarchy
L. Suter
Yale NPA Seminar
4
NOvA physics goals
•  Designed to measure 4 channels in
NuMI beam
•  νe appearance (ν
_µ
_
•  νe appearance (νµ
ν_e)
νe)
•  νµ disappearance (νµ
ν_µ)
_
_
•  νµ disappearance (νµ
νµ)
_
Inverted hierarchy
_
•  Probability of νe(νe) appearance in νµ(νµ)
beam depends on θ13, θ23, δCP, Δm231,
Δm232
•  Results in sensitivity to the neutrino
mass hierarchy, CP violation and the
θ23 octant
•  Broad non-oscillation program
•  Neutrino cross-sections
•  Sterile neutrino searches
•  Supernova searches
•  Searches for monopoles and other exotics
Normal hierarchy
δCP
L. Suter
Yale NPA Seminar
5
NOvA physics goals
•  Designed to measure 4 channels in
NuMI beam
•  νe appearance (ν
_µ
_
•  νe appearance (νµ
ν_e)
νe)
•  νµ disappearance (νµ
ν_µ)
_
_
•  νµ disappearance (νµ
νµ)
_
O
Inverted hierarchy θ23 > 45
_
•  Probability of νe(νe) appearance in νµ(νµ)
Normal hierarchy
beam depends on θ13, θ23, δCP, Δm231,
Δm232
•  Results in sensitivity to the neutrino
mass hierarchy, CP violation and the
θ23 octant
•  Broad non-oscillation program
•  Neutrino cross-sections
•  Sterile neutrino searches
•  Supernova searches
•  Searches for monopoles and other exotics
θ23 < 45O
L. Suter
Yale NPA Seminar
6
NOvA physics goals
•  Designed to measure 4 channels in
NuMI beam
•  νe appearance (ν
_µ
_
•  νe appearance (νµ
ν_e)
νe)
•  νµ disappearance (νµ
ν_µ)
_
_
•  νµ disappearance (νµ
νµ)
_
_
•  Probability of νe(νe) appearance in νµ(νµ)
beam depends on θ13, θ23, δCP, Δm231,
Δm232
•  Results in sensitivity to the neutrino
mass hierarchy, CP violation and the
θ23 octant
•  Broad non-oscillation program
•  Neutrino cross-sections
•  Sterile neutrino searches
•  Supernova searches
•  Searches for monopoles and other exotics
Inverted hierarchy
Normal hierarchy
L. Suter
Yale NPA Seminar
7
Off-axis long-baseline neutrino oscillation experiment
MINOS,
Sudan
L. Suter
Yale NPA Seminar
8
Off-Axis long-baseline neutrino oscillation experiment
MINOS,
Sudan
Longest baseline of any
running accelerator neutrino
experiment at 810km
Increased baseline
increases sensitivity to mass
hierarchy
L. Suter
9
Yale NPA Seminar
Off-Axis long-baseline neutrino oscillation experiment
Detectors located 0.8o of NuMI
beam axis
MINOS,
Sudan
νµ CC / 6E20 POT / kton / 0.1 GeV
Far Detector flux
NOνA Simulation
On-Axis
FLUKA11
Narrowly peaked ν flux centered
at 2 GeV
7 mrad Off-Axis
15
14.6 mrad Off-Axis (NOν A)
21 mrad Off-Axis
E⌫ =
10
| m232 |L
2⇡
⇡ 2 GeV
Achieves maximum oscillation
5
Suppresses high energy tail
0
5
2 GeV
Eν [GeV]
10
15
L. Suter
10
Yale NPA Seminar
•  The ND and FD have similar but
8
θ = 0 mrad
θ = 7 mrad
Eν [GeV]
not identical spectrum
•  Neutrino energy relies on the
angle between π decay and ν
interaction in detector
10
•  Off-axis the dependence on pion
energy becomes flat
•  The ND sees decays from a
θ = 14.6 mrad (NOν A)
6
θ = 21 mrad
4
2
0
0
10
20
30
40
Eπ [GeV]
broader range of angles, whereas
the FD sees a point source
Decay pipe
p
target
νµ
ND
FD
π+
809 km
L. Suter
Yale NPA Seminar
11
Need to design a detector
with excellent νe identification
and background rejection
Want a huge, low-Z, totally
active, tracking calorimeter
11
L. Suter
Yale NPA Seminar
12
To 1 APD pixel
γ
L15.8 m
15.8 m
typical
charged
particle
path
Need to design a detector
with excellent νe identification
and background rejection
Want a huge, low-Z, totally
active, tracking calorimeter
W
D
4 cm
PVC cell filled with liquid scintillator (mineral oil doped with 5% pseudocumene)
12
L. Suter
Yale NPA Seminar
13
To 1 APD pixel
L15.8 m
γ
typical
charged
particle
path
W
D
4 cm
344,064 identical cells
constructed into 896
alternating horizontal and
vertical planes, each of 384
cells long
Cells constructed into horizontal and virtual planes for 3D reconstruction
13
L. Suter
Yale NPA Seminar
14
32 pixel APD
32 WSF loops
Each cell has loop of wavelength shifting fiber read out in groups of 32 by a 32
pixel Avalanche Photodiode
14
L. Suter
Yale NPA Seminar
15
60 m
15.6 m
Far Detector, on surface
14 kTons
896 layers
344,064 pixels
m
Bea
4.2 m
ND, 0.3 kTons
214 layers
20,192 pixels
4.2 m
15.8 m
Two identical 65% active low-Z tracking calorimeters
15.6 m
L. Suter
Yale NPA Seminar
Far Detector
16
Near Detector
Two identical detectors
•  Measure ν rates after oscillation
•  Use of a ratio measurement allows for
cancelation of most systematics
•  Large flux used to characterize ν
beam before oscillation
•  Use data to predict expected rate
at FD
L. Suter
17
Yale NPA Seminar
1m
Proton
1m
νμ + n → μ + p
dE/dx ≈ 12.9 MeV/cm (MIP through cell)
Muon
νμ Charged Current
Proton
Michel eνe + n → e + p
Electron
νe Charged Current
Proton
π0 (→γγ)
Neutral Current
gap
ν + X → ν + X'
Low-Z to enhance electron
photon separation, each
plane is just 0.15 X0
L. Suter
Yale NPA Seminar
•  Far Detector started data taking in
March 2013, with both detectors
completed August 2014
•  NOvA running in medium energy
NuMI beam with a beam power from
330-500 kW
•  Slip-stacking Booster batches since
March 2015
•  Upgrade to 700 kW, NOvA design
power, dependent on a Booster
upgrade and 4+6 slip-stacking
Number hits
Double the intensity
in the first two
Booster batches
Time of ND hits
18
L. Suter
19
Yale NPA Seminar
Delivered/Recorded POT
Rapid intensity increase since February
1.8"
1.7"
Up to a beam power of ~480kW
1.6"
Beam
shutdowns
1.5"
1.4"
1.3"
1.2"
1e18$POT$Per$Day$
1.1"
1"
0.9"
0.8"
0.7"
0.6"
0.5"
0.4"
0.3"
0.2"
28.day"average"delivered"
0.1"
5/1/15"
4/1/15"
3/1/15"
2/1/15"
1/1/15"
12/1/14"
11/1/14"
10/1/14"
9/1/14"
8/1/14"
7/1/14"
6/1/14"
5/1/14"
4/1/14"
3/1/14"
2/1/14"
1/1/14"
12/1/13"
11/1/13"
10/1/13"
9/1/13"
0"
28.day"average"recorded"
L. Suter
Yale NPA Seminar
20
First analysis data
•  We began collecting physics
data during the construction
phase
•  Modular nature of detector
means as soon as each Far
Detector “diblock” (1 kton) was
fully commissioned it was added
•  FD size is not static throughout
our data set
•  Full suite of FD configurations is
simulated in our analyses
3.45×1020 POT recorded
Average 79.4% of full detector mass
2.74×1020 POT-equiv
L. Suter
Yale NPA Seminar
•  Biggest calibration correction
applied to the NOvA detectors is
due to attenuation in the
wavelength shifting fiber
calibration
window
Detector Calibration
21
Far Detector
Data
•  Light levels drop by a factor of 8
across a FD cell
•  Muons (cosmic or ν induced) are
used to probe detector response
•  Stopping muons provide
standard candle for setting
absolute energy scale
•  Also have multiple probes of the
energy scale which all agree
within 5%
•  Michele e- spectrum, π0 mass, dE/dx of
µ, p
Data
MC 𝜋0 signal
MC bkgd
Data 𝜇: 134.2 ± 2.9 MeV
Data 𝜎: 50.9 ± 2.1 MeV
MC 𝜇: 136.3 ± 0.6 MeV
MC 𝜎: 47.0 ± 0.7 MeV
L. Suter
Yale NPA Seminar
22
Surface far detector, rate is driven
by cosmic muons
Beam
Record 10 µs beam window ± 270 µs side band
Color denotes charge deposited
L. Suter
23
Yale NPA Seminar
Far detector 550µs NOvA event
0
1000
2000
3000
4000
5000
6000
x (cm)
500
0
− 500
Top view
Beam direction
y (cm)
500
0
− 500
Color denotes
deposited charge
Side view
0
1000
2000
3000
UTC Fri Jan 9, 2015
00:13:53.087341608
102
10
1
1043
102
10
10
1
5000
6000
z (cm)
hits
Run: 18620 / 13
Event: 178402 / --
hits
NOvA - FNAL E929
4000
0
100
200
300
400
500
t (µsec)
10
102
3
10
q (ADC)
L. Suter
24
Yale NPA Seminar
Zoomed FD 10µs NOvA event
0
1000
2000
3000
4000
5000
6000
0
− 500
Top view
y (cm)
500
Beam direction
0
− 500
Color denotes
deposited charge
Side view
0
Run: 18620 / 13
Event: 178402 / -UTC Fri Jan 9, 2015
00:13:53.087341608
102
10
1
218
2000
3000
4000
hits
NOvA - FNAL E929
1000
5000
6000
z (cm)
hits
x (cm)
500
102
10
1
219
220
221
222
223
224
225
226
227
228
t (µsec)
10
102
3
10
q (ADC)
L. Suter
25
Yale NPA Seminar
Zoomed FD 10µs NOvA event
3000
3500
4000
4500
5000
5500
x (cm)
200
0
− 200
− 400
y (cm)
400
Top view
Beam direction
200
Color denotes
deposited charge
0
Side view
Run: 18620 / 13
Event: 178402 / -UTC Fri Jan 9, 2015
00:13:53.087341608
4000
4500
102
10
1
222 222.2 222.4 222.6 222.8 223 223.2 223.4 223.6 223.8 224
hits
NOvA - FNAL E929
3500
t (µsec)
hits
3000
5000
5500
z (cm)
10
1
10
102
3
10
q (ADC)
L. Suter
26
Yale NPA Seminar
Zoomed ND 10µs NOvA event
200
0
200
400
600
800
1000
1200
1400
1600
x (cm)
100
0
− 100
Top view
− 200
200
Beam direction
y (cm)
100
0
Color denotes
deposited time
−100
−200
Side view
0
200
400
600
800
1000
1200
1400
1600
z (cm)
UTC Mon Nov 10, 2014
00:10:35.977151232
102
10
1
218
hits
Run: 10542 / 1
Event: 113296 / --
hits
NOvA - FNAL E929
10
1
219
220
221
222
223
224
225
226
227
228
t (µsec)
10
102
3
10
q (ADC)
L. Suter
Yale NPA Seminar
Muon Neutrino
Disappearance
27
L. Suter
Yale NPA Seminar
Selecting Muon Neutrinos
•  Select νµ charged current
events below 5 GeV
•  Suppress NC (both detectors) and
cosmic backgrounds (FD only)
•  Containment cuts require a veto
region around event
•  4-variable k-nearest-neighbor
used to identify muons
•  track length
•  dE/dx along track
•  scattering along track
•  track-only plane fraction
•  Large ND data sample used to
validate simulation
28
L. Suter
Yale NPA Seminar
29
Cosmic Rejection
Rejection factor from
beam timing: 105
event topology: 107
Cosmic rate measured
directly with FD data.
Output of cosmic rejection
decision tree after all other cuts
Based on reconstructed track
direction, position, and length;
and energy and number of hits
in event
L. Suter
Yale NPA Seminar
30
Energy Estimation
Muon energy determined
from track length
Fractional energy resolution
Eν= Eµ+Ehad
NOνA Preliminary
0.15
FD
ND
Hadronic energy
contribution is
calometric sum of offtrack hits
Neutrino spectrum
0.10
0.05
0.00
1
2
3
4
True Neutrino Energy (GeV)
5
Energy resolution
at 2 GeV beam
peak is ~7%
L. Suter
Energy Estimation
•  See good agreement for muon
simulation but the simulated
hadronic system has 21% more
energy than in data
•  The hadronic energy scale is
recalibrated so the total energy
peak of the data matches the MC
•  Correction taken as a systematic on
the absolute energy scale
•  This results in 6% overall neutrino
energy scale uncertainty
•  ND reconstructed energy
distribution is used to produce a
data driven prediction of the FD
spectrum
Yale NPA Seminar
31
L. Suter
32
Yale NPA Seminar
Extrapolation method
•  Having a ND enables a data-driven method to predict the νµ
energy spectrum at the FD
•  Removes the dependence on MC simulation of the flux
•  Identical detector construction cancels detector dependent systematic
uncertainties
•  Bin-by-bin direct extrapolation using Far/Near ratio method
0.2
NO A Simulation
×10-3
Ratio(FD/ND)
0.18
0.16
Ratio
0.14
0.12
0.1
0.08
CC events
6e20 POT, FHC, 14kton
µ
0.06
0.04
1
2
3
4
5
6
7
True Energy (GeV)
8
9
10
L. Suter
33
Yale NPA Seminar
The Far/Near ratio extrapolation method
20
6
×10
2.
3.
4.
5.
reconstructed energy
spectrum
Use simulated ND migration
matrix to transform to true
energy spectrum
Apply FD/ND flux ratio
Apply oscillation prediction
(or null prediction)
Use FD migration matrix to
translate back to
reconstructed energy
Events
1.  Starting from observed ND
ND, 1.66×10 POT
NOνA Preliminary
0.12
Simulated ν µ CC
Simulated Background
0.10
Data
0.08
0.06
0.04
0.02
0.00
0
1
2
3
4
Reconstructed Neutrino Energy (GeV)
5
L. Suter
34
Yale NPA Seminar
The Far/Near ratio extrapolation method
1.  Starting from observed ND
4.
5.
90000
80000
4
70000
60000
3
50000
40000
2
30000
20000
1
6e20 POT, FHC, 14kton
0
0
µ
1
10000
CC Contained QE ND events
2
3
4
Reconstructed Energy (GeV)
5
0
Events/6e20 POT/GeV
3.
NO A Simulation
5
True Energy (GeV)
2.
reconstructed energy
spectrum
Use simulated ND migration
matrix to transform to true
energy spectrum
Apply FD/ND flux ratio
Apply oscillation prediction
(or null prediction)
Use FD migration matrix to
translate back to
reconstructed energy
L. Suter
35
Yale NPA Seminar
The Far/Near ratio extrapolation method
1.  Starting from observed ND
3.
4.
5.
0.2
NO A Simulation
×10-3
Ratio(FD/ND)
0.18
0.16
0.14
Ratio
2.
reconstructed energy
spectrum
Use simulated ND migration
matrix to transform to true
energy spectrum
Apply FD/ND flux ratio
Apply oscillation prediction
(or null prediction)
Use FD migration matrix to
translate back to
reconstructed energy
0.12
0.1
0.08
µ
0.06
0.04
CC events
6e20 POT, FHC, 14kton
1
2
3
4
5
6
7
True Energy (GeV)
8
9
10
L. Suter
Yale NPA Seminar
36
The Far/Near ratio extrapolation method
1.  Starting from observed ND
4.
5.
NO A Simulation
5
100
4
80
3
60
2
40
1
0
0
20
6e20 POT, FHC, 14kton
µ CC Contained QE FD events
1
2
3
4
Reconstructed Energy (GeV)
5
0
Events/6e20 POT/GeV
3.
True Energy (GeV)
2.
reconstructed energy
spectrum
Use simulated ND migration
matrix to transform to true
energy spectrum
Apply FD/ND flux ratio
Apply oscillation prediction
(or null prediction)
Use FD migration matrix to
translate back to
reconstructed energy
L. Suter
37
Yale NPA Seminar
Systematic uncertainties
NOνA Preliminary
3.5
Hadronic modeling systematic is
only one with a noticeable effect
Abs. had E scale syst.
3
All systs/osc. par. marg.
2.5
2
Uncertainties assessed
POT-equiv. 90% sensitivity
No systematics
∆m232 (10-3 eV2)
Analysis statistically limited and
all systematic uncertainties
dominated by the absolute
hadronic energy scale
20
NOν A 2.74 × 10
0.3
MC sensitivity
0.4
0.5
2
0.6
0.7
sin θ23
- NC and 𝜈𝜏 CC background rate
- Hadronic energy
(100% each)
(21%, ~equiv. to 6% on E𝜈 )
- Multiple calibration and light-level systematics
- Neutrino flux
(Hit energy, fiber attenuation, threshold effects)
(NA49 + beam transport model)
- Oscillation parameter uncertainties
- Absolute, relative normalization
(current world knowledge)
(1%, 2%)
- Neutrino interactions
(GENIE / Intranuke model)
L. Suter
38
Yale NPA Seminar
Selected νµ event
− 2002200
2400
x (cm)
− 400
2800
3000
3200
7 meters
3 meters
− 300
2600
− 500
Top view
− 100
Beam direction
y (cm)
− 200
− 300
Color denotes
deposited charge
− 400
Side view
− 500
2200
2400
2600
2800
3000
3200
z (cm)
UTC Tue Apr 22, 2014
21:41:51.422846016
hits
Run: 14828 / 38
Event: 192569 / --
hits
NOvA - FNAL E929
102
10
1
102
10
1
218
220
222
224
226
228
t (µsec)
10
102
3
10
q (ADC)
L. Suter
39
Yale NPA Seminar
Selected νµ event
− 200
4200
4400
4600
4800
x (cm)
5200
5400
5600
5800
15 meters
1 meter
− 300
5000
− 400
Top view
− 200
Color denotes
deposited charge
y (cm)
− 300
Beam direction
− 400
Side view
4200
4400
4600
4800
5000
5200
5400
5600
5800
z (cm)
UTC Sun Jan 11, 2015
15:46:53.665326720
hits
Run: 18639 / 10
Event: 141206 / --
hits
NOvA - FNAL E929
102
10
1
102
10
1
218
220
222
224
226
228
t (µsec)
10
102
3
10
q (ADC)
L. Suter
40
Yale NPA Seminar
Selected νµ event
4000
4200
4400
4600
4800
5000
5200
− 100
1.5 meters
x (cm)
− 200
− 300
− 400
Top view
Color denotes
deposited charge
Beam direction
y (cm)
− 500
10 meters
− 600
− 700
Side view
4000
4200
4400
4600
4800
5000
5200
z (cm)
UTC Wed Oct 29, 2014
14:17:32.565656512
102
10
1
hits
Run: 17953 / 38
Event: 256887 / --
hits
NOvA - FNAL E929
102
10
1
218
220
222
224
226
228
t (µsec)
10
102
3
10
q (ADC)
L. Suter
Yale NPA Seminar
νµ Disappearance Results
•  201 events predicted without oscillations
•  Including 2 beam background and 1.4 cosmic events
•  33 events observed!
Oscillations fit
the data well,
𝜒2/Ndof=12.6/16
41
L. Suter
Yale NPA Seminar
νµ Disappearance Results
•  201 events predicted without oscillations
•  Including 2 beam background and 1.4 cosmic events
•  33 events observed!
Oscillations fit
the data well,
𝜒2/Ndof=12.6/16
42
L. Suter
43
Yale NPA Seminar
NOνA Preliminary
7
Selected candidates
6
Background
Spill boundaries
Events / µ s
5
4
3
2
1
0
200
220
240
260
280
300
Main track start Y (cm)
∆t from t0 (µ s)
800
1
600
0.9
0.8
400
0.7
200
0.6
0
0.5
−200
0.4
0.3
−400
0.2
−600
−800
0.1
0
1000
2000
3000
4000
5000
6000
0
Main track start Z (cm)
Note 1: Second timing window at +64 µs required for some of the early data
Note 2: Colors show relative efficiency. Not weighted by time variation in detector size.
L. Suter
Yale NPA Seminar
44
L. Suter
45
Yale NPA Seminar
νµ oscillation probability
NOνA Preliminary
3.5
–2.40 +0.14
-0.17
Normal Hierarchy
90% CL
×10-3 eV2
sin2(𝜃23) = 0.51 ± 0.10
6.5%
measurement
uncertainty
∆m232 (10-3 eV2)
Δm2 =
+0.16
+2.37 -0.15
NOν A 2.74 ×1020 POT-equiv.
68% CL
3
2.5
2
0.3
0.4
0.5
2
0.6
sin θ23
•  Fully quantifying the hadronic response will be essential for the
next generation of results
0.7
L. Suter
Yale NPA Seminar
46
νµ oscillation probability
Δm2 =
+2.37
+0.16
-0.15
–2.40 +0.14
-0.17
×10-3 eV2
sin2(𝜃23) = 0.51 ± 0.10
6.5%
measurement
uncertainty
•  Good compatibility with both MINOS and T2K
•  With less than 10% of the nominal final statistics NOνA is
already competitive with the world limits
L. Suter
Yale NPA Seminar
Electron Neutrino
Appearance
47
L. Suter
Yale NPA Seminar
Selecting Electron Neutrinos
•  Aim to select a pure sample of νe charged
current events at both detectors
•  Select events with electromagnetic showers
•  Suppress backgrounds from
NC/νµCC/beam νe (for FD) and cosmic events
•  Basic cuts to remove obvious backgrounds:
•  Fiducial and Containment
•  Reconstructed pT/p
•  Remove very vertical events
•  Shower length
•  Number of hits
•  Calorimetric energy
48
L. Suter
Yale NPA Seminar
Selecting Electron Neutrinos
Two complimentary electron identification algorithms
Likelihood Identification:
Compare dE/dx in transverse and
longitudinal slices to simulated
e/µ/pi/p+ distributions
First plane will look MIP
like but as move away
from vertex dE/dx for e
and µ will differ
49
L. Suter
Yale NPA Seminar
50
Selecting Electron Neutrinos
Two complimentary electron identification algorithms
Library Event Matching:
Pattern of energy deposition of
entire event compared to a
simulated event library
•  Compare an trial event to a MC library, using individual cell hits rather than highlevel reconstructed variables.
•  Extract a pattern function for the trail event cell by cell, including both position and
charge information
•  Variables based on ‘goodness’ of the match to the best matching events, along
with the calorimetric energy of the trial event are trained in a BDT
L. Suter
51
Yale NPA Seminar
Selecting Electron Neutrinos
Near Detector Data
Validate simulation using beam intrinsic νe’s
Extrapolate beam intrinsic νe’s in the ND to predict background νe rate at the FD
NOνA Preliminary
5
10
4
ND Data
MC NC
MC ν µ CC
MC Beam ν e
Total Beam MC
Flux+Stat. uncertainty
4
10
Events/4.87×1019 POT
Events/4.87×1019 POT
NOνA Preliminary
5
10
3
10
102
10
ND Data
MC NC
MC ν µ CC
MC Beam ν e
Total Beam MC
Flux+Stat. uncertainty
1
0.5
LID
Likelihood Identification
103
102
10
1
10−1
10−1
0
10
1
0
0.2
0.4
0.6
LEM
0.8
Library Event Matching
1
L. Suter
52
Yale NPA Seminar
Selecting Electron Neutrinos
Near Detector Data
NOνA Preliminary
LID
LID
ND Data
Events / 1.66 × 1020 POT
Events / 1.66 × 1020 POT
500
Total MC
(Flux + stat. uncert.)
400
MC Beam ν e
MC NC
300
MC ν µ CC
200
100
0
0
NOνA Preliminary
1
2
3
4
5
Calorimetric Energy (GeV)
1000
LEM
LEM
ND Data
Total MC
(Flux + stat. uncert.)
800
MC Beam ν e
MC NC
600
MC ν µ CC
400
200
0
0
1
2
3
Calorimetric Energy (GeV)
•  Good agreement after selection
•  Data selects ~5% more events than simulation
4
5
L. Suter
Yale NPA Seminar
Selecting Electron Neutrinos
Far Detector Cosmic rejection
•  We achieve better than 1
part in 108 rejection of
cosmic ray interactions
•  We measure this
background using data
collected outside of the
beam spill
•  We expect 0.06 cosmic
events to pass all
selections
53
L. Suter
54
Yale NPA Seminar
Predicting the FD events
NOνA Preliminary
•  Use simulated F/N ratio in bins of
•  Predicted oscillated νe from ND νµ
events
•  Extrapolate beam νe and NC events
to predict FD beam backgrounds
•  FD cosmic rate determined from
out of time FD data
•  Shape information not used just
integrated number of events due
to small number predicted
events at FD
Events / 1.66 × 1020 POT
energy to extrapolate ND data
LID
500
ND Data
Total MC
(Flux + stat. uncert.)
400
MC Beam ν e
MC NC
300
MC ν µ CC
200
100
0
0
1
2
3
4
Calorimetric Energy (GeV)
Before unblinding,
we chose the more
traditional LID as the
primary selector
5
L. Suter
Yale NPA Seminar
55
Systematic Uncertainties
•  Systematics assessed by modifying the simulation used in
the extrapolation
•  Variation in the background and signal prediction taken as
the size of the systematic
signal
background
Calibration
𝜈 Interaction
LID
Scint. Saturation
Normalization
𝜈 Flux
ND BG composition
Other
Total Uncertainty
9.6%
0%
5%
10%
12.9%
LEM has similar systematic
uncertainties
L. Suter
56
Yale NPA Seminar
FD event prediction
We expect about 1 background event
•  Small dependence on oscillation parameters
Total Bkg
Beam νe
NC
νµ CC
ν𝝉 CC
Cosmic
LID
0.94 ± 0.09
0.47
0.36
0.05
0.02
0.06
LEM
1.00 ± 0.11
0.46
0.40
0.07
0.02
0.06
L. Suter
57
Yale NPA Seminar
FD event prediction
We expect about 1 background event
•  Small dependence on oscillation parameters
Total Bkg
Beam νe
NC
νµ CC
ν𝝉 CC
Cosmic
LID
0.94 ± 0.09
0.47
0.36
0.05
0.02
0.06
LEM
1.00 ± 0.11
0.46
0.40
0.07
0.02
0.06
Signal prediction depends on oscillation parameters
NH 𝜹CP=3π/2
IH 𝜹CP=π/2
LID
5.62 ± 0.72
2.24 ± 0.29
LEM
5.91 ± 0.59
2.34 ± 0.23
58
Yale NPA Seminar
L. Suter
νe Appearance Results
NOνA Preliminary
LID
0.8
2.74×1020 POT-equiv.
FD data
LID
Best-fit prediction
Events / 0.25 GeV
Events / 0.25 GeV
2.74×1020 POT-equiv.
0.6
Background
0.4
0.2
0
1
NOνA Preliminary
LEM
0.8
1.5
2
2.5
3
Calorimetric energy (GeV)
LID: Selected 6 events
3.3σ significance for νe appearance
0.6
FD data
LEM
Best-fit prediction
Background
0.4
0.2
0
1
1.5
2
2.5
3
Calorimetric energy (GeV)
LEM: Select 11 events
5.5σ significance for νe appearance
•  All 6 LID events selected by LEM
•  Trinomial probability of selecting this combination (11:6/5/0) is 9.2%
L. Suter
Yale NPA Seminar
59
Selected νe event
3 meters
0.5 meter
Top view
Beam direction
Side view
Color denotes
deposited charge
L. Suter
Yale NPA Seminar
60
Selected νe event
2.5 meters
1.5 meters
Top view
Beam direction
Side view
Color denotes
deposited charge
L. Suter
61
Yale NPA Seminar
Selected νe event
4 meters
2.5 meters
Top view
Beam direction
Side view
Color denotes
deposited charge
L. Suter
Yale NPA Seminar
Event time (𝜇s)
62
L. Suter
νe Appearance Results
LID
sin θ23 = 0.50
2π
Normal hierarchy
Best fit
3π
2
68% C.L.
•  Contours determined using
90% C.L.
δCP
Feldman-Cousins
procedure
63 2
Yale NPA Seminar
LID 2.74×1020 POT equiv.
Reactor 68% C.L.
π
π
2
•  Include errors on solar
parameters
•  Atmospheric Δm2 varied within
new NOvA errors
•  sin2θ23 held fixed at 0.5
NOvA Preliminary
2π
Inverted hierarchy
3π
2
LID
agreement with reactor
measurements
δCP
•  LID results in good
π
π
2
LID: Selected 6 events
3.3σ significance for νe appearance
0
0
0.1
0.2
0.3
sin22θ13
0.4
0.5
L. Suter
νe Appearance Results
LEM
right
•  Some tension with reactor
results, particularly in IH
sin2θ23 = 0.50
2π
Normal hierarchy
NOvA
Preliminary
3π
2
δCP
•  LEM curves shift to the
64
Yale NPA Seminar
LEM 2.74×1020 POT equiv.
LEM
π
π
2
2π
Inverted hierarchy
δCP
3π
2
π
π
2
LEM: Select 11 events
5.5σ significance for νe appearance
0
0
0.1
0.2
0.3
sin22θ13
0.4
0.5
65
Yale NPA Seminar
L. Suter
νe Appearance Results
Can additionally apply the global reactor constraint assuming
sin22θ13= 0.086 ± 0.05
•  LID shows mild tension with IH, 0 < 𝜹CP < 0.8π
•  LEM disfavors IH at greater than 2σ for all 𝜹CP
Both statements hold
for 0.4 < sin2θ23 < 0.6
3σ
Inverted hierarchy
Inverted hierarchy
4σ
NOvA Preliminary
2σ
Normal hierarchy
LEM
90%
Significance
Significance
5σ
Normal hierarchy
LID
sin2θ23 = 0.50
LEM 2.74×1020 POT equiv.
sin2θ23 = 0.50
LID 2.74×1020 POT equiv.
NOvA Preliminary
3σ
2σ
90%
1σ
1σ
0
0
π/2
π
δCP
3π/2
2π
0
0
π/2
π
δCP
3π/2
Jagged structure due to discrete nature of counting experiment
2π
L. Suter
66
Yale NPA Seminar
Summary
•  NOvA observes muon neutrino disappearance
•  6.5% measurement of the atmospheric mass splitting
•  θ23 consistent with maximal mixing
•  Only 1/13th of proposed exposure
•  NOvA observes electron neutrino appearance
selector, 5.5𝜎 for secondary selector.
•  Some preference for normal hierarchy
•  Beam returned in October
•  Plan to increase to 700kW by March 2016
•  Double the exposure by next summer
•  Lots more data to come!
at 3.3𝜎 for primary
L. Suter
67
Yale NPA Seminar
Octant determination after 3+3 years of running
Lower Octant
Upper Octant
2
5
NOνA octant determination
2
20
sin 2θ13=0.095, sin 2θ23=0.95, θ23<π/4
36×10 POT
Inverted
4.5
Normal
4
3.5
3
2.5
2
1.5
1
0.5
0
0
0.2
0.4
0.6
0.8
1
δCP / π
1.2
1.4
1.6
1.8
2
significance of octant determination (σ)
significance of octant determination (σ)
NOνA octant determination
5
sin22θ13=0.095, sin22θ23=0.95, θ23>π/4
36×1020 POT
Inverted
4.5
Normal
4
3.5
3
2.5
2
1.5
1
0.5
0
0
0.2
0.4
0.6
0.8
1
δCP / π
1.2
1.4
1.6
1.8
With 3 FHC + 3 RHC years NOvA has a potential sensitivity to the θ23 octant
up to 2.5σ
2
L. Suter
68
Yale NPA Seminar
νe sensitivities after 3+3 years of running with T2k
CP violation significance
36×1020 POT NOνA
significance of CP violation (σ)
sin22θ13=0.095, sin22θ23=1.00
+ 7×1021 POT T2K
Inverted
3
Normal
2.5
2
1.5
1
0.5
0
0
0.2
0.4
0.6
0.8
1
δCP / π
1.2
1.4
1.6
1.8
2
36×1020 POT NOνA
NOνA+T2K hierarchy resolution
significance of hierarchy resolution (σ)
NOνA+T2K CPV determination
3.5
Mass hierarchy significance
4
sin22θ13=0.095, sin22θ23=1.00
+ 7×1021 POT T2K
Inverted
3.5
Normal
3
2.5
2
1.5
1
0.5
0
0
0.2
0.4
0.6
0.8
1
δCP / π
1.2
1.4
1.6
With 3 FHC + 3 RHC years NOvA has a potential neutrino mass hierarchy
sensitivity up to 3σ and neutrino sector CP violation up to 2σ
1.8
2
L. Suter
69
Yale NPA Seminar
Comparison of cosmic data and MC
NOνA Preliminary
NOνA Preliminary
Events/1 year exposure
Events/1 year exposure
15000
60000
40000
Far Detector Data
20000
Cosmic Simulation
0
-1
-0.5
0
Cosine of Angle to Beam
0.5
Far Detector Data
Cosmic Simulation
10000
5000
0
0
1
2
4
6
Calorimetric Energy (GeV)
NOνA Preliminary
Events/1 year exposure
Events/1 year exposure
4
Cosmic Simulation
10
103
-500
0
Vertex Y (cm)
500
10
NOνA Preliminary
105
Far Detector Data
8
10000
Far Detector Data
8000
Cosmic Simulation
6000
4000
2000
0
0
100
200
300
Number of Hits per Slice
400
500
L. Suter
70
Yale NPA Seminar
Comparison of cosmic data and MC
NOνA Preliminary
×103
NOνA Preliminary
3
5
×10
1.5
Cosmic data
4
Events/6e20 POT
Events/6e20 POT
Cosmic data
Cosmic MC (scaled to data)
1
0.5
Cosmic MC (scaled to data)
3
2
1
0
0
0.2
0.4
0.6
0.8
Cosine of angle between track and beam
1
0
0.2
0.4
0.6
νµ CC Event ID
0.8
1
L. Suter
Yale NPA Seminar
Reverse Horn current
71
L. Suter
Yale NPA Seminar
Forward Horn current
72
L. Suter
Yale NPA Seminar
73
344,064 channels!
99.5% operational
L. Suter
74
Yale NPA Seminar
Chi2 vs Delta
12
12
Normal hierarchy
Inverted hierarchy
10
Inverted hierarchy
8
-2∆logL
-2∆logL
Normal hierarchy
10
8
6
6
4
4
2
2
00
sin2θ23 = 0.50
LEM 2.74×1020 POT equiv.
sin2θ23 = 0.50
LEM 2.74×1020 POT equiv.
π/2
π
δCP
3π/2
2π
00
π/2
π
δCP
3π/2
2π
L. Suter
75
Yale NPA Seminar
LID 2.74×1020 POT equiv.
sin2θ23 = 0.40
2π
Normal hierarchy
LID
Best fit
3π
2
68% C.L.
δCP
90% C.L.
Reactor 68% C.L.
π
π
2
NOvA Preliminary
2π
Inverted hierarchy
δCP
3π
2
π
π
2
0
0
0.1
0.2
0.3
sin22θ13
0.4
0.5
L. Suter
76
Yale NPA Seminar
LID 2.74×1020 POT equiv.
sin2θ23 = 0.60
2π
Normal hierarchy
LID
Best fit
3π
2
68% C.L.
δCP
90% C.L.
Reactor 68% C.L.
π
π
2
NOvA Preliminary
2π
Inverted hierarchy
δCP
3π
2
π
π
2
0
0
0.1
0.2
0.3
sin22θ13
0.4
0.5
L. Suter
77
Yale NPA Seminar
LEM 2.74×1020 POT equiv.
sin2θ23 = 0.60
2π
Normal hierarchy
LEM
δCP
3π
2
NOvA Preliminary
π
π
2
2π
Inverted hierarchy
δCP
3π
2
π
π
2
0
0
0.1
0.2
0.3
sin22θ13
0.4
0.5
L. Suter
78
Yale NPA Seminar
sin2θ23 = 0.40
LID 2.74×1020 POT equiv.
3σ
Normal hierarchy
Significance
Inverted hierarchy
NOvA Preliminary
2σ
90%
1σ
sin2θ23 = 0.40
LEM 2.74×1020 POT equiv.
5σ
0
π/2
π
δCP
3π/2
Normal hierarchy
2π
Inverted hierarchy
4σ
Significance
0
NOvA Preliminary
3σ
2σ
90%
1σ
0
0
π/2
π
δCP
3π/2
2π
L. Suter
79
Yale NPA Seminar
sin2θ23 = 0.60
LID 2.74×1020 POT equiv.
3σ
Normal hierarchy
Significance
Inverted hierarchy
NOvA Preliminary
2σ
90%
1σ
sin2θ23 = 0.60
LEM 2.74×1020 POT equiv.
5σ
0
π/2
π
δCP
3π/2
Normal hierarchy
2π
Inverted hierarchy
4σ
Significance
0
NOvA Preliminary
3σ
2σ
90%
1σ
0
0
π/2
π
δCP
3π/2
2π
L. Suter
80
Yale NPA Seminar
νµ oscillation probability
NOνA Preliminary
12
Data
NOν A normal hierarchy
Events / 0.25 GeV
10
Best fit prediction (no systs)
20
2.74×10 POT-equiv
Expected 1-σ syst range
Best fit χ2=12.64 / 16 dof
Best fit prediction (systs)
8
Backgrounds
6
4
2
0
0
1
2
3
4
Reconstructed Neutrino Energy (GeV)
•  Systematics included as nuisance parameters
•  Hadronic energy
•  Flux, normalization, cross sections, NC bkg rate, calibration
5
L. Suter
Yale NPA Seminar
Flux uncertainties
•  NOνA flux modeled using FLUKA/
FLUGG
•  For each detector the flux
uncertainty is large (~20% at 2
GeV peak) and dominated by the
hadron production uncertainties
•  Estimated by comparing the NuMI
NOνA flux peak
target MC predictions to the the thintarget data from NA49
•  Hadron transport uncertainties
were also investigated
•  NuMI target and horn positions, horn
current and magnetic field
uncertainties, and beam spot size and
position
•  Determined to be small compared
to hadron production uncertainties
NOνA flux peak
81
L. Suter
Yale NPA Seminar
82
NOνA flux peak
Flux uncertainties are highly correlated between the two detectors.
In F/N ratio flux uncertainty reduced to percent level
L. Suter
83
Yale NPA Seminar
Systematic uncertainties summary
Systematic
Value @ 1σ
Best fit (σ)
Bkg. (NC and ντ)
100%
0.06
Absolute Normalization
1.3%
0.0004
Absolute Hadronic energy scale 22%
-0.67
Absolute energy scale
1%
0.06
Beam
Energy dependent
(20% @ 2 GeV)
-0.02
Relative Normalization
1.3%
-0.03
Relative Hadronic energy scale
5.4%
0.05
GENIE Ma
15-25%
-0.06
GENIE Mv
10%
-0.06
Oscillation parameters marginalized over in fit
δCP = Unconstrained, Δm221 = (7.53 ± 0.18)×10-5 , sin22θ13 = 0.086 ± 0.005,
sin2θ12 = 0.846 ± 0.021
L. Suter
Energy resolution
Yale NPA Seminar
84
L. Suter
NOνA Preliminary
20
3
×10
ND, 1.66×10 POT
ND, 1.66×10 POT
NOνA Preliminary
Simulated ν µ CC
Simulated Background
Simulated ν µ CC
Simulated Background
0.15
Data
Events
60
Events
20
6
×10
Hadronic energy
80
85
Yale NPA Seminar
40
Data
0.10
0.05
20
0
0
0.00
0
20
40
Hadronic NHit (GeV)
20
×106
60
ND, 1.66×10 POT
0.20
80
1
2
3
4
5
Hadronic energy (GeV)
100
NOνA Preliminary
NOνA Preliminary
20
×106
ND, 1.66×10 POT
0.15
Simulated ν µ CC
Simulated Background
Simulated ν µ CC
Simulated Background
Data
Data
0.15
Events
Events
0.10
0.10
0.05
0.05
0.00
0
1
2
3
4
Slice Calorimetric Energy (GeV)
5
0.00
0
0.01
0.02
0.03
Average Hadronic Energy Per Hit (GeV)
0.04
L. Suter
Uncertainties on
hadron
production from
NA49
Yale NPA Seminar
86
L. Suter
87
Yale NPA Seminar
Signal Prediction
•  Signal predictions based on ND νµCC energy
spectrum
•  No direct benchmark of simulation of signal events
•  Independent EM samples show good data/MC
agreement
Data
MC 𝜋0 signal
MC bkgd
Data 𝜇: 134.2 ± 2.9 MeV
Data 𝜎: 50.9 ± 2.1 MeV
MC 𝜇: 136.3 ± 0.6 MeV
MC 𝜎: 47.0 ± 0.7 MeV
L. Suter
88
Yale NPA Seminar
Background characteristics
Both selection techniques
achieve good sensitivity to νe
appearance
NOνA FD, Background, LID
NOνA Simulation
•  35% signal selection efficiency
Selected background
dominated by beam νe and
NC DIS events
•  Most NC events have an
energetic π0
Events/3.52×1020 POT
(wrt containment)
•  Reject 99.7% of NC
backgrounds
•  62% expected overlap of the
signal
0.25
ν µCC QE
ν µCC Res
ν µCC DIS
ν µCC Coh
ν eCC QE
ν eCC Res
ν eCC DIS
ν eCC Coh
NC QE
NC Res
NC DIS
NC Coh
0.2
0.15
0.1
0.05
0
1
1.5
2
2.5
Calorimetric Energy (GeV)
3
L. Suter
89
Yale NPA Seminar
Calibration uncertainty
Hadronic energy scale
•  Determine 21% hadronic energy correction (6% on Eν) using ND data
•  Hadronic energy scale determined from tuning data to very well known
off-axis Eν energy peak
•  We conservatively take a 100% absolute uncertainty on this correction
•  The ND and FD have different acceptances due to their different sizes
•  This is our largest systematic uncertainty
# events / 1.66e20 POT
# events / 1.66e20 POT
NOν A Preliminary
50000
All ν µ cuts applied
ND MC
ND data (recalibrated)
40000
30000
20000
NOν A Preliminary
20000
All ν µ cuts applied
ND MC
15000
ND data (recalibrated)
10000
5000
10000
0
0
0.5
1
1.5
2
2.5
3
Reconstructed Hadronic E (GeV)
0
0
1
2
3
4
5
Reconstructed neutrino E (GeV)
Combined to give a 22% total absolute hadronic energy uncertainty
L. Suter
90
Yale NPA Seminar
Calibration uncertainty
Hadronic energy scale
•  Estimate relative uncertainty due to the different detector acceptances
•  As 21% scale is calculated using ND data may be optimized for only
ND
•  Investigated by allowing the normalization and the energy scale of
DIS, RES and QE events (as defined by GENIE) to float
•  Do a three parameter simultaneous fit of Eν, Eµ and Ehad
•  Take the difference between the one-parameter scaling used and this
interaction-dependent scaling to determine the relative uncertainty
Determine a 2% relative uncertainty and 1% relative normalization uncertainty
NOν A Preliminary
NOν A Preliminary
All ν µ cuts applied
40000
All ν µ cuts applied
Events / 1.66e20 POT
Events / 1.66e20 POT
ND data
ND data
15000
ND MC 3-par fit
30000
ND MC 3-par fit
10000
20000
10000
0
0
0.5
1
1.5
2
Reconstructed hadronic energy (GeV)
2.5
3
5000
0
0
1
2
3
Reconstructed neutrino energy (GeV)
4
Combined to give a 5% total relative hadronic energy uncertainty
5
L. Suter
Yale NPA Seminar
Simulation
91
Highly detailed end-to-end simulation chain
•  Beam hadron production, propagation; neutrino flux: FLUKA/FLUGG
•  Cosmic ray flux: CRY
•  Neutrino interactions and FSI modeling: GENIE
•  Detector simulation: GEANT4
•  Readout electronics and DAQ: Custom simulation routines
X (m)
Simulation: Locations of neutrino interactions
that produce activity in the Near Detector
Ryan Patterson, Caltech
Fermilab JETP, August 6, 2015
L. Suter
Energy Scale
Yale NPA Seminar
NC 𝜋0
events
•  All agree to 5%
Data
MC 𝜋0 signal
MC bkgd
Data 𝜇: 134.2 ± 2.9 MeV
Data 𝜎: 50.9 ± 2.1 MeV
MC 𝜇: 136.3 ± 0.6 MeV
MC 𝜎: 47.0 ± 0.7 MeV
•  Near Detector
•  cosmic 𝜇 dE/dx [~vertical]
•  beam 𝜇 dE/dx [~horizontal]
•  Michel e- spectrum
•  𝜋0 mass
•  hadronic shower E-per-hit
•  Far Detector
•  cosmic 𝜇 dE/dx [~vertical]
•  beam 𝜇 dE/dx [~horizontal]
•  Michel e- spectrum
92
𝜈𝜇 CC
events
L. Suter
Reconstruction
Yale NPA Seminar
93
Vertexing: Find lines of energy
depositions w/ Hough transform
CC events: 11 cm resolution
Clustering: Find clusters in
angular space around vertex.
Merge views via topology and
prong dE/dx
Tracking: Trace particle trajectories with Kalman filter tracker (below).
Also have a cosmic ray tracker: lightweight, very fast, and useful for large
calibration samples and online monitoring tools.
Ryan Patterson, Caltech
L. Suter
94
Yale NPA Seminar
Making a Neutrino Beam
Production
•
•
•
•
•
Focusing
Decay
Monitoring
Enhanced 700kW NuMI beam line
Cycle time from 2.2 s to 1.3 s using Recycler slip-stacking
Increased intensity: 12 Booster batches up from 11
New high power target
New horn, reconfigured for higher energy beam