nova first results
TRANSCRIPT
Introduc?on• Long-baselineappearancemeasurement
• NuMIbeam• NOvAdetectors
• Muonneutrinodisappearance
• Electronneutrinoappearance
JeffHartnell,MoriondEW2016 2
JeffHartnell,MoriondEW2016 3
Long-baseline 𝜈𝜇→𝜈e For fixed L/E = 0.4 km/MeV A more quantitative sketch…
At right: P(𝜈⎺𝜇→ 𝜈⎺e) vs. P(𝜈𝜇→𝜈e) plotted for a single neutrino energy and baseline
Ryan Patterson, Caltech Fermilab JETP, August 6, 2015 4
JeffHartnell,MoriondEW2016 4
Long-baseline 𝜈𝜇→𝜈e For fixed L/E = 0.4 km/MeV A more quantitative sketch…
At right: P(𝜈⎺𝜇→ 𝜈⎺e) vs. P(𝜈𝜇→𝜈e) plotted for a single neutrino energy and baseline Measure these probabilities (an example measurement of each shown) Also: Both probabilities ∝ sin2𝜃23
Ryan Patterson, Caltech Fermilab JETP, August 6, 2015 5
JeffHartnell,MoriondEW2016 5
NOvAOverview• “Conven?onal”beam• Two-detectorexperiment:
• Neardetector– measurebeamcomposi?on
– energyspectrum
• Fardetector– measureoscilla?onsandsearchfornewphysics
Ash River
Ash River
810 km
KeyFeaturesof2ndGenera?onExpt• Narrowband(off-axis)beam
• Detectorsop?misedfor– νeflavouriden?fica?on– νeappearancemaximum(L/E)
• Higherpowerbeam
• NOvAhasabouttriplethema\ereffectofT2Kandhigherrela?vean?neutrinoxsec
JeffHartnell,MoriondEW2016 6
JeffHartnell,MoriondEW2016 9
To APD
4 cm ⨯ 6 cm
1560 cm
A NO𝜈A cell NO𝜈A detectors
Fiber pairs from 32 cells
32-pixel APD
Far detector: 14-kton, fine-grained, low-Z, highly-active tracking calorimeter → 344,000 channels
Near detector: 0.3-kton version of the same → 20,000 channels
Extruded PVC cells filled with 11M liters of scintillator
instrumented with 𝜆-shifting fiber and APDs
Ryan Patterson, Caltech Fermilab JETP, August 6, 2015 10
JeffHartnell,MoriondEW2016 11
𝜈𝜇 disappearance
(simulated 𝜈𝜇 CC event)
• Identify contained 𝜈𝜇 CC events in each detector • Measure their energies • Extract oscillation information from differences between
the Far and Near energy spectra
NearDetectorNuMuCCSpectrum
JeffHartnell,MoriondEW2016 15
5
Length of Primary Muon Track (m)0 5 10 15
)3Ev
ents
(x10
0
5
10
15
20
25
30
Simulated Selected EventsSimulated BackgroundData
POT20 10×ND, 1.66
)ZθPrimary Muon Track cos(0 0.2 0.4 0.6 0.8 1
Even
ts
1
10
210
310
410
510
610
Simulated Selected EventsSimulated BackgroundData
POT20 10×ND, 1.66
FIG. 2. Reconstructed track length (left) and track angle ✓Z relative to the detector longitudinal axis, along the beam direction(right) for the primary muons in selected ⌫µ CC interactions in the ND. The simulated distributions follow the conventions ofFig. 1.
Hadronic Energy (GeV)0 0.5 1 1.5 2 2.5 3
)3Ev
ents
(x10
0
20
40
60
80
100DataData (w/o 14% offset)Simulated Selected EventsSimulated Background
POT20 10×ND, 1.66
FIG. 3. Reconstructed hadronic energy Ehad for selected⌫µ CC interactions in the ND, both with (black circles) andwithout (gray squares) the 14% o↵set described in the text.The simulated distributions follow the conventions of Fig. 1.
Figure 2 shows the reconstructed muon track param-eters for ⌫µ CC events in the ND. Figure 3 shows theEhad distribution both with and without the 14% dif-ference in Ehad calibration scale between data and sim-ulation. A corresponding ±14% uncertainty is assessedon the hadronic energy scale, and is included in all ofthe uncertainty bands shown. Figure 4 shows the finalE⌫ distribution. The energy resolution for reconstructed⌫µ CC events is estimated from simulation to be 7%.
The prediction for the FD neutrino energy spectrumis based on the observed ND neutrino energy spectrum,with corrections for acceptance and flux di↵erences de-rived from simulation. First, the small NC background,estimated from simulation, is subtracted from the NDdata spectrum. The resulting background-subtractedspectrum is then converted into a true neutrino energy
Reconstructed Neutrino Energy (GeV)0 1 2 3 4 5
)3Ev
ents
(x10
0
20
40
60
80
100
Simulated Selected EventsSimulated BackgroundData
POT20 10×ND, 1.66
FIG. 4. Reconstructed neutrino energy E⌫ for selected ⌫µ CCinteractions in the ND. The simulated distributions follow theconventions of Fig. 1.
spectrum via a mapping derived from simulation. Thistrue neutrino energy spectrum is then used to constructa spectrum in the FD by multiplying it by the energy-dependent ratio of FD-to-ND selected events from sim-ulation. Oscillation probabilities for a given set of os-cillation parameters are then applied, by energy bin, tothe predicted true FD energy spectrum, which is thenmapped to a reconstructed neutrino energy spectrum us-ing FD simulation. The extrapolated ⌫µ CC energy spec-trum is then combined with beam-induced backgrounds(NC, ⌫e CC, and ⌫⌧ CC) predicted from simulation, andthe background spectrum measured using events selectedfrom outside of the beam spill window.
Systematic uncertainties in the calibration, flux esti-mate, cross sections, hadronization modeling, particle-transport modeling and exposure di↵erences between thetwo detectors are assessed by varying these aspects of the
MC scaled up by 7.2% Shape only systematics
JeffHartnell,MoriondEW2016 16Evan Niner I Results from NOvA 02/11/16
νμ Spectrum
• 211.8 ± 12.5 (syst.) events predicted in the absence of oscillations.
• 33 candidate events between 0 and 5 GeV observed.
12
Reconstructed Neutrino Energy (GeV)0 1 2 3 4 5
Even
ts /
0.25
GeV
0
10
20
30
40
50 DataUnoscillated predictionBest fit prediction (no systs)
syst. rangeσExpected 1-Best fit prediction (systs)Backgrounds
Normal Hierarchy POT-equiv.2010×2.74
=19.0/16dof/N2χBest fit
FarDetector
νμDisappearanceResults
JeffHartnell,MoriondEW2016 17Evan Niner I Results from NOvA 02/11/16
Best Fit Oscillation Parameters
13
23θ2sin0.3 0.4 0.5 0.6 0.7
)2 e
V-3
(10
322m
∆
2.0
2.5
3.0
3.5 Normal Hierarchy, 90% CLAνNO
T2K 2014MINOS 2014
arXiv:1601.05037
�m232 =
�2.52+0.20
�0.18
�⇥ 10�3eV2 �m2
32 = (�2.52± 0.19)⇥ 10�3eV2
sin2 (✓23) = [0.38, 0.65] sin2 (✓23) = [0.37, 0.64]
Normal Hierarchy Inverted Hierarchy
(68% CL) (68% CL)Degenerate best fit points at 0.43 and 0.60 Degenerate best fit points at 0.44 and 0.59Evan Niner I Results from NOvA 02/11/16
Best Fit Oscillation Parameters
13
23θ2sin0.3 0.4 0.5 0.6 0.7
)2 e
V-3
(10
322m
∆
2.0
2.5
3.0
3.5 Normal Hierarchy, 90% CLAνNO
T2K 2014MINOS 2014
arXiv:1601.05037
�m232 =
�2.52+0.20
�0.18
�⇥ 10�3eV2 �m2
32 = (�2.52± 0.19)⇥ 10�3eV2
sin2 (✓23) = [0.38, 0.65] sin2 (✓23) = [0.37, 0.64]
Normal Hierarchy Inverted Hierarchy
(68% CL) (68% CL)Degenerate best fit points at 0.43 and 0.60 Degenerate best fit points at 0.44 and 0.59
JeffHartnell,MoriondEW2016 18
𝜈e appearance
(simulated 𝜈e CC event)
• Identify contained 𝜈e CC candidates in each detector • Use Near Det. candidates to predict beam backgrounds
in the Far Detector • Interpret any Far Det. excess over predicted backgrounds
as 𝜈e appearance
νeSelec?on
JeffHartnell,MoriondEW2016 19Evan Niner I Results from NOvA 02/11/16
νe Identification• Likelihood Identifier (LID)
- Compare longitudinal and transverse dE/dx in leading shower to template histograms for e/p/n/μ/π±/π0/γ.
- Build neural net from these inputs and reconstructed quantities.
• Library Event Matching (LEM)- Compares input event to simulated
event library.- Properties from most similar events
fed into decision tree.
• 62% event overlap between selectors. • LID chosen before unblinding as
primary selector.
15
νeAppearance
JeffHartnell,MoriondEW2016 23Evan Niner I Results from NOvA 02/11/16
νe Appearance Analysis Results• LID observed 6 events on
a background prediction of 0.99±0.11(syst), 3.3σ excess.
• LEM observed 11 events on a background of 1.07±0.14(syst), 5.3σ excess.
• All LID events in LEM set.
• 7.8% probability of this overlap configuration or one less likely.
17
JeffHartnell,MoriondEW2016 24
Evan Niner I Results from NOvA 02/11/16
νe Appearance Analysis Results
19
0 0.1 0.2 0.3 0.4 0.5
CPδCPδ
13θ22sin
0
2π
π
2π3
π2
2π
π
2π3
π2 POT equiv.2010× 2.74LID = 0.5023θ2sin
A Preliminary
νNO
Best fit68% C.L.90% C.L.Reactor 68% C.L.
Normal hierarchy
Inverted hierarchy
Normal
2mΔ
221mΔ
Inverted
eν µν τν
Global bestfit from reactor experiments
LIDSelector
JeffHartnell,MoriondEW2016 25Evan Niner I Results from NOvA 02/11/16
νe Appearance Analysis Results
20
0 0.1 0.2 0.3 0.4 0.5
CPδCPδ
13θ22sin
0
2π
π
2π3
π2
2π
π
2π3
π2 POT equiv.2010× 2.74LID = 0.5023θ2sin
A Preliminary
νNO
Best fit68% C.L.90% C.L.Reactor 68% C.L.
Normal hierarchy
Inverted hierarchy
Normal
2mΔ
221mΔ
Inverted
eν µν τν
Global bestfit from reactor experiments
0 0.1 0.2 0.3 0.4 0.5
CPδCPδ
13θ22sin
0
2π
π
2π3
π2
2π
π
2π3
π2 POT equiv.2010× 2.74LEM = 0.5023θ2sin
A Preliminary
νNONormal hierarchy
Inverted hierarchy
LEMSelector
ResultswithReactorConstraint
JeffHartnell,MoriondEW2016 26
Evan Niner I Results from NOvA 02/11/16
Significance with Reactor Constraint
• Apply global reactor constraint- sin2θ13=0.086±0.05
• Marginalize over θ23.• Both selectors weakly
prefer normal mass hierarchy and π<δcp<2π.
• This preference is consistent with T2K (arXiv:1502.01550)
21
CPδ
Sign
ifican
ce
0 /2π π /2π3 π20
σ1
σ2
σ3
σ4
σ5
90%
POT equiv.2010×2.74
NH LID NH LEMIH LID IH LEM
arXiv:1601.05022
Evan Niner I Results from NOvA 02/11/16
Significance with Reactor Constraint
• Apply global reactor constraint- sin2θ13=0.086±0.05
• Marginalize over θ23.• Both selectors weakly
prefer normal mass hierarchy and π<δcp<2π.
• This preference is consistent with T2K (arXiv:1502.01550)
21
CPδ
Sign
ifican
ce
0 /2π π /2π3 π20
σ1
σ2
σ3
σ4
σ5
90%
POT equiv.2010×2.74
NH LID NH LEMIH LID IH LEM
arXiv:1601.05022
Conclusions• FirstNOvAoscilla?onresultswith7.6%ofplannedexposure
• νμdisappearanceconsistentw/MINOS&T2K• νeappearanceresulthintsatnormalhierarchyandπ<δCP<2π,consistentwithT2K
• Cross-sec?onstudiesinprogress,νeCCandcoherentπ0resultsshownatNuINT
• Planning2ndresultwithdoublethesta?s?csforthesummer
• Staytuned!JeffHartnell,MoriondEW2016 27
JeffHartnell,MoriondEW2016 29Evan Niner I Results from NOvA 02/11/16
Relation of Oscillation Parameters in NOvA
25
inverted%hierarchy
normal%hierarchy
Θ23%<%45 o
Θ23%>%45 o
%
Normal Inverted
eν µν τν
2mΔ
221mΔ
JeffHartnell,MoriondEW2016 30Evan Niner I Results from NOvA 02/11/16
νe Appearance Analysis Strategy
• FD background prediction extrapolated from ND - ND selects ~7% more background
in data relative to simulation. • Combination of containment,
topology and event classifier achieve cosmic rejection factor >108. Effective FD fidicual volume of 10 kT.
• “Cut and count” analysis between 1.5 and 2.7 GeV in FD for the primary selector.
16
Calorimetric Energy (GeV)0 1 2 3 4 5
PO
T20
10
×Ev
ents
/ 1.
66
0
100
200
300
400
500A PreliminaryνNOLID
ND Data
(Flux + stat. uncert.)Total MC
eνMC Beam
MC NC
CCµνMC
LID0.0 0.5 1.0 1.5
PO
T20
10
×Ev
ents
/ 1.
66
1−10
1
10
210
310
410
510
610
A PreliminaryνNOND Data eνMC Beam
(Flux + stat. uncert.)Total MC MC NC
CCµνMC
ND energy spectrum with LID>0.95
JeffHartnell,MoriondEW2016 31Evan Niner I Results from NOvA 02/11/1626
(1) Estimate the underlying true energy distribution of selected ND events(2) Multiply by expected Far/Near event ratio and !"→!" oscillation probability
as a function of true energy(3) Convert FD true energy distribution into predicted FD reco energy distributionSystematic uncertainties assessed by varying all MC-based steps
JeffHartnell,MoriondEW2016 32
Evan Niner I Results from NOvA 02/11/1627
Hadronic Energy (GeV)0 0.5 1 1.5 2 2.5 3
)3Ev
ents
(x10
0
20
40
60
80
100DataData (w/o 14% offset)Simulated Selected EventsSimulated Background
POT20 10×ND, 1.66
Length of Primary Muon Track (m)0 5 10 15
)3Ev
ents
(x10
0
5
10
15
20
25
30
Simulated Selected EventsSimulated BackgroundData
POT20 10×ND, 1.66
E⌫ = Eµ + Ehadrons
• Muon variables in agreement• Best fit to hadronic energy prefers 14% increase in data
JeffHartnell,MoriondEW2016 33Evan Niner I Results from NOvA 02/11/1629
• Correct%for%attenuation%in%each%cell%using%throughGgoing%cosmic%muons%(right)%
• Set%absolute%energy%scale%using%stopping%muons%in%data%and%tuned%Monte%Carlo%(below)
calib
ratio
n w
indo
w
Far Detector Data
calib
ratio
n w
indo
w
JeffHartnell,MoriondEW2016 34[G. Feldman]
NOvAFarDetector
Total mass of 14 ktons
An admirer
TASD: Totally Active Scintillator Design Longitudinal sampling is ~0.15 X0, which gives: -- excellent µ-e separation -- π0 rejection capability
Full-size Modules