Neutrino Mass Hierarchy Determination with PINGU
Ken ClarkUniversity of Toronto
K Clark - NuFACT 2013
IceCube
AMANDA
DeepCore
scattering
DOM: Digital Optical Module
• First need to introduce IceCube
• Instrumented ~1km3 of ice with ~5000 DOMs
• 78 vertical strings, 60 DOMs per string
2
K Clark - NuFACT 2013
IceCube Events
3
K Clark - NuFACT 2013
IceCube
X (m)-100 -50 0 50 100 150 200
Y (m
)
-150
-100
-50
0
50
100 IceCube
PINGU Geometry - 26m String Spacing
125m
• 78 Strings
• 125m string spacing
• 17m DOM spacing
• Add 8 strings
• 75m string spacing
• 7m DOM spacing
• Add 20 strings
• 26m string spacing
• 5m DOM spacing10 TeV 1 EeV1 TeV100 GeV10 GeV1 GeV100 MeV
IceCube
10 MeV
4
K Clark - NuFACT 2013
IceCube Results
• Designed to look for high energy (TeV - PeV) events
• Much success in recent studies at these energies
1.04 + 0.16 PeV 1.14 + 0.17 PeV
5
K Clark - NuFACT 2013
IceCube
X (m)-100 -50 0 50 100 150 200
Y (m
)
-150
-100
-50
0
50
100 IceCube
PINGU Geometry - 26m String Spacing
• 78 Strings
• 125m string spacing
• 17m DOM spacing
• Add 8 strings
• 75m string spacing
• 7m DOM spacing
• Add 20 strings
• 26m string spacing
• 5m DOM spacing10 TeV 1 EeV1 TeV100 GeV10 GeV1 GeV100 MeV
IceCube
10 MeV
6
125m
K Clark - NuFACT 2013
IceCube + DeepCore
• 78 Strings
• 125m string spacing
• 17m DOM spacing
• Add 8 strings
• 75m string spacing
• 7m DOM spacing
• Add 20 strings
• 26m string spacing
• 5m DOM spacing
X (m)-100 -50 0 50 100 150 200
Y (m
)
-150
-100
-50
0
50
100 IceCubeDeepCore
PINGU Geometry - 26m String Spacing
125m
10 TeV 1 EeV1 TeV100 GeV10 GeV1 GeV100 MeV
IceCube
10 MeV
DeepCore
7
75m
K Clark - NuFACT 2013
IceCube + DeepCore
• Addition of extra strings in closer proximity lowers the detection threshold energy
• Volumes shown are calculated at trigger level
8
K Clark - NuFACT 2013
IceCube + DeepCore
• Addition of extra strings in closer proximity lowers the detection threshold energy
• This allows for sensitivity at the energy of an oscillation minimum
9
K Clark - NuFACT 2013
DeepCore Results
• Approximately 1 year of data analyzed
• High rate in detector provides large event sample
• High energy sample constrains uncertainties in fit
10
Phys. Rev. Lett. 111, 081801 (2013)
K Clark - NuFACT 2013
• Oscillation parameters extracted
• Very good agreement with world averages
• Possible improvement with more advanced reconstructions and event selection
DeepCore Results
11
K Clark - NuFACT 2013
IceCube + DeepCore
• 78 Strings
• 125m string spacing
• 17m DOM spacing
• Add 8 strings
• 75m string spacing
• 7m DOM spacing
• Add 20 strings
• 26m string spacing
• 5m DOM spacing
X (m)-100 -50 0 50 100 150 200
Y (m
)
-150
-100
-50
0
50
100 IceCubeDeepCore
PINGU Geometry - 26m String Spacing
125m
75m
10 TeV 1 EeV1 TeV100 GeV10 GeV1 GeV100 MeV
IceCube
10 MeV
DeepCore
12
K Clark - NuFACT 2013
X (m)-100 -50 0 50 100 150 200
Y (m
)-150
-100
-50
0
50
100 IceCubeDeepCorePINGU
PINGU Geometry - 26m String Spacing
IceCube + DeepCore + PINGU• 78 Strings
• 125m string spacing
• 17m DOM spacing
• Add 8 strings
• 75m string spacing
• 7m DOM spacing
• Add >=20 strings
• 26m string spacing
• 5m DOM spacing
125m
75m
26m
10 TeV 1 EeV1 TeV100 GeV10 GeV1 GeV100 MeV
IceCube
10 MeV
DeepCorePINGU
13
K Clark - NuFACT 2013
• 9.28 GeV Neutrino, 4.9 GeV muon, 4.5 GeV cascade
IceCube + DeepCore
Improvement with PINGU
14
K Clark - NuFACT 2013
• 9.28 GeV Neutrino, 4.9 GeV muon, 4.5 GeV cascade
IceCube + DeepCore
Improvement with PINGU
IceCube + DeepCore + PINGU15
K Clark - NuFACT 2013
PINGU Physics• Many topics opened up by the lowering of
the energy threshold• muon neutrino disappearance• tau neutrino appearance• maximal θ23
• dark matter• neutrino mass hierarchy
16
K Clark - NuFACT 2013
PINGU Physics• Many topics opened up by the lowering of
the energy threshold• muon neutrino disappearance• tau neutrino appearance• maximal θ23
• dark matter• neutrino mass hierarchy
17
K Clark - NuFACT 2013
Neutrino Mass Hierarchy• Three flavour
oscillations mean disappearance/appearance of νμ are both important
P⌫µ!⌫µ = 1�
cos
2✓
m13sin
22✓23 ⇥ sin
2
1.27
✓�m
231 +A+ (�m
231)
m
2
◆L
E
�
�sin2✓m13sin22✓23 ⇥ sin2
1.27
✓�m2
31 +A� (�m231)
m
2
◆L
E
�
�sin4✓23sin22✓m13sin
2
1.27(�m2
31)m L
E
�
• Can be used to distinguish mass hieararchy
18
K Clark - NuFACT 2013
1Length (km)
0 2000 4000 6000 8000 10000 12000
Eart
h D
ensi
ty (g
/cm
^3)
0
2
4
6
8
10
12
14Normal Hierarchy
Inverted Hierarchy
°) with Travel Through the Earth - 10 GeV, 179µν→µνP(
)µν
→µν
P(
00.10.20.30.40.50.60.70.80.91
Why Atmospheric Neutrinos?
• Oscillation probability P(νμ→νμ) directly related to L/E• L measured using angle through the Earth
19
K Clark - NuFACT 2013
1
1
2
2Length (km)
0 2000 4000 6000 8000 10000 12000
Eart
h D
ensi
ty (g
/cm
^3)
0
2
4
6
8
10
12
14Normal Hierarchy
Inverted Hierarchy
°) with Travel Through the Earth - 10 GeV, 179µν→µνP(
)µν
→µν
P(
00.10.20.30.40.50.60.70.80.91
Length (km)0 1000 2000 3000 4000 5000 6000 7000
Eart
h D
ensi
ty (g
/cm
^3)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Normal Hierarchy
Inverted Hierarchy
°) with Travel Through the Earth - 6 GeV, 126µν→µνP(
)µν
→µν
P(
00.10.20.30.40.50.60.70.80.91
1
2
Why Atmospheric Neutrinos?
20
K Clark - NuFACT 2013
PINGU Significance
i = cos(zenith)j = energy
21
• Atmospheric neutrinos provide a large flux of incoming neutrinos at a range of incoming angles and energies
• Non-resonance MSW effects make up roughly 2/3 of PINGU’s sensitivity as (anti-)neutrinos travel through the mantle
• Parametric resonance occurs for neutrinos traveling through the core to make up the remaining 1/3
• Resonances are due to matter-induced oscillation phase changes
Core
K Clark - NuFACT 2013
Calculating Significance• Counts in one year are
binned depending on energy and zenith angle
• This includes both νμ and νμ
• There are differences in the number of counts per bin
• These differences create the patterns which give the distinguishability
N IHij �NNH
ijqNNH
ij
i = cos(zenith)j = energy
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Ener
gy (G
eV)
2
4
6
8
10
12
14
16
18
20
200
400
600
800
1000
1200
Counts in PINGU 1 Year - NH
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Ener
gy (G
eV)
2
4
6
8
10
12
14
16
18
20
200
400
600
800
1000
1200
1400Counts in PINGU 1 Year - IH
22
K Clark - NuFACT 2013
Calculating Significance
N IHij �NNH
ijqNNH
ij
i = cos(zenith)j = energy
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Ener
gy (G
eV)
2
4
6
8
10
12
14
16
18
20
-6
-4
-2
0
2
4
6]1/2Distinguishability Metric [(IH-NH)/NH
23
• As an illustration, a scaled subtraction between hierarchies can be done
• Not the way significance is calculated, but illustrative
• As an example of systematics, look at resolutions
Perfect detector resolution
K Clark - NuFACT 2013
Calculating Significance
N IHij �NNH
ijqNNH
ij
i = cos(zenith)j = energy
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Ener
gy (G
eV)
2
4
6
8
10
12
14
16
18
20
-6
-4
-2
0
2
4
6]1/2Distinguishability Metric [(IH-NH)/NH
• As an illustration, a scaled subtraction between hierarchies can be done
• Not the way significance is calculated, but illustrative
• As an example of systematics, look at resolutions
24
Zenith resolution: 10° Energy resolution 1 GeV
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Ener
gy (G
eV)
2
4
6
8
10
12
14
16
18
20
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1]1/2Distinguishability Metric [(IH-NH)/NH
K Clark - NuFACT 2013
Calculating Significance
N IHij �NNH
ijqNNH
ij
i = cos(zenith)j = energy
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Ener
gy (G
eV)
2
4
6
8
10
12
14
16
18
20
-6
-4
-2
0
2
4
6]1/2Distinguishability Metric [(IH-NH)/NH
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Ener
gy (G
eV)
2
4
6
8
10
12
14
16
18
20
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1]1/2Distinguishability Metric [(IH-NH)/NH
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Ener
gy (G
eV)
2
4
6
8
10
12
14
16
18
20
-1
-0.5
0
0.5
1
]1/2Distinguishability Metric [(IH-NH)/NH
• As an illustration, a scaled subtraction between hierarchies can be done
• Not the way significance is calculated, but illustrative
• As an example of systematics, look at resolutions
25
Zenith resolution: 12.5° Energy resolution 3 GeV
K Clark - NuFACT 2013
PINGU Systematics• Reconstructions are obviously a
significant factor in the hierarchy determination
• Currently using DeepCore algorithms
• New more computationally intensive algorithms may improve resolutions
• Also studying many other sytematics
• θ23, θ13 Δm2atm, δCP
• Atmospheric ν spectrum, efficiency errors, etc.
26
Events with Selection Applied - Energy
Events with Selection Applied - Zenith Angle
K Clark - NuFACT 2013
PINGU Hierarchy Analysis• For the final result, two different analysis
methods are being pursued• Both rely on the creation of “templates”
from simulated data
27
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
(Ene
rgy
(GeV
))10
Log
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0
20
40
60
80
100
Inverted Hierarchy Template - Event Selection Applied
Cos(zenith angle)-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
(Ene
rgy
(GeV
))10
Log
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0
20
40
60
80
100
Normal Hierarchy Template - Event Selection Applied
K Clark - NuFACT 2013
Analysis Method 1• Uses the “Asimov” method using the most likely dataset at
each value of the oscillation parameters
• Calculate the χ2 as a function of these parameters and minimize for Δm2>0 and Δm2<0
28
K Clark - NuFACT 2013
Analysis Method 2
• Throw many pseudo-experiments varying the bin contents with statistical errors, allow oscillation parameters to be fit
• Generate likelihood ratios, determine significance from these
29
lambdaInvExpInvCosHist_244_tot
Entries 1e+07Mean 15.84RMS 5.527
LLR-40 -20 0 20 40
1
10
210
310
410
510lambdaInvExpInvCosHist_244_tot
Entries 1e+07Mean 15.84RMS 5.527
= 2.44132MΔ(Observed Inv Exp Inv) / (Obs Inv Exp Norm)
lambdaNormExpInvCosHist_244_tot
Entries 1e+07Mean -13.57RMS 5.416
lambdaNormExpInvCosHist_244_tot
Entries 1e+07Mean -13.57RMS 5.416
lambdaNormExpInvCosHist_244_tot
Entries 1e+07Mean -13.57RMS 5.416
132MΔ
0.2100.2120.2140.2160.2180.2200.2220.2240.2260.2280.2300.2320.2340.2360.2380.2400.2420.2440.2460.2480.2500.2520.2540.2560.2580.2600.2620.2640.2660.2680
1000
2000
3000
4000
5000
310× with Trial Sets Pulled at 0.24413
2MΔBest Likelihood value for
Template:IH Data:IH
Template:NH Data:NH
Template:NH Data:IH
Template:IH Data:NH
K Clark - NuFACT 2013
Supplemental Methods• Two relatively fast analysis methods used to
study effects of systematics & parameters
30
K Clark - NuFACT 2013
Hierarchy Sensitivity
31
• Calculated sensitivity shown under several assumptions• Event selection, efficiency and detector geometry varied
• Note that all systematics have not been included here
arXiv: 1306.5846
K Clark - NuFACT 2013
PINGU Current Activities• Clearly no data to show yet• Studies underway for detailed LoI• Example: choosing a detector geometry
X (m)-100 -50 0 50 100 150 200
Y (m
)
-150
-100
-50
0
50
100PINGU Geometry V6 (Dozier)
IceCube
DeepCore
PINGU (HQE)
PINGU Geometry V6 (Dozier)
32
X (m)-100 -50 0 50 100 150 200
Y (m
)
-150
-100
-50
0
50
100 IceCube
DeepCore
PINGU
87 88 89 90 91 92 93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124 125 126
PINGU Geometry V15 (Ellett)
X (m)-100 -50 0 50 100 150 200
Y (m
)
-150
-100
-50
0
50
100 IceCube
DeepCore
PINGU
87 88 89 90 91 92 93 94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131 132 133
PINGU Geometry V14 (Domi)
K Clark - NuFACT 2013
PINGU Advantages• Relatively fast
• Deployment could begin in the 2016-2017 season, takes 2-3 years
• Relatively inexpensive• Start up costs on the order of $10M plus ~$1.25M
per string
• IceCube provides well understood veto • Well understood technology
• IceCube techniques have proven to be robust33
K Clark - NuFACT 2013
Conclusions
• IceCube and DeepCore have shown the viability of neutrino oscillation physics in the ice
• PINGU will extend the reach of these analyses to lower energies
• Detailed LoI and full proposal will be available soon
34
K Clark - NuFACT 2013
The IceCube/PINGU Collaboration44 institutions - 4 continents - ~250 Physicists
35
University of Toronto
K Clark - NuFACT 201336
K Clark - NuFACT 201337
Oscillograms
K Clark - NuFACT 2013
Monte Carlo Procedure• Neutrino interactions
modeled with GENIE• Results passed to
Geant4 to create light-generating particles
• Custom GPU software propagates light through detector
38
K Clark - NuFACT 2013
PINGU Advantages• This is all known
technology
• Drilling, installation can be done “quickly”
• Low cost (relative to other experiments)
• “We know how to do this”
39