A 3D Track Fit with Multiple Scattering for the Mu3e Experiment
Moritz Kiehn for the Mu3e Collaboration
Physikalisches Institut, Universität Heidelberg
Connecting The Dots, Berkeley, 2015-02-10
The Mu3e Experiment 2
• Precision experiment
• Search for µ+→e+e−e+ In this talk
• Experimental concept
• Track fitting in different regimes
• Triplet fit for Mu3e
• Performance comparisons
µ → eee in the Standard Model 3
Features
• Charged lepton flavor violating
• Expected BR(µ→eee)≪10−50
• Current limit from Sindrum
BR(µ→eee)<1·10−12 @90 % CL
Nucl.Phys. B299(1) 1988 (1–6)
• Our Sensitivity: 1 in 1016 decays Importance
• Observable rate only from New Physics
Signal and Backgrounds 4
Signal
e+
e+ e-
Backgrounds Internal Conversion
e- e+
e+ ν
ν
Accidental
e+
e+ e-
• Common vertex
•
Pp~i =0
• p <53 MeV
• Common vertex
•
P~pi 6=0
• In-time
• No common vertex
• Out-of-time Requiresσp<0.3 MeV
σt<1 ns
Detector Concept 5
Target μ Beam
28 MeV/c
Magnetic field∼1 T
Environment
• >109 µ+ decays/s (continous)
• Electrons p <53 MeV/c
Pixel Tracker
• Monolithic active pixels
• 50 µm silicon, 80 µm pixel size
• Continous readout
Detector Concept 5
Target Inner pixel layers
Scintillating fibres
Outer pixel layers μ Beam
28 MeV/c
Magnetic field∼1 T
Environment
• >109 µ+ decays/s (continous)
• Electrons p <53 MeV/c
Pixel Tracker
• Monolithic active pixels
• 50 µm silicon, 80 µm pixel size
• Continous readout
Detector Concept 5
Target Inner pixel layers
Scintillating fibres
Outer pixel layers Recurl pixel layers
Scintillator tiles
μ Beam
28 MeV/c
Magnetic field∼1 T
Environment
• >109 µ+ decays/s (continous)
• Electrons p <53 MeV/c
Pixel Tracker
• Monolithic active pixels
• 50 µm silicon, 80 µm pixel size
• Continous readout
Multiple Scattering 6
Ω MS
θMS
B
θ
MS∼
1pp x /X
0Mu3e Example
• p = 35 MeV/c
• 50 µm Si
• ΩR=5 cm
→ ∆y ≈320 µm
→ Scattering dominates
Tracking Models 7
Position Resolution
σ
ΘMS Scattering
d
Scattering scale factor:
d ·θMS
σ ≪1
Dominated byhit resolution Reconstruction
• Helix fit
1. Circle fit in xy-plane
e.g. Karimaeki 1991
2. Straight line in sz-space
• 2.5D fit, global parameters
Tracking Models (cont’d) 8
Position Resolution
σ
ΘMS Scattering
d
Scattering scale factor:
d ·θMS
σ ≈1
Reconstruction
• Kalman filter
e.g. Fruewirth 1987
• General Broken Lines
Blobel,Kleinwort 2011
• 3D fit
Tracking Models / General Broken Lines 9
☓ ☓ ☓ ☓
ui ui+1 ui-1
Θi
reference trajectory
reference trajectory Global Trajectory (3D)
Local Trajectory (2D)
☓ ☓ ☓ ☓
linearize around reference
Minimize ui andθi
see Kleinwort 2012
Tracking Models (cont’d) 10
Position Resolution
σ
ΘMS Scattering
d
Scattering scale factor:
d ·θMS
σ ≫1
Dominated byscattering Reconstruction
• Kalman filter
• General Broken Lines
• Anything else?
A Triplet of Hits 11
Sensor 1 Sensor 2
Sensor 3
# parameters
+ 9 data points
- 3 start position - 2 direction angles - 1 curvature / radius - 2 path lengths
+ 1 constraints
A Triplet of Hits (cont’d) 12
Sensor 1 Sensor 2
Sensor 3
θMS,2 θMS,1
# parameters
+ 9 data points
- 3 start position - 2 direction angles - 1 curvature / radius - 2 path lengths - 2 scattering angles
- 1 constraints
Additional constraints
• < θMS,i >=0
• < θ2MS,i >=PDG
• ∆E ≈0
A Triplet of Hits (cont’d) 12
Sensor 1 Sensor 2
Sensor 3
θMS,2 θMS,1
# parameters
+ 9 data points
- 3 start position - 2 direction angles - 1 curvature / radius - 2 path lengths - 2 scattering angles
- 1 constraints
Additional constraints
• < θMS,i >=0
• < θ2MS,i >=PDG
• ∆E ≈0
Triplet Fit 13
s01 s12
c2 c1
r01 r12
h0 h2
h1
ϕ01x
ϕ12x
ϕM S
d01 d12
ϕ01
ϕ12
x y
z s
z01 z12
h2
h1
h0
θ12
θ01
s01
s12
θM S
Assumptions:
• No position error
• No energy loss
• Thin scatterer at middle hit
Minimize:
χ2i(R3D) = ϕMS(R3D)2
σϕ2 +θMS(R3D)2 σθ2 Problem: highly non-linear
Solution: linearize around circle.
Triplet Fit 13
s01 s12
c2 c1
r01 r12
h0 h2
h1
ϕ01x
ϕ12x
ϕM S
d01 d12
ϕ01
ϕ12
x y
z s
z01 z12
h2
h1
h0
θ12
θ01
s01
s12
θM S
Assumptions:
• No position error
• No energy loss
• Thin scatterer at middle hit
Minimize:
χ2i(R3D) = ϕMS(R3D)2
σϕ2 +θMS(R3D)2 σθ2 Problem: highly non-linear
Solution: linearize around circle.
Triplet Fit (cont’d) 14
triplet 1
triplet 2
1. Define overlapping triplets χ2( ¯R3D) =X
χ2i 2a. Minimizeχ2 globally
R¯3D =arg min
x
χ2(x)
2b. Equivalent: minimize each triplet R¯3D =
PwiR3D,i Pwi
Compared Track Fits 15
What is considered?
Positions Scattering
Helix ✓ ✗
Triplet ✗ ✓
GeneralBrokenLines ✓ ✓
Coordinates and Track Parameters 16
Coordinate System
x y
z
B~ θ φ
λ
Track Parameters
r
z λ x
x⊥ y Φ
All parameters defined at inner-most layer
Mu3e Geometry 17
0 10 20 30 40 50 60 70 80
x position / mm
0 10 20 30 40 50 60 70 80
y position / mm
• 4 layers
• B = 1 T
• x/X0 = 1 ‰
• σ = 23 µm
• p = 15–53 MeV
Momentum Resolution 18
20 25 30 35 40 45 50
Momentum / MeV/c
0.026 0.028 0.030 0.032 0.034 0.036 0.038 0.040
Relative momentum resolution
= 10°
Helix Triplet
GeneralBrokenLines
7.1
Scattering scale factor f = d
4.5 3.3 MS2.6/
2.1Dip Angle λ Resolution 19
20 25 30 35 40 45 50
Momentum / MeV/c
0.000 0.005 0.010 0.015 0.020
re so lut ion / r ad
= 10°
Helix Triplet
GeneralBrokenLines
7.1
Scattering scale factor f = d
4.5 3.3 MS2.6/
2.1Azimuthal Angle φ Resolution 20
20 25 30 35 40 45 50
Momentum / MeV/c
0.000 0.005 0.010 0.015 0.020 0.025
re so lut ion / r ad
= 10°
Helix Triplet
GeneralBrokenLines
7.1
Scattering scale factor f = d
4.5 3.3 MS2.6/
2.1Local Offset x
⊥Resolution 21
20 25 30 35 40 45 50
Momentum / MeV/c
0.00 0.01 0.02 0.03 0.04 0.05 0.06
x re so lut ion / m m
= 10°
Helix Triplet
GeneralBrokenLines
7.1
Scattering scale factor f = d
4.5 3.3 MS2.6/
2.1Mu3e Geometry with Recurlers 22
0 50 100 150 200 250 300 350
x position / mm
200 150 100 50 0 50 100 150
y position / mm
• 4 layers, but>4 hits
• B = 1 T
• x/X0 = 1 ‰
• σ = 23 µm
• p = 15–53 MeV
Momentum Resolution 23
20 25 30 35 40 45 50
Momentum / MeV/c
0.000 0.005 0.010 0.015 0.020 0.025 0.030
Relative momentum resolution
= 10°
Helix Triplet GBL/Helix GBL/Triplet
7.1
Scattering scale factor f = d
4.5 3.3 MS2.6/
2.1LHC-like Geometry 24
0 50 100 150 200
x position / mm
0 50 100 150 200
y position / mm
• 6 layers
• B = 2 T
• x/X0 = 2 %
• σ = 25 µm
• p = 100–2000 MeV
Momentum Resolution 25
500 1000 1500
Momentum / MeV/c
0.04 0.06 0.08 0.10 0.12
Relative momentum resolution
= 10°
Helix Triplet
GeneralBrokenLines
4.8
Scattering scale factor f = d
1.1 0.63 0.44MS/
0.34Speed 26
2·109µ/s 50 ns integration
Prototype online reconstruction Single Nvidia GTX 680
>109Triplets/s
Summary and Outlook 27
Mu3e triplet fit
• 3D fit w/ scattering
• Fast, non-iterative
• Works for us Applications
• Low momentum, dominating scattering
• Fast online reconstruction
• Reference for refit
Next?
• Full paper coming soon
http://www.psi.ch/mu3e
Backup
The Mu3e Collaboration A1
University Geneva Heidelberg University
Karlsruhe Institute of Technology Mainz University
Paul Scherrer Institute ETH Zürich
University Zürich
Performance Full Mu3e Simulation A2
4 Hits
[MeV/c]
- pmc prec
-5 -4 -3 -2 -1 0 1 2 3 4 5
0 10000 20000 30000 40000 50000 60000
70000 µ = -0.24
= 1.43 σ
6 Hits (Recurler)
[MeV/c]
- pmc prec
-5 -4 -3 -2 -1 0 1 2 3 4 5
0 20 40 60 80 100 120
103
×
= -0.35 µ
= 0.22 σ
Geant4 simulation w/ complete detector No energy loss correction