tracker software update
DESCRIPTION
Tracker Software Update. Adam Dobbs, MICE CM37 7 th Nov 2013. Contents. People Data structure reminder Reconstruction Reminder Pattern Recognition Results Reducer output Momentum Residuals Momentum Resolution Kalman Results Current Status and To Do. People. Ken Long. Hardware - PowerPoint PPT PresentationTRANSCRIPT
Tracker Software Update
Adam Dobbs, MICE CM377th Nov 2013
A. Dobbs, Tracker Software Update 2
Contents• People• Data structure reminder• Reconstruction Reminder• Pattern Recognition Results
• Reducer output• Momentum Residuals• Momentum Resolution
• Kalman Results• Current Status and To Do
07/11/2013
A. Dobbs, Tracker Software Update 3
People
07/11/2013
HardwareGeoff Barber (IC)
Kevin Ladhams (IC)Adam Dobbs (IC)
Melissa George (IC)Craig MacWaters (RAL)Edward Overton (Shef)
Chris Heidt (UCR)
DAQ / Electronics / Controls
David Adey (FNAL)Alan Bross (FNAL)
Matt Robinson (Shef)
SoftwareAdam Dobbs (IC)
Edward Santos (IC)Melissa George (IC)
Chris Hunt (IC)Chris Heidt (UCR)
Ken Long
A. Dobbs, Tracker Software Update 4
Data Structure I
07/11/2013
Spill
Recon EventMC Event
SciFi EventSciFi Hit
PRec Straight Track
PRec Helical Track
Full Straight Track
Full Helical TrackDigit Cluster Spacepoint
A. Dobbs, Tracker Software Update 5
Reconstruction• Digitisation – unpack the real data or digitise MC data• Clustering – look for adjacent channel hits and group them• Spacepoints Reconstruction – look for intersecting clusters on
different planes• Pattern Recognition – use a linear least squares circle fit in x-y, and
straight line fit in s-z to associate spacepoints with tracks• Final track fit – use a Kalman filter to smooth and filter the tracks,
accounting with multiple coulomb scattering and energy loss
07/11/2013
A. Dobbs, Tracker Software Update 6
Results I: Pattern Recognition
07/11/2013
Helical Pattern Recognition tracks in T2, shown using a Reducer
x (m m )-10 0 10 2 0 30 4 0 5 0
y(m
m)
-4 0
-2 00
2 0
4 0
6 0
T racke r 2 X -Y P ro je c tio n
z (m m )0 2 0 0 4 00 60 0 8 00 1 0 0 0 1 2 00
x(m
m)
-1 00
1 0
2 0
3 0
4 0
5 0
T ra cke r 2 Z -X P ro je c tio n
z (m m )0 20 0 4 00 6 0 0 8 00 10 0 0 12 00
y(m
m)
-4 0
-2 00
2 0
4 0
6 0
T ra cke r 2 Z -Y P ro je c tio n
A. Dobbs, Tracker Software Update 7
T1 Momentum Residual Histogramst1 _ p t_ re s
E n trie s 99 52
M e an -0 .03 0 07
R M S 0.44 41
U n d erflo w 3
O ve rflow 0
(M e V /c )T
- pM C
Tp
-5 -4 -3 -2 -1 0 1 2 3 4 50
2 0 0
4 0 0
6 0 0
8 0 0
1 0 0 0t1 _ p t_ re s
E n trie s 99 52
M e an -0 .03 0 07
R M S 0.44 41
U n d erflo w 3
O ve rflow 0
R e s id u a lt
T 1 p
t1 _ p z_ re sE n tr ies 99 52
M ea n -0 .26 8 3
R M S 3 .62 7
U nd e rflow 4 4
O ve rflo w 2 1
(M e V /c )z
- pM C
zp
-30 -2 0 -1 0 0 1 0 2 0 3 00
2 0 0
4 0 0
6 0 0
8 0 0
1 0 0 0
1 2 0 0
1 4 0 0 t1 _ p z_ re sE n tr ies 99 52
M ea n -0 .26 8 3
R M S 3 .62 7
U nd e rflow 4 4
O ve rflo w 2 1
R e s idu a lz
T 1 p
t1 _ p z_ re s_ lo gE n trie s 99 52
M ean -0 .3648
R M S 8.995
U n de rflo w 5
O verflow 0
(M e V /c )z
- pM C
zp
-5 0 0-4 0 0-3 00-2 0 0-1 0 0 0 1 0 0 2 0 0 30 0 4 00 5 0 0
1
1 0
21 0
31 0
t1 _ p z_ re s_ lo gE n trie s 99 52
M ean -0 .3648
R M S 8.995
U n de rflo w 5
O verflow 0
R e s id u a lz
T 1 p
07/11/2013
Pattern Recognition momentum residual plots for µ- for 10,000 events• Generally good, some outliers need
investigating• Small amount of charge mis-id (<1 in 500
tracks), effect removed from histos
A. Dobbs, Tracker Software Update 8
T2 Momentum Residual Histograms
07/11/2013
Pattern Recognition momentum residual plots for µ- for 10,000 events • Same comments as before
t2 _p t_ re sE n trie s 9 8 97
M ea n -0 .0 0 7 23 9
R M S 0.43
U n d erflo w 0
O ve rflow 0
(M eV /c )T
- pM C
Tp
-5 -4 -3 -2 -1 0 1 2 3 4 50
2 0 0
4 0 0
6 0 0
8 0 0
1 0 0 0
t2 _p t_ re sE n trie s 9 8 97
M ea n -0 .0 0 7 23 9
R M S 0.43
U n d erflo w 0
O ve rflow 0
R e s idu a lt
T 2 p
t2 _p z_ re sE n trie s 98 97
M ea n 0 .0 4 0 35
R M S 3 .3 68
U n de rflo w 3 0
O v e rflow 1 9
(M e V /c )z
- pM C
zp
-3 0 -2 0 -1 0 0 1 0 2 0 3 00
2 0 0
4 0 0
6 0 0
8 0 0
10 0 0
12 0 0
14 0 0
16 0 0 t2 _p z_ re sE n trie s 98 97
M ea n 0 .0 4 0 35
R M S 3 .3 68
U n de rflo w 3 0
O v e rflow 1 9
R e s id ua lz
T 2 p
t2_ pz_ re s_ logE n tr ie s 98 9 7
M ea n 0 .01 41
R M S 7 .3 5 8
U n d e rflo w 2
O ve rflo w 0
(M e V /c )z
- pM C
zp
-5 0 0-40 0-3 0 0-2 00-1 0 0 0 1 0 0 2 0 0 3 0 0 4 00 5 0 0
1
1 0
21 0
31 0
t2_ pz_ re s_ logE n tr ie s 98 9 7
M ea n 0 .01 41
R M S 7 .3 5 8
U n d e rflo w 2
O ve rflo w 0
R es id u a lz
T 2 p
A. Dobbs, Tracker Software Update 9
T1 Momentum Residual Graphs
07/11/2013
Pt residual vs Pt Pz residual vs Pt
Pattern Recognition momentum residual plots for µ- for 10,000 events
(M e V /c )M C
tp
0 2 0 4 0 6 0 8 0 1 00 1 20
(Me
V/c
)t
- p
MC
tp
-20
-15
-10
-5
0
M C
tres vs . p
tT 1 p M C
tres vs . p
tT 1 p
(M e V /c )M C
tp
0 2 0 40 60 8 0 1 0 0 1 2 0
(Me
V/c
)z
- p
MC
zp
-2 0 0
-1 5 0
-1 0 0
-50
0
50
1 0 0
1 5 0
2 0 0
M C
tre s vs . p
zT 1 p M C
tre s vs . p
zT 1 p
A. Dobbs, Tracker Software Update 10
T2 Momentum Residual Graphs
07/11/2013
Pt residual vs Pt Pz residual vs Pt
Pattern Recognition momentum residual plots for µ- for 10,000 events
(M eV /c )t
p
(M eV /c )M C
tp
0 2 0 4 0 60 80 10 0 1 20
(Me
V/c
)t
- p
MC
tp
-4
-3
-2
-1
0
1
2
3
M C
tre s vs . p
tT 2 p M C
tre s vs . p
tT 2 p
(Me
V/c
)z
- p
MC
zp
(M e V /c )M C
tp
0 2 0 4 0 6 0 8 0 1 0 0 1 2 0
(Me
V/c
)z
- p
MC
zp
-2 0 0
-1 5 0
-1 0 0
-5 0
0
5 0
1 0 0
1 5 0
2 0 0
M C
tre s vs . p
zT 2 p M C
tre s vs . p
zT 2 p
A. Dobbs, Tracker Software Update 11
PR Longitudinal Momentum Resolution
07/11/2013
(M eV /c )M C
tp
0 2 0 4 0 60 8 0 1 0 0
Re
solu
tion
(M
eV
/c)
zp
2
4
6
8
1 0
R eso lu tio nz
T 1 p
(M e V /c )M C
tp
0 20 4 0 60 80 100
Re
solu
tion
(M
eV
/c)
zp
0
2
4
6
8
10
12
R e so lu tio nz
T 2 p
Pattern Recognition longitudinal momentum residual plots for µ- for 10,000 eventsResolution found by forming histograms of pz_mc – pz, for 20 MeV/c pt_mc intervals, followed by gaussian fit, fit sigma gives resolution
NB: Hot off the press (yesterday…) so view with caution for now
Anomalous point off scale
A. Dobbs, Tracker Software Update 12
Results II: Kalman
07/11/2013
• Resolution of the track parameters computed as the difference between MC truth and reconstruction values
• The distribution RMS to beam RMS ratio is shown
• Requirement to measure 10% change in emittance to 1% accuracy → transverse momentum resolution must be better than 10% of the beam RMS → Distro RMS / Beam RMS < 10%
• Results show we are well within this requirement! (0.93% and 2.13%)
En tries 9076 6
M ean -0 .0 053 5
R M S 0 .3557
x P os itio n R e so lu tion , m m-2 -1 .5 -1 -0 .5 0 0 .5 1 1 .5 2
0
200
400
600
800
10 00
12 00
14 00
16 00
18 00
20 00
En tries 9076 6
M ean -0 .0 053 5
R M S 0 .3557
E ntrie s 9 0 76 6
M e a n -0 .00 0 5 02 2
R M S 0 .3 60 5
y P os itio n R eso lu t ion , m m-2 -1 .5 -1 -0 .5 0 0 .5 1 1 .5 2
0
200
400
600
800
10 00
12 00
14 00
16 00
18 00
20 00
22 00E ntrie s 9 0 76 6
M e a n -0 .00 0 5 02 2
R M S 0 .3 60 5
E ntrie s 90 7 6 6
M e a n 0 .0 00 1 58 4
R M S 0 .73 5 9
X M o m en tu m R eso lu tion , M eV /c-5 -4 -3 -2 -1 0 1 2 3 4 5
0
10 00
20 00
30 00
40 00
50 00
60 00
E ntrie s 90 7 6 6
M e a n 0 .0 00 1 58 4
R M S 0 .73 5 9
E n tries 90 766
M ean 0 .00124 5
R M S 0 .703 9
Y M o m en tu m R e so lu t ion , M eV /c-5 -4 -3 -2 -1 0 1 2 3 4 5
0
10 00
20 00
30 00
40 00
50 00
60 00
E n tries 90 766
M ean 0 .00124 5
R M S 0 .703 9
RMS Ratio:2.14%
RMS Ratio:2.14%
RMS Ratio:0.93%
RMS Ratio:0.93%
A. Dobbs, Tracker Software Update 13
Current StatusItem Owner Functionality Tests Doc Other To Do
Kalman Santos Trunk Unit No Clean upIntegration tests
Pattern Recognition Dobbs Trunk Unit Some RefactorClean upSpacepoint efficiency studyCheck outliersIntegration tests
Recon to Spacepoints Santos Trunk Unit Some Integration testsDigitisation Santos Trunk Unit NoMC Digitisation Heidt Trunk Unit No Clean upError propagation Santos Development No SomeAlignment Santos Development No
07/11/2013
A. Dobbs, Tracker Software Update 14
Current Status (cont)
07/11/2013
Item Owner Functionality Tests Doc Other To Do
Noise Heidt Development No NoGeometry Heidt Development No No Check geometry in CDB
Access from CDB
Calibrations Heidt / Santos?
Development No No
Performance Hunt Not started No NoReducer / Online Dobbs Development Some No Make fast for online
A. Dobbs, Tracker Software Update 15
Questions
07/11/2013
A. Dobbs, Tracker Software Update 16
Data Structure II• Hits – Monte Carlo objects formed when a particle traverse a detector• Digits – Detector response to a channel hit• Clusters – Groups of digits from adjacent channels in same plane• Spacepoints – 2 or 3 clusters from different planes on the same
station, giving an (x,y) position• PRec Tracks – Tracks found by Pattern Recognition• Full Tracks – The final tracks produced by the Kalman fitter
07/11/2013