Status od LocalReco open issues Conclusion
DT Local Reco
Stefano Lacaprara
INFN LNL
Joint DT-DPG MuonDet PyhsCERN, 3-Apr-2008
Status od LocalReco open issues Conclusion
Outline
1 Status od LocalRecoHitsSegments
2 open issuesPossible solution
3 Conclusion
Status od LocalReco open issues Conclusion
Note on Local reco
A CMS Note on current status of DT Local Reco (Hit andSegment) has been submitted
Reference point for studies on MC simulation
Pending for approval
Status od LocalReco open issues Conclusion
Hits
Hit reco status
Hit reconstruction is stable since some time
Different reco algo are available:
one based on parametrized time-to-distance relationship,taking into account B field and impact angle, from detailedGARFIELD studies. Optimum results on MC studies, used forNotea constant drift velocity, used for Cosmic Data reconstruction
Resolution µm for different wheel at following step ofreconstruction
Superlayer type r-φ r-z, W 0 r-z, W ±1 r-z, W ±2Step 1 237 250 271 308Step 2 231 250 271 305Step 3 207 196 210 228
Status od LocalReco open issues Conclusion
Hits
Hit resolution r -φprojection, different wheel
Entries 3562444
/ ndf 2χ 4.70e+04 / 27
Constant 399± 5.66e+05
Mean 1.30e-05± -2.36e-03
Sigma 1.07e-05± 2.38e-02
)sim-dreco
(d-0.15 -0.1 -0.05 0 0.05 0.1 0.15
# e
ve
nts
0
100
200
300
400
500
600
310×Entries 3562444
/ ndf 2χ 4.70e+04 / 27
Constant 399± 5.66e+05
Mean 1.30e-05± -2.36e-03
Sigma 1.07e-05± 2.38e-02
Entries 3405697
/ ndf 2χ 5.59e+04 / 27
Constant 403± 5.61e+05
Mean 1.29e-05± 9.82e-04
Sigma 1.06e-05± 2.32e-02
)sim-dreco
(d-0.15 -0.1 -0.05 0 0.05 0.1 0.15
# e
ve
nts
0
100
200
300
400
500
600310×
Entries 3405697
/ ndf 2χ 5.59e+04 / 27
Constant 403± 5.61e+05
Mean 1.29e-05± 9.82e-04
Sigma 1.06e-05± 2.32e-02
Entries 3447732
/ ndf 2χ 5.70e+04 / 27
Constant 454± 6.35e+05
Mean 1.15e-05± -2.39e-04
Sigma 9.58e-06± 2.09e-02
)sim-dreco
(d-0.15 -0.1 -0.05 0 0.05 0.1 0.15
# e
ve
nts
0
100
200
300
400
500
600
310×Entries 3447732
/ ndf 2χ 5.70e+04 / 27
Constant 454± 6.35e+05
Mean 1.15e-05± -2.39e-04
Sigma 9.58e-06± 2.09e-02
(cm) (cm) (cm)
Status od LocalReco open issues Conclusion
Hits
Hit resolution r -φprojection vs impact angle
track angle (rad)0 0.1 0.2 0.3 0.4 0.5 0.6
) (cm
)sim-d
reco
(d
-0.2
-0.15
-0.1
-0.05
-0
0.05
0.1
0.15
0.2
track angle (rad)0 0.1 0.2 0.3 0.4 0.5 0.6
) (cm
)sim-d
reco
(d
-0.2
-0.15
-0.1
-0.05
-0
0.05
0.1
0.15
0.2
track angle (rad)0 0.1 0.2 0.3 0.4 0.5 0.6
) (cm
)sim-d
reco
(d
-0.2
-0.15
-0.1
-0.05
-0
0.05
0.1
0.15
0.2
After step 2, when impact angle effect is taken into account byreconstruction SL segments, angle effect is removed
Status od LocalReco open issues Conclusion
Hits
Hit resolution r -φprojection vs distance from FE
distance from FE (cm)0 50 100 150 200 250 300 350 400
) (c
m)
sim
-dre
co(d
-0.2
-0.15
-0.1
-0.05
-0
0.05
0.1
0.15
0.2
distance from FE (cm)0 50 100 150 200 250 300 350 400
) (c
m)
sim
-dre
co(d
-0.2
-0.15
-0.1
-0.05
-0
0.05
0.1
0.15
0.2
distance from FE (cm)0 50 100 150 200 250 300 350 400
) (c
m)
sim
-dre
co(d
-0.2
-0.15
-0.1
-0.05
-0
0.05
0.1
0.15
0.2
After step 3, when coordinate along the wire is reconstructed with3D segments, sistematic shift due to wire propagation time iscured.
Status od LocalReco open issues Conclusion
Hits
Hit Pulls r -φand r -z
Entries 3562444
/ ndf 2χ 3.47e+04 / 97
Constant 95± 1.35e+05
Mean 0.0005± -0.0995
Sigma 0.000± 0.997
recoσ)/sim-dreco
(d-5 -4 -3 -2 -1 0 1 2 3 4 5
# ev
ents
0
20
40
60
80
100
120
140
310×Entries 3562444
/ ndf 2χ 3.47e+04 / 97
Constant 95± 1.35e+05
Mean 0.0005± -0.0995
Sigma 0.000± 0.997
Entries 3405697
/ ndf 2χ 4.03e+04 / 97
Constant 94± 1.31e+05
Mean 0.0006± 0.0425
Sigma 0.000± 0.996
recoσ)/sim-dreco
(d-5 -4 -3 -2 -1 0 1 2 3 4 5
# ev
ents
0
20
40
60
80
100
120
140
310×Entries 3405697
/ ndf 2χ 4.03e+04 / 97
Constant 94± 1.31e+05
Mean 0.0006± 0.0425
Sigma 0.000± 0.996
Entries 3447732
/ ndf 2χ 3.75e+04 / 97
Constant 98± 1.37e+05
Mean 0.0005± -0.0109
Sigma 0.000± 0.968
recoσ)/sim-dreco
(d-5 -4 -3 -2 -1 0 1 2 3 4 5
# ev
ents
0
20
40
60
80
100
120
140
310×Entries 3447732
/ ndf 2χ 3.75e+04 / 97
Constant 98± 1.37e+05
Mean 0.0005± -0.0109
Sigma 0.000± 0.968
Entries 1378880
/ ndf 2χ 1.34e+04 / 97
Constant 57± 5.07e+04
Mean 0.00091± 0.00854
Sigma 0.00± 1.03
recoσ)/sim-dreco
(d-5 -4 -3 -2 -1 0 1 2 3 4 5
# ev
ents
0
10000
20000
30000
40000
50000
Entries 1378880
/ ndf 2χ 1.34e+04 / 97
Constant 57± 5.07e+04
Mean 0.00091± 0.00854
Sigma 0.00± 1.03
Entries 1325584
/ ndf 2χ 1.28e+04 / 97
Constant 56± 4.95e+04
Mean 0.0009± 0.0132
Sigma 0.00± 1.04
recoσ)/sim-dreco
(d-5 -4 -3 -2 -1 0 1 2 3 4 5
# ev
ents
0
10000
20000
30000
40000
50000
Entries 1325584
/ ndf 2χ 1.28e+04 / 97
Constant 56± 4.95e+04
Mean 0.0009± 0.0132
Sigma 0.00± 1.04
Entries 1325584
/ ndf 2χ 1.68e+04 / 97
Constant 58± 4.99e+04
Mean 0.0009± -0.0174
Sigma 0.00± 1.02
recoσ)/sim-dreco
(d-5 -4 -3 -2 -1 0 1 2 3 4 5
# ev
ents
0
10000
20000
30000
40000
50000
Entries 1325584
/ ndf 2χ 1.68e+04 / 97
Constant 58± 4.99e+04
Mean 0.0009± -0.0174
Sigma 0.00± 1.02
Pull under control at all three steps
Status od LocalReco open issues Conclusion
Hits
Hit residuals for param and constant drift time
Distance from wire (cm)0 0.5 1 1.5 2 2.5
(cm
)si
m -
dre
cod
-0.4
-0.2
0
0.2
0.4
Entries 27874Entries 27874
Distance from wire (cm)0 0.5 1 1.5 2 2.5
(cm
)si
m -
dre
cod
-0.4
-0.2
0
0.2
0.4
Entries 27879Entries 27879
Of course, cell non lineary affects constant drift time reconstruction
Status od LocalReco open issues Conclusion
Segments
Segment reco status
Segment reconstruction is stable as well;
still some open issues (see later)
Different reco algo are available/proposed:
Standard combinatorics reconstruction (described in note)Addition to compute vdrift and t0 from the fit itself (Anna M.)MeanTimer tecnique: being developed for low β particlestudies (Piotr T.)noDrift fast reco using just wire position (Martijn M.)
3 steps reconstruction:1 pattern reco for hits, with L/R ambiguity2 refit using track angle3 refit with B and position along wire known by combinig two SL
Status od LocalReco open issues Conclusion
Segments
Segment reco position resolution
Constant 73± 2.87e+04
Mean 1.56e-05± 3.82e-05
Sigma 0.00002± 0.00882
(cm)sim-xrecx-0.15 -0.1 -0.05 0 0.05 0.1 0.150
5000
10000
15000
20000
25000
30000
Constant 73± 2.87e+04
Mean 1.56e-05± 3.82e-05
Sigma 0.00002± 0.00882
Constant 50± 2.17e+04
Mean 0.000021± 0.000152
Sigma 0.0000± 0.0123
(cm)sim
-yrec
y-0.1 -0.05 0 0.05 0.10
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
22000
24000 Constant 50± 2.17e+04
Mean 0.000021± 0.000152
Sigma 0.0000± 0.0123
Status od LocalReco open issues Conclusion
Segments
Segment reco angle resolution
Constant 68± 2.65e+04
Mean 1.27e-06± 3.45e-07
Sigma 0.000001± 0.000715
(rad)simxα-rec
xα-0.015 -0.01 -0.005 0 0.005 0.01 0.0150
5000
10000
15000
20000
25000
30000Constant 68± 2.65e+04
Mean 1.27e-06± 3.45e-07
Sigma 0.000001± 0.000715
Constant 54± 2.21e+04
Mean 1.04e-05± 1.05e-05
Sigma 0.00001± 0.00597
(rad)simyα-rec
yα-0.1-0.08-0.06-0.04-0.02 0 0.020.040.060.08 0.10
5000
10000
15000
20000
25000Constant 54± 2.21e+04
Mean 1.04e-05± 1.05e-05
Sigma 0.00001± 0.00597
Status od LocalReco open issues Conclusion
Segments
Segment reco Pulls
Constant 19± 8.2e+03
Mean 0.0020± 0.0039
Sigma 0.0± 1.1
σ)sim-xrec
(x-5 -4 -3 -2 -1 0 1 2 3 4 50
1000
2000
3000
4000
5000
6000
7000
8000
9000 Constant 19± 8.2e+03
Mean 0.0020± 0.0039
Sigma 0.0± 1.1
Constant 19± 8.2e+03
Mean 0.002± 0.009
Sigma 0.0± 1.1
σ)/sim
-yrec
(y-5 -4 -3 -2 -1 0 1 2 3 4 50
1000
2000
3000
4000
5000
6000
7000
8000
Constant 19± 8.2e+03
Mean 0.002± 0.009
Sigma 0.0± 1.1
Constant 15± 6.4e+03
Mean 0.00191± 0.00091
Sigma 0.0± 1
σ)/simxα-rec
xα(-5 -4 -3 -2 -1 0 1 2 3 4 50
1000
2000
3000
4000
5000
6000
7000 Constant 15± 6.4e+03
Mean 0.00191± 0.00091
Sigma 0.0± 1
Constant 10± 3e+03
Mean 0.0029± 0.0005
Sigma 0.0± 1.1
σ)/simyα-rec
yα(-5 -4 -3 -2 -1 0 1 2 3 4 50
500
1000
1500
2000
2500
3000
Constant 10± 3e+03
Mean 0.0029± 0.0005
Sigma 0.0± 1.1
Status od LocalReco open issues Conclusion
Segments
Segment reco Efficiency
x (cm)-60 -40 -20 0 20 40 60 80 100 120
∈
0.8
0.85
0.9
0.95
1
1.05
Eff
Full chamber
Excluding chamber edge
y (cm)-150 -100 -50 0 50
∈
0.8
0.85
0.9
0.95
1
1.05
Eff
Full chamber
Excluding chamber edge
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x (cm)-50 0 50 100 150
y (
cm
)
-140
-120
-100
-80
-60
-40
-20
0
20
40
x (cm)-150 -100 -50 0 50
∈
0.86
0.88
0.9
0.92
0.94
0.96
0.98
1
1.02
1.04
Eff
Full chamber
Excluding chamber edge
y (cm)-100 -50 0 50 100
∈
0.8
0.85
0.9
0.95
1
1.05
Eff
Full chamber
Excluding chamber edge
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x (cm)-150 -100 -50 0 50 100
y (
cm
)
-120
-100
-80
-60
-40
-20
0
20
40
60
80
x (cm)-150 -100 -50 0 50 100 150
∈
0.8
0.85
0.9
0.95
1
1.05
Eff
Full chamber
Excluding chamber edge
y (cm)-50 0 50 100 150
∈
0.8
0.85
0.9
0.95
1
1.05
Eff
Full chamber
Excluding chamber edge
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x (cm)-150 -100 -50 0 50 100 150
y (
cm
)
-50
0
50
100
150
x (cm)-100 -50 0 50 100 150 200
∈
0.8
0.85
0.9
0.95
1
1.05
Eff
Full chamber
Excluding chamber edge
y (cm)-150 -100 -50 0 50 100 150
∈0.8
0.85
0.9
0.95
1
1.05
Eff
Full chamber
Excluding chamber edge
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x (cm)-50 0 50 100 150 200 250
y (
cm
)
-150
-100
-50
0
50
100
150
Estimated from GREN Data (not MC) by using a sandwitch ofchambers
Status od LocalReco open issues Conclusion
Outline
1 Status od LocalRecoHitsSegments
2 open issuesPossible solution
3 Conclusion
Status od LocalReco open issues Conclusion
Still problem in pattern recongition for r -zSL, where,sometime, a wrong pattern is selectd resulting in wrongdirection of segments
Affects seed generation as well as StandAlone/Globalreconstruction
If a “wrong” r -zsegment is used, other correct segment canbe rejected as incompatible with track candidate
More serius on Cosmic data than on p − p (MC) data due tolack of IP constraint
Code optimization for speed and memory footprint
Task force
A dedicated task force (4 people) is being setup in order to studyand possible solve these problems.
Status od LocalReco open issues Conclusion
Possible solution
Possible Solution being discussed
The problem is not in pattern reco (which does find correctsegments) but in segment cleaning, where sometime we rejectthe right segments and retain the bad onedifficult to fine tune cleaner...
in r -zthe track are straight!Try a reconstruction using simultaneously more than one station inorder to increase the lever arm
Prototype algo using hits from layers of different stationfirst results are promising but there are several issues to beaddressed
which chambers to considercombinatoric increase and can be problematicshow to use these multi SL segments in track reconstruction
Status od LocalReco open issues Conclusion
Possible solution
Other approach
reconstruct clusters of hits in each SL
can be useful also for high Pt Muons with large showering,where segment parretrn reco is hard/impossible
Use hits from a SL in a station (as now) plus a clusters “largehit” from close by SL in other station
much less combinatorics, still longer lever arm, avoid problemin tuning error for track propagation in magnet-joke
can reuse much of the current code, need a “clusterer” for DT
...
Status od LocalReco open issues Conclusion
Outline
1 Status od LocalRecoHitsSegments
2 open issuesPossible solution
3 Conclusion
Status od LocalReco open issues Conclusion
Conclusion
Reconstruction Status
hit and segment reconstruction is rather stable
other recon algos are beeing developed and are under test
still issues on r -zreconstruction, important mostly for cosmicreconstruction
dedicated task force to cope with these issues setup: somepossible smart idea are available.