MdcPatRec Tracking Status Zhang Yao , Zhang Xueyao Shandong University
2
Outline
1. Main Drift Chamber (MDC) geometry
2. Tracking background
3. Tracking algorithm
4. Modification of algorithms
5. Tracking Performance
6. Summary
3
Main Drift Chamber (MDC) geometry
Interaction point
Cosθ=0.83
Cosθ=0.93
• 43 sense wire layers group to 11 super-layer
• cosθ from -0.93 to 0.93
• Cell is nearly square in shape
4
MDC Tracking software MdcPatRec
• MDC tracking software purpose– Precise momentum measurement– Efficiency find tracks short and not form IP
• Migrated from BaBar Drifit Chamber Tracking software
• Personpower
– Zhang Xueyao of Shangdong University
– Zhang Yao of Shangdong University
5
MdcPatRec Tracking algorithm
Make Hits
Track Finding : 2-D tracking
Segment finding
Track Finding : 3-D tracking
MdcDigi
MdcHit
Segment
Circle
Helix
6
Tracking algorithm(1) --- Segment finding
4 3
2
1 0
7 6 5
10101001
01234567
Pattern No.0
(2,0)(2,1)(2,2)(2,3)(2,4)(2,5)
(1,0)(1,1)(1,2)(1,3)(1,4)(1,5)
(3,0)(3,1)(3,2)(3,3)(3,4)(3,5)
(0,0)(0,1)(0,2)(0,3)(0,4)(0,5)
clockwise
One of segment patternOne 4-hit segment pattern
Map to a group-word to match exist pattern
We have 8 4-hit patterns and 20 3-hit patterns in all
7
Tracking algorithm(2) --- Track finding
Combine axial segments
Circle fitting
Combine stereo segments
Helix fitting
8
Modification of algorithm
1. MDC GeometryCreate a new interface through MdcGeomSvc to MdcPatRec
reconstruction algorithm
2. Segment constructionChange the segment construction
3. Stereo segment fittingUse new algorithm of fitting suit to MDC geometry
4. Track parameter conversionConvert BaBar tracking parameter to BesIII parameter Add
some new features to tracking result.
9
• Original segment finding algorithm– construct 8 bit group-word according to one reference wire by wire Id
– But the super-layer1 ~ 5 don’t have equal numbers of cell in each layer
– The neighbors in wire Id is not the neighbors in position
– CAN NOT construct by wire id in super-layer 1~5
… … … …
Modification of algorithm(1) --- segment construction
clockwise
(2,0)(2,1)(2,2)(2,3)(2,4)(2,5)
(1,0)(1,1)(1,2)(1,3)(1,4)(1,5)
(3,0)(3,1)(3,2)(3,3)(3,4)(3,5)
(0,0)(0,1)(0,2)(0,3)(0,4)(0,5)
Referencewire
Efficiency of segment finding rely on “neighbor”s finding
10
• Use φ (azimuthal angle) of reference wire get “neighbors”
• Take 2 reference wire– Layer 2 and layer 0: take No. 2 wire as reference wire
– Layer 3 : take No. 4 wire as reference wire
• Verify the shape of segment 4 3
2
1 0
7 6 5
0
1
2
3Layer id
Referencewires
23
12
10
layerlayer
layerlayer
layerlayer
•All segments can pass this test•Almost all hits can form segment
Modification of algorithm(1) --- segment construction
11
Modification of algorithm(2) --- stereo segment fitting
• Lost hits in stereo super-layer (especially inner chamber)
• Stereo layers don’t share rotate angle in super-layer of inner chamber
Calculate approximate z0 and cotθ by “stereo angle of super-layer”
Calculate position of every hits in super- layers, do a linear fit, get z0 and cotθ
Boss V5.0.0 New st. fitting algorithm
12
Modification of algorithm(3) --- Track parameter conversion
Tbabar
babar
z
z
babar
t
babar
babar
SXSEYE
zd
Bp
dd
)()(
tantan
567.33312
0
00
0
Five helix parameters and error matrix
Other parameters of trackand hits converted for TDS
ambiguity flag left - 1 => left 0 right 1 => right 1 don’t know 0 => don’t know 2
Nstereo hits
number of hit in stereo layer
φterminal
φ of hit with max flight lengthMdcRecHit Status If hit is used in helix fitting
•Convert BaBar parameters to BesIII parameters and stored to TDS
13
Tracking Performance
• Use new MDC geometry– Rotate cell number of stereo layer changed
– Number of wires in super-layer No. 6 changed
• Single track event generated by – SingleParticleGun
– fixed in transverse momentum
14
Tracking Performance(1) --- New stereo fitting
Boss V5.0.0 New st fitting algorithm
•Efficiency of stereo super-layer tracking improved•Momentum resolution, spatial resolution improved
Inner chamber hits increase!
Hit distribution on layers
15
Tracking Performance(2) --- spatial distribution
μ- at pt = 1GeV/cDistribution of d0
σ=0.15 mm
μ- at pt = 1GeV/cDistribution of z0
σ=0.8 mm
•d0: signed distance from the pivot to track in x-y plane•z0: signed distance from the pivot to track in z direction
Note: d0 and z0 fitted with double Gaussian
d0 (cm) z0 (cm)
16
Tracking Performance(3) --- Tracking uniformity
μ- at pt = 1GeV/c, cos θ [–1 ,1], [0, 2π]
θ of Mc Truth
φ of Mc Truth
θ after recon.
φ after recon.
φ (rads) φ (rads)
θ (rads) d0 (rads)
Tra
cks
Tra
cks
Tra
cks
Tra
cks
17
Tracking Performance(4) --- Momentum resolution
Note: Fitted with single Gaussian in each bin.
μ- at pt = 1GeV/cMomentum resolutionσ = 0.4%
Momentum resolution Vs Pt (e-,μ-,π,p)
18
Tracking Performance(5) --- spatial residual
μ- at pt = 1GeV/cspatial residualσ = 110μm
spatial residual Vs Pt (e-,μ-,π,p)
pt (GeV/c)Note: Fitted with double Gaussian in each bin.
19
Tracking Performance(6) --- Tracking efficiency for single track
Efficiency vs Pt (e-,μ-,π,p)
•Have good efficiency with decrease of Pt •Can keep efficiency with 99% at noise level of 20% with
noise type1 and type2
95% 95%95%
pt (GeV/c) Noise level (%)
Noise level type0: = C type1: 1/r type2: 1/r2
Noise level (%)
Eff
.
Eff
.
Eff
.
Efficiency vs Noisee- (type0,1,2)
Efficiency vs Noise type0 (e-,k,π-)
20
Summary
1. Released in Boss 5.0.0 , can be used by analysis codes. 2. Segment tracking and stereo fitting improved3. Requirements satisfied from tracking point of view:
• Good momentum and spatial resolution : σxy = 110 μm, σ( δpt/pt ) = 0.4%. (muons)
• High tracking efficiency: efficiency > 99% for single track pt > 300MeV
• Good tracking performance at high noise level
4. Work with MDC calibration next5. Need further checking with low momentum , not form IP,
and multi-track event