Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://zebu.uoregon.edu/~uochep/talks/talks00/Iwasaki.pdf Äàòà èçìåíåíèÿ: Fri Apr 13 00:56:12 2001 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 10:33:10 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï

LCD Root analysis and simulation tools

10/26/2000 M. Iwasaki University of Oregon


LCD Root Simulation/Analysis Flow
Generator PANDORA-PYTHIA (C++), PYBMS, PYTHIA, ISAJET,... Output: stdHEP (HEPEVT common) Fast Simulator For physics analysis Simulator Full Simulator:GISMO For detector study Output: Ascii (FastMC) or SIO binary (Full) Þ Convert to root data Event Analysis We use Root for FastMC and Analysis


Why Root? 1) There are many experiment groups using Root It is very easy to get use to... 2) Many software are written by C++ Currently: GEANT4, GISMO, Pandora, ... Future: CERNLIB, PYTHIA, STDHEP, ... 3) Root is maintained by many peoples in the world and there are many useful classes Vector Matrix operations, Lorentz Boost, Rotation, ..
For example: TLorentzVector pJet1(px1,py1,pz1,E1), pJet2(px2,py2,pz2,E2); TLorentzVector pW = pJet1 + pJet2; double MassW = sqrt(pW*pW);

Operator Overloading


LCD Simulation/Analysis with Root
FastMC ·Track .. Smear & bend charged particles Set 5 parameter error matrix (B.Schumm) ·Cluster .. Smear particle position & Energy Cluster merging ·IP .. Smear position FullMC ·Track .. Smear & bend charged particles Set 5 parameter error matrix (B.Schumm) Apply min-Hit & min-PT cut Tracking .. Not yet ·Cluster .. Make clusters by Cheater Algorithm (gather cal hits which are from the same particle)


LCD Simulation/Analysis with Root
Analysis tools ·Jet Finder ... 3 kinds of algorithms ·Thrust Finder ·Particle extrapolator ·Topological Vertexing ... transrated from SLD ZVTOP (T.Abe)


Tracks at Root
[c] /5 1800 1600 1400 1200 1000 800 600 400 200 0 0 0.5 1 1.5 2 2.5 3 3.5 4
2

Maintained by T. Abe

Now we installed B.Schumm's 5 parameter Covariant matrix for track smearing
hchi2 Nent = 45907 Mean = 0.9962 RMS = 0.6306

4.5

5

Conf. Level 500

hcnfl Nent = 45907 Mean = 0.5019 RMS = 0.2889

400

300

200

100

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Topological vertex finder (Translated from SLD ZVTOP: See T.Abe's talk) Reconstruct secondary vertices in a jet .. Find points of high overlap tracks
10000

8000

uds

1400 1200 1000

c

2000 1800 1600 1400 1200

b

6000 800

1000 4000 600 400 2000 200 0 0 0 0 800 600 400 200 1 2 3 4 5 6 7 8 9 10 0 0 1 2 3 4 5 6 7 8 9 10

1

2

3

4

5

6

7

8

9

10

#Ver tices reconstructed


Using the information of reconstructed vertices, we can do 1) Heavy-flavor tagging Mass tag method:: See T.Abe's talk

1. Reconstruct Second Vertex 2. Form 'PT-corrected mass' of SV Mcorr. = (M
vtx 2

+|PTvtx| 2) + |P
SV eIP

T

vtx

|

PTvtx Pvtx

e+

3. Identify heavy-quark signals Charm 0.6

bottom charm uds

400 350 300 250 200 150 100 50 0 0

Ver tex Momentum (GeV)

450

45 40 35 30 25 20 15 10 5 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

1

2

3

4

5

6

7

PT corrected Ver tex Mass (GeV)

Ver tex Mass (GeV)

b-quark efficiency 60% (50% :SLD) purity 98% (98%) c-quark efficiency 30% (16%) purity 80% (70%)


Analysis example:
Reconstruction of t-quark mass in tt 6jets event Without b-tag
1400 1200 1000 800 300 600 200 400 200 0 100 100 0 100 600 500 400

With b-tag

Signal BG
120 140 160 180 200 220 240 260

120

140

160

180

200

220

240

260

Mass (GeV/c2)

Mass (GeV/c2)

(FastMC)


2) Charge separation See also T.Abe's talk
1000

BB B
+ 0

800

600

400

200

0

-4

-3

-2

-1 0 1 Vertex Charge

2

3

4


Clusters at Root
Calorimeter Hits
250 250 200 200

Clusters

(cm)

100

R
-300 -200 -100 0 100 200 300

cluster

150

(cm)

150

R

hit

100

50

50

0

0 -300 -200 -100 0 100 200 300

Zhit (cm)

Zcluster (cm)

(Full MC)


Clusters at Root
Fast MC
450 400 1600 350 1400 300 1200 250 1000 200 150 100 50 0 0 800 600 400 200 0 0
ClsE Nent = 9358 Mean = 2.527 RMS = 2.273

Full MC
1800

ClsE Nent = 18473 Mean = 1.865 RMS = 2

2

4

6

8

10

2

4

6

8

10

Cluster Energy(GeV) Energy scale for Full :: determined by µ


Cluster merging effect The fast simulator does not have Cluster width yet Near clusters might be merged and regard as one cluster

Merging probability (transverse only) 13mrad 20mrad(NLC) 30mrad(JLC) Small 2.7% 5.4% 10.0% Large 5.1% 9.4%


For Energy flow analysis We'll see Track-Cluster matching to separate charged/neutral clusters

We need realistic · Cluster position resolution · Cluster width(spread) in our Simulator


Fast MC
1000

Track = Cluster

dL_t Nent = 3533 Mean = 1.198 RMS = 2.342

Photon
60

dL_g Nent = 4844 Mean = 8.806 RMS = 5.24

800 50

600

40

30

20

10

0 0

2

4

6

8

10

12

14

16

18

20

0 0

2

4

6

8

10

12

14

16

18

20

Full MC
Track = Cluster
350 300
dL_s Nent = 7461 Mean = 8.868 RMS = 8.669

Photon
200 180 160 140 120

dL_g Nent = 9869 Mean = 15.05 RMS = 9.985

250

200 100 150 80 60 40 50 20 0 0

100

0 0

5

10

15

20

25

30

35

40

5

10

15

20

25

30

35

40


Charged/Neutral cluster separation performance
1

Fast MC Large
Neutral cluster efficiency
0.9

Fast MC Small

0.8

0.7

0.6

Full MC Large Full MC Small

0.5

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Charged Cluster rejection factor

Very much different between FastMC and Full MC Þ Need tune up the FastMC


Summary
1) Root analysis tools work well Especially Topological Vertexing is the excellent tool for Heavy-flavor tagging 2) For Fast MC Þ Need detail studies using FullMC to input the realistic detector parameters In particular for Calorimeter (Need to make Cal hits, like JLC quick simulator??) There is a LCD ROOT Analysis tutorial page! (Under construction) URL: http://www-sldnt.slac.stanford.edu/nld/New/Docs /LCD_Root/root.htm