Документ взят из кэша поисковой машины. Адрес оригинального документа : http://zebu.uoregon.edu/~uochep/talks/talks00/frey.pdf Дата изменения: Fri Apr 13 00:55:32 2001 Дата индексирования: Tue Oct 2 09:51:47 2012 Кодировка: Поисковые слова: m 5

Study of Energy Flow in Jet Reconstruction
R. Frey & M. Iwasaki, Univ. of Oregon

Good jet reconstruction essential to explore and make use of all decay modes multi-jet masses: e.g. Zh vs ZZ vs WW reconstruct parton angles to extract quantum numbers, anomalous moments, e.g. WW, tt, t ! bq q 0 Use combination of tracker and calorimeter which provides best resolution: 0 tracker for h , EM cal. for 0 , HAD cal. for KL, etc. Requires excellent h id. EM Cal. segmentation Realistic modelling requires more-than-primitive cal. clustering algorithms This Study: Develop EFlowtechnique in LCD simulation Implications for detector design in terms of physics benchmarks Compare to other techniques for jet recon. Start with LCD Fast Simulation MovetoFull Sim. Gizmo GEANT 4, clustering alg.

c.f.

N. Graf talk


The LCD Fast Simulation Cal. The L, S and P detectors di erent optimized for ease of simulation description not cost L and S with highly segmented EM Cal for EFlow One Cal. shower cluster" per MC particle Energies and momenta smeared by standard gaussian parameterizations positions also smeared in 3-D. helical extrapolation of charged particles through tracker and calorimeter Capability to merge clusters to produce a little realism Same framework as Full Sim. root or JAS


Ident. and measurementofPhotons Here, used e+e, ! ZZ ! 4q Start by looking at all Cal. clusters. Use to id. photons: Longitudinal depth of shower max. cluster max. or shower start No charged tracks overlap with cluster helical extrapolation of tracks to cluster position 2-D separation bend, non-bend Nearest charged track does not give p = E Combine these photon candidates with charged tracks ! nd jets


R = cluster radial position S detector top photons; bottom :
R - photons (cm) 700 600 500 400 300 200 100 0 0

, inner wall of cal.
dRph Nent = 1033 Mean = 75.87 RMS = 0.7432

20

40

60

80

100

120

140

160

180

200

R - pi+- (cm)

dRpi Nent = 779 Mean = 97.03 RMS = 23.91

70 60 50 40 30 20 10 0 0

20

40

60

80

100

120

140

160

180

200

L detector top photons; bottom :
R - photons (cm) 1000
dRph Nent = 1035 Mean = 200.8 RMS = 0.6366

800

600

400

200

0 0

50

100

150

200

250

300

350

400

450

500

R - pi+- (cm)

dRpi Nent = 710 Mean = 231.7 RMS = 30.31

140 120 100 80 60 40 20 0 0

50

100

150

200

250

300

350

400

450

500


Separation of Cluster and nearest charged track extrapolated Small Detector: BR2 =3:4T-m2 , Rm =0:9cm Cluster is due to a :
photon clus-trk: dz vs drphi 14 12 10 8 6 4 100 2 0 0 0 0 300
d2D Nent = 779 Mean x = 0.6505 Mean y = 0.2787 RMS x = 1.631 RMS y = 0.6634

dz (cm) 500

dZ Nent = 779 Mean = 0.3568 RMS = 1.045

400

200

2

4

6

8

10

12

14

2

4

6

8

10

12

14

drphi (cm) 400 350 300 250 200 150 100 50 0 0 2 4 6 8 10

dRPhi Nent = 779 Mean = 0.6665 RMS = 1.658

12

14

Cluster is due to a :
photon clus-trk: dz vs drphi 14 12 10 8 6 4 20 2 0 0 0 0 60
d2D Nent = 1033 Mean x = 3.124 Mean y = 2.602 RMS x = 3.137 RMS y = 2.984

dz (cm) 100

dZ Nent = 1033 Mean = 2.745 RMS = 3.037

80

40

2

4

6

8

10

12

14

2

4

6

8

10

12

14

drphi (cm) 60 50 40 30 20 10 0 0

dRPhi Nent = 1033 Mean = 3.225 RMS = 3.221

2

4

6

8

10

12

14


Separation between Cluster and nearest charged track extrapolated Large Detector: BR2 =12 T-m2 , Rm =1:6cm Cluster is due to a :
photon clus-trk: dz vs drphi 14 12 10 8 250 6 4 2 0 0 200 150 100 50 2 4 6 8 10 12 14 0 0 2 4 6 8 10 12 14
d2D Nent = 710 Mean x = 0.6962 Mean y = 0.2795 RMS x = 1.919 RMS y = 0.7595

dz (cm)

450 400 350 300

dZ Nent = 710 Mean = 0.3429 RMS = 1.012

drphi (cm) 400 350 300 250 200 150 100 50 0 0 2 4 6 8 10

dRPhi Nent = 710 Mean = 0.6962 RMS = 1.919

12

14

Cluster is due to a :
photon clus-trk: dz vs drphi 14 12 10 25 8 20 6 4 2 0 0 15 10 5 0 0
d2D Nent = 1035 Mean x = 5.094 Mean y = 3.859 RMS x = 3.775 RMS y = 3.755

dz (cm)

35 30

dZ Nent = 1035 Mean = 4.313 RMS = 3.859

2

4

6

8

10

12

14

2

4

6

8

10

12

14

drphi (cm) 35 30 25 20 15 10 5 0 0

dRPhi Nent = 1035 Mean = 5.251 RMS = 3.854

2

4

6

8

10

12

14


Combine bend non-bend d2D Also look at dE cluster E , p of nearest track Small Detector Cluster is due to a :
dr trk-clus (cm) 240 220 200 180 160 140 120 100 80 60 40 20 0 0 2 4 6 8 10 12 14
d2D Nent = 779 Mean = 0.8047 RMS = 1.827

E-p, best match (GeV)

dE Nent = 779 Mean = -0.07155 RMS = 1.696

250

200

150

100

50

0 -25

-20

-15

-10

-5

0

5

10

15

20

25

Cluster is due to a :
dr trk-clus (cm) 18 16 14 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14
d2D Nent = 1033 Mean = 6.043 RMS = 3.636

E-p, best match (GeV)

140 120 100 80 60 40 20 0 -25

dE Nent = 1033 Mean = -4.742 RMS = 6.07

-20

-15

-10

-5

0

5

10

15

20

25


Combine bend non-bend d2D Also look at dE cluster E , p of nearest track Large Detector Cluster is due to a :
dr trk-clus (cm) 180 160 140 120 100 80 60 40 20 0 0 2 4 6 8 10 12 14
d2D Nent = 710 Mean = 1.122 RMS = 2.057

E-p, best match (GeV)

dE Nent = 710 Mean = -0.0388 RMS = 1.624

250

200

150

100

50

0 -25

-20

-15

-10

-5

0

5

10

15

20

25

Cluster is due to a :
dr trk-clus (cm) 9 8 7 6 5 4 3 2 1 0 0 2 4 6 8 10 12 14
d2D Nent = 1035 Mean = 8.321 RMS = 3.411

E-p, best match (GeV) 140 120 100 80 60 40 20 0 -25

dE Nent = 1035 Mean = -5.308 RMS = 6.284

-20

-15

-10

-5

0

5

10

15

20

25


Jet-Jet Mass in e+e,

!

ZZ

!

jets

Use thrust axis to divide event: 2 jets vs 2 jets typical simply add additional jets if 2 per hemisphere no extra" jet-jet combinations Start with unsmeared MC particles:
JJ Mass
60
JJM Nent = 118 Mean = 90.82 RMS = 7.089

50

40

30

20

10

0 0

20

40

60

80 100 120 140 160 180 200


Exclude nal-state neutrinos:
JJ Mass
JJM Nent = 90 Mean = 88.28 RMS = 9.211

35 30 25 20 15 10 5 0 0

20

40

60

80 100 120 140 160 180 200

0 Also exclude KL's :
JJ Mass
JJM Nent = 90 Mean = 82.94 RMS = 11.95

18 16 14 12 10 8 6 4 2 0 0 20 40 60

80 100 120 140 160 180 200


Fast MC Simulation Charged Tracks Only::
JJ Mass
12 10 8 6 4 2 0 0
JJM Nent = 76 Mean = 53.7 RMS = 15.78

20

40

60

80 100 120 140 160 180 200

Fast MC Simulation Cal. Clusters Only::
JJ Mass
14 12 10 8 6 4 2 0 0
JJM Nent = 74 Mean = 85.52 RMS = 20.25

20

40

60

80 100 120 140 160 180 200


Energy Flow - Detector S; d2D
JJ Mass

0:5cm,

dE

5 GeV, no
JJM Nent = 70 Mean = 97.5 RMS = 15.41

R

cut:

12 10 8 6 4 2 0 0

20

40

60

80 100 120 140 160 180 200

Energy Flow - Detector L; d2
JJ Mass
14 12 10 8 6 4 2 0 0

D

1:5 cm, dE

5 GeV, no
JJM Nent = 64 Mean = 90.28 RMS = 15.8

R

cut::

20

40

60

80 100 120 140 160 180 200


As expected, E Flowgives better resolution than clusters or tracks alone No parameter optimization or wide study of inputs, but EFlow multiplicity charged mult. R cut shower position cut does not help, since excludes some fake photons, but also excludes neutral hadrons S detector requires track cluster separation d2D of 1 cm or less L is more forgiving - broad minimum up to 5cm


Summary
A start : : : Some optimization possible to nudge Fast Sim. EFlow ! unsmeared 4vector But clearly most important step is to use fully simulated MC and a realistic clustering algorithm Expect challenges with pattern recognition Is it better than using cal.only? Important implications for detector cost and size gure of merit is BR2=Rm Can one construct su ciently ne granularityat modest R and cost ?