muon hlt: status of the algorithms and performance

18
Muon HLT: status of the algorithms and performance Sergio Grancagnolo for the Muon HLT group

Upload: eliza

Post on 21-Jan-2016

41 views

Category:

Documents


0 download

DESCRIPTION

Muon HLT: status of the algorithms and performance. Sergio Grancagnolo for the Muon HLT group. Summary. Performances Efficiencies of muFast, muComb, TrigMoore Efficiencies and p T resolutions for MuidSA and MuidCB EF Comparison TrigMuonEF vs TrigMoore - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Muon HLT: status of the algorithms and performance

Muon HLT: status of the algorithms and performance

Sergio Grancagnolo

for the Muon HLT group

Page 2: Muon HLT: status of the algorithms and performance

Summary

• Performances– Efficiencies of muFast, muComb, TrigMoore

– Efficiencies and pT resolutions for MuidSA and MuidCB

• EF– Comparison TrigMuonEF vs TrigMoore– TRT only chains: comparison with all ID

reconstruction

Page 3: Muon HLT: status of the algorithms and performance

LVL2: muFast efficiency wrt LVL1

e322_s461_r534 OFLCOND-SIM-00-00-03

e322_s461_r535 known problems of the real detector(dead channels, noisy channels, etc.)

vs. pT vs. eta

Page 4: Muon HLT: status of the algorithms and performance

LVL2: muComb efficiency wrt LVL1

vs. pT vs. eta

e322_s461_r534e322_s461_r535

Page 5: Muon HLT: status of the algorithms and performance

EF: MS reconstruction wrt LVL2

vs. pT vs. eta

e322_s461_r534e322_s461_r535

Page 6: Muon HLT: status of the algorithms and performance

EF: Muon extrapolated and combined reconstruction wrt LVL2

e322_s461_r534e322_s461_r535

Extrapolation Combination

Page 7: Muon HLT: status of the algorithms and performance

Muon EF: pT resolution

Extrapolation Combination

e322_s461_r534e322_s461_r535

Page 8: Muon HLT: status of the algorithms and performance

TrigMuonEF - Introduction• Validation of TrigMuonEF wrt TrigMoore performance

• Based on AODs produced in 14.2.20, bugfix + tags to fix bug #39808 – MuonRdoToPrepData-01-03-14– RecExCond-00-01-39– MuonRecExample-00-13-20

• Only mu6 (TrigMoore) and mu6_ef (TrigMuonEF) enabled – Consistent threshold cuts at L2

• valid3.005200.T1_McAtNlo_Jimmy.digit.RDO.e322_s46

• ~19k ttbar events 8

Page 9: Muon HLT: status of the algorithms and performance

Efficiency vs eta TrigMoore

TrigMuonEF

Effi

cien

cy Effi

cien

cy

Extrapolator (MuidStandalone)MS extrapolated @IP

Combined (MuidCombined)

Efficiency w.r.t. L2 muComb

9

Page 10: Muon HLT: status of the algorithms and performance

Turn-on curves

No cuts

pT > 6 GeV

pT > 20 GeV

pT > 40 GeV

Full circles = TrigMoore

Open circles = TrigMuonEF

Efficiency w.r.t. L2 muComb

Effi

cienc

y

pt pt

pt pt

Effi

cienc

y

Effi

cienc

y Effi

cienc

y

Extrapolatorbarrel

Extrapolatorendcaps

Combinedendcaps

Combinedbarrel

10

Page 11: Muon HLT: status of the algorithms and performance

Fake and good tracks

Extrapolation

Combined Extrapolation

Combined

fakes good

fakes good

fakes good

fakes good

mu0 196 (2.1%)

9040

80 (0.8%)

9034

228 (2.5%)

9062

71 (0.8%)

9066

mu6 129 (1.4%)

8933

74 (0.8%)

8970

142 (1.6%)

8928

61(0.7%)

8994

mu20

22 (0.4%)

5919

8 (0.1%)

5951

29 (0.5%)

5905

6 (0.1%)

5953

18600 ttbar events

Same R (<0.1) cut for both extrapolation and combined

R distributionextrapolator

R distributioncombined

TrigMoore

TrigMuonEF

TrigMoore

TrigMuonEF

TrigMoore TrigMuonEF

11

Page 12: Muon HLT: status of the algorithms and performance

eta-phi resolutions vs pt• Each resolution (eta,phi) at vertex is comparable• The TrigMuonEF Extrapolator has worse phi resolution than

TrigMooreTrigMuonEF extrapolator

TrigMoore extrapolator

TrigMuonEF combined

TrigMoore combined

12

Page 13: Muon HLT: status of the algorithms and performance

pT Resolutions

TrigMoore

TrigMuonEF

• The TrigMuonEF Extrapolator has sligthly better pT resolution than TrigMoore for pT<40GeV

pt pt

PT

reso

luti

on

PT

reso

luti

onExtrapolator (MuidStandalone) Combined (MuidCombined)

13

Page 14: Muon HLT: status of the algorithms and performance

14

LVL2 ID tracking with TRT

Using TRTSegFinder_Muon on a cosmic runNo tracks combinedNeed to pass to the cosmic configuration for ID algos

• muComb code with TRT only ready (TrigmuComb-00-00-42)

• Not yet in the release• Require a LVL2 chain under

developement

Page 15: Muon HLT: status of the algorithms and performance

Setup to run TrigMuonEF with TRT only

• Tested on val,rel_2,14.2.30 (last week) with testMuonSliceAthenaModernRDO.py on tt events

• Tags required – Reconstruction/MuonIdentification/MuidCombinedTools-00-03-01– Trigger/TrigAlgorithms/TrigMuonEF-00-00-48

• Instead of– TrigEFIDInsideOut_Muon()– TrigMuonEFCombinerConfig()

use– TrigEFIDOutsideInTRTOnly_Muon()– TrigMuonEFCombinerConfigTRTOnly()

within Trigger/TriggerCommon/TriggerMenuPython/python/Muon.py • TrigMuonEFCombiner configured with

– MuidMatchMaker.MatchTrtTracks = True– self.IdTracks = "InDetTrigTrackSlimmerTRTOnly_Muon_EFID"

• ID tracks name changed from InDetTrigParticleCreation_Muon_EFID to InDetTrigParticleCreationTRTOnly_Muon_EFID

Page 16: Muon HLT: status of the algorithms and performance

Comparison TRT only / all ID

• Number of TRT hits shown– More TRT only tracks reconstructed

Entries 4365Entries 4143

All ID

Tracks that do not reack TRT

TRT only

Page 17: Muon HLT: status of the algorithms and performance

Comparison TRT only / all ID (II)

• eta distribution for MS-ID combined tracks– Fake TRT only tracks killed in combination– Lost tracks are outside the TRT acceptance

Entries 811Entries 884

All ID TRT only

Page 18: Muon HLT: status of the algorithms and performance

Conclusions

• Muon HLT is performing well

• Transition to TrigMuonEF ready– To be enabled the pT hypo algo if requested

• First results with TRT only tracks encouraging– To complete configuration for LVL2