20
20 Jun 2022 Muon HLT Meeting C. Diez L2 Muon Trigger Study Status Report B. de la Cruz , C. Diez (CIEMAT, Madrid) Muon HLT Meeting 20th March, 2009

L2 Muon Trigger Study Status Report

  • Upload
    kishi

  • View
    51

  • Download
    0

Embed Size (px)

DESCRIPTION

L2 Muon Trigger Study Status Report. B. de la Cruz , C. Diez (CIEMAT, Madrid) Muon HLT Meeting 20th March, 2009. Motivation for the study. L2 Muon Trigger is already established but there may be still room for improvement optimizing the 2 step algorithm: Seeding - PowerPoint PPT Presentation

Citation preview

Page 1: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

L2 Muon Trigger Study Status Report

B. de la Cruz , C. Diez

(CIEMAT, Madrid)

Muon HLT Meeting 20th March, 2009

Page 2: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

L2 Muon Trigger is already established but there may be still room for improvement optimizing the 2 step algorithm:

Seeding Pattern Recognition in the muon system and the tuning of

the L2 parameters.

Motivation for the study

FirstOverview: http://indico.cern.ch/getFile.py/access?contribId=3&resId=0&materialId=slides&confId=52841

Where we reproduced validation histograms with official HLT Validation package (MultiTrackValidator Analyzer for residuals, pulls and efficiencies of track reconstruction) running on MC samples.

Page 3: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

According to Riccardo and J.Roch, several L2 optimizations possible:

Use of FastSteppingHelixPropagator (code implemented some time ago by Slava Krutelyov but never activated) instead of usual SteppingHelixPropagator, in the L2 pattern recognition and reconstruction (bigger extrapolation steps in L2 propagation)

Add a final refit of the L2 muon track, considering all the final hits selected after a pattern recognition inside-out (FW) and outside-in (BW)

In the case of Refitting, drop the outside-in (BW) pattern recognition, to save time

Use CMSSW_3_1_0_pre3 to test optimizations above and Z MonteCarlo sample:

/RelValZMM/CMSSW_3_1_0_pre3_STARTUP_30X_v1/GEN-SIM-DIGI-RAW-HLTDEBUG (9000 evts)

Results comparing distributions of several variables (pt, eta, phi), their pulls and resolutions will be shown here.

First Tests on L2MuonProducer

Page 4: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Tests on L2MuonProducer

Possible improvements

Usual SteppingHelix Propagator

(default config)

Fast SteppingHelix

Propagator

ReFit = False(default config)

Default configuration Test

FastSteppingHelix

ReFit = True Test Refit Test

FastSteppingHelix+ReFit

BW pattern reco = False

(ReFit = True)

Test pattern reco = False

(ReFit = True)

Test FastSteppingHelix

+Refit+NoBW

Options presented in previous slides tested alone and in combination

Page 5: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

In HLTrigger/Configuration/python/HLT_2E30_cff.py in L2MuonProducer, activate DoRefit = cms.bool( True)

adding as Parameters those from RecoMuon/StandAloneMuonProducer/python/StandAloneMuons_cfi.py

Use CMSSW_3_1_0_pre3 to test optimizations above and Z MonteCarlo sample:

/RelValZMM/CMSSW_3_1_0_pre3_STARTUP_30X_v1/GEN-SIM-DIGI-RAW-HLTDEBUG (9000 evts)

ReFit implemented in L2

Page 6: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Pt distributions

Removing BW pattern recognition, Pt distributionIs still “discrete”

Comparing with Pt L1 distribution:

L1 Pt values in 2 bins: Correction shift by 2 GeV in Barrel and 3GeV in EndCaps

Page 7: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

distributions

only for “standard” SteppingHelixPropagator (similar for FastStep.)

Removing BW pattern recognition affects to the values in the barrel: Is that the expected behaviour?

No mayor differences between distributions with/ without reffiting

Page 8: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Pt resolution

Refit improves resolution

(see next slide) by 3%

No BW filter: resolution

gets worse by 10% (respect

to the “original” L2 reco)

Implementing the

FastSteppingHelix

propagator does not the

resolution

Page 9: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Pt resolution

~20%

~17%

~31%

Page 10: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Pt resolution vs Pt &

Page 11: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Comment on L2 vs L1

Page 12: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

resolution

Changes in resolution smaller than in the case of Pt

Flat distributions in in all cases (not shown)

Page 13: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

resolution

Page 14: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

resolution

Resolution similar in all casesremoving BW patter recogition worsens the tail of the distributions, similar peak

Page 15: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

resolution

No changes with new propagator in resolution, refit doesn´t improves itH resolution with BW pattern recognition only slightly worse (~5% difference)

Page 16: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Pull distributions

mean

Page 17: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Pull distributions

Page 18: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Sigma :Pull distributions vs

Pulls do not change with the FastStepping Helix PropagatorFlatter distributions with Refit for distributions as a function of

Page 19: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Pull distributions vs

Pulls do not change with the FastStepping Helix PropagatorFlatter distributions with Refit for distributions as a function of

Page 20: L2 Muon Trigger Study  Status Report

21 Apr 2023 Muon HLT Meeting C. Diez

Comments & Outlook

The improvement on the “SteppingHelixPropagator ” does not worsen the resolution

Compare these set of validation plots to existing ones in http://cern.ch/cms-muonhltval (Jeff. Klukas).

The work will go on, paying attention to the different L2 parameters being used