-- AndreasKorn - 16 Apr 2009

Validation Samples

A small 10k cosmic sample has been simulated and digitized with release AtlasProduction 14.5.1.4 and reconstructed with 14.5.2.Y_rel_5 plus fixes. The samples make use of the new TRT selection volume that includes the endcap region. Geometry tag ATLAS-GEO-07-00-00 with field on was used. The produced RDO & ESD files are available here:

/castor/cern.ch/grid/atlas/atlasgroupdisk/proj-simcos/rel14/ATLAS-GEO-07-00-00
Commissioning ntuples are also availble:
/afs/cern.ch/user/g/gencomm/w0/korn/ESD2/[1-9]*/myCommission*root
List of Fixes (Recipe)
See Jamies Page here:
cmt co -r TRT_DriftFunctionTool-00-01-27-01 InnerDetector/InDetRecTools/TRT_DriftFunctionTool
cmt co -r SiLorentzAngleSvc-00-00-07-02 InnerDetector/InDetConditions/SiLorentzAngleSvc  
cmt co -r RecExCommission-00-04-00-02  Reconstruction/RecExample/RecExCommission
cmt co -r TrackIsolationTools-00-05-05-01  Reconstruction/RecoTools/TrackIsolationTools 
cmt co -r MuonIsolationTools-00-07-15-01 Reconstruction/MuonIdentification/MuonIsolationTools
cmt co -r InDetRecExample-01-16-07-15 InnerDetector/InDetExample/InDetRecExample  
cmt co -r TrigT1CaloMonitoring-00-04-10-01 /Trigger/TrigT1/TrigT1CaloMonitoring
cmt co -r TrigT1Monitoring-00-00-22-01 /Trigger/TrigT1/TrigT1Monitoring  
cmt co -r TileEvent-00-08-12 /TileCalorimeter/TileEvent
cmt co -r TileEventAthenaPool-00-04-07 /TileCalorimeter/TileSvc/TileEventAthenaPool

Validation Monitoring Plots

DQMF monitoring plots have been produced:

a good data run using the new reprocessing for comparison can be found here

The reconstruction has been retested with 14.5.2.10 plus fixes (to become 14.5.2.11):

The first produced digits (no field, Pixel volume: ) have been reconstructed with 14.5.2.10: a possible nofield data run using the new reprocessing for comparison can be found here

Official Production Samples

Monitoring is available for the official production samples. As reference the old official MC and the validation MC have been used. Data reference are from runs 91890 and 96903.

The new larger size IDVol samples
Files used:

689 for valid2.108867.CosSimIDVolSolOnTorOn.recon.HIST.s540_d167_r676_tid067157
693 for valid2.108866.CosSimIDVolSolOffTorOff.recon.HIST.s541_d168_r677_tid067158

Latest Production Sample, please use this!

Older Reconstruction Interation, please don't use!

Technical Details

Detailed Validation Comparisons

Tae Jeong and I looked through the produced plots and contacted experts. An e-mail was send to tracking, muon and e-gamma performance convenors to ask for further comparisons. Here we list observed differences between old and new MC with possible explanations. Old MC refers to the ATLAS_GEO_03_00_00 TRT Barrel sample generated by Else and Jamie. Both MC samples were reconstructed with 14.5.2.Y_rel_5. When appropriate we also list comparisons between distributions in new MC and data.

Pixel

Pixel Clusters
The number of pixel clusters is reduced and there are more larger clusters in the new MC. This is the expected behavior due to a fix in digitization, that spreads charge more evenly.

Pixel Timing
The behavior in the old MC is not understood. In the new MC a propper peak as in data is observed.

TRT

TRT hits on track
There is a tendency towards fewer TRT hits on tracks in the new vs the old MC. This is not fully explained yet, but thought to be related to the inclusion of the TRT endcap volume in filtering.

TRT occupancy
When comparing new with old MC, the TRT high occupancy is down for the barrel and up for the Endcap. This is likely due to the different track illumination caused by the larger filter volume that includes TRT Endcaps. Low threshold occupancy is dominated by noise and does not show this effect. Barrel A Barrel C EndCap A EndCap C

TRT rt relation
The rt relation seems consistent with xenon.

TRT eventphase
A difference between MC and data is observed in the TRT eventphase. But it agrees with when comparing with the old MC.

Explanation by Sasa and James: EP distribution is not simulated in MC, all tracks come at the same time within 3BC and the spread is only due to EP resolution. The actual value does not matter so much as long as LE && TE distributions are reasonable?

Muons

Muon phi reconstruction
Changes in the phi distribution with respect to older reconstruction is due to an updated definition in the muon reco. The same can be observed for the new data reprocessing.

Muon number of chambers attached to tracks
The number of muon chambers attached to the track is larger in the new MC than in data. This is due to lower HV settings and chambers timing problems in the data.

MC data MC data

This effect is a bit more clearer in the RPC plots here:

MC data

MC data

MC data

MC data

Muon Chamber MDT residuals

Maria showed here a widening of MDT residuals. I had a closer look and there seems to be also a problem with the drift radius I suspected a t0 problem, Domizia explains, that t0 refinement has likely not been run for the simulation.

MET

There seems to be a change in MET reco as pointed out by Hideki. This is due to the new muon angle definitions that is not propagated to th MET reco yet. But the distributions seem to agree between old, new MC and reprocessed data.

TJ pointed out a difference between data and MC:. This is likely a selection effect due tracks away fro the interaction point, but needs to better understood.

magnetic field

Domizia calculated the bending power and finds a disagreement between MC and data (3.2 Tm s 4.4 Tm and 4.6 Tm expected). This could also explain differences in the angular distributions seen in the ID cosmics (taking shafts etc into account). Checks are underway to see if this could be cured with the new GEO-06 field map ...

Validation Analysis

A few selected people have been asked directly to test there analysis on the new sample. A mail has been send tracking, muon and e-gamma performance convenors to ask for further comparisons. If you have contributions, please add them here!!!
  • extrapolation of tracks for validation MC
    toppos_MCnew2.gif
  • extrapolation of tracks for official dataset valid2.108867.CosSimIDVolSolOnTorOn.recon.ESD.s533_d167_r676
    toppos_MC_108867.gif
  • From Tommaso: Pixel cluster size vs incidence angle on the module. The agreement with cosmics data is quite good.
    LorentzDataVsSim26may.eps LorentzDataVsSim26may-1.gif

Known Problems

  • possible event number overlap, due to bug in event offset setting

Release Tests

A variety of test were run to check fixes and qualify releases.

Simulation/Digitisation

  • 14.5.1.4
  • 14.5.1.5 (XML trigger database problems)
  • 14.5.1.6

Reconstruction

  • 14.5.2.Y_rel_5 plus fixes:
  • 14.5.2.Y-VAL_rel_3
  • 14.5.2.Y-VAL_rel_4
  • 14.5.2.Y_rel_5
  • 14.5.2.10 (rare crashes in muon & egamma reco)
  • 14.5.2.10 plus fixes:
  • 14.5.2.11 (rare crashes in muon reco, calo entry layer not stored)
  • 14.5.2.11 plus fixes:

-- AndreasKorn - 16 Apr 2009 -- AndreasKorn - 28 Apr 2009 -- AndreasKorn - 15 May 2009

Topic attachments
I Attachment History Action Size Date Who Comment
GIFgif LorentzDataVsSim26may-1.gif r1 manage 73.9 K 2009-06-02 - 18:02 AndreasKorn  
Unknown file formateps LorentzDataVsSim26may.eps r1 manage 24.6 K 2009-05-31 - 17:06 TommasoLari  
GIFgif toppos_MC_108867.gif r1 manage 47.1 K 2009-05-15 - 13:02 AndreasKorn  
GIFgif toppos_MCnew2.gif r1 manage 21.2 K 2009-04-16 - 13:51 AndreasKorn  
Edit | Attach | Watch | Print version | History: r15 < r14 < r13 < r12 < r11 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r15 - 2009-06-02 - AndreasKorn
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Main All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback