-- AmnonHarel - 24 Feb 2009

Introduction

See index in AmnonHarelDataMixing

Testing Mike's DataMixer, circa 20-26Feb, with his "V4a", "V4b", "V5", "V6", & "V7" files. Goal was to see a plot of HPD discharge noise overlayed near / over MC signal. Like this overlayed event = [data event + MC QCD event, where the plots show the HCal energy in recHits (endcaps are displaced two iEtas towards the edges). At the digi level, the most interesting is HB-, iPhi=14 overlay = data + little form MC (no recHits above threshold there).

  • Caveat: data plots show all recHits, above CRAFT threhsolds, MC and Mixed plots only show recHits above planned collision thresholds (0.9 & 1.4GeV in HB & HE).

Conclusions (on the overlaying process, and reconstruction)

  • Reconstruction of mixed events is similar to what I did in AmnonHarelDataMixingRound2, but needed new InputTag-s for the HCal digis ("mix")
    • see /uscms/home/aharel/CMSSW_3_1_0_pre1/src/SimGeneral/DataMixingModule/python/recoMixed_310pre1_cff.py
  • Mike and Rick found a bug in the code that produced "V4a", hence "V4b".
  • energies in V4b too low, hence V5
  • V5 is swamped with fake energy. E.g. event 306 and some of its digis
    • same with V6
    • Rick found a bug that starts at event #2 (I was looking at #300). Hence V7
  • V7 looks reasonable smile
    • but why are the overlayed digis sometimes lower than the data inputs? e.g.
      • leading digi in example above
      • saw other cases with ~2 GeV (e.g. data event 2331132 = MC # 360, ~2GeV at iEta=-27 --> shown as -29 in my plots, iPhi=69)
      • Rick & Mike: so far all can be explained by the different zero suppresion in data and MC reconstruction
    • Need to examine examples with high Es being added up
      • Can we made non-random overlay to produce such overlaps (suggested patricular data & MC events to Mike)

Details

  • work area: /uscms/home/aharel/CMSSW_3_1_0_pre/src/RecoJets/JetAnalyzers
  • data area: contains both .log and .root files /uscms_data/d1/aharel
  • V4a MC events: /uscms_data/d1/mikeh/QCD_Pt_50_80_cfi_GEN_SIM_DIGI_L1_DIGI2RAW_NoNoise.root
    • examined using gdQCD4new.py --> gdQCD4.newlog * this file contains calorimeter digis, but no reco. calo jets
    • raw2digi, reco & translate to my sc.c format using prepQCD4.py --> prepQCD4.log, prepQCD4events.root (in CMSSW format) & prepQCD4.root (in my format)
      • see comments about raw2digi below
  • Non-collision data events: same as "V4" (but for real this time) /uscms_data/d1/mikeh/promptRecoHCalNoise.root (from run 68021)
    • examined using gdData4.py --> gdData4.log
    • this file contains calorimeter digis, but no reco. calo jets
    • raw2digi, reco & translate to my sc.c format using prepData4.py --> prepData4.log, prepData4event.root (in CMSSW format) & prepData4.root (in my format)
      • Finally update to a CRAFT global tag: CRAFT_30X::All. Still pretending the magnetic field is 0 though (don't really care).
      • Looking for anomolous events:
        • #14 (i.e. 645870, on MC #306) looks odd at HB-, 70. Very long low signal (like RBX) but only a few channels in the same HPD.
        • #28 (i.e. 1082063, on MC #320) is notable for having a very short energy deposit (confirmed by ECal, so probably a malformed muon deposit)
        • #364 is the most HPD-discharge like, and there's also no ECal energy. Yet, it just looks like a muon hitting HCal.
        • all the rest look like uncorrelated noise + muons.
        • #46 (i.e. 1649543, on MC #339) has a ~40GeV hit.
      • Not sure why the raw2digi step is needed. Without it complains about:
              Looking for type: edm::SortedCollection<ZDCDataFrame,edm::StrictWeakOrdering<ZDCDataFrame> >
              Looking for module label: hcalDigis
              Looking for productInstanceName:
* Overlayed events "V7": Reco version: /uscms_data/d1/mikeh/CRAFT_on_QCD_NewHcal_RECO_v7.root
    • examined using gdO7.py --> gdO57log
      • contains both mixed digis and recHits smile
    • dump in my sc.c format for plots with dumpMix7.py --> dumpMix7.root

  • Overlayed events "V6": Reco version: /uscms_data/d1/mikeh/CRAFT_on_QCD_NewHcal_RECO_v6.root
    • examined using gdO6.py --> gdO6.log
      • File is corrupt

  • Overlayed events "V6": Digis version: /uscms_data/d1/mikeh/CRAFT_on_QCD_NewHcal_Digis_v6.root
    • examined using gdOD6.py --> gdOD6.log
        • as promised, contains mixer digis but no reconstruction
      • reconstructed events 300-330 using reco6.py --> reco6.log, reco6Events.root
      • dump in my sc.c format for plots with dumpMix6.py --> dumpMix6.root
        • saw same problem as with V5

  • Overlayed events "V5": Reco version: /uscms_data/d1/mikeh/CRAFT_on_QCD_NewHcal_RECO_v5.root
    • examined using gdO5.py --> gdO5.log
      • contains both mixed digis and recHits smile
    • dump in my sc.c format for plots with dumpMix5.py --> dumpMix5.root

  • Overlayed events "V4b": Reco version: /uscms_data/d1/mikeh/CRAFT_on_QCD_NewHcal_RECO_v4b_sub.root
    • examined using gdO4bv1.py --> gdO4bv1.log
      • contains both mixed digis and recHits smile
    • dump in my sc.c format for plots with dumpMix4b.py --> dumpMix4b.root
      • did not see the data (except perhaps with much reduced energies, which makes sense in retrospect given Rick's resulting bug fix)

  • Overlayed events "V4a": Reco version: /uscms_data/d1/mikeh/CRAFT_on_QCD_NewHcal_RECO_v4a.root
    • examined using gdO4aReco.py --> gdO4aReco.log
      • this file is corrupt? edm::RootTree (in setCacheSize, called in constructor) crashes.
        • See below on why that happened
        • Mike's remaking it for 100 events
    • "processed to Digi stage": /uscms_data/d1/mikeh/CRAFT_on_QCD_NewHcal_Digis_v4.root
      • examined using gdO4aDigi.py --> gdO4aDigi.log
        • as promised, contains mixer digis but no reconstruction
      • reconstructed using reco4a.py --> reco4a.log, reco4aEvents.root
        • this crashes on event 132 in CkfTrackCandidateMaker
        • I rerun on events 300-330, where my interesting examples are.
      • dump in my sc.c format for plots with dumpMix4a.py --> dumpMix4a.root
        • updated DigiInputTag, it's now "mix"
    • who's where: /uscms_data/d1/mikeh/Craft_on_QCD_NewHcal_v4a_evlist.txt
Topic attachments
I Attachment History Action Size Date Who Comment
PowerPointppt overlay1.ppt r2 r1 manage 646.0 K 2009-02-25 - 18:57 AmnonHarel Talk draft
GIFgif sad4_14.gif r1 manage 17.7 K 2009-03-01 - 10:47 AmnonHarel Data event #14 - energy scan
GIFgif sad7_28.gif r1 manage 17.6 K 2009-03-08 - 15:51 AmnonHarel Data event #28, hcal energy in recHits
GIFgif sad7_68.gif r1 manage 16.6 K 2009-03-09 - 21:44 AmnonHarel Data event #68 - hcal energy in recHits
GIFgif sam5_156.gif r1 manage 45.2 K 2009-03-01 - 10:48 AmnonHarel Mixed event #306 - V5 - hcal energy in recHits
GIFgif sam7_320.gif r1 manage 16.7 K 2009-03-08 - 16:13 AmnonHarel Mixed event #320, hcal energy in recHits
GIFgif sam7_360.gif r1 manage 16.3 K 2009-03-09 - 21:45 AmnonHarel Mixed event #360 - hcal energy in recHits
GIFgif saq7_320.gif r1 manage 16.5 K 2009-03-08 - 16:12 AmnonHarel QCD event #320, hcal energy in recHits
PNGpng sxd4_14.png r1 manage 37.5 K 2009-03-01 - 10:49 AmnonHarel Data event #14 - selected digis
PNGpng sxd7_28h214.png r1 manage 36.2 K 2009-03-08 - 15:50 AmnonHarel Data event #28, digis at HB-, iphi=14
PNGpng sxd7_28h349.png r1 manage 26.2 K 2009-03-08 - 15:50 AmnonHarel Data event #28 - digis at HB+, iphi=49
PNGpng sxm5_156.png r2 r1 manage 99.8 K 2009-03-01 - 10:53 AmnonHarel Mixed event #306 - V5 - selected digis
PNGpng sxm7_320h214.png r1 manage 37.7 K 2009-03-08 - 16:16 AmnonHarel Mixed event #320, digis at HB-, iphi=14
PNGpng sxm7_320h349.png r1 manage 34.3 K 2009-03-08 - 16:18 AmnonHarel Mixed event #320 - digis at HB+, iphi=49
Edit | Attach | Watch | Print version | History: r15 < r14 < r13 < r12 < r11 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r15 - 2009-08-12 - AmnonHarel
 
    • 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