-- AmnonHarel - 12 Dec 2008



Testing Mike's DataMixer, version V00-01-18. Goal was to see a plot of HPD discharge noise overlayed near / over MC signal, such as this: sarm1.gif

To understand this event, look at the attached pure MC, overlayed data, and the HPD noise analysis of the data and the mixed MC. The HPD analyzed is the one chosen by my HPD discharge noise ID, which uses only the geometry of the HCal energy in eta-phi space. Here it picks the right HPD even in the mixed event.

Conclusions (on the overlaying process)

  • made a new reconstruction config python/recoMixed_cff.py (in the work area given below)
  • the overlayed MC is too noisy: where did the 10s of GeV deposits around myiEta=+/-30 (i.e. +/-28 in standard iEta) come from?
    • myIEta is motivated in e.g. slide 5 (+ 6 & 29) of this HPD discharge ID talk. HB is in 1-16, HE is shifted by 2, with depth 1 of |iEta|=29 shifted by an extra place, so it's 18-32.
  • mostly things I learned about CMSSW, configs, etc. For example:
    • when working with intermediate files, take care to actually store the digis. I.e., choose the appropriate event content
  • for now need to read reco digis rather than sim digis (see hacked python/mixOne_data_on_sim_cfi.py)
  • improve data mixer's error and warning messages
    • diffrentiate between "pileup", "source", and "total" logs (e.g. #HBHE hits from each)
    • issue errors (or at least warnings) when expected input is missing in the event
    • can NDEBUG (see http://www.uscms.org/LPC/lpc_offl/MessageLogger/debug.html#ML_NDEBUG) be used instead of DEBUG, simplifying the code a bit (and be undefined for now, allowing user control without recompiling smile )?
  • don't forget to reconstruction on the new digis
  • to read in the mixed digis, change the InputTag (just a label is not enough) from the usual HBHEDigis to mix:HBHEDigiCollectionDM

How I got there

  • work area: /uscms/home/aharel/CMSSW_2_1_10/src/SimGeneral/DataMixingModule
  • data area: contains both .log and .root files /uscms_data/d1/aharel
  • MC events:
    • used copyRawQCDMC.py to pick up some events from summer08 dijet sample (80-120GEV).
      • /store/mc/Summer08/QCDDiJetPt80to120/GEN-SIM-RAW/IDEAL_V9_v1/0000/06A708D1-FD96-DD11-A79A-0015C5EC47A2.root --> rawQCDMCevents.root
    • used prepMC.py to add digi and (to ease comparison plotting) reco to it. It also dumps some digi for my plots with sc.c.
      • --> preparedQCDMCevents.root
      • plotting uses "AHtest" to be placed in RecoJets/JetAnalyzers. Tiny update needed for JetToDigiDump for package to compile.
      • dump for plots in prepMC.root
  • Non-collision data events:
    • used copyRaw57795.py to pick up some events from cruzet 4 run 57797.
      • both HCal and muon triggers
      • /store/data/Commissioning08/Cosmics/RAW/CRUZET4_v1/000/057/795/083559CB-B96E-DD11-B611-001617E30D06.root --> raw57795events.root
    • used prep57795.py to add digi and (to ease comparison plotting) reco to it. It also dumps some digi for my plots with sc.c.
      • --> prepared57795events.root
      • dump for plots in prep57795.root
      • set all threshold to 0 so all plotting information is available for no-collision data
      • looking at the plots: events 20 & 25 have in-time HPD discharges (see attachments)
      • BTW: first run with an old tag (CRUZET4_V2P::All, instead of with CRUZET4_V4P::All), had no effects on plots (identical fC, MET, E_HCal)
      • Started writing out all the branches (see below). Can figure out more precise treatment according to SWGuideSelectingBranchesForOutput & JetMETSamples#Format
  • Mixed events:
    • used test/mm_deb_overlay_cff.py and python/mixOne_data_on_sim_cfi.py (latter included in former) --> /uscms_data/d1/aharel/MixTest.root
    • re-reconstructed with recoMixTest.py --> recoMixTest.root, recoMixTest.log
      • turns out Mike had a couple of similar configs, the best being python/Reco_from_DataMix_Digis_CalRecHits_cff.py. But they start from calorimetry rechits, and have a few mistakes.
      • I only got the python to work by placing a copy of the standard reconstruction inside my python/recoMixed_cff.py. Got to be a better way...
    • dump for plots with dumpMix.py --> mix.root
    • added debug prints and warning messages to DataMixingHcalDigiWorker.cc
      • turns out that getByLabel (hcalDigis, HBHEDigiCollection) failed (for the pileup). But it is looking in my file (adding pileups from event  run: 57795...) ---> Fixed prep57795.py which didn't store the digis (see above) Same problem with prepMC.py and the signal...
    • studying output -
      • mixer expected InputTag "simHcalDigis", which I don't have despite using keep * in prepMC. How can I get them?
        • for now, hacked the mixer configs to use the usual reco digis instead (no "sim") in name.
      • first attempts to run recoMixTest.py failed on missing non-calorimetry branches, but it works now that the mixer gets the right calorimetry inputs. Not sure how the two are connected.
      • outdated note: the expected problem rears its head: reco still uses the old inputs --> same recHits (though I see the new merged digis smile compare sxmix4 and sxmc4). How can we get reco to read the "DM" digis?
      • another odd problem: even in the MC input, I see recHits (with E>threshold) that I can't match any digis to. E.g. compare samc4 & sxmx4
Topic attachments
I Attachment History Action Size Date Who Comment
GIFgif samc1.gif r1 manage 16.0 K 2009-01-01 - 22:23 AmnonHarel HCal energy plot of 1st event in pure MC
GIFgif samc4.gif r1 manage 15.8 K 2008-12-30 - 16:33 AmnonHarel HCal energy plot of 4th event in pure MC
GIFgif saod21.gif r1 manage 15.7 K 2008-12-14 - 17:15 AmnonHarel HCal energy plot of 21th data (CRuZeT) event
GIFgif saod26.gif r1 manage 15.4 K 2008-12-14 - 17:14 AmnonHarel HCal energy plot of 26th data (CRuZeT) event
GIFgif sarm1.gif r1 manage 35.3 K 2009-01-01 - 22:25 AmnonHarel HCal energy plot of 1st mixed event = 1st MC + 21 data
PNGpng sxmc1.png r1 manage 14.7 K 2009-01-01 - 22:22 AmnonHarel HPD analysis of 1st event in pure MC
PNGpng sxmc4.png r1 manage 34.8 K 2008-12-30 - 16:29 AmnonHarel HPD analysis of 4th event in pure MC
PNGpng sxmix4.png r1 manage 35.4 K 2008-12-30 - 16:28 AmnonHarel HPD analysis of 4th event in mixer output
PNGpng sxod21.png r1 manage 47.3 K 2008-12-14 - 17:13 AmnonHarel HPD analysis of 21th data (CRuZeT) event
PNGpng sxod26.png r1 manage 35.1 K 2008-12-14 - 17:13 AmnonHarel HPD analysis of 26th data (CRuZeT) event
PNGpng sxrm1.png r1 manage 51.3 K 2009-01-01 - 22:26 AmnonHarel HPD analysis of 1st mixed event = 1st MC + 21 data
Edit | Attach | Watch | Print version | History: r26 < r25 < r24 < r23 < r22 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r26 - 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 & 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback