Date post: | 13-Jan-2016 |
Category: |
Documents |
Upload: | randolf-phelps |
View: | 213 times |
Download: | 0 times |
73X Validation, Feb. 23rd, 2015 Robert Schöfbeck, Mariarosaria D’Alfonso, Matthias Weber
report on 73X rereco Feb. 23th, 2015
Robert Schoefbeck, Mariarosaria D’Alfonso, Matthias Weber
2
73X Validation, Feb. 23rd, 2015 Robert Schöfbeck, Mariarosaria D’Alfonso, Matthias Weber
8 TeV samples
o 1M JetHT, 1.6M zMu skim of DoubleMu Run2012D• rerecoed with HCAL method 2 in 7_3_2_patch1
o DAS linkhttps://cmsweb.cern.ch/das/request?view=plain&limit=10&instance=prod%2Fglobal&input=dataset%3D%2F*%2F*HcalExtValid*%2F*
3
73X Validation, Feb. 23rd, 2015 Robert Schöfbeck, Mariarosaria D’Alfonso, Matthias Weber
o study MET resolution in Z to μμo more plots:https://dalfonso.web.cern.ch/dalfonso/MET_73X_validation/
o resolution (RMS/scale) comparable between 53X and 73X
o ~6-10% lower scale attributed to calorimetry changeso not the final PF calibration, small trend in tkMET under study
o more plots: https://dalfonso.web.cern.ch/dalfonso/MET_73X_validation/
DoubleMu zSkim rereco’d
53X
73X
4
73X Validation, Feb. 23rd, 2015 Robert Schöfbeck, Mariarosaria D’Alfonso, Matthias Weber
MET tail: JetHT 53X vs. 73X
o picked run 203835o comparing 73X JetHT HcalExtValid v2 RECO against
53X Jan22 rereco in AODo MET filters applied in both 53X and 73Xo JSON applied: gives ~6k eventso these data are available on EOS in 53X and 73Xo aim of the study: identify sumET and MET outliers
• also for sub-sums according to pfCand species
5
MET and sum(ET) scatter plots
immediate observations:1. seemingly smallish correlation in bulk MET region (blue arrow)2. different events in the MET tails (red arrows)3. sum(ET) relatively well under control (right plot)
6
o look at φ(MET) in order to understand theseemingly small correlation in bulk MET
o Conclusion: correlation is OK, looks as expected
MET phi
MET(53X)>50all events
7
o What is the energy composition of events in the tails?o First, disentangle pfCandidate species and look at sum(Pt)
and sub-MET scatter plots
o While there are sum(ET) outliers from h0, the MET looks under control. Suggest DPGs investigatea few off-diagonal events.
MET outliers from neutrals
neutrals‘h0’
8
o What is the energy composition of events in the tails?o charged sumPt much higher (as usual)
o sum(Pt) relatively well under control, few outliers in MET, suggest DPGs investigate a few off-diagonal events.
MET outliers from charged
charged‘h’
9
o What is the energy composition of events in the tails?o HF (here showing charged component) seems fine
o No issues observed in HF whatsoever
MET outliers from HF
‘h_HF’
10
MET outliers from gammas/e
‘gamma’
‘e’
same eventsas on p3
Energy is going back and forth between e and gamma. Correlated with MET outliers.
11
event lists
o created lists of outliers in MET and sumPt for all species: http://www.hephy.at/schoefbeck/etc/eventlist.txt
(open file w/o line wrap)o example: outliers in total MET
(c/p the event numbers from the text file)
12
debugging MET tails
Quoting always “53X→73X”1. pfMET 28.1→421.7: sumPt(h0) 142.0→46.0 (MET(h0) small), sumPt(gamma)
250→570, MET(gamma): 25.6 →424.42. pfMET 11.7→308.1 (similar characteristics), MET(gamma): 12.8→309.93. pfMET 42.0→211.6; MET(h) 131.5→208.6, MET(h0) 125.2→37.7,MET(gamma)
241.0→50.9sumPt(h)1063.3→1353.4
4. pfMET 4.7→174.3 (h ~ unchanged, MET(h0) unchanged, small) , MET(gamma) 12.3→154.7
5. pfMET 84.4→219.8 MET(h) 117.1→201.0, MET(mu) 140→06. pfMET 66.8→191.6 (?) 200 GeV change in sumPt(h),
several smaller changes7. pfMET 104.2→216.0
MET(gamma) 20.1→142.58. pfMET 12.9→121.5
MET(gamma) 21.8→139.49. pfMET 34.8→136.1 small changes
in h0, h, gamma are adding up10.pfMET 137.9→33.2, MET(h) 240.0→90.311.pfMET 197.1→43.8, MET(h) 262.8→73.112.pfMET 167.1→12.6, MET(h) 24.6→223.413.pfMET 178.3→8.7, MET(gamma) 145.9→43.7, MET(mu) 47.2→0
1.2.3.4.5.6.7.8.9.10.11.12.13.
73X>
53X73X<
53X
event numbers of prominent outliers
13
o Energy going back and forth between e and gamma.This feature is related to the most prominent outliersin the total MET.
o one event (208352:15:20368765) with a 140 GeV muon apparently lost in 73X (seems to create MET)
o several events have significantly less MET in 73X (seems related to charged hadrons ‘h’)
o retrieved 73X outliers on next slide for further study in cmsShow etc./eos/cms/store/group/phys_jetmet/schoef/pickEvents/73X-RECO-pickEvents
more on MET tails
14
list of events with large MET(gamma)
o top half: more MET(gamma) in 73X. Up to 400 GeV difference.
• picture is similar for sumPt and for electrons
o bottom half: less MET(gamma) in 73X. Up to 230 GeV diff.
o go here to c/p list: http://www.hephy.at/schoefbeck/etc/eventlist.txt
15
Summary
o Spotted several differences in the 53X and 73XMET tails related to e/gamma, mu, ho apparently less issues with h0, no problems with
HF
o event lists are ready for DPGs to study
o MET scale and resolution seem under control
16
Backup
17
73X Validation, Feb. 23rd, 2015 Robert Schöfbeck, Mariarosaria D’Alfonso, Matthias Weber
sanity check: DQM
o compare with 71X relvals and compare broad characteristics of reconstruction methods.
o reference sample: /JetHT/CMSSW_7_1_0-GR_R_71_V6_RelVal_jet2012D-v1/DQMExpect to see changes related to:o timing HCAL and ECAL o PFcalibration (hadrons and egamma)
o link to central DQM GUI (thanks to Matthias!!)https://cmsweb.cern.ch/dqm/relval/start?runnr=208307;dataset=/JetHT/CMSSW_7_3_2_patch1-GR_R_73_V0_HcalExtValid_RelVal_jet2012D-v1/DQMIO;sampletype=offline_data;filter=all;referencepos=overlay;referenceshow=all;referenceobj1=other::/JetHT/CMSSW_7_1_0-GR_R_71_V6_RelVal_jet2012D-v1/DQM:;referenceobj2=none;referenceobj3=none;referenceobj4=none;search=;striptype=object;stripruns=;stripaxis=run;stripomit=none;workspace=Everything;size=M;root=JetMET/MET/pfMet/Cleaned;focus=JetMET/MET/pfMet/Cleaned/PfNeutralHadronEt;zoom=no;
18
sanity check vs. 71XDQM/MET
JetHT run 208307HcalExtValid
JetHT run 208307CMSSW_7_1_0-GR_R_71_V6_RelVal
19
73X Validation, Feb. 23rd, 2015 Robert Schöfbeck, Mariarosaria D’Alfonso, Matthias Weber
o MET agrees within stat. MET tail not worrisome o plot on previous slide is for
a dijet selectiono lower photon ET (as expected)o lower neutral ET (as expected)o higher HF hadron ET
o was this expected?o sumET reduced by ~1.5%o other fractions vary consistently
sanity check vs. 71XDQM/MET
20
73X Validation, Feb. 23rd, 2015 Robert Schöfbeck, Mariarosaria D’Alfonso, Matthias Weber
JetHT run 208307HcalExtValid
JetHT run 208307CMSSW_7_1_0-GR_R_71_V6_RelVal
sanity check vs. 71XDQM/Jets (AK4PF)
o lower neutral hadron energyo higher HF energyo improvement of eta ‘horns’
21
jetHT rereco’d
o looking for spectacular mis-recoo comparing: 1. HLT_HT750 triggered data
2. applying recommended MET filters
3. applying offline HT + dijet requirement
filters removehigh MET noise
noiseremoved
22
jetHT rereco’d
o left: pT of leading jeto middle: neutral had. e.f. (nhef) of leading jeto right: max (nhef) per event for all jets > 100 GeV
o Summary: HCAL noise effectively removed, no signof residual noise. No hints of unforseen effects found. Note: This study is not sensitive to % level effects in calo reco
o more plots: http://www.hephy.at/schoefbeck/pngHCAL/
23
73X Validation, Feb. 23rd, 2015 Robert Schöfbeck, Mariarosaria D’Alfonso, Matthias Weber
Conclusion
o Checked JetHT and DoubleMu zSkim rereco’d data
o JetHTo Nothing worrying found, observed changes in line
with reconstruction
o Double Mu zSkim Run2012Do MET resolution comparableo MET scale 6-10% lower in 73X