cms-kr / tzwi Goto Github PK

• 추출된 histogram들을 combine tool에 입력.

We need to change the 2016 MC tune soon: CUETP8M2T4 (CUETP8M1) -> CP5.
(Maybe... in this year?)

From the PAG meeting on 2020 Feb. 5th,
https://indico.cern.ch/event/885400/contributions/3732106/attachments/1982229/3301574/topSyst_05022020.pdf

Top systematic page recommended changing the 2016 MC tune from the old to the new one.

In the topSystematics twiki page,
https://twiki.cern.ch/twiki/bin/viewauth/CMS/TopSystematics#Parton_shower_Color_Reconnection

They note about the tune:
"If you are using the CUETP8M2T4 (CUETP8M1) tune for the top quark pair (single top) production in the 2016 MC, you should consider switching to the MC samples with the CP5 tune."

• Reference about comparing to CUETP8M1 and CP5 in tX meeting, single-top s-channel production:
  https://indico.cern.ch/event/881651/contributions/3714874/attachments/1977804/3292371/tX_2020-01-29_s-channel.pdf

Synchronization study

• cross section과 numEvents에 맞춰 scale해서 그림
• 각종 syst. 를 고려해 syst. band그림
• 최종 결과에 사용될 plot 몇 개 선정해 histogram들과 syst band 추출

• DY scale factor추출
• QCD scale factor추출

Estimate computing resources to produce ntuples

• CPU time (RD2017, ttbarDoubleLepton) = 2.9kHz
• Disk space (RD2017, ttbarDoubleLepton friend tree) = 70MB
  • Input file size = 1.5GB

Rough estimation: 1.5GB/970s => (388GB/70h for SingleMuon dataset)

Conclusion: Submit condor jobs to produce ntuples

voms-proxy 없이 사용하면 빈 디텍토리만 만들어집니다.

반드시 DBS 질의를 위해 proxy 인증서가 필요합니다만, 없어도 에러가 발생하지 않습니다.

Finalize ntuple structure for the dilepton analysis

• Dilepton "good" physics objects branches
• Combined trigger bits
• Combined MET filter flags
• Cut flow bits (or step)
• Gen level top quarks and additional jets
• nVertex
• PU weights
• b tag weights
• lepton efficiency scale factors
• HLT scale factors
• gen weight, pdf weights, scale weights, etc

NanoAOD production step

Some failures during ntupling.

Traceback (most recent call last):
  File "/cms/ldap_home/jhgoh/work/CMS/NanoAOD/CMSSW_10_2_11/bin/slc6_amd64_gcc700/nano_postproc.py", line 51, in <module>
    p.run()
  File "/cms/ldap_home/jhgoh/work/CMS/NanoAOD/CMSSW_10_2_11/python/PhysicsTools/NanoAODTools/postprocessing/framework/postprocessor.py", line 131, in run
    (nall, npass, timeLoop) = eventLoop(self.modules, inFile, outFile, inTree, outTree)
  File "/cms/ldap_home/jhgoh/work/CMS/NanoAOD/CMSSW_10_2_11/python/PhysicsTools/NanoAODTools/postprocessing/framework/eventloop.py", line 48, in eventLoop
    m.beginFile(inputFile, outputFile, inputTree, wrappedOutputTree)
  File "/cms/ldap_home/jhgoh/work/CMS/NanoAOD/CMSSW_10_2_11/python/TZWi/TopAnalysis/postprocessing/CombineHLT.py", line 62, in beginFile
    self.initReaders(inputTree)
  File "/cms/ldap_home/jhgoh/work/CMS/NanoAOD/CMSSW_10_2_11/python/TZWi/TopAnalysis/postprocessing/CombineHLT.py", line 69, in initReaders
    setattr(self, "b_"+name, tree.valueReader(name))
  File "/cms/ldap_home/jhgoh/work/CMS/NanoAOD/CMSSW_10_2_11/python/PhysicsTools/NanoAODTools/postprocessing/framework/treeReaderArrayTools.py", line 38, in getValueReader
    if not tree.GetBranch(branchName): raise RuntimeError, "Can't find branch '%s'" % branchName
RuntimeError: Can't find branch 'HLT_Mu23_TrkIsoVVL_Ele12_CaloIdL_TrackIdL_IsoVL'

Synchronization study

Complete developing ntuple production step

• TopAnalysis/test/토픽/prod_ntuple에서 각 샘플, 파일별로 run.sh 실행

• 적당한 /xrootd 유저 디렉토리에 저장해서 위치 공유하기

• Apply lumi JSON file

• reco level postprocessor: object selection

• reco level postprocessor: event selection / cut flow

• gen level postprocessor: gen level object selector

• gen weights

• puweight + systematics

• JES/JER on the fly + systematics

• lepton SF event weight + systematics

• b tag SF event weight + systematics

• nEvent histogram

• 각 cut flow별 정의된 histogram들 만들기 cross section normalization은 여기서는 하지 않음
• num event로 normalization도 하지 않음
• 각종 event weight는 곱한 상태
• systematics를 위한 다양한 조합별로 histogram들을 모두 생성