Debug Mixed Data Cutflow
- Trying run on c++
- Container with:
> cmssw-el7 -p --bind `readlink $HOME` --bind `readlink -f ${HOME}/nobackup/` --bind /uscms_data --bind /cvmfs -- /bin/bash -l
> source setupHH4b.sh
- Commands to run the c++ on mixed analysis in:
debug_py.sh
nTupleAnalysis ZZ4b/nTupleAnalysis/scripts/nTupleAnalysis_cfg.py -i closureTests/ULTrig//fileLists/mixed2018_3bDvTMix4bDvT_wJCM_v0.txt --friends closureTests/ULTrig//fileLists/mixed2018_3bDvTMix4bDvT_wJCM_friends_3bDvTMix4bDvT_v0_SeedAll45.txt -o root://cmseos.fnal.gov//store/user/jda102/condor/ZH4b/ULTrig/ -p NONE -y 2018 --bTag 0.6 --histFile hists_wFvT_3bDvTMix4bDvT_v0_weights_newSBDefSeedAll45.root --histDetailLevel fourTag.passPreSel.passMjjOth.HHSR.bdtStudy --FvTName FvT_3bDvTMix4bDvT_v0_newSBDefSeedAve --unBlind --isDataMCMix --runKlBdt
- Need to uncomment SRevents lines
- make with scram b -j 5
- Dumping
py ZZ4b/nTupleAnalysis/scripts/dumpSRInfo.py --in root://cmseos.fnal.gov//store/user/jda102/condor/ZH4b/ULTrig/mixed2018_3bDvTMix4bDvT_wJCM_v0/hists_wFvT_3bDvTMix4bDvT_v0_weights_newSBDefSeedAll45.root --out SR_info_mixed2018_3bDvTMix4bDvT_wJCM_v0
- Added SvBMAps to the output
- Now checking the python
- Read in pkl files with
debugMixedData.py
./run_container python analysis/debugMixedData.py hists/histMixedData.coffea
- Running on py
./run_container python runner.py -o histMixedData.coffea -d mixeddata -p analysis/processors/processor_HH4b.py -y 2018 -op output/ -m metadata/datasets_HH4b.yml -c analysis/metadata/HH4b_nottcheck.yml
- Going back to the previous SvB files.
SvB_MA_file = f'{path}/SvB_MA_newSBDef.root' if 'mix' in self.dataset else f'{fname.replace("picoAOD", "SvB_MA")}'
- Read in the c++ and the python with
./run_container python analysis/debugMixedData.py output//histMixedData.coffea
- Start with:
N c++ events 21697
In c++, but not py 1613 frac 0.07
In py but not c++ 684 frac 0.03
- Not making PU cut in c++ brings us to
N c++ events 21697
In c++, but not py 708 frac 0.03
In py but not c++ 701 frac 0.03
- Residual effect probably from random numbers
- AE will try with this change