/PhaseTwoAnalysis

Primary LanguageC++

Repository for collecting recipes for standard physics objects for analysis of simulated events with the CMS phase 2 detector.

Contains (in the future)

a) scripts/recipes to set up the correct CMSSW environment

b) a collection of latest recipes and a CMSSW config to apply them to get collections of standard objects suitable for physics analysis as CMSSW producers (e.g. recommendedTightMuons = tightMuonProducer(slimmedMuons) etc.)

c) an ntuple production configuration file that produces an ntuple that can be analysed in a similar manner als Delphes samples (with the DAnalysis framework)

Installation

cmsrel CMSSW_9_3_2
cd CMSSW_9_3_2/src
cmsenv
git cms-init
git cms-merge-topic -u nsmith-:EgammaFromMultiCl_932
mkdir -p RecoEgamma && pushd RecoEgamma
git clone -b integrated https://github.com/nsmith-/Phase2InterimID.git
popd
git clone https://github.com/jkiesele/PhaseTwoAnalysis.git
scram b -j 8

As the global tag contains Run-2 JEC, you might want to download an SQLite file to rerun JEC. See the TWiki for more details.

How to run PAT on RECO datasets

The PatProducer folder contains a configuration file to produce miniAOD files from RECO files. Interactively, after updating the list of input files, one can run

cmsRun miniAOD-prod_PAT.py

A skeleton of crab configuration file is also provided in this folder. The following fields need to be updated:

  • config.General.requestName
  • config.Data.inputDataset
  • config.Data.outLFNDirBase

before running

source /cvmfs/cms.cern.ch/crab3/crab.sh
crab submit crabConfig.py

Plotting basic distributions from RECO collections

A basic EDAnalyzer is available in the BasicRecoDistrib folder. Several private functions handle electron and forward muon ID. Lepton isolation is computed with a simple loop over neighbouring particles and there is no b-tagging information. Normalization to luminosity is not handled. More details are given in the implementation section of the .cc file. After updating the list of input files, the analyzer can be run interactively from the test subfolder :

cmsRun ConfFile_cfg.py

If you want to rerun JEC, you can use the updateJEC argument with the path to the SQLite file. Befor the EDAnalyzer, PUPPI is run on the fly and jets are re-clustered. The MET is also recomputed but not exactly with the official recipe (that needs PAT collections).

Plots in a pdf format can be obtained by running:

root -l 
.L plotIt.C++
plotIt()

A skeleton of crab configuration file is also provided. The following fields need to be updated:

  • config.General.requestName
  • config.Data.inputDataset
  • config.Data.outLFNDirBase

before running

source /cvmfs/cms.cern.ch/crab3/crab.sh
crab submit crabConfig.py

Plotting basic distributions from PAT collections

A basic EDAnalyzer is available in the BasicPatDistrib folder. Several private functions handle central electron and forward muon ID. A flag (useDeepCSV) can be set to true in the configuration file to use deepCSV rather than MVAv2 as b-tagging discriminant. Normalization to luminosity is not handled. More details are given in the implementation section of the .cc file. After updating the list of input files, the analyzer can be run interactively from the test subfolder :

cmsRun ConfFile_cfg.py

If you want to rerun JEC, you can use the updateJEC argument with the path to the SQLite file.

Plots in a pdf format can be obtained by running:

root -l 
.L plotIt.C++
plotIt()

A skeleton of crab configuration file is also provided. The following fields need to be updated:

  • config.General.requestName
  • config.Data.inputDataset
  • config.Data.outLFNDirBase

before running

source /cvmfs/cms.cern.ch/crab3/crab.sh
crab submit crabConfig.py

Producing flat ntuples

Flat ntuples can be produced in the NTupler folder, either from PAT or RECO events, by running interactively:

cmsRun scripts/produceNtuples_cfg.py skim=False/True outFilename=MiniEvents.root inputFormat=RECO/PAT

If you want to rerun JEC, you can use the updateJEC argument with the path to the SQLite file.

The skim flag can be used to reduce the size of the output files. A histogram containing the number of events before the skim is then stored in the output files. By default, events are required to contain at least 1 lepton and 2 jets, but this can be easily modified ll.71-97 of src/produceNtuples_cfg.py.

The structure of the output tree can be seen/modified in interface/MiniEvent.h and src/MiniEvent.cc.

The main analyzers are:

  • plugins/MiniFromPat.cc -- to run over PAT events
  • plugins/MiniFromReco.cc -- to run over RECO events

Details on the object definitions are given in the implementation section.

A skeleton of crab configuration file is also provided. The following fields need to be updated:

  • config.General.requestName
  • config.Data.inputDataset
  • config.Data.outLFNDirBase

before running

source /cvmfs/cms.cern.ch/crab3/crab.sh
crab submit crabConfig.py

To adjust the input parameters of scripts/produceNtuples_cfg.py, the three following fields need to be modified consistently:

  • config.JobType.pyCfgParams
  • config.JobType.inputFiles
  • config.JobType.outputFiles

Producing edm ntuples

CMS edm ntuples can be produced in the NTupler folder too, either from PAT or RECO events, by running interactively:

cmsRun scripts/edmFilter_cfg.py outFilename=FilteredEvents.root inputFormat=RECO/PAT

whether the input file format is RECO or miniAOD. If you want to rerun JEC, you can use the updateJEC argument with the path to the SQLite file.

To run over PAT events, the main producers are:

  • ../Electrons/plugins/PatElectronFilter.cc
  • ../Muons/plugins/PatMuonFilter.cc
  • ../Jets/plugins/PatJetFilter.cc

while for RECO events, they are:

  • ../Electrons/plugins/RecoElectronFilter.cc
  • ../Muons/plugins/RecoMuonFilter.cc
  • ../Jets/plugins/RecoJetFilter.cc

Details on the object definitions are given in the implementation section, but, please, already note that no JEC is applied and ak4 and PUPPI algorithms are considered everywhere.

When running over PAT events, the following collections are produced:

  • doubles_electronfilter_LooseElectronRelIso_EDMFilter
  • doubles_electronfilter_MediumElectronRelIso_EDMFilter
  • doubles_electronfilter_TightElectronRelIso_EDMFilter
  • patElectrons_electronfilter_LooseElectrons_EDMFilter
  • patElectrons_electronfilter_MediumElectrons_EDMFilter
  • patElectrons_electronfilter_TightElectrons_EDMFilter
  • doubles_muonfilter_LooseMuonRelIso_EDMFilter
  • doubles_muonfilter_MediumMuonRelIso_EDMFilter
  • doubles_muonfilter_TightMuonRelIso_EDMFilter
  • patMuons_muonfilter_LooseMuons_EDMFilter
  • patMuons_muonfilter_MediumMuons_EDMFilter
  • patMuons_muonfilter_TightMuons_EDMFilter
  • patJets_jetfilter_Jets_EDMFilter
  • patJets_jetfilter_LooseMVAv2Jets_EDMFilter
  • patJets_jetfilter_MediumMVAv2Jets_EDMFilter
  • patJets_jetfilter_TightMVAv2Jets_EDMFilter
  • patJets_jetfilter_LooseDeepCSVJets_EDMFilter
  • patJets_jetfilter_MediumDeepCSVJets_EDMFilter
  • patJets_jetfilter_TightDeepCSVJets_EDMFilter

while for RECO events, they are:

  • doubles_electronfilter_LooseElectronRelIso_EDMFilter
  • doubles_electronfilter_MediumElectronRelIso_EDMFilter
  • doubles_electronfilter_TightElectronRelIso_EDMFilter
  • recoGsfElectrons_electronfilter_LooseElectrons_EDMFilter
  • recoGsfElectrons_electronfilter_MediumElectrons_EDMFilter
  • recoGsfElectrons_electronfilter_TightElectrons_EDMFilter
  • doubles_muonfilter_LooseMuonRelIso_EDMFilter
  • doubles_muonfilter_MediumMuonRelIso_EDMFilter
  • doubles_muonfilter_TightMuonRelIso_EDMFilter
  • recoMuons_muonfilter_LooseMuons_EDMFilter
  • recoMuons_muonfilter_MediumMuons_EDMFilter
  • recoMuons_muonfilter_TightMuons_EDMFilter
  • recoPFJets_jetfilter_Jets_EDMFilter
  • recoPFMETs_puppiMet__EDMFilter

The initial vectors of electrons, muons, jets (and PFMETs) are dropped to avoid any confusion.