TWiki> CMSPublic Web>SWGuide>WorkBook>WorkBookMiniAOD>WorkBookMiniAOD2014 (2015-05-26, AndreaRizzi) EditAttachPDF

4.7 MiniAOD Analysis Documentation

This page documents the MiniAOD data tier as it is implemented in Phys14 (CMSSW release 7_2_0) and CSA14 (CMSSW release 7_0_6_patch1 or later).

Introduction

The MiniAOD format

The MiniAOD is a new high-level data tier introduced in Spring 2014 to serve the needs of the mainstream physics analyses while keeping a small event size (30-50 kb/event).

The main contents of the MiniAOD are:

  • High level physics objects (leptons, photons, jets, ETmiss), with detailed information in order to allow e.g. retuning of identification criteria, saved using PAT dataformats.
    Some preselection requirements are applied on the objects, and objects failing these requirements are either not stored or stored only with a more limited set of information.
    Some high level corrections are applied: L1+L2+L3(+residual) corrections to jets, type1 corrections to ETmiss.
  • The full list of particles reconstructed by the ParticleFlow, though only storing the most basic quantities for each object (4-vector, impact parameter, pdg id, some quality flags), and with reduced numerical precision; these are useful to recompute isolation, or to perform jet substructure studies.
    For charged particles with pT > 0.9 GeV, more information about the associated track is saved, including the covariance matrix, so that they can be used for b-tagging purposes.
  • MC Truth information: a subset of the genParticles enough to describe the hard scattering process, jet flavour information, and final state leptons and photons; GenJets with pT > 8 GeV are also stored, and so are the other mc summary information (e.g event weight, LHE header, PDF, PU information).
    In addition, all the stable genParticles with mc status code 1 are also saved, to allow reclustering of GenJets with different algorithms and substructure studies.
  • Trigger information: MiniAOD contains the trigger bits associated to all paths, and all the trigger objects that have contributed to firing at least one filter within the trigger. In addition, we store all objects reconstructed at L1 and the L1 global trigger summary, and the prescale values of all the triggers.

Structure of this documentation

This documentation will focus on analyzing MiniAOD files, since the production of MiniAODs is normally done centrally (but in the last part some instructions will be given also on how to produce MiniAOD files privately). The examples will be given both in the context of full CMSSW running using EDAnalyzers and in the context of FWLite with python whenever applicable. FWLite in C++ is of course also supported, but it's up to the reader to guess what the code should be starting from the two examples.

The color scheme of the twikipage is as follows:

  • Shell commands will be embedded in grey box, e.g. 
    cmsrel CMSSW_7_0_6_patch1
  • Python code will be embedded in light blue box, e.g. 
from DataFormats.FWLite import Handle, Events
  • C++ code will be embedded in light green box, e.g. 
iEvent.getByToken(muons_, muons);
  • Example output will be in a yellow box, e.g. 
    boh?

Setting up your environment

The documentation will use the CMSSW_7_0_6_patch1 release and assume you're working on a SLC6 node (e.g. lxplus6.cern.ch). To setup the release, you can do 
cmsrel CMSSW_7_0_6_patch1
cd CMSSW_7_0_6_patch1/src
cmsenv

As input test files in the examples we will use /store/cmst3/user/gpetrucc/miniAOD/v1/TT_Tune4C_13TeV-pythia8-tauola_PU_S14_PAT.root (available at CERN), which contains ~8k events of ttbar production, but any other MINIAODSIM file will work equally well.

Skeleton for CMSSW Analysis

To create a skeleton of c++ analyzer, you can do the following 
mkdir Test
cd Test
mkedanlzr MiniAnalyzer

New package "MiniAnalyzer" of EDAnalyzer type is successfully generated
MiniAnalyzer/
|  plugins/
|  |-- BuildFile.xml
|  |-- MiniAnalyzer.cc
|  python/
|  |-- CfiFile_cfi.py
|  |-- ConfFile_cfg.py
|  test/
|  doc/
Total: 4 directories, 4 files

Then you'll need to edit the BuildFile.xml to add a <use name="DataFormats/PatCandidates"/>

Skeleton for Python FWLite Analysis 

# import ROOT in batch mode
import sys
oldargv = sys.argv[:]
sys.argv = [ '-b-' ]
import ROOT
ROOT.gROOT.SetBatch(True)
sys.argv = oldargv

# load FWLite C++ libraries
ROOT.gSystem.Load("libFWCoreFWLite.so");
ROOT.gSystem.Load("libDataFormatsFWLite.so");
ROOT.AutoLibraryLoader.enable()

# load FWlite python libraries
from DataFormats.FWLite import Handle, Events

# open file (you can use 'edmFileUtil -d /store/whatever.root' to get the physical file name)
events = Events("root://eoscms//eos/cms/store/cmst3/user/gpetrucc/miniAOD/v1/TT_Tune4C_13TeV-pythia8-tauola_PU_S14_PAT.root")

for i,event in enumerate(events):
    print "\nEvent", i
    if i > 10: break

Analyzing MiniAOD

High level physics objects

High level physics objects in miniAOD are saved in the PAT dataformats, which in turn derive from the RECO/AOD formats, so almost everything that you can find in the WorkBook documentation on PAT and physics objects applies to them direcly.

Object Label C++ class Selection Detailed information
Muons slimmedMuons std::vector<pat::Muon> includes muons with pT > 5 GeV or that pass the PF muon ID (see below) All standard muon ID and isolations, the associated tracker, outer and global tracks, muonBestTrack and typePMuonBestTrack
Electrons slimmedElectrons std::vector<pat::Electron> all gedGsfElectrons electrons For electrons of pT > 5 GeV, full information is saved (supercluster, seed cluster, interesting rechits, isolation and id variables).
For electrons below the threshold, only the superCluster and seedCluster are provided, and the id and isolation variables are zeroed out to save space
Taus slimmedTaus std::vector<pat::Tau> taus from hpsPFTauProducer with pT > 18 GeV, and passing the basic decayModeFinding id All POG-supported tau id discriminators are included. Links to the PF candidates are also provided.
Photons slimmedPhotons std::vector<pat::Photon> gedPhotons with pT > 14 GeV and hadTowOverEm() < 0.15 For photons that pass a minimal r9 or isolation cut, full information is saved (supercluster, seed cluster, interesting rechits, isolation and id variables).
The requirement is the logical or of the three conditions r9()>0.8 , chargedHadronIso()<20, chargedHadronIso()<0.3*pt()
Jets slimmedJets std::vector<pat::Jet> ak4PFJetsCHS with pT > 10 GeV L1+L2+L3+residual corrections are applied; b-tagging and pileup jet id information are embedded. Links are provided to the constituent PF candidates.
slimmedGenJets std::vector<reco::GenJet> ak4GenJets with pT > 8 GeV links to constituents (packed gen particles) are available
slimmedJetsAK8 std::vector<pat::Jet> ak8PFJetsCHS with pT > 100 GeV (for substructure) L1+L2+L3+residual corrections are applied (AK7PFchs corrections used); some precomputed substructure info is provided, and also links to the constituent PF candidates.
MET slimmedMETs std::vector<pat::MET> the type1 PF MET MET uncertainties are provided
note: type1 corrections are computed from non-CHS ak4PFJets , as prescribed by the JetMET group.

Muons

Muons from the AOD muons collection are included in MiniAOD if they pass at least one of these three requirements:

  • pT > 5 GeV
  • pT > 3 GeV and any of the following IDs: PF, global, arbitrated tracker, standalone, RPCMuLoose
  • any pT, if they pass the PF ID.

Muon id information is available and accessible as usual through the isLooseMuon, isTightMuon, isSoftMuon, isHighPtMuon and muonID methods as provided in the pat::Muon class.

The following tracks are embedded by value in the pat::Muon, if available: tracker, outer and global tracks, muonBestTrack and typePMuonBestTrack.
DYT and TeV-muon refits are not embedded in 70X and 72X (though if the muonBestTrack or tunePMuonBestTrack selects any of those tracks, they'll be available), but they are be available for muons with pt > 100 GeV starting from CMSSW 73X.

known Features in CSA14 and PHYS14

  • Due to a bug, in some muons the reference tunePMuonBestTrack not available (but the muonBestTrack always is). A fix exists in newer releases (PR #8568), that can be applied as is also in CMSSW 72X, but it requires the miniAODs to be reproduced starting from AOD.

Electrons

same All gedGsfElectrons electrons are saved.

For electrons of pT > 5 GeV, detailed information is saved:

  • Supercluster, basic clusters and seed clusters, accessible through the reco::GsfElectron information (though they're stored externally in the reducedEgamma collections). The final superclusters and clusters are saved, not also the intermediate unrefined "PF" superclusters and clusters.
  • Interesting rechits, saved in reducedEGamma:reducedEBRecHits, reducedEGamma:reducedEERecHits, reducedEGamma:reducedESRecHits for barrel, endcap and pre-showers respectively.
  • Shower shape variables computed from the standard superclusters are provided in the basic reco::GsfElectron interface, e.g. sigmaIetaIeta() (the only missing one in AOD is sigmaIetaIphi() which is added in pat::Electron interface).
    In addition, for backwards compatibility MiniAOD also includes the shower shape variables computed using the full 5x5 superclusters like in Run 1, which are accessible through the pat::Electron methods, e.g. full5x5_sigmaIetaIeta().

For electrons that don't satisfy this cut, only the supercluster and seed clusters are provided.

Photons

For photons, the source collection is the gedPhotons, with two levels of selection:

  • A basic selection defining which objects are kept: pT > 14 GeV and hadTowOverEm() < 0.15
  • A tighter selection defining for which objects the detailed information is kept: it's the logical or of a R9 requirement r9()>0.8 , and an absolute and relative isolation cuts done with charged hadrons only, chargedHadronIso()<20, chargedHadronIso()<0.3*pt()

Similarly to electrons, the basic information is only the supercluster and seed clusters, while the detailed information includes basic clusters and interesting rechits.

Note that electrons and photons share a common collection of interesting clusters and rechits, and so if the same supercluster gives rise to both an electron and a photon the two high-level physics objects will really reference to the same object (as it happens in AOD). For more technical details, refer to reducedEgamma_cfi.py

Taus

For taus, MiniAOD includes all objects with pT > 18 GeV and passing the decayModeFinding discriminator. Tighter tau ID requirements can be applied using the tauID(name) method, all POG-supported discriminators are included. The tau ID discriminators recommended for CSA14 and Phys14 studies are described on this wiki.

Additional information on the taus, e.g. the impact parameter of the reconstructed tau vertex, are saved in the TauPFEssential format which is embedded in the pat::Tau object.

Links to the packed PF candidates corresponding to the PFCandidates used to reconstruct a given tau are accessible through the methods like leadChargedHadrCand(), leadNeutralCand(), leadCand(), signalCands(), isolationCands(), ...
Note instead that the older methods with similar name but with a PF inside, like signalPFCands(), are not usable on MiniAOD since they're not compatible with the packed PF candidate format used in MiniAOD.

Jets

Two collections of jets are saved: one general purpose collection of jets (from ak4PFJetsCHS ), and one for dedicated studies of substructure jets (from ak8PFJetsCHS). Both are made using anti-kT clustering out of Particle Flow candidates, the only difference is the distance parameter, which approximately corresponds to the radius of the jet cone.

For both jet collections, links to the daughters are available through the daughter(idx) and numberOfDaughters() methods, and they will point into the collection of packed PF candidates described below.
It should be noted that the getPFConstituent and getPFConstituents methods instead will NOT work, since they explicitly require the daughter to be a full reco::PFCandidate and not a packed PF one. One subtle point for AK8 jets is that, to save space, the first two constituents of the AK8 jets are the soft-dropped subjets. The rest of the constituents store the unclustered =pat::PackedCandidate=s.

For the AK4 jets, the following information is provided:

  • b-tagging discriminators, available from the bDiscriminator(name) method, for: jetBProbabilityBJetTags, jetProbabilityBJetTags, trackCountingHighPurBJetTags, trackCountingHighEffBJetTags, simpleSecondaryVertexHighEffBJetTags, simpleSecondaryVertexHighPurBJetTags, combinedSecondaryVertexBJetTags , combinedInclusiveSecondaryVertexBJetTags
  • additional variables related to the associated secondary vertex if there is one, stored as userFloats: the mass of the vertex (vtxMass), the number of tracks in it (vtxNtracks), and the decay length value and significance (vtx3DVal, vtx3DSig)
  • the discriminator for the MVA PileUp id, saved as userFloat("pileupJetId:fullDiscriminant"). It should be noted that the training used is for ak5PFJetsCHS in CMSSW 5.3.X and Run 1 pileup, so it's probably not optimal for Run 2.

For AK8 jets, the following information is included:

  • pruned, trimmed and filtered masses, available as userFloats with labels ak8PFJetsCHSPrunedLinks, ak8PFJetsCHSTrimmedLinks, ak8PFJetsCHSFilteredLinks
  • NEW for 72x : n-subjettiness variables tau1, tau2, and tau3, available as userFloats with labels NjettinessAK8:tau1, NjettinessAK8:tau2, and NjettinessAK8:tau3
  • NEW for 72x : the CMS top tagger reclusters the AK8 constituents with the Cambridge-Aachen algorithm with D=0.8 (CA8). It outputs between one and four subjets corresponding to the decay products of the top quark sequential decay (quarks from W decay, and the b quark). This information is available as tagInfo("caTop").properties().topMass, tagInfo("caTop").properties().minMass, tagInfo("caTop").properties().nSubJets
  • The documentation to use userFloats is highlighted here. A simple snippet to perform a very simple W/Z/H tagging selection would be :
Snippet to perform simple W/Z/H tagging Hide example CMSSW code 
double tau1 = jet.userFloat("NjettinessAK8:tau1");    //
double tau2 = jet.userFloat("NjettinessAK8:tau2");    //  Access the n-subjettiness variables
double tau3 = jet.userFloat("NjettinessAK8:tau3");    // 

double trimmed_mass = jet.userFloat("ak8PFJetsCHSTrimmedLinks");   // access to trimmed mass
double pruned_mass = jet.userFloat("ak8PFJetsCHSPrunedLinks");     // access to pruned mass
double filtered_mass = jet.userFloat("ak8PFJetsCHSFilteredLinks"); // access to filtered mass

bool mySimpleWTagger = (tau2/tau1) < 0.6 && pruned_mass > 50.0;
  • The documentation to use TagInfos is highlighted here. The CMS top tagger utilizes this functionality as well. To implement a simple top tagger, a simple snippet would be :
Snippet to perform simple top tagging Hide example CMSSW code 

reco::CATopTag const * tagInfo =  dynamic_cast<reco::CATopTagInfo const *>( jet.tagInfo("caTop"));
bool topTagged = false;
if ( tagInfo != 0 ) {
   double minMass = tagInfo->properties().minMass;
   double topMass = tagInfo->properties().topMass;
   int nSubJets = tagInfo->properties().nSubJets;

   if ( nSubJets > 2 && minMass > 50.0 && topMass > 150.0 )
      topTagged = true;
}
  • To access all of the constituents of the soft-drop groomed jet:
Snippet to access all constituents of the soft-drop groomed jet Hide example CMSSW code 

std::vector< pat::PackedCandidate const *> constituents;
for ( unsigned int i = 0; i < jet.numberOfDaughters(); ++i ) {
   auto const * subjet = jet.daughter(i);
   for ( unsigned int j = 0; j < subjet->numberOfDaughters(); ++j ) {
      constituents.push_back( subjet.daughter(j) );
   }
}
  • To access all of the constituents of the ungroomed jet :
Snippet to access all constituents of the ungroomed jet Hide example CMSSW code 

std::vector< pat::PackedCandidate const *> constituents;
// First get the constituents from the subjets
for ( unsigned int i = 0; i < 2; ++i ) {
   auto const & subjet = jet.daughter(i);
   for ( unsigned int j = 0; j < subjet.numberOfDaughters(); ++j ) {
      constituents.push_back( subjet.daughter(j) );
   }
}
// Then get the constituents that were groomed away
for ( unsigned int i = 2; i < jet.numberOfDaughters(); ++i ) {
   constituents.push_back( jet.daughter(i) );
}

ETmiss

The type1 corrected ETmiss is used in the MiniAOD. Also, on MC the generated ETmiss is provided. The caloMet is not stored (though it will be accessible from the pat::MET in 74X) Note that the type1 corrections are computed from non-CHS ak4PFJets , as prescribed by the JetMET group.

The more sophisticated algorithm like MVAmet is not available at the moment. The metsignificance stored refers to the algorithm used at the beginning of run1 and the one used at the end of run1 is not available at the moment. Standalone recipes to run on top of miniAOD are needed to recalculate those.

The pat::MET object also contains the information related to the type of corrections and uncertainties from the standard POG recipe of shifting and smearing objects, from the shiftedXyz(uncertainty, level) methods:

  • Xyz can be any of P2, P3, P4, Px, Py, Pz, Phi, SumEt
  • uncertainty can be any of the 12 variations considered (e.g. UnclusteredEnUp or JetResDown)
  • level is Type1 (default), Raw, Type1p2

The default Type1 and the shiftedXyz(uncertainty, level) methods correct for the scale of ak4PFJets (not for the ak4PFJetsCHS). As example: you can get the raw MET with the slimmetMET.shiftedPt(pat::MET::NoShift, pat::MET::Raw) (in python slimmetMET.shiftedPt(slimmetMET.NoShift, slimmetMET.Raw) )

known Features in PHYS14 In the PHYS14 miniAOD there are four know features

  • 0) method uncorrectedPt() is broken: see hn link
  • 1) MET significance empty: see hn link
  • 2) same exactly value for "up" and for "down" variation (affect JetEn and UnclusteredEn) hn link
  • 3) MET uncertainties with C++ : if the pat::MET object is copied by value instead of taking a const pat::MET & reference to the one in the event, one has errors when using the shiftedP** methods . See HN question, HN answer
  • 4) met.Pt() value and pat::MET::NoShift values does not return the same exact value. This is due to the differnece in the muon cleaning applied at the RECO and PAT level hn link
0) 2) ad 3) 4) will be fixed in the next version of the miniAOD: see PR#8159; extra things fixed in the PR#8416

known Features in CSA14 In the CSA miniAOD first version, due to an implementation issue in the pat::MET class, it is not possible to undo the type1 corrections to retrieve the raw PF ETmiss; however, that can be recalculated with very good accuracy from the packed PF candidates according to the recipes below (which in the end do nothing more than taking the negative vector sum of all pf candidates, as per the pf met definition).

How to re-make the raw pfMet in CMSSW (EDProducer) Hide example CMSSW code
just put this in your CMSSW configuration file 
from RecoMET.METProducers.PFMET_cfi import pfMet

process.pfMet = pfMet.clone(src = "packedPFCandidates")
process.pfMet.calculateSignificance = False # this can't be easily implemented on packed PF candidates at the moment
and then put the process.pfMet in your path.

How to re-make the raw pfMet in FWLite (python) Hide example python code 
from math import atan2, hypot

# define the handle
pfs = Handle("std::vector<pat::PackedCandidate>")

# within the event loop, do
event.getByLabel("packedPFCandidates", pfs)
mypx,mypy,mysumet = 0,0,0
for pf in pfs.product():
        mypx -= pf.px()
        mypy -= pf.py()
        mysumet += pf.pt()
mymet = hypot(py,px)
myphi = atan2(py,px)

Example code accessing all high-level physics objects

Show example CMSSW code Hide example CMSSW code 
// system include files
#include <memory>

// user include files
#include "FWCore/Framework/interface/Frameworkfwd.h"
#include "FWCore/Framework/interface/EDAnalyzer.h"

#include "FWCore/Framework/interface/Event.h"
#include "FWCore/Framework/interface/MakerMacros.h"

#include "FWCore/ParameterSet/interface/ParameterSet.h"

#include "DataFormats/VertexReco/interface/VertexFwd.h"
#include "DataFormats/VertexReco/interface/Vertex.h"
#include "DataFormats/PatCandidates/interface/Muon.h"
#include "DataFormats/PatCandidates/interface/Electron.h"
#include "DataFormats/PatCandidates/interface/Tau.h"
#include "DataFormats/PatCandidates/interface/Photon.h"
#include "DataFormats/PatCandidates/interface/Jet.h"
#include "DataFormats/PatCandidates/interface/MET.h"
#include "DataFormats/PatCandidates/interface/PackedCandidate.h"
//
// class declaration
//

class MiniAnalyzer : public edm::EDAnalyzer {
   public:
      explicit MiniAnalyzer(const edm::ParameterSet&);
      ~MiniAnalyzer();

   private:
      virtual void analyze(const edm::Event&, const edm::EventSetup&) override;

      // ----------member data ---------------------------
      edm::EDGetTokenT<reco::VertexCollection> vtxToken_;
      edm::EDGetTokenT<pat::MuonCollection> muonToken_;
      edm::EDGetTokenT<pat::ElectronCollection> electronToken_;
      edm::EDGetTokenT<pat::TauCollection> tauToken_;
      edm::EDGetTokenT<pat::PhotonCollection> photonToken_;
      edm::EDGetTokenT<pat::JetCollection> jetToken_;
      edm::EDGetTokenT<pat::JetCollection> fatjetToken_;
      edm::EDGetTokenT<pat::METCollection> metToken_;
};

MiniAnalyzer::MiniAnalyzer(const edm::ParameterSet& iConfig):
    vtxToken_(consumes<reco::VertexCollection>(iConfig.getParameter<edm::InputTag>("vertices"))),
    muonToken_(consumes<pat::MuonCollection>(iConfig.getParameter<edm::InputTag>("muons"))),
    electronToken_(consumes<pat::ElectronCollection>(iConfig.getParameter<edm::InputTag>("electrons"))),
    tauToken_(consumes<pat::TauCollection>(iConfig.getParameter<edm::InputTag>("taus"))),
    photonToken_(consumes<pat::PhotonCollection>(iConfig.getParameter<edm::InputTag>("photons"))),
    jetToken_(consumes<pat::JetCollection>(iConfig.getParameter<edm::InputTag>("jets"))),
    fatjetToken_(consumes<pat::JetCollection>(iConfig.getParameter<edm::InputTag>("fatjets"))),
    metToken_(consumes<pat::METCollection>(iConfig.getParameter<edm::InputTag>("mets")))
{
}

MiniAnalyzer::~MiniAnalyzer()
{
}


void
MiniAnalyzer::analyze(const edm::Event& iEvent, const edm::EventSetup& iSetup)
{
    edm::Handle<reco::VertexCollection> vertices;
    iEvent.getByToken(vtxToken_, vertices);
    if (vertices->empty()) return; // skip the event if no PV found
    const reco::Vertex &PV = vertices->front();

    edm::Handle<pat::MuonCollection> muons;
    iEvent.getByToken(muonToken_, muons);
    for (const pat::Muon &mu : *muons) {
        if (mu.pt() < 5 || !mu.isLooseMuon()) continue;
        printf("muon with pt %4.1f, dz(PV) %+5.3f, POG loose id %d, tight id %d\n",
                mu.pt(), mu.muonBestTrack()->dz(PV.position()), mu.isLooseMuon(), mu.isTightMuon(PV));
    }

    edm::Handle<pat::ElectronCollection> electrons;
    iEvent.getByToken(electronToken_, electrons);
    for (const pat::Electron &el : *electrons) {
        if (el.pt() < 5) continue;
        printf("elec with pt %4.1f, supercluster eta %+5.3f, sigmaIetaIeta %.3f (%.3f with full5x5 shower shapes), lost hits %d, pass conv veto %d\n",
                    el.pt(), el.superCluster()->eta(), el.sigmaIetaIeta(), el.full5x5_sigmaIetaIeta(), el.gsfTrack()->trackerExpectedHitsInner().numberOfLostHits(), el.passConversionVeto());
    }

    edm::Handle<pat::PhotonCollection> photons;
    iEvent.getByToken(photonToken_, photons);
    for (const pat::Photon &pho : *photons) {
        if (pho.pt() < 20 or pho.chargedHadronIso()/pho.pt() > 0.3) continue;
        printf("phot with pt %4.1f, supercluster eta %+5.3f, sigmaIetaIeta %.3f (%.3f with full5x5 shower shapes)\n",
                    pho.pt(), pho.superCluster()->eta(), pho.sigmaIetaIeta(), pho.full5x5_sigmaIetaIeta());
    }


    edm::Handle<pat::TauCollection> taus;
    iEvent.getByToken(tauToken_, taus);
    for (const pat::Tau &tau : *taus) {
        if (tau.pt() < 20) continue;
        printf("tau  with pt %4.1f, dxy signif %.1f, ID(byMediumCombinedIsolationDeltaBetaCorr3Hits) %.1f, lead candidate pt %.1f, pdgId %d \n",
                    tau.pt(), tau.dxy_Sig(), tau.tauID("byMediumCombinedIsolationDeltaBetaCorr3Hits"), tau.leadCand()->pt(), tau.leadCand()->pdgId());
    }


    edm::Handle<pat::JetCollection> jets;
    iEvent.getByToken(jetToken_, jets);
    int ijet = 0;
    for (const pat::Jet &j : *jets) {
        if (j.pt() < 20) continue;
        printf("jet  with pt %5.1f (raw pt %5.1f), eta %+4.2f, btag CSV %.3f, CISV %.3f, pileup mva disc %+.2f\n",
            j.pt(), j.pt()*j.jecFactor("Uncorrected"), j.eta(), std::max(0.f,j.bDiscriminator("combinedSecondaryVertexBJetTags")), std::max(0.f,j.bDiscriminator("combinedInclusiveSecondaryVertexBJetTags")), j.userFloat("pileupJetId:fullDiscriminant"));
        if ((++ijet) == 1) { // for the first jet, let's print the leading constituents
            std::vector<reco::CandidatePtr> daus(j.daughterPtrVector());
            std::sort(daus.begin(), daus.end(), [](const reco::CandidatePtr &p1, const reco::CandidatePtr &p2) { return p1->pt() > p2->pt(); }); // the joys of C++11
            for (unsigned int i2 = 0, n = daus.size(); i2 < n && i2 <= 3; ++i2) {
                const pat::PackedCandidate &cand = dynamic_cast<const pat::PackedCandidate &>(*daus[i2]);
                printf("         constituent %3d: pt %6.2f, dz(pv) %+.3f, pdgId %+3d\n", i2,cand.pt(),cand.dz(PV.position()),cand.pdgId());
            }
        }
    }


    edm::Handle<pat::JetCollection> fatjets;
    iEvent.getByToken(fatjetToken_, fatjets);
    for (const pat::Jet &j : *fatjets) {
        printf("AK8j with pt %5.1f (raw pt %5.1f), eta %+4.2f, mass %5.1f ungroomed, %5.1f pruned, %5.1f trimmed, %5.1f filtered. CMS TopTagger %.1f\n",
            j.pt(), j.pt()*j.jecFactor("Uncorrected"), j.eta(), j.mass(), j.userFloat("ak8PFJetsCHSPrunedLinks"), j.userFloat("ak8PFJetsCHSTrimmedLinks"), j.userFloat("ak8PFJetsCHSFilteredLinks"), j.userFloat("cmsTopTagPFJetsCHSLinksAK8"));
    }
 
    edm::Handle<pat::METCollection> mets;
    iEvent.getByToken(metToken_, mets);
    const pat::MET &met = mets->front();
    printf("MET: pt %5.1f, phi %+4.2f, sumEt (%.1f). genMET %.1f. MET with JES up/down: %.1f/%.1f\n",
        met.pt(), met.phi(), met.sumEt(),
        met.genMET()->pt(),
        met.shiftedPt(pat::MET::JetEnUp), met.shiftedPt(pat::MET::JetEnDown));

    printf("\n");
}

//define this as a plug-in
DEFINE_FWK_MODULE(MiniAnalyzer);
python configuration to run it 
import FWCore.ParameterSet.Config as cms

process = cms.Process("Demo")

process.load("FWCore.MessageService.MessageLogger_cfi")
process.maxEvents = cms.untracked.PSet( input = cms.untracked.int32(10) )

process.source = cms.Source("PoolSource",
    fileNames = cms.untracked.vstring(
        '/store/cmst3/user/gpetrucc/miniAOD/v1/TT_Tune4C_13TeV-pythia8-tauola_PU_S14_PAT.root'
    )
)

process.demo = cms.EDAnalyzer("MiniAnalyzer",
    vertices = cms.InputTag("offlineSlimmedPrimaryVertices"),
    muons = cms.InputTag("slimmedMuons"),
    electrons = cms.InputTag("slimmedElectrons"),
    taus = cms.InputTag("slimmedTaus"),
    photons = cms.InputTag("slimmedPhotons"),
    jets = cms.InputTag("slimmedJets"),
    fatjets = cms.InputTag("slimmedJetsAK8"),
    mets = cms.InputTag("slimmedMETs"),
)


process.p = cms.Path(process.demo)

Show example python code Hide example python code 
# import ROOT in batch mode
import sys
oldargv = sys.argv[:]
sys.argv = [ '-b-' ]
import ROOT
ROOT.gROOT.SetBatch(True)
sys.argv = oldargv

# load FWLite C++ libraries
ROOT.gSystem.Load("libFWCoreFWLite.so");
ROOT.gSystem.Load("libDataFormatsFWLite.so");
ROOT.AutoLibraryLoader.enable()

# load FWlite python libraries
from DataFormats.FWLite import Handle, Events

muons, muonLabel = Handle("std::vector<pat::Muon>"), "slimmedMuons"
electrons, electronLabel = Handle("std::vector<pat::Electron>"), "slimmedElectrons"
photons, photonLabel = Handle("std::vector<pat::Photon>"), "slimmedPhotons"
taus, tauLabel = Handle("std::vector<pat::Tau>"), "slimmedTaus"
jets, jetLabel = Handle("std::vector<pat::Jet>"), "slimmedJets"
fatjets, fatjetLabel = Handle("std::vector<pat::Jet>"), "slimmedJetsAK8"
mets, metLabel = Handle("std::vector<pat::MET>"), "slimmedMETs"
vertices, vertexLabel = Handle("std::vector<reco::Vertex>"), "offlineSlimmedPrimaryVertices"

# open file (you can use 'edmFileUtil -d /store/whatever.root' to get the physical file name)
events = Events("root://eoscms//eos/cms/store/cmst3/user/gpetrucc/miniAOD/v1/TT_Tune4C_13TeV-pythia8-tauola_PU_S14_PAT.root")

for iev,event in enumerate(events):
    if iev >= 10: break
    event.getByLabel(muonLabel, muons)
    event.getByLabel(electronLabel, electrons)
    event.getByLabel(photonLabel, photons)
    event.getByLabel(tauLabel, taus)
    event.getByLabel(jetLabel, jets)
    event.getByLabel(fatjetLabel, fatjets)
    event.getByLabel(metLabel, mets)
    event.getByLabel(vertexLabel, vertices)

    print "\nEvent %d: run %6d, lumi %4d, event %12d" % (iev,event.eventAuxiliary().run(), event.eventAuxiliary().luminosityBlock(),event.eventAuxiliary().event())

    # Vertices
    if len(vertices.product()) == 0 or vertices.product()[0].ndof() < 4:
        print "Event has no good primary vertex."
        continue
    else:
        PV = vertices.product()[0]
        print "PV at x,y,z = %+5.3f, %+5.3f, %+6.3f (ndof %.1f)" % (PV.x(), PV.y(), PV.z(), PV.ndof())

    # Muons
    for i,mu in enumerate(muons.product()): 
        if mu.pt() < 5 or not mu.isLooseMuon(): continue
        print "muon %2d: pt %4.1f, dz(PV) %+5.3f, POG loose id %d, tight id %d." % (
            i, mu.pt(), mu.muonBestTrack().dz(PV.position()), mu.isLooseMuon(), mu.isTightMuon(PV))

    # Electrons
    for i,el in enumerate(electrons.product()):
        if el.pt() < 5: continue
        print "elec %2d: pt %4.1f, supercluster eta %+5.3f, sigmaIetaIeta %.3f (%.3f with full5x5 shower shapes), lost hits %d, pass conv veto %d" % (
                    i, el.pt(), el.superCluster().eta(), el.sigmaIetaIeta(), el.full5x5_sigmaIetaIeta(), el.gsfTrack().trackerExpectedHitsInner().numberOfLostHits(), el.passConversionVeto())

    # Photon
    for i,pho in enumerate(photons.product()):
        if pho.pt() < 20 or pho.chargedHadronIso()/pho.pt() > 0.3: continue
        print "phot %2d: pt %4.1f, supercluster eta %+5.3f, sigmaIetaIeta %.3f (%.3f with full5x5 shower shapes)" % (
                    i, pho.pt(), pho.superCluster().eta(), pho.sigmaIetaIeta(), pho.full5x5_sigmaIetaIeta())

    # Tau
    for i,tau in enumerate(taus.product()):
        if tau.pt() < 20: continue
        print "tau  %2d: pt %4.1f, dxy signif %.1f, ID(byMediumCombinedIsolationDeltaBetaCorr3Hits) %.1f, lead candidate pt %.1f, pdgId %d " % (
                    i, tau.pt(), tau.dxy_Sig(), tau.tauID("byMediumCombinedIsolationDeltaBetaCorr3Hits"), tau.leadCand().pt(), tau.leadCand().pdgId()) 


    # Jets (standard AK4)
    for i,j in enumerate(jets.product()):
        if j.pt() < 20: continue
        print "jet %3d: pt %5.1f (raw pt %5.1f), eta %+4.2f, btag CSV %.3f, CISV %.3f, pileup mva disc %+.2f" % (
            i, j.pt(), j.pt()*j.jecFactor('Uncorrected'), j.eta(), max(0,j.bDiscriminator("combinedSecondaryVertexBJetTags")), max(0,j.bDiscriminator("combinedInclusiveSecondaryVertexBJetTags")), j.userFloat("pileupJetId:fullDiscriminant"))
        if i == 0: # for the first jet, let's print the leading constituents
            constituents = [ j.daughter(i2) for i2 in xrange(j.numberOfDaughters()) ]
            constituents.sort(key = lambda c:c.pt(), reverse=True)
            for i2, cand in enumerate(constituents):
                print "         constituent %3d: pt %6.2f, dz(pv) %+.3f, pdgId %+3d" % (i2,cand.pt(),cand.dz(PV.position()),cand.pdgId()) 
                if i2 > 3: break

    # Fat AK8 Jets
    for i,j in enumerate(fatjets.product()):
        print "jet %3d: pt %5.1f (raw pt %5.1f), eta %+4.2f, mass %5.1f ungroomed, %5.1f pruned, %5.1f trimmed, %5.1f filtered. CMS TopTagger %.1f" % (
            i, j.pt(), j.pt()*j.jecFactor('Uncorrected'), j.eta(), j.mass(), j.userFloat('ak8PFJetsCHSPrunedLinks'), j.userFloat('ak8PFJetsCHSTrimmedLinks'), j.userFloat('ak8PFJetsCHSFilteredLinks'), j.userFloat("cmsTopTagPFJetsCHSLinksAK8"))

    # MET:
    met = mets.product().front()
    print "MET: pt %5.1f, phi %+4.2f, sumEt (%.1f). rawMET: %.1f, genMET %.1f. MET with JES up/down: %.1f/%.1f" % (
        met.pt(), met.phi(), met.sumEt(),
        met.uncorrectedPt(), #<<?-- seems to return the same as pt(): help needed from JetMET experts!
        met.genMET().pt(),
        met.shiftedPt(ROOT.pat.MET.JetEnUp), met.shiftedPt(ROOT.pat.MET.JetEnDown));

Packed ParticleFlow Candidates

All candidates reconstructed by the ParticleFlow algorithm are saved in MiniAOD in the packedPFCandidates collection, using as dataformat pat::PackedCandidate.

For all packed PF candidates, this information is stored:

  • the momentum 4-vector, accessible through the usual methods like pt(), eta(), phi(), mass(), energy(), px(), p4(), polarP4().
    An additional method phiAtVtx() is provided, returning the phi of the candidate's track at the vertex; this is identical to phi() for the vast majority of the particles, but the two might differ for some of them if the calorimeters had contributed significantly in defining the 4-vector of the particle.
  • the particle charge and pdgId: 11, 13, 22 for ele/mu/gamma, 211 for charged hadrons, 130 for neutral hadrons, 1 and 2 for hadronic and em particles in HF.
  • the longitudinal and transverse impact parameters with respect to the PV: dxz(), dz().
    The impact parameters can also be recomputed with respect to any other position by calling the methods dxz(point), dz(point)).
    The vertex() method returns the position of the point of closest approach to the PV. This shouldn't be confused with the vertexRef() method, that returns a reference to the PV itself.
  • quality flags:
    • Association to the first PV, fromPV(): returns a number between 3 and 0 to define how tight the association with the first PV is: the tighest, 3 (PVUsedInFit), is if the track is used in the first PV fit; 2 (PVTight) is if the track is not used in the fit of any of the PVs, and is closest in z to the first PV, while 1 (PVLoose) is if the track is closest in z to a PV other then the first one. 0 (NoPV) is returned if the track is used in the fit of another PV.
      The definition normally used for isolation calculations is fromPV() > 1; the definition used for CHS subtraction in jets is fromPV() > 0.
    • Inner hit information: lostInnerHits() will return -1 (validHitInFirstPixelBarrelLayer) if the track has a valid hit in the first pixel barrel layer, 0 (noLostInnerHits) if it doesn't have such hit because of geometrical or detector inefficiencies (i.e. the hit wasn't expected to be there), 1 (oneLostHit) if the track extrapolation towards the beamline crosses an active detector but no hit is found there, 2 (moreLostHits) if there are at least two missing expected inner hits.
      For electrons, this information refers to the gsfTrack(), i.e. the usual cuts electron.gsfTrack()->trackerExpectedHitsInner().numberOfLostHits() <= X can be implemented as cand.lostInnerHits() <= X (but only for X equal to 0 or 1)
    • High-purity track: trackHighPurity() will return true for charged candidates whose reco::Teack had the highPurity quality flag set.

For charged packed PF candidates with pT > 0.95 GeV, additional information is stored:

  • the uncertainty on the impact parameter dzError(), dxyError()
  • the number of hits and pixel hits on the track (numberOfHits(), numberOfPixelHits())
  • the reco::Track of the candidate is provided by the pseudoTrack() method, with:
    • the momentum of the particle
    • an approximate covariance matrix of the track state at the vertex
    • approximate hitPattern() and trackerExpectedHitsInner() that yield the correct number of hits, pixel hits and the information returned by lostInnerHits()
      (but nothing more, so e.g. you can't rely on methods counting layers, or expected outer hits, or lost hits within the track)
    • the track normalized chisquare (truncated to an integer)
    • the highPurity quality flag set, if the original track had it.

All physics objects except the MET contain pointers to the packed PF candidates corresponding to the PFCandidates they were made from:

  • For all objects, this information is returned by the methods numberOfSourceCandidatePtrs() and sourceCandidatePtr(index) methods.
  • For jets, they are also accessible through the natural interface daughter(index) and numberOfDaughters()
  • For taus, they are also accessible through the more specific methods like leadChargedHadrCand(), leadNeutralCand(), leadCand(), signalCands(), isolationCands(), ...
    (note instead that the methods with similar name but with a PF inside, like signalPFCands(), are not usable on MiniAOD)
  • For muons and for electrons or photons associated to a single PF candidate, the originalObjectRef() also points to that PF candidate.
  • For electrons and photons, they're also accessible through the associatedPackedPFCandidates() method.

The packed PF candidates can be used either directly in the analysis code, e.g. to recompute isolation variables, or it is possible to use them as input to standard CMSSW tools in order to re-reconstruct jets with different algorithms or after masking the leptons, to run b-tagging, and so on.

In the code examples below we will show how to use the packed PF candidates to compute some isolation variable or some quantity based on the constituents of a jet. The more advanced CMSSW patterns are instead covered in the last section of this tutorial.

Show example CMSSW code Hide example CMSSW code
Code snippets for an EDAnalyzer 
// system include files
#include <memory>
#include <cmath>

// user include files
#include "FWCore/Framework/interface/Frameworkfwd.h"
#include "FWCore/Framework/interface/EDAnalyzer.h"
#include "FWCore/Framework/interface/Event.h"
#include "FWCore/Framework/interface/MakerMacros.h"
#include "FWCore/ParameterSet/interface/ParameterSet.h"

#include "DataFormats/Math/interface/deltaR.h"
#include "DataFormats/PatCandidates/interface/PackedCandidate.h"
#include "DataFormats/PatCandidates/interface/Muon.h"
#include "DataFormats/PatCandidates/interface/Electron.h"
#include "DataFormats/PatCandidates/interface/Jet.h"

class PackedCandAnalyzer : public edm::EDAnalyzer {
   public:
      explicit PackedCandAnalyzer(const edm::ParameterSet&);
      ~PackedCandAnalyzer() {}

   private:
      virtual void analyze(const edm::Event&, const edm::EventSetup&) override;

      edm::EDGetTokenT<pat::ElectronCollection> electronToken_;
      edm::EDGetTokenT<pat::MuonCollection> muonToken_;
      edm::EDGetTokenT<pat::JetCollection> jetToken_;
      edm::EDGetTokenT<pat::PackedCandidateCollection> pfToken_;
};

PackedCandAnalyzer::PackedCandAnalyzer(const edm::ParameterSet& iConfig):
    electronToken_(consumes<pat::ElectronCollection>(iConfig.getParameter<edm::InputTag>("electrons"))),
    muonToken_(consumes<pat::MuonCollection>(iConfig.getParameter<edm::InputTag>("muons"))),
    jetToken_(consumes<pat::JetCollection>(iConfig.getParameter<edm::InputTag>("jets"))),
    pfToken_(consumes<pat::PackedCandidateCollection>(iConfig.getParameter<edm::InputTag>("pfCands")))
{
}

void PackedCandAnalyzer::analyze(const edm::Event& iEvent, const edm::EventSetup& iSetup)
{
    edm::Handle<pat::MuonCollection> muons;
    iEvent.getByToken(muonToken_, muons);
    edm::Handle<pat::ElectronCollection> electrons;
    iEvent.getByToken(electronToken_, electrons);
    edm::Handle<pat::PackedCandidateCollection> pfs;
    iEvent.getByToken(pfToken_, pfs);
    edm::Handle<pat::JetCollection> jets;
    iEvent.getByToken(jetToken_, jets);

    std::vector<const reco::Candidate *> leptons;
    for (const pat::Muon &mu : *muons) leptons.push_back(&mu);
    for (const pat::Electron &el : *electrons) leptons.push_back(&el);

    for (const reco::Candidate *lep : leptons) {
        if (lep->pt() < 5) continue;
        // initialize sums
        double charged = 0, neutral = 0, pileup  = 0;
        // now get a list of the PF candidates used to build this lepton, so to exclude them
        std::vector<reco::CandidatePtr> footprint;
        for (unsigned int i = 0, n = lep->numberOfSourceCandidatePtrs(); i < n; ++i) {
            footprint.push_back(lep->sourceCandidatePtr(i));
        }
        // now loop on pf candidates
        for (unsigned int i = 0, n = pfs->size(); i < n; ++i) {
            const pat::PackedCandidate &pf = (*pfs)[i];
            if (deltaR(pf,*lep) < 0.2) {
                // pfcandidate-based footprint removal
                if (std::find(footprint.begin(), footprint.end(), reco::CandidatePtr(pfs,i)) != footprint.end()) {
                    continue;
                }
                if (pf.charge() == 0) {
                    if (pf.pt() > 0.5) neutral += pf.pt();
                } else if (pf.fromPV() >= 2) {
                    charged += pf.pt();
                } else {
                    if (pf.pt() > 0.5) pileup += pf.pt();
                }
            }
        }
        // do deltaBeta
        double iso = charged + std::max(0.0, neutral-0.5*pileup);
        printf("%-8s of pt %6.1f, eta %+4.2f: relIso = %5.2f\n",
                    abs(lep->pdgId())==13 ? "muon" : "electron",
                    lep->pt(), lep->eta(), iso/lep->pt());

    }

    // Let's compute the fraction of charged pt from particles with dz < 0.1 cm
    for (const pat::Jet &j :  *jets) {
        if (j.pt() < 40 || fabs(j.eta()) > 2.4) continue;
        double in = 0, out = 0; 
        for (unsigned int id = 0, nd = j.numberOfDaughters(); id < nd; ++id) {
            const pat::PackedCandidate &dau = dynamic_cast<const pat::PackedCandidate &>(*j.daughter(id));
            if (dau.charge() == 0) continue;
            (fabs(dau.dz())<0.1 ? in : out) += dau.pt();
        }
        double sum = in + out;
        printf("Jet with pt %6.1f, eta %+4.2f, beta(0.1) = %+5.3f, pileup mva disc %+.2f\n",
                j.pt(),j.eta(), sum ? in/sum : 0, j.userFloat("pileupJetId:fullDiscriminant"));
    }
}

//define this as a plug-in
DEFINE_FWK_MODULE(PackedCandAnalyzer);
Python configuration 
import FWCore.ParameterSet.Config as cms

process = cms.Process("Demo")

process.load("FWCore.MessageService.MessageLogger_cfi")
process.maxEvents = cms.untracked.PSet( input = cms.untracked.int32(10) )

process.source = cms.Source("PoolSource",
    fileNames = cms.untracked.vstring(
        '/store/cmst3/user/gpetrucc/miniAOD/v1/TT_Tune4C_13TeV-pythia8-tauola_PU_S14_PAT.root'
    )
)

process.demo = cms.EDAnalyzer("PackedCandAnalyzer",
    electrons = cms.InputTag("slimmedElectrons"),
    muons = cms.InputTag("slimmedMuons"),
    jets = cms.InputTag("slimmedJets"),
    pfCands = cms.InputTag("packedPFCandidates"),
)

process.p = cms.Path(process.demo)

Show example python code Hide example python code 
# after loading ROOT and FWlite libraries as described in the introduction
# define deltaR
from math import hypot, pi
def deltaR(a,b):
    dphi = abs(a.phi()-b.phi());
    if dphi > pi: dphi = 2*pi-dphi
    return hypot(a.eta()-b.eta(),dphi)

muons, muonLabel = Handle("std::vector<pat::Muon>"), "slimmedMuons"
electrons, electronLabel = Handle("std::vector<pat::Electron>"), "slimmedElectrons"
jets, jetLabel = Handle("std::vector<pat::Jet>"), "slimmedJets"
pfs, pfLabel = Handle("std::vector<pat::PackedCandidate>"), "packedPFCandidates"

# open file (you can use 'edmFileUtil -d /store/whatever.root' to get the physical file name)
events = Events("root://eoscms//eos/cms/store/cmst3/user/gpetrucc/miniAOD/v1/TT_Tune4C_13TeV-pythia8-tauola_PU_S14_PAT.root")

for iev,event in enumerate(events):
    event.getByLabel(muonLabel, muons)
    event.getByLabel(electronLabel, electrons)
    event.getByLabel(pfLabel, pfs)
    event.getByLabel(jetLabel, jets)

    print "\nEvent %d: run %6d, lumi %4d, event %12d" % (iev,event.eventAuxiliary().run(), event.eventAuxiliary().luminosityBlock(),event.eventAuxiliary().event())

    # Let's compute lepton PF Isolation with R=0.2, 0.5 GeV threshold on neutrals, and deltaBeta corrections
    leps  = [ p for p in muons.product() ] + [ p for p in electrons.product() ]
    for lep in leps:
        # skip those below 5 GeV, which we don't care about
        if lep.pt() < 5: continue
        # initialize sums
        charged = 0
        neutral = 0
        pileup  = 0
        # now get a list of the PF candidates used to build this lepton, so to exclude them
        footprint = set()
        for i in xrange(lep.numberOfSourceCandidatePtrs()):
            footprint.add(lep.sourceCandidatePtr(i).key()) # the key is the index in the pf collection
        # now loop on pf candidates
        for ipf,pf in enumerate(pfs.product()):
            if deltaR(pf,lep) < 0.2:
                # pfcandidate-based footprint removal
                if ipf in footprint: continue
                # add up
                if (pf.charge() == 0):
                    if pf.pt() > 0.5: neutral += pf.pt()
                elif pf.fromPV() >= 2:
                    charged += pf.pt()
                else:
                    if pf.pt() > 0.5: pileup += pf.pt()
        # do deltaBeta
        iso = charged + max(0, neutral-0.5*pileup)
        print "%-8s of pt %6.1f, eta %+4.2f: relIso = %5.2f" % (
                    "muon" if abs(lep.pdgId())==13 else "electron",
                    lep.pt(), lep.eta(), iso/lep.pt())

    # Let's compute the fraction of charged pt from particles with dz < 0.1 cm
    for i,j in enumerate(jets.product()):
        if j.pt() < 40 or abs(j.eta()) > 2.4: continue
        sums = [0,0]
        for id in xrange(j.numberOfDaughters()):
            dau = j.daughter(id)
            if (dau.charge() == 0): continue
            sums[ abs(dau.dz())<0.1 ] += dau.pt()
        sum = sums[0]+sums[1]
        print "Jet with pt %6.1f, eta %+4.2f, beta(0.1) = %+5.3f, pileup mva disc %+.2f" % (
                j.pt(),j.eta(), sums[1]/sum if sum else 0, j.userFloat("pileupJetId:fullDiscriminant"))

    if iev > 10: break

known Features in CSA14 and PHYS14

  • the charge() method returns always negative for PF candidates that are electrons or muons (the pdgId is instead correct, +11/+13 for negative leptons, -11/-13 for positive leptons)
    This can be fixed as in pull request 8584.

MC Truth

The standard PAT methods to access the MC truth are working with miniAOD, in particular you can for example use 
pat::Electron::genParticle()
pat::Muon::genParticle()
pat::Jet::partonFlavour()

The only important difference wrt normal PAT is that not all gen particles are stored. In fact gen particles are stored in two different collections as explained below.

The prunedGenParticles are standard GenParticle selected with the GenParticlePruner, in particular

  • only a subset (about 50-100 per event) of the gen particle are stored
  • the selection is made trying to keep the most important information.
  • the navigation between the particles is working but it takes "shortcuts" (i.e. you do not go through the dropped particles)
  • due to switch to Pythia8 (no more status = 3), the exact content is still being tuned
The pruned genParticles are the ones pointed by the MC matching of the high level patObjectes (e.g. pat::Electron::genParticle())

The packedGenParticles collection contains all status=1 GenParticle in a reduced format similar to the one used for PFCandidates. Besides the 4-vector a link to the first ancestor is saved in the prunedGenParticles collection. This collection is used to re-cluster the MC truth (e.g. if you want to study a different jet algorithm or some substructure techniques). Caveat: current imlpementation of the ref to first ancestor is missing a few accessors, the only access is via PackedGenParticle::mother(0) and returns a pointer. This has to be fixed in later version to give full access to the motherRef and a proper implementation of numberOfMothers (currently returning always 0)

The following examples show how to navigate between the two collections: the B mesons (500 < pdg ID < 600) are looked for in the pruned collection (because B are important particles, but they are not stable) and then we search which of the stable (status = 1) particles originated from a B decay. In order to do so we first ask to each packed candidate its "pruned" mother (NB, it is actually the first ancestor that survived the pruning), then we navigate back the pruned collection from such initial mother and see if we find the B meson.

Show example CMSSW python FWLite analyzer Hide example CMSSW python FWLite analyzer 
import ROOT
import sys
from DataFormats.FWLite import Events, Handle
from math import *

def isAncestor(a,p) :
        if a == p : 
                return True
        for i in xrange(0,p.numberOfMothers()) :
                if isAncestor(a,p.mother(i)) :
                         return True
        return False



events = Events (['root://cms-xrd-global.cern.ch//store/mc/Spring14miniaod/TTJets_MSDecaysCKM_central_Tune4C_13TeV-madgraph-tauola/MINIAODSIM/PU20bx25_POSTLS170_V5-v1/00000/F6EDDC10-8DFC-E311-BC5D-0025905A60D6.root'])

handlePruned  = Handle ("std::vector<reco::GenParticle>")
handlePacked  = Handle ("std::vector<pat::PackedGenParticle>")
labelPruned = ("prunedGenParticles")
labelPacked = ("packedGenParticles")

# loop over events
count= 0
for event in events:
    event.getByLabel (labelPacked, handlePacked)
    event.getByLabel (labelPruned, handlePruned)
    # get the product
    packed = handlePacked.product()
    pruned = handlePruned.product()
    
    for p in pruned :
        if abs(p.pdgId()) > 500 and abs(p.pdgId()) < 600 :
                print "PdgId : %s   pt : %s  eta : %s   phi : %s" %(p.pdgId(),p.pt(),p.eta(),p.phi())    
                print "     daughters"
                for pa in packed:
                        mother = pa.mother(0)
                        if mother and isAncestor(p,mother) :
                              print "     PdgId : %s   pt : %s  eta : %s   phi : %s" %(pa.pdgId(),pa.pt(),pa.eta(),pa.phi())

Show example CMSSW C++ analyzer Hide example CMSSW C++ analyzer 
// Original Author:  Andrea RIZZI
//         Created:  Mon, 07 Jul 2014 07:56:38 GMT

// system include files
#include <memory>

// user include files
#include "FWCore/Framework/interface/Frameworkfwd.h"
#include "FWCore/Framework/interface/EDAnalyzer.h"

#include "FWCore/Framework/interface/Event.h"
#include "FWCore/Framework/interface/MakerMacros.h"

#include "FWCore/ParameterSet/interface/ParameterSet.h"
#include "DataFormats/PatCandidates/interface/PackedGenParticle.h"
#include "DataFormats/Candidate/interface/Candidate.h"
#include "DataFormats/HepMCCandidate/interface/GenParticle.h"
//
// class declaration
//

class MiniAODGenPartAnalyzer : public edm::EDAnalyzer {
   public:
      explicit MiniAODGenPartAnalyzer(const edm::ParameterSet&);
      ~MiniAODGenPartAnalyzer();
      bool isAncestor(const reco::Candidate * ancestor, const reco::Candidate * particle);



   private:
      virtual void beginJob() override;
      virtual void analyze(const edm::Event&, const edm::EventSetup&) override;
      virtual void endJob() override;

      edm::EDGetTokenT<edm::View<reco::GenParticle> > prunedGenToken_;
      edm::EDGetTokenT<edm::View<pat::PackedGenParticle> > packedGenToken_;
};

MiniAODGenPartAnalyzer::MiniAODGenPartAnalyzer(const edm::ParameterSet& iConfig):
  prunedGenToken_(consumes<edm::View<reco::GenParticle> >(iConfig.getParameter<edm::InputTag>("pruned"))),
  packedGenToken_(consumes<edm::View<pat::PackedGenParticle> >(iConfig.getParameter<edm::InputTag>("packed")))
{
}


MiniAODGenPartAnalyzer::~MiniAODGenPartAnalyzer()
{
}

//Check recursively if any ancestor of particle is the given one

bool MiniAODGenPartAnalyzer::isAncestor(const reco::Candidate* ancestor, const reco::Candidate * particle)
{
//particle is already the ancestor
        if(ancestor == particle ) return true;

//otherwise loop on mothers, if any and return true if the ancestor is found
        for(size_t i=0;i< particle->numberOfMothers();i++)
        {
                if(isAncestor(ancestor,particle->mother(i))) return true;
        }
//if we did not return yet, then particle and ancestor are not relatives
        return false;
}

void
MiniAODGenPartAnalyzer::analyze(const edm::Event& iEvent, const edm::EventSetup& iSetup)
{
        using namespace edm;
        using namespace reco;
        using namespace pat;

        // Pruned particles are the one containing "important" stuff
        Handle<edm::View<reco::GenParticle> > pruned;
        iEvent.getByToken(prunedGenToken_,pruned);

        // Packed particles are all the status 1, so usable to remake jets
        // The navigation from status 1 to pruned is possible (the other direction should be made by hand)
        Handle<edm::View<pat::PackedGenParticle> > packed;
        iEvent.getByToken(packedGenToken_,packed);

        //let's try to find all status1 originating directly from a B meson decay 

        for(size_t i=0; i<pruned->size();i++){
                if(abs((*pruned)[i].pdgId()) > 500 && abs((*pruned)[i].pdgId()) <600){
                        const Candidate * bMeson = &(*pruned)[i];
                        std::cout << "PdgID: " << bMeson->pdgId() << " pt " << bMeson->pt() << " eta: " << bMeson->eta() << " phi: " << bMeson->phi() << std::endl;
                        std::cout << "  found daugthers: " << std::endl;
                        for(size_t j=0; j<packed->size();j++){
//get the pointer to the first survied ancestor of a given packed GenParticle in the prunedCollection 
                                const Candidate * motherInPrunedCollection = (*packed)[j].mother(0) ;
                                if(motherInPrunedCollection != nullptr && isAncestor( bMeson , motherInPrunedCollection)){
                                        std::cout << "     PdgID: " << (*packed)[j].pdgId() << " pt " << (*packed)[j].pt() << " eta: " << (*packed)[j].eta() << " phi: " << (*packed)[j].phi() << std::endl;
                                }
                        }
                }

        }


}


// ------------ method called once each job just before starting event loop  ------------
        void 
MiniAODGenPartAnalyzer::beginJob()
{
}

// ------------ method called once each job just after ending the event loop  ------------
        void 
MiniAODGenPartAnalyzer::endJob() 
{
}

DEFINE_FWK_MODULE(MiniAODGenPartAnalyzer);

Python cfg for cmsRun 

import FWCore.ParameterSet.Config as cms

process = cms.Process("Demo")

process.load("FWCore.MessageService.MessageLogger_cfi")

process.maxEvents = cms.untracked.PSet( input = cms.untracked.int32(-1) )

process.source = cms.Source("PoolSource",
    fileNames = cms.untracked.vstring("root://cms-xrd-global.cern.ch//store/mc/Spring14miniaod/TTJets_MSDecaysCKM_central_Tune4C_13TeV-madgraph-tauola/MINIAODSIM/PU20bx25_POSTLS170_V5-v1/00000/F6EDDC10-8DFC-E311-BC5D-0025905A60D6.root")
)

process.demo = cms.EDAnalyzer('MiniAODGenPartAnalyzer',
packed = cms.InputTag("packedGenParticles"),
pruned = cms.InputTag("prunedGenParticles")
)


process.p = cms.Path(process.demo)

Please note that you need to the following libraries in your BuildFile.xml 

DataFormats/PatCandidates
DataFormats/HepMCCandidate

GenJets

One collections of gen jets is saved, slimmedGenJets, made from ak4GenJets, i.e. using anti-kT clustering out of stable gen particles. In CMSSW 70X (CSA14) and 72X (Phys14) campaign, neutrinos are included in the gen jets, while starting from 74X neutrinos and other invisible BSM particles are excluded.

Links to the daughters are available through the daughter(idx) and numberOfDaughters() methods, and they will point into the collection of packed gen candidates described below.
It should be noted that the getGenConstituents method instead will NOT work, since they explicitly require the daughter to be a full reco::GenParticle and not a packed one.

known Features in CSA14 and PHYS14

  • in 72X and earlier, packed gen particles are saved after a cut |η|<6, but the jet clustering uses rapidity instead of η, so slimmed gen jets within the detector acceptance can still contain some daughter references that are not saved in the packed gen particles. The isAvailable method of the reco::CandidatePtr returned by daughterPtr(index) can be used to check if a gen particle is available.
    This is fixed in 74X and later, in which packed gen particles are saved after |y|<6 instead of |η|<6.

Trigger

Trigger bits for each path are stored as edm::TriggerResults, as standard in CMSSW.

In order to access paths by trigger name, you can get an edm::TriggerNames instance from the Event (both in full framework and in fwlite, see example at the bottom of the section), which translates indices in names. The mapping is the same in all events sharing the same trigger configuration (which is identified by the parameterSetID() method of the trigger results object).

Trigger objects associated to filters in the HLT are saved as instances of pat::TriggerObjectStandAlone, so that they can be used e.g. for trigger matching and measurements of trigger efficiency with the tag-and-probe method. The trigger names stored in the objects are packed, but they can be unpacked by passing an edm::TriggerNames instance, as shown in the example code at the bottom of this section.

Trigger prescales are encoded for each path into an pat::PackedTriggerPrescales, which can be unpacked with an edm::TriggerNames if access by path name is needed, just like for the TriggerResults.

Some extra trigger-level information copied directly from the AOD could also be available, namely the l1extra and L1GlobalTriggerReadoutRecord, though they might be missing in some versions of the miniAODs (CMSSW_7_0_6_patch1 selects them only if they're produced in the HLT process, which might not be the case in production; this will be addressed in future miniAOD versions)

Show example CMSSW code Hide example CMSSW code
EDAnalyzer code 
// system include files
#include <memory>
#include <cmath>

// user include files
#include "FWCore/Framework/interface/Frameworkfwd.h"
#include "FWCore/Framework/interface/EDAnalyzer.h"
#include "FWCore/Framework/interface/Event.h"
#include "FWCore/Framework/interface/MakerMacros.h"
#include "FWCore/ParameterSet/interface/ParameterSet.h"

#include "DataFormats/Math/interface/deltaR.h"
#include "FWCore/Common/interface/TriggerNames.h"
#include "DataFormats/Common/interface/TriggerResults.h"
#include "DataFormats/PatCandidates/interface/TriggerObjectStandAlone.h"
#include "DataFormats/PatCandidates/interface/PackedTriggerPrescales.h"

class MiniAODTriggerAnalyzer : public edm::EDAnalyzer {
   public:
      explicit MiniAODTriggerAnalyzer(const edm::ParameterSet&);
      ~MiniAODTriggerAnalyzer() {}

   private:
      virtual void analyze(const edm::Event&, const edm::EventSetup&) override;

      edm::EDGetTokenT<edm::TriggerResults> triggerBits_;
      edm::EDGetTokenT<pat::TriggerObjectStandAloneCollection> triggerObjects_;
      edm::EDGetTokenT<pat::PackedTriggerPrescales> triggerPrescales_;
};

MiniAODTriggerAnalyzer::MiniAODTriggerAnalyzer(const edm::ParameterSet& iConfig):
    triggerBits_(consumes<edm::TriggerResults>(iConfig.getParameter<edm::InputTag>("bits"))),
    triggerObjects_(consumes<pat::TriggerObjectStandAloneCollection>(iConfig.getParameter<edm::InputTag>("objects"))),
    triggerPrescales_(consumes<pat::PackedTriggerPrescales>(iConfig.getParameter<edm::InputTag>("prescales")))
{
}

void MiniAODTriggerAnalyzer::analyze(const edm::Event& iEvent, const edm::EventSetup& iSetup)
{
    edm::Handle<edm::TriggerResults> triggerBits;
    edm::Handle<pat::TriggerObjectStandAloneCollection> triggerObjects;
    edm::Handle<pat::PackedTriggerPrescales> triggerPrescales;

    iEvent.getByToken(triggerBits_, triggerBits);
    iEvent.getByToken(triggerObjects_, triggerObjects);
    iEvent.getByToken(triggerPrescales_, triggerPrescales);

    const edm::TriggerNames &names = iEvent.triggerNames(*triggerBits);
    std::cout << "\n === TRIGGER PATHS === " << std::endl;
    for (unsigned int i = 0, n = triggerBits->size(); i < n; ++i) {
        std::cout << "Trigger " << names.triggerName(i) << 
                ", prescale " << triggerPrescales->getPrescaleForIndex(i) <<
                ": " << (triggerBits->accept(i) ? "PASS" : "fail (or not run)") 
                << std::endl;
    }
    std::cout << "\n === TRIGGER OBJECTS === " << std::endl;
    for (pat::TriggerObjectStandAlone obj : *triggerObjects) { // note: not "const &" since we want to call unpackPathNames
        obj.unpackPathNames(names);
        std::cout << "\tTrigger object:  pt " << obj.pt() << ", eta " << obj.eta() << ", phi " << obj.phi() << std::endl;
        // Print trigger object collection and type
        std::cout << "\t   Collection: " << obj.collection() << std::endl;
        std::cout << "\t   Type IDs:   ";
        for (unsigned h = 0; h < obj.filterIds().size(); ++h) std::cout << " " << obj.filterIds()[h] ;
        std::cout << std::endl;
        // Print associated trigger filters
        std::cout << "\t   Filters:    ";
        for (unsigned h = 0; h < obj.filterLabels().size(); ++h) std::cout << " " << obj.filterLabels()[h];
        std::cout << std::endl;
        std::vector<std::string> pathNamesAll  = obj.pathNames(false);
        std::vector<std::string> pathNamesLast = obj.pathNames(true);
        // Print all trigger paths, for each one record also if the object is associated to a 'l3' filter (always true for the
        // definition used in the PAT trigger producer) and if it's associated to the last filter of a successfull path (which
        // means that this object did cause this trigger to succeed; however, it doesn't work on some multi-object triggers)
        std::cout << "\t   Paths (" << pathNamesAll.size()<<"/"<<pathNamesLast.size()<<"):    ";
        for (unsigned h = 0, n = pathNamesAll.size(); h < n; ++h) {
            bool isBoth = obj.hasPathName( pathNamesAll[h], true, true ); 
            bool isL3   = obj.hasPathName( pathNamesAll[h], false, true ); 
            bool isLF   = obj.hasPathName( pathNamesAll[h], true, false ); 
            bool isNone = obj.hasPathName( pathNamesAll[h], false, false ); 
            std::cout << "   " << pathNamesAll[h];
            if (isBoth) std::cout << "(L,3)";
            if (isL3 && !isBoth) std::cout << "(*,3)";
            if (isLF && !isBoth) std::cout << "(L,*)";
            if (isNone && !isBoth && !isL3 && !isLF) std::cout << "(*,*)";
        }
        std::cout << std::endl;
    }
    std::cout << std::endl;

}

//define this as a plug-in
DEFINE_FWK_MODULE(MiniAODTriggerAnalyzer);
Python configuration 
import FWCore.ParameterSet.Config as cms

process = cms.Process("Demo")

process.load("FWCore.MessageService.MessageLogger_cfi")
process.maxEvents = cms.untracked.PSet( input = cms.untracked.int32(10) )

process.source = cms.Source("PoolSource",
    fileNames = cms.untracked.vstring(
        '/store/cmst3/user/gpetrucc/miniAOD/v1/TT_Tune4C_13TeV-pythia8-tauola_PU_S14_PAT.root'
    )
)

process.demo = cms.EDAnalyzer("MiniAODTriggerAnalyzer",
    bits = cms.InputTag("TriggerResults","","HLT"),
    prescales = cms.InputTag("patTrigger"),
    objects = cms.InputTag("selectedPatTrigger"),
)

process.p = cms.Path(process.demo)

Show example python code Hide example python code 
# import ROOT in batch mode
import sys
oldargv = sys.argv[:]
sys.argv = [ '-b-' ]
import ROOT
ROOT.gROOT.SetBatch(True)
sys.argv = oldargv

# load FWLite C++ libraries
ROOT.gSystem.Load("libFWCoreFWLite.so");
ROOT.gSystem.Load("libDataFormatsFWLite.so");
ROOT.AutoLibraryLoader.enable()

# load FWlite python libraries
from DataFormats.FWLite import Handle, Events

triggerBits, triggerBitLabel = Handle("edm::TriggerResults"), ("TriggerResults","","HLT")
triggerObjects, triggerObjectLabel  = Handle("std::vector<pat::TriggerObjectStandAlone>"), "selectedPatTrigger"
triggerPrescales, triggerPrescaleLabel  = Handle("pat::PackedTriggerPrescales"), "patTrigger"

# open file (you can use 'edmFileUtil -d /store/whatever.root' to get the physical file name)
events = Events("root://eoscms//eos/cms/store/cmst3/user/gpetrucc/miniAOD/v1/TT_Tune4C_13TeV-pythia8-tauola_PU_S14_PAT.root")

for iev,event in enumerate(events):
    event.getByLabel(triggerBitLabel, triggerBits)
    event.getByLabel(triggerObjectLabel, triggerObjects)
    event.getByLabel(triggerPrescaleLabel, triggerPrescales)

    print "\nEvent %d: run %6d, lumi %4d, event %12d" % (iev,event.eventAuxiliary().run(), event.eventAuxiliary().luminosityBlock(),event.eventAuxiliary().event())
    print "\n === TRIGGER PATHS ==="
    names = event.object().triggerNames(triggerBits.product())
    for i in xrange(triggerBits.product().size()):
        print "Trigger ", names.triggerName(i), ", prescale ", triggerPrescales.product().getPrescaleForIndex(i), ": ", ("PASS" if triggerBits.product().accept(i) else "fail (or not run)") 

    print "\n === TRIGGER OBJECTS ==="
    for j,to in enumerate(triggerObjects.product()):
        to.unpackPathNames(names);
        print "Trigger object pt %6.2f eta %+5.3f phi %+5.3f  " % (to.pt(),to.eta(),to.phi())
        print "         collection: ", to.collection()
        print "         type ids: ", ", ".join([str(f) for f in to.filterIds()])
        print "         filters: ", ", ".join([str(f) for f in to.filterLabels()])
        pathslast = set(to.pathNames(True))
        print "         paths:   ", ", ".join([("%s*" if f in pathslast else "%s")%f for f in to.filterLabels()]) 
    if iev > 10: break

Miscellanea

Primary vertices and BeamSpot

The BeamSpot is copied as is from the AOD in the offlineBeamSpot collection.

For the primary vertices, a collection offlineSlimmedPrimaryVertices is provided, that has the same content as the AOD offlinePrimaryVertices collection except that vertices don't contain track references anymore (since there's not tracks in MiniAOD anyway).

ETmiss filters

Some filters for ETmiss cleaning are run at MiniAOD production stage, and saved as a trigger results (an edm::TriggerResults of the PAT process); the individual filters can be accessed with the same code as for the HLT paths, except of course for the process name.

The list of all the paths and filters included and their definitions can be found in PhysicsTools/PatAlgos/python/slimming/metFilterPaths_cff.py.

In addition, just like in AOD the HCal noise information is provided in a HcalNoiseSummary object hcalnoise .

Advanced topics: re-clustering, event interpretation

The following example shows how to re-run jet reconstruction with a different algorithm (in the example using AK5 instead of miniAOD default that is ak4). The example also shows how to rerun b-tagging: in order to rerun btagging a dedicate unpacker is used to re-create a generalTracks like collection and to properly relink PV and IVF vertices collections to those tracks (this will not be needed anymore in 720)

Show example CMSSW pytohn configuration Hide example CMSSW python configuration 
import FWCore.ParameterSet.Config as cms


process = cms.Process("EX")
process.source = cms.Source("PoolSource",
    fileNames = cms.untracked.vstring("root://cms-xrd-global.cern.ch//store/mc/Spring14miniaod/TTJets_MSDecaysCKM_central_Tune4C_13TeV-madgraph-tauola/MINIAODSIM/PU20bx25_POSTLS170_V5-v1/00000/F6EDDC10-8DFC-E311-BC5D-0025905A60D6.root")
)
process.maxEvents = cms.untracked.PSet( input = cms.untracked.int32(10) )

process.OUT = cms.OutputModule("PoolOutputModule",
    fileName = cms.untracked.string('test.root'),
    outputCommands = cms.untracked.vstring(['drop *','keep *_*_*_PAT']) #keep all input miniAOD content, more stuff added below
)
process.endpath= cms.EndPath(process.OUT)

from RecoJets.JetProducers.ak5PFJets_cfi import ak5PFJets
from RecoJets.JetProducers.ak5GenJets_cfi import ak5GenJets
from RecoMET.METProducers.PFMET_cfi import pfMet

# Select candidates that would pass CHS requirements
process.chs = cms.EDFilter("CandPtrSelector", src = cms.InputTag("packedPFCandidates"), cut = cms.string("fromPV"))

#makes chs ak5 jets   (instead of ak4 that are default in miniAOD 70X)
process.ak5PFJetsCHS = ak5PFJets.clone(src = 'chs')
#let's also make non-chs ak5 jets
process.ak5PFJets = ak5PFJets.clone(src = 'packedPFCandidates') 
process.OUT.outputCommands.append("keep *_ak5PFJets_*_EX")
process.OUT.outputCommands.append("keep *_ak5PFJetsCHS_*_EX")

# if we really want to study the ak5 vs ak4 we may need to remake gen Jets in ak5 too
# in order to do so we use the "all status 1" gen particles (packedGenParticles)
process.ak5GenJets = ak5GenJets.clone(src = 'packedGenParticles')
process.OUT.outputCommands.append("keep *_ak5GenJets_*_EX")

# the following part is needed if you want to run b-tagging on the freshly made jets
# CAVEAT: it is not 100% the same b-tagging as in RECO, but performance plots are almost identical

# As tracks are not stored in miniAOD, and b-tag fwk for CMSSW < 72X does not accept candidates
# we need to recreate tracks and pv for btagging in standard reco format:
process.load('RecoBTag.Configuration.RecoBTag_cff')
process.load('RecoJets.Configuration.RecoJetAssociations_cff')
process.load('PhysicsTools.PatAlgos.slimming.unpackedTracksAndVertices_cfi')
process.load('Configuration.StandardSequences.FrontierConditions_GlobalTag_cff')
process.load('Configuration.StandardSequences.Geometry_cff')
process.load('Configuration.StandardSequences.MagneticField_38T_cff')
#process.load('TrackingTools.TransientTrack.TransientTrackBuilder_cfi')
process.GlobalTag.globaltag = 'PLS170_V7AN1::All'

process.ak5JetTracksAssociatorAtVertexPF.jets = cms.InputTag("ak5PFJetsCHS")
process.ak5JetTracksAssociatorAtVertexPF.tracks = cms.InputTag("unpackedTracksAndVertices")
process.impactParameterTagInfos.primaryVertex = cms.InputTag("unpackedTracksAndVertices")
process.inclusiveSecondaryVertexFinderTagInfos.extSVCollection = cms.InputTag("unpackedTracksAndVertices","secondary","")
process.combinedSecondaryVertex.trackMultiplicityMin = 1

process.OUT.outputCommands.append("keep *_*VertexBJetTags_*_EX") #runs all Vertex btaggers 



process.options = cms.untracked.PSet( 
        wantSummary = cms.untracked.bool(True), # while the timing of this is not reliable in unscheduled mode, it still helps understanding what was actually run 
        allowUnscheduled = cms.untracked.bool(True)
)

Event interpretations can be easily run starting from miniAOD with the following differences with respect to standard EI

  • The input candidates are the packedCandidates
  • The selector should be CandPtrSelector so that the Ptr to the original candidates are always kept
  • The projector to use should be CandPtrProjector
In addition, in order to run btagging for CMSSW < 72X, a dedicated unpacker (unpackedTracksAndVertices) is needed to re-create tracks and vertices in a format compatible with b-tagging framework.

The following code shows a commented example of EI configuration.

Show example CMSSW pytohn configuration Hide example CMSSW python configuration 
import FWCore.ParameterSet.Config as cms

process = cms.Process("S2")
# Input source
process.source = cms.Source("PoolSource",
    fileNames = cms.untracked.vstring("root://cms-xrd-global.cern.ch//store/mc/Spring14miniaod/TTJets_MSDecaysCKM_central_Tune4C_13TeV-madgraph-tauola/MINIAODSIM/PU20bx25_POSTLS170_V5-v1/00000/F6EDDC10-8DFC-E311-BC5D-0025905A60D6.root")
)

process.maxEvents = cms.untracked.PSet( input = cms.untracked.int32(-1) )



#select isolated  muons and electrons collections
#tune the requirements to whatever ID and isolation you prefer 

process.selectedMuons = cms.EDFilter("CandPtrSelector", src = cms.InputTag("slimmedMuons"), cut = cms.string('''abs(eta)<2.5 && pt>10. &&
   (pfIsolationR04().sumChargedHadronPt+
    max(0.,pfIsolationR04().sumNeutralHadronEt+
    pfIsolationR04().sumPhotonEt-
    0.50*pfIsolationR04().sumPUPt))/pt < 0.20 && 
    (isPFMuon && (isGlobalMuon || isTrackerMuon) )'''))

process.selectedElectrons = cms.EDFilter("CandPtrSelector", src = cms.InputTag("slimmedElectrons"), cut = cms.string('''abs(eta)<2.5 && pt>20. &&
    gsfTrack.isAvailable() &&
    gsfTrack.trackerExpectedHitsInner.numberOfLostHits<2 &&
    (pfIsolationVariables().sumChargedHadronPt+
    max(0.,pfIsolationVariables().sumNeutralHadronEt+
    pfIsolationVariables().sumPhotonEt-
    0.5*pfIsolationVariables().sumPUPt))/pt < 0.15'''))

### Do "projections"
# first select the packedCandidates passing the loose "fromPV()" requirement (equivalent to CHS definition used for Jets in Run I)
process.pfCHS = cms.EDFilter("CandPtrSelector", src = cms.InputTag("packedPFCandidates"), cut = cms.string("fromPV"))
# then remove the previously selected muons
process.pfNoMuonCHS =  cms.EDProducer("CandPtrProjector", src = cms.InputTag("pfCHS"), veto = cms.InputTag("selectedMuons"))
# then remove the previously selected electrons
process.pfNoElectronsCHS = cms.EDProducer("CandPtrProjector", src = cms.InputTag("pfNoMuonCHS"), veto =  cms.InputTag("selectedElectrons"))

#Import RECO jet producer for ak4 PF and GEN jet
from RecoJets.JetProducers.ak4PFJets_cfi import ak4PFJets
from RecoJets.JetProducers.ak4GenJets_cfi import ak4GenJets
process.ak4PFJetsCHS = ak4PFJets.clone(src = 'pfNoElectronsCHS', doAreaFastjet = True)
process.ak4GenJets = ak4GenJets.clone(src = 'packedGenParticles')


# The following is make patJets, but EI is done with the above
process.load("PhysicsTools.PatAlgos.producersLayer1.patCandidates_cff")
process.load("Configuration.EventContent.EventContent_cff")
process.load('Configuration.StandardSequences.Geometry_cff')
process.load('Configuration.StandardSequences.MagneticField_38T_cff')
process.load('Configuration.StandardSequences.FrontierConditions_GlobalTag_cff')
process.GlobalTag.globaltag = 'PLS170_V7AN1::All'


from PhysicsTools.PatAlgos.tools.jetTools import addJetCollection
addJetCollection(
   process,
   postfix   = "",
   labelName = 'AK4PFCHS',
   jetSource = cms.InputTag('ak4PFJetsCHS'),
   trackSource = cms.InputTag('unpackedTracksAndVertices'), 
   pvSource = cms.InputTag('unpackedTracksAndVertices'), 
   jetCorrections = ('AK4PFchs', cms.vstring(['L1FastJet', 'L2Relative', 'L3Absolute']), 'Type-2'),
   btagDiscriminators = [      'combinedSecondaryVertexBJetTags'     ]
   ,algo= 'AK', rParam = 0.4
   )

#adjust MC matching
process.patJetGenJetMatchAK4PFCHS.matched = "ak4GenJets"
process.patJetPartonMatchAK4PFCHS.matched = "prunedGenParticles"
process.patJetPartons.particles = "prunedGenParticles"

#adjust PV used for Jet Corrections
process.patJetCorrFactorsAK4PFCHS.primaryVertices = "offlineSlimmedPrimaryVertices"

# the following part is needed if you want to run b-tagging on the freshly made jets
# CAVEAT: it is not 100% the same b-tagging as in RECO, but performance plots are almost identical

# As tracks are not stored in miniAOD, and b-tag fwk for CMSSW < 72X does not accept candidates
# we need to recreate tracks and pv for btagging in standard reco format:
process.load('PhysicsTools.PatAlgos.slimming.unpackedTracksAndVertices_cfi')
process.combinedSecondaryVertex.trackMultiplicityMin = 1  #needed for CMSSW < 71X


#new PAT default running is "unscheduled" so we just need to say in the outputCommands what we want to store
process.options = cms.untracked.PSet(
        allowUnscheduled = cms.untracked.bool(True)
)

process.OUT = cms.OutputModule("PoolOutputModule",
    fileName = cms.untracked.string('test.root'),
    outputCommands = cms.untracked.vstring(['drop *','keep patJets_patJetsAK4PFCHS_*_*','keep *_*_*_PAT'])
)
process.endpath= cms.EndPath(process.OUT)

Producing MiniAOD

To run miniAOD, use cmsDriver to create a cfg file, specifying the kind of data and the global tag

cmsDriver.py miniAOD-prod -s PAT --eventcontent [ MINIAOD | MINIAODSIM ] --runUnscheduled [ --data | --mc [ --fast ] ] --filein xxx  --conditions yyy --no_exec
(the --fast is needed on fast sim to switch off some MET filters)

for example, for Phys14 MC production in CMSSW_7_2_0 

cmsDriver.py miniAOD-prod -s PAT --eventcontent MINIAODSIM --runUnscheduled --mc --filein /store/mc/Phys14DR/TTJets_MSDecaysCKM_central_Tune4C_13TeV-madgraph-tauola/AODSIM/PU20bx25_PHYS14_25_V1-v1/00000/0008EBB4-2475-E411-9F25-0025907DC9F8.root --conditions PHYS14_25_V1 -n 100 --no_exec
note: apparently in 7.2.X and later, the --conditions option take global tags without the extra ::All postifx that was there in older releases

The global tags, from https://twiki.cern.ch/twiki/bin/view/CMSPublic/SWGuideFrontierConditions#miniAOD_production are:

GT name Comments
PHYS14_25_V2 Phys14 GT for 25ns in 72X (asymptotic alignment and calibration scenario), with the new Phys14 jet energy
PHYS14_25_V1 Phys14 GT for 25ns in 72X (asymptotic alignment and calibration scenario), with CSA14 jet energy corrections as used in the central Phys14 production
PLS170_V7AN1 CSA14 GT for 25ns in 70X (asymptotic alignment and calibration scenario), as POSTLS170_V7 + updated JEC
PLS170_V6AN1 CSA14 GT for 50ns in 70X (more pessimistic alignment and calibration scenario), as POSTLS170_V6 + updated JEC
GR_70_V2_AN1 CSA14 GT for data in 70X, as GR_R_70_V2 + updated JEC

-- AndreaRizzi - 2015-05-26

Edit | Attach | Watch | Print version | History: r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r1 - 2015-05-26 - AndreaRizzi
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright &© 2008-2023 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