def arbitrate(self):
pi1 = ROOT.Math.LorentzVector('<ROOT::Math::PxPyPzM4D<double> >')(
self.tk1_.px(), self.tk1_.py(), self.tk1_.pz(), m_pi_ch)
pi2 = ROOT.Math.LorentzVector('<ROOT::Math::PxPyPzM4D<double> >')(
self.tk2_.px(), self.tk2_.py(), self.tk2_.pz(), m_pi_ch)
k1 = ROOT.Math.LorentzVector('<ROOT::Math::PxPyPzM4D<double> >')(
self.tk1_.px(), self.tk1_.py(), self.tk1_.pz(), m_k_ch)
k2 = ROOT.Math.LorentzVector('<ROOT::Math::PxPyPzM4D<double> >')(
self.tk2_.px(), self.tk2_.py(), self.tk2_.pz(), m_k_ch)
# assign the mass hypotheses
if abs((pi1 + k2).mass() - m_k_st_zero) < abs((pi2 + k1).mass() -
m_k_st_zero):
self.pi_ = PhysicsObject(
ROOT.pat.PackedCandidate(self.tk1_.physObj))
self.k_ = PhysicsObject(ROOT.pat.PackedCandidate(
self.tk2_.physObj))
else:
self.pi_ = PhysicsObject(
ROOT.pat.PackedCandidate(self.tk2_.physObj))
self.k_ = PhysicsObject(ROOT.pat.PackedCandidate(
self.tk1_.physObj))
self.pi_.setMass(m_pi_ch)
self.k_.setMass(m_k_ch)
self.pi_.setPdgId(self.pi_.charge() * 211)
self.k_.setPdgId(self.k_.charge() * 321)
def __init__(self, leg1, leg2, pdgid, status=3, sort_by_pt=False, mass1=-1, mass2=-1):
'''
Parameters (stored as attributes):
leg1,2 : first and second leg.
pdgid : pdg code of the resonance
status : status code of the resonance
'''
if isinstance(leg1.physObj, ROOT.pat.PackedCandidate): leg1 = PhysicsObject(ROOT.pat.PackedCandidate(leg1.physObj))
if isinstance(leg2.physObj, ROOT.pat.PackedCandidate): leg2 = PhysicsObject(ROOT.pat.PackedCandidate(leg2.physObj))
if sort_by_pt:
self.leg1 = leg1 if leg1.pt() >= leg2.pt() else leg2
self.leg2 = leg2 if leg1.pt() >= leg2.pt() else leg1
else:
self.leg1 = leg1
self.leg2 = leg2
if mass1>0: self.leg1.setMass(mass1)
if mass2>0: self.leg2.setMass(mass2)
self._p4 = leg1.p4() + leg2.p4()
self._charge = leg1.charge() + leg2.charge()
self._pdgid = pdgid
self._status = status
def process(self, event):
run = event.input.eventAuxiliary().id().run()
lumi = event.input.eventAuxiliary().id().luminosityBlock()
eventId = event.input.eventAuxiliary().id().event()
self.readCollections( event.input )
# Will need who for jet calibration later
rho = self.handles["rho"].product()[0]
########
# AK8 Jets from MiniAOD + Subjet btags
########
setattr(event, "ak08", map(PhysicsObject, self.handles["ak08"].product()))
setattr(event, "ak0softdropsubjets", map(PhysicsObject, self.handles["ak08softdropsubjets"].product()))
# bb-tag Output
newtags = self.handles['ak08bbtag'].product()
# Loop over jets
for ij, jet in enumerate(getattr(event, "ak08")):
# Fill bb-tag
for i in xrange(len(newtags)) :
if jet.physObj == newtags.key(i).get():
jet.bbtag = newtags.value(i)
# bb-tag Inputs
muonTagInfos = self.handles['ak08muonTagInfos'].product()[ij]
elecTagInfos = self.handles['ak08elecTagInfos'].product()[ij]
ipTagInfo = self.handles['ak08ipTagInfos'].product()[ij]
svTagInfo = self.handles['ak08svTagInfos'].product()[ij]
calcBBTagVariables(jet,
muonTagInfos,
elecTagInfos,
ipTagInfo,
svTagInfo,
njettiness_08,
maxSVDeltaRToJet = 0.7,)
# end of loop over jets
########
# Ungroomed Fatjets + NSubjettiness + Hbb Tagging
########
for prefix in ["ca15"]:
if self.skip_ca15 and ("ca15" in prefix):
continue
# N-Subjettiness
tau1 = self.handles[prefix+'tau1'].product()
tau2 = self.handles[prefix+'tau2'].product()
tau3 = self.handles[prefix+'tau3'].product()
# bb-tag Output
newtags = self.handles[prefix+'bbtag'].product()
# Four Vector
setattr(event, prefix+"ungroomed", map(PhysicsObject, self.handles[prefix+'ungroomed'].product()))
# Loop over jets
for ij, jet in enumerate(getattr(event, prefix+"ungroomed")):
# Fill N-Subjettiness
jet.tau1 = tau1.get(ij)
jet.tau2 = tau2.get(ij)
jet.tau3 = tau3.get(ij)
# Fill bb-tag
for i in xrange(len(newtags)) :
if jet.physObj == newtags.key(i).get():
jet.bbtag = newtags.value(i)
# bb-tag Inputs
muonTagInfos = self.handles['ca15muonTagInfos'].product()[ij]
elecTagInfos = self.handles['ca15elecTagInfos'].product()[ij]
ipTagInfo = self.handles['ca15ipTagInfos'].product()[ij]
svTagInfo = self.handles['ca15svTagInfos'].product()[ij]
calcBBTagVariables(jet,
muonTagInfos,
elecTagInfos,
ipTagInfo,
svTagInfo,
njettiness_15,
maxSVDeltaRToJet = 1.3)
# end of loop over jets
########
# Softdrop Fatjets + NSubjettiness
########
for fj_name in ["ca15softdropz2b1"]:
if self.skip_ca15 and ("ca15" in fj_name):
continue
# Set the four-vector
setattr(event, fj_name, map(PhysicsObject, self.handles[fj_name].product()))
# N-Subjettiness
tau1 = self.handles[fj_name+'tau1'].product()
tau2 = self.handles[fj_name+'tau2'].product()
tau3 = self.handles[fj_name+'tau3'].product()
# Loop over jets
for ij, jet in enumerate(getattr(event, fj_name)):
# Fill N-Subjettiness
jet.tau1 = tau1.get(ij)
jet.tau2 = tau2.get(ij)
jet.tau3 = tau3.get(ij)
# end of loop over jets
########
# Groomed Uncalibrated Fatjets
########
for fj_name in ['ak08pruned', 'ca15trimmed', 'ca15softdrop', 'ca15pruned']:
if self.skip_ca15 and ("ca15" in fj_name):
continue
setattr(event, fj_name, map(PhysicsObject, self.handles[fj_name].product()))
########
# Groomed Fatjets to calibrate
########
pruned_cal_jets = []
for groomed_fj in self.handles['ak08pruned'].product():
# We need the closest ungroomed fatjet to get the JEC:
# - Make a list of pairs: deltaR(ungroomed fj, groomed fj) for all ungroomed fatjets
# - Sort by deltaR
# - And take the minimum
if len(getattr(event, "ak08")):
closest_ung_fj_and_dr = sorted(
[(ung_fj, deltaR2(ung_fj, groomed_fj)) for ung_fj in getattr(event, "ak08")],
key=lambda x:x[1])[0]
else:
print "WARNING: No ungroomed fatjets found in event with groomed fatjet. Skipping"
continue
# Use the jet cone size for matching
minimal_dr_groomed_ungroomed = 0.8
if closest_ung_fj_and_dr[1] > minimal_dr_groomed_ungroomed:
print "WARNING: No ungroomed fatjet found close to groomed fatjet. Skipping"
continue
ungroomed_jet = Jet(closest_ung_fj_and_dr[0])
c = self.jetReCalibrator.getCorrection(ungroomed_jet, rho)
# Need to do a deep-copy. Otherwise the original jet will be modified
cal_groomed_fj = PhysicsObject(groomed_fj).__copy__()
cal_groomed_fj.scaleEnergy(c)
pruned_cal_jets.append(cal_groomed_fj)
setattr(event, 'ak08prunedcal', pruned_cal_jets)
########
# AK08 Regression
#######
self.regressionsAK08[0].evaluateRegressionAK08(event)
########
# Subjets
########
for fj_name in ['ak08pruned','ca15pruned']:
if self.skip_ca15 and ("ca15" in fj_name):
continue
setattr(event, fj_name + "subjets", map(PhysicsObject, self.handles[fj_name+"subjets"].product()))
newtags = self.handles[fj_name+'subjetbtag'].product()
for i in xrange(0,len(newtags)) :
for j in getattr(event, fj_name+"subjets"):
if j.physObj == newtags.key(i).get():
j.btag = newtags.value(i)
########
# HEPTopTagger
########
if not self.skip_ca15:
candJets = self.handles['httCandJets'].product()
candInfos = self.handles['httCandInfos'].product()
event.httCandidates = map(PhysicsObject, candJets)
sjbtags = self.handles['httSubjetBtags'].product()
for i in xrange(0, len(candJets)):
event.httCandidates[i].fRec = candInfos[i].properties().fRec
event.httCandidates[i].Ropt = candInfos[i].properties().Ropt
event.httCandidates[i].RoptCalc = candInfos[i].properties().RoptCalc
event.httCandidates[i].ptForRoptCalc = candInfos[i].properties().ptForRoptCalc
# HTT return the subjet-pair closest to the W-mass as W-subjets
# Could be improved by b-tagging if we run into a problem
[sj_w1, sj_w2, sj_nonw] = [con.__deref__() for con in candJets[i].getJetConstituents() if not con.isNull()]
event.httCandidates[i].sjW1pt = sj_w1.pt()
event.httCandidates[i].sjW1eta = sj_w1.eta()
event.httCandidates[i].sjW1phi = sj_w1.phi()
event.httCandidates[i].sjW1mass = sj_w1.mass()
# Get the correct b-tag
for ib in xrange(0, len(sjbtags)) :
if sj_w1 == sjbtags.key(ib).get():
event.httCandidates[i].sjW1btag = sjbtags.value(ib)
event.httCandidates[i].sjW2pt = sj_w2.pt()
event.httCandidates[i].sjW2eta = sj_w2.eta()
event.httCandidates[i].sjW2phi = sj_w2.phi()
event.httCandidates[i].sjW2mass = sj_w2.mass()
# Get the correct b-tag
for ib in xrange(0, len(sjbtags)) :
if sj_w2 == sjbtags.key(ib).get():
event.httCandidates[i].sjW2btag = sjbtags.value(ib)
event.httCandidates[i].sjNonWpt = sj_nonw.pt()
event.httCandidates[i].sjNonWeta = sj_nonw.eta()
event.httCandidates[i].sjNonWphi = sj_nonw.phi()
event.httCandidates[i].sjNonWmass = sj_nonw.mass()
# Get the correct b-tag
for ib in xrange(0, len(sjbtags)) :
if sj_nonw == sjbtags.key(ib).get():
event.httCandidates[i].sjNonWbtag = sjbtags.value(ib)
return True
def genMatch(event,
leg,
ptSelGentauleps,
ptSelGenleps,
ptSelGenSummary,
dR=0.2,
matchAll=True):
dR2 = dR * dR
leg.isTauHad = False
leg.isTauLep = False
leg.isPromptLep = False
leg.genp = None
best_dr2 = dR2
# if hasattr(leg, 'genJet') and leg.genJet():
# if leg.genJet().pt() > 15.:
# dr2 = deltaR2(leg.eta(), leg.phi(), leg.genJet().eta(), leg.genJet().phi())
# if dr2 < best_dr2:
# best_dr2 = dr2
# leg.genp = leg.genJet()
# leg.genp.setPdgId(-15 * leg.genp.charge())
# leg.isTauHad = True
l1match, dR2best = bestMatch(leg, event.genTauJets)
if dR2best < best_dr2:
best_dr2 = dR2best
leg.genp = GenParticle(l1match)
leg.genp.setPdgId(-15 * leg.genp.charge())
leg.isTauHad = True
# to generated leptons from taus
l1match, dR2best = bestMatch(leg, ptSelGentauleps)
if dR2best < best_dr2:
best_dr2 = dR2best
leg.genp = l1match
leg.isTauLep = True
leg.isTauHad = False
# to generated prompt leptons
l1match, dR2best = bestMatch(leg, ptSelGenleps)
if dR2best < best_dr2:
best_dr2 = dR2best
leg.genp = l1match
leg.isPromptLep = True
leg.isTauLep = False
leg.isTauHad = False
if best_dr2 < dR2:
return
# match with any other relevant gen particle
if matchAll:
l1match, dR2best = bestMatch(leg, ptSelGenSummary)
if dR2best < best_dr2:
leg.genp = l1match
return
# Ok do one more Pythia 8 trick...
# This is to overcome that the GenAnalyzer doesn't like particles
# that have daughters with same pdgId and status 71
if not hasattr(event, 'pythiaQuarksGluons'):
event.pythiaQuarksGluons = []
for gen in event.genParticles:
pdg = abs(gen.pdgId())
status = gen.status()
if pdg in [1, 2, 3, 4, 5, 21] and status > 3:
if gen.isMostlyLikePythia6Status3():
event.pythiaQuarksGluons.append(gen)
l1match, dR2best = bestMatch(leg, event.pythiaQuarksGluons)
if dR2best < best_dr2:
leg.genp = l1match
return
# Now this may be a pileup lepton, or one whose ancestor doesn't
# appear in the gen summary because it's an unclear case in Pythia 8
# To check the latter, match against jets as well...
l1match, dR2best = bestMatch(leg, event.genJets)
# Check if there's a gen jet with pT > 10 GeV (otherwise it's PU)
if dR2best < dR2 and l1match.pt() > 10.:
leg.genp = PhysicsObject(l1match)
jet, dR2best = bestMatch(l1match, event.jets)
if dR2best < dR2:
leg.genp.detFlavour = jet.partonFlavour()
else:
print 'no match found', leg.pt(), leg.eta()
def genMatch(event, leg, ptSelGentauleps, ptSelGenleps, ptSelGenSummary,
dR=0.2, matchAll=True):
dR2 = dR * dR
leg.isTauHad = False
leg.isTauLep = False
leg.isPromptLep = False
leg.genp = None
best_dr2 = dR2
# The following would work for pat::Taus, but we also want to flag a
# muon/electron as coming from a hadronic tau with the usual definition
# if this happens
# if hasattr(leg, 'genJet') and leg.genJet():
# if leg.genJet().pt() > 15.:
# dr2 = deltaR2(leg.eta(), leg.phi(), leg.genJet().eta(), leg.genJet().phi())
# if dr2 < best_dr2:
# best_dr2 = dr2
# leg.genp = leg.genJet()
# leg.genp.setPdgId(-15 * leg.genp.charge())
# leg.isTauHad = True
# RM: needed to append genTauJets to the events,
# when genMatch is used as a static method
if not hasattr(event, 'genTauJets'):
HTTGenAnalyzer.getGenTauJets(event)
l1match, dR2best = bestMatch(leg, event.genTauJets)
if dR2best < best_dr2:
best_dr2 = dR2best
# leg.genp = GenParticle(l1match)
leg.genp = l1match
leg.genp.setPdgId(-15 * leg.genp.charge())
leg.isTauHad = True
# if not leg.genJet():
# print 'Warning, tau does not have matched gen tau'
# elif leg.genJet().pt() < 15.:
# print 'Warning, tau has matched gen jet but with pt =', leg.genJet().pt()
# to generated leptons from taus
l1match, dR2best = bestMatch(leg, ptSelGentauleps)
if dR2best < best_dr2:
best_dr2 = dR2best
leg.genp = l1match
leg.isTauLep = True
leg.isTauHad = False
# to generated prompt leptons
l1match, dR2best = bestMatch(leg, ptSelGenleps)
if dR2best < best_dr2:
best_dr2 = dR2best
leg.genp = l1match
leg.isPromptLep = True
leg.isTauLep = False
leg.isTauHad = False
if best_dr2 < dR2:
return
# match with any other relevant gen particle
if matchAll:
l1match, dR2best = bestMatch(leg, ptSelGenSummary)
if dR2best < best_dr2:
leg.genp = l1match
return
# Ok do one more Pythia 8 trick...
# This is to overcome that the GenAnalyzer doesn't like particles
# that have daughters with same pdgId and status 71
if not hasattr(event, 'pythiaQuarksGluons'):
event.pythiaQuarksGluons = []
for gen in event.genParticles:
pdg = abs(gen.pdgId())
status = gen.status()
if pdg in [1, 2, 3, 4, 5, 21] and status > 3:
if gen.isMostlyLikePythia6Status3():
event.pythiaQuarksGluons.append(gen)
l1match, dR2best = bestMatch(leg, event.pythiaQuarksGluons)
if dR2best < best_dr2:
leg.genp = l1match
return
# Now this may be a pileup lepton, or one whose ancestor doesn't
# appear in the gen summary because it's an unclear case in Pythia 8
# To check the latter, match against jets as well...
l1match, dR2best = bestMatch(leg, getattr(event, 'genJets', []))
# Check if there's a gen jet with pT > 10 GeV (otherwise it's PU)
if dR2best < dR2 and l1match.pt() > 10.:
leg.genp = PhysicsObject(l1match)
jet, dR2best = bestMatch(l1match, event.jets)
if dR2best < dR2:
leg.genp.detFlavour = jet.partonFlavour()
else:
print 'no match found', leg.pt(), leg.eta()
def evaluateRegressionAK08(self, event):
#self.readCollections( event.input )
reg_fj = []
for ung_fj in getattr(event, "ak08"):
# We need the closest ungroomed fatjet to get the JEC: # - Make a list of pairs: deltaR(ungroomed fj, groomed fj) for all ungroomed fatjets # - Sort by deltaR # - And take the minimum
if len(getattr(event, "ak08pruned")):
closest_pr_fj_and_dr = sorted(
[(pr_fj, deltaR2(ung_fj, pr_fj))
for pr_fj in getattr(event, "ak08pruned")],
key=lambda x: x[1])[0]
else:
print "WARNING: No pruned fatjets found in event with ungroomed fatjet. Skipping"
continue
# Use the jet cone size for matching
minimal_dr_groomed_ungroomed = 0.8
if closest_pr_fj_and_dr[1] > minimal_dr_groomed_ungroomed:
print "WARNING: No pruned fatjet found close to ungroomed fatjet. Skipping"
continue
pr_jet = Jet(closest_pr_fj_and_dr[0])
if len(getattr(event, "ak08prunedcal")):
closest_cal_fj_and_dr = sorted(
[(cal_fj, deltaR2(ung_fj, cal_fj))
for cal_fj in getattr(event, "ak08prunedcal")],
key=lambda x: x[1])[0]
else:
print "WARNING: No calib groomed fatjets found in event with ungroomed fatjet. Skipping"
continue
if closest_cal_fj_and_dr[1] > minimal_dr_groomed_ungroomed:
print "WARNING: No calib fatjet found close to ungroomed fatjet. Skipping"
continue
cal_jet = Jet(closest_cal_fj_and_dr[0])
# now check the AK08 jet is not a lepton
# if len(getattr(event, "selectedLeptons")):
# closest_pr_lep_and_dr = sorted( [(lep, deltaR2(cal_jet,lep)) for lep in getattr(event, "vLeptons")], key=lambda x:x[1])[0]
# if closest_pr_lep_and_dr[1] < (0.4 * 0.4):
# print "WARNING: No groomed fatjet is overlapping with a vLepton. Skipping"
# continue
# Need to do a deep-copy. Otherwise the original jet will be modified
reg_groomed_fj = PhysicsObject(closest_cal_fj_and_dr[0]).__copy__()
# for j in event.FatjetAK08pruned:
self.FatjetAK08ungroomed_pt[0] = ung_fj.pt()
# print 'ung_fj.pt() ', ung_fj.pt()
self.FatjetAK08pruned_pt[0] = pr_jet.pt()
# print 'pr_jet.pt() ', pr_jet.pt()
self.FatjetAK08prunedCal_pt[0] = cal_jet.pt()
# print 'cal_jet.pt() ', cal_jet.pt()
self.FatjetAK08prunedCal_eta[0] = cal_jet.eta()
self.FatjetAK08ungroomed_vertexNTracks[0] = ung_fj.vertexNTracks
self.FatjetAK08ungroomed_SV_mass_0[0] = ung_fj.SV_mass_0
self.FatjetAK08ungroomed_SV_EnergyRatio_0[
0] = ung_fj.SV_EnergyRatio_0
self.FatjetAK08ungroomed_SV_EnergyRatio_1[
0] = ung_fj.SV_EnergyRatio_1
self.FatjetAK08ungroomed_PFLepton_ptrel[0] = ung_fj.PFLepton_ptrel
self.FatjetAK08ungroomed_nSL[0] = ung_fj.nSL
# print 'ung_fj.nSL ', ung_fj.nSL
reg_groomed_fj.scaleEnergy(
self.reader.EvaluateRegression(self.name)[0])
reg_fj.append(reg_groomed_fj)
# print 'reg_groomed_fj.pt() ', reg_groomed_fj.pt()
setattr(event, 'ak08prunedreg', reg_fj)
