def fillHistosAndCount(histos, files, lls, njs, testRun=False) :
"Fill the histograms, and provide a dict of event counters[sample][sel] for the summary"
treename = 'SusySel'
counts = dict()
for sample, filename in files.iteritems() :
countsSample = collections.defaultdict(float)
histosSample = histos[sample]
file = r.TFile.Open(filename)
tree = file.Get(treename)
nEvents = tree.GetEntries()
nEventsToProcess = nEvents if not testRun else nEvents/10
print "processing %s (%d entries %s) %s"%(sample, nEventsToProcess, ", 10% test" if testRun else "", datetime.datetime.now())
for iEvent, event in enumerate(tree) :
if iEvent > nEventsToProcess : break
l0, l1, met, pars = addTlv(event.l0), addTlv(event.l1), addTlv(event.met), event.pars
jets, lepts = [addTlv(j) for j in event.jets], [addTlv(l) for l in event.lepts]
ll = getDilepType(l0, l1)
nJets = len(jets)
nj = 'eq1j' if nJets==1 else 'ge2j'
assert nJets>0,"messed something up in the selection upstream"
if ll not in lls or nj not in njs : continue
pt0 = l0.p4.Pt()
pt1 = l1.p4.Pt()
j0 = jets[0]
mll = (l0.p4 + l1.p4).M()
mtllmet = computeMt(l0.p4 + l1.p4, met.p4)
ht = computeHt(met.p4, [l0.p4, l1.p4]+[j.p4 for j in jets])
metrel = computeMetRel(met.p4, [l0.p4, l1.p4]+[j.p4 for j in jets])
mtl0 = computeMt(l0.p4, met.p4)
mtl1 = computeMt(l1.p4, met.p4)
mtmin = min([mtl0, mtl1])
mtmax = max([mtl0, mtl1])
mlj = computeMlj(l0.p4, l1.p4, j0.p4)
dphill = abs(phi_mpi_pi(l0.p4.DeltaPhi(l1.p4)))
detall = fabs(l0.p4.Eta() - l1.p4.Eta())
l3Veto = not thirdLepZcandidateIsInWindow(l0, l1, lepts)
mljj = None
if nJets >1 :
j0, j1 = jets[0], jets[1]
mt2j = computeMt2j(l0.p4, l1.p4, j0.p4, j1.p4, met.p4)
mljj = computeMljj(l0.p4, l1.p4, j0.p4, j1.p4)
dphijj = fabs(phi_mpi_pi(j0.p4.DeltaPhi(j1.p4)))
detajj = fabs(j0.p4.Eta() - j1.p4.Eta())
if passSelection(pt0, pt1, mll, mtllmet, ht, metrel, l3Veto,
detall, mtmax, mlj, mljj,
ll, nj) :
llnj = llnjKey(ll, nj)
weight = pars.weight
varHistos = histosSample[llnj]
varValues = dict([(v, eval(v)) for v in variablesToPlot()])
fillVarHistos(varHistos, varValues, weight, nj)
countsSample[llnj] += weight
file.Close()
file.Delete()
counts[sample] = countsSample
return counts
def createOutTree(filenames, dilepChan, nJetChan, tag='', overwrite=False) :
assert dilepChan in ['ee','mm','em']
assert nJetChan in ['eq1j', 'ge2j']
outFilenames = dict()
for sample, filename in filenames.iteritems() :
outFilename = '/tmp/'+sample+'_'+dilepChan+'_'+nJetChan+'.root'
if os.path.exists(outFilename) and not overwrite :
outFilenames[sample] = outFilename
continue
outFile = r.TFile.Open(outFilename, 'recreate')
outTree = r.TTree("training","Training tree")
outTree.Branch('vars', vars, '/F:'.join(leafNames))
outTree.SetDirectory(outFile)
file = r.TFile.Open(filename)
tree = file.Get(treename)
print "processing %s %s %s (%d entries)"%(sample, dilepChan, nJetChan, tree.GetEntries())
for iEvent, event in enumerate(tree) :
resetVars(vars)
l0 = addTlv(event.l0)
l1 = addTlv(event.l1)
met = addTlv(event.met)
jets = [addTlv(j) for j in event.jets]
lepts = [addTlv(l) for l in event.lepts]
pars = event.pars
dilepType = getDilepType(event.l0, event.l1)
nJets = len(jets)
if dilepType != dilepChan : continue
if nJets<1 or (nJets>1 and nJetChan=='eq1j') : continue
if thirdLepZcandidateIsInWindow(l0, l1, lepts, 20.0) : continue
mt0, mt1 = computeMt(l0.p4, met.p4), computeMt(l1.p4, met.p4)
vars.pt0 = l0.p4.Pt()
vars.pt1 = l1.p4.Pt()
vars.mll = (l0.p4+l1.p4).M()
vars.mtmin = min([mt0, mt1])
vars.mtmax = max([mt0, mt1])
vars.mtllmet = computeMt(l0.p4 + l1.p4, met.p4)
vars.ht = computeHt(met.p4, [l0.p4, l1.p4]+[j.p4 for j in jets])
vars.metrel = computeMetRel(met.p4, [l0.p4, l1.p4]+[j.p4 for j in jets])
vars.dphill = fabs(phi_mpi_pi(l0.p4.DeltaPhi(l1.p4)))
vars.detall = fabs(l0.p4.Eta() - l1.p4.Eta())
if nJets >1 :
j0, j1 = jets[0], jets[1]
vars.mt2j = computeMt2j(l0.p4, l1.p4, j0.p4, j1.p4, met.p4)
vars.mljj = computeMljj(l0.p4, l1.p4, j0.p4, j1.p4)
vars.dphijj = fabs(phi_mpi_pi(j0.p4.DeltaPhi(j1.p4)))
vars.detajj = fabs(j0.p4.Eta() - j1.p4.Eta())
outTree.Fill()
print "filled ",outTree.GetEntries()," entries"
outFile.Write()
outFile.Close()
outFilenames[sample] = outFile.GetName()
return outFilenames
def fillHistos() :
filenames = getInputFiles()
hs_nlep = dict()
hs_nillep = dict()
hs_npairs = dict()
hs_masses = dict()
hs_bestm = dict()
hs_bestlpt = dict()
hs_bestleta= dict()
hs_bestdphi= dict()
hs_bestdrj = dict()
for sample, filename in filenames.iteritems() :
file = r.TFile.Open(filename)
tree = file.Get(treename)
print "processing %s (%d entries)"%(sample, tree.GetEntries())
iEvent = 0
h_nlep = r.TH1F('h_nlep_'+sample, '; N_{lep, low-p_{T}};entries/bin', 5, -0.5, 4.5)
h_nillep = r.TH1F('h_nillep_'+sample,
'separated from signal leptons; N_{lep, iso-lep, low-p_{T}};entries/bin',
5, -0.5, 4.5)
h_npairs = r.TH1F('h_npairs_'+sample, '; N_{valid pairs}; entries/bin', 5, -0.5, 4.5)
h_masses = r.TH1F('h_masses_'+sample, '; m_{ll}, any valid pair [GeV]; entries/bin', 25, 0.0, 200.0)
h_bestm = r.TH1F('h_bestm_'+sample, '; m_{ll}, best pair [GeV]; entries/bin', 25, 0.0, 200.0)
h_bestlpt = r.TH1F('h_bestlpt_'+sample, '; p_{T}, best soft lepton [GeV]; entries/bin', 25, 0.0, 200.0)
h_bestleta= r.TH1F('h_bestleta_'+sample, '; #eta, best soft lepton; entries/bin', 25, -3.0, +3.0)
h_bestdphi= r.TH1F('h_bestdphi'+sample, '; |#Delta#phi(l_{hard},l_{soft})|, best pair [rad]; entries/bin',
25, 0.0, +2.*math.pi)
h_bestdrj = r.TH1F('h_bestdrj'+sample, '; min#DeltaR(l_{soft}), best soft lep; entries/bin',
20, 0.0, 2.0)
for event in tree :
l0 = addTlv(event.l0)
l1 = addTlv(event.l1)
met = addTlv(event.met)
jets = [addTlv(j) for j in event.jets]
lepts = [addTlv(l) for l in event.lepts]
pars = event.pars
weight, evtN, runN = pars.weight, pars.eventNumber, pars.runNumber
h_nlep.Fill(float(len(lepts)), weight)
if len(lepts) < 1 : continue
lepts = filter(lambda l : lepIsSeparatedFromOther(l.p4, [l0.p4, l1.p4]), lepts)
h_nillep.Fill(float(len(lepts)), weight)
validPairs = [(lh, ls) for lh in [l0, l1] for ls in lepts if lepPairIsZcand(lh, ls)]
h_npairs.Fill(float(len(validPairs)), weight)
pairsSortedByBestM = sorted(validPairs, key=deltaMZ0)
for i,(lh,ls) in enumerate(pairsSortedByBestM) :
m = (lh.p4+ls.p4).M()
h_masses.Fill(m, weight)
if i==0 :
h_bestm.Fill(m, weight)
h_bestlpt.Fill(ls.p4.Pt(), weight)
h_bestleta.Fill(ls.p4.Eta(), weight)
h_bestdphi.Fill(abs(phi_mpi_pi(ls.p4.DeltaPhi(lh.p4))), weight)
if len(jets) :
minDrJ = sorted([ls.p4.DeltaR(j.p4) for j in jets])[0]
h_bestdrj.Fill(minDrJ, weight)
if iEvent<10 : print "jets [%d], lepts[%d]"%(len(jets), len(lepts))
iEvent += 1
hs_nlep [sample] = h_nlep
hs_nillep [sample] = h_nillep
hs_npairs [sample] = h_npairs
hs_masses [sample] = h_masses
hs_bestm [sample] = h_bestm
hs_bestlpt [sample] = h_bestlpt
hs_bestleta[sample] = h_bestleta
hs_bestdphi[sample] = h_bestdphi
hs_bestdrj [sample] = h_bestdrj
return {'hs_nlep' : hs_nlep,
'hs_nillep' : hs_nillep,
'hs_npairs' : hs_npairs,
'hs_masses' : hs_masses,
'hs_bestm' : hs_bestm,
'hs_bestlpt' : hs_bestlpt,
'hs_bestleta': hs_bestleta,
'hs_bestdphi': hs_bestdphi,
'hs_bestdrj' : hs_bestdrj
}
