def fillHistos(chain, histosThisGroupPerSource, isData, lepton, group, selection, verbose=False):
"expect histos[sel][source][var][loose,tight]"
normFactor = 3.2 if group=='heavyflavor' else 1.0 # bb/cc hand-waving normalization factor, see notes 2014-04-17
nLepFilled = 0
num_processed_entries = 0
for event in chain :
num_processed_entries += 1
pars = event.pars
weight, evtN, runN = pars.weight, pars.eventNumber, pars.runNumber
weight = weight*normFactor
l0, l1 = kin.addTlv(event.l0), kin.addTlv(event.l1)
met = kin.addTlv(event.met)
jets = [kin.addTlv(j) for j in event.jets]
dphi_ll_vBetaT, mdeltar = kin.computeRazor(l0.p4, l1.p4, met.p4)
sourceReal = 3 # see FakeLeptonSources.h
l0IsFake = l0.source!=sourceReal and not isData
l1IsFake = l0.source!=sourceReal and not isData
atLeastOneIsFake = l0IsFake or l1IsFake
if not atLeastOneIsFake : continue
is_same_sign = int((l0.charge * l1.charge)>0)
is_opp_sign = not is_same_sign
l0_pt, l1_pt = l0.p4.Pt(), l1.p4.Pt()
dphi_l0_met = abs(l0.p4.DeltaPhi(met.p4))
dphi_l1_met = abs(l1.p4.DeltaPhi(met.p4))
dphi_l0_l1 = abs(l0.p4.DeltaPhi(l1.p4))
pass_selection = (is_opp_sign and l0_pt>45.0 and l1_pt>12.0)
# and
# dphi_l0_l1>2.3 and dphi_l1_met<0.7 and dphi_l0_met>2.5
# and (l0_pt-l1_pt)>7.0)
if not pass_selection : continue
def fillHistosBySource(lep):
isTight = lep.isTight
source = lep.source
leptonSource = enum2source(lep)
isReal = source==sourceReal and not isData
isFake = not isReal and not isData
sourceIsKnown = not isData
isRightLep = lep.isMu and lepton=='mu' or lep.isEl and lepton=='el'
def fill():
pt, eta, mt = lep.p4.Pt(), abs(lep.p4.Eta()), computeMt(lep.p4, met.p4)
pt, eta, mt = shiftWithinRange(pt, eta, mt) # avoid loosing entries due to over/underflow
histosThisGroupPerSource[selection][leptonSource]['mt' ]['loose'].Fill(mt, weight)
histosThisGroupPerSource[selection][leptonSource]['pt' ]['loose'].Fill(pt, weight)
histosThisGroupPerSource[selection][leptonSource]['eta' ]['loose'].Fill(eta, weight)
histosThisGroupPerSource[selection][leptonSource]['mdeltar']['loose'].Fill(mdeltar, weight)
histosThisGroupPerSource[selection][leptonSource]['pt_eta' ]['loose'].Fill(pt, eta, weight)
filled = False
# if isRightLep and sourceIsKnown and isFake: # DG-2014-08-08: pt cut still needed? and lep.p4.Pt()>20.0:
if isRightLep and sourceIsKnown and isFake: # and mdeltar>20.0: # DG-2014-10-17: try razor mod sel
fill()
filled = True
return filled
if fillHistosBySource(l0) : nLepFilled += 1
if fillHistosBySource(l1) : nLepFilled += 1
if verbose : print "filled histos for %d leptons"%nLepFilled
return num_processed_entries
def count_and_fill(chain, sample='', syst='', verbose=False, debug=False, blinded=True,
onthefly_tight_def=None, tightight=False, quicktest=False,
cached_cut=None, noncached_cuts=[]):
"""
count and fill for one sample (or group), one syst.
"""
sysGroup = systUtils.Group(sample).setSyst(syst)
is_mc = systUtils.Group(sample).isMc
is_data = systUtils.Group(sample).isData
is_qflip_sample = dataset.DatasetGroup(sample).is_qflip
assert bool(cached_cut) != bool(noncached_cuts),"must choose either cached selection or non-cached selections: {}, {}".format(cached_cut, noncached_cuts)
cuts = [cached_cut] if cached_cut else noncached_cuts
if noncached_cuts:
chain.preselect(None)
selections = [c.GetName() for c in cuts]
counters = book_counters(selections)
histos = book_histograms(sample_name=sample, variables=variables_to_fill(),
systematics=[syst], selections=selections
)[syst]
if is_qflip_sample : # for qflip, only fill ss histos
selections = [s for s in selections if s.endswith('_ss')]
weight_expr = 'event.pars.weight'
weight_expr = sysGroup.weightLeafname
qflip_expr = 'event.pars.qflipWeight'
print 'weight_expr: ',weight_expr
print 'selections: ','\n'.join(["%d) %s : %s"%(i, cut.GetName(), cut.GetTitle()) for i, cut in enumerate(cuts)])
start_time = time.clock()
num_total_entries = chain.GetEntries()
num_processed_entries = 0
fields_to_print = ['l0_pt', 'l1_pt', 'l0_eta', 'l1_eta',
'met_pt',
'm_ll', 'pt_ll', 'dpt_l0_l1',
'dphi_l0_met', 'dphi_l1_met', 'dphi_l0_l1',
'mt0', 'mt1',
'n_soft_jets',
'eta_csj0', 'phi_csj0', 'eta_csj1', 'phi_csj1']
if debug : print ",".join(fields_to_print)
for iEntry, event in enumerate(chain):
if quicktest and 100*iEntry > num_total_entries: break
run_num = event.pars.runNumber
evt_num = event.pars.eventNumber
l0 = addTlv(event.l0)
l1 = addTlv(event.l1)
met = addTlv(event.met)
l0_is_el, l0_is_mu = l0.isEl, l0.isMu
l1_is_el, l1_is_mu = l1.isEl, l1.isMu
l0_is_t = onthefly_tight_def(l0) if onthefly_tight_def else l0.isTight
l1_is_t = onthefly_tight_def(l1) if onthefly_tight_def else l1.isTight
is_emu = int(l0_is_el and l1_is_mu)
is_mue = int(l0_is_mu and l1_is_el)
is_mumu = int(l0_is_mu and l1_is_mu)
is_ee = int(l0_is_el and l1_is_el)
is_same_sign = int((l0.charge * l1.charge)>0)
is_opp_sign = not is_same_sign
is_qflippable = is_opp_sign and (l0_is_el or l1_is_el) and is_mc
weight = eval(weight_expr)
qflip_prob = eval(qflip_expr)
# print "event : same sign {0}, opp_sign {1}, qflippable {2}, qflip_prob {3}".format(is_same_sign, is_opp_sign, is_qflippable, eval(qflip_expr))
l0_pt, l1_pt = l0.p4.Pt(), l1.p4.Pt()
d_pt0_pt1 = l0_pt - l1_pt
l0_eta, l1_eta = l0.p4.Eta(), l1.p4.Eta()
l0_phi, l1_phi = l0.p4.Phi(), l1.p4.Phi()
met_pt = met.p4.Pt()
m_ll = (l0.p4 + l1.p4).M()
pt_ll = (l0.p4 + l1.p4).Pt()
dphi_l0_met = abs(l0.p4.DeltaPhi(met.p4))
dphi_l1_met = abs(l1.p4.DeltaPhi(met.p4))
dphi_l0_l1 = abs(l0.p4.DeltaPhi(l1.p4))
dpt_l0_l1 = l0.p4.Pt()-l1.p4.Pt()
m_coll = computeCollinearMassLepTau(l0.p4, l1.p4, met.p4)
mt0, mt1 = computeMt(l0.p4, met.p4), computeMt(l1.p4, met.p4)
dphillbeta, mdr = computeRazor(l0.p4, l1.p4, met.p4)
def jet_pt2(j) : return j.px*j.px+j.py*j.py
cl_jets = [addTlv(j) for j in event.jets if jet_pt2(j)>30.*30.]
n_cl_jets = len(cl_jets)
n_b_jets = event.pars.numBjets
n_f_jets = event.pars.numFjets
n_bf_jets = n_b_jets + n_f_jets
n_jets = n_cl_jets + n_b_jets + n_f_jets
# n_jets = event.pars.numFjets + event.pars.numBjets
soft_jets = [addTlv(j) for j in event.jets if jet_pt2(j)<30.**2] # todo: merge with cl_jets loop
n_soft_jets = len(soft_jets)
csj0 = first(sorted(soft_jets, key=lambda j : j.p4.DeltaR(l0.p4)))
csj1 = first(sorted(soft_jets, key=lambda j : j.p4.DeltaR(l1.p4)))
eta_csj0 = csj0.p4.Eta() if csj0 else -5.0
phi_csj0 = csj0.p4.Phi() if csj0 else -5.0
eta_csj1 = csj1.p4.Eta() if csj1 else -5.0
phi_csj1 = csj1.p4.Phi() if csj1 else -5.0
drl0csj = csj0.p4.DeltaR(l0.p4) if csj0 else None
drl1csj = csj1.p4.DeltaR(l1.p4) if csj1 else None
m_jj = (cl_jets[0].p4 + cl_jets[1].p4).M() if n_cl_jets>1 else None
deta_jj = abs(cl_jets[0].p4.Eta() - cl_jets[1].p4.Eta()) if n_cl_jets>1 else None
pass_sels = {}
if tightight and not (l0_is_t and l1_is_t) : continue
for cut in cuts:
sel = cut.GetName()
sel_expr = cut.GetTitle()
pass_sel = eval(sel_expr)# and (l0_pt>60.0 and dphi_l1_met<0.7)
pass_sels[sel] = pass_sel
is_ss_sel = sel.endswith('_ss')
as_qflip = is_qflippable and (is_opp_sign and is_ss_sel)
if is_qflip_sample and not as_qflip : pass_sel = False
if not is_qflip_sample and as_qflip : pass_sel = False
if not pass_sel : continue
if pass_sel and not cached_cut : chain.add_entry_to_list(cut, iEntry)
# <isElectron 1> <isElectron 2> <isTight 1> <isTight 2> <pt 1> <pt 2> <eta 1> <eta 2>
lltype = "{0}{1}".format('e' if l0_is_el else 'mu', 'e' if l1_is_el else 'mu')
qqtype = "{0}{1}".format('T' if l0_is_t else 'L', 'T' if l1_is_t else 'L')
if debug : print ','.join([str(eval(_)) for _ in fields_to_print])
def fmt(b) : return '1' if b else '0'
# --- begin dbg
# print "event: {0:12s} {1} {2} {3} {4} {5} {6} {7} {8}".format(lltype+' '+qqtype, #+' '+sel,
# fmt(l0_is_el), fmt(l1_is_el),
# fmt(l0_is_t), fmt(l1_is_t),
# l0_pt, l1_pt,
# l0.p4.Eta(), l1.p4.Eta())
# print "event: {0:12s} {1} {2} {3:.2f} {4:.2f}".format(lltype+' '+qqtype+' '+sel,
# run_num, evt_num,
# l0_pt, l1_pt)
# --- end dbg
fill_weight = (weight * qflip_prob) if as_qflip else weight
h = histos[sel]
h['onebin' ].Fill(1.0, fill_weight)
h['njets' ].Fill(n_jets, fill_weight)
h['pt0' ].Fill(l0_pt, fill_weight)
h['pt1' ].Fill(l1_pt, fill_weight)
h['d_pt0_pt1'].Fill(d_pt0_pt1, fill_weight)
h['eta0' ].Fill(l0_eta, fill_weight)
h['eta1' ].Fill(l1_eta, fill_weight)
h['phi0' ].Fill(l0_phi, fill_weight)
h['phi1' ].Fill(l1_phi, fill_weight)
h['mll' ].Fill(m_ll, fill_weight)
h['ptll' ].Fill(pt_ll, fill_weight)
h['met' ].Fill(met_pt, fill_weight)
h['dphil0met'].Fill(dphi_l0_met, fill_weight)
h['dphil1met'].Fill(dphi_l1_met, fill_weight)
h['nsj' ].Fill(n_soft_jets, fill_weight)
h['pt0_vs_pt1' ].Fill(l1_pt, l0_pt, fill_weight)
h['met_vs_pt1' ].Fill(l1_pt, met.p4.Pt(), fill_weight)
h['dphil0l1_vs_pt1' ].Fill(l1_pt, dphi_l0_l1, fill_weight)
h['dphil0met_vs_pt1'].Fill(l1_pt, dphi_l0_met, fill_weight)
h['dphil1met_vs_pt1'].Fill(l1_pt, dphi_l1_met, fill_weight)
if n_soft_jets:
h['drl0csj'].Fill(drl0csj, fill_weight)
h['drl1csj'].Fill(drl1csj, fill_weight)
if n_jets==2 and n_cl_jets==2: # fixme: f jets are not saved, but we need them for vbf
h['m_jj' ].Fill(m_jj, fill_weight)
h['deta_jj'].Fill(deta_jj, fill_weight)
if is_data and (blinded and 100.0<m_coll and m_coll<150.0) : pass
else :
h['mcoll'].Fill(m_coll, fill_weight)
h['mcollcoarse'].Fill(m_coll, fill_weight)
h['mcoll_vs_pt1'].Fill(l1_pt, m_coll, fill_weight)
counters[sel] += (fill_weight)
# print ('e' if l0_is_el else 'm'),('e' if l1_is_el else 'm'),' : ',
# print ' is_opp_sign: ',is_opp_sign,
# print ' is_qflippable: ',is_qflippable,
# print pass_sels
num_processed_entries += 1
end_time = time.clock()
delta_time = end_time - start_time
if verbose:
print ("processed {0:d} entries ".format(num_processed_entries)
+"in "+("{0:d} min ".format(int(delta_time/60)) if delta_time>60 else
"{0:.1f} s ".format(delta_time))
+"({0:.1f} kHz)".format((num_processed_entries/delta_time) if delta_time else 1.0e9)
)
if verbose:
for v in ['onebin']: #, 'pt0', 'pt1']:
for sel in selections:
h = histos[sel][v]
print "{0}: integral {1}, entries {2}".format(h.GetName(), h.Integral(), h.GetEntries())
return counters, histos
def count_and_fill(chain,
sample='',
syst='',
verbose=False,
debug=False,
blinded=True,
onthefly_tight_def=None,
tightight=False,
quicktest=False,
cached_cut=None,
noncached_cuts=[]):
"""
count and fill for one sample (or group), one syst.
"""
sysGroup = systUtils.Group(sample).setSyst(syst)
is_mc = systUtils.Group(sample).isMc
is_data = systUtils.Group(sample).isData
is_qflip_sample = dataset.DatasetGroup(sample).is_qflip
assert bool(cached_cut) != bool(
noncached_cuts
), "must choose either cached selection or non-cached selections: {}, {}".format(
cached_cut, noncached_cuts)
cuts = [cached_cut] if cached_cut else noncached_cuts
if noncached_cuts:
chain.preselect(None)
selections = [c.GetName() for c in cuts]
counters = book_counters(selections)
histos = book_histograms(sample_name=sample,
variables=variables_to_fill(),
systematics=[syst],
selections=selections)[syst]
if is_qflip_sample: # for qflip, only fill ss histos
selections = [s for s in selections if s.endswith('_ss')]
weight_expr = 'event.pars.weight'
weight_expr = sysGroup.weightLeafname
qflip_expr = 'event.pars.qflipWeight'
print 'weight_expr: ', weight_expr
print 'selections: ', '\n'.join([
"%d) %s : %s" % (i, cut.GetName(), cut.GetTitle())
for i, cut in enumerate(cuts)
])
start_time = time.clock()
num_total_entries = chain.GetEntries()
num_processed_entries = 0
fields_to_print = [
'l0_pt', 'l1_pt', 'l0_eta', 'l1_eta', 'met_pt', 'm_ll', 'pt_ll',
'dpt_l0_l1', 'dphi_l0_met', 'dphi_l1_met', 'dphi_l0_l1', 'mt0', 'mt1',
'n_soft_jets', 'eta_csj0', 'phi_csj0', 'eta_csj1', 'phi_csj1'
]
if debug: print ",".join(fields_to_print)
for iEntry, event in enumerate(chain):
if quicktest and 100 * iEntry > num_total_entries: break
run_num = event.pars.runNumber
evt_num = event.pars.eventNumber
l0 = addTlv(event.l0)
l1 = addTlv(event.l1)
met = addTlv(event.met)
l0_is_el, l0_is_mu = l0.isEl, l0.isMu
l1_is_el, l1_is_mu = l1.isEl, l1.isMu
l0_is_t = onthefly_tight_def(l0) if onthefly_tight_def else l0.isTight
l1_is_t = onthefly_tight_def(l1) if onthefly_tight_def else l1.isTight
is_emu = int(l0_is_el and l1_is_mu)
is_mue = int(l0_is_mu and l1_is_el)
is_mumu = int(l0_is_mu and l1_is_mu)
is_ee = int(l0_is_el and l1_is_el)
is_same_sign = int((l0.charge * l1.charge) > 0)
is_opp_sign = not is_same_sign
is_qflippable = is_opp_sign and (l0_is_el or l1_is_el) and is_mc
weight = eval(weight_expr)
qflip_prob = eval(qflip_expr)
# print "event : same sign {0}, opp_sign {1}, qflippable {2}, qflip_prob {3}".format(is_same_sign, is_opp_sign, is_qflippable, eval(qflip_expr))
l0_pt, l1_pt = l0.p4.Pt(), l1.p4.Pt()
d_pt0_pt1 = l0_pt - l1_pt
l0_eta, l1_eta = l0.p4.Eta(), l1.p4.Eta()
l0_phi, l1_phi = l0.p4.Phi(), l1.p4.Phi()
met_pt = met.p4.Pt()
m_ll = (l0.p4 + l1.p4).M()
pt_ll = (l0.p4 + l1.p4).Pt()
dphi_l0_met = abs(l0.p4.DeltaPhi(met.p4))
dphi_l1_met = abs(l1.p4.DeltaPhi(met.p4))
dphi_l0_l1 = abs(l0.p4.DeltaPhi(l1.p4))
dpt_l0_l1 = l0.p4.Pt() - l1.p4.Pt()
m_coll = computeCollinearMassLepTau(l0.p4, l1.p4, met.p4)
mt0, mt1 = computeMt(l0.p4, met.p4), computeMt(l1.p4, met.p4)
dphillbeta, mdr = computeRazor(l0.p4, l1.p4, met.p4)
def jet_pt2(j):
return j.px * j.px + j.py * j.py
cl_jets = [addTlv(j) for j in event.jets if jet_pt2(j) > 30. * 30.]
n_cl_jets = len(cl_jets)
n_b_jets = event.pars.numBjets
n_f_jets = event.pars.numFjets
n_bf_jets = n_b_jets + n_f_jets
n_jets = n_cl_jets + n_b_jets + n_f_jets
# n_jets = event.pars.numFjets + event.pars.numBjets
soft_jets = [addTlv(j) for j in event.jets
if jet_pt2(j) < 30.**2] # todo: merge with cl_jets loop
n_soft_jets = len(soft_jets)
csj0 = first(sorted(soft_jets, key=lambda j: j.p4.DeltaR(l0.p4)))
csj1 = first(sorted(soft_jets, key=lambda j: j.p4.DeltaR(l1.p4)))
eta_csj0 = csj0.p4.Eta() if csj0 else -5.0
phi_csj0 = csj0.p4.Phi() if csj0 else -5.0
eta_csj1 = csj1.p4.Eta() if csj1 else -5.0
phi_csj1 = csj1.p4.Phi() if csj1 else -5.0
drl0csj = csj0.p4.DeltaR(l0.p4) if csj0 else None
drl1csj = csj1.p4.DeltaR(l1.p4) if csj1 else None
m_jj = (cl_jets[0].p4 + cl_jets[1].p4).M() if n_cl_jets > 1 else None
deta_jj = abs(cl_jets[0].p4.Eta() -
cl_jets[1].p4.Eta()) if n_cl_jets > 1 else None
pass_sels = {}
if tightight and not (l0_is_t and l1_is_t): continue
for cut in cuts:
sel = cut.GetName()
sel_expr = cut.GetTitle()
pass_sel = eval(sel_expr) # and (l0_pt>60.0 and dphi_l1_met<0.7)
pass_sels[sel] = pass_sel
is_ss_sel = sel.endswith('_ss')
as_qflip = is_qflippable and (is_opp_sign and is_ss_sel)
if is_qflip_sample and not as_qflip: pass_sel = False
if not is_qflip_sample and as_qflip: pass_sel = False
if not pass_sel: continue
if pass_sel and not cached_cut:
chain.add_entry_to_list(cut, iEntry)
# <isElectron 1> <isElectron 2> <isTight 1> <isTight 2> <pt 1> <pt 2> <eta 1> <eta 2>
lltype = "{0}{1}".format('e' if l0_is_el else 'mu',
'e' if l1_is_el else 'mu')
qqtype = "{0}{1}".format('T' if l0_is_t else 'L',
'T' if l1_is_t else 'L')
if debug: print ','.join([str(eval(_)) for _ in fields_to_print])
def fmt(b):
return '1' if b else '0'
# --- begin dbg
# print "event: {0:12s} {1} {2} {3} {4} {5} {6} {7} {8}".format(lltype+' '+qqtype, #+' '+sel,
# fmt(l0_is_el), fmt(l1_is_el),
# fmt(l0_is_t), fmt(l1_is_t),
# l0_pt, l1_pt,
# l0.p4.Eta(), l1.p4.Eta())
# print "event: {0:12s} {1} {2} {3:.2f} {4:.2f}".format(lltype+' '+qqtype+' '+sel,
# run_num, evt_num,
# l0_pt, l1_pt)
# --- end dbg
fill_weight = (weight * qflip_prob) if as_qflip else weight
h = histos[sel]
h['onebin'].Fill(1.0, fill_weight)
h['njets'].Fill(n_jets, fill_weight)
h['pt0'].Fill(l0_pt, fill_weight)
h['pt1'].Fill(l1_pt, fill_weight)
h['d_pt0_pt1'].Fill(d_pt0_pt1, fill_weight)
h['eta0'].Fill(l0_eta, fill_weight)
h['eta1'].Fill(l1_eta, fill_weight)
h['phi0'].Fill(l0_phi, fill_weight)
h['phi1'].Fill(l1_phi, fill_weight)
h['mll'].Fill(m_ll, fill_weight)
h['ptll'].Fill(pt_ll, fill_weight)
h['met'].Fill(met_pt, fill_weight)
h['dphil0met'].Fill(dphi_l0_met, fill_weight)
h['dphil1met'].Fill(dphi_l1_met, fill_weight)
h['nsj'].Fill(n_soft_jets, fill_weight)
h['pt0_vs_pt1'].Fill(l1_pt, l0_pt, fill_weight)
h['met_vs_pt1'].Fill(l1_pt, met.p4.Pt(), fill_weight)
h['dphil0l1_vs_pt1'].Fill(l1_pt, dphi_l0_l1, fill_weight)
h['dphil0met_vs_pt1'].Fill(l1_pt, dphi_l0_met, fill_weight)
h['dphil1met_vs_pt1'].Fill(l1_pt, dphi_l1_met, fill_weight)
if n_soft_jets:
h['drl0csj'].Fill(drl0csj, fill_weight)
h['drl1csj'].Fill(drl1csj, fill_weight)
if n_jets == 2 and n_cl_jets == 2: # fixme: f jets are not saved, but we need them for vbf
h['m_jj'].Fill(m_jj, fill_weight)
h['deta_jj'].Fill(deta_jj, fill_weight)
if is_data and (blinded and 100.0 < m_coll and m_coll < 150.0):
pass
else:
h['mcoll'].Fill(m_coll, fill_weight)
h['mcollcoarse'].Fill(m_coll, fill_weight)
h['mcoll_vs_pt1'].Fill(l1_pt, m_coll, fill_weight)
counters[sel] += (fill_weight)
# print ('e' if l0_is_el else 'm'),('e' if l1_is_el else 'm'),' : ',
# print ' is_opp_sign: ',is_opp_sign,
# print ' is_qflippable: ',is_qflippable,
# print pass_sels
num_processed_entries += 1
end_time = time.clock()
delta_time = end_time - start_time
if verbose:
print(
"processed {0:d} entries ".format(num_processed_entries) + "in " +
("{0:d} min ".format(int(delta_time / 60))
if delta_time > 60 else "{0:.1f} s ".format(delta_time)) +
"({0:.1f} kHz)".format((num_processed_entries /
delta_time) if delta_time else 1.0e9))
if verbose:
for v in ['onebin']: #, 'pt0', 'pt1']:
for sel in selections:
h = histos[sel][v]
print "{0}: integral {1}, entries {2}".format(
h.GetName(), h.Integral(), h.GetEntries())
return counters, histos