def mkRegression(name,weights,variabs):
pset = cms.PSet(
variables=cms.VPSet(),
classifier=cms.string(name),
weights=cms.FileInPath(weights),
regression=cms.bool(True),
)
cfgTools.addVariables(pset.variables,variabs)
return pset
def mkRegression(name, weights, variabs):
pset = cms.PSet(
variables=cms.VPSet(),
classifier=cms.string(name),
weights=cms.FileInPath(weights),
regression=cms.bool(True),
)
cfgTools.addVariables(pset.variables, variabs)
return pset
ttHScoreThreshold=cms.double(0.0), #to be updated
# For standardization
ttHKiller_mean=ttHKiller_mean,
ttHKiller_std=ttHKiller_std,
ttHKiller_listmean=ttHKiller_listmean,
ttHKiller_liststd=ttHKiller_liststd)
cfgTools.addVariables(
flashggDoubleHTag.MVAConfig.variables,
# here the syntax is VarNameInTMVA := expression
[
"Mjj := dijet().M()",
"leadingJet_DeepCSV := leadJet().bDiscriminator('pfDeepCSVJetTags:probb')+leadJet().bDiscriminator('pfDeepCSVJetTags:probbb')", #FIXME make the btag type configurable?
"subleadingJet_DeepCSV := subleadJet().bDiscriminator('pfDeepCSVJetTags:probb')+subleadJet().bDiscriminator('pfDeepCSVJetTags:probbb')",
"absCosThetaStar_CS := abs(getCosThetaStar_CS)",
"absCosTheta_bb := abs(CosThetaAngles()[1])",
"absCosTheta_gg := abs(CosThetaAngles()[0])",
"diphotonCandidatePtOverdiHiggsM := diphotonPtOverM()",
"dijetCandidatePtOverdiHiggsM := dijetPtOverM()",
"customLeadingPhotonIDMVA := diPhoton.leadingView.phoIdMvaWrtChosenVtx",
"customSubLeadingPhotonIDMVA := diPhoton.subLeadingView.phoIdMvaWrtChosenVtx",
"leadingPhotonSigOverE := diPhoton.leadingPhoton.sigEOverE",
"subleadingPhotonSigOverE := diPhoton.subLeadingPhoton.sigEOverE",
"sigmaMOverM := sqrt(0.5*(diPhoton.leadingPhoton.sigEOverE*diPhoton.leadingPhoton.sigEOverE + diPhoton.subLeadingPhoton.sigEOverE*diPhoton.subLeadingPhoton.sigEOverE))",
"PhoJetMinDr := getPhoJetMinDr()",
"rho := global.rho",
"(leadingJet_bRegNNResolution*1.4826) := leadJet().userFloat('bRegNNResolution')*1.4826",
"(subleadingJet_bRegNNResolution*1.4826) := subleadJet().userFloat('bRegNNResolution')*1.4826",
"(sigmaMJets*1.4826) := getSigmaMOverMJets()*1.4826"
])
def setup_flashggDifferentialPhoIdInputsCorrection( process, metaConditions ):
process.RandomNumberGeneratorService = cms.Service("RandomNumberGeneratorService",
flashggDifferentialPhoIdInputsCorrection=cms.PSet(
initialSeed=cms.untracked.uint32(
90)
)
)
process.load("flashgg.Taggers.flashggDifferentialPhoIdInputsCorrection_cfi")
# print(process)
setattr(flashggDifferentialPhoIdInputsCorrection, "photonIdMVAweightfile_EB", cms.FileInPath(str(metaConditions['flashggPhotons']['photonIdMVAweightfile_EB'])))
setattr(flashggDifferentialPhoIdInputsCorrection, "photonIdMVAweightfile_EE", cms.FileInPath(str(metaConditions['flashggPhotons']['photonIdMVAweightfile_EE'])))
setattr(flashggDifferentialPhoIdInputsCorrection, "effAreasConfigFile", cms.FileInPath(str(metaConditions['flashggPhotons']['effAreasConfigFile'])))
setattr(flashggDifferentialPhoIdInputsCorrection, "is2017", cms.bool(metaConditions['flashggPhotons']['is2017']))
corrections_summary = {}
with open(os.path.expandvars('$CMSSW_BASE/src/'+metaConditions['PhoIdInputCorrections']['corrections_summary'])) as json_file:
corrections_summary = json.load(json_file)
#---Shower shapes
for var in corrections_summary['shower_shapes'].keys():
for subdet in ['EB', 'EE']:
ss_summary = corrections_summary['shower_shapes'][var][subdet]
setattr(flashggDifferentialPhoIdInputsCorrection, var+'_corrector_config_'+subdet,
cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(ss_summary['weights'])),
regr_output_scaling = cms.string('x[0]*(%f)+(%f)' % (ss_summary['scale'], ss_summary['center'])),
regression = cms.bool(True),
classifier = cms.string('BDTG_{}_{}'.format(var, subdet))
)
)
xgb_config = getattr(flashggDifferentialPhoIdInputsCorrection, var+'_corrector_config_'+subdet)
# print(list(metaConditions['PhoIdInputCorrections']['SS_variables']))
cfgTools.addVariables(xgb_config.variables, [str(st) for st in metaConditions['PhoIdInputCorrections']['SS_variables']])
# ['f0 := pt',
# 'f1 := superCluster.eta',
# 'f2 := phi',
# 'f3 := global.rho',
# 'f4 := sieip',
# 'f5 := s4',
# 'f6 := full5x5_r9',
# 'f7 := superCluster.phiWidth',
# 'f8 := full5x5_sigmaIetaIeta',
# 'f9 := superCluster.etaWidth'
# ]
# )
#---Isolations
for subdet in ['EB', 'EE']:
#-photon isolation
iso_summary = corrections_summary['isolations']['phoIso'][subdet]
setattr(flashggDifferentialPhoIdInputsCorrection, 'phoIso_corrector_config_'+subdet,
cms.PSet(
clf_data = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['peak_tail_clfs']['weights_data'])),
classifier = cms.string('BDTG_{}_{}_p0t_data'.format("phoIso", subdet))
),
clf_mc = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['peak_tail_clfs']['weights_mc'])),
classifier = cms.string('BDTG_{}_{}_p0t_mc'.format("phoIso", subdet))
),
peak2tail = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['peak2tail']['weights'])),
regression = cms.bool(True),
classifier = cms.string('BDTG_{}_{}_p2t'.format("phoIso", subdet))
),
morphing = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['morphing']['weights'])),
regression = cms.bool(True),
classifier = cms.string('BDTG_{}_{}'.format("phoIso", subdet)),
regr_output_scaling = cms.string('x[0]*(%f)+(%f)' % (iso_summary['morphing']['scale'], iso_summary['morphing']['center']))
)
)
)
phoIso_corrector_config = getattr(flashggDifferentialPhoIdInputsCorrection, 'phoIso_corrector_config_'+subdet)
cfgTools.addVariables(phoIso_corrector_config.clf_data.variables, [str(x) for x in iso_summary['peak_tail_clfs']['variables']])
cfgTools.addVariables(phoIso_corrector_config.clf_mc.variables, [str(x) for x in iso_summary['peak_tail_clfs']['variables']])
cfgTools.addVariables(phoIso_corrector_config.peak2tail.variables, [str(x) for x in iso_summary['peak2tail']['variables']])
cfgTools.addVariables(phoIso_corrector_config.morphing.variables, [str(x) for x in iso_summary['morphing']['variables']])
#-charged isolations
iso_summary = corrections_summary['isolations']['chIso'][subdet]
setattr(flashggDifferentialPhoIdInputsCorrection, 'chIso_corrector_config_'+subdet,
cms.PSet(
clf_data = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['peak_tail_clfs']['weights_data'])),
multiclass = cms.bool(True),
classifier = cms.string('BDTG_{}_{}_3Cat_data'.format("chIso", subdet))
),
clf_mc = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['peak_tail_clfs']['weights_mc'])),
multiclass = cms.bool(True),
classifier = cms.string('BDTG_{}_{}_3Cat_mc'.format("chIso", subdet))
),
chIso_peak2tail = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['chIso_peak2tail']['weights'])),
regression = cms.bool(True),
classifier = cms.string('BDTG_{}_{}_p2t'.format("chIso", subdet))
),
chIso_morphing = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['chIso_morphing']['weights'])),
classifier = cms.string('BDTG_{}_{}'.format("chIso", subdet)),
regression = cms.bool(True),
regr_output_scaling = cms.string('x[0]*(%f)+(%f)' % (iso_summary['chIso_morphing']['scale'],
iso_summary['chIso_morphing']['center']))
),
chIsoWorst_peak2tail = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['chIsoWorst_peak2tail']['weights'])),
regression = cms.bool(True),
classifier = cms.string('BDTG_{}_{}_p2t'.format("chIsoWorst", subdet))
),
chIsoWorst_morphing = cms.PSet(
variables = cms.VPSet(),
weights = cms.FileInPath(str(iso_summary['chIsoWorst_morphing']['weights'])),
classifier = cms.string('BDTG_{}_{}'.format("chIsoWorst", subdet)),
regression = cms.bool(True),
regr_output_scaling = cms.string('x[0]*(%f)+(%f)' % (iso_summary['chIsoWorst_morphing']['scale'],
iso_summary['chIsoWorst_morphing']['center']))
)
)
)
chIso_corrector_config = getattr(flashggDifferentialPhoIdInputsCorrection, 'chIso_corrector_config_'+subdet)
cfgTools.addVariables(chIso_corrector_config.clf_data.variables, [str(x) for x in iso_summary['peak_tail_clfs']['variables']])
cfgTools.addVariables(chIso_corrector_config.clf_mc.variables, [str(x) for x in iso_summary['peak_tail_clfs']['variables']])
cfgTools.addVariables(chIso_corrector_config.chIso_peak2tail.variables, [str(x) for x in iso_summary['chIso_peak2tail']['variables']])
cfgTools.addVariables(chIso_corrector_config.chIso_morphing.variables, [str(x) for x in iso_summary['chIso_morphing']['variables']])
cfgTools.addVariables(chIso_corrector_config.chIsoWorst_peak2tail.variables, [str(x) for x in iso_summary['chIsoWorst_peak2tail']['variables']])
cfgTools.addVariables(chIso_corrector_config.chIsoWorst_morphing.variables, [str(x) for x in iso_summary['chIsoWorst_morphing']['variables']])
import flashgg.Taggers.dumperConfigTools as cfgTools
cShapeMVA_EB = cms.PSet(
variables=cms.VPSet(),
classifier=cms.string("BDTCShape_EB"),
weights=cms.FileInPath("diphotons/Analysis/data/TMVAClassification_BDTCShape_EB.weights.xml")
## classifier=cms.string("BDTCShape_test_EB"),
## weights=cms.FileInPath("diphotons/Analysis/data/TMVAClassification_BDTCShape_test_EB.weights.xml")
)
cShapeMVA_EE = cms.PSet(
variables=cms.VPSet(),
classifier=cms.string("BDTCShape_EE"),
weights=cms.FileInPath("diphotons/Analysis/data/TMVAClassification_BDTCShape_EE.weights.xml")
## classifier=cms.string("BDTCShape_test_EE"),
## weights=cms.FileInPath("diphotons/Analysis/data/TMVAClassification_BDTCShape_test_EE.weights.xml")
)
variables = ["full5x5_sigmaIetaIeta","etaWidth := superCluster.etaWidth","phiWidth := superCluster.phiWidth",
"covarianceIetaIphi := sieip","r9","scEta := superCluster.eta","pt", "s4"
"hadronicOverEm"
]
## variables = ["full5x5_sigmaIetaIeta","etaWidth := superCluster.etaWidth","phiWidth := superCluster.phiWidth",
## "covarianceIetaIphi := sieip","r9","scEta := superCluster.eta","scRawEnergy := superCluster.rawEnergy","hadronicOverEm"
## ]
cfgTools.addVariables(cShapeMVA_EB.variables,variables)
cfgTools.addVariables(cShapeMVA_EE.variables,variables)