def getCounter(self, datasetName, name):
if not datasetName in self.residualNames:
return self.datasetsEmb.getCounter(datasetName, name)
# Ugly hack
sigName = name
if isinstance(sigName, basestring):
sigName = sigName.replace(self.analysisEmb, self.analysisSig)
else:
sigName = sigName.clone(tree=sigName.tree.replace(self.analysisEmb, self.analysisSig))
# Get properly normalized embedded data, embedded DY and normal DY histograms
(embHisto, tmp) = self.datasetsEmb.getHistogram(datasetName, name)
sigHisto = self.datasetsSig.getDataset(datasetName).getDatasetRootHisto(sigName) # DatasetRootHisto
sigHisto.normalizeToLuminosity(self.datasetsEmb.getLuminosity())
sigHisto = sigHisto.getHistogram() # ROOT.TH1
table = counter.CounterTable()
table.appendColumn(counter.HistoCounter("Embedded", embHisto))
table.appendColumn(counter.HistoCounter("Normal", sigHisto))
table.removeNonFullRows()
embColumn = table.getColumn(name="Embedded")
sigColumn = table.getColumn(name="Normal")
residual = counter.subtractColumn(datasetName+" residual", sigColumn, embColumn)
return residual
def getCounter(self, datasetName, name):
if not datasetName in self.residualNames:
return self.datasetsEmb.getCounter(datasetName, name)
# Ugly hack
sigName = name
if isinstance(sigName, basestring):
sigName = sigName.replace(self.analysisEmb, self.analysisSig)
else:
sigName = sigName.clone(
tree=sigName.tree.replace(self.analysisEmb, self.analysisSig))
# Get properly normalized embedded data, embedded DY and normal DY histograms
(embHisto, tmp) = self.datasetsEmb.getHistogram(datasetName, name)
sigHisto = self.datasetsSig.getDataset(
datasetName).getDatasetRootHisto(sigName) # DatasetRootHisto
sigHisto.normalizeToLuminosity(self.datasetsEmb.getLuminosity())
sigHisto = sigHisto.getHistogram() # ROOT.TH1
table = counter.CounterTable()
table.appendColumn(counter.HistoCounter("Embedded", embHisto))
table.appendColumn(counter.HistoCounter("Normal", sigHisto))
table.removeNonFullRows()
embColumn = table.getColumn(name="Embedded")
sigColumn = table.getColumn(name="Normal")
residual = counter.subtractColumn(datasetName + " residual", sigColumn,
embColumn)
return residual
def doCounters(datasets, MyCuts, MyCutsName):
### Define the counters to be used
#eventCounter = counter.EventCounter(datasets, counters=myAnalysis + myDataEra + "/counters")
eventCounter = counter.EventCounter(datasets)
### Normalise the MC sample to a luminosity before creating a table
eventCounter.normalizeMCToLuminosity(GetLumi(datasets))
myRowNames = []
for iCut, iCutName in zip(MyCuts, MyCutsName):
#print "*** Cut: %s\n*** CutName: %s" % (iCut, iCutName)
print "*** Processing TCut with:\n Name = \"%s\" \n Expr = \"%s\"" % (iCutName, iCut)
### Define the event "weight" to be used
EvtWeight = GetEventWeight(iCut)
### Define the TTree to be used
treePath = "tree" # treePath = myAnalysis+"/tree"
treeDraw = dataset.TreeDraw(treePath, weight=EvtWeight, selection=iCut)
### Append custom rows to the event counter. An asterisk denotes that the counter row was added. Informative and makes things easier
myRowName = iCutName #"*" + iCutName
myRowNames.append(myRowName)
eventCounter.getMainCounter().appendRow(myRowName, treeDraw)
### Get table with all default rows removed and manage the table format
myTable = GetCustomTable(eventCounter, myRowNames)
if getBool("bMergeEwk"):
DataMinusEwkMc = counter.subtractColumn("DataMinusEwkMc", myTable.getColumn(name="Data"), myTable.getColumn(name="EWK MC"))
QcdPurity = counter.divideColumn("QCD Purity", DataMinusEwkMc, myTable.getColumn(name="Data"))
myTable.appendColumn(QcdPurity)
### See http://docs.python.org/2/library/string.html for string format
cellTextFormat = counter.TableFormatText(counter.CellFormatTeX(valueFormat='%.2E', valueOnly=True)) #%.2e
cellLaTeXFormat = counter.TableFormatLaTeX(counter.CellFormatTeX(valueFormat='%.2E', valueOnly=True)) #%.2e
purityFormat = counter.CellFormatTeX(valueFormat='%.2f', valueOnly=True)
### Customise the "QCD Purity" column
cellTextFormat.setColumnFormat(purityFormat, name="QCD Purity") #does nothing
cellLaTeXFormat.setColumnFormat(purityFormat, name="QCD Purity") #does nothing
# between construction of table format and table format
### Print the final table with the desired format
print "============================================================"
print "Data-Era: %s (%s pb-1)" % (myDataEra, GetLumi(datasets))
print "============================================================"
print myTable.format(cellTextFormat)
print "============================================================"
print "Data-Era: %s (%s pb-1)" % (myDataEra, GetLumi(datasets))
print "============================================================"
print myTable.format(cellLaTeXFormat)
return
def _calculateResidual(self, table, sigTable):
columnNames = table.getColumnNames()
for name in columnNames:
if name in self.residualNames:
i = columnNames.index(name)
col = table.getColumn(index=i)
table.removeColumn(i)
col = counter.subtractColumn(name+" residual", sigTable.getColumn(name=name), col)
table.insertColumn(i, col)
return table
def _calculateResidual(self, table, sigTable):
columnNames = table.getColumnNames()
for name in columnNames:
if name in self.residualNames:
i = columnNames.index(name)
col = table.getColumn(index=i)
table.removeColumn(i)
col = counter.subtractColumn(name + " residual",
sigTable.getColumn(name=name),
col)
table.insertColumn(i, col)
return table
def writeRootHisto(self, directory, dstName, srcEmbName, srcSigName,
isCounter):
if not self.datasetsEmb.hasHistogram(
self.datasetName,
srcEmbName) or not self.datasetSig.hasRootHisto(srcSigName):
return
histo = None
# Get properly normalized embedded and normal MC
(embMcHisto,
tmp) = self.datasetsEmb.getHistogram(self.datasetName, srcEmbName)
sigMcHisto = self.datasetSig.getDatasetRootHisto(
srcSigName).getHistogram() # ROOT.TH1
#print embMcHisto, sigMcHisto
if self.normalOnly:
histo = sigMcHisto
elif self.embeddedOnly:
histo = embMcHisto
else:
if isCounter:
embMcCounter = counter.HistoCounter("EmbMc", embMcHisto)
sigMcCounter = counter.HistoCounter("SigMc", sigMcHisto)
table = counter.CounterTable()
table.appendColumn(embMcCounter)
table.appendColumn(sigMcCounter)
table.removeNonFullRows()
if table.getNrows() == 0:
return
embMcColumn = table.getColumn(name="EmbMc")
sigMcColumn = table.getColumn(name="SigMc")
residual = counter.subtractColumn("Correction", sigMcColumn,
embMcColumn)
histo = dataset._counterToHisto(dstName,
residual.getPairList())
else:
# Residual MC: normal-embedded
sigMcHisto.Add(embMcHisto, -1)
histo = sigMcHisto
histo.SetName(dstName)
histo.SetDirectory(directory)
histo.Write()
histo.Delete() # could this help???
print "%s/%s" % (directory.GetPath(), dstName)
def getSubCounterTable(self, name):
table = self.eventCounterEmb.getSubCounterTable(name)
sigTable = self.eventCounterSig.getSubCounterTable(name)
sigDyColumn = sigTable.getColumn(name="DYJetsToLL")
embDyColumn = table.getColumn(name="DYJetsToLL")
dyCorrection = counter.subtractColumn("Correction", sigDyColumn, embDyColumn)
if "Data" in table.getColumnNames():
self._correctColumn(table, "Data", dyCorrection)
self._correctColumn(table, "DYJetsToLL", dyCorrection)
return table
def getMainCounterTable(self):
table = self.eventCounterEmb.getMainCounterTable()
sigTable = self.eventCounterSig.getMainCounterTable()
sigDyColumn = sigTable.getColumn(name="DYJetsToLL")
embDyColumn = table.getColumn(name="DYJetsToLL")
dyCorrection = counter.subtractColumn("Correction", sigDyColumn,
embDyColumn)
if "Data" in table.getColumnNames():
self._correctColumn(table, "Data", dyCorrection)
self._correctColumn(table, "DYJetsToLL", dyCorrection)
return table
def writeRootHisto(self, directory, dstName, srcEmbName, srcSigName, isCounter):
if not self.datasetsEmb.hasHistogram(self.datasetName, srcEmbName) or not self.datasetSig.hasRootHisto(srcSigName):
return
histo = None
# Get properly normalized embedded and normal MC
(embMcHisto, tmp) = self.datasetsEmb.getHistogram(self.datasetName, srcEmbName)
sigMcHisto = self.datasetSig.getDatasetRootHisto(srcSigName).getHistogram() # ROOT.TH1
#print embMcHisto, sigMcHisto
if self.normalOnly:
histo = sigMcHisto
elif self.embeddedOnly:
histo = embMcHisto
else:
if isCounter:
embMcCounter = counter.HistoCounter("EmbMc", embMcHisto)
sigMcCounter = counter.HistoCounter("SigMc", sigMcHisto)
table = counter.CounterTable()
table.appendColumn(embMcCounter)
table.appendColumn(sigMcCounter)
table.removeNonFullRows()
if table.getNrows() == 0:
return
embMcColumn = table.getColumn(name="EmbMc")
sigMcColumn = table.getColumn(name="SigMc")
residual = counter.subtractColumn("Correction", sigMcColumn, embMcColumn)
histo = dataset._counterToHisto(dstName, residual.getPairList())
else:
# Residual MC: normal-embedded
sigMcHisto.Add(embMcHisto, -1)
histo = sigMcHisto
histo.SetName(dstName)
histo.SetDirectory(directory)
histo.Write()
histo.Delete() # could this help???
print "%s/%s" % (directory.GetPath(), dstName)
def getCounter(self, datasetName, name):
if not datasetName in ["Data", "EWKMC", "DYJetsToLL"]:
(embDataHisto,
tmp) = self.datasetsEmb.getHistogram(datasetName, name)
return counter.HistoCounter(datasetName, embDataHisto)
# Ugly hack
sigName = name
if isinstance(sigName, basestring):
sigName = sigName.replace(self.analysisEmb, self.analysisSig)
else:
sigName = sigName.clone(
tree=sigName.tree.replace(self.analysisEmb, self.analysisSig))
# Get properly normalized embedded data, embedded DY and normal DY histograms
(embDataHisto, tmp) = self.datasetsEmb.getHistogram(datasetName, name)
(embDyHisto, tmp) = self.datasetsEmb.getHistogram("DYJetsToLL", name)
sigDyHisto = self.datasetsSig.getDataset(
"DYJetsToLL").getDatasetRootHisto(sigName) # DatasetRootHisto
sigDyHisto.normalizeToLuminosity(self.datasetsEmb.getLuminosity())
sigDyHisto = sigDyHisto.getHistogram() # ROOT.TH1
embDataCounter = counter.HistoCounter("EmbData", embDataHisto)
embDyCounter = counter.HistoCounter("EmbDy", embDyHisto)
sigDyCounter = counter.HistoCounter("SigDy", sigDyHisto)
table = counter.CounterTable()
table.appendColumn(embDataCounter)
table.appendColumn(embDyCounter)
table.appendColumn(sigDyCounter)
table.removeNonFullRows()
column = table.getColumn(name="EmbData")
embDyColumn = table.getColumn(name="EmbDy")
sigDyColumn = table.getColumn(name="SigDy")
dyCorrection = counter.subtractColumn("Correction", sigDyColumn,
embDyColumn)
column = counter.sumColumn(datasetName, [column, dyCorrection])
return column
def getCounter(self, datasetName, name):
if not datasetName in ["Data", "EWKMC", "DYJetsToLL"]:
(embDataHisto, tmp) = self.datasetsEmb.getHistogram(datasetName, name)
return counter.HistoCounter(datasetName, embDataHisto)
# Ugly hack
sigName = name
if isinstance(sigName, basestring):
sigName = sigName.replace(self.analysisEmb, self.analysisSig)
else:
sigName = sigName.clone(tree=sigName.tree.replace(self.analysisEmb, self.analysisSig))
# Get properly normalized embedded data, embedded DY and normal DY histograms
(embDataHisto, tmp) = self.datasetsEmb.getHistogram(datasetName, name)
(embDyHisto, tmp) = self.datasetsEmb.getHistogram("DYJetsToLL", name)
sigDyHisto = self.datasetsSig.getDataset("DYJetsToLL").getDatasetRootHisto(sigName) # DatasetRootHisto
sigDyHisto.normalizeToLuminosity(self.datasetsEmb.getLuminosity())
sigDyHisto = sigDyHisto.getHistogram() # ROOT.TH1
embDataCounter = counter.HistoCounter("EmbData", embDataHisto)
embDyCounter = counter.HistoCounter("EmbDy", embDyHisto)
sigDyCounter = counter.HistoCounter("SigDy", sigDyHisto)
table = counter.CounterTable()
table.appendColumn(embDataCounter)
table.appendColumn(embDyCounter)
table.appendColumn(sigDyCounter)
table.removeNonFullRows()
column = table.getColumn(name="EmbData")
embDyColumn = table.getColumn(name="EmbDy")
sigDyColumn = table.getColumn(name="SigDy")
dyCorrection = counter.subtractColumn("Correction", sigDyColumn, embDyColumn)
column = counter.sumColumn(datasetName, [column, dyCorrection])
return column
def doCounters(datasets, MyCuts, MyCutsName):
### Define the counters to be used
#eventCounter = counter.EventCounter(datasets, counters=myAnalysis + myDataEra + "/counters")
eventCounter = counter.EventCounter(datasets)
### Normalise the MC sample to a luminosity before creating a table
eventCounter.normalizeMCToLuminosity(GetLumi(datasets))
myRowNames = []
for iCut, iCutName in zip(MyCuts, MyCutsName):
#print "*** Cut: %s\n*** CutName: %s" % (iCut, iCutName)
print "*** Processing TCut with:\n Name = \"%s\" \n Expr = \"%s\"" % (
iCutName, iCut)
### Define the event "weight" to be used
EvtWeight = GetEventWeight(iCut)
### Define the TTree to be used
treePath = "tree" # treePath = myAnalysis+"/tree"
treeDraw = dataset.TreeDraw(treePath, weight=EvtWeight, selection=iCut)
### Append custom rows to the event counter. An asterisk denotes that the counter row was added. Informative and makes things easier
myRowName = iCutName #"*" + iCutName
myRowNames.append(myRowName)
eventCounter.getMainCounter().appendRow(myRowName, treeDraw)
### Get table with all default rows removed and manage the table format
myTable = GetCustomTable(eventCounter, myRowNames)
if getBool("bMergeEwk"):
DataMinusEwkMc = counter.subtractColumn(
"DataMinusEwkMc", myTable.getColumn(name="Data"),
myTable.getColumn(name="EWK MC"))
QcdPurity = counter.divideColumn("QCD Purity", DataMinusEwkMc,
myTable.getColumn(name="Data"))
myTable.appendColumn(QcdPurity)
### See http://docs.python.org/2/library/string.html for string format
cellTextFormat = counter.TableFormatText(
counter.CellFormatTeX(valueFormat='%.2E', valueOnly=True)) #%.2e
cellLaTeXFormat = counter.TableFormatLaTeX(
counter.CellFormatTeX(valueFormat='%.2E', valueOnly=True)) #%.2e
purityFormat = counter.CellFormatTeX(valueFormat='%.2f', valueOnly=True)
### Customise the "QCD Purity" column
cellTextFormat.setColumnFormat(purityFormat,
name="QCD Purity") #does nothing
cellLaTeXFormat.setColumnFormat(purityFormat,
name="QCD Purity") #does nothing
# between construction of table format and table format
### Print the final table with the desired format
print "============================================================"
print "Data-Era: %s (%s pb-1)" % (myDataEra, GetLumi(datasets))
print "============================================================"
print myTable.format(cellTextFormat)
print "============================================================"
print "Data-Era: %s (%s pb-1)" % (myDataEra, GetLumi(datasets))
print "============================================================"
print myTable.format(cellLaTeXFormat)
return
def processDirectory(dset, srcDirName, dstDir, scaleBy, dsetSubtractFrom):
# Get directories, recurse to them
dirs = dset.getDirectoryContent(srcDirName,
lambda o: isinstance(o, ROOT.TDirectory))
dirs = filter(lambda n: n != "configInfo", dirs)
for d in dirs:
newDir = dstDir.mkdir(d)
processDirectory(dset, os.path.join(srcDirName, d), newDir, scaleBy,
dsetSubtractFrom)
# Then process histograms
histos = dset.getDirectoryContent(srcDirName,
lambda o: isinstance(o, ROOT.TH1))
dstDir.cd()
shouldScale = True
if srcDirName == "counters":
# Don't touch unweighted counters
shouldScale = False
isCounter = srcDirName in ["counters", "counters/weighted"]
for hname in histos:
# drh = dset.getDatasetRootHisto(os.path.join(srcDirName, hname))
# hnew = drh.getHistogram() # TH1
hnew = dset.getAverageHistogram(
os.path.join(srcDirName, hname),
normalizeMCByCrossSection=(dsetSubtractFrom is not None))
hnew.SetName(hname)
if shouldScale and hname not in "SplittedBinInfo":
tauEmbedding.scaleTauBRNormalization(hnew)
if scaleBy is not None:
hnew.Scale(scaleBy)
if dsetSubtractFrom is not None:
drh = dsetSubtractFrom.getDatasetRootHisto(
os.path.join(srcDirName, hname))
if dsetSubtractFrom.isMC():
drh.normalizeByCrossSection()
hsub = drh.getHistogram()
if not isCounter:
# hnew = hsub-hnew
hnew.Scale(-1)
hnew.Add(hsub)
else:
cnew = counter.HistoCounter("Emb", hnew)
csub = counter.HistoCounter("Norm", hsub)
table = counter.CounterTable()
table.appendColumn(cnew)
table.appendColumn(csub)
cres = counter.subtractColumn("Result",
table.getColumn(name="Norm"),
table.getColumn(name="Emb"))
hnew2 = dataset._counterToHisto(hnew.GetName(),
cres.getPairList())
hnew2.SetTitle(hnew.GetTitle())
hnew = hnew2
if srcDirName == "counters" and hname == "counter" and hnew.GetBinContent(
1) < 0:
hnew.SetBinContent(1, 0)
hnew.SetDirectory(dstDir)
# # set the first count in main counters to 0 if it is negative,
# # this is to circumvent certain assumptions made elsewhere in
# # the code
# if hname == "counter" and (srcDirName == "counters" or srcDirName == "counters/weighted") and hnew.GetBinContent(1) < 0:
# hnew.SetBinContent(1, 0)
# hnew.SetBinError(1, 0)
hnew.Write()
# ROOT.gDirectory.Delete(hname)
hnew.Delete()
def processDirectory(dset, srcDirName, dstDir, scaleBy, dsetSubtractFrom):
# Get directories, recurse to them
dirs = dset.getDirectoryContent(srcDirName, lambda o: isinstance(o, ROOT.TDirectory))
dirs = filter(lambda n: n != "configInfo", dirs)
for d in dirs:
newDir = dstDir.mkdir(d)
processDirectory(dset, os.path.join(srcDirName, d), newDir, scaleBy, dsetSubtractFrom)
# Then process histograms
histos = dset.getDirectoryContent(srcDirName, lambda o: isinstance(o, ROOT.TH1))
dstDir.cd()
shouldScale = True
if srcDirName == "counters":
# Don't touch unweighted counters
shouldScale = False
isCounter = srcDirName in ["counters", "counters/weighted"]
for hname in histos:
# drh = dset.getDatasetRootHisto(os.path.join(srcDirName, hname))
# hnew = drh.getHistogram() # TH1
hnew = dset.getAverageHistogram(os.path.join(srcDirName, hname), normalizeMCByCrossSection = (dsetSubtractFrom is not None))
hnew.SetName(hname)
if shouldScale and hname not in "SplittedBinInfo":
tauEmbedding.scaleTauBRNormalization(hnew)
if scaleBy is not None:
hnew.Scale(scaleBy)
if dsetSubtractFrom is not None:
drh = dsetSubtractFrom.getDatasetRootHisto(os.path.join(srcDirName, hname))
if dsetSubtractFrom.isMC():
drh.normalizeByCrossSection()
hsub = drh.getHistogram()
if not isCounter:
# hnew = hsub-hnew
hnew.Scale(-1)
hnew.Add(hsub)
else:
cnew = counter.HistoCounter("Emb", hnew)
csub = counter.HistoCounter("Norm", hsub)
table = counter.CounterTable()
table.appendColumn(cnew)
table.appendColumn(csub)
cres = counter.subtractColumn("Result", table.getColumn(name="Norm"), table.getColumn(name="Emb"))
hnew2 = dataset._counterToHisto(hnew.GetName(), cres.getPairList())
hnew2.SetTitle(hnew.GetTitle())
hnew = hnew2
if srcDirName == "counters" and hname == "counter" and hnew.GetBinContent(1) < 0:
hnew.SetBinContent(1, 0)
hnew.SetDirectory(dstDir)
# # set the first count in main counters to 0 if it is negative,
# # this is to circumvent certain assumptions made elsewhere in
# # the code
# if hname == "counter" and (srcDirName == "counters" or srcDirName == "counters/weighted") and hnew.GetBinContent(1) < 0:
# hnew.SetBinContent(1, 0)
# hnew.SetBinError(1, 0)
hnew.Write()
# ROOT.gDirectory.Delete(hname)
hnew.Delete()