def doNeyman(responseMatrix,comphep1000Matrix,comphep1100Matrix,comphep0100Matrix,N):
result=[]
comphep1000Gen=comphep1000Matrix.ProjectionX("comphep1000Gen")
comphep1100Gen=comphep1100Matrix.ProjectionX("comphep1100Gen")
comphep0100Gen=comphep0100Matrix.ProjectionX("comphep0100Gen")
comphep1000Reco=comphep1000Matrix.ProjectionY("comphep1000Reco")
comphep1100Reco=comphep1100Matrix.ProjectionY("comphep1100Reco")
comphep0100Reco=comphep0100Matrix.ProjectionY("comphep0100Reco")
'''
cv=ROOT.TCanvas("cv","",800,600)
cv.Divide(1,2)
cv.cd(1)
comphep1000Gen.Draw()
comphep1100Gen.Draw("Same")
comphep0100Gen.Draw("Same")
cv.cd(2)
comphep1000Reco.Draw()
comphep1100Reco.Draw("Same")
comphep0100Reco.Draw("Same")
cv.Update()
cv.WaitPrimitive()
'''
tmHistinverted=u2b.readResponseFromHistogramAndInvert(responseMatrix)
for i in range(N):
vl=1.0
vr=1.0*i/(N-1)
genDist=generateWeightedHist(comphep1000Gen,comphep1100Gen,comphep0100Gen,vl,vr)
recoDist=generateWeightedHist(comphep1000Reco,comphep1100Reco,comphep0100Reco,vl,vr)
measured = u2b.DataDistribution.createFromHistogram(recoDist,True,False,0.0)
unfolded= u2b.CompoundDistribution(u2b.Unfolding(tmHistinverted),measured)
asymmetry=u2b.CompoundDistribution(u2b.Asymmetry(),unfolded)
result.append({
"vl":vl,
"vr":vr,
"generated":calcAsymmetry(genDist),
"unfolded":asymmetry.getMean(0),
"uncertainty":asymmetry.getUncertainty(0),
"bias":math.fabs(calcAsymmetry(genDist)-asymmetry.getMean(0))
})
return result
def doUnfolding(
histFiles,
signalHistName,
backgroundHistNames,
dataHistNames,
responseFiles,
responseHistName,
fitResult,
systematic="nominal",
useStatUnc=True,
useMCStatUnc=True,
useFitUnc=True,
):
measured = None
if options.mc_only:
# fake data by adding up all nominal histograms
fakeData = None
for histName in fitResult.keys():
if histName == signalHistName or histName in backgroundHistNames:
if fakeData == None:
fakeData = readHist1d(
histFiles, HISTPREFIX + histName, "nominal", fitResult[histName]["scale"], REBIN_RECO
)
else:
fakeData.Add(
readHist1d(
histFiles, HISTPREFIX + histName, "nominal", fitResult[histName]["scale"], REBIN_RECO
)
)
measured = u2b.DataDistribution.createFromHistogram(fakeData, useStatUnc, False, 0.0)
else:
# read data histograms
data = None
for histName in dataHistNames:
if data == None:
data = readHist1d(histFiles, HISTPREFIX + histName, "nominal", 1, REBIN_RECO)
measured = u2b.DataDistribution.createFromHistogram(data, useStatUnc, False, 0.0)
# TODO: decorrelate before subtracting - or define correlation between bgs
for histName in fitResult.keys():
if histName in backgroundHistNames:
backgroundHist = readHist1d(
histFiles, HISTPREFIX + histName, systematic, fitResult[histName]["scale"], REBIN_RECO
)
backgroundDist = u2b.DataDistribution.createFromHistogram(
backgroundHist, False, useMCStatUnc, fitResult[histName]["uncertainty"] if useFitUnc else 0.0
)
# subtract background from measured data
measured = u2b.CompoundDistribution(u2b.Subtraction(), measured, backgroundDist)
tmHist = readHist2d(
responseFiles, responseHistName, systematic, fitResult[signalHistName]["scale"], REBIN_GEN, REBIN_RECO
)
tmHistinverted = u2b.readResponseFromHistogramAndInvert(tmHist)
if verbose:
print "measured after bg subtraction:"
print " ... m1=", measured.getMean(0), " +- ", measured.getUncertainty(0)
print " ... m1=", measured.getMeanSampled(0), " +- ", measured.getUncertaintySampled(0), " (toys)"
print " ... m2=", measured.getMean(1), " +- ", measured.getUncertainty(1)
print " ... m1=", measured.getMeanSampled(1), " +- ", measured.getUncertaintySampled(1), " (toys)"
print " ... rho=", measured.getCorrelation(0, 1)
print " ... rho=", measured.getCorrelationSampled(0, 1), " (toys)"
print
unfolded = u2b.CompoundDistribution(u2b.Unfolding(tmHistinverted), measured)
if verbose:
print "unfolded distribution:"
print " ... u1=", unfolded.getMean(0), " +- ", unfolded.getUncertainty(0)
print " ... u1=", unfolded.getMeanSampled(0), " +- ", unfolded.getUncertaintySampled(0), " (toys)"
print " ... u2=", unfolded.getMean(1), " +- ", unfolded.getUncertainty(1)
print " ... u2=", unfolded.getMeanSampled(1), " +- ", unfolded.getUncertaintySampled(1), " (toys)"
print " ... rho=", unfolded.getCorrelation(0, 1)
print " ... rho=", unfolded.getCorrelationSampled(0, 1), " (toys)"
print
# calculate asymmetry
asymmetry = u2b.CompoundDistribution(u2b.Asymmetry(), unfolded)
if verbose:
print "final result:"
print " ... A=", asymmetry.getMean(0), " +- ", asymmetry.getUncertainty(0)
return {"mean": asymmetry.getMean(0), "uncertainty": asymmetry.getUncertainty(0)}
