def computeCrossRatioFromQCDMC(self):
'''
Study the correlation of two variables computing the ratio
r = N_A * ND / N_B * N_C
'''
quantity = Quantity(name_flat='hnl_mass', nbins=1, bin_min=0, bin_max=5.4)
#hist_mc_tot_A = self.getHistoMC(quantity=quantity, selection=self.selection+' && b_mass<6.27 && hnl_charge==0')
#hist_mc_tot_A = self.getHistoMC(quantity=quantity, selection=self.selection+' && hnl_cos2d>0.993 && hnl_charge==0')
#hist_mc_tot_A = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && b_mass<6.27 && hnl_charge==0')
hist_mc_tot_A = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && hnl_cos2d>0.993 && sv_prob>0.05')
#hist_mc_tot_B = self.getHistoMC(quantity=quantity, selection=self.selection+' && b_mass<6.27 && hnl_charge!=0')
#hist_mc_tot_B = self.getHistoMC(quantity=quantity, selection=self.selection+' && hnl_cos2d>0.993 && hnl_charge!=0')
#hist_mc_tot_B = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && b_mass<6.27 && hnl_charge!=0')
hist_mc_tot_B = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && hnl_cos2d>0.993 && sv_prob<0.05')
#hist_mc_tot_C = self.getHistoMC(quantity=quantity, selection=self.selection+' && b_mass>6.27 && hnl_charge==0')
#hist_mc_tot_C = self.getHistoMC(quantity=quantity, selection=self.selection+' && hnl_cos2d<0.993 && hnl_charge==0')
#hist_mc_tot_C = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && b_mass>6.27 && hnl_charge==0')
hist_mc_tot_C = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && hnl_cos2d<0.993 && sv_prob>0.05')
#hist_mc_tot_D = self.getHistoMC(quantity=quantity, selection=self.selection+' && b_mass>6.27 && hnl_charge!=0')
#hist_mc_tot_D = self.getHistoMC(quantity=quantity, selection=self.selection+' && hnl_cos2d<0.993 && hnl_charge!=0')
#hist_mc_tot_D = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && b_mass>6.27 && hnl_charge!=0')
hist_mc_tot_D = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && hnl_cos2d<0.993 && sv_prob<0.05')
n_obs_mc_A = hist_mc_tot_A.GetBinContent(1)
n_obs_mc_B = hist_mc_tot_B.GetBinContent(1)
n_obs_mc_C = hist_mc_tot_C.GetBinContent(1)
n_obs_mc_D = hist_mc_tot_D.GetBinContent(1)
n_err_mc_A = hist_mc_tot_A.GetBinError(1) #math.sqrt(n_obs_mc_A)
n_err_mc_B = hist_mc_tot_B.GetBinError(1) #math.sqrt(n_obs_mc_B)
n_err_mc_C = hist_mc_tot_C.GetBinError(1) #math.sqrt(n_obs_mc_C)
n_err_mc_D = hist_mc_tot_D.GetBinError(1) #math.sqrt(n_obs_mc_D)
ratio = n_obs_mc_A*n_obs_mc_D/(n_obs_mc_B*n_obs_mc_C)
err = ratio * (n_err_mc_A/n_obs_mc_A + n_err_mc_B/n_obs_mc_B + n_err_mc_C/n_obs_mc_C + n_err_mc_D/n_obs_mc_D)
return ratio, err
def computeBkgYieldsFromMC(self):
'''
QCD MC (background) yields are computed as N_exp(SR) = N_obs(SR) * weight(CR)
'''
#quantity = Quantity(name_flat='hnl_mass', nbins=1, bin_min=0, bin_max=1000)
quantity = Quantity(name_flat='hnl_mass', nbins=1, bin_min=2.83268, bin_max=3.13932)
selection_extra = self.selection
# we compute the weight in the control region
weight, err_weight = self.computeApproxWeightQCDMCtoData(quantity, selection=('hnl_charge!=0' if selection_extra=='' else 'hnl_charge!=0 &&' + selection_extra))
#weight, err_weight = self.computeWeightQCDMCtoData(lumi_data=774)
#hist_mc_tot = self.getHistoMC(quantity=quantity, selection='hnl_charge==0' if selection_extra=='' else 'hnl_charge==0 &&' + selection_extra)
hist_mc_tot = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection='hnl_charge==0' if selection_extra=='' else 'hnl_charge==0 &&' + selection_extra)
n_obs_mc = hist_mc_tot.GetBinContent(1)
n_err_mc = math.sqrt(n_obs_mc) #hist_mc_tot.GetBinError(1)
n_exp_mc = n_obs_mc * weight
err = n_exp_mc * (n_err_mc / n_obs_mc + err_weight / weight)
return int(n_exp_mc), int(err), weight
def computeApproxWeightQCDMCtoData(self, quantity, selection):
'''
weight = lumi_data / lumi_mc = N_data * sigma_mc / (N_mc * sigma_data) estimated as N_data / N_mc
'''
f_data = ROOT.TFile.Open('root://t3dcachedb.psi.ch:1094/'+self.data_file.filename, 'READ')
hist_data = PlottingTools.createHisto(self, f_data, 'signal_tree', quantity, branchname='flat', selection=selection)
hist_data.Sumw2()
n_obs_data = hist_data.GetBinContent(1)
n_err_data = math.sqrt(n_obs_data) #hist_data.GetBinError(1)
hist_mc_tot = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=selection)
n_obs_mc = hist_mc_tot.GetBinContent(1)
n_err_mc = math.sqrt(n_obs_mc) #hist_mc_tot.GetBinError(1)
weight = n_obs_data / n_obs_mc
if n_obs_data != 0 and n_obs_mc != 0:
err = weight* (n_err_data / n_obs_data + n_err_mc / n_obs_mc)
else:
err = 0
return weight, err
def validateABCDOnQCDMC(self, plot_name='closure', label=''):
'''
Get mupi mass ditribution from data using the ABCD method
A = b_mass < 6.27 && hnl_charge == 0 (SR)
B = b_mass < 6.27 && hnl_charge != 0
C = b_mass > 6.27 && hnl_charge == 0
D = b_mass > 6.27 && hnl_charge != 0
and compare it to the actual distribution in the SR
'''
#quantity = Quantity(name_flat='hnl_mass', nbins=100, bin_min=0, bin_max=5.6)
quantity = Quantity(name_flat='hnl_mass', nbins=50, bin_min=0, bin_max=5.37)
bin_selection = self.selection
#hist_mc_tot_A = self.getHistoMC(quantity=quantity, selection=self.selection+' && hnl_cos2d>0.993 && sv_prob>0.05')
#hist_mc_tot_B = self.getHistoMC(quantity=quantity, selection=self.selection+' && b_mass<6.27 && hnl_charge!=0')
#hist_mc_tot_C = self.getHistoMC(quantity=quantity, selection=self.selection+' && b_mass>6.27 && hnl_charge==0')
#hist_mc_tot_D = self.getHistoMC(quantity=quantity, selection=self.selection+' && b_mass>6.27 && hnl_charge!=0')
#hist_mc_tot_A = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && b_mass<6.27 && hnl_charge==0')
#hist_mc_tot_B = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && b_mass<6.27 && hnl_charge!=0')
#hist_mc_tot_C = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && b_mass>6.27 && hnl_charge==0')
#hist_mc_tot_D = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && b_mass>6.27 && hnl_charge!=0')
hist_mc_tot_A = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && hnl_cos2d>0.993 && sv_prob>0.05')
hist_mc_tot_B = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && hnl_cos2d>0.993 && sv_prob<0.05')
hist_mc_tot_C = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && hnl_cos2d<0.993 && sv_prob>0.05')
hist_mc_tot_D = PlottingTools.createWeightedHistoQCDMC(self, self.qcd_files, self.white_list, quantity=quantity, selection=self.selection+' && hnl_cos2d<0.993 && sv_prob<0.05')
#hist_estimate = ROOT.TH1D('hist_estimate', 'hist_estimate', 100, 0, 5.6)
hist_estimate = ROOT.TH1D('hist_estimate', 'hist_estimate', 50, 0, 5.37)
for ibin in range(1, hist_mc_tot_B.GetNbinsX()+1):
n_obs_mc_B = hist_mc_tot_B.GetBinContent(ibin)
n_obs_mc_C = hist_mc_tot_C.GetBinContent(ibin)
n_obs_mc_D = hist_mc_tot_D.GetBinContent(ibin)
n_err_mc_B = hist_mc_tot_B.GetBinError(ibin)
n_err_mc_C = hist_mc_tot_C.GetBinError(ibin)
n_err_mc_D = hist_mc_tot_D.GetBinError(ibin)
content = n_obs_mc_B * (n_obs_mc_C / n_obs_mc_D) if n_obs_mc_C != 0 and n_obs_mc_D != 0 else 0
err = content * (n_err_mc_B / n_obs_mc_B + n_err_mc_C / n_obs_mc_C + n_err_mc_D / n_obs_mc_D) if n_obs_mc_B!=0 and n_obs_mc_C!=0 and n_obs_mc_D!=0 else 0
hist_estimate.SetBinContent(ibin, content)
hist_estimate.SetBinError(ibin, err)
ROOT.gStyle.SetOptStat(0)
canv = PlottingTools.getTCanvas(self, 'canv', 800, 900)
pad_up = ROOT.TPad("pad_up","pad_up",0,0.25,1,1)
pad_up.SetBottomMargin(0.03)
pad_up.Draw()
canv.cd()
pad_down = ROOT.TPad("pad_down","pad_down",0,0,1,0.25)
pad_down.SetBottomMargin(0.25)
pad_down.Draw()
hist_estimate.SetLineWidth(3)
hist_estimate.SetLineColor(ROOT.kOrange)
#hist_estimate.SetTitle('Closure of the ABCD method in the Signal Region')
hist_estimate.SetTitle('')
#hist_estimate.GetXaxis().SetTitle('#mu#pi invariant mass [GeV]')
hist_estimate.GetXaxis().SetLabelSize(0.0)
hist_estimate.GetXaxis().SetTitleSize(0.0)
#hist_estimate.GetXaxis().SetTitleOffset(1.1)
hist_estimate.GetYaxis().SetTitle('Entries')
hist_estimate.GetYaxis().SetLabelSize(0.037)
hist_estimate.GetYaxis().SetTitleSize(0.042)
hist_estimate.GetYaxis().SetTitleOffset(1.1)
#hist_estimate.GetYaxis().SetRangeUser(1e-9, self.getMaxRangeY(hist_estimate, hist_mc_stack, do_log))
hist_mc_tot_A.SetLineWidth(3)
hist_mc_tot_A.SetLineColor(ROOT.kBlue)
int_estimate = hist_estimate.Integral()
hist_estimate.Scale(1/int_estimate)
int_A = hist_mc_tot_A.Integral()
hist_mc_tot_A.Scale(1/int_A)
pad_up.cd()
hist_estimate.Draw('histo')
hist_mc_tot_A.Draw('histo same')
legend = PlottingTools.getRootTLegend(self, xmin=0.55, ymin=0.65, xmax=0.8, ymax=0.85, size=0.043)
legend.AddEntry(hist_estimate, 'QCD ABCD estimate')
legend.AddEntry(hist_mc_tot_A, 'QCD true')
legend.Draw()
label_box = ROOT.TPaveText(0.65,0.4,0.8,0.6, "brNDC")
label_box.SetBorderSize(0)
label_box.SetFillColor(ROOT.kWhite)
label_box.SetTextSize(0.042)
label_box.SetTextAlign(11)
label_box.SetTextFont(42)
label_box.AddText(label)
label_box.Draw()
pad_down.cd()
hist_ratio = PlottingTools.getRatioHistogram(self, hist_estimate, hist_mc_tot_A)
hist_ratio.Sumw2()
for ibin in range(0, hist_ratio.GetNbinsX()+1):
if hist_estimate.GetBinContent(ibin) != 0 and hist_mc_tot_A.GetBinContent(ibin) != 0:
#err = hist_ratio.GetBinContent(ibin) * (math.sqrt(hist_data.GetBinContent(ibin))/hist_data.GetBinContent(ibin) + math.sqrt(hist_mc_tot.GetBinContent(ibin))/hist_mc_tot.GetBinContent(ibin))
#err = math.sqrt((math.sqrt(hist_data.GetBinContent(ibin))/hist_mc_tot.GetBinContent(ibin))**2 + (math.sqrt(hist_mc_tot.GetBinContent(ibin))*hist_data.GetBinContent(ibin)/(hist_mc_tot.GetBinContent(ibin))**2)**2)
#err = math.sqrt((hist_data.GetBinError(ibin)/hist_mc_tot.GetBinContent(ibin))**2 + (hist_mc_tot.GetBinError(ibin)*hist_data.GetBinContent(ibin)/(hist_mc_tot.GetBinContent(ibin))**2)**2)
err = hist_ratio.GetBinContent(ibin) * (hist_estimate.GetBinError(ibin) / hist_estimate.GetBinContent(ibin) + hist_mc_tot_A.GetBinError(ibin) / hist_mc_tot_A.GetBinContent(ibin))
else:
err = 0
if hist_ratio.GetBinContent(ibin) != 0: hist_ratio.SetBinError(ibin, err)
hist_ratio.SetLineWidth(2)
hist_ratio.SetLineColor(ROOT.kBlack)
hist_ratio.SetMarkerStyle(20)
hist_ratio.SetTitle('')
hist_ratio.GetXaxis().SetTitle('#mu#pi invariant mass [GeV]')
hist_ratio.GetXaxis().SetLabelSize(0.1)
hist_ratio.GetXaxis().SetTitleSize(0.15)
hist_ratio.GetXaxis().SetTitleOffset(0.73)
hist_ratio.GetYaxis().SetTitle('Estimate/True')
hist_ratio.GetYaxis().SetLabelSize(0.1)
hist_ratio.GetYaxis().SetTitleSize(0.13)
hist_ratio.GetYaxis().SetTitleOffset(0.345)
#val_min = hist_ratio.GetBinContent(hist_ratio.GetMinimumBin())
#val_max = hist_ratio.GetBinContent(hist_ratio.GetMaximumBin())
hist_ratio.GetYaxis().SetRangeUser(-1, 3)
hist_ratio.Draw('PE')
line = ROOT.TLine(0, 1, 5.6, 1)
line.SetLineColor(4)
line.SetLineWidth(2)
line.Draw('same')
canv.cd()
if not path.exists('./myPlots/ABCDClosure'):
os.system('mkdir -p ./myPlots/ABCDClosure')
canv.SaveAs('./myPlots/ABCDClosure/{}.png'.format(plot_name))
canv.SaveAs('./myPlots/ABCDClosure/{}.pdf'.format(plot_name))
canv.SaveAs('./myPlots/ABCDClosure/{}.C'.format(plot_name))