def getSummedShapes(path, tag, chan=''):
## FIXME get the following numbers from the file
lxy_max = 5.
lxy_bins = 50
if chan != '':
chan = '_' + chan
## prepare a summed histogram
summed_cat = []
## open the corresponfing file
tfile = mRF.openTFile(path)
# nevents = mRF.getHist(tfile,'map_counter').Integral(0,2)
nevents = 0.
for i in xrange(ybin - 1):
map_cat = mRF.getHist(tfile, 'map' + chan + '_cat' + str(i))
# n = map_cat.Integral(0,map_cat.GetNbinsX()+1)
n = map_cat.Integral()
nevents += n
checksum = 0.
checkweight = 0.
## loop over categories
for i in xrange(ybin - 1):
summed = TH1F(tag + '_summed_lxy' + chan + '_cat' + str(i),
tag + '_summed_lxy' + chan + '_cat' + str(i), lxy_bins,
0., lxy_max).Clone()
summed.Sumw2()
## get the map histo
map_cat = mRF.getHist(tfile, 'map' + chan + '_cat' + str(i))
map_cat.Sumw2()
checksum += map_cat.GetEntries()
## get the gamma binning per category
nbins = map_cat.GetNbinsX()
for j in xrange(nbins):
w = float(map_cat.GetBinContent(j + 1))
checkweight += w
if tag == 'templ':
hist = mRF.getHist(tfile,
tag + '_lxy_cat' + str(i) + '_bin' + str(j))
else:
hist = mRF.getHist(
tfile,
tag + '_lxy' + chan + '_cat' + str(i) + '_bin' + str(j))
hist.Sumw2()
## scale the histogram
integr = float(hist.Integral(0, hist.GetNbinsX() + 1))
if integr > 0 and tag == 'templ':
hist.Scale(w / integr)
## sum the histograms
summed.Add(hist.Clone())
summed_cat.append(summed)
print 'checksum (map): ', checksum
print 'checksum: ', sum([h.GetEntries() for h in summed_cat])
print 'checkweight: ', checkweight
return summed_cat
def getSummedShapes(path, tag, chan=''):
## FIXME get the following numbers from the file
lxy_max = 5.
lxy_bins = 50
if chan != '':
chan = '_'+chan
## prepare a summed histogram
summed_cat = []
## open the corresponfing file
tfile = mRF.openTFile(path)
# nevents = mRF.getHist(tfile,'map_counter').Integral(0,2)
nevents = 0.
for i in xrange(ybin-1):
map_cat = mRF.getHist(tfile,'map'+chan+'_cat'+str(i))
# n = map_cat.Integral(0,map_cat.GetNbinsX()+1)
n = map_cat.Integral()
nevents+=n
checksum = 0.
checkweight = 0.
## loop over categories
for i in xrange(ybin-1):
summed = TH1F(tag+'_summed_lxy'+chan+'_cat'+str(i),tag+'_summed_lxy'+chan+'_cat'+str(i),lxy_bins, 0., lxy_max).Clone()
summed.Sumw2()
## get the map histo
map_cat = mRF.getHist(tfile,'map'+chan+'_cat'+str(i))
map_cat.Sumw2()
checksum+=map_cat.GetEntries()
## get the gamma binning per category
nbins = map_cat.GetNbinsX()
for j in xrange(nbins):
w = float(map_cat.GetBinContent(j+1))
checkweight+=w
if tag=='templ':
hist = mRF.getHist(tfile,tag+'_lxy_cat'+str(i)+'_bin'+str(j))
else:
hist = mRF.getHist(tfile,tag+'_lxy'+chan+'_cat'+str(i)+'_bin'+str(j))
hist.Sumw2()
## scale the histogram
integr = float(hist.Integral(0, hist.GetNbinsX()+1))
if integr > 0 and tag=='templ':
hist.Scale(w/integr)
## sum the histograms
summed.Add(hist.Clone())
summed_cat.append(summed)
print 'checksum (map): ',checksum
print 'checksum: ',sum([h.GetEntries() for h in summed_cat])
print 'checkweight: ',checkweight
return summed_cat
def getNormalization(f,chan,tag,verbose=0):
h_wjets = mRF.getHist(f,'W+jets/'+tag)
h_dyll = mRF.getHist(f,'Z#rightarrow ll/'+tag)
h_stop = mRF.getHist(f,'Single top/'+tag)
h_ttbar = mRF.getHist(f, 't#bar{t}172.5/'+tag)
h_qcd = mRF.getHist(f, 'QCD/'+tag)
h_vv = mRF.getHist(f, 'VV/'+tag)
h_ttv = mRF.getHist(f, 'other t#bar{t}/'+tag)
h_others = h_qcd.Clone("others")
h_others.Add(h_vv)
h_others.Add(h_ttv)
h_data = mRF.getHist(f,'data/'+tag)
mcsum = h_wjets.Integral()+h_stop.Integral()+h_dyll.Integral()+h_qcd.Integral()+h_vv.Integral()+h_ttv.Integral()+h_ttbar.Integral()
if verbose > 0:
print '========================================================================='
print ' %-25s ' % tag
print '-----------------------------'
print ' %16s | %8.0f (total yields)' % ('Single Top', h_stop.Integral())
print ' %16s | %8.0f ' % ('Drell-Yan', h_dyll.Integral())
print ' %16s | %8.0f ' % ('QCD', h_qcd.Integral())
print ' %16s | %8.0f ' % ('VV', h_vv.Integral())
print ' %16s | %8.0f ' % ('ttW/Z', h_ttv.Integral())
print ' %16s | %8.0f ' % ('ttbar', h_ttbar.Integral())
print ' %16s | %8.0f ' % ('W+jets', h_wjets.Integral())
print '-----------------------------'
print ' %16s | %8.0f ' % ('Sum MC', mcsum)
print ' %16s | %8.0f ' % ('Data', h_data.Integral())
## sum single top to it
# st_samples = ['T_s-channel','T_t-channel','T_tW-channel','Tbar_s-channel','Tbar_t-channel','Tbar_tW-channel']
# st_weight = 1
# for st_sample in st_samples:
# f_st = mRF.openTFile(path+'/'+st_sample+'.root')
# h_st = mRF.getHist(f_st,'hist_charge'+chan)
# for key in topinfo.samples.keys():
# if st_sample in key:
# nevents = topinfo.samples[key][0]
# xsec = topinfo.samples[key][1]
# st_weight = topinfo.lumi[chan.replace('_','')] * xsec / nevents
# h_st.Scale(st_weight)
# h_wjets.Add(h_st)
## Yields
n_wjets_plus = h_wjets.GetBinContent(3)
n_wjets_minus = h_wjets.GetBinContent(1)
n_data_plus = h_data.GetBinContent(3)
n_data_minus = h_data.GetBinContent(1)
n_dyll_plus = h_dyll.GetBinContent(3)
n_dyll_minus = h_dyll.GetBinContent(1)
n_stop_plus = h_stop.GetBinContent(3)
n_stop_minus = h_stop.GetBinContent(1)
n_others_plus = h_others.GetBinContent(3)
n_others_minus = h_others.GetBinContent(1)
n_ttbar_plus = h_ttbar.GetBinContent(3)
n_ttbar_minus = h_ttbar.GetBinContent(1)
n_mc_plus = n_dyll_plus+n_stop_plus+n_others_plus+n_ttbar_plus #+ n_wjets_plus
n_mc_minus = n_dyll_minus+n_stop_minus+n_others_minus+n_ttbar_minus #+ n_wjets_minus
## Asymmetries
a_wjets = float(n_wjets_plus-n_wjets_minus)/float(n_wjets_plus+n_wjets_minus)
a_data = float(n_data_plus-n_data_minus)/float(n_data_plus+n_data_minus)
a_dyll = float(n_dyll_plus-n_dyll_minus)/float(n_dyll_plus+n_dyll_minus)
a_stop = float(n_stop_plus-n_stop_minus)/float(n_stop_plus+n_stop_minus)
a_others = float(n_others_plus-n_others_minus)/float(n_others_plus+n_others_minus)
a_ttbar = float(n_ttbar_plus-n_ttbar_minus)/float(n_ttbar_plus+n_ttbar_minus)
## Scale factors
n_corr = float(n_data_plus-n_data_minus) / a_wjets
n_uncorr = n_wjets_plus+n_wjets_minus
# n_uncorr = h_wjets.Integral(0,4) ## Juerg?
weight = float(h_wjets.Integral())/n_uncorr ## is always == 1.0?
## We're scaling actually the DIFFERENCE, not the asymmetry
w_sf = (n_data_plus-n_data_minus) / (n_wjets_plus-n_wjets_minus) ## == n_corr/n_uncorr
w_frac = n_corr / (n_data_plus+n_data_minus) ## Equal to a_data/a_wjets
## Errors
n_wjets_plus_err = h_wjets.GetBinError(3)
n_wjets_minus_err = h_wjets.GetBinError(1)
n_data_plus_err = h_data.GetBinError(3)
n_data_minus_err = h_data.GetBinError(1)
n_dyll_plus_err = h_dyll.GetBinError(3)
n_dyll_minus_err = h_dyll.GetBinError(1)
n_stop_plus_err = h_stop.GetBinError(3)
n_stop_minus_err = h_stop.GetBinError(1)
n_others_plus_err = h_others.GetBinError(3)
n_others_minus_err = h_others.GetBinError(1)
n_ttbar_plus_err = h_ttbar.GetBinError(3)
n_ttbar_minus_err = h_ttbar.GetBinError(1)
sum_wjets_err = sqrt(pow(n_wjets_plus_err,2)+pow(n_wjets_minus_err,2))
sum_data_err = sqrt(pow(n_data_plus_err,2)+pow(n_data_minus_err,2))
sum_dyll_err = sqrt(pow(n_dyll_plus_err,2)+pow(n_dyll_minus_err,2))
sum_stop_err = sqrt(pow(n_stop_plus_err,2)+pow(n_stop_minus_err,2))
sum_others_err = sqrt(pow(n_others_plus_err,2)+pow(n_others_minus_err,2))
sum_ttbar_err = sqrt(pow(n_ttbar_plus_err,2)+pow(n_ttbar_minus_err,2))
a_wjets_err = a_wjets * sqrt(pow((sum_wjets_err /float(n_wjets_plus-n_wjets_minus)),2) +pow((sum_wjets_err /float(n_wjets_plus+n_wjets_minus)),2))
a_data_err = a_data * sqrt(pow((sum_data_err/float(n_data_plus-n_data_minus)),2)+pow((sum_data_err/float(n_data_plus+n_data_minus)),2))
a_dyll_err = a_dyll * sqrt(pow((sum_dyll_err /float(n_dyll_plus-n_dyll_minus)),2) +pow((sum_dyll_err /float(n_dyll_plus+n_dyll_minus)),2))
a_stop_err = a_stop * sqrt(pow((sum_stop_err /float(n_stop_plus-n_stop_minus)),2) +pow((sum_stop_err /float(n_stop_plus+n_stop_minus)),2))
a_others_err = a_others * sqrt(pow((sum_others_err /float(n_others_plus-n_others_minus)),2) +pow((sum_others_err /float(n_others_plus+n_others_minus)),2))
a_ttbar_err = a_ttbar * sqrt(pow((sum_ttbar_err /float(n_ttbar_plus-n_ttbar_minus)),2) +pow((sum_ttbar_err /float(n_ttbar_plus+n_ttbar_minus)),2))
n_corr_err = n_corr * sqrt(pow(sum_data_err/float(n_data_plus-n_data_minus),2)+pow(a_wjets_err/a_wjets,2))
n_uncorr_err = sqrt(h_wjets.Integral())*weight
w_sf_err = w_sf * sqrt(pow(n_uncorr_err/n_uncorr,2)+pow(n_corr_err/n_corr,2))
if verbose == 0:
line = '%-16s : %5.2f+-%4.2f | %9.0f | %9.0f | %9.0f | %5.2f+-%4.2f | %5.2f+-%4.2f | %5.2f%% | %5.2f '
print line % (tag[:-7], w_sf, w_sf_err, n_data_plus-n_data_minus, n_mc_plus-n_mc_minus, n_wjets_plus-n_wjets_minus, a_data*100, a_data_err*100, a_wjets*100, a_wjets_err*100, (h_wjets.Integral()/mcsum)*100, h_data.Integral()/mcsum)
if verbose > 0:
print '-------------------------------------------------------------------'
print ' sample | N+ | N- | N+ - N- | Asymmetry |'
print '-------------------------------------------------------------------'
print ' Wjets | %9.0f | %9.0f | %9.0f | %7.4f +- %7.4f |' % ( n_wjets_plus, n_wjets_minus, (n_wjets_plus -n_wjets_minus), a_wjets, a_wjets_err)
print ' ttbar | %9.0f | %9.0f | %9.0f | %7.4f +- %7.4f |' % ( n_ttbar_plus, n_ttbar_minus, (n_ttbar_plus-n_ttbar_minus), a_ttbar, a_ttbar_err)
print ' s-top | %9.0f | %9.0f | %9.0f | %7.4f +- %7.4f |' % ( n_stop_plus, n_stop_minus, (n_stop_plus-n_stop_minus), a_stop, a_stop_err)
print ' DY+jets | %9.0f | %9.0f | %9.0f | %7.4f +- %7.4f |' % ( n_dyll_plus, n_dyll_minus, (n_dyll_plus-n_dyll_minus), a_dyll, a_dyll_err)
print ' others | %9.0f | %9.0f | %9.0f | %7.4f +- %7.4f |' % ( n_others_plus, n_others_minus, (n_others_plus-n_others_minus), a_others, a_others_err)
print ' Data | %9.0f | %9.0f | %9.0f | %7.4f +- %7.4f |' % ( n_data_plus, n_data_minus, (n_data_plus-n_data_minus), a_data, a_data_err)
print '-------------------------------------------------------------------'
print ' W_sf | %6.3f +- %6.3f ' % (w_sf, w_sf_err)
# print '----------------------------------------------------------------'
# print ' Ncorr | (%5.2f +- %5.2f) x 10^3' % (n_corr/1e3, n_corr_err/1e3)
print '----------------------------------------------------------------'
print ' This SF corresponds to an estimated fraction of Wjets of %5.2f%%' % (w_frac*100)
print ' (uncorrected fraction on MC was: %5.2f%%' % ((h_wjets.Integral()/mcsum)*100)
print '----------------------------------------------------------------'
n_wjets = [n_wjets_plus, n_wjets_minus, n_wjets_plus_err, n_wjets_minus_err]
n_data = [n_data_plus, n_data_minus, n_data_plus_err, n_data_minus_err]
a_wjets = [a_wjets, a_wjets_err]
a_data = [a_data, a_data_err]
n_corr = [n_corr, n_corr_err]
n_uncorr = [n_uncorr, n_uncorr]
return [w_sf, w_sf_err]
def main():
usage = 'usage: %prog [options]'
parser = optparse.OptionParser(usage)
parser.add_option('-i', '--inputfile' , dest='inputfile' , help='Name of the input file containing the histograms.' , default=None)
parser.add_option('-o', '--outputfile', dest='outputfile' , help='Name of the output file containing the histograms.', default='lxyhist.root')
parser.add_option('-b', '--batch' , dest='batch' , help='Use this flag to run root in batch mode.' , default=False, action='store_true')
(opt, args) = parser.parse_args()
if opt.inputfile is None:
parser.error('No input file defined!')
inputfile = opt.inputfile
outputfile = opt.outputfile
batch = opt.batch
if batch:
sys.argv.append( '-b' )
ROOT.gROOT.SetBatch()
i_file = mRF.openTFile(inputfile)
channels = ['e','mu']
procs = [k for k in mRF.GetKeyNames(i_file,'')]
hists = {}
for proc in procs:
keys = [k for k in mRF.GetKeyNames(i_file,proc)]
for key in keys:
## Select jetlxy or secvtxmass histograms for e and mu channels:
try: chan, var = key.split('_', 1)
except ValueError: continue
if not (chan == 'e' or chan == 'mu'): continue
if not (var == 'jetlxy' or var == 'secvtxmass'): continue
hist = mRF.getHist(i_file, proc + '/' + key)
if var == 'jetlxy':
var = var.replace('jet','max')
if var == 'secvtxmass':
var = var.replace('secvtx','jet')
## Extract generated top mass
if not 'ttbar' in proc_dict[proc]:
mass = '0'
else:
try:
mass = proc.split('}')[-1].replace('.','')
except IndexError:
mass = '1725'
if mass == '' or 'syst' in mass:
mass = '1725'
name = chan+'_'+var+'_'+proc_dict[proc]+'_'+mass
# print name
hists[name] = hist.Clone() ## FIXME (name.replace(' ',''))
o_file = ROOT.TFile.Open(outputfile,'recreate')
for chan in channels:
o_file.mkdir(chan)
for h in hists:
channel = h.split('_')[0]
if channel == 'e': o_file.cd(channel)
elif channel == 'mu': o_file.cd(channel)
else: continue
hists[h].Write()
o_file.Close()
def getNormalization(i_file,chan,level,outputdir=''):
postfix = chan+'_'+level
postfix_qcd = chan+'_antiiso_'+level
# data = mRF.getHist(i_file, chan+'/'+chan+'_'+level+'_data_0')
data = mRF.getHist(i_file,'data/'+postfix)
data.SetMarkerStyle(20)
data.GetXaxis().SetTitle('#slashed{E}_{T} [GeV]')
qcd_shape_hist = mRF.getHist(i_file,'data/'+postfix_qcd) ## take the qcd templates from antiiso data
# qcd = mRF.getHist(i_file, chan+'/'+chan+'_'+level+'_qcd_0')
qcd = mRF.getHist(i_file,'QCD/'+postfix)
qcd.SetLineWidth(2)
qcd.SetLineColor(qcd_color)
qcd.SetLineStyle(qcd_color)
qcd.GetXaxis().SetTitle('#slashed{E}_{T} [GeV]')
# wjets = mRF.getHist(i_file, chan+'/'+chan+'_'+level+'_wjets_0').Clone()
# ttbar = mRF.getHist(i_file, chan+'/'+chan+'_'+level+'_ttbar_1725').Clone()
# zjets = mRF.getHist(i_file, chan+'/'+chan+'_'+level+'_zjets_0').Clone()
# stop = mRF.getHist(i_file, chan+'/'+chan+'_'+level+'_singletop_0').Clone()
wjets = mRF.getHist(i_file,'W+jets/'+postfix).Clone()
ttbar = mRF.getHist(i_file,'t#bar{t}172.5/'+postfix).Clone()
zjets = mRF.getHist(i_file,'Z#rightarrow ll/'+postfix).Clone()
stop = mRF.getHist(i_file,'Single top/'+postfix).Clone()
vv = mRF.getHist(i_file,'VV/'+postfix).Clone()
mc = wjets.Clone('mc')
mc.SetLineWidth(2)
mc.SetLineColor(mc_color)
mc.SetLineStyle(mc_color)
mc.GetXaxis().SetTitle('#slashed{E}_{T} [GeV]')
mc.Add(zjets)
mc.Add(ttbar)
mc.Add(stop)
print 'Summary: ',level
print '-'*25
print 'ttbar : %9.0f' % ttbar.Integral()
print 'singletop : %9.0f' % stop.Integral()
print 'wjets : %9.0f' % wjets.Integral()
print 'zjets : %9.0f' % zjets.Integral()
print '-'*25
print 'mc sum : %9.0f' % mc.Integral()
print 'qcd : %9.0f' % qcd.Integral()
print '-'*25
print 'data : %9.0f' % data.Integral()
print '-'*25
frac_start = qcd.Integral(0, qcd.GetNbinsX()+1) / ( mc.Integral(0, mc.GetNbinsX()+1) + qcd.Integral(0, qcd.GetNbinsX()+1) )
start = 0.2
frac = RooRealVar('frac','frac',start,0.0,1.)
met = RooRealVar('met','met',0.,300.)
data_shape = RooDataHist('data','data', RooArgList(met),data)
mc_shape = RooDataHist('mc_shape','mc_shape', RooArgList(met),mc)
qcd_shape = RooDataHist('qcd_shape','qcd_shape',RooArgList(met),qcd_shape_hist)
mc_pdf = RooHistPdf('mc_pdf' ,'mc_pdf' ,RooArgSet(met),mc_shape)
qcd_pdf = RooHistPdf('qcd_pdf','qcd_pdf',RooArgSet(met),qcd_shape)
model = RooAddPdf('model', 'model', RooArgList(qcd_pdf, mc_pdf), RooArgList(frac))
model.fitTo(data_shape)
c = TCanvas('qcd_'+chan,'qcd_'+chan,600,600)
ROOT.gStyle.SetOptTitle(0)
ROOT.gStyle.SetOptStat(0)
frame = met.frame()
data_shape.plotOn(frame,RooFit.Name('data'))
model.plotOn(frame,RooFit.Name('model'))
model.plotOn(frame,RooFit.Components('mc_pdf'),RooFit.LineStyle(mc_color),RooFit.LineColor(mc_color))
model.plotOn(frame,RooFit.Components('qcd_pdf'),RooFit.LineStyle(qcd_color),RooFit.LineColor(qcd_color))
c.SetLeftMargin(0.2)
frame.GetXaxis().SetTitle('#slash{E}_{T} [GeV]')
frame.GetYaxis().SetTitleOffset(2.5)
frame.Draw()
leg = TLegend(0.6,0.65,0.85,0.85)
leg.SetFillColor(0)
leg.SetLineColor(0)
leg.SetHeader(chandict[chan]+'-Channel')
leg.AddEntry(data,'Data','p')
leg.AddEntry(mc,'t#bar{t},t,W+jets,DY','l')
leg.AddEntry(qcd,'QCD','l')
leg.Draw()
pave = TPaveText(0.6,0.5,0.85,0.65,"NDC")
pave.SetBorderSize(0)
pave.SetFillStyle(0)
# pave.SetTextAlign(32)
# pave.SetTextFont(42)
pave.AddText('f_{QCD} = %.3f' % frac.getVal()+' #pm %.3f ' % frac.getError())
pave.AddText('#chi^{2}/N_{dof} = %.2f' % frame.chiSquare('model','data'))
pave.AddText('SF = %.2f' % float(frac.getVal()/frac_start))
pave.Draw()
outdir = ''
if len(outputdir)>0:
outdir = createDir(outputdir)
c.Update()
c.Print(outdir+'qcd_'+chan+'_'+level+'.png')
c.Print(outdir+'qcd_'+chan+'_'+level+'.pdf')
# c.Print('qcd_'+chan+'_'+level+'.eps')
# c.Print('qcd_'+chan+'_'+level+'.C')
return float(frac.getVal()/frac_start),0.