def Plot2d(dataType, outputDir, datasets, mcsets, histlist):
#
# this function is basically the 2d analogon to Plot1d: create 2d plots according to the histograms in the list
# NOTE: ALL SAMPLES MUST CONTAIN THE EXACT SAME LIST OF HISTOGRAMS, OTHERWISE THIS FUNCTION WILL NOT WORK
#
# this function takes the following parameters:
# dataType....type of the lepton ('mu', 'el')
# outputDir...basic output directory
# datasets....list of data samples [datasample1, datasample2, ..]
# mcsets......list of mc samples [mcsample1, mcsample2, ..]
# histlist....list of 1d histogram names ['h_Loose_LepIso', 'h_Loose_LepPhi', ..]
# define canvas and pads
canv = helper.makeCanvas(900, 675, 'c2d')
pad_plot = helper.makePad('tot')
pad_plot.cd()
pad_plot.SetTicks(1,1)
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for hist in datasets[0].hists:
# get index of the histogram
i = datasets[0].hists.index(hist)
# only histograms in the parameter histlist are plotted
if not hist.GetName() in histlist: continue
# pre- and postpends
prepend = ''
postpend = ''
if '_Loose_' in hist.GetName(): prepend = 'Loose_'
if '_Tight_' in hist.GetName(): prepend = 'Tight_'
# we stack both data and mc samples (i.e. one may use several data and several mc samples)
data = ROOT.THStack()
mc = ROOT.THStack()
for dataset in datasets: data.Add(dataset.hists[i])
for mcset in mcsets: mc .Add(mcset .hists[i])
# call make2dPlot to make plots for data, total MC, each mc sample
make2dPlot(dataType, canv, pad_plot, outputDir, data.GetStack().Last(), 'data', prepend + helper.getSaveName(hist) + postpend)
make2dPlot(dataType, canv, pad_plot, outputDir, mc .GetStack().Last(), 'MC' , prepend + helper.getSaveName(hist) + postpend)
for mcset in mcsets: make2dPlot(dataType, canv, pad_plot, outputDir, mcset.hists[i], mcset.GetName(), prepend + helper.getSaveName(hist) + postpend)
def make2dFRPlot(dataType, canv, outputDir, dataset, hist, title_indeces, name='', exportinroot = False, folder = 'fakerates_2d/', isprompt = False):
#
# produces all 2d fake rate maps
#
# this function takes the following paramters:
# dataType..........type of the lepton ('mu', 'el')
# canv..............canvas to be plotted
# outputDir.........basic output directory
# dataset...........dataset (only to take the title of the X and Y axis from)
# hist..............histogram to be plotted
# title_indeces.....list of two elements which are the indeces of the histograms that carry the X anc Y axis titles in their X axis title
# name..............name of the output file
# exportinroot......True if we want to export the canvas in a root file
# folder............one may change the last outputfolder to something else than fakerates_2d
# isprompt..........temporary solution to plot prompt ratio
# define pad
pad_plot = helper.makePad('tot')
pad_plot.cd()
pad_plot.SetTicks(1,1)
# draw histogram
hist.Draw("text colz e")
# cosmetics
hist.SetMarkerColor(ROOT.kBlack)
hist.SetMarkerSize(1.8)
hist.GetXaxis().SetTitle(helper.getXTitle(dataType, dataset.hists[title_indeces[0]]))
hist.GetYaxis().SetTitle(helper.getXTitle(dataType, dataset.hists[title_indeces[1]]))
hist.GetXaxis().SetTitleSize(0.07)
hist.GetXaxis().SetLabelSize(0.07)
hist.GetYaxis().SetTitleSize(0.07)
hist.GetYaxis().SetLabelSize(0.07)
hist.SetMinimum(0.0)
if dataType == 'el': hist.SetMaximum(0.6)
else : hist.SetMaximum(0.4)
if isprompt == True: prepend = 'PR'
else : prepend = 'FR'
hist.SetTitle(prepend + " 2d Map (" + helper.getLegendName(name) + ")")
# save plot
helper.saveCanvas(canv, pad_plot, outputDir + folder, prepend + "_2dmap_" + name.lower().replace(" ", "_"), False, exportinroot)
pad_plot.Close()
import ROOT, copy, array
import lib as helper
ROOT.gROOT.SetBatch(1)
ROOT.gStyle.SetOptStat(0)
ROOT.gStyle.SetPaintTextFormat("4.3f")
ROOT.TGaxis.SetMaxDigits(3)
file = ROOT.TFile('stufffortalk/ssFR_data_ewkcor_sync16Apr2014f.root', 'read')
canv = helper.makeCanvas(900, 675)
pad_plot = helper.makePad('tot')
pad_plot.cd()
pad_plot.SetTicks(1,1)
hold = file.Get('h_mufr40c')
etabin = array.array('d', [0.0, 0.5, 1.0, 1.5, 2.0, 2.5])
ptbins = array.array('d', [10., 15., 20., 25., 30., 35., 45., 50.])
hist = ROOT.TH2F('hist', 'hist', len(ptbins)-1, ptbins, len(etabin)-1, etabin)
print hold.GetNbinsX()
print hold.GetNbinsY()
print hist.GetNbinsX()
print hist.GetNbinsY()
for y in range(1, hold.GetNbinsY()+1):
for x in range(1, hold.GetNbinsX()+1):
hist.SetBinContent(y,x,hold.GetBinContent(x,y))
print 'x=' + str(x) + ', y=' + str(y) + ', value=' + str(hold.GetBinContent(x,y))
def canvasWithRatio(stack, histo, legend):
c1 = ROOT.TCanvas('canvas', 'canvas', 675, 675)
pad_plot = lib.makePad('plot')
pad_ratio = lib.makePad('ratio')
pad_plot.SetTicks(1,1)
pad_ratio.SetTicks(1,1)
stackcp = copy.deepcopy(stack)
stackhisto = copy.deepcopy(stackcp.GetStack().Last())
hlist = stackcp.GetHists()
newstack = ROOT.THStack('st', 'st')
for i in hlist:
i.Scale(1./stackhisto.Integral())
newstack.Add(copy.deepcopy(i))
pad_plot.cd()
stackcp.Draw('hist')
stackcp.GetYaxis().SetLabelSize(0.05)
histoerr = copy.deepcopy(stackhisto)
histoerr.SetFillColor(ROOT.kGray+3)
histoerr.SetLineColor(ROOT.kGray+3)
histoerr.SetFillStyle(3004)
histoerr.Draw('same e2')
#newstack.Draw('hist')
histo.SetMarkerStyle(20)
histo.SetMarkerSize(0.9)
histo.SetMarkerColor(ROOT.kBlack)
histo.SetLineColor(ROOT.kBlack)
histonorm = copy.deepcopy(histo)
#histonorm.Scale(1./histonorm.Integral())
histonorm.Draw('same pe')
stackmax = stackcp.GetMaximum()
histomax = histo.GetMaximum()
stackcp.SetMaximum(1.15* max(stackmax, histomax) )
legend.Draw('same')
pad_ratio.cd()
#stackhisto.Scale(1./stackhisto.Integral())
histocp = copy.deepcopy(histo)
histocp.SetName('ratio')
histocp.SetStats(0)
#histocp.Scale(1./histocp.Integral())
histocp.Divide(stackhisto)
histocp.GetYaxis().SetRangeUser(0.00, 2.0)
histocp.Draw('pe')
histocp.GetYaxis().SetNdivisions(505)
histocp.GetYaxis().SetLabelSize(0.08)
histocp.GetXaxis().SetLabelSize(0.12)
histocp.GetXaxis().SetTitle(histocp.GetName().split('_')[0])
histocp.GetXaxis().SetTitleOffset(1.15)
histocp.GetXaxis().SetTitleSize(0.15)
histocp.SetTitle('')
fl = ROOT.TF1("fl","[0]*x+[1]",0.,1.)
histocp.Fit(fl)
fl.Draw('same')
line = lib.makeLine(histocp.GetXaxis().GetXmin(), 1., histocp.GetXaxis().GetXmax(), 1.)
line.Draw('same')
pad_ratio.Draw()
pad_ratio.Update()
c1.Update()
return c1, stackhisto, histocp, stackcp, newstack, histonorm, line, histoerr
if module == 'fakerates_2dct' or module == 'all':
FR.Plot2dFRMapClosureTest(dataType, outputDir, module, data_samples, mc_samples, closure_samples, [wjets, dyjets50, dyjets10], qcd_samples)
if module == 'closure' or module == 'all':
Plot.PlotProvenance(dataType, outputDir, closure_samples, [3,4,5,6])
# hard-coded to produce FR vs LepEta and LepPt plots for different jet cuts and different Jet Pts
if module == 'fakerates_1d' or module == 'all':
canv = helper.makeCanvas(900, 675)
pad_plot = helper.makePad('plot')
pad_ratio = helper.makePad('ratio')
for hist in data.hists:
i = data.hists.index(hist)
# Get Numerator Plots
if hist.GetName() == 'h_Tight_LepEta_30':
histindex_eta = i
data_numerator_eta30 = copy.deepcopy(hist)
if hist.GetName() == 'h_Tight_LepEta_40':
data_numerator_eta40 = copy.deepcopy(hist)
if hist.GetName() == 'h_Tight_LepEta_50':
data_numerator_eta50 = copy.deepcopy(hist)
def Plot2dFRMap(dataType, outputDir, module, datasets, mcsets, mcsetsplot = [], mcsubtract = [], mcsubtractplot = [], doProjection = False, mcsubtractscales = False):
#
# produce the fake rate 2d maps (incl. ewk subtraction) for given samples
# NOTE: ALL SAMPLES MUST CONTAIN THE EXACT SAME LIST OF HISTOGRAMS, OTHERWISE THIS FUNCTION WILL NOT WORK
#
# this function takes the following paramters:
# dataType..........type of the lepton ('mu', 'el')
# outputDir.........basic output directory
# module............module (the 1d projections on lepton eta and lepton pt are only created for the module 'all')
# datasets..........list of data samples [datasample1, datasample2, ..]
# mcsets............list of mc samples [mcsample1, mcsample2, ..] to enter bg stack
# histlist..........list of 1d histogram names ['h_Loose_LepIso', 'h_Loose_LepPhi', ..]
# mcsetsplot........list of mc samples [mcsample1, mcsample2, ..] to be drawn seperately
# mcsubtract........list of mc samples [mcsample1, mcsample2, ..] to enter ewk subtraction
# mcsubtractplot....list of mc samples [mcsample1, mcsample2, ..] to be drawn in comparison with ewk subtraction
# doProjection......True if the 1d projections on lepton eta and lepton pt bins shall be produced as well
# mcsubtractscales..True if we first want to scale ewk according to ETH/UCSx methods
# default for ewk scales (central1 is ewk scale according to ETH method with lower and upper limit, central2 is scale according to UCSx)
central1 = 1.0
lower = 1.0
upper = 1.0
central2 = 1.0
# compute the scales according to ETH and UCSx method
# this part needs improvement
if mcsubtractscales:
scfirst = fit.getMCScaleFactorSimultaneouslyWithErrors(datasets, mcsubtractplot, mcsubtract)
central1 = scfirst[0][1]
lower = scfirst[1][1]
upper = scfirst[2][1]
scsecond = fit.getMCScaleFactorMutually(mcsubtract, 'h_Tight_MTMET30', datasets, mcsubtractplot, 60, 100)
central2 = scsecond[0]
print "------**------"
print "qcd = " + str(scfirst[0][0])
print "central 1 = " + str(central1)
print "lower 1 = " + str(lower)
print "upper 1 = " + str(upper)
print "central 2 = " + str(central2)
print "------++------"
# define canvas
canv = helper.makeCanvas(900, 675, 'c2dFR')
# we add data and mc in stacks for both numerator and denominator
index_numerator = 0
index_denominator = 0
FR_mcplot = [{} for i in range(len(mcsetsplot))]
FR_mcplot_copy = [{} for i in range(len(mcsetsplot))]
mcplot_denominator = [{} for i in range(len(mcsetsplot))]
title_indeces = [0, 0]
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for hist in datasets[0].hists:
# get index of histogram
i = datasets[0].hists.index(hist)
# get indeces of histograms which carry title for X axis (0) and Y axis (1)
if hist.GetName() == 'h_Loose_LepEta':
title_indeces[1] = i
if hist.GetName() == 'h_Loose_LepPt':
title_indeces[0] = i
# get numerator histograms
if hist.GetName()[-8:] == 'h_FTight':
index_numerator = i
data_numerator = ROOT.THStack()
mc_numerator = ROOT.THStack()
mcsub_numerator = ROOT.THStack()
for data in datasets: data_numerator .Add(copy.deepcopy(data.hists[index_numerator]))
for mc in mcsets: mc_numerator .Add(copy.deepcopy(mc .hists[index_numerator]))
for j, mc in enumerate(mcsetsplot): FR_mcplot[j] = copy.deepcopy(mc .hists[index_numerator])
for mc in mcsubtractplot: mcsub_numerator .Add(copy.deepcopy(mc .hists[index_numerator]))
# get denominator histograms
if hist.GetName()[-8:] == 'h_PTight':
mc_numerator_p = ROOT.THStack()
for mc in mcsets:
if 'dyjets50' in mc.GetName():
mc_numerator_p .Add(copy.deepcopy(mc .hists[i]))
# get denominator histograms
if hist.GetName()[-8:] == 'h_FLoose':
index_denominator = i
data_denominator = ROOT.THStack()
mc_denominator = ROOT.THStack()
mcsub_denominator = ROOT.THStack()
for data in datasets: data_denominator .Add(copy.deepcopy(data.hists[index_denominator]))
for mc in mcsets: mc_denominator .Add(copy.deepcopy(mc .hists[index_denominator]))
for j, mc in enumerate(mcsetsplot): mcplot_denominator[j] = copy.deepcopy(mc .hists[index_denominator])
for mc in mcsubtractplot: mcsub_denominator .Add(copy.deepcopy(mc .hists[index_denominator]))
# get denominator histograms
if hist.GetName()[-8:] == 'h_PLoose':
mc_denominator_p = ROOT.THStack()
for mc in mcsets:
if 'dyjets50' in mc.GetName():
mc_denominator_p .Add(copy.deepcopy(mc .hists[i]))
# Get Numerator CERN Histogram
if hist.GetName()[-19:] == 'h_FTight_CERN_small':
data_numerator_CERN_small = ROOT.THStack()
for data in datasets: data_numerator_CERN_small .Add(copy.deepcopy(data.hists[i]))
# Get Numerator CERN Histogram
if hist.GetName()[-19:] == 'h_FTight_CERN_large':
data_numerator_CERN_large = ROOT.THStack()
for data in datasets: data_numerator_CERN_large .Add(copy.deepcopy(data.hists[i]))
# Get Numerator CERN Histogram
if hist.GetName()[-19:] == 'h_FLoose_CERN_small':
data_denominator_CERN_small = ROOT.THStack()
for data in datasets: data_denominator_CERN_small .Add(copy.deepcopy(data.hists[i]))
# Get Numerator CERN Histogram
if hist.GetName()[-19:] == 'h_FLoose_CERN_small':
data_denominator_CERN_large = ROOT.THStack()
for data in datasets: data_denominator_CERN_large .Add(copy.deepcopy(data.hists[i]))
# get numerator histograms from stack
FR_data = copy.deepcopy(data_numerator .GetStack().Last())
FR_mc = copy.deepcopy(mc_numerator .GetStack().Last())
PR_mc = copy.deepcopy(mc_numerator_p .GetStack().Last())
FR_mcsub = copy.deepcopy(mcsub_numerator .GetStack().Last())
# we copy the numerators for the projections
FR_data_copy = copy.deepcopy(FR_data)
FR_mc_copy = copy.deepcopy(FR_mc)
PR_mc_copy = copy.deepcopy(PR_mc)
FR_mcsub_copy = copy.deepcopy(FR_mcsub)
for j in range(len(mcsetsplot)): FR_mcplot_copy[j] = copy.deepcopy(FR_mcplot[j])
# ewk numerators
FR_data_mcsub = copy.deepcopy(FR_data)
FR_data_mcsub_c1 = copy.deepcopy(FR_data)
FR_data_mcsub_l1 = copy.deepcopy(FR_data)
FR_data_mcsub_u1 = copy.deepcopy(FR_data)
FR_data_mcsub_c2 = copy.deepcopy(FR_data)
# CERN method (not working yet)
FR_data_CERN_small = copy.deepcopy(data_numerator_CERN_small.GetStack().Last())
FR_data_CERN_large = copy.deepcopy(data_numerator_CERN_large.GetStack().Last())
FR_data_CERN_small.Divide(FR_data_CERN_small, copy.deepcopy(data_denominator_CERN_small.GetStack().Last()), 1, 1, 'B')
FR_data_CERN_large.Divide(FR_data_CERN_large, copy.deepcopy(data_denominator_CERN_large.GetStack().Last()), 1, 1, 'B')
FR_data_mcsub_c3 = copy.deepcopy(FR_data)
FR_data_mcsub_c3.Divide(FR_data_mcsub_c3, copy.deepcopy(data_denominator.GetStack().Last()), 1, 1, 'B')
FR_data_mcsub_c3 = DoMCSubCERN(FR_data_mcsub_c3, FR_data_CERN_small, FR_data_CERN_large, 447731, 517016, 23680, 10089) # numbers from loose MET
#FR_data_mcsub_c3 = DoMCSubCERN(FR_data_mcsub_c3, FR_data_CERN_small, FR_data_CERN_large, 108963, 27751, 847, 270) # numbers from tight MET
# you gotta fill in the numbers by hand (i know, not very nice indeed) from the counters fCounter_CERN_small/-large from Fakerates.cc
# ewk denominators
data_denominator_mcsub = copy.deepcopy(data_denominator.GetStack().Last())
data_denominator_mcsub_c1 = copy.deepcopy(data_denominator.GetStack().Last())
data_denominator_mcsub_l1 = copy.deepcopy(data_denominator.GetStack().Last())
data_denominator_mcsub_u1 = copy.deepcopy(data_denominator.GetStack().Last())
data_denominator_mcsub_c2 = copy.deepcopy(data_denominator.GetStack().Last())
# create 2d PLOT
# we subtract every mc set in mcsubtract from the data 5 times (unscaled, ETH, lower, upper, UCSx)
# both for numerator and denominator (WATCH OUT FOR THE SCALING!)
if len(mcsubtract)>0:
for mc in mcsets:
# numerator unscaled
FR_data_mcsub.Add(mc.hists[index_numerator], -1)
# numerator ETH central
mc.hists[index_numerator].Scale(central1)
FR_data_mcsub_c1.Add(mc.hists[index_numerator], -1)
# numerator ETH lower bound
mc.hists[index_numerator].Scale(lower/central1)
FR_data_mcsub_l1.Add(mc.hists[index_numerator], -1)
# numerator ETH upper bound
mc.hists[index_numerator].Scale(upper/lower)
FR_data_mcsub_u1.Add(mc.hists[index_numerator], -1)
# numerator UCSx central
mc.hists[index_numerator].Scale(central2/upper)
FR_data_mcsub_c2.Add(mc.hists[index_numerator], -1)
# rescale it to 1 again
mc.hists[index_numerator].Scale(1/central2)
# denominator unscaled
data_denominator_mcsub.Add(mc.hists[index_denominator], -1)
# denominator ETH central
mc.hists[index_denominator].Scale(central1)
data_denominator_mcsub_c1.Add(mc.hists[index_denominator],-1)
# denominator ETH lower bound
mc.hists[index_denominator].Scale(lower/central1)
data_denominator_mcsub_l1.Add(mc.hists[index_denominator],-1)
# denominator ETH upper bound
mc.hists[index_denominator].Scale(upper/lower)
data_denominator_mcsub_u1.Add(mc.hists[index_denominator],-1)
# denominator UCSx central
mc.hists[index_denominator].Scale(central2/upper)
data_denominator_mcsub_c2.Add(mc.hists[index_denominator],-1)
# rescale it to 1 again
mc.hists[index_denominator].Scale(1/central2)
# ewk subtracted numerators
data_numerator_mcsub = copy.deepcopy(FR_data_mcsub )
data_numerator_mcsub_c1 = copy.deepcopy(FR_data_mcsub_c1)
data_numerator_mcsub_l1 = copy.deepcopy(FR_data_mcsub_l1)
data_numerator_mcsub_u1 = copy.deepcopy(FR_data_mcsub_u1)
data_numerator_mcsub_c2 = copy.deepcopy(FR_data_mcsub_c2)
# ewk subtracted fake rates
FR_data_mcsub .Divide(FR_data_mcsub , data_denominator_mcsub , 1, 1, '')
FR_data_mcsub_c1.Divide(FR_data_mcsub_c1, data_denominator_mcsub_c1, 1, 1, '')
FR_data_mcsub_l1.Divide(FR_data_mcsub_l1, data_denominator_mcsub_l1, 1, 1, '')
FR_data_mcsub_u1.Divide(FR_data_mcsub_u1, data_denominator_mcsub_u1, 1, 1, '')
FR_data_mcsub_c2.Divide(FR_data_mcsub_c2, data_denominator_mcsub_c2, 1, 1, '')
data_den = data_denominator .GetStack().Last()
#make2dFRPlot(dataType, canv, outputDir, datasets[0], data_den , title_indeces, 'data_den', True)
# fake rates for pure data, total mc and every single mc sample
FR_data .Divide(FR_data , copy.deepcopy(data_denominator .GetStack().Last()), 1, 1, 'B')
FR_mc .Divide(FR_mc , copy.deepcopy(mc_denominator .GetStack().Last()), 1, 1, 'B')
PR_mc .Divide(PR_mc , copy.deepcopy(mc_denominator_p .GetStack().Last()), 1, 1, 'B')
FR_mcsub .Divide(FR_mcsub , copy.deepcopy(mcsub_denominator .GetStack().Last()), 1, 1, 'B')
for j in range(len(mcsetsplot)): FR_mcplot[j].Divide(FR_mcplot[j], copy.deepcopy(mcplot_denominator[j]), 1, 1, 'B')
# call make2dFRPlot to produce fake rate maps
# pure data
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_data , title_indeces, 'data')
# total mc
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_mc , title_indeces, 'mc' )
# total mc (prompt ratio)
make2dFRPlot(dataType, canv, outputDir, datasets[0], PR_mc , title_indeces, 'mc' , False, 'fakerates_2d/', True)
# mcsubtractplot (i.e. qcd)
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_mcsub , title_indeces, mcsubtractplot[0].GetName())
# every single mc sample
if len(mcsetsplot)>0:
for j in range(len(mcsetsplot)):
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_mcplot[j], title_indeces, mcsetsplot[j].GetName())
# ewk subtracted unscaled
if len(mcsubtract)>0:
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_data_mcsub, title_indeces, 'datamcsub')
# ewk subtracted ETH, lower, upper, UCSx, CERN
if mcsubtractscales:
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_data_mcsub_c1, title_indeces, 'datamcsub_central1')
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_data_mcsub_l1, title_indeces, 'datamcsub_lower1' )
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_data_mcsub_u1, title_indeces, 'datamcsub_upper1' )
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_data_mcsub_c2, title_indeces, 'datamcsub_central2')
make2dFRPlot(dataType, canv, outputDir, datasets[0], FR_data_mcsub_c3, title_indeces, 'datamcsub_central3')
# do make the lepton eta and lepton pt projections if module == 'all'
if doProjection == True and module == 'all':
# re-adjust the canvas margin and create some pads as we're goin 1d now, bro!
canv.SetRightMargin(0.0)
pad_plot = helper.makePad('plot')
pad_ratio = helper.makePad('ratio')
# projection on lepton pt goes first
# ewk subtracted numerators
FR_data_px_mcsub = copy.deepcopy(data_numerator_mcsub .ProjectionX())
FR_data_px_mcsub_c1 = copy.deepcopy(data_numerator_mcsub_c1.ProjectionX())
FR_data_px_mcsub_l1 = copy.deepcopy(data_numerator_mcsub_l1.ProjectionX())
FR_data_px_mcsub_u1 = copy.deepcopy(data_numerator_mcsub_u1.ProjectionX())
FR_data_px_mcsub_c2 = copy.deepcopy(data_numerator_mcsub_c2.ProjectionX())
# ewk subtracted fake rates
FR_data_px_mcsub .Divide(FR_data_px_mcsub , data_denominator_mcsub .ProjectionX(), 1, 1, '')
FR_data_px_mcsub_c1.Divide(FR_data_px_mcsub_c1, data_denominator_mcsub_c1.ProjectionX(), 1, 1, '')
FR_data_px_mcsub_l1.Divide(FR_data_px_mcsub_l1, data_denominator_mcsub_l1.ProjectionX(), 1, 1, '')
FR_data_px_mcsub_u1.Divide(FR_data_px_mcsub_u1, data_denominator_mcsub_u1.ProjectionX(), 1, 1, '')
FR_data_px_mcsub_c2.Divide(FR_data_px_mcsub_c2, data_denominator_mcsub_c2.ProjectionX(), 1, 1, '')
# normal numerators
FR_data_px = copy.deepcopy(FR_data_copy .ProjectionX())
FR_mc_px = copy.deepcopy(FR_mc_copy .ProjectionX())
PR_mc_px = copy.deepcopy(PR_mc_copy .ProjectionX())
FR_mcsub_px = copy.deepcopy(FR_mcsub_copy .ProjectionX())
# normal fake rates
FR_data_px .Divide(FR_data_px , copy.deepcopy(data_denominator .GetStack().Last().ProjectionX()), 1, 1, 'B')
FR_mc_px .Divide(FR_mc_px , copy.deepcopy(mc_denominator .GetStack().Last().ProjectionX()), 1, 1, 'B')
PR_mc_px .Divide(PR_mc_px , copy.deepcopy(mc_denominator_p .GetStack().Last().ProjectionX()), 1, 1, 'B')
FR_mcsub_px .Divide(FR_mcsub_px , copy.deepcopy(mcsub_denominator .GetStack().Last().ProjectionX()), 1, 1, 'B')
for j in range(len(mcsetsplot)):
FR_mcplot_px[j] = copy.deepcopy(FR_mcplot_copy[j].ProjectionX())
FR_mcplot_px[j].Divide(FR_mcplot_px[j], copy.deepcopy(mcplot_denominator[j].ProjectionX()), 1, 1, 'B')
# here comes the lepton eta projection
# ewk subtracted numerators
FR_data_py_mcsub = copy.deepcopy(data_numerator_mcsub .ProjectionY())
FR_data_py_mcsub_c1 = copy.deepcopy(data_numerator_mcsub_c1.ProjectionY())
FR_data_py_mcsub_l1 = copy.deepcopy(data_numerator_mcsub_l1.ProjectionY())
FR_data_py_mcsub_u1 = copy.deepcopy(data_numerator_mcsub_u1.ProjectionY())
FR_data_py_mcsub_c2 = copy.deepcopy(data_numerator_mcsub_c2.ProjectionY())
# ewk subtracted fake rates
FR_data_py_mcsub .Divide(FR_data_py_mcsub , data_denominator_mcsub .ProjectionY(), 1, 1, '')
FR_data_py_mcsub_c1.Divide(FR_data_py_mcsub_c1, data_denominator_mcsub_c1.ProjectionY(), 1, 1, '')
FR_data_py_mcsub_l1.Divide(FR_data_py_mcsub_l1, data_denominator_mcsub_l1.ProjectionY(), 1, 1, '')
FR_data_py_mcsub_u1.Divide(FR_data_py_mcsub_u1, data_denominator_mcsub_u1.ProjectionY(), 1, 1, '')
FR_data_py_mcsub_c2.Divide(FR_data_py_mcsub_c2, data_denominator_mcsub_c2.ProjectionY(), 1, 1, '')
# normal numerators
FR_data_py = copy.deepcopy(FR_data_copy .ProjectionY())
FR_mc_py = copy.deepcopy(FR_mc_copy .ProjectionY())
PR_mc_py = copy.deepcopy(PR_mc_copy .ProjectionY())
FR_mcsub_py = copy.deepcopy(FR_mcsub_copy .ProjectionY())
# normal fake rates
FR_data_py .Divide(FR_data_py , copy.deepcopy(data_denominator .GetStack().Last().ProjectionY()), 1, 1, 'B')
FR_mc_py .Divide(FR_mc_py , copy.deepcopy(mc_denominator .GetStack().Last().ProjectionY()), 1, 1, 'B')
PR_mc_py .Divide(PR_mc_py , copy.deepcopy(mc_denominator_p .GetStack().Last().ProjectionY()), 1, 1, 'B')
FR_mcsub_py .Divide(FR_mcsub_py , copy.deepcopy(mcsub_denominator .GetStack().Last().ProjectionY()), 1, 1, 'B')
for j in range(len(mcsetsplot)):
FR_mcplot_py[j] = copy.deepcopy(FR_mcplot_copy[j].ProjectionY())
FR_mcplot_py[j].Divide(FR_mcplot_py[j], copy.deepcopy(mcplot_denominator[j].ProjectionY()), 1, 1, 'B')
# plottin some nice plots, man! hell yeah!
# define the list of histograms which shall be plotted
# i.e. one has to take the histogram of the stack, not the stack itself
# projections on lepton pt
# data vs. total mc
histstoplot = []
histstoplot.append([FR_data_px, 'data'])
histstoplot.append([FR_mc_px, 'totbg'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[0]], 'FR_proj_Pt_data')
histstoplot = []
histstoplot.append([FR_data_px, 'data'])
histstoplot.append([PR_mc_px, 'totbg'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[0]], 'PR_proj_Pt_dyjets50')
# data vs. every mc sample
if len(mcsetsplot)>0:
for j in range(len(mcsetsplot)):
histstoplot = []
histstoplot.append([FR_data_px, 'data'])
histstoplot.append([FR_mcplot_px[j], mcsetsplot[j].GetName()])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[0]], 'FR_proj_Pt_data-' + mcsetsplot[j].GetName().lstrip(dataType + '_'))
# ewk subtracted unscaled vs. mcsubtractplot (i.e. qcd)
# THIS PART WANTS TO BE IMPROVED: we treat all mcsubplot samples as qcd!
if len(mcsubtract)>0:
histstoplot = []
histstoplot.append([FR_data_px_mcsub, 'datamcsub'])
if len(mcsubtractplot)>0:
histstoplot.append([FR_mcsub_px, 'mu_qcdmuenr'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[0]], 'FR_proj_Pt_data-ew')
# ewk subtraction methods vs. other methods
if mcsubtractscales:
# data vs. ETH
histstoplot = []
histstoplot.append([FR_data_px, 'data'])
histstoplot.append([FR_data_px_mcsub_c1, 'datamcsub_central1'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[0]], 'FR_proj_Pt_data-ew_data-eth', True, 'Data/ETH')
# ETH vs. UCSx
histstoplot = []
histstoplot.append([FR_data_px_mcsub_c1, 'datamcsub_central1'])
histstoplot.append([FR_data_px_mcsub_c2, 'datamcsub_central2'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[0]], 'FR_proj_Pt_data-ew_eth-ucsx', True, 'ETH/UCSx', 1.01, 0.99)
# ETH vs. mcsubtractplot (i.e. qcd)
# THIS PART WANTS TO BE IMPROVED: we treat all mcsubplot samples as qcd!
histstoplot = []
histstoplot.append([FR_data_px_mcsub_c1, 'datamcsub_central1'])
if len(mcsubtractplot)>0:
histstoplot.append([FR_mcsub_px, 'mu_qcdmuenr'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[0]], 'FR_proj_Pt_data-ew_data-eth-qcd', True, 'ETH/QCD')
# projections on lepton eta
# data vs. total mc
histstoplot = []
histstoplot.append([FR_data_py, 'data'])
histstoplot.append([FR_mc_py, 'totbg'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[1]], 'FR_proj_Eta_data')
histstoplot = []
histstoplot.append([FR_data_py, 'data'])
histstoplot.append([PR_mc_py, 'totbg'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[1]], 'PR_proj_Eta_dyjets50')
# data vs. every single mc
if len(mcsetsplot)>0:
for j in range(len(mcsetsplot)):
histstoplot = []
histstoplot.append([FR_data_py, 'data'])
histstoplot.append([FR_mcplot_py[j], mcsetsplot[j].GetName()])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[1]], 'FR_proj_Pt_data-' + mcsetsplot[j].GetName().lstrip(dataType + '_'))
# ewk subtraction unscaled vs. mcsubtractplot (i.e. qcd) vs. data
# THIS PART WANTS TO BE IMPROVED: we treat all mcsubplot samples as qcd!
if len(mcsubtract)>0:
histstoplot = []
histstoplot.append([FR_data_py, 'data'])
histstoplot.append([FR_data_py_mcsub, 'datamcsub'])
if len(mcsubtractplot)>0:
histstoplot.append([FR_mcsub_py, 'mu_qcdmuenr'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[1]], 'FR_proj_Eta_data-ew')
# ewk subtraction methods compared
if mcsubtractscales:
# data vs. ETH
histstoplot = []
histstoplot.append([FR_data_py, 'data'])
histstoplot.append([FR_data_py_mcsub_c1, 'datamcsub_central1'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[1]], 'FR_proj_Eta_data-ew_data-eth', True, 'Data/ETH')
# ETH vs. UCSx
histstoplot = []
histstoplot.append([FR_data_py_mcsub_c1, 'datamcsub_central1'])
histstoplot.append([FR_data_py_mcsub_c2, 'datamcsub_central2'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[1]], 'FR_proj_Eta_data-ew_eth-ucsx', True, 'ETH/UCSx', 1.01, 0.99)
# ETH vs. mcsubtractplot (i.e. qcd) vs. data
# THIS PART WANTS TO BE IMPROVED: we treat all mcsubplot samples as qcd!
histstoplot = []
histstoplot.append([FR_data_py_mcsub_c1, 'datamcsub_central1'])
if len(mcsubtractplot)>0:
histstoplot.append([FR_mcsub_py, 'mu_qcdmuenr'])
histstoplot.append([FR_data_py, 'data'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, datasets[0].hists[title_indeces[1]], 'FR_proj_Eta_data-ew_data-eth-qcd', True, 'ETH/QCD')
return True
def PlotFR(dataType, outputDir, datasets, mcsets, histlist, mcsetsplot = [], mcsubtract = [], mcsubtractplot = [], mcsubtractscales = False, bgestimation = False):
#
# produce the fake rate plots (incl. ewk subtraction) for given histograms and given samples
# NOTE: ALL SAMPLES MUST CONTAIN THE EXACT SAME LIST OF HISTOGRAMS, OTHERWISE THIS FUNCTION WILL NOT WORK
#
# this function takes the following paramters:
# dataType..........type of the lepton ('mu', 'el')
# outputDir.........basic output directory
# datasets..........list of data samples [datasample1, datasample2, ..]
# mcsets............list of mc samples [mcsample1, mcsample2, ..] to enter bg stack
# histlist..........list of 1d histogram names ['h_Loose_LepIso', 'h_Loose_LepPhi', ..]
# mcsetsplot........list of mc samples [mcsample1, mcsample2, ..] to be drawn seperately
# mcsubtract........list of mc samples [mcsample1, mcsample2, ..] to enter ewk subtraction
# mcsubtractplot....list of mc samples [mcsample1, mcsample2, ..] to be drawn in comparison with ewk subtraction
# mcsubtractscales..True if we first want to scale ewk according to ETH/UCSx methods
# produce canvas and pads
canv = helper.makeCanvas(900, 675, 'c1dFR')
pad_plot = helper.makePad('plot')
pad_ratio = helper.makePad('ratio')
pad_plot.SetTicks(1,1)
pad_ratio.SetTicks(1,1)
# default for ewk scales (central1 is ewk scale according to ETH method with lower and upper limit, central2 is scale according to UCSx)
central1 = 1.0
lower = 1.0
upper = 1.0
central2 = 1.0
# compute the scales according to ETH and UCSx method
# this part needs improvement
if mcsubtractscales:
scfirst = fit.getMCScaleFactorSimultaneouslyWithErrors(datasets, mcsubtractplot, mcsubtract, 50, 120, 'h_Tight_MTMET30')
qcd2 = scfirst[0][0]
central12= scfirst[0][1]
lower2 = scfirst[1][1]
upper2 = scfirst[2][1]
scfirst = fit.getMCScaleFactorSimultaneouslyWithErrors(datasets, mcsubtractplot, mcsubtract)
central1 = scfirst[0][1]
lower = scfirst[1][1]
upper = scfirst[2][1]
scsecond = fit.getMCScaleFactorMutually(mcsubtract, 'h_Tight_MTMET30', datasets, mcsubtractplot, 60, 100)
central2 = scsecond[0]
scsecond = fit.getMCScaleFactorMutually(mcsubtract, 'h_Tight_MTMET20', datasets, mcsubtractplot, 60, 100)
central22= scsecond[0]
print "------**------"
print "MET30:"
print "qcd = " + str(qcd2)
print "central 1 = " + str(central12)
print "lower 1 = " + str(lower2)
print "upper 1 = " + str(upper2)
print "central 2 = " + str(central2)
print "------**------"
print "MET20:"
print "qcd = " + str(scfirst[0][0])
print "central 1 = " + str(central1)
print "lower 1 = " + str(lower)
print "upper 1 = " + str(upper)
print "central 2 = " + str(central22)
print "------++------"
# we add data and mc in stacks for both numerator and denominator
m = -1
n = -1
index_numerators = []
index_denominators = []
data_numerators = []
data_denominators = []
mc_numerators = []
mc_denominators = []
FRs_mcplot = []
mcplot_denominators = []
mcsub_numerators = []
mcsub_denominators = []
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for hist in datasets[0].hists:
# get index of the histogram
i = datasets[0].hists.index(hist)
# only histograms in the parameter histlist are plotted
if not hist.GetName() in histlist: continue
# there are two NVertices plots in the list with different binning
# we only plot the NVertices, but not the NVertices1 (this we use for the fake rate plots, see lib_FakeRate.py)
if hist.GetName()[-9:] == "NVertices": continue # we take "NVertices1" histograms for Fakerate Plots
# get numerator stacks
if 'h_Tight_' in hist.GetName():
index_numerators .append(i)
data_numerators .append(ROOT.THStack())
mc_numerators .append(ROOT.THStack())
FRs_mcplot .append([])
mcsub_numerators .append(ROOT.THStack())
m += 1
for data in datasets: data_numerators[m] .Add(copy.deepcopy(data.hists[index_numerators[m]]))
for mc in mcsets: mc_numerators[m] .Add(copy.deepcopy(mc.hists [index_numerators[m]]))
for j, mc in enumerate(mcsetsplot): FRs_mcplot[m][j] = copy.deepcopy(mc.hists [index_numerators[m]])
for mc in mcsubtractplot: mcsub_numerators[m] .Add(copy.deepcopy(mc.hists [index_numerators[m]]))
# get denominator stacks
if 'h_Loose_' in hist.GetName():
index_denominators .append(i)
data_denominators .append(ROOT.THStack())
mc_denominators .append(ROOT.THStack())
mcplot_denominators.append([])
mcsub_denominators .append(ROOT.THStack())
n += 1
for data in datasets: data_denominators[n] .Add(copy.deepcopy(data.hists[index_denominators[n]]))
for mc in mcsets: mc_denominators[n] .Add(copy.deepcopy(mc.hists [index_denominators[n]]))
for j, mc in enumerate(mcsetsplot): mcplot_denominators[n][j] = copy.deepcopy(mc.hists [index_denominators[n]])
for mc in mcsubtractplot: mcsub_denominators[n] .Add(copy.deepcopy(mc.hists [index_denominators[n]]))
# get numerators from stack for data, mc and mcsubtract
FRs_data = [copy.deepcopy(data_numerator .GetStack().Last()) for data_numerator in data_numerators ]
FRs_mc = [copy.deepcopy(mc_numerator .GetStack().Last()) for mc_numerator in mc_numerators ]
FRs_mcsub = [copy.deepcopy(mcsub_numerator .GetStack().Last()) for mcsub_numerator in mcsub_numerators ]
# perform ewk subtraction, once unscaled, once ETH method (central1, lower, upper) and once UCSx method
for i in range(len(FRs_data)):
FR_data_mcsub = copy.deepcopy(FRs_data[i])
FR_data_mcsub_c1 = copy.deepcopy(FRs_data[i])
FR_data_mcsub_l1 = copy.deepcopy(FRs_data[i])
FR_data_mcsub_u1 = copy.deepcopy(FRs_data[i])
FR_data_mcsub_c2 = copy.deepcopy(FRs_data[i])
data_denominator_mcsub = copy.deepcopy(data_denominators[i].GetStack().Last())
data_denominator_mcsub_c1 = copy.deepcopy(data_denominators[i].GetStack().Last())
data_denominator_mcsub_l1 = copy.deepcopy(data_denominators[i].GetStack().Last())
data_denominator_mcsub_u1 = copy.deepcopy(data_denominators[i].GetStack().Last())
data_denominator_mcsub_c2 = copy.deepcopy(data_denominators[i].GetStack().Last())
# we subtract every mc set in mcsubtract from the data 5 times (unscaled, ETH, lower, upper, UCSx)
# both for numerator and denominator (WATCH OUT FOR THE SCALING!)
if len(mcsubtract)>0:
for mc in mcsubtract:
# numerator unscaled
FR_data_mcsub.Add(mc.hists[index_numerators[i]], -1)
# numerator ETH central
mc.hists[index_numerators[i]].Scale(central1)
FR_data_mcsub_c1.Add(mc.hists[index_numerators[i]], -1)
# numerator ETH lower bound
mc.hists[index_numerators[i]].Scale(lower/central1)
FR_data_mcsub_l1.Add(mc.hists[index_numerators[i]], -1)
# numerator ETH upper bound
mc.hists[index_numerators[i]].Scale(upper/lower)
FR_data_mcsub_u1.Add(mc.hists[index_numerators[i]], -1)
# numerator UCSx central
mc.hists[index_numerators[i]].Scale(central2/upper)
FR_data_mcsub_c2.Add(mc.hists[index_numerators[i]], -1)
# rescale it to 1 again
mc.hists[index_numerators[i]].Scale(1/central2)
# denominator unscaled
data_denominator_mcsub.Add(mc.hists[index_denominators[i]],-1)
# denominator ETH central
mc.hists[index_denominators[i]].Scale(central1)
data_denominator_mcsub_c1.Add(mc.hists[index_denominators[i]],-1)
# denominator ETH lower bound
mc.hists[index_denominators[i]].Scale(lower/central1)
data_denominator_mcsub_l1.Add(mc.hists[index_denominators[i]],-1)
# denominator ETH upper bound
mc.hists[index_denominators[i]].Scale(upper/lower)
data_denominator_mcsub_u1.Add(mc.hists[index_denominators[i]],-1)
# denomiantor UCSx central
mc.hists[index_denominators[i]].Scale(central2/upper)
data_denominator_mcsub_c2.Add(mc.hists[index_denominators[i]],-1)
# rescale it to 1 again
mc.hists[index_denominators[i]].Scale(1/central2)
# ewk subtracted fake rates (non-correlated error propagation)
FR_data_mcsub .Divide(FR_data_mcsub , data_denominator_mcsub , 1, 1, '')
FR_data_mcsub_c1.Divide(FR_data_mcsub_c1, data_denominator_mcsub_c1, 1, 1, '')
FR_data_mcsub_l1.Divide(FR_data_mcsub_l1, data_denominator_mcsub_l1, 1, 1, '')
FR_data_mcsub_u1.Divide(FR_data_mcsub_u1, data_denominator_mcsub_u1, 1, 1, '')
FR_data_mcsub_c2.Divide(FR_data_mcsub_c2, data_denominator_mcsub_c2, 1, 1, '')
# fake rates for pure data, total mc, single mc (correlated error propagation)
FRs_data[i] .Divide(FRs_data[i] , data_denominators[i] .GetStack().Last(), 1, 1, 'B')
FRs_mc[i] .Divide(FRs_mc[i] , mc_denominators[i] .GetStack().Last(), 1, 1, 'B')
FRs_mcsub[i] .Divide(FRs_mcsub[i] , mcsub_denominators[i] .GetStack().Last(), 1, 1, 'B')
if len(mcsetsplot)>0:
for j in range(len(mcsetsplot)):
FRs_mcplot[i][j].Divide(FRs_mcplot[i][j], mcplot_denominators[i][j], 1, 1, 'B')
# define the list of histograms which shall be plotted
# i.e. one has to take the histogram of the stack, not the stack itself
# data vs. total mc
histstoplot = []
histstoplot.append([FRs_data[i], 'data'])
histstoplot.append([FRs_mc[i], 'totbg'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, FRs_data[i], 'FR_' + FRs_data[i].GetName().lstrip('h_Tight_') + '_data')
# fake rate for every single mc
if len(mcsetsplot)>0:
for j in range(len(mcsetsplot)):
histstoplot = []
histstoplot.append([FRs_data[i], 'data'])
histstoplot.append([FRs_mcplot[i][j], mcsetsplot[j].GetName()])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, FRs_data[i], 'FR_' + FRs_data[i].GetName().lstrip('h_Tight_') + '_data-' + mcsetsplot[j].GetName().lstrip(dataType + '_'))
# ewk subtracted vs. mcsubtractplot
# THIS PART WANTS TO BE IMPROVED: we treat all mcsubplot samples as qcd!
if len(mcsubtract)>0:
histstoplot = []
histstoplot.append([FR_data_mcsub, 'datamcsub'])
if len(mcsubtractplot)>0:
histstoplot.append([FRs_mcsub[i], 'mu_qcdmuenr'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, FRs_data[i], 'FR_' + FRs_data[i].GetName().lstrip('h_Tight_') + '_data-ew')
# ewk subtracted one method vs. the other
if len(mcsubtract)>0 and mcsubtractscales:
# data vs. ETH
histstoplot = []
histstoplot.append([FRs_data[i], 'data'])
histstoplot.append([FR_data_mcsub_c1, 'datamcsub_central1'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, FRs_data[i], 'FR_' + FRs_data[i].GetName().lstrip('h_Tight_') + '_data-ew_data-eth', True, 'Data/ETH')
# ETH vs. UCSx
histstoplot = []
histstoplot.append([FR_data_mcsub_c1, 'datamcsub_central1'])
histstoplot.append([FR_data_mcsub_c2, 'datamcsub_central2'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, FRs_data[i], 'FR_' + FRs_data[i].GetName().lstrip('h_Tight_') + '_data-ew_eth-ucsx', True, 'ETH/UCSx', 1.01, 0.99)
# ETH vs. mcsubtractplot vs. data
# THIS PART WANTS TO BE IMPROVED: we treat all mcsubplot samples as qcd!
histstoplot = []
histstoplot.append([FR_data_mcsub_c1, 'datamcsub_central1'])
if len(mcsubtractplot)>0:
histstoplot.append([FRs_mcsub[i], 'mu_qcdmuenr'])
histstoplot.append([FRs_data[i], 'data'])
make1dFRPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, FRs_data[i], 'FR_' + FRs_data[i].GetName().lstrip('h_Tight_') + '_data-ew_data-eth-qcd', True, 'ETH/QCD')
return True
def PlotProvenance(dataType, outputDir, mcsets, binstoplot = []):
#
# produce and plot the provenance plot from a list of bins
# NOTE: ALL SAMPLES MUST CONTAIN THE EXACT SAME LIST OF HISTOGRAMS, OTHERWISE THIS FUNCTION WILL NOT WORK
#
# this function takes the following parameters:
# dataType....type of the lepton ('mu', 'el')
# outputDir...basic output directory
# mcsets......list of mc samples [mcsample1, mcsample2, ..]
# binstoplot..list of bin numbers (matching origins from Fakerates.cc) we want to plot
# define canvas and pads
canv = helper.makeCanvas(900, 675, 'c1dZ')
pad_plot = helper.makePad('tot')
pad_plot.SetTicks(1,1)
pad_plot.cd()
# define some numbers and lists
nbins = len(binstoplot)
index_loose = 0
index_tight = 0
index_lnt = 0
labels = ['all', 'W', 'B', 'C', 'U/D/S', 'unm.']
hist_plot = []
mcgroups_loose = []
mcgroups_loose_lnt = []
mcgroups_loose_aes = []
mcgroups_tight = []
mcgroups_tight_aes = []
mcnames = []
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for hist in mcsets[0].hists:
# get index of the histogram
i = mcsets[0].hists.index(hist)
# only consider plots with 'Provenance' in their names
if not 'Provenance' in hist.GetName(): continue
# h_Loose_Provenance is the denominator after lepton selection only
# h_Loose_ProvenanceLNT is the numerator-subtracted denominator after lepton selection only
# h_Loose_ProvenanceAES is the denominator after event and lepton selection
# h_Tight_Provenance is the numerator after lepton selection only
# h_Tight_ProvenanceAES is the numerator after event and lepton selection
if hist.GetName() == 'h_Loose_Provenance' : index_loose = i
if hist.GetName() == 'h_Loose_ProvenanceLNT': index_loose_lnt = i
if hist.GetName() == 'h_Loose_ProvenanceAES': index_loose_aes = i
if hist.GetName() == 'h_Tight_Provenance' : index_tight = i
if hist.GetName() == 'h_Tight_ProvenanceAES': index_tight_aes = i
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for mcset in mcsets:
# strip off the digits from the sample name
label = ''.join([j for j in mcset.GetName() if not j.isdigit()])
foundat = -1
for j, mcname in enumerate(mcnames):
if label == mcname: foundat = j
# we group all provenance plots from samples which have the same label (e.g. dyjets10 and dyjets50)
if foundat == -1:
mcgroups_loose .append(ROOT.THStack())
mcgroups_loose_lnt.append(ROOT.THStack())
mcgroups_loose_aes.append(ROOT.THStack())
mcgroups_tight .append(ROOT.THStack())
mcgroups_tight_aes.append(ROOT.THStack())
mcgroups_loose [len(mcgroups_loose) -1].Add(mcset.hists[index_loose] )
mcgroups_loose_lnt[len(mcgroups_loose_lnt)-1].Add(mcset.hists[index_loose_lnt])
mcgroups_loose_aes[len(mcgroups_loose_aes)-1].Add(mcset.hists[index_loose_aes])
mcgroups_tight [len(mcgroups_tight) -1].Add(mcset.hists[index_tight] )
mcgroups_tight_aes[len(mcgroups_tight_aes)-1].Add(mcset.hists[index_tight_aes])
mcnames.append(label)
else:
mcgroups_loose [foundat].Add(mcset.hists[index_loose] )
mcgroups_loose_lnt[foundat].Add(mcset.hists[index_loose_lnt])
mcgroups_loose_aes[foundat].Add(mcset.hists[index_loose_aes])
mcgroups_tight [foundat].Add(mcset.hists[index_tight] )
mcgroups_tight_aes[foundat].Add(mcset.hists[index_tight_aes])
# fake rate provenance are created with and without event selection
hist_den = [{} for j in range(len(mcnames))]
hist_den_aes = [{} for j in range(len(mcnames))]
hist_num = [{} for j in range(len(mcnames))]
hist_num_aes = [{} for j in range(len(mcnames))]
# create the histograms for each provenance plot (loose, lnt, loose_aes, tight, tight_aes)
for i, plotgroup in enumerate([mcgroups_loose, mcgroups_loose_lnt, mcgroups_loose_aes, mcgroups_tight, mcgroups_tight_aes]):
hist_plot.append([])
for j, group in enumerate(plotgroup):
group.Draw('hist')
histogram = group.GetStack().Last()
if nbins == 0: nbins = histogram.GetNbinsX()
if histogram.GetName() == 'h_Loose_Provenance' : postpend = 'loose'
if histogram.GetName() == 'h_Loose_ProvenanceLNT': postpend = 'loose_lnt'
if histogram.GetName() == 'h_Loose_ProvenanceAES': postpend = 'loose_aes'
if histogram.GetName() == 'h_Tight_Provenance' : postpend = 'tight'
if histogram.GetName() == 'h_Tight_ProvenanceAES': postpend = 'tight_aes'
# we create as many histograms as we want to plot origins and we store them all in hist_plot
hist_plot[i].append(ROOT.TH1F(histogram.GetName() + '_plot' + str(j), histogram.GetName(), nbins, 0, nbins))
# each of these histograms we fill in such a way that only one bin is non-zero
n = 1
for k in range(1, histogram.GetNbinsX()+1):
if binstoplot == [] or k in binstoplot:
hist_plot[i][j].SetBinContent(n, histogram.GetBinContent(k))
n +=1
# copy numerators and denominators for subsequent fake rate production
if hist_plot[i][j].GetName() == 'h_Loose_Provenance_plot' + str(j) : hist_den[j] = copy.deepcopy(hist_plot[i][j])
if hist_plot[i][j].GetName() == 'h_Loose_ProvenanceAES_plot' + str(j): hist_den_aes[j] = copy.deepcopy(hist_plot[i][j])
if hist_plot[i][j].GetName() == 'h_Tight_Provenance_plot' + str(j) : hist_num[j] = copy.deepcopy(hist_plot[i][j])
if hist_plot[i][j].GetName() == 'h_Tight_ProvenanceAES_plot' + str(j): hist_num_aes[j] = copy.deepcopy(hist_plot[i][j])
# produce nice lits of histograms and plot them
hist_plots = makeProvenancePlots(hist_plot[i][j], labels, binstoplot)
hist_plots[0].Draw('hist text')
for k in range(1,len(hist_plots)): hist_plots[k].Draw('hist text same')
helper.saveCanvas(canv, pad_plot, outputDir + "closure/", mcnames[j] + '_' + postpend)
# we create the fake rate provenances with and without event selection
for j in range(len(mcnames)):
hist_num[j].Divide(hist_num[j], hist_den[j], 1, 1, 'B')
hist_nums = makeProvenancePlots(hist_num[j], labels, binstoplot, False)
hist_nums[0].Draw('hist text')
for k in range(1, len(hist_nums)): hist_nums[k].Draw('hist text same')
helper.saveCanvas(canv, pad_plot, outputDir + "closure/", mcnames[j] + '_rate')
hist_num_aes[j].Divide(hist_num_aes[j], hist_den_aes[j], 1, 1, 'B')
hist_nums = makeProvenancePlots(hist_num_aes[j], labels, binstoplot, False)
hist_nums[0].Draw('hist text')
for k in range(1, len(hist_nums)): hist_nums[k].Draw('hist text same')
helper.saveCanvas(canv, pad_plot, outputDir + "closure/", mcnames[j] + '_rate_aes')
def PlotCompare(dataType, outputDir, mcsets, histname, leg, cutoff = -1, precise = False):
#
# plot a set of histograms defined by histname and comparing mc samples
# NOTE: ALL SAMPLES MUST CONTAIN THE EXACT SAME LIST OF HISTOGRAMS, OTHERWISE THIS FUNCTION WILL NOT WORK
#
# this function takes the following parameters:
# dataType....type of the lepton ('mu', 'el')
# outputDir...basic output directory
# mcsets......list of mc samples [mcsample1, mcsample2, ..]
# leg.........legend to be plotted
# cutoff......cutoff for the histogram name (see function getSaveName() in lib.py)
# precise.....True if only the histogram with the name matching histname exactly shall be plotted
# define canvas and pads
canv = helper.makeCanvas(900, 675)
pad_plot = helper.makePad('plot')
pad_ratio = helper.makePad('ratio')
pad_ratio.cd()
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for hist in mcsets[0].hists:
# get index of histogram
i = mcsets[0].hists.index(hist)
pad_plot.cd()
# if precise is True, we only plot the histogram with the name matching histname exactly
# else, we plot all histograms which have histname in their name (useful for several versions or bins of a histogram)
if precise and not histname == hist.GetName(): continue
if not precise and not histname in hist.GetName(): continue
# pre- and postpends
prepend = ''
postpend = '_compare'
if '_Loose_' in hist.GetName(): prepend = 'Loose_'
if '_Tight_' in hist.GetName(): prepend = 'Tight_'
# draw first histogram
hist.Draw('hist')
hist.SetFillStyle(0)
hist.Scale(1.0/hist.Integral())
max = hist.GetMaximum()
for j in range(1,len(mcsets)):
mcsets[j].hists[i].Draw('hist same')
mcsets[j].hists[i].SetFillStyle(0)
mcsets[j].hists[i].SetLineStyle(2)
mcsets[j].hists[i].Scale(1.0/mcsets[j].hists[i].Integral())
if max < mcsets[j].hists[i].GetMaximum(): max = mcsets[j].hists[i].GetMaximum()
# do some cosmetics
hist.SetMinimum(0.0001)
hist.SetMaximum(1.5 * max)
hist = helper.set1dPlotStyle(dataType, hist, helper.getColor(mcsets[0].GetName()), '', hist, '1/Integral')
# draw legend
leg.Draw()
# draw ratio plot
pad_ratio.cd()
hist_ratio = copy.deepcopy(hist)
hist_ratio.Divide(copy.deepcopy(mcsets[1].hists[i]))
hist_ratio.Draw("p e1")
hist_ratio = helper.setRatioStyle(dataType, hist_ratio, hist)
line = helper.makeLine(hist_ratio.GetXaxis().GetXmin(), 1.00, hist_ratio.GetXaxis().GetXmax(), 1.00)
line.Draw()
# make plot
helper.saveCanvas(canv, pad_plot, outputDir + "compare/", prepend + helper.getSaveName(hist, cutoff) + postpend)
def PlotJPtZooms(dataType, outputDir, dataset, mcsets, leg):
#
# plotting JetPt and JetRawPt distributions in bins of lepton eta and lepton pt (binning according to fake rate maps)
# NOTE: ALL SAMPLES MUST CONTAIN THE EXACT SAME LIST OF HISTOGRAMS, OTHERWISE THIS FUNCTION WILL NOT WORK
#
# this function takes the following parameters:
# dataType....type of the lepton ('mu', 'el')
# outputDir...basic output directory
# datasets....list of data samples [datasample1, datasample2, ..]
# mcsets......list of mc samples [mcsample1, mcsample2, ..]
# leg.........legend to be plotted
# define canvas and pads
canv = helper.makeCanvas(900, 675, 'c1dZ')
pad_plot = helper.makePad('tot')
pad_plot.SetTicks(1,1)
pad_plot.cd()
# text for eta binning
t_eta = ROOT.TLatex()
t_eta.SetNDC()
t_eta.SetTextSize(0.05)
t_eta.SetTextAlign(11)
t_eta.SetTextColor(ROOT.kBlack)
# text for pt binning
t_pt = ROOT.TLatex()
t_pt.SetNDC()
t_pt.SetTextSize(0.05)
t_pt.SetTextAlign(11)
t_pt.SetTextColor(ROOT.kBlack)
# bins in eta and pt, with total number of bins
bins_eta = [0.0, 1.0, 2.4]
#bins_pt = [10.0, 20.0, 30.0, 35.0, 37.5, 40.0, 42.5, 45.0, 47.5, 50.0, 55.0, 60.0, 70.0] # old
bins_pt = [10.0, 20.0, 22.5, 25.0, 27.5, 30.0, 32.5, 35.0, 40.0, 50.0, 60.0, 70.0] # new
bins_tot = (len(bins_eta)-1)*(len(bins_pt)-1)
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for hist in dataset.hists:
# get index of the histogram
i = dataset.hists.index(hist)
# only histograms in the parameter histlist are plotted
if not "JPtZoom" in hist.GetName(): continue
# pre- and postpends
prepend = ''
postpend = ''
if '_Loose_' in hist.GetName(): prepend = 'Loose_'
if '_Tight_' in hist.GetName(): prepend = 'Tight_'
# get the id of the plot, i.e. the last number in the name (e.g. '0' in h_Loose_DJPtZoomC_0)
id = hist.GetName().split('_')[-1]
# draw data
hist.Scale(1.0/hist.Integral())
hist.SetFillStyle(0)
hist.SetLineStyle(2)
hist.Draw("HIST")
max = hist.GetMaximum()
# draw mc samples
for mc in mcsets:
mc.hists[i].Scale(1.0/mc.hists[i].Integral())
mc.hists[i].SetFillStyle(0)
mc.hists[i].Draw("HIST SAME")
if mc.hists[i].GetMaximum()>max: max = mc.hists[i].GetMaximum()
# do some cosmetics
hist.SetMaximum(1.5*max)
hist.GetXaxis().SetTitle(helper.getXTitle(dataType, hist))
hist.GetYaxis().SetTitle("1/Integral")
hist.SetTitle("")
hist.GetXaxis().SetTitleSize(0.07)
hist.GetXaxis().SetLabelSize(0.07)
hist.GetYaxis().SetTitleSize(0.07)
hist.GetYaxis().SetLabelSize(0.07)
hist.GetXaxis().SetNdivisions(505)
hist.GetYaxis().SetNdivisions(505)
# draw legend
leg.Draw()
# write bin texts
m = int(id)//(len(bins_pt)-1)
n = int(id)%(len(bins_pt)-1)
if "ZoomC" in hist.GetName(): write = "corr."
else: write = "raw"
text_eta = str(bins_eta[m]) + " #leq jet-|#eta| < " + str(bins_eta[m+1])
text_pt = str(bins_pt[n]) + " #leq jet-p_{T} (" + write + ") < " + str(bins_pt[n+1])
t_eta.DrawLatex(0.22, 0.8, text_eta)
t_pt.DrawLatex(0.22, 0.73, text_pt)
# draw plots
helper.saveCanvas(canv, pad_plot, outputDir + "zoom_jpt/", prepend + helper.getSaveName(hist, '-2:') + postpend, False)
def PlotMETZooms(dataType, outputDir, datasets, mcsets, leg, grouping = False):
#
# plotting MET distribution in bins of lepton eta and lepton pt (binning according to fake rate maps)
# NOTE: ALL SAMPLES MUST CONTAIN THE EXACT SAME LIST OF HISTOGRAMS, OTHERWISE THIS FUNCTION WILL NOT WORK
#
# dataType....type of the lepton ('mu', 'el')
# outputDir...basic output directory
# datasets....list of data samples [datasample1, datasample2, ..]
# mcsets......list of mc samples [mcsample1, mcsample2, ..]
# leg.........legend to be plotted
# grouping....True if mc samples should be grouped before stacking (useful in case of e.g. several QCD files)
# define canvas and pads
canv = helper.makeCanvas(900, 675, 'c1dM')
pad_plot = helper.makePad('plot')
pad_ratio = helper.makePad('ratio')
pad_plot.SetTicks(1,1)
pad_ratio.SetTicks(1,1)
# text for eta binning
t_eta = ROOT.TLatex()
t_eta.SetNDC()
t_eta.SetTextSize(0.05)
t_eta.SetTextAlign(11)
t_eta.SetTextColor(ROOT.kBlack)
# text for pt binning
t_pt = ROOT.TLatex()
t_pt.SetNDC()
t_pt.SetTextSize(0.05)
t_pt.SetTextAlign(11)
t_pt.SetTextColor(ROOT.kBlack)
# bins in eta and pt, with total number of bins
bins_eta = [0.0, 0.5, 1.0, 1.5, 2.0, 2.5]
bins_pt = [20.0, 25.0, 30.0, 35.0, 40.0, 45.0, 50.0]
bins_tot = (len(bins_eta)-1)*(len(bins_pt)-1)
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for hist in datasets[0].hists:
# get index of the histogram
i = datasets[0].hists.index(hist)
pad_plot.cd()
# only histograms in the parameter histlist are plotted
if not "METZoom" in hist.GetName(): continue
# pre- and postpends
prepend = ''
postpend = ''
if '_Loose_' in hist.GetName(): prepend = 'Loose_'
if '_Tight_' in hist.GetName(): prepend = 'Tight_'
# get the id of the plot, i.e. the last number in the name (e.g. '0' in h_Loose_FRMETZoom_0)
id = hist.GetName().split('_')[-1]
# we stack both data and mc samples (i.e. one may use several data and several mc samples)
data = ROOT.THStack()
mc = ROOT.THStack()
for dataset in datasets:
data.Add(dataset.hists[i])
# if grouping is enabled, we first sum all 'similar' mc samples before adding them to the stack
# similar means, that we sum all samples which have the same name up to a few digits at the end
# in this way, e.g., the samples dyjets50 and dyjets10 are stacked together
# if grouping is not enabled, we just stack all mc samples in one stack
if grouping:
mcgroups = []
mcnames = []
for mcset in mcsets:
label = ''.join([j for j in mcset.GetName() if not j.isdigit()])
foundat = -1
for j, mcname in enumerate(mcnames):
if label == mcname: foundat = j
if foundat == -1:
mcgroups.append(ROOT.THStack())
mcgroups[len(mcgroups)-1].Add(mcset.hists[i])
mcnames.append(label)
else:
mcgroups[foundat].Add(mcset.hists[i])
for j, group in enumerate(mcgroups):
group.Draw('hist')
histogram = group.GetStack().Last()
mc.Add(histogram)
else:
for mcset in mcsets:
mc.Add(mcset.hists[i])
# draw histogram
mc.Draw('hist')
# define the list of histograms which shall be plotted
# i.e. one has to take the histogram of the stack, not the stack itself
hists = []
hists.append([data.GetStack().Last(), 'data' ])
hists.append([mc , 'totbg'])
# plot pad_plot first, then we add the texts and plot ratio afterwards
make1dPlot_plot(dataType, pad_plot, hists, hists[0][0], leg)
# write bin texts
m = int(id)//(len(bins_pt)-1)
n = int(id)%(len(bins_pt)-1)
if dataType == 'el': lepton = 'e'
else : lepton = '#mu'
text_eta = str(bins_eta[m]) + " #leq " + lepton + "-|#eta| < " + str(bins_eta[m+1])
text_pt = str(bins_pt[n]) + " #leq " + lepton + "-p_{T} < " + str(bins_pt[n+1])
t_eta.DrawLatex(0.22, 0.8, text_eta)
t_pt.DrawLatex(0.22, 0.73, text_pt)
# plot ratio
#make1dPlot_ratio(dataType, pad_ratio, hists, hists[0][0])
# calling the function to produce the ratio plot results in a malfunction i do not understand properly
# so we copy the lines from make1dPlot_ratio() while we still keep that function for other purposes
pad_ratio.cd()
data_bg_ratio = copy.deepcopy(hists[0][0])
data_bg_ratio.Divide(copy.deepcopy(hists[1][0].GetStack().Last()))
data_bg_ratio.Draw("p e")
data_bg_ratio = helper.setRatioStyle(dataType, data_bg_ratio, hists[0][0])
line = helper.makeLine(data_bg_ratio.GetXaxis().GetXmin(), 1.00, data_bg_ratio.GetXaxis().GetXmax(), 1.00)
line.Draw()
# save plot
ROOT.gPad.RedrawAxis()
helper.saveCanvas(canv, pad_plot, outputDir + "zoom_met/", prepend + helper.getSaveName(hist, '-2:') + postpend)
def Plot1d(dataType, outputDir, datasets, mcsets, histlist, leg, grouping = False):
#
# given data and mc samples, all histograms in histlist are produced with mc samples stacked
# NOTE: ALL SAMPLES MUST CONTAIN THE EXACT SAME LIST OF HISTOGRAMS, OTHERWISE THIS FUNCTION WILL NOT WORK
#
# this function takes the following parameters:
# dataType....type of the lepton ('mu', 'el')
# outputDir...basic output directory
# datasets....list of data samples [datasample1, datasample2, ..]
# mcsets......list of mc samples [mcsample1, mcsample2, ..]
# histlist....list of 1d histogram names ['h_Loose_LepIso', 'h_Loose_LepPhi', ..]
# leg.........legend to be drawn
# grouping....True if mc samples should be grouped before stacking (useful in case of e.g. several QCD files)
# define canvas and pads
canv = helper.makeCanvas(900, 675, 'c1d')
pad_plot = helper.makePad('plot')
pad_ratio = helper.makePad('ratio')
pad_plot.SetTicks(1,1)
pad_ratio.SetTicks(1,1)
# iterate over all histograms in root files
# it does not matter which sample we iterate on, as all samples contain the same list of histograms
for hist in datasets[0].hists:
# get index of the histogram
i = datasets[0].hists.index(hist)
pad_plot.cd()
# only histograms in the parameter histlist are plotted
if not hist.GetName() in histlist: continue
# there are two NVertices plots in the list with different binning
# we only plot the NVertices, but not the NVertices1 (this we use for the fake rate plots, see lib_FakeRate.py)
if hist.GetName()[-10:] == "NVertices1": continue
# pre- and postpends
prepend = ''
postpend = ''
if '_Loose_' in hist.GetName(): prepend = 'Loose_'
if '_Tight_' in hist.GetName(): prepend = 'Tight_'
# we stack both data and mc samples (i.e. one may use several data and several mc samples)
data = ROOT.THStack()
mc = ROOT.THStack()
for dataset in datasets:
data.Add(dataset.hists[i])
# if grouping is enabled, we first sum all 'similar' mc samples before adding them to the stack
# similar means, that we sum all samples which have the same name up to a few digits at the end
# in this way, e.g., the samples dyjets50 and dyjets10 are stacked together
# if grouping is not enabled, we just stack all mc samples in one stack
if grouping:
mcgroups = []
mcnames = []
for mcset in mcsets:
label = ''.join([j for j in mcset.GetName() if not j.isdigit()])
foundat = -1
for j, mcname in enumerate(mcnames):
if label == mcname: foundat = j
if foundat == -1:
mcgroups.append(ROOT.THStack())
mcgroups[len(mcgroups)-1].Add(mcset.hists[i])
mcnames.append(label)
else:
mcgroups[foundat].Add(mcset.hists[i])
for j, group in enumerate(mcgroups):
group.Draw('hist')
histogram = group.GetStack().Last()
mc.Add(histogram)
else:
for mc in mcsets:
mc.Add(mc.hists[i])
# define the list of histograms which shall be plotted
# i.e. one has to take the histogram of the stack, not the stack itself
histstoplot = []
histstoplot.append([data.GetStack().Last(), 'data' ])
histstoplot.append([mc , 'totbg'])
# call make1dPlot to create the plot
make1dPlot(dataType, canv, pad_plot, pad_ratio, outputDir, histstoplot, hist, prepend + helper.getSaveName(hist) + postpend, leg)