def Direct_Estimator(var, cut, year):
from root_numpy import root2array, fill_hist, array2root
import numpy.lib.recfunctions as rfn
### Preliminary Operations ###
treeRead = not cut in [
"nnqq", "en", "enqq", "mn", "mnqq", "ee", "eeqq", "mm", "mmqq", "em",
"emqq", "qqqq"
] # Read from tree
channel = cut
unit = ''
if "GeV" in variable[var]['title']: unit = ' GeV'
isBlind = BLIND and 'SR' in channel
isAH = False #'qqqq' in channel or 'hp' in channel or 'lp' in channel
showSignal = False if 'SB' in cut or 'TR' in cut else True #'SR' in channel or channel=='qqqq'#or len(channel)==5
stype = "HVT model B"
if len(sign) > 0 and 'AZh' in sign[0]: stype = "2HDM"
elif len(sign) > 0 and 'monoH' in sign[0]: stype = "Z'-2HDM m_{A}=300 GeV"
if treeRead:
for k in sorted(alias.keys(), key=len, reverse=True):
if BTAGGING == 'semimedium':
if k in cut:
cut = cut.replace(k, aliasSM[k])
else:
if k in cut:
cut = cut.replace(
k, alias[k].format(WP=working_points[BTAGGING]))
print "Plotting from", ("tree" if treeRead else
"file"), var, "in", channel, "channel with:"
print " cut :", cut
if var == 'jj_deltaEta_widejet':
if "jj_deltaEta_widejet<1.1 && " in cut:
print
print "omitting jj_deltaEta_widejet<1.1 cut to draw the deltaEta distribution"
print
cut = cut.replace("jj_deltaEta_widejet<1.1 && ", "")
else:
print
print "no 'jj_deltaEta_widejet<1.1 && ' in the cut string detected, so it cannot be ommited explicitly"
print
### Create and fill MC histograms ###
# Create dict
file = {}
tree = {}
hist = {}
### Create and fill MC histograms ###
for i, s in enumerate(back + sign):
if True: #FIXME
if variable[var]['nbins'] > 0:
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events / ( " + str(
(variable[var]['max'] - variable[var]['min']) /
variable[var]['nbins']) + unit + " );" +
('log' if variable[var]['log'] else ''),
variable[var]['nbins'], variable[var]['min'],
variable[var]['max'])
else:
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events" +
('log' if variable[var]['log'] else ''),
len(variable[var]['bins']) - 1,
array('f', variable[var]['bins']))
hist[s].Sumw2()
for j, ss in enumerate(sample[s]['files']):
if not 'data' in s:
if year == "run2" or year in ss:
arr = root2array(
NTUPLEDIR + ss + ".root",
branches=[
var, "jpt_1", "jpt_2", "eventWeightLumi",
"TMath::Abs(jflavour_1)==5 && TMath::Abs(jflavour_2)==5",
"TMath::Abs(jflavour_1)==5 && TMath::Abs(jflavour_2)!=5",
"TMath::Abs(jflavour_1)!=5 && TMath::Abs(jflavour_2)==5",
"TMath::Abs(jflavour_1)!=5 && TMath::Abs(jflavour_2)!=5"
],
selection=cut if len(cut) > 0 else "")
print "imported " + NTUPLEDIR + ss + ".root"
arr.dtype.names = [
var, "jpt_1", "jpt_2", "eventWeightLumi", "bb",
"bq", "qb", "qq"
]
MANtag_eff1 = np.array(map(MANtag_eff, arr["jpt_1"]))
MANtag_eff2 = np.array(map(MANtag_eff, arr["jpt_2"]))
MANtag_mis1 = np.array(map(MANtag_mis, arr["jpt_1"]))
MANtag_mis2 = np.array(map(MANtag_mis, arr["jpt_2"]))
MANtag_weight = np.multiply(
arr["eventWeightLumi"],
np.multiply(arr['bb'],
np.multiply(MANtag_eff1, MANtag_eff2))
+ np.multiply(
arr['bq'], np.multiply(MANtag_eff1,
MANtag_mis2)) +
np.multiply(arr['qb'],
np.multiply(MANtag_mis1,
MANtag_eff2)) +
np.multiply(arr['qq'],
np.multiply(MANtag_mis1, MANtag_mis2)))
fill_hist(hist[s], arr[var], weights=MANtag_weight)
deepCSV_eff1 = np.array(map(deepCSV_eff, arr["jpt_1"]))
deepCSV_eff2 = np.array(map(deepCSV_eff, arr["jpt_2"]))
deepCSV_mis1 = np.array(map(deepCSV_mis, arr["jpt_1"]))
deepCSV_mis2 = np.array(map(deepCSV_mis, arr["jpt_2"]))
deepCSV_weight = np.multiply(
arr["eventWeightLumi"],
np.multiply(
arr['bb'],
np.multiply(deepCSV_eff1, deepCSV_eff2)) +
np.multiply(
arr['bq'],
np.multiply(deepCSV_eff1, deepCSV_mis2)) +
np.multiply(
arr['qb'],
np.multiply(deepCSV_mis1, deepCSV_eff2)) +
np.multiply(
arr['qq'],
np.multiply(deepCSV_mis1, deepCSV_mis2)))
if var == "jj_mass_widejet" and options.save and not "data" in ss:
arr = rfn.append_fields(arr,
"MANtag_weight",
MANtag_weight,
usemask=False)
arr = rfn.append_fields(arr,
"deepCSV_weight",
deepCSV_weight,
usemask=False)
array2root(arr,
NTUPLEDIR + "MANtag/" + ss + "_" +
BTAGGING + ".root",
treename="tree",
mode='recreate')
print "saved as", NTUPLEDIR + "MANtag/" + ss + "_" + BTAGGING + ".root"
arr = None
hist[s].Scale(sample[s]['weight'] if hist[s].Integral() >= 0 else 0)
hist[s].SetFillColor(sample[s]['fillcolor'])
hist[s].SetFillStyle(sample[s]['fillstyle'])
hist[s].SetLineColor(sample[s]['linecolor'])
hist[s].SetLineStyle(sample[s]['linestyle'])
if channel.endswith('TR') and channel.replace('TR', '') in topSF:
hist['TTbarSL'].Scale(topSF[channel.replace('TR', '')][0])
hist['ST'].Scale(topSF[channel.replace('TR', '')][0])
hist['BkgSum'] = hist['data_obs'].Clone(
"BkgSum") if 'data_obs' in hist else hist[back[0]].Clone("BkgSum")
hist['BkgSum'].Reset("MICES")
hist['BkgSum'].SetFillStyle(3003)
hist['BkgSum'].SetFillColor(1)
for i, s in enumerate(back):
hist['BkgSum'].Add(hist[s])
# Create data and Bkg sum histograms
if options.blind or 'SR' in channel:
hist['data_obs'] = hist['BkgSum'].Clone("data_obs")
hist['data_obs'].Reset("MICES")
# Set histogram style
hist['data_obs'].SetMarkerStyle(20)
hist['data_obs'].SetMarkerSize(1.25)
for i, s in enumerate(back + sign + ['BkgSum']):
addOverflow(hist[s], False) # Add overflow
for i, s in enumerate(sign):
hist[s].SetLineWidth(3)
for i, s in enumerate(sign):
sample[s][
'plot'] = True #sample[s]['plot'] and s.startswith(channel[:2])
if isAH:
for i, s in enumerate(back):
hist[s].SetFillStyle(3005)
hist[s].SetLineWidth(2)
#for i, s in enumerate(sign):
# hist[s].SetFillStyle(0)
if not var == "Events":
sfnorm = hist[data[0]].Integral() / hist['BkgSum'].Integral()
print "Applying SF:", sfnorm
for i, s in enumerate(back + ['BkgSum']):
hist[s].Scale(sfnorm)
if BLIND and var.endswith("Mass"):
for i, s in enumerate(data + back + ['BkgSum']):
first, last = hist[s].FindBin(65), hist[s].FindBin(135)
for j in range(first, last):
hist[s].SetBinContent(j, -1.e-4)
if BLIND and var.endswith("Tau21"):
for i, s in enumerate(data):
first, last = hist[s].FindBin(0), hist[s].FindBin(0.6)
for j in range(first, last):
hist[s].SetBinContent(j, -1.e-4)
# Create stack
if variable[var]['nbins'] > 0:
bkg = THStack(
"Bkg",
";" + hist['BkgSum'].GetXaxis().GetTitle() + ";Events / ( " + str(
(variable[var]['max'] - variable[var]['min']) /
variable[var]['nbins']) + unit + " )")
else:
bkg = THStack("Bkg",
";" + hist['BkgSum'].GetXaxis().GetTitle() + ";Events; ")
for i, s in enumerate(back):
bkg.Add(hist[s])
# Legend
leg = TLegend(0.65, 0.6, 0.95, 0.9)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
if len(data) > 0:
leg.AddEntry(hist[data[0]], sample[data[0]]['label'], "pe")
for i, s in reversed(list(enumerate(['BkgSum'] + back))):
leg.AddEntry(hist[s], sample[s]['label'], "f")
if showSignal:
for i, s in enumerate(sign):
if sample[s]['plot']:
leg.AddEntry(hist[s], sample[s]['label'], "fl")
leg.SetY1(0.9 - leg.GetNRows() * 0.05)
# --- Display ---
c1 = TCanvas("c1",
hist.values()[0].GetXaxis().GetTitle(), 800,
800 if RATIO else 600)
if RATIO:
c1.Divide(1, 2)
setTopPad(c1.GetPad(1), RATIO)
setBotPad(c1.GetPad(2), RATIO)
c1.cd(1)
c1.GetPad(bool(RATIO)).SetTopMargin(0.06)
c1.GetPad(bool(RATIO)).SetRightMargin(0.05)
c1.GetPad(bool(RATIO)).SetTicks(1, 1)
log = variable[var]['log'] #"log" in hist['BkgSum'].GetZaxis().GetTitle()
if log: c1.GetPad(bool(RATIO)).SetLogy()
# Draw
bkg.Draw("HIST") # stack
hist['BkgSum'].Draw("SAME, E2") # sum of bkg
if not isBlind and len(data) > 0: hist['data_obs'].Draw("SAME, PE") # data
if 'sync' in hist: hist['sync'].Draw("SAME, PE")
#data_graph.Draw("SAME, PE")
if showSignal:
smagn = 1. #if treeRead else 1.e2 #if log else 1.e2
for i, s in enumerate(sign):
# if sample[s]['plot']:
hist[s].Scale(smagn)
hist[s].Draw(
"SAME, HIST"
) # signals Normalized, hist[s].Integral()*sample[s]['weight']
textS = drawText(0.80, 0.9 - leg.GetNRows() * 0.05 - 0.02,
stype + " (x%d)" % smagn, True)
#bkg.GetYaxis().SetTitleOffset(bkg.GetYaxis().GetTitleOffset()*1.075)
bkg.GetYaxis().SetTitleOffset(0.9)
#bkg.GetYaxis().SetTitleOffset(2.)
bkg.SetMaximum((5. if log else 1.25) * max(
bkg.GetMaximum(),
hist['data_obs'].GetBinContent(hist['data_obs'].GetMaximumBin()) +
hist['data_obs'].GetBinError(hist['data_obs'].GetMaximumBin())))
#if bkg.GetMaximum() < max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum()): bkg.SetMaximum(max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum())*1.25)
bkg.SetMinimum(
max(
min(hist['BkgSum'].GetBinContent(hist['BkgSum'].GetMinimumBin(
)), hist['data_obs'].GetMinimum()), 5.e-1) if log else 0.)
if log:
bkg.GetYaxis().SetNoExponent(bkg.GetMaximum() < 1.e4)
#bkg.GetYaxis().SetMoreLogLabels(True)
bkg.GetXaxis().SetRangeUser(variable[var]['min'], variable[var]['max'])
#if log: bkg.SetMinimum(1)
leg.Draw()
#drawCMS(LUMI[year], "Preliminary")
drawCMS(LUMI[year], "Work in Progress", suppressCMS=True)
drawRegion('XVH' + channel, True)
drawAnalysis(channel)
setHistStyle(bkg, 1.2 if RATIO else 1.1)
setHistStyle(hist['BkgSum'], 1.2 if RATIO else 1.1)
if RATIO:
c1.cd(2)
err = hist['BkgSum'].Clone("BkgErr;")
err.SetTitle("")
err.GetYaxis().SetTitle("Data / MC")
err.GetYaxis().SetTitleOffset(0.9)
err.GetXaxis().SetRangeUser(variable[var]['min'], variable[var]['max'])
for i in range(1, err.GetNbinsX() + 1):
err.SetBinContent(i, 1)
if hist['BkgSum'].GetBinContent(i) > 0:
err.SetBinError(
i, hist['BkgSum'].GetBinError(i) /
hist['BkgSum'].GetBinContent(i))
setBotStyle(err)
errLine = err.Clone("errLine")
errLine.SetLineWidth(1)
errLine.SetFillStyle(0)
res = hist['data_obs'].Clone("Residues")
for i in range(0, res.GetNbinsX() + 1):
if hist['BkgSum'].GetBinContent(i) > 0:
res.SetBinContent(
i,
res.GetBinContent(i) / hist['BkgSum'].GetBinContent(i))
res.SetBinError(
i,
res.GetBinError(i) / hist['BkgSum'].GetBinContent(i))
if 'sync' in hist:
res.SetMarkerColor(2)
res.SetMarkerStyle(31)
res.Reset()
for i in range(0, res.GetNbinsX() + 1):
x = hist['data_obs'].GetXaxis().GetBinCenter(i)
if hist['sync'].GetBinContent(hist['sync'].FindBin(x)) > 0:
res.SetBinContent(
i, hist['data_obs'].GetBinContent(
hist['data_obs'].FindBin(x)) /
hist['sync'].GetBinContent(hist['sync'].FindBin(x)))
res.SetBinError(
i, hist['data_obs'].GetBinError(
hist['data_obs'].FindBin(x)) /
hist['sync'].GetBinContent(hist['sync'].FindBin(x)))
setBotStyle(res)
#err.GetXaxis().SetLabelOffset(err.GetXaxis().GetLabelOffset()*5)
#err.GetXaxis().SetTitleOffset(err.GetXaxis().GetTitleOffset()*2)
err.Draw("E2")
errLine.Draw("SAME, HIST")
if not isBlind and len(data) > 0:
res.Draw("SAME, PE0")
#res_graph.Draw("SAME, PE0")
if len(err.GetXaxis().GetBinLabel(
1)) == 0: # Bin labels: not a ordinary plot
drawRatio(hist['data_obs'], hist['BkgSum'])
drawStat(hist['data_obs'], hist['BkgSum'])
c1.Update()
if gROOT.IsBatch():
if channel == "": channel = "nocut"
varname = var.replace('.', '_').replace('()', '')
if not os.path.exists("plots/" + channel):
os.makedirs("plots/" + channel)
suffix = ''
if "b" in channel or 'mu' in channel: suffix += "_" + BTAGGING
c1.Print("plots/MANtag_study/" + channel + "/" + varname + "_" + year +
suffix + ".png")
c1.Print("plots/MANtag_study/" + channel + "/" + varname + "_" + year +
suffix + ".pdf")
# Print table
printTable(hist, sign)
def DeepCSV_pt_distribution(
year): ## everything below is jsut copy&past from above
from root_numpy import root2array, fill_hist, array2root
import numpy.lib.recfunctions as rfn
from aliases import alias_deepCSV, WP_deepCSV
### Preliminary Operations ###
treeRead = True
var = 'jpt_1'
channel = 'preselection'
cut = alias_deepCSV['preselection']
unit = ''
if "GeV" in variable[var]['title']: unit = ' GeV'
isBlind = BLIND and 'SR' in channel
isAH = False
showSignal = True
stype = "HVT model B"
if len(sign) > 0 and 'AZh' in sign[0]: stype = "2HDM"
elif len(sign) > 0 and 'monoH' in sign[0]: stype = "Z'-2HDM m_{A}=300 GeV"
if treeRead:
for k in sorted(alias_deepCSV.keys(), key=len, reverse=True):
if k in cut:
cut = cut.replace(
k, alias_deepCSV[k].format(WP=WP_deepCSV[BTAGGING][year]))
print "Plotting from", ("tree" if treeRead else
"file"), var, "in", channel, "channel with:"
print " cut :", cut
### Create and fill MC histograms ###
# Create dict
file = {}
tree = {}
hist = {}
N_signal_tot = 0.
N_signal_tag = 0.
### Create and fill MC histograms ###
for i, s in enumerate(back + sign):
if variable[var]['nbins'] > 0:
hist[s] = TH1F(
s, ";jet p_{T};Events / ( " + str(
(variable[var]['max'] - variable[var]['min']) /
variable[var]['nbins']) + unit + " );" +
('log' if variable[var]['log'] else ''),
variable[var]['nbins'], variable[var]['min'],
variable[var]['max'])
else:
hist[s] = TH1F(
s,
";jet p_{T};Events" + ('log' if variable[var]['log'] else ''),
len(variable[var]['bins']) - 1,
array('f', variable[var]['bins']))
hist[s].Sumw2()
for j, ss in enumerate(sample[s]['files']):
if not 'data' in s:
if year == "run2" or year in ss:
arr = root2array(NTUPLEDIR + ss + ".root",
branches=["jpt_1", "eventWeightLumi"],
selection=cut + " && jdeepCSV_1>" +
str(WP_deepCSV[BTAGGING][year]))
if 'signal' in ss.lower():
N_signal_tag += len(arr['jpt_1'][arr['jpt_1'] > 3500])
print "imported " + NTUPLEDIR + ss + ".root"
fill_hist(hist[s],
arr["jpt_1"],
weights=arr["eventWeightLumi"])
arr = None
arr = root2array(NTUPLEDIR + ss + ".root",
branches=["jpt_2", "eventWeightLumi"],
selection=cut + " && jdeepCSV_2>" +
str(WP_deepCSV[BTAGGING][year]))
print "imported " + NTUPLEDIR + ss + ".root"
if 'signal' in ss.lower():
N_signal_tag += len(arr['jpt_2'][arr['jpt_2'] > 3500])
fill_hist(hist[s],
arr["jpt_2"],
weights=arr["eventWeightLumi"])
arr = None
if 'signal' in ss.lower():
arr = root2array(NTUPLEDIR + ss + ".root",
branches=["jpt_1", "eventWeightLumi"],
selection=cut)
N_signal_tot += len(arr['jpt_1'][arr['jpt_1'] > 3500])
arr = None
arr = root2array(NTUPLEDIR + ss + ".root",
branches=["jpt_2", "eventWeightLumi"],
selection=cut)
N_signal_tot += len(arr['jpt_2'][arr['jpt_2'] > 3500])
arr = None
hist[s].Scale(sample[s]['weight'] if hist[s].Integral() >= 0 else 0)
hist[s].SetFillColor(sample[s]['fillcolor'])
hist[s].SetFillStyle(sample[s]['fillstyle'])
hist[s].SetLineColor(sample[s]['linecolor'])
hist[s].SetLineStyle(sample[s]['linestyle'])
if channel.endswith('TR') and channel.replace('TR', '') in topSF:
hist['TTbarSL'].Scale(topSF[channel.replace('TR', '')][0])
hist['ST'].Scale(topSF[channel.replace('TR', '')][0])
hist['BkgSum'] = hist['data_obs'].Clone(
"BkgSum") if 'data_obs' in hist else hist[back[0]].Clone("BkgSum")
hist['BkgSum'].Reset("MICES")
hist['BkgSum'].SetFillStyle(3003)
hist['BkgSum'].SetFillColor(1)
for i, s in enumerate(back):
hist['BkgSum'].Add(hist[s])
# Create data and Bkg sum histograms
if options.blind or 'SR' in channel:
hist['data_obs'] = hist['BkgSum'].Clone("data_obs")
hist['data_obs'].Reset("MICES")
# Set histogram style
hist['data_obs'].SetMarkerStyle(20)
hist['data_obs'].SetMarkerSize(1.25)
for i, s in enumerate(back + sign + ['BkgSum']):
addOverflow(hist[s], False) # Add overflow
for i, s in enumerate(sign):
hist[s].SetLineWidth(3)
for i, s in enumerate(sign):
sample[s][
'plot'] = True #sample[s]['plot'] and s.startswith(channel[:2])
if isAH:
for i, s in enumerate(back):
hist[s].SetFillStyle(3005)
hist[s].SetLineWidth(2)
#for i, s in enumerate(sign):
# hist[s].SetFillStyle(0)
if not var == "Events":
sfnorm = hist[data[0]].Integral() / hist['BkgSum'].Integral()
print "Applying SF:", sfnorm
for i, s in enumerate(back + ['BkgSum']):
hist[s].Scale(sfnorm)
if BLIND and var.endswith("Mass"):
for i, s in enumerate(data + back + ['BkgSum']):
first, last = hist[s].FindBin(65), hist[s].FindBin(135)
for j in range(first, last):
hist[s].SetBinContent(j, -1.e-4)
if BLIND and var.endswith("Tau21"):
for i, s in enumerate(data):
first, last = hist[s].FindBin(0), hist[s].FindBin(0.6)
for j in range(first, last):
hist[s].SetBinContent(j, -1.e-4)
# Create stack
if variable[var]['nbins'] > 0:
bkg = THStack(
"Bkg",
";" + hist['BkgSum'].GetXaxis().GetTitle() + ";Events / ( " + str(
(variable[var]['max'] - variable[var]['min']) /
variable[var]['nbins']) + unit + " )")
else:
bkg = THStack("Bkg",
";" + hist['BkgSum'].GetXaxis().GetTitle() + ";Events; ")
for i, s in enumerate(back):
bkg.Add(hist[s])
# Legend
leg = TLegend(0.65, 0.6, 0.95, 0.9)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
if len(data) > 0:
leg.AddEntry(hist[data[0]], sample[data[0]]['label'], "pe")
for i, s in reversed(list(enumerate(['BkgSum'] + back))):
leg.AddEntry(hist[s], sample[s]['label'], "f")
if showSignal:
for i, s in enumerate(sign):
if sample[s]['plot']:
leg.AddEntry(hist[s], sample[s]['label'], "fl")
leg.SetY1(0.9 - leg.GetNRows() * 0.05)
# --- Display ---
c1 = TCanvas("c1",
hist.values()[0].GetXaxis().GetTitle(), 800,
800 if RATIO else 600)
if RATIO:
c1.Divide(1, 2)
setTopPad(c1.GetPad(1), RATIO)
setBotPad(c1.GetPad(2), RATIO)
c1.cd(1)
c1.GetPad(bool(RATIO)).SetTopMargin(0.06)
c1.GetPad(bool(RATIO)).SetRightMargin(0.05)
c1.GetPad(bool(RATIO)).SetTicks(1, 1)
log = variable[var]['log'] #"log" in hist['BkgSum'].GetZaxis().GetTitle()
if log: c1.GetPad(bool(RATIO)).SetLogy()
# Draw
bkg.Draw("HIST") # stack
hist['BkgSum'].Draw("SAME, E2") # sum of bkg
if not isBlind and len(data) > 0: hist['data_obs'].Draw("SAME, PE") # data
if 'sync' in hist: hist['sync'].Draw("SAME, PE")
#data_graph.Draw("SAME, PE")
if showSignal:
smagn = 1. #if treeRead else 1.e2 #if log else 1.e2
for i, s in enumerate(sign):
# if sample[s]['plot']:
hist[s].Scale(smagn)
hist[s].Draw(
"SAME, HIST"
) # signals Normalized, hist[s].Integral()*sample[s]['weight']
textS = drawText(0.80, 0.9 - leg.GetNRows() * 0.05 - 0.02,
stype + " (x%d)" % smagn, True)
#bkg.GetYaxis().SetTitleOffset(bkg.GetYaxis().GetTitleOffset()*1.075)
bkg.GetYaxis().SetTitleOffset(0.9)
#bkg.GetYaxis().SetTitleOffset(2.)
bkg.SetMaximum((5. if log else 1.25) * max(
bkg.GetMaximum(),
hist['data_obs'].GetBinContent(hist['data_obs'].GetMaximumBin()) +
hist['data_obs'].GetBinError(hist['data_obs'].GetMaximumBin())))
#if bkg.GetMaximum() < max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum()): bkg.SetMaximum(max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum())*1.25)
bkg.SetMinimum(
max(
min(hist['BkgSum'].GetBinContent(hist['BkgSum'].GetMinimumBin(
)), hist['data_obs'].GetMinimum()), 5.e-1) if log else 0.)
if log:
bkg.GetYaxis().SetNoExponent(bkg.GetMaximum() < 1.e4)
#bkg.GetYaxis().SetMoreLogLabels(True)
bkg.GetXaxis().SetRangeUser(variable[var]['min'], variable[var]['max'])
#if log: bkg.SetMinimum(1)
leg.Draw()
#drawCMS(LUMI[year], "Preliminary")
drawCMS(LUMI[year], "", suppressCMS=True)
drawRegion('XVH' + channel, True)
drawAnalysis(channel)
setHistStyle(bkg, 1.2 if RATIO else 1.1)
setHistStyle(hist['BkgSum'], 1.2 if RATIO else 1.1)
if RATIO:
c1.cd(2)
err = hist['BkgSum'].Clone("BkgErr;")
err.SetTitle("")
err.GetYaxis().SetTitle("Data / MC")
err.GetYaxis().SetTitleOffset(0.9)
err.GetXaxis().SetRangeUser(variable[var]['min'], variable[var]['max'])
for i in range(1, err.GetNbinsX() + 1):
err.SetBinContent(i, 1)
if hist['BkgSum'].GetBinContent(i) > 0:
err.SetBinError(
i, hist['BkgSum'].GetBinError(i) /
hist['BkgSum'].GetBinContent(i))
setBotStyle(err)
errLine = err.Clone("errLine")
errLine.SetLineWidth(1)
errLine.SetFillStyle(0)
res = hist['data_obs'].Clone("Residues")
for i in range(0, res.GetNbinsX() + 1):
if hist['BkgSum'].GetBinContent(i) > 0:
res.SetBinContent(
i,
res.GetBinContent(i) / hist['BkgSum'].GetBinContent(i))
res.SetBinError(
i,
res.GetBinError(i) / hist['BkgSum'].GetBinContent(i))
if 'sync' in hist:
res.SetMarkerColor(2)
res.SetMarkerStyle(31)
res.Reset()
for i in range(0, res.GetNbinsX() + 1):
x = hist['data_obs'].GetXaxis().GetBinCenter(i)
if hist['sync'].GetBinContent(hist['sync'].FindBin(x)) > 0:
res.SetBinContent(
i, hist['data_obs'].GetBinContent(
hist['data_obs'].FindBin(x)) /
hist['sync'].GetBinContent(hist['sync'].FindBin(x)))
res.SetBinError(
i, hist['data_obs'].GetBinError(
hist['data_obs'].FindBin(x)) /
hist['sync'].GetBinContent(hist['sync'].FindBin(x)))
setBotStyle(res)
#err.GetXaxis().SetLabelOffset(err.GetXaxis().GetLabelOffset()*5)
#err.GetXaxis().SetTitleOffset(err.GetXaxis().GetTitleOffset()*2)
err.Draw("E2")
errLine.Draw("SAME, HIST")
if not isBlind and len(data) > 0:
res.Draw("SAME, PE0")
#res_graph.Draw("SAME, PE0")
if len(err.GetXaxis().GetBinLabel(
1)) == 0: # Bin labels: not a ordinary plot
drawRatio(hist['data_obs'], hist['BkgSum'])
drawStat(hist['data_obs'], hist['BkgSum'])
c1.Update()
if gROOT.IsBatch():
if channel == "": channel = "nocut"
varname = var.replace('.', '_').replace('()', '')
if not os.path.exists("plots/" + channel):
os.makedirs("plots/" + channel)
suffix = ''
if "b" in channel or 'mu' in channel: suffix += "_" + BTAGGING
c1.Print("plots/MANtag_study/deepCSV_plots/pt_" + year + suffix +
".png")
c1.Print("plots/MANtag_study/deepCSV_plots/pt_" + year + suffix +
".pdf")
# Print table
printTable(hist, sign)
print 'deepCSV efficiency:', N_signal_tag / N_signal_tot
def limit2HDM():
global signals
signals = range(800, 2000 + 1, 50)
multF = HTOBB
THEORY = ['T1', 'T2']
mass, val = fillValues("./combine/AZh/AZh_M%d.txt")
Obs0s = TGraph()
Exp0s = TGraph()
Exp1s = TGraphAsymmErrors()
Exp2s = TGraphAsymmErrors()
massB, valB = fillValues("./combine/BBAZh/BBAZh_M%d.txt")
Obs0sB = TGraph()
Exp0sB = TGraph()
Exp1sB = TGraphAsymmErrors()
Exp2sB = TGraphAsymmErrors()
for i, m in enumerate(mass):
if not m in val:
print "Key Error:", m, "not in value map"
continue
n = Exp0s.GetN()
Obs0s.SetPoint(n, m, val[m][0] * multF)
Exp0s.SetPoint(n, m, val[m][3] * multF)
Exp1s.SetPoint(n, m, val[m][3] * multF)
Exp1s.SetPointError(n, 0., 0., val[m][3] * multF - val[m][2] * multF,
val[m][4] * multF - val[m][3] * multF)
Exp2s.SetPoint(n, m, val[m][3] * multF)
Exp2s.SetPointError(n, 0., 0., val[m][3] * multF - val[m][1] * multF,
val[m][5] * multF - val[m][3] * multF)
Obs0sB.SetPoint(n, m, valB[m][0] * multF)
Exp0sB.SetPoint(n, m, valB[m][3] * multF)
Exp1sB.SetPoint(n, m, valB[m][3] * multF)
Exp1sB.SetPointError(n, 0., 0.,
valB[m][3] * multF - valB[m][2] * multF,
valB[m][4] * multF - valB[m][3] * multF)
Exp2sB.SetPoint(n, m, valB[m][3] * multF)
Exp2sB.SetPointError(n, 0., 0.,
valB[m][3] * multF - valB[m][1] * multF,
valB[m][5] * multF - valB[m][3] * multF)
col = 629
Exp2s.SetLineWidth(2)
Exp2s.SetLineStyle(1)
Obs0s.SetLineWidth(3)
Obs0s.SetMarkerStyle(0)
Obs0s.SetLineColor(1)
Exp0s.SetLineStyle(2)
Exp0s.SetLineWidth(3)
Exp0s.SetLineColor(1)
# Exp1s.SetFillColorAlpha(col, 0.4) #kGreen+1
# Exp1s.SetLineColorAlpha(col, 0.4)
# Exp2s.SetFillColorAlpha(col, 0.2) #kOrange
# Exp2s.SetLineColorAlpha(col, 0.2)
Exp1s.SetFillColor(417)
Exp1s.SetLineColor(417)
Exp2s.SetFillColor(800)
Exp2s.SetLineColor(800)
colB = 922
Exp2sB.SetLineWidth(2)
Obs0sB.SetLineStyle(9)
Obs0sB.SetLineWidth(3)
Obs0sB.SetMarkerStyle(0)
Obs0sB.SetLineColor(colB)
Exp0sB.SetLineStyle(8)
Exp0sB.SetLineWidth(3)
Exp0sB.SetLineColor(colB)
Exp1sB.SetFillColorAlpha(colB, 0.4) #kGreen+1
Exp1sB.SetLineColorAlpha(colB, 0.4)
Exp2sB.SetFillColorAlpha(colB, 0.2) #kOrange
Exp2sB.SetLineColorAlpha(colB, 0.2)
Exp2s.GetXaxis().SetTitle("m_{A} (GeV)")
Exp2s.GetXaxis().SetTitleSize(Exp2s.GetXaxis().GetTitleSize() * 1.25)
Exp2s.GetXaxis().SetNoExponent(True)
Exp2s.GetXaxis().SetMoreLogLabels(True)
Exp2s.GetYaxis().SetTitle(
"#sigma(A) #bf{#it{#Beta}}(A #rightarrow Zh) #bf{#it{#Beta}}(h #rightarrow bb) (fb)"
)
Exp2s.GetYaxis().SetTitleOffset(1.5)
Exp2s.GetYaxis().SetNoExponent(True)
Exp2s.GetYaxis().SetMoreLogLabels()
Theory = {}
#for t in THEORY:
# Theory[t] = TGraphAsymmErrors()
# for m in sorted(THDM[t]['ggA'].keys()):
# if m < mass[0] or m > mass[-1]: continue
# Xs, Xs_Up, Xs_Down = 0., 0., 0.
# Xs = THDM[t]['ggA'][m]
# Xs_Up = Xs*(1.+math.sqrt((THDM['PDF']['ggA'][m][0]-1.)**2 + (THDM['QCD']['ggA'][m][0]-1.)**2))
# Xs_Down = Xs*(1.-math.sqrt((1.-THDM['PDF']['ggA'][m][1])**2 + (1.-THDM['QCD']['ggA'][m][1])**2))
# n = Theory[t].GetN()
# Theory[t].SetPoint(n, m, Xs)
# Theory[t].SetPointError(n, 0., 0., (Xs-Xs_Down), (Xs_Up-Xs))
# Theory[t].SetLineColor(theoryLineColor[t])
# Theory[t].SetFillColor(theoryFillColor[t])
# Theory[t].SetFillStyle(theoryFillStyle[t])
# Theory[t].SetLineWidth(2)
# #Theory[t].SetLineStyle(7)
c1 = TCanvas("c1", "Exclusion Limits", 800, 600)
c1.cd()
#SetPad(c1.GetPad(0))
c1.GetPad(0).SetTopMargin(0.06)
c1.GetPad(0).SetRightMargin(0.05)
c1.GetPad(0).SetLeftMargin(0.12)
c1.GetPad(0).SetTicks(1, 1)
c1.GetPad(0).SetLogy()
Exp2s.Draw("A3")
Exp1s.Draw("SAME, 3")
Exp0s.Draw("SAME, L")
# Exp2sB.Draw("SAME, 3")
# Exp1sB.Draw("SAME, 3")
Exp0sB.Draw("SAME, L")
if not options.blind:
Obs0s.Draw("SAME, L")
Obs0sB.Draw("SAME, L")
for t in THEORY:
Theory[t].Draw("SAME, L3")
Theory[t].Draw("SAME, L3X0Y0")
#setHistStyle(Exp2s)
# Exp2s.GetXaxis().SetTitleSize(0.045)
# Exp2s.GetYaxis().SetTitleSize(0.04)
# Exp2s.GetXaxis().SetLabelSize(0.04)
# Exp2s.GetYaxis().SetLabelSize(0.04)
# Exp2s.GetXaxis().SetTitleOffset(1)
# Exp2s.GetYaxis().SetTitleOffset(1.25)
Exp2s.GetXaxis().SetTitleSize(0.050)
Exp2s.GetYaxis().SetTitleSize(0.050)
Exp2s.GetXaxis().SetLabelSize(0.045)
Exp2s.GetYaxis().SetLabelSize(0.045)
Exp2s.GetXaxis().SetTitleOffset(0.90)
Exp2s.GetYaxis().SetTitleOffset(1.25)
Exp2s.GetYaxis().SetMoreLogLabels(True)
Exp2s.GetYaxis().SetNoExponent(True)
Exp2s.GetYaxis().SetRangeUser(0.5, 1.e3)
Exp2s.GetXaxis().SetRangeUser(mass[0], mass[-1])
drawAnalysis('AZh')
drawRegion('AZHsl', True)
drawCMS(LUMI, "") #Preliminary
#drawCMS(LUMI, "Work in Progress", suppressCMS=True)
# legend
leg = TLegend(0.6, 0.90, 0.99, 0.90)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
leg.SetHeader("95% CL upper limits")
leg.AddEntry(None, "gg #rightarrow A #rightarrow Zh",
"") #"95% CL upper limits"
leg.AddEntry(Obs0s, "Observed", "l")
leg.AddEntry(Exp0s, "Expected", "l")
leg.AddEntry(Exp1s, "#pm 1 std. deviation", "f")
leg.AddEntry(Exp2s, "#pm 2 std. deviation", "f")
leg.AddEntry(None, "", "")
leg.AddEntry(None, "bbA #rightarrow Zh", "")
leg.AddEntry(Obs0sB, "Observed", "l")
leg.AddEntry(Exp0sB, "Expected", "l")
leg.SetY1(leg.GetY2() - leg.GetNRows() * 0.045)
leg.Draw()
# latex = TLatex()
# latex.SetNDC()
# latex.SetTextSize(0.040)
# latex.SetTextFont(42)
# latex.DrawLatex(0.65, leg.GetY1()-0.045, "cos(#beta-#alpha)=0.25, tan(#beta)=1")
# legB = TLegend(0.12, 0.4-4*0.3/5., 0.65, 0.4)
legB = TLegend(0.15, 0.27, 0.68, 0.27)
legB.SetBorderSize(0)
legB.SetFillStyle(0) #1001
legB.SetFillColor(0)
for t in THEORY:
legB.AddEntry(Theory[t], theoryLabel[t], "fl")
legB.AddEntry(None, "cos(#beta-#alpha)=0.25, tan(#beta)=1", "")
legB.SetY1(legB.GetY2() - legB.GetNRows() * 0.045)
legB.Draw()
c1.GetPad(0).RedrawAxis()
leg.Draw()
c1.Update()
if not gROOT.IsBatch(): raw_input("Press Enter to continue...")
c1.Print("plotsLimit/Exclusion/THDM.png")
c1.Print("plotsLimit/Exclusion/THDM.pdf")
def make_plot_all_rods(error_dict, rod_dict, name):
leg = TLegend(0,0,0,0)
if 'lock' in name:
leg = TLegend(0.18,0.85,0.50,0.55)
else:
leg = TLegend(0.18,0.85,0.40,0.55)
leg.SetLineColor(0)
leg.SetFillStyle(0)
leg.SetShadowColor(0)
leg.SetBorderSize(0)
leg.SetNColumns(3)
gStyle.SetLegendTextSize(0.045)
v_hists = []
#for e,c in zip(error_bits, error_colors):
for e in error_bits:
h = TH1F('h'+e,'h'+e, len(rod_dict), 0, len(rod_dict))
h.SetFillStyle(1001)
h.SetLineWidth(0)
v_hists.append(h)
v_hists[-1].SetDirectory(0)
if bin(int('0x'+e, 16))[2:].zfill(4) == '0111':
leg.AddEntry(v_hists[-1],'GOL 3',"f");
elif bin(int('0x'+e, 16))[2:].zfill(4) == '1011':
leg.AddEntry(v_hists[-1],'GOL 2',"f");
elif bin(int('0x'+e, 16))[2:].zfill(4) == '1101':
leg.AddEntry(v_hists[-1],'GOL 1',"f");
elif bin(int('0x'+e, 16))[2:].zfill(4) == '1110':
leg.AddEntry(v_hists[-1],'GOL 0',"f");
else:
leg.AddEntry(v_hists[-1],bin(int('0x'+e, 16))[2:].zfill(4),"f");
h = leg.GetY2()-leg.GetY1();
w = leg.GetX2()-leg.GetX1()*.6;
leg.SetMargin(leg.GetNColumns()*h/(leg.GetNRows()*w))
for key,val in error_dict.items():
idx_rod = 0
for i, key2 in enumerate(rod_dict):
if key2 == key[:8]:
idx_rod = i
v_hists[int(key[11:12], 16)].Fill(idx_rod, val)
stack = THStack("stack","stack")
for hist in v_hists:
stack.Add(hist)
if 'buff' in name:
c1 = TCanvas( 'c1', 'c1', 2000, 500)
else:
c1 = TCanvas( 'c1', 'c1', 1000, 500)
h1 = TH1F('h_1','h_1', len(rod_dict), 0, len(rod_dict))
for i, key in enumerate(rod_dict):
h1.GetXaxis().SetBinLabel(i+1,key)
h1.SetBinContent(i+1,rod_dict[key])
h1.GetXaxis().LabelsOption("v")
h1.GetXaxis().SetTitle("ROD")
h1.GetXaxis().SetTitleOffset(2.2)
h1.GetYaxis().SetTitle("# of rocketio errors")
h1.SetLineColor(kRed)
h1.SetLineWidth(1)
leg.AddEntry(h1,'total',"l");
c1.SetBottomMargin(0.23)
h1.GetXaxis().SetTitle("ROD")
h1.Draw("HIST")
stack.Draw("PFC PLC SAME HIST")
h1.Draw("SAMEHIST")
AtlasStyle.ATLAS_LABEL(0.19,.88, 1, "Internal")
leg.Draw()
c1.Update()
c1.Print("plots/"+name +".pdf")
c1.Clear()
def addSys(var, cut, sys):
binLow = ""
binHigh = ""
binName = ""
if "binned" in cut:
binLow = cut[cut.find("LowVal") + 6:cut.find("HighVal") - 1]
binHigh = cut[cut.find("HighVal") + 7:]
binName = "bin_" + binLow + "_" + binHigh
cut = cut[:cut.find("binned")]
channel = cut
weight = "eventWeightLumi" #+ ("*stitchWeight" if any([x for x in back if x.endswith('b')]) else "")
cut = selection[cut]
if not binLow == "":
cut = cut + " && " + var + " > " + binLow + " && " + var + " < " + binHigh
weightUp = weightDown = weight
varUp = varDown = var
cutUp = cutDown = cut
# Systematics
if sys == 'CMS_scale_j':
if var != "MET_sign": varUp = var.replace('pt', 'ptScaleUp')
else: varUp = var.replace('sign', 'signScaleUp')
if var != "MET_sign": varDown = var.replace('pt', 'ptScaleDown')
else: varDown = var.replace('sign', 'signScaleDown')
cutUp = cut.replace('MET_pt', 'MET_ptScaleUp')
cutUp = cutUp.replace('Jets', 'JetsScaleUp')
cutUp = cutUp.replace('12', '12ScaleUp')
cutUp = cutUp.replace('mT>', 'mTScaleUp>')
cutUp = cutUp.replace('mT2', 'mT2ScaleUp')
cutDown = cut.replace('MET_pt', 'MET_ptScaleDown')
cutDown = cutDown.replace('Jets', 'JetsScaleDown')
cutDown = cutDown.replace('12', '12ScaleDown')
cutDown = cutDown.replace('mT>', 'mTScaleDown>')
cutDown = cutDown.replace('mT2', 'mT2ScaleDown')
elif sys == 'CMS_res_j':
if var != "MET_sign": varUp = var.replace('pt', 'ptResUp')
else: varUp = var.replace('sign', 'signResUp')
if var != "MET_sign": varDown = var.replace('pt', 'ptResDown')
else: varDown = var.replace('sign', 'signResDown')
cutUp = cut.replace('MET_pt', 'MET_ptResUp')
cutUp = cutUp.replace('Jets', 'JetsResUp')
cutUp = cutUp.replace('12', '12ResUp')
cutUp = cutUp.replace('mT>', 'mTResUp>')
cutUp = cutUp.replace('mT2', 'mT2ResUp')
cutDown = cut.replace('MET_pt', 'MET_ptResDown')
cutDown = cutDown.replace('Jets', 'JetsResDown')
cutDown = cutDown.replace('12', '12ResDown')
cutDown = cutDown.replace('mT>', 'mTResDown>')
cutDown = cutDown.replace('mT2', 'mT2ResDown')
elif sys == 'CMS_WqcdWeightRen':
weightUp += "*WqcdWeightRenUp/WqcdWeight"
weightDown += "*WqcdWeightRenDown/WqcdWeight"
elif sys == 'CMS_WqcdWeightFac':
weightUp += "*WqcdWeightFacUp/WqcdWeight"
weightDown += "*WqcdWeightFacDown/WqcdWeight"
elif sys == 'CMS_ZqcdWeightRen':
weightUp += "*ZqcdWeightRenUp/ZqcdWeight"
weightDown += "*ZqcdWeightRenDown/ZqcdWeight"
elif sys == 'CMS_ZqcdWeightFac':
weightUp += "*ZqcdWeightFacUp/ZqcdWeight"
weightDown += "*ZqcdWeightFacDown/ZqcdWeight"
elif sys == 'CMS_WewkWeight':
weightUp += "/WewkWeight"
weightDown += ""
elif sys == 'CMS_ZewkWeight':
weightUp += "/ZewkWeight"
weightDown += ""
elif sys == 'CMS_pdf':
weightUp += "*PDFWeightUp/eventWeight"
weightDown += "*PDFWeightDown/eventWeight"
elif sys == 'CMS_HF':
weightUp += "*1.20"
weightDown += "*0.8"
elif sys == 'CMS_eff_b':
weightUp += "*bTagWeightUp/bTagWeight"
weightDown += "*bTagWeightDown/bTagWeight"
elif sys == 'CMS_scale_pu':
weightUp += "*puWeightUp/puWeight"
weightDown += "*puWeightDown/puWeight"
elif sys == 'CMS_scale_top':
weightUp += "/TopWeight"
weightDown += ""
elif sys == 'CMS_eff_trigger':
weightUp += "*triggerWeightUp/triggerWeight"
weightDown += "*triggerWeightDown/triggerWeight"
elif sys == 'CMS_eff_e' and '2e' in cut or '1e' in channel:
weightUp += "*leptonWeightUp/leptonWeight"
weightDown += "*leptonWeightDown/leptonW\
eight"
elif sys == 'CMS_eff_m' and '2m' in cut or '1m' in channel:
weightUp += "*leptonWeightUp/leptonWeight"
weightDown += "*leptonWeightDown/leptonW\
eight"
elif sys == 'QCDscale_ren':
weightUp += "*QCDRenWeightUp"
weightDown += "*QCDRenWeightDown"
elif sys == 'QCDscale_fac':
weightUp += "*QCDFacWeightUp"
weightDown += "*QCDFacWeightDown"
# elif sys=='EWKscale_Z': weightDown += "/ZewkWeight"
# elif sys=='EWKscale_W': weightDown += "/WewkWeight"
else:
print "Systematic", sys, "not applicable or not recognized."
### Create and fill MC histograms ###
file = {}
tree = {}
hist = {}
histUp = {}
histDown = {}
isBlind = BLIND and 'SR' in channel
for i, s in enumerate(back + sign):
tree[s] = TChain("tree")
for j, ss in enumerate(sample[s]['files']):
tree[s].Add(NTUPLEDIR + ss + ".root")
if not binLow == "":
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events;" +
('log' if variable[var]['log'] else ''), 1, float(binLow),
float(binHigh))
elif binLow == "" and variable[var]['nbins'] > 0:
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events;" +
('log' if variable[var]['log'] else ''),
variable[var]['nbins'], variable[var]['min'],
variable[var]['max'])
else:
hist[s] = TH1F(s, ";" + variable[var]['title'],
len(variable[var]['bins']) - 1,
array('f', variable[var]['bins']))
hist[s].Sumw2()
histUp[s] = hist[s].Clone(s + 'Up')
histDown[s] = hist[s].Clone(s + 'Down')
redFactorString = ""
redFactorValue = ""
#if isBlind and 'data' not in s and options.limit:
if isBlind and 'data' not in s:
redFactorValue = " / 15"
cutstring = ("*(" + cut + ")" if len(cut) > 0 else "")
cutstringUp = ("*(" + cutUp + ")" if len(cut) > 0 else "")
cutstringDown = ("*(" + cutDown + ")" if len(cut) > 0 else "")
if '-' in s:
cutstring = cutstring.replace(
cut, cut + "&& nBQuarks==" + s.split('-')[1][0])
tree[s].Project(s, var,
"(" + weight + redFactorValue + ")" + cutstring)
if 'HF' not in sys or 'QCDscale' not in sys:
tree[s].Project(s + 'Up', varUp,
"(" + weightUp + redFactorValue + ")" + cutstring)
tree[s].Project(
s + 'Down', varDown,
"(" + weightDown + redFactorValue + ")" + cutstring)
if 'HF' in sys:
if s.startswith('WJ') or s.startswith('ZJ') or s.startswith(
'DYJets'):
tree[s].Project(
s + 'Up', varUp,
"(" + weightUp + redFactorValue + ")" + cutstringUp)
tree[s].Project(
s + 'Down', varDown,
"(" + weightDown + redFactorValue + ")" + cutstringDown)
else:
tree[s].Project(
s + 'Up', varUp,
"(" + weight + redFactorValue + ")" + cutstringUp)
tree[s].Project(
s + 'Down', varDown,
"(" + weight + redFactorValue + ")" + cutstringDown)
if 'QCDscale' in sys:
if s.startswith('WJ') or s.startswith('ZJ') or s.startswith(
'DYJets'):
tree[s].Project(
s + 'Up', varUp,
"(" + weight + redFactorValue + ")" + cutstringUp)
tree[s].Project(
s + 'Down', varDown,
"(" + weight + redFactorValue + ")" + cutstringDown)
else:
tree[s].Project(
s + 'Up', varUp,
"(" + weightUp + redFactorValue + ")" + cutstringUp)
tree[s].Project(
s + 'Down', varDown,
"(" + weightDown + redFactorValue + ")" + cutstringDown)
hist[s].Scale(sample[s]['weight'] if hist[s].Integral() >= 0 else 0)
hist[s].SetLineWidth(2)
histUp[s].SetLineWidth(2)
histDown[s].SetLineWidth(2)
hist[s].SetLineColor(1)
histUp[s].SetLineColor(629)
histDown[s].SetLineColor(602)
# Rescale normalization for QCD scales FIXME
if 'QCDscale' in sys:
for s in back + sign: #['TTbar', 'TTbarSL', 'ST']:
if s in hist and histUp[s].Integral(
) > 0. and histDown[s].Integral() > 0.:
histUp[s].Scale(hist[s].Integral() / histUp[s].Integral())
histDown[s].Scale(hist[s].Integral() / histDown[s].Integral())
hist['BkgSum'] = hist[back[0]].Clone("BkgSum")
hist['BkgSum'].Reset()
histUp['BkgSum'] = hist['BkgSum'].Clone("BkgSumUp")
histUp['BkgSum'].SetLineColor(629)
histUp['BkgSum'].Reset()
histDown['BkgSum'] = hist['BkgSum'].Clone("BkgSumDown")
histDown['BkgSum'].SetLineColor(602)
histDown['BkgSum'].Reset()
for i, s in enumerate(back):
hist['BkgSum'].Add(hist[s], 1)
histUp['BkgSum'].Add(histUp[s], 1)
histDown['BkgSum'].Add(histDown[s], 1)
for i, s in enumerate(back + sign + ['BkgSum']):
addOverflow(hist[s], False)
addOverflow(histUp[s], False)
addOverflow(histDown[s], False)
c1 = TCanvas("c1", "Signals", 800, 600)
c1.cd()
gStyle.SetOptStat(0)
gStyle.SetOptTitle(0)
if RATIO:
c1.Divide(1, 2)
setTopPad(c1.GetPad(1), RATIO)
setBotPad(c1.GetPad(2), RATIO)
c1.cd(1)
c1.GetPad(bool(RATIO)).SetTopMargin(0.06)
c1.GetPad(bool(RATIO)).SetRightMargin(0.06)
c1.GetPad(bool(RATIO)).SetTicks(1, 1)
c1.GetPad(bool(RATIO)).SetLogy()
histUp['BkgSum'].SetMaximum(histUp['BkgSum'].GetMaximum() * 5)
histUp['BkgSum'].Draw("HIST")
histDown['BkgSum'].Draw("SAME, HIST")
hist['BkgSum'].Draw("SAME, HIST")
drawCMS(-1, "Simulation", False)
setHistStyle(histUp['BkgSum'], 1.2 if RATIO else 1.1)
if RATIO:
c1.cd(2)
errUp = histUp['BkgSum'].Clone("BkgUp;")
errUp.Add(hist['BkgSum'], -1)
errUp.Divide(hist['BkgSum'])
errUp.SetTitle("")
errUp.GetYaxis().SetTitle("#frac{shifted-central}{central}")
errUp.GetYaxis().SetNdivisions(503)
setBotStyle(errUp)
errUp.GetYaxis().SetRangeUser(-0.3, 0.3)
errUp.Draw("HIST")
errDown = histDown['BkgSum'].Clone("BkgDown;")
errDown.Add(hist['BkgSum'], -1)
errDown.Divide(hist['BkgSum'])
errDown.Draw("SAME, HIST")
f1 = TF1("myfunc", "[0]", -100000, 10000)
f1.SetLineColor(1)
f1.SetLineStyle(7)
f1.SetLineWidth(1)
f1.SetParameter(0, 0)
f1.Draw("same")
leg = TLegend(0.65, 0.80, 0.95, 0.80)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetHeader(sys.replace('CMS', '').replace('_', ' '))
leg.AddEntry(histUp['BkgSum'], "Up", "l")
leg.AddEntry(hist['BkgSum'], "Central", "l")
leg.AddEntry(histDown['BkgSum'], "Down", "l")
leg.SetY1(0.75 - leg.GetNRows() * 0.045)
c1.cd(1)
leg.Draw()
if options.saveplots:
if not os.path.exists("plotsSys_" + options.name + "/" + channel +
binName):
os.makedirs("plotsSys_" + options.name + "/" + channel + binName)
c1.Print("plotsSys_" + options.name + "/" + channel + binName + "/" +
sys + ".png")
c1.Print("plotsSys_" + options.name + "/" + channel + binName + "/" +
sys + ".pdf")
for i, s in enumerate(back + sign):
c2 = TCanvas(s + "canvas", "Signals", 800, 600)
c2.cd()
gStyle.SetOptStat(0)
gStyle.SetOptTitle(0)
c2.GetPad(0).SetTopMargin(0.06)
c2.GetPad(0).SetRightMargin(0.06)
c2.GetPad(0).SetTicky(2)
c2.GetPad(0).SetLogy()
histUp[s].SetMaximum(histUp[s].GetMaximum() * 5)
histUp[s].Draw("HIST")
histDown[s].Draw("SAME, HIST")
hist[s].Draw("SAME, HIST")
drawCMS(-1, "Simulation", False)
if options.saveplots:
c2.Print("plotsSys_" + options.name + "/" + channel + binName +
"/" + sys + "_" + s + ".png")
c2.Print("plotsSys_" + options.name + "/" + channel + binName +
"/" + sys + "_" + s + ".pdf")
saveHist(histUp, channel + binName, sys + 'Up')
saveHist(histDown, channel + binName, sys + 'Down')
print "Added systematic", sys, "to channel", channel
def MakeLegend(can,
x1=.8,
y1=.8,
x2=.9,
y2=.9,
textsize=18,
ncolumns=1,
totalentries=0,
option='f',
skip=[]):
from ROOT import TLegend, TH1, gStyle, TGraph
if can.GetPrimitive('pad_top'):
MakeLegend(can.GetPrimitive('pad_top'),
x1,
y1,
x2,
y2,
textsize,
ncolumns,
totalentries,
skip=skip)
return
if CanvasEmpty(can):
print 'Error: trying to make legend from canvas with 0 plots. Will do nothing.'
return
#
# if a previous version exists from this function, delete it
#
if can.GetPrimitive('legend'):
can.GetPrimitive('legend').Delete()
leg = TLegend(x1, y1, x2, y2)
leg.SetName('legend')
tobject_collector.append(leg)
leg.SetTextFont(43)
leg.SetTextSize(textsize)
leg.SetTextFont(43)
leg.SetBorderSize(0)
leg.SetFillStyle(0)
leg.SetNColumns(ncolumns)
#
# Add by TH1 GetTitle()
#
the_primitives = can.GetListOfPrimitives()
if can.GetPrimitive('pad_top'):
the_primitives = can.GetPrimitive('pad_top').GetListOfPrimitives()
if can.GetPrimitive('stack'):
the_stack = list(reversed(list(can.GetPrimitive('stack').GetHists())))
the_primitives = the_stack + list(the_primitives)
if type(option) == type(''):
option = [option] * 100
total = 0
for i in the_primitives:
if i.GetName() == 'stack': continue
drawopt = i.GetDrawOption()
if issubclass(type(i), TH1) or issubclass(type(i), TGraph):
if i.GetTitle() in skip:
continue
leg.AddEntry(i, i.GetTitle(), option[total]) # plef
total += 1
#
# Add empty entries to ensure a standard layout
#
for i in range(100):
if totalentries == 0: break
if total >= totalentries: break
leg.AddEntry(None, '', '')
total += 1
# recipe for making roughly square boxes
h = leg.GetY2() - leg.GetY1()
w = leg.GetX2() - leg.GetX1()
leg.SetMargin(leg.GetNColumns() * h / float(leg.GetNRows() * w))
can.cd()
if can.GetPrimitive('pad_top'):
can.GetPrimitive('pad_top').cd()
leg.Draw()
can.Modified()
can.Update()
return
def plot(var, cut, nm1=False):
### Preliminary Operations ###
treeRead = True if not FILE else False # Read from tree
channel = cut
isBlind = BLIND
showSignal = False if 'SB' in cut or 'TR' in cut else True
# Determine explicit cut
if treeRead:
for k in sorted(alias.keys(), key=len, reverse=True):
if k in cut: cut = cut.replace(k, alias[k])
# Determine Primary Dataset
pd = []
if "isSingleMuonPhotonTrigger" in cut:
pd = [x for x in sample['data_obs']['files'] if "MuonEG" in x]
elif "isJPsiTrigger" in cut:
pd = [x for x in sample['data_obs']['files'] if "Charmonium" in x]
else:
print "Cannot determine Primary Dataset."
exit()
print "Plotting from", ("tree" if treeRead else
"file"), var, "in", channel, "channel with:"
print " dataset:", pd
print " cut :", cut
if isBlind and "SR" in channel and var in ["H_mass"]:
cut += " && ( isMC ? 1 : !(H_mass > 86 && H_mass < 96) && !(H_mass > 120 && H_mass < 130) )"
### Create and fill MC histograms ###
# Create dict
file = {}
tree = {}
hist = {}
cutstring = "(eventWeightLumi)" + ("*(" + cut +
")" if len(cut) > 0 else "")
### Create and fill MC histograms ###
for i, s in enumerate(data + back + sign):
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events;" +
('logx' if variable[var]['logx'] else '') +
('logy' if variable[var]['logy'] else ''), variable[var]['nbins'],
variable[var]['min'], variable[var]['max'])
hist[s].Sumw2()
tree[s] = TChain("Events")
for j, ss in enumerate(sample[s]['files']):
if s in data and not ss in pd: continue
if YEAR == 2016 and not ('Run2016' in ss or 'Summer16' in ss):
continue
if YEAR == 2017 and not ('Run2017' in ss or 'Fall17' in ss):
continue
if YEAR == 2018 and not ('Run2018' in ss or 'Autumn18' in ss):
continue
for f in os.listdir(NTUPLEDIR + '/' + ss):
tree[s].Add(NTUPLEDIR + '/' + ss + '/' + f)
tree[s].Project(s, var, cutstring)
if not tree[s].GetTree() == None:
hist[s].SetOption("%s" % tree[s].GetTree().GetEntriesFast())
# jobs = []
# queue = multiprocessing.Queue()
# for i, s in enumerate(data+back+sign):
# for j, ss in enumerate(sample[s]['files']):
# if s in data and not ss in pd: continue
# if YEAR == 2016 and not ('Run2016' in ss or 'Summer16' in ss): continue
# if YEAR == 2017 and not ('Run2017' in ss or 'Fall17' in ss): continue
# if YEAR == 2018 and not ('Run2018' in ss or 'Autumn18' in ss): continue
# if treeRead: # Project from tree
## hist[s] = loopProject(s, ss, variable[var], cutstring, True)
# p = multiprocessing.Process(target=parallelProject, args=(queue, s, ss, variable[var], cutstring, ))
# jobs.append(p)
# p.start()
# else: # Histogram written to file
# hist[s] = readhist(FILE, s, var, cut)
#
# # Wait for all jobs to finish
# for job in jobs:
# h = queue.get()
# if not h.GetOption() in hist: hist[h.GetOption()] = h
# else: hist[h.GetOption()].Add(h)
# for job in jobs:
# job.join()
# Histogram style
for i, s in enumerate(data + back + sign):
hist[s].Scale(sample[s]['weight'] if hist[s].Integral() >= 0 else 0)
hist[s].SetFillColor(sample[s]['fillcolor'])
hist[s].SetFillStyle(sample[s]['fillstyle'] if not options.norm else 0)
hist[s].SetLineColor(sample[s]['linecolor'])
hist[s].SetLineStyle(sample[s]['linestyle'])
hist[s].SetLineWidth(sample[s]['linewidth'])
### Create Bkg Sum histogram ###
hist['BkgSum'] = hist['data_obs'].Clone(
"BkgSum") if 'data_obs' in hist else hist[back[0]].Clone("BkgSum")
hist['BkgSum'].Reset("MICES")
hist['BkgSum'].SetFillStyle(3003)
hist['BkgSum'].SetFillColor(1)
for i, s in enumerate(back):
hist['BkgSum'].Add(hist[s])
if options.norm:
for i, s in enumerate(back + ['BkgSum']):
hist[s].Scale(hist[data[0]].Integral() / hist['BkgSum'].Integral())
for i, s in enumerate(sign):
hist[s].Scale(hist[data[0]].Integral() / hist[s].Integral())
# Create data and Bkg sum histograms
# if BLIND: # or 'SR' in channel:
# hist['data_obs'] = hist['BkgSum'].Clone("data_obs")
# hist['data_obs'].Reset("MICES")
# Set histogram style
hist['data_obs'].SetMarkerStyle(20)
hist['data_obs'].SetMarkerSize(1.25)
# for i, s in enumerate(data+back+sign+['BkgSum']): addOverflow(hist[s], False) # Add overflow
for i, s in enumerate(sign):
hist[s].SetLineWidth(3)
for i, s in enumerate(sign):
sample[s]['plot'] = True
# Create stack
bkg = THStack("Bkg",
";" + hist['BkgSum'].GetXaxis().GetTitle() + ";Events")
for i, s in enumerate(back):
bkg.Add(hist[s])
# Legend
leg = TLegend(0.65, 0.6, 0.95, 0.9)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
if len(data) > 0:
leg.AddEntry(hist[data[0]], sample[data[0]]['label'], "pe")
for i, s in reversed(list(enumerate(['BkgSum'] + back))):
leg.AddEntry(hist[s], sample[s]['label'], "f")
if showSignal:
for i, s in enumerate(sign):
if sample[s]['plot']:
leg.AddEntry(hist[s], sample[s]['label'], "fl")
leg.SetY1(0.9 - leg.GetNRows() * 0.04)
# --- Display ---
c1 = TCanvas("c1",
hist.values()[0].GetXaxis().GetTitle(), 800,
800 if RATIO else 600)
if RATIO:
c1.Divide(1, 2)
setTopPad(c1.GetPad(1), RATIO)
setBotPad(c1.GetPad(2), RATIO)
c1.cd(1)
c1.GetPad(bool(RATIO)).SetTopMargin(0.06)
c1.GetPad(bool(RATIO)).SetRightMargin(0.05)
c1.GetPad(bool(RATIO)).SetTicks(1, 1)
logX, logY = "logx" in hist['BkgSum'].GetZaxis().GetTitle(
), "logy" in hist['BkgSum'].GetZaxis().GetTitle()
if logY: c1.GetPad(bool(RATIO)).SetLogy()
if logX: c1.GetPad(bool(RATIO)).SetLogx()
# Draw
bkg.Draw("HIST") # stack
hist['BkgSum'].Draw("SAME, E2") # sum of bkg
if len(data) > 0: hist['data_obs'].Draw("SAME, PE") # data
#data_graph.Draw("SAME, PE")
# if showSignal:
# smagn = 1. #if treeRead else 1.e2 #if logY else 1.e2
for i, s in enumerate(sign):
if sample[s]['plot']: hist[s].Draw("SAME, HIST")
# hist[s].Scale(smagn)
# hist[s].Draw("SAME, HIST") # signals Normalized, hist[s].Integral()*sample[s]['weight']
# #textS = drawText(0.80, 0.9-leg.GetNRows()*0.05 - 0.02, stype+" (x%d)" % smagn, True)
bkg.GetYaxis().SetTitleOffset(bkg.GetYaxis().GetTitleOffset() * 1.075)
bkg.SetMaximum((5. if logY else 1.25) * max(
bkg.GetMaximum(),
hist['data_obs'].GetBinContent(hist['data_obs'].GetMaximumBin()) +
hist['data_obs'].GetBinError(hist['data_obs'].GetMaximumBin())))
#if bkg.GetMaximum() < max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum()): bkg.SetMaximum(max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum())*1.25)
bkg.SetMinimum(
max(
min(hist['BkgSum'].GetBinContent(hist['BkgSum'].GetMinimumBin(
)), hist['data_obs'].GetMinimum()), 5.e-1) if logY else 0.)
if logY:
bkg.GetYaxis().SetNoExponent(bkg.GetMaximum() < 1.e4)
bkg.GetYaxis().SetMoreLogLabels(True)
#if logY: bkg.SetMinimum(1)
leg.Draw()
drawCMS(LUMI[YEAR], "Preliminary")
if channel in aliasNames: drawRegion(aliasNames[channel], True)
#drawAnalysis(channel)
#if nm1 and not cutValue is None: drawCut(cutValue, bkg.GetMinimum(), bkg.GetMaximum()) #FIXME
#if len(sign) > 0:
# if channel.startswith('X') and len(sign)>0: drawNorm(0.9-0.05*(leg.GetNRows()+1), "#sigma(X) = %.1f pb" % 1.)
setHistStyle(bkg, 1.2 if RATIO else 1.1)
setHistStyle(hist['BkgSum'], 1.2 if RATIO else 1.1)
if RATIO:
c1.cd(2)
if logX: c1.GetPad(2).SetLogx()
err = hist['BkgSum'].Clone("BkgErr;")
err.SetTitle("")
err.GetYaxis().SetTitle("Data / Bkg")
for i in range(1, err.GetNbinsX() + 1):
err.SetBinContent(i, 1)
if hist['BkgSum'].GetBinContent(i) > 0:
err.SetBinError(
i, hist['BkgSum'].GetBinError(i) /
hist['BkgSum'].GetBinContent(i))
setBotStyle(err)
errLine = err.Clone("errLine")
errLine.SetLineWidth(1)
errLine.SetFillStyle(0)
res = hist['data_obs'].Clone("Residues")
for i in range(0, res.GetNbinsX() + 1):
if hist['BkgSum'].GetBinContent(i) > 0:
res.SetBinContent(
i,
res.GetBinContent(i) / hist['BkgSum'].GetBinContent(i))
res.SetBinError(
i,
res.GetBinError(i) / hist['BkgSum'].GetBinContent(i))
setBotStyle(res)
#err.GetXaxis().SetLabelOffset(err.GetXaxis().GetLabelOffset()*5)
#err.GetXaxis().SetTitleOffset(err.GetXaxis().GetTitleOffset()*2)
err.Draw("E2")
errLine.Draw("SAME, HIST")
if len(data) > 0:
res.Draw("SAME, PE0")
#res_graph.Draw("SAME, PE0")
if len(err.GetXaxis().GetBinLabel(
1)) == 0: # Bin labels: not a ordinary plot
drawRatio(hist['data_obs'], hist['BkgSum'])
drawStat(hist['data_obs'], hist['BkgSum'])
if var in ["H_mass"]:
c1.cd(bool(RATIO))
boxZ = drawBox(XZMIN, hist['data_obs'].GetMinimum(), XZMAX,
hist['data_obs'].GetMaximum() / 1.30, "Z")
boxH = drawBox(XHMIN, hist['data_obs'].GetMinimum(), XHMAX,
hist['data_obs'].GetMaximum() / 1.30, "H")
c1.Update()
if True: #gROOT.IsBatch():
varname = var.replace('.', '_').replace('()', '')
if not os.path.exists("plots/" + channel):
os.makedirs("plots/" + channel)
c1.Print("plots/" + channel + "/" + varname + ".png")
c1.Print("plots/" + channel + "/" + varname + ".pdf")
# Print table
printTable(hist, sign)
if not gROOT.IsBatch(): raw_input("Press Enter to continue...")
def plot(var, cut, norm=False, nm1=False):
### Preliminary Operations ###
doBinned = False
if options.mode == "binned": doBinned = True
fileRead = os.path.exists("combinedCards_" + options.name +
"/fitDiagnostics_" + options.file + ".root")
treeRead = not any(
x == cut
for x in ['0l', '1e', '1m', '2e', '2m', '1e1m', 'Gen', 'Trigger'
]) #(var in variable.keys()) # Read from tree
#signal definition
if fileRead:
sign = ['ttDM_MChi1_MPhi200_scalar',
'tDM_MChi1_MPhi200_scalar'] #for postfit plot
if not fileRead and not options.limit:
sign = ['ttDM_MChi1_MPhi100_scalar',
'tDM_MChi1_MPhi100_scalar'] #for normal plotting
#bkg definition
if fileRead or options.limit:
back = [
"QCD", "DYJetsToNuNu_HT", "DYJetsToLL_HT", "VV", "ST",
"WJetsToLNu_HT", "TTbarSL"
] #for postfit or limit
if (cut).find('>250') or (cut).startswith('AH'): #for hadronic selections
back = [
"QCD", "DYJetsToLL_HT", "VV", "ST", "WJetsToLNu_HT", "TTbarV",
"TTbar2L", "TTbar1L", "DYJetsToNuNu_HT"
]
if fileRead or options.limit:
back = [
"QCD", "DYJetsToLL_HT", "VV", "ST", "WJetsToLNu_HT", "TTbarSL",
"DYJetsToNuNu_HT"
] #for postfit or limit
binLow = ""
binHigh = ""
binName = ""
if "binned" in cut:
binLow = cut[cut.find("LowVal") + 6:cut.find("HighVal") - 1]
binHigh = cut[cut.find("HighVal") + 7:]
binName = "bin_" + binLow + "_" + binHigh
cut = cut[:cut.find("binned")]
useformula = False
if 'formula' in variable[var]:
useformula = True
channel = cut
plotdir = cut
plotname = var
weight = "eventWeightLumi" #*(2.2/35.9)
isBlind = BLIND and ('SR' in channel or 'ps' in channel)
if fileRead:
isBlind = False
options.saveplots = True
RESIDUAL = True
elif isBlind:
RATIO = 0
SIGNAL = 20
else:
RATIO = 4
SIGNAL = 1
RESIDUAL = False
showSignal = True #('SR' in channel)
cutSplit = cut.split()
for s in cutSplit:
if s in selection.keys():
plotdir = s
cut = cut.replace(s, selection[s])
if not binLow == "":
cut = cut + " && " + var + " > " + binLow + " && " + var + " < " + binHigh
#if treeRead and cut in selection: cut = cut.replace(cut, selection[cut])
# Determine Primary Dataset
pd = []
if any(w in cut
for w in ['1l', '1m', '2m', 'isWtoMN', 'isZtoMM', 'isTtoEM']):
pd += [x for x in sample['data_obs']['files'] if 'SingleMuon' in x]
if any(w in cut for w in ['1l', '1e', '2e', 'isWtoEN', 'isZtoEE']):
pd += [x for x in sample['data_obs']['files'] if 'SingleElectron' in x]
if any(w in cut for w in ['0l', 'isZtoNN']):
pd += [x for x in sample['data_obs']['files'] if 'MET' in x]
if len(pd) == 0:
raw_input("Warning: Primary Dataset not recognized, continue?")
print "Plotting from", ("tree" if treeRead else
"file"), var, "in", channel, "channel with:"
print " dataset:", pd
print " cut :", cut
print " cut :", weight
### Create and fill MC histograms ###
# Create dict
file = {}
tree = {}
hist = {}
### Create and fill MC histograms ###
for i, s in enumerate(data + back + sign):
if fileRead:
var = 'MET_pt'
if channel.startswith('SL'): var = 'MET_sign'
if channel.endswith('ZR'): var = 'FakeMET_pt'
plotname = var
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events;" +
('log' if variable[var]['log'] else ''),
variable[var]['nbins'], variable[var]['min'],
variable[var]['max'])
if doBinned:
bins = np.array([])
if 'bins' in variable[var].keys():
bins = np.array(variable[var]['bins'])
else:
binsize = (variable[var]['max'] -
variable[var]['min']) / variable[var]['nbins']
bins = np.arange(variable[var]['min'],
variable[var]['max'] + binsize, binsize)
bins = np.append(bins, 10000)
for i in range(0, len(bins) - 1):
rbin = str(bins[i]) + "_" + str(bins[i + 1])
fileName = "combinedCards_" + options.name + "/fitDiagnostics_" + options.file + ".root" if not any(
t in s for t in ['data', 'tDM']
) else "rootfiles_" + options.name + "/" + channel + "bin_" + rbin + ".root"
histName = "shapes_fit_b/" + channel + "bin_" + rbin + "/" + s if not any(
t in s for t in ['data', 'tDM']) else s
file[s] = TFile(fileName, "READ")
tmphist = file[s].Get(histName)
if 'data' not in s: hist[s].SetMarkerSize(0)
if tmphist:
hist[s].SetBinContent(i + 1, tmphist.GetBinContent(1))
hist[s].SetBinError(i + 1, tmphist.GetBinError(1))
else:
hist[s].SetBinContent(i + 1, 0.)
hist[s].SetBinError(i + 1, 0.)
else:
fileName = "combinedCards_" + options.name + "/fitDiagnostics_" + options.file + ".root" if not s == 'data_obs' else "rootfiles_" + options.name + "/" + channel + binName + ".root"
histName = "shapes_fit_b/" + channel + "/" + s if not s == 'data_obs' else s
file[s] = TFile(fileName, "READ")
tmphist = file[s].Get(histName)
if tmphist == None:
tmphist = hist[back[0]].Clone(s)
tmphist.Reset("MICES")
print "Histogram", histName, "not found in file", fileName
if s == 'data_obs': hist[s] = tmphist
else:
hist[s] = hist['data_obs'].Clone(s)
hist[s].SetMarkerSize(0)
for i in range(tmphist.GetNbinsX() + 1):
hist[s].SetBinContent(i + 1,
tmphist.GetBinContent(i + 1))
elif treeRead: # Project from tree
tree[s] = TChain("tree")
for j, ss in enumerate(sample[s]['files']):
if not 'data' in s or ('data' in s and ss in pd):
tree[s].Add(NTUPLEDIR + ss + ".root")
if not binLow == "":
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events;" +
('log' if variable[var]['log'] else ''), 1, float(binLow),
float(binHigh))
elif binLow == "" and variable[var]['nbins'] > 0:
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events;" +
('log' if variable[var]['log'] else ''),
variable[var]['nbins'], variable[var]['min'],
variable[var]['max'])
else:
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events;" +
('log' if variable[var]['log'] else ''),
len(variable[var]['bins']) - 1,
array('f', variable[var]['bins']))
hist[s].Sumw2()
redFactorString = ""
redFactorValue = ""
#if isBlind and 'data' in s:
if isBlind and 'data' in s and options.limit:
redFactorString = " && Entry$ % 15 == 1"
#if isBlind and 'data' not in s:
if isBlind and 'data' not in s and options.limit:
redFactorValue = " / 15"
cutstring = "(" + weight + redFactorValue + ")" + (
"*(" + cut + redFactorString + ")" if len(cut) > 0 else "")
if '-' in s:
cutstring = cutstring.replace(
cut, cut + "&& nBQuarks==" + s.split('-')[1][0])
if useformula == True:
tree[s].Project(s, variable[var]['formula'], cutstring)
else:
tree[s].Project(s, var, cutstring)
if not tree[s].GetTree() == None:
hist[s].SetOption("%s" % tree[s].GetTree().GetEntriesFast())
else: # Histogram written to file
for j, ss in enumerate(sample[s]['files']):
if not 'data' in s or ('data' in s and ss in pd):
file[ss] = TFile(NTUPLEDIR + ss + ".root", "R")
if file[ss].IsZombie():
print "WARNING: file", NTUPLEDIR + ss + ".root", "does not exist"
continue
tmphist = file[ss].Get(cut + "/" + var)
if tmphist == None: continue
if not s in hist.keys(): hist[s] = tmphist
else: hist[s].Add(tmphist)
if hist[s].Integral() < 0: hist[s].Scale(0)
hist[s].SetFillColor(sample[s]['fillcolor'])
hist[s].SetFillStyle(sample[s]['fillstyle'])
hist[s].SetLineColor(sample[s]['linecolor'])
hist[s].SetLineStyle(sample[s]['linestyle'])
#if 'WJetsToLNu' in s and 'SL' in channel and 'WR' in channel: hist[s].Scale(1.30)
#if 'TTbar' in s and 'SL' in channel and 'TR' in channel: hist[s].Scale(0.91)
hist['BkgSum'] = hist[back[0]].Clone("BkgSum")
hist['BkgSum'].Reset("MICES")
for i, s in enumerate(back):
hist['BkgSum'].Add(hist[s], 1)
if fileRead:
hist['PreFit'] = hist['BkgSum'].Clone("PreFit")
if doBinned:
for i in range(0, len(bins) - 1):
rbin = str(bins[i]) + "_" + str(bins[i + 1])
tmphist = file[back[0]].Get("shapes_prefit/" + channel +
"bin_" + rbin + "/" +
"total_background")
if tmphist:
hist['PreFit'].SetBinContent(i + 1,
tmphist.GetBinContent(1))
else:
hist['PreFit'].SetBinContent(i + 1, 0.)
else:
tmphist = file[back[0]].Get("shapes_prefit/" + channel + "/" +
"total_background")
for i in range(tmphist.GetNbinsX() + 1):
hist['PreFit'].SetBinContent(i + 1,
tmphist.GetBinContent(i + 1))
addOverflow(hist['PreFit'], False)
hist['PreFit'].SetLineStyle(2)
hist['PreFit'].SetLineColor(617) #923
hist['PreFit'].SetLineWidth(3)
hist['PreFit'].SetFillStyle(0)
hist['BkgSum'].SetFillStyle(3002)
hist['BkgSum'].SetFillColor(1)
# Create data and Bkg sum histograms
# if options.blind or 'SR' in channel:
# hist['data_obs'] = hist['BkgSum'].Clone("data_obs")
# hist['data_obs'].Reset("MICES")
# Set histogram style
hist[data[0]].SetMarkerStyle(20)
hist[data[0]].SetMarkerSize(1.25)
for i, s in enumerate(data + back + sign + ['BkgSum']):
addOverflow(hist[s], False) # Add overflow
for i, s in enumerate(sign):
hist[s].SetLineWidth(3)
#for i, s in enumerate(sign): sample[s]['plot'] = True#sample[s]['plot'] and s.startswith(channel[:2])
if norm:
for i, s in enumerate(sign):
hist[s].Scale(hist['BkgSum'].Integral() / hist[s].Integral())
# for i, s in enumerate(back):
# hist[s].SetFillStyle(3005)
# hist[s].SetLineWidth(2)
# #for i, s in enumerate(sign):
# # hist[s].SetFillStyle(0)
# if not var=="Events":
# sfnorm = hist[data[0]].Integral()/hist['BkgSum'].Integral()
# print "Applying SF:", sfnorm
# for i, s in enumerate(back+['BkgSum']): hist[s].Scale(sfnorm)
if SIGNAL > 1:
if not var == "Events":
for i, s in enumerate(sign):
hist[s].Scale(SIGNAL)
# Create stack
bkg = THStack("Bkg",
";" + hist['BkgSum'].GetXaxis().GetTitle() + ";Events")
for i, s in enumerate(back):
bkg.Add(hist[s])
# Legend
#leg = TLegend(0.65, 0.6, 0.95, 0.9)
leg = TLegend(0.45, 0.63, 0.93, 0.92)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
leg.SetNColumns(3)
leg.SetTextFont(42)
if len(data) > 0:
leg.AddEntry(hist[data[0]], sample[data[0]]['label'], "pe")
for i, s in reversed(list(enumerate(back))):
leg.AddEntry(hist[s], sample[s]['label'], "f")
if 'PreFit' not in hist:
leg.AddEntry(hist['BkgSum'], sample['BkgSum']['label'], "f")
else:
leg.AddEntry(hist['BkgSum'], 'MC unc.', "l")
leg.AddEntry(hist['PreFit'], sample['PreFit']['label'], "l")
if showSignal:
for i, s in enumerate(sign):
if SIGNAL > 1:
if sample[s]['plot']:
leg.AddEntry(hist[s],
'%s (x%d)' % (sample[s]['label'], SIGNAL),
"l")
else:
if sample[s]['plot']:
leg.AddEntry(hist[s], sample[s]['label'], "l")
leg.SetY1(0.9 - leg.GetNRows() * 0.05)
# --- Display ---
c1 = TCanvas("c1",
hist.values()[0].GetXaxis().GetTitle(), 800,
800 if RATIO else 600)
if RATIO:
if RESIDUAL:
c1.Divide(1, 3)
setFitTopPad(c1.GetPad(1), RATIO)
setFitBotPad(c1.GetPad(2), RATIO)
setFitResPad(c1.GetPad(3), RATIO)
else:
c1.Divide(1, 2)
setTopPad(c1.GetPad(1), RATIO)
setBotPad(c1.GetPad(2), RATIO)
c1.cd(1)
c1.GetPad(bool(RATIO)).SetTopMargin(0.06)
c1.GetPad(bool(RATIO)).SetRightMargin(0.05)
c1.GetPad(bool(RATIO)).SetTicks(1, 1)
log = ("log" in hist['BkgSum'].GetZaxis().GetTitle())
if log: c1.GetPad(bool(RATIO)).SetLogy()
# Draw
bkg.Draw("HIST") # stack
hist['BkgSum'].Draw("SAME, E2") # sum of bkg
if not isBlind and len(data) > 0:
graph = fixData(hist[data[0]], USEGARWOOD)
graph.Draw("SAME, PE")
#data_graph.Draw("SAME, PE")
if 'PreFit' in hist: hist['PreFit'].Draw("SAME, HIST")
if showSignal:
for i, s in enumerate(sign):
if sample[s]['plot']:
hist[s].Draw(
"SAME, HIST"
) # signals Normalized, hist[s].Integral()*sample[s]['weight']
bkg.GetYaxis().SetTitleOffset(bkg.GetYaxis().GetTitleOffset() * 1.075)
bkg.SetMaximum((5. if log else 1.25) * max(
bkg.GetMaximum(),
hist[data[0]].GetBinContent(hist[data[0]].GetMaximumBin()) +
hist[data[0]].GetBinError(hist[data[0]].GetMaximumBin())))
if len(sign) > 0 and bkg.GetMaximum() < max(hist[sign[0]].GetMaximum(),
hist[sign[-1]].GetMaximum()):
bkg.SetMaximum(
max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum()) *
1.25)
bkg.SetMinimum(
max(
min(hist['BkgSum'].GetBinContent(hist['BkgSum'].GetMinimumBin(
)), hist[data[0]].GetMinimum()), 5.e-1) if log else 0.)
if log:
#bkg.GetYaxis().SetNoExponent(bkg.GetMaximum() < 1.e4)
bkg.GetYaxis().SetNoExponent(bkg.GetMaximum() < 1.e3)
bkg.GetYaxis().SetMoreLogLabels(True)
else:
bkg.GetYaxis().SetNoExponent(bkg.GetMaximum() < 1.e3)
leg.Draw()
if fileRead and 'SR' in channel:
drawCMS(LUMI / 15., "Preliminary")
else:
drawCMS(LUMI, "Preliminary")
drawRegion(channel, True)
drawAnalysis("DM" + channel[:2])
drawOverflow()
setHistStyle(bkg, 1.2 if RATIO else 1.1)
setHistStyle(hist['BkgSum'], 1.2 if RATIO else 1.1)
if RATIO:
c1.cd(2)
err = hist['BkgSum'].Clone("BkgErr;")
err.SetTitle("")
err.GetYaxis().SetTitle("Data / Bkg")
for i in range(1, err.GetNbinsX() + 1):
err.SetBinContent(i, 1)
if hist['BkgSum'].GetBinContent(i) > 0:
err.SetBinError(
i, hist['BkgSum'].GetBinError(i) /
hist['BkgSum'].GetBinContent(i))
if RESIDUAL: setFitBotStyle(err)
else: setBotStyle(err)
errLine = err.Clone("errLine")
errLine.SetLineWidth(1)
errLine.SetFillStyle(0)
res = hist[data[0]].Clone("Residues")
for i in range(0, res.GetNbinsX() + 1):
if hist['BkgSum'].GetBinContent(i) > 0:
res.SetBinContent(
i,
res.GetBinContent(i) / hist['BkgSum'].GetBinContent(i))
res.SetBinError(
i,
res.GetBinError(i) / hist['BkgSum'].GetBinContent(i))
if RESIDUAL: setFitBotStyle(res)
else: setBotStyle(res)
#err.GetXaxis().SetLabelOffset(err.GetXaxis().GetLabelOffset()*5)
#err.GetXaxis().SetTitleOffset(err.GetXaxis().GetTitleOffset()*2)
err.Draw("E2")
if 'PreFit' in hist:
respre = hist[data[0]].Clone("ResiduesPreFit")
respre.Divide(hist['PreFit'])
respre.SetLineStyle(2)
respre.SetLineColor(617) #923
respre.SetLineWidth(3)
respre.SetFillStyle(0)
respre.Draw("SAME, HIST")
errLine.Draw("SAME, HIST")
if not isBlind and len(data) > 0:
res.Draw("SAME, PE0")
#res_graph.Draw("SAME, PE0")
if len(err.GetXaxis().GetBinLabel(
1)) == 0: # Bin labels: not a ordinary plot
drawRatio(hist['data_obs'], hist['BkgSum'])
drawStat(hist['data_obs'], hist['BkgSum'])
c1.Update()
if RATIO and RESIDUAL:
c1.cd(3)
c1.SetGrid(1, 0)
resFit = hist[data[0]].Clone("Residues")
resFit.Reset("MICES")
resFit.SetTitle("")
#resFit.GetYaxis().SetTitle("Residuals")
resFit.GetYaxis().SetTitle(
"#frac{Data - Bkg}{#sqrt{#sigma_{Data}^{2}+#sigma_{Bkg}^{2}}}")
for i in range(0, res.GetNbinsX() + 1):
if hist['BkgSum'].GetBinContent(i) > 0:
resFit.SetBinContent(
i, (hist[data[0]].GetBinContent(i) -
hist['BkgSum'].GetBinContent(i)) / (math.sqrt(
math.pow(hist['BkgSum'].GetBinError(i), 2) +
math.pow(hist[data[0]].GetBinError(i), 2))))
setFitResStyle(resFit)
resFit.SetLineColor(15)
resFit.SetFillColor(15)
resFit.SetFillStyle(1001)
resFit.Draw("HIST")
resFitLine = resFit.Clone("resFitLine")
resFitLine.SetLineWidth(1)
resFitLine.SetFillStyle(0)
resFitLine.Draw("SAME, HIST")
c1.Update()
if gROOT.IsBatch(
) and options.saveplots: # and (treeRead and channel in selection.keys()):
AddString = ""
if not os.path.exists("plots_" + options.name + "/" + plotdir):
os.makedirs("plots_" + options.name + "/" + plotdir)
if fileRead:
if RESIDUAL: AddString = "_PostFit_Residual"
else: AddString = "_PostFit"
#c1.Print("plots_"+options.name+"/"+plotdir+"/"+plotname+binName+AddString+".png")
c1.Print("plots_" + options.name + "/" + plotdir + "/" + plotname +
binName + AddString + ".pdf")
# Print table
printTable(hist, sign)
if not gROOT.IsBatch(): raw_input("Press Enter to continue...")
if gROOT.IsBatch() and not fileRead and (
var == 'MET_pt' or
(channel.startswith('SL') and var == 'MET_sign') or
(channel.endswith('ZR') and var == 'FakeMET_pt')):
saveHist(hist, channel + binName)
def plot(var, cut, norm=False, nm1=False):
### Preliminary Operations ###
fileRead = os.path.exists(options.file)
treeRead = not any(x==cut for x in ['0l', '1e', '1m', '2e', '2m', '1e1m', 'Gen', 'Trigger'])#(var in variable.keys()) # Read from tree
binLow = ""
binHigh = ""
binName = ""
if "binned" in cut:
binLow = cut[cut.find("LowVal")+6:cut.find("HighVal")-1]
binHigh = cut[cut.find("HighVal")+7:]
binName = "bin_"+binLow+"_"+binHigh
cut = cut[:cut.find("binned")]
useformula = True
if 'formula' in variable[var]:
print variable[var]['formula']
useformula = True
channel = cut
plotdir = cut
plotname = var
weight = "eventWeightLumi" #*(2.2/35.9)
isBlind = BLIND and 'SR' in channel
showSignal = True#('SR' in channel)
cutSplit = cut.split()
for s in cutSplit:
if s in selection.keys():
plotdir = s
cut = cut.replace(s, selection[s])
if not binLow == "":
cut = cut + " && " + var + " > " + binLow + " && " + var + " < " + binHigh
#if treeRead and cut in selection: cut = cut.replace(cut, selection[cut])
# Determine Primary Dataset
pd = []
print cut
if any(w in cut for w in ['1l', '1m', '2m', 'isWtoMN', 'isZtoMM', 'isTtoEM']): pd += [x for x in sample['data_obs']['files'] if 'SingleMuon' in x]
if any(w in cut for w in ['1l', '1e', '2e', 'isWtoEN', 'isZtoEE']): pd += [x for x in sample['data_obs']['files'] if 'SingleElectron' in x]
if any(w in cut for w in ['0l', 'isZtoNN']): pd += [x for x in sample['data_obs']['files'] if 'MET' in x]
if len(pd)==0: raw_input("Warning: Primary Dataset not recognized, continue?")
print "Plotting from", ("tree" if treeRead else "file"), var, "in", channel, "channel with:"
print " dataset:", pd
print " cut :", cut
print " cut :", weight
### Create and fill MC histograms ###
# Create dict
file = {}
tree = {}
hist = {}
### Create and fill MC histograms ###
for i, s in enumerate(data+back+sign):
if fileRead:
fileName = options.file if not s=='data_obs' else "rootfiles_"+options.name+"/"+channel+binName+".root"
histName = "shapes_fit_b/"+channel+"/"+s if not s=='data_obs' else s
file[s] = TFile(fileName, "READ")
tmphist = file[s].Get(histName)
if tmphist==None:
tmphist = hist[back[0]].Clone(s)
tmphist.Reset("MICES")
print "Histogram", histName, "not found in file", fileName
if s=='data_obs': hist[s] = tmphist
else:
hist[s] = hist['data_obs'].Clone(s)
#hist[s].Reset("MICES")
hist[s].SetMarkerSize(0)
for i in range(tmphist.GetNbinsX()+1): hist[s].SetBinContent(i+1, tmphist.GetBinContent(i+1))
elif treeRead: # Project from tree
tree[s] = TChain("tree")
for j, ss in enumerate(sample[s]['files']):
if not 'data' in s or ('data' in s and ss in pd):
tree[s].Add(NTUPLEDIR + ss + ".root")
if not binLow == "":
hist[s] = TH1F(s, ";"+variable[var]['title']+";Events;"+('log' if variable[var]['log'] else ''), 1, float(binLow), float(binHigh))
elif binLow == "" and variable[var]['nbins']>0:
hist[s] = TH1F(s, ";"+variable[var]['title']+";Events;"+('log' if variable[var]['log'] else ''), variable[var]['nbins'], variable[var]['min'], variable[var]['max'])
else:
hist[s] = TH1F(s, ";"+variable[var]['title']+";Events;"+('log' if variable[var]['log'] else ''), len(variable[var]['bins'])-1, array('f', variable[var]['bins']))
hist[s].Sumw2()
redFactorString = ""
redFactorValue = ""
if isBlind and 'data' in s:
redFactorString = " && Entry$ % 15 == 0"
if isBlind and 'data' not in s:
redFactorValue = " / 15"
cutstring = "("+weight+redFactorValue+")" + ("*("+cut+redFactorString+")" if len(cut)>0 else "")
if '-' in s: cutstring = cutstring.replace(cut, cut + "&& nBQuarks==" + s.split('-')[1][0])
if useformula == True:
tree[s].Project(s, variable[var]['formula'], cutstring)
else:
tree[s].Project(s, var, cutstring)
if not tree[s].GetTree()==None: hist[s].SetOption("%s" % tree[s].GetTree().GetEntriesFast())
else: # Histogram written to file
for j, ss in enumerate(sample[s]['files']):
if not 'data' in s or ('data' in s and ss in pd):
file[ss] = TFile(NTUPLEDIR + ss + ".root", "R")
if file[ss].IsZombie():
print "WARNING: file", NTUPLEDIR + ss + ".root", "does not exist"
continue
tmphist = file[ss].Get(cut+"/"+var)
if tmphist==None: continue
if not s in hist.keys(): hist[s] = tmphist
else: hist[s].Add(tmphist)
if hist[s].Integral() < 0: hist[s].Scale(0)
hist[s].SetFillColor(sample[s]['fillcolor'])
hist[s].SetFillStyle(sample[s]['fillstyle'])
hist[s].SetLineColor(sample[s]['linecolor'])
hist[s].SetLineStyle(sample[s]['linestyle'])
#if 'WJetsToLNu' in s and 'SL' in channel and 'WR' in channel: hist[s].Scale(1.30)
#if 'TTbar' in s and 'SL' in channel and 'TR' in channel: hist[s].Scale(0.91)
hist['BkgSum'] = hist[back[0]].Clone("BkgSum")
hist['BkgSum'].Reset("MICES")
for i, s in enumerate(back): hist['BkgSum'].Add(hist[s], 1)
if fileRead:
#hist['BkgSum'] = file[back[0]].Get("shapes_fit_b/"+channel+"/"+"total_background")
tmphist = file[back[0]].Get("shapes_prefit/"+channel+"/"+"total_background")
hist['PreFit'] = hist['BkgSum'].Clone("PreFit")
for i in range(tmphist.GetNbinsX()+1): hist['PreFit'].SetBinContent(i+1, tmphist.GetBinContent(i+1))
hist['PreFit'].SetLineStyle(2)
hist['PreFit'].SetLineColor(923)
hist['PreFit'].SetLineWidth(3)
hist['PreFit'].SetFillStyle(0)
hist['BkgSum'].SetFillStyle(3003)
hist['BkgSum'].SetFillColor(1)
# Create data and Bkg sum histograms
# if options.blind or 'SR' in channel:
# hist['data_obs'] = hist['BkgSum'].Clone("data_obs")
# hist['data_obs'].Reset("MICES")
# Set histogram style
hist[data[0]].SetMarkerStyle(20)
hist[data[0]].SetMarkerSize(1.25)
for i, s in enumerate(data+back+sign+['BkgSum']): addOverflow(hist[s], False) # Add overflow
for i, s in enumerate(sign): hist[s].SetLineWidth(3)
#for i, s in enumerate(sign): sample[s]['plot'] = True#sample[s]['plot'] and s.startswith(channel[:2])
if norm:
for i, s in enumerate(sign):
hist[s].Scale(hist['BkgSum'].Integral()/hist[s].Integral())
# for i, s in enumerate(back):
# hist[s].SetFillStyle(3005)
# hist[s].SetLineWidth(2)
# #for i, s in enumerate(sign):
# # hist[s].SetFillStyle(0)
# if not var=="Events":
# sfnorm = hist[data[0]].Integral()/hist['BkgSum'].Integral()
# print "Applying SF:", sfnorm
# for i, s in enumerate(back+['BkgSum']): hist[s].Scale(sfnorm)
# Create stack
bkg = THStack("Bkg", ";"+hist['BkgSum'].GetXaxis().GetTitle()+";Events")
for i, s in enumerate(back): bkg.Add(hist[s])
# Legend
leg = TLegend(0.65, 0.6, 0.95, 0.9)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
if len(data) > 0:
leg.AddEntry(hist[data[0]], sample[data[0]]['label'], "pe")
for i, s in reversed(list(enumerate(['BkgSum']+back))):
leg.AddEntry(hist[s], sample[s]['label'], "f")
if 'PreFit' in hist: leg.AddEntry(hist['PreFit'], sample['PreFit']['label'], "l")
if showSignal:
for i, s in enumerate(sign):
if sample[s]['plot']: leg.AddEntry(hist[s], sample[s]['label'], "fl")
leg.SetY1(0.9-leg.GetNRows()*0.05)
# --- Display ---
c1 = TCanvas("c1", hist.values()[0].GetXaxis().GetTitle(), 800, 800 if RATIO else 600)
if RATIO:
c1.Divide(1, 2)
setTopPad(c1.GetPad(1), RATIO)
setBotPad(c1.GetPad(2), RATIO)
c1.cd(1)
c1.GetPad(bool(RATIO)).SetTopMargin(0.06)
c1.GetPad(bool(RATIO)).SetRightMargin(0.05)
c1.GetPad(bool(RATIO)).SetTicks(1, 1)
log = ("log" in hist['BkgSum'].GetZaxis().GetTitle())
if log: c1.GetPad(bool(RATIO)).SetLogy()
# Draw
bkg.Draw("HIST") # stack
hist['BkgSum'].Draw("SAME, E2") # sum of bkg
if not isBlind and len(data) > 0: hist[data[0]].Draw("SAME, PE") # data
#data_graph.Draw("SAME, PE")
if 'PreFit' in hist: hist['PreFit'].Draw("SAME, HIST")
if showSignal:
for i, s in enumerate(sign):
if sample[s]['plot']: hist[s].Draw("SAME, HIST") # signals Normalized, hist[s].Integral()*sample[s]['weight']
bkg.GetYaxis().SetTitleOffset(bkg.GetYaxis().GetTitleOffset()*1.075)
bkg.SetMaximum((5. if log else 1.25)*max(bkg.GetMaximum(), hist[data[0]].GetBinContent(hist[data[0]].GetMaximumBin())+hist[data[0]].GetBinError(hist[data[0]].GetMaximumBin())))
if len(sign) > 0 and bkg.GetMaximum() < max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum()): bkg.SetMaximum(max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum())*1.25)
bkg.SetMinimum(max(min(hist['BkgSum'].GetBinContent(hist['BkgSum'].GetMinimumBin()), hist[data[0]].GetMinimum()), 5.e-1) if log else 0.)
if log:
bkg.GetYaxis().SetNoExponent(bkg.GetMaximum() < 1.e4)
bkg.GetYaxis().SetMoreLogLabels(True)
leg.Draw()
drawCMS(LUMI, "Preliminary")
drawRegion(channel, True)
drawAnalysis("DM"+channel[:2])
drawOverflow()
setHistStyle(bkg, 1.2 if RATIO else 1.1)
setHistStyle(hist['BkgSum'], 1.2 if RATIO else 1.1)
if RATIO:
c1.cd(2)
err = hist['BkgSum'].Clone("BkgErr;")
err.SetTitle("")
err.GetYaxis().SetTitle("Data / Bkg")
for i in range(1, err.GetNbinsX()+1):
err.SetBinContent(i, 1)
if hist['BkgSum'].GetBinContent(i) > 0:
err.SetBinError(i, hist['BkgSum'].GetBinError(i)/hist['BkgSum'].GetBinContent(i))
setBotStyle(err)
errLine = err.Clone("errLine")
errLine.SetLineWidth(1)
errLine.SetFillStyle(0)
res = hist[data[0]].Clone("Residues")
for i in range(0, res.GetNbinsX()+1):
if hist['BkgSum'].GetBinContent(i) > 0:
res.SetBinContent(i, res.GetBinContent(i)/hist['BkgSum'].GetBinContent(i))
res.SetBinError(i, res.GetBinError(i)/hist['BkgSum'].GetBinContent(i))
setBotStyle(res)
#err.GetXaxis().SetLabelOffset(err.GetXaxis().GetLabelOffset()*5)
#err.GetXaxis().SetTitleOffset(err.GetXaxis().GetTitleOffset()*2)
err.Draw("E2")
if 'PreFit' in hist:
respre = hist['PreFit'].Clone("ResiduesPreFit")
respre.Divide(hist['BkgSum'])
respre.Draw("SAME, HIST")
errLine.Draw("SAME, HIST")
if not isBlind and len(data) > 0:
res.Draw("SAME, PE0")
#res_graph.Draw("SAME, PE0")
if len(err.GetXaxis().GetBinLabel(1))==0: # Bin labels: not a ordinary plot
drawRatio(hist['data_obs'], hist['BkgSum'])
drawStat(hist['data_obs'], hist['BkgSum'])
c1.Update()
if gROOT.IsBatch() and options.saveplots: # and (treeRead and channel in selection.keys()):
if not os.path.exists("plots_"+options.name+"/"+plotdir): os.makedirs("plots_"+options.name+"/"+plotdir)
c1.Print("plots_"+options.name+"/"+plotdir+"/"+plotname+binName+".png")
c1.Print("plots_"+options.name+"/"+plotdir+"/"+plotname+binName+".pdf")
# Print table
printTable(hist, sign)
if not gROOT.IsBatch(): raw_input("Press Enter to continue...")
if gROOT.IsBatch() and not fileRead and (var == 'MET_pt' or (channel.startswith('SL') and var == 'MET_sign') or (channel.endswith('ZR') and var == 'FakeMET_pt')):
saveHist(hist, channel+binName)
def dijet(category):
channel = 'bb'
stype = channel
isSB = True # relict from using Alberto's more complex script
isData = not ISMC
nTupleDir = NTUPLEDIR
samples = data if isData else back
pd = []
for sample_name in samples:
if YEAR == 'run2':
pd += sample[sample_name]['files']
else:
pd += [x for x in sample[sample_name]['files'] if YEAR in x]
print "datasets:", pd
if not os.path.exists(PLOTDIR): os.makedirs(PLOTDIR)
if BIAS: print "Running in BIAS mode"
order = 0
RSS = {}
X_mass = RooRealVar("jj_mass_widejet", "m_{jj}", X_min, X_max, "GeV")
weight = RooRealVar("MANtag_weight", "", -1.e9, 1.e9)
variables = RooArgSet(X_mass)
variables.add(RooArgSet(weight))
if VARBINS:
binsXmass = RooBinning(len(abins) - 1, abins)
X_mass.setBinning(RooBinning(len(abins_narrow) - 1, abins_narrow))
plot_binning = RooBinning(
int((X_mass.getMax() - X_mass.getMin()) / 100), X_mass.getMin(),
X_mass.getMax())
else:
X_mass.setBins(int((X_mass.getMax() - X_mass.getMin()) / 10))
binsXmass = RooBinning(int((X_mass.getMax() - X_mass.getMin()) / 100),
X_mass.getMin(), X_mass.getMax())
plot_binning = binsXmass
baseCut = ""
print stype, "|", baseCut
print " - Reading from Tree"
treeBkg = TChain("tree")
for ss in pd:
if os.path.exists(nTupleDir + ss + "_" + BTAGGING + ".root"):
treeBkg.Add(nTupleDir + ss + "_" + BTAGGING + ".root")
else:
print "found no file for sample:", ss
setData = RooDataSet("setData", "Data (QCD+TTbar MC)", variables,
RooFit.Cut(baseCut), RooFit.WeightVar(weight),
RooFit.Import(treeBkg))
nevents = setData.sumEntries()
dataMin, dataMax = array('d', [0.]), array('d', [0.])
setData.getRange(X_mass, dataMin, dataMax)
xmin, xmax = dataMin[0], dataMax[0]
lastBin = X_mass.getMax()
if VARBINS:
for b in narrow_bins:
if b > xmax:
lastBin = b
break
print "Imported", (
"data" if isData else "MC"
), "RooDataSet with", nevents, "events between [%.1f, %.1f]" % (xmin, xmax)
#xmax = xmax+binsXmass.averageBinWidth() # start form next bin
# 1 parameter
print "fitting 1 parameter model"
p1_1 = RooRealVar("CMS" + YEAR + "_" + category + "_p1_1", "p1", 7.0, 0.,
2000.)
modelBkg1 = RooGenericPdf("Bkg1", "Bkg. fit (2 par.)",
"1./pow(@0/13000, @1)", RooArgList(X_mass, p1_1))
normzBkg1 = RooRealVar(
modelBkg1.GetName() + "_norm", "Number of background events", nevents,
0., 5. * nevents) #range dependent of actual number of events!
modelExt1 = RooExtendPdf(modelBkg1.GetName() + "_ext",
modelBkg1.GetTitle(), modelBkg1, normzBkg1)
fitRes1 = modelExt1.fitTo(setData, RooFit.Extended(True), RooFit.Save(1),
RooFit.SumW2Error(not isData),
RooFit.Strategy(2), RooFit.Minimizer("Minuit2"),
RooFit.PrintLevel(1 if VERBOSE else -1))
fitRes1.Print()
RSS[1] = drawFit("Bkg1", category, X_mass, modelBkg1, setData, binsXmass,
[fitRes1], normzBkg1.getVal())
# 2 parameters
print "fitting 2 parameter model"
p2_1 = RooRealVar("CMS" + YEAR + "_" + category + "_p2_1", "p1", 0., -100.,
1000.)
p2_2 = RooRealVar("CMS" + YEAR + "_" + category + "_p2_2", "p2",
p1_1.getVal(), -100., 600.)
modelBkg2 = RooGenericPdf("Bkg2", "Bkg. fit (3 par.)",
"pow(1-@0/13000, @1) / pow(@0/13000, @2)",
RooArgList(X_mass, p2_1, p2_2))
normzBkg2 = RooRealVar(modelBkg2.GetName() + "_norm",
"Number of background events", nevents, 0.,
5. * nevents)
modelExt2 = RooExtendPdf(modelBkg2.GetName() + "_ext",
modelBkg2.GetTitle(), modelBkg2, normzBkg2)
fitRes2 = modelExt2.fitTo(setData, RooFit.Extended(True), RooFit.Save(1),
RooFit.SumW2Error(not isData),
RooFit.Strategy(2), RooFit.Minimizer("Minuit2"),
RooFit.PrintLevel(1 if VERBOSE else -1))
fitRes2.Print()
RSS[2] = drawFit("Bkg2", category, X_mass, modelBkg2, setData, binsXmass,
[fitRes2], normzBkg2.getVal())
# 3 parameters
print "fitting 3 parameter model"
p3_1 = RooRealVar("CMS" + YEAR + "_" + category + "_p3_1", "p1",
p2_1.getVal(), -2000., 2000.)
p3_2 = RooRealVar("CMS" + YEAR + "_" + category + "_p3_2", "p2",
p2_2.getVal(), -400., 2000.)
p3_3 = RooRealVar("CMS" + YEAR + "_" + category + "_p3_3", "p3", -2.5,
-500., 500.)
modelBkg3 = RooGenericPdf(
"Bkg3", "Bkg. fit (4 par.)",
"pow(1-@0/13000, @1) / pow(@0/13000, @2+@3*log(@0/13000))",
RooArgList(X_mass, p3_1, p3_2, p3_3))
normzBkg3 = RooRealVar(modelBkg3.GetName() + "_norm",
"Number of background events", nevents, 0.,
5. * nevents)
modelExt3 = RooExtendPdf(modelBkg3.GetName() + "_ext",
modelBkg3.GetTitle(), modelBkg3, normzBkg3)
fitRes3 = modelExt3.fitTo(setData, RooFit.Extended(True), RooFit.Save(1),
RooFit.SumW2Error(not isData),
RooFit.Strategy(2), RooFit.Minimizer("Minuit2"),
RooFit.PrintLevel(1 if VERBOSE else -1))
fitRes3.Print()
RSS[3] = drawFit("Bkg3", category, X_mass, modelBkg3, setData, binsXmass,
[fitRes3], normzBkg3.getVal())
# 4 parameters
print "fitting 4 parameter model"
p4_1 = RooRealVar("CMS" + YEAR + "_" + category + "_p4_1", "p1",
p3_1.getVal(), -2000., 2000.)
p4_2 = RooRealVar("CMS" + YEAR + "_" + category + "_p4_2", "p2",
p3_2.getVal(), -2000., 2000.)
p4_3 = RooRealVar("CMS" + YEAR + "_" + category + "_p4_3", "p3",
p3_3.getVal(), -50., 50.)
p4_4 = RooRealVar("CMS" + YEAR + "_" + category + "_p4_4", "p4", 0.1, -50.,
50.)
modelBkg4 = RooGenericPdf(
"Bkg4", "Bkg. fit (5 par.)",
"pow(1 - @0/13000, @1) / pow(@0/13000, @2+@3*log(@0/13000)+@4*pow(log(@0/13000), 2))",
RooArgList(X_mass, p4_1, p4_2, p4_3, p4_4))
normzBkg4 = RooRealVar(modelBkg4.GetName() + "_norm",
"Number of background events", nevents, 0.,
5. * nevents)
modelExt4 = RooExtendPdf(modelBkg4.GetName() + "_ext",
modelBkg4.GetTitle(), modelBkg4, normzBkg4)
fitRes4 = modelExt4.fitTo(setData, RooFit.Extended(True), RooFit.Save(1),
RooFit.SumW2Error(not isData),
RooFit.Strategy(2), RooFit.Minimizer("Minuit2"),
RooFit.PrintLevel(1 if VERBOSE else -1))
fitRes4.Print()
RSS[4] = drawFit("Bkg4", category, X_mass, modelBkg4, setData, binsXmass,
[fitRes4], normzBkg4.getVal())
# Normalization parameters are should be set constant, but shape ones should not
# if BIAS:
# p1_1.setConstant(True)
# p2_1.setConstant(True)
# p2_2.setConstant(True)
# p3_1.setConstant(True)
# p3_2.setConstant(True)
# p3_3.setConstant(True)
# p4_1.setConstant(True)
# p4_2.setConstant(True)
# p4_3.setConstant(True)
# p4_4.setConstant(True)
normzBkg1.setConstant(True)
normzBkg2.setConstant(True)
normzBkg3.setConstant(True)
normzBkg4.setConstant(True)
#*******************************************************#
# #
# Fisher #
# #
#*******************************************************#
# Fisher test
with open(PLOTDIR + "/Fisher_" + category + ".tex", 'w') as fout:
fout.write(r"\begin{tabular}{c|c|c|c|c}")
fout.write("\n")
fout.write(r"function & $\chi^2$ & RSS & ndof & F-test \\")
fout.write("\n")
fout.write("\hline")
fout.write("\n")
CL_high = False
for o1 in range(1, 5):
o2 = min(o1 + 1, 5)
fout.write("%d par & %.2f & %.2f & %d & " %
(o1 + 1, RSS[o1]["chi2"], RSS[o1]["rss"],
RSS[o1]["nbins"] - RSS[o1]["npar"]))
if o2 > len(RSS):
fout.write(r"\\")
fout.write("\n")
continue #order==0 and
CL = fisherTest(RSS[o1]['rss'], RSS[o2]['rss'], o1 + 1., o2 + 1.,
RSS[o1]["nbins"])
fout.write("CL=%.3f " % (CL))
if CL > 0.10: # The function with less parameters is enough
if not CL_high:
order = o1
#fout.write( "%d par are sufficient " % (o1+1))
CL_high = True
else:
#fout.write( "%d par are needed " % (o2+1))
if not CL_high:
order = o2
fout.write(r"\\")
fout.write("\n")
fout.write("\hline")
fout.write("\n")
fout.write(r"\end{tabular}")
print "saved F-test table as", PLOTDIR + "/Fisher_" + category + ".tex"
#print "-"*25
#print "function & $\\chi^2$ & RSS & ndof & F-test & result \\\\"
#print "\\multicolumn{6}{c}{", "Zprime_to_bb", "} \\\\"
#print "\\hline"
#CL_high = False
#for o1 in range(1, 5):
# o2 = min(o1 + 1, 5)
# print "%d par & %.2f & %.2f & %d & " % (o1+1, RSS[o1]["chi2"], RSS[o1]["rss"], RSS[o1]["nbins"]-RSS[o1]["npar"]),
# if o2 > len(RSS):
# print "\\\\"
# continue #order==0 and
# CL = fisherTest(RSS[o1]['rss'], RSS[o2]['rss'], o1+1., o2+1., RSS[o1]["nbins"])
# print "%d par vs %d par CL=%f & " % (o1+1, o2+1, CL),
# if CL > 0.10: # The function with less parameters is enough
# if not CL_high:
# order = o1
# print "%d par are sufficient" % (o1+1),
# CL_high=True
# else:
# print "%d par are needed" % (o2+1),
# if not CL_high:
# order = o2
# print "\\\\"
#print "\\hline"
#print "-"*25
#print "@ Order is", order, "("+category+")"
#order = min(3, order)
#order = 2
if order == 1:
modelBkg = modelBkg1 #.Clone("Bkg")
modelAlt = modelBkg2 #.Clone("BkgAlt")
normzBkg = normzBkg1 #.Clone("Bkg_norm")
fitRes = fitRes1
elif order == 2:
modelBkg = modelBkg2 #.Clone("Bkg")
modelAlt = modelBkg3 #.Clone("BkgAlt")
normzBkg = normzBkg2 #.Clone("Bkg_norm")
fitRes = fitRes2
elif order == 3:
modelBkg = modelBkg3 #.Clone("Bkg")
modelAlt = modelBkg4 #.Clone("BkgAlt")
normzBkg = normzBkg3 #.Clone("Bkg_norm")
fitRes = fitRes3
elif order == 4:
modelBkg = modelBkg4 #.Clone("Bkg")
modelAlt = modelBkg3 #.Clone("BkgAlt")
normzBkg = normzBkg4 #.Clone("Bkg_norm")
fitRes = fitRes4
else:
print "Functions with", order + 1, "or more parameters are needed to fit the background"
exit()
modelBkg.SetName("Bkg_" + YEAR + "_" + category)
modelAlt.SetName("Alt_" + YEAR + "_" + category)
normzBkg.SetName("Bkg_" + YEAR + "_" + category + "_norm")
print "-" * 25
# Generate pseudo data
setToys = RooDataSet()
setToys.SetName("data_toys")
setToys.SetTitle("Data (toys)")
if not isData:
print " - Generating", nevents, "events for toy data"
setToys = modelBkg.generate(RooArgSet(X_mass), nevents)
#setToys = modelAlt.generate(RooArgSet(X_mass), nevents)
print "toy data generated"
if VERBOSE: raw_input("Press Enter to continue...")
#*******************************************************#
# #
# Plot #
# #
#*******************************************************#
print "starting to plot"
c = TCanvas("c_" + category, category, 800, 800)
c.Divide(1, 2)
setTopPad(c.GetPad(1), RATIO)
setBotPad(c.GetPad(2), RATIO)
c.cd(1)
frame = X_mass.frame()
setPadStyle(frame, 1.25, True)
if VARBINS: frame.GetXaxis().SetRangeUser(X_mass.getMin(), lastBin)
signal = getSignal(
category, stype,
2000) #replacing Alberto's getSignal by own dummy function
graphData = setData.plotOn(frame, RooFit.Binning(plot_binning),
RooFit.Scaling(False), RooFit.Invisible())
modelBkg.plotOn(frame, RooFit.VisualizeError(fitRes, 1, False),
RooFit.LineColor(602), RooFit.FillColor(590),
RooFit.FillStyle(1001), RooFit.DrawOption("FL"),
RooFit.Name("1sigma"))
modelBkg.plotOn(frame, RooFit.LineColor(602), RooFit.FillColor(590),
RooFit.FillStyle(1001), RooFit.DrawOption("L"),
RooFit.Name(modelBkg.GetName()))
modelAlt.plotOn(frame, RooFit.LineStyle(7), RooFit.LineColor(613),
RooFit.FillColor(609), RooFit.FillStyle(1001),
RooFit.DrawOption("L"), RooFit.Name(modelAlt.GetName()))
if not isSB and signal[0] is not None: # FIXME remove /(2./3.)
signal[0].plotOn(
frame,
RooFit.Normalization(signal[1] * signal[2], RooAbsReal.NumEvent),
RooFit.LineStyle(3), RooFit.LineWidth(6), RooFit.LineColor(629),
RooFit.DrawOption("L"), RooFit.Name("Signal"))
graphData = setData.plotOn(
frame, RooFit.Binning(plot_binning), RooFit.Scaling(False),
RooFit.XErrorSize(0 if not VARBINS else 1),
RooFit.DataError(RooAbsData.Poisson if isData else RooAbsData.SumW2),
RooFit.DrawOption("PE0"), RooFit.Name(setData.GetName()))
fixData(graphData.getHist(), True, True, not isData)
pulls = frame.pullHist(setData.GetName(), modelBkg.GetName(), True)
chi = frame.chiSquare(setData.GetName(), modelBkg.GetName(), True)
#setToys.plotOn(frame, RooFit.DataError(RooAbsData.Poisson), RooFit.DrawOption("PE0"), RooFit.MarkerColor(2))
frame.GetYaxis().SetTitle("Events / ( 100 GeV )")
frame.GetYaxis().SetTitleOffset(1.05)
frame.Draw()
#print "frame drawn"
# Get Chi2
# chi2[1] = frame.chiSquare(modelBkg1.GetName(), setData.GetName())
# chi2[2] = frame.chiSquare(modelBkg2.GetName(), setData.GetName())
# chi2[3] = frame.chiSquare(modelBkg3.GetName(), setData.GetName())
# chi2[4] = frame.chiSquare(modelBkg4.GetName(), setData.GetName())
frame.SetMaximum(frame.GetMaximum() * 10)
frame.SetMinimum(max(frame.GetMinimum(), 1.e-1))
c.GetPad(1).SetLogy()
drawAnalysis(category)
drawRegion(category, True)
#drawCMS(LUMI, "Simulation Preliminary")
drawCMS(LUMI, "Work in Progress", suppressCMS=True)
leg = TLegend(0.575, 0.6, 0.95, 0.9)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
leg.AddEntry(setData.GetName(),
setData.GetTitle() + " (%d events)" % nevents, "PEL")
leg.AddEntry(modelBkg.GetName(), modelBkg.GetTitle(),
"FL") #.SetTextColor(629)
leg.AddEntry(modelAlt.GetName(), modelAlt.GetTitle(), "L")
if not isSB and signal[0] is not None:
leg.AddEntry("Signal", signal[0].GetTitle(), "L")
leg.SetY1(0.9 - leg.GetNRows() * 0.05)
leg.Draw()
latex = TLatex()
latex.SetNDC()
latex.SetTextSize(0.04)
latex.SetTextFont(42)
if not isSB:
latex.DrawLatex(leg.GetX1() * 1.16,
leg.GetY1() - 0.04, "HVT model B (g_{V}=3)")
# latex.DrawLatex(0.67, leg.GetY1()-0.045, "#sigma_{X} = 1.0 pb")
c.cd(2)
frame_res = X_mass.frame()
setPadStyle(frame_res, 1.25)
frame_res.addPlotable(pulls, "P")
setBotStyle(frame_res, RATIO, False)
if VARBINS: frame_res.GetXaxis().SetRangeUser(X_mass.getMin(), lastBin)
frame_res.GetYaxis().SetRangeUser(-5, 5)
frame_res.GetYaxis().SetTitle("pulls(#sigma)")
frame_res.GetYaxis().SetTitleOffset(0.3)
frame_res.Draw()
fixData(pulls, False, True, False)
drawChi2(RSS[order]["chi2"], RSS[order]["nbins"] - (order + 1), True)
line = drawLine(X_mass.getMin(), 0, lastBin, 0)
if VARBINS:
c.SaveAs(PLOTDIR + "/BkgSR_" + category + ".pdf")
c.SaveAs(PLOTDIR + "/BkgSR_" + category + ".png")
else:
c.SaveAs(PLOTDIR + "/BkgSR_" + category + ".pdf")
c.SaveAs(PLOTDIR + "/BkgSR_" + category + ".png")
#*******************************************************#
# #
# Generate workspace #
# #
#*******************************************************#
if BIAS:
gSystem.Load("libHiggsAnalysisCombinedLimit.so")
from ROOT import RooMultiPdf
cat = RooCategory("pdf_index", "Index of Pdf which is active")
pdfs = RooArgList(modelBkg, modelAlt)
roomultipdf = RooMultiPdf("roomultipdf", "All Pdfs", cat, pdfs)
normulti = RooRealVar("roomultipdf_norm",
"Number of background events", nevents, 0., 1.e6)
normzBkg.setConstant(
False
) ## newly put here to ensure it's freely floating in the combine fit
# create workspace
w = RooWorkspace("Zprime_" + YEAR, "workspace")
# Dataset
if isData: getattr(w, "import")(setData, RooFit.Rename("data_obs"))
else: getattr(w, "import")(setToys, RooFit.Rename("data_obs"))
#getattr(w, "import")(setData, RooFit.Rename("data_obs"))
if BIAS:
getattr(w, "import")(cat, RooFit.Rename(cat.GetName()))
getattr(w, "import")(normulti, RooFit.Rename(normulti.GetName()))
getattr(w, "import")(roomultipdf, RooFit.Rename(roomultipdf.GetName()))
getattr(w, "import")(modelBkg, RooFit.Rename(modelBkg.GetName()))
getattr(w, "import")(modelAlt, RooFit.Rename(modelAlt.GetName()))
getattr(w, "import")(normzBkg, RooFit.Rename(normzBkg.GetName()))
w.writeToFile(WORKDIR + "%s_%s.root" % (DATA_TYPE + "_" + YEAR, category),
True)
print "Workspace", WORKDIR + "%s_%s.root" % (
DATA_TYPE + "_" + YEAR, category), "saved successfully"
if VERBOSE: raw_input("Press Enter to continue...")
def plot(var, cut, year, norm=False, nm1=False):
### Preliminary Operations ###
treeRead = not cut in [
"nnqq", "en", "enqq", "mn", "mnqq", "ee", "eeqq", "mm", "mmqq", "em",
"emqq", "qqqq"
] # Read from tree
channel = cut
unit = ''
if "GeV" in variable[var]['title']: unit = ' GeV'
isBlind = BLIND and 'SR' in channel
isAH = False #'qqqq' in channel or 'hp' in channel or 'lp' in channel
showSignal = False if 'SB' in cut or 'TR' in cut else True #'SR' in channel or channel=='qqqq'#or len(channel)==5
stype = "HVT model B"
if len(sign) > 0 and 'AZh' in sign[0]: stype = "2HDM"
elif len(sign) > 0 and 'monoH' in sign[0]: stype = "Z'-2HDM m_{A}=300 GeV"
if treeRead:
for k in sorted(alias.keys(), key=len, reverse=True):
if BTAGGING == 'semimedium':
if k in cut:
if ADDSELECTION:
cut = cut.replace(
k, aliasSM[k] + SELECTIONS[options.selection])
else:
cut = cut.replace(k, aliasSM[k])
else:
if k in cut:
if ADDSELECTION:
cut = cut.replace(
k, alias[k].format(WP=working_points[BTAGGING]) +
SELECTIONS[options.selection])
else:
cut = cut.replace(
k, alias[k].format(WP=working_points[BTAGGING]))
# Determine Primary Dataset
pd = sample['data_obs']['files']
print "Plotting from", ("tree" if treeRead else
"file"), var, "in", channel, "channel with:"
print " dataset:", pd
print " cut :", cut
if var == 'jj_deltaEta_widejet':
if "jj_deltaEta_widejet<1.1 && " in cut:
print
print "omitting jj_deltaEta_widejet<1.1 cut to draw the deltaEta distribution"
print
cut = cut.replace("jj_deltaEta_widejet<1.1 && ", "")
else:
print
print "no 'jj_deltaEta_widejet<1.1 && ' in the cut string detected, so it cannot be ommited explicitly"
print
### Create and fill MC histograms ###
# Create dict
file = {}
tree = {}
hist = {}
### Create and fill MC histograms ###
for i, s in enumerate(data + back + sign):
if treeRead: # Project from tree
tree[s] = TChain("tree")
for j, ss in enumerate(sample[s]['files']):
if not 'data' in s or ('data' in s and ss in pd):
if year == "run2" or year in ss:
tree[s].Add(NTUPLEDIR + ss + ".root")
if variable[var]['nbins'] > 0:
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events / ( " + str(
(variable[var]['max'] - variable[var]['min']) /
variable[var]['nbins']) + unit + " );" +
('log' if variable[var]['log'] else ''),
variable[var]['nbins'], variable[var]['min'],
variable[var]['max'])
else:
hist[s] = TH1F(
s, ";" + variable[var]['title'] + ";Events" +
('log' if variable[var]['log'] else ''),
len(variable[var]['bins']) - 1,
array('f', variable[var]['bins']))
hist[s].Sumw2()
cutstring = "(eventWeightLumi)" + ("*(" + cut +
")" if len(cut) > 0 else "")
tree[s].Project(s, var, cutstring)
if not tree[s].GetTree() == None:
hist[s].SetOption("%s" % tree[s].GetTree().GetEntriesFast())
else: # Histogram written to file
for j, ss in enumerate(sample[s]['files']):
if not 'data' in s or ('data' in s and ss in pd):
file[ss] = TFile(NTUPLEDIR + ss + ".root", "R")
if file[ss].IsZombie():
print "WARNING: file", NTUPLEDIR + ss + ".root", "does not exist"
continue
tmphist = file[ss].Get(cut + "/" + var)
if tmphist == None: continue
if not s in hist.keys(): hist[s] = tmphist
else: hist[s].Add(tmphist)
hist[s].Scale(sample[s]['weight'] if hist[s].Integral() >= 0 else 0)
hist[s].SetFillColor(sample[s]['fillcolor'])
hist[s].SetFillStyle(sample[s]['fillstyle'])
hist[s].SetLineColor(sample[s]['linecolor'])
hist[s].SetLineStyle(sample[s]['linestyle'])
if channel.endswith('TR') and channel.replace('TR', '') in topSF:
hist['TTbarSL'].Scale(topSF[channel.replace('TR', '')][0])
hist['ST'].Scale(topSF[channel.replace('TR', '')][0])
hist['BkgSum'] = hist['data_obs'].Clone(
"BkgSum") if 'data_obs' in hist else hist[back[0]].Clone("BkgSum")
hist['BkgSum'].Reset("MICES")
hist['BkgSum'].SetFillStyle(3003)
hist['BkgSum'].SetFillColor(1)
for i, s in enumerate(back):
hist['BkgSum'].Add(hist[s])
if options.norm:
for i, s in enumerate(back + ['BkgSum']):
hist[s].Scale(hist[data[0]].Integral() / hist['BkgSum'].Integral())
# Create data and Bkg sum histograms
if options.blind or 'SR' in channel:
hist['data_obs'] = hist['BkgSum'].Clone("data_obs")
hist['data_obs'].Reset("MICES")
# Set histogram style
hist['data_obs'].SetMarkerStyle(20)
hist['data_obs'].SetMarkerSize(1.25)
for i, s in enumerate(data + back + sign + ['BkgSum']):
addOverflow(hist[s], False) # Add overflow
for i, s in enumerate(sign):
hist[s].SetLineWidth(3)
for i, s in enumerate(sign):
sample[s][
'plot'] = True #sample[s]['plot'] and s.startswith(channel[:2])
if isAH:
for i, s in enumerate(back):
hist[s].SetFillStyle(3005)
hist[s].SetLineWidth(2)
#for i, s in enumerate(sign):
# hist[s].SetFillStyle(0)
if not var == "Events":
sfnorm = hist[data[0]].Integral() / hist['BkgSum'].Integral()
print "Applying SF:", sfnorm
for i, s in enumerate(back + ['BkgSum']):
hist[s].Scale(sfnorm)
if BLIND and var.endswith("Mass"):
for i, s in enumerate(data + back + ['BkgSum']):
first, last = hist[s].FindBin(65), hist[s].FindBin(135)
for j in range(first, last):
hist[s].SetBinContent(j, -1.e-4)
if BLIND and var.endswith("Tau21"):
for i, s in enumerate(data):
first, last = hist[s].FindBin(0), hist[s].FindBin(0.6)
for j in range(first, last):
hist[s].SetBinContent(j, -1.e-4)
if SYNC and var == "jj_mass_widejet" and year in ["2016", "2017", "2018"]:
#iFile = TFile("sync/JetHT_run" + year + "_red_cert_scan.root", "READ")
#hist['sync'] = iFile.Get("Mjj")
if year == '2016':
iFile = TFile("sync/2016/2016_07Aug2017_1246_1p1.root", "READ")
hist['sync'] = iFile.Get("h_mjj_data")
elif year == '2017':
iFile = TFile(
"sync/2017/histos_Run2017BCDEF_17Nov2017_JEC2017_mjj1530_cemf_lt_0p8_deltaETA_lt_1p1.root",
"READ")
hist['sync'] = iFile.Get("h_mjj_data")
elif year == '2018':
iFile = TFile(
"sync/2018/Double_sideband_inputs_18v10_preliminary_v2.root",
"READ")
hist['sync'] = iFile.Get("h_mjj")
# hist['sync'] = tmp.Rebin(len(dijet_bins)-1, "sync", array('d', dijet_bins))
# hist['sync'] = tmp.Rebin(100, "sync")
hist['sync'].SetMarkerStyle(31)
hist['sync'].SetMarkerSize(1.25)
hist['sync'].SetMarkerColor(2)
print "Imported and drawing sync file"
# Create stack
if variable[var]['nbins'] > 0:
bkg = THStack(
"Bkg",
";" + hist['BkgSum'].GetXaxis().GetTitle() + ";Events / ( " + str(
(variable[var]['max'] - variable[var]['min']) /
variable[var]['nbins']) + unit + " )")
else:
bkg = THStack("Bkg",
";" + hist['BkgSum'].GetXaxis().GetTitle() + ";Events; ")
for i, s in enumerate(back):
bkg.Add(hist[s])
# Legend
leg = TLegend(0.65, 0.6, 0.95, 0.9)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
if len(data) > 0:
leg.AddEntry(hist[data[0]], sample[data[0]]['label'], "pe")
for i, s in reversed(list(enumerate(['BkgSum'] + back))):
leg.AddEntry(hist[s], sample[s]['label'], "f")
if showSignal:
for i, s in enumerate(sign):
if sample[s]['plot']:
leg.AddEntry(hist[s], sample[s]['label'], "fl")
leg.SetY1(0.9 - leg.GetNRows() * 0.05)
# --- Display ---
c1 = TCanvas("c1",
hist.values()[0].GetXaxis().GetTitle(), 800,
800 if RATIO else 600)
if RATIO:
c1.Divide(1, 2)
setTopPad(c1.GetPad(1), RATIO)
setBotPad(c1.GetPad(2), RATIO)
c1.cd(1)
c1.GetPad(bool(RATIO)).SetTopMargin(0.06)
c1.GetPad(bool(RATIO)).SetRightMargin(0.05)
c1.GetPad(bool(RATIO)).SetTicks(1, 1)
log = variable[var]['log'] #"log" in hist['BkgSum'].GetZaxis().GetTitle()
if log: c1.GetPad(bool(RATIO)).SetLogy()
# Draw
bkg.Draw("HIST") # stack
hist['BkgSum'].Draw("SAME, E2") # sum of bkg
if not isBlind and len(data) > 0: hist['data_obs'].Draw("SAME, PE") # data
if 'sync' in hist: hist['sync'].Draw("SAME, PE")
#data_graph.Draw("SAME, PE")
if showSignal:
smagn = 1. #if treeRead else 1.e2 #if log else 1.e2
for i, s in enumerate(sign):
# if sample[s]['plot']:
hist[s].Scale(smagn)
hist[s].Draw(
"SAME, HIST"
) # signals Normalized, hist[s].Integral()*sample[s]['weight']
textS = drawText(0.80, 0.9 - leg.GetNRows() * 0.05 - 0.02,
stype + " (x%d)" % smagn, True)
#bkg.GetYaxis().SetTitleOffset(bkg.GetYaxis().GetTitleOffset()*1.075)
bkg.GetYaxis().SetTitleOffset(0.9)
#bkg.GetYaxis().SetTitleOffset(2.)
bkg.SetMaximum((5. if log else 1.25) * max(
bkg.GetMaximum(),
hist['data_obs'].GetBinContent(hist['data_obs'].GetMaximumBin()) +
hist['data_obs'].GetBinError(hist['data_obs'].GetMaximumBin())))
#if bkg.GetMaximum() < max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum()): bkg.SetMaximum(max(hist[sign[0]].GetMaximum(), hist[sign[-1]].GetMaximum())*1.25)
bkg.SetMinimum(
max(
min(hist['BkgSum'].GetBinContent(hist['BkgSum'].GetMinimumBin(
)), hist['data_obs'].GetMinimum()), 5.e-1) if log else 0.)
if log:
bkg.GetYaxis().SetNoExponent(bkg.GetMaximum() < 1.e4)
#bkg.GetYaxis().SetMoreLogLabels(True)
bkg.GetXaxis().SetRangeUser(variable[var]['min'], variable[var]['max'])
#if log: bkg.SetMinimum(1)
leg.Draw()
#drawCMS(LUMI[year], "Preliminary")
#drawCMS(LUMI[year], "Work in Progress", suppressCMS=True)
drawCMS(LUMI[year], "", suppressCMS=True)
drawRegion('XVH' + channel, True)
drawAnalysis(channel)
setHistStyle(bkg, 1.2 if RATIO else 1.1)
setHistStyle(hist['BkgSum'], 1.2 if RATIO else 1.1)
if RATIO:
c1.cd(2)
err = hist['BkgSum'].Clone("BkgErr;")
err.SetTitle("")
if SYNC:
err.GetYaxis().SetTitle("Nano/Mini")
else:
err.GetYaxis().SetTitle("Data / MC")
err.GetYaxis().SetTitleOffset(0.9)
err.GetXaxis().SetRangeUser(variable[var]['min'], variable[var]['max'])
for i in range(1, err.GetNbinsX() + 1):
err.SetBinContent(i, 1)
if hist['BkgSum'].GetBinContent(i) > 0:
err.SetBinError(
i, hist['BkgSum'].GetBinError(i) /
hist['BkgSum'].GetBinContent(i))
setBotStyle(err)
errLine = err.Clone("errLine")
errLine.SetLineWidth(1)
errLine.SetFillStyle(0)
res = hist['data_obs'].Clone("Residues")
for i in range(0, res.GetNbinsX() + 1):
if hist['BkgSum'].GetBinContent(i) > 0:
res.SetBinContent(
i,
res.GetBinContent(i) / hist['BkgSum'].GetBinContent(i))
res.SetBinError(
i,
res.GetBinError(i) / hist['BkgSum'].GetBinContent(i))
if 'sync' in hist:
res.SetMarkerColor(1)
res.SetMarkerStyle(20)
res.Reset()
for i in range(0, res.GetNbinsX() + 1):
x = hist['data_obs'].GetXaxis().GetBinCenter(i)
if hist['sync'].GetBinContent(hist['sync'].FindBin(x)) > 0:
res.SetBinContent(
i, hist['data_obs'].GetBinContent(
hist['data_obs'].FindBin(x)) /
hist['sync'].GetBinContent(hist['sync'].FindBin(x)))
res.SetBinError(
i, hist['data_obs'].GetBinError(
hist['data_obs'].FindBin(x)) /
hist['sync'].GetBinContent(hist['sync'].FindBin(x)))
setBotStyle(res)
#err.GetXaxis().SetLabelOffset(err.GetXaxis().GetLabelOffset()*5)
#err.GetXaxis().SetTitleOffset(err.GetXaxis().GetTitleOffset()*2)
err.Draw("E2")
errLine.Draw("SAME, HIST")
if not isBlind and len(data) > 0:
res.Draw("SAME, PE0")
#res_graph.Draw("SAME, PE0")
if len(err.GetXaxis().GetBinLabel(
1)) == 0: # Bin labels: not a ordinary plot
drawRatio(hist['data_obs'], hist['BkgSum'])
drawStat(hist['data_obs'], hist['BkgSum'])
if SYNC: err.GetYaxis().SetRangeUser(0.9, 1.1)
c1.Update()
if gROOT.IsBatch():
if channel == "": channel = "nocut"
varname = var.replace('.', '_').replace('()', '')
if not os.path.exists("plots/" + channel):
os.makedirs("plots/" + channel)
suffix = ''
if "b" in channel or 'mu' in channel: suffix += "_" + BTAGGING
if ADDSELECTION: suffix += "_" + options.selection
c1.Print("plots/" + channel + "/" + varname + "_" + year + suffix +
".png")
c1.Print("plots/" + channel + "/" + varname + "_" + year + suffix +
".pdf")
# Print table
printTable(hist, sign)
# if True:
# sFile = TFile("sync/data_2016.root", "RECREATE")
# sFile.cd()
# hist['data_obs'].
if not gROOT.IsBatch(): raw_input("Press Enter to continue...")
def plotPrePost(category, category2):
if len(category) == 0:
print "Please select a category with the -c option"
exit()
is2 = len(category2) > 0
if 'l' in category:
category, category2, is2 = category.replace('l',
'e'), category.replace(
'l', 'm'), True
isAH = len(category) >= 5
X_name = "JJ_mass" if isAH else "VH_mass"
signalName = signal + category + "_M%d" % mass
data = ["data_obs"]
back = ["Bkg_" + category] if isAH else [
"VV_" + category, "Top_" + category, "Vjets_" + category
]
sign = [signalName] if len(signal) > 0 else []
fitr = ["total", "total_signal", "total_background"]
labels = {
"data_obs": "Data",
"VV_" + category: "VV, Vh",
"Top_" + category: "t#bar{t}, t+X",
"Vjets_" + category: "V+jets",
"Bkg_" + category: "Bkg fit",
signalName: "Signal"
}
if 'ee' in category or 'mm' in category:
labels["Vjets_" + category] = "Z(ll)+jets"
elif 'en' in category or 'mn' in category:
labels["Vjets_" + category] = "W(l#nu)+jets"
elif 'nn' in category:
labels["Vjets_" + category] = "Z(#nu#nu),W(l#nu)+jets"
lastBin = 3500. if 'ee' in category or 'mm' in category or 'nn' in category else 4500.
# xs = 0.
# if 'XWH' in signal or 'XVH' in signal: xs += HVT['B3']['W']['XS'][mass]*HVT['B3']['W']['BR'][mass]
# if 'XZH' in signal or 'XVH' in signal: xs += HVT['B3']['Z']['XS'][mass]*HVT['B3']['Z']['BR'][mass]
xs = getCrossSection(signal, category, mass)
#if len(signal2)>0: xs[signal2] = getCrossSection(signal2, category, 0)
histData, histPre, histPost, graphPre, graphPost = None, {}, {}, {}, {}
if readData:
dataFile = TFile("workspace/" + category + ".root", "READ")
workspace = dataFile.Get("VH_2016")
variable = workspace.var(X_name)
dataset = workspace.data(data[0])
if is2:
dataFile2 = TFile("workspace/" + category2 + ".root", "READ")
workspace2 = dataFile2.Get("VH_2016")
dataset2 = workspace2.data(data[0])
dataset.append(dataset2)
data2 = dataset.createHistogram(variable, variable,
variable.getBinning().numBins() / 10,
1)
data1 = data2.ProjectionX()
data1.SetMarkerStyle(20)
data1.SetMarkerSize(1.25)
data1.SetLineColor(1)
histData = data1
graphData = convertHistToGraph(histData, True)
width = histData.GetXaxis().GetBinWidth(1)
inFile = TFile(options.fileName, "READ")
if inFile == None:
print "File", options.fileName, "not found"
return
if not inFile.GetDirectory("shapes_prefit/" + category):
print "Category", category, "not recognized"
return
for i, h in enumerate(back + sign + fitr):
histPre[h] = inFile.Get("shapes_prefit/" + category + "/" + h)
if is2:
histPre[h].Add(
inFile.Get("shapes_prefit/" + category2 + "/" +
h.replace(category, category2)))
histPre[h].SetName(h + '_pre')
histPre[h].SetLineColor(getColor(h, category))
histPre[h].SetFillColor(getColor(h, category))
histPre[h].SetLineWidth(3)
if h in back: histPre[h].SetFillStyle(1001)
elif h in fitr: histPre[h].SetFillStyle(3002)
elif h in sign:
histPre[h].SetTitle("m_{%s'} = %d GeV" % (signal[1], mass))
histPre[h].SetOption("HVT model B g_{V}=3")
histPre[h].SetFillStyle(1)
histPre[h].SetLineStyle(3)
histPre[h].SetLineWidth(6)
if xs > 0.: histPre[h].Scale(xs * 1000.)
# histPre[h].Rebin(10)
if readData:
histPre[h].Scale(width)
histPre[h].GetYaxis().SetTitle("Events / ( %d GeV )" % width)
if 'nn' in category:
histPre[h].GetXaxis().SetTitle(
histPre[h].GetXaxis().GetTitle().replace(
'm_{VH}', 'm^{T}_{VH}'))
for i, h in enumerate(back + sign + fitr):
histPost[h] = inFile.Get("shapes_fit_b/" + category + "/" + h)
if is2:
histPost[h].Add(
inFile.Get("shapes_fit_b/" + category2 + "/" +
h.replace(category, category2)))
histPost[h].SetName(h + '_post')
histPost[h].SetLineColor(getColor(h, category))
histPost[h].SetFillColor(getColor(h, category))
histPost[h].SetLineWidth(3)
if h in back: histPost[h].SetFillStyle(1001)
elif h in fitr: histPost[h].SetFillStyle(3002)
elif h in sign:
histPost[h].SetTitle("m_{%s'} = %d GeV" % (signal[1], mass))
histPost[h].SetOption("HVT model B g_{V}=3")
histPost[h].SetFillStyle(1)
histPost[h].SetLineStyle(3)
histPost[h].SetLineWidth(6)
if xs > 0.: histPost[h].Scale(xs * 1000.)
# histPost[h].Rebin(10)
if readData:
histPost[h].Scale(width)
histPost[h].GetYaxis().SetTitle("Events / ( %d GeV )" % width)
if 'nn' in category:
histPost[h].GetXaxis().SetTitle(
histPost[h].GetXaxis().GetTitle().replace(
'm_{VH}', 'm^{T}_{VH}'))
# Set errors ot zero to have smooth curves
for i, h in enumerate(back + sign):
for i in range(histPre[h].GetNbinsX()):
histPre[h].SetBinError(i + 1, 0.)
for i in range(histPost[h].GetNbinsX()):
histPost[h].SetBinError(i + 1, 0.)
stackPre = THStack(
"Pre", ";" + histPre['total'].GetXaxis().GetTitle() + ";" +
histPre['total'].GetYaxis().GetTitle())
for i, s in enumerate(back):
stackPre.Add(histPre[s])
stackPost = THStack(
"Post", ";" + histPost['total'].GetXaxis().GetTitle() + ";" +
histPost['total'].GetYaxis().GetTitle())
for i, s in enumerate(back):
stackPost.Add(histPost[s])
for i, h in enumerate(back):
tmpPre = histPre[back[i]].Clone(back[i] + "_stack_pre")
for j in range(i):
tmpPre.Add(histPre[back[j]])
graphPre[back[i]] = convertHistToGraph(tmpPre)
tmpPost = histPost[back[i]].Clone(back[i] + "_stack_post")
for j in range(i):
tmpPost.Add(histPost[back[j]])
graphPost[back[i]] = convertHistToGraph(tmpPost)
# Additional signal, if present
inFiles = {}
for i, f in enumerate(
signals): #combine/test/mlfit_monoHnn_MZ3000_MA300.root
signalName2 = f.replace('combine/test/mlfit_', '').replace('.root', '')
if 'AZh' in signalName2:
signalName2 = signalName2.replace('AZh', 'AZh' + category)
signalSmpl2 = signalName2.replace(category, '')
signal2 = signalName2.split('_M')[0].replace(category, '')
try:
mass2, mass2A = int(signalName2.split('_M')[1]), 0
except:
mass2, mass2A = int(
signalName2.split('_MZ')[1].split('_MA')[0]), int(
signalName2.split('_MA')[1])
#
inFiles[signalName2] = TFile(f, "READ")
histSign2 = inFiles[signalName2].Get("shapes_prefit/" + category +
"/" + signalName2)
if is2:
histSign2.Add(inFiles[signalName2].Get(
"shapes_prefit/" + category2 + "/" +
signalName2.replace(category, category2)))
histSign2.SetName(signalName2 + '_pre')
if signal2.startswith('X'):
histSign2.SetTitle("m_{V'} = %d GeV" % mass2)
elif signal2.startswith('A'):
histSign2.SetTitle("m_{A} = %d GeV" % mass2)
elif not mass2 == 0:
histSign2.SetTitle("m_{Z'} = %d GeV" % mass2)
if i == 0:
histSign2.SetOption(
"Z'-2HDM\nm_{A}=300 GeV" if 'monoH' in signalName2 else
"Type-II 2HDM\ncos(#beta-#alpha) = 0.25\ntan#beta = 1")
histSign2.SetLineColor(getColor(signalName2, category))
histSign2.SetFillColor(getColor(signalName2, category))
histSign2.SetFillStyle(1)
histSign2.SetLineStyle(5 + i)
histSign2.SetLineWidth(5)
if readData: histSign2.Scale(width)
xs2 = getCrossSection(signalName2, category, 0)
if xs2 > 0.: histSign2.Scale(xs2 * 1000.)
sign += [signalName2]
histPre[signalName2] = histSign2
histPost[signalName2] = histSign2
leg = TLegend(0.6 - 0.005, 0.6, 0.925, 0.9)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
if readData:
leg.AddEntry(graphData, labels[data[0]], "pe")
for i, s in reversed(list(enumerate(back))):
leg.AddEntry(histPre[s], labels[s], "f")
#for i, s in enumerate(sign):
# leg.AddEntry(histPre[s], labels[s], "fl")
###
c1 = TCanvas("c1", "Pre-Post Fit", 800, 800)
c1.Divide(1, 2)
c1.cd(1)
setTopPad(c1.GetPad(1), RATIO)
setBotPad(c1.GetPad(2), RATIO)
c1.GetPad(1).SetTopMargin(0.06)
c1.GetPad(1).SetRightMargin(0.05)
c1.GetPad(1).SetBottomMargin(0.01)
c1.GetPad(1).SetTicks(1, 1)
#histData.Draw("APE" if d==0 else "SAME, PE")
#for i, h in enumerate(back): graphPost[h].Draw("ACL" if i==0 else "C")
stackPost.Draw("C")
setHistStyle(stackPost, 1.2)
stackPost.SetMaximum(stackPost.GetMaximum() * 5.)
stackPost.SetMinimum(max(stackPost.GetMinimum(), 0.2))
stackPost.GetXaxis().SetRangeUser(stackPost.GetXaxis().GetXmin(), lastBin)
histPost['total'].SetLineWidth(1)
histPost['total'].Draw("SAME, E3")
histPre['total_background'].SetLineColor(921)
histPre['total_background'].SetLineStyle(2)
histPre['total_background'].SetLineWidth(3)
histPre['total_background'].SetFillColor(1)
histPre['total_background'].SetFillStyle(0)
histPre['total_background'].Draw("SAME, HIST")
for i, s in enumerate(sign):
histPre[s].Draw("SAME, L")
if readData:
graphData.Draw("SAME, PE0")
leg.AddEntry(histPost['total'], "Bkg. unc.", "f")
leg.AddEntry(histPre['total_background'], "Pre-fit", "l")
for i, s in enumerate(sign):
for o in histPre[s].GetOption().split('\n'):
if len(o) > 0.: leg.AddEntry(None, o, "")
leg.AddEntry(histPre[s], histPre[s].GetTitle(), "l")
leg.SetY1(0.9 - leg.GetNRows() * 0.060)
leg.Draw()
# if len(sign)>0:
# latex = TLatex()
# latex.SetNDC()
# latex.SetTextSize(0.045)
# latex.SetTextFont(42)
# latex.DrawLatex(0.67, leg.GetY1()-0.045, "HVT model B g_{V}=3")
drawCMS(LUMI, "") #Preliminary
drawRegion('XVH' + options.category, True)
drawAnalysis(category)
c1.GetPad(1).SetLogy()
c1.cd(2)
err = histPost['total'].Clone("BkgErr;")
err.SetTitle("")
err.Reset("MICES")
err.GetYaxis().SetTitle("(N^{data}-N^{bkg})/#sigma")
setBotStyle(err, 4 + 1)
err.GetXaxis().SetTitleSize(0.16)
#err.GetXaxis().SetTitleOffset(1.25);
err.GetYaxis().SetTitleOffset(0.33)
err.GetXaxis().SetRangeUser(err.GetXaxis().GetXmin(), lastBin)
err.GetYaxis().SetRangeUser(-5., 5.)
err.SetLineWidth(2)
err.SetLineStyle(2)
err.SetFillStyle(0)
err.Draw("L") #"E2"
if readData:
pulls = makeResidHist(graphData, histPost['total'])
#setBotStyle(pulls, RATIO, False)
pulls.Draw("SAME, PE0")
#drawRatio(hist['data_obs'], hist['BkgSum'])
#drawStat(hist['data_obs'], hist['BkgSum'])
chi2, nbins, npar = 0., 0, 0
for i in range(0, pulls.GetN()):
if graphData.GetY()[i] > 1.e-3:
nbins = nbins + 1
chi2 += pulls.GetY()[i]**2
#drawChi2(chi2, nbins-npar, True)
c1.Update()
if not gROOT.IsBatch(): raw_input("Press Enter to continue...")
c1.Print("plotsPrePost/BkgSR_" + options.category + ".png")
c1.Print("plotsPrePost/BkgSR_" + options.category + ".pdf")
###
if VERBOSE:
c2 = TCanvas("c2", "Pre-Post Fit", 1200, 600)
c2.Divide(2, 1)
c2.cd(1)
c2.GetPad(1).SetTopMargin(0.06)
c2.GetPad(1).SetRightMargin(0.05)
c2.GetPad(1).SetBottomMargin(0.10)
c2.GetPad(1).SetTicks(1, 1)
#histData.Draw("APE" if d==0 else "SAME, PE")
#for i, h in enumerate(back): graphPre[h].Draw("FL" if i==0 else "SAME, FL")
stackPre.Draw("HIST")
histPre['total'].Draw("SAME, E2")
stackPre.SetMaximum(stackPre.GetMaximum() * 5.)
stackPre.SetMinimum(max(stackPre.GetMinimum(), 0.01))
histData.Draw("SAME, PE")
leg.Draw()
drawCMS(LUMI, "Preliminary")
drawRegion('XVH' + category, True)
drawAnalysis(category)
c2.cd(2)
c2.GetPad(2).SetTopMargin(0.06)
c2.GetPad(2).SetRightMargin(0.05)
c2.GetPad(2).SetBottomMargin(0.10)
c2.GetPad(2).SetTicks(1, 1)
#histData.Draw("APE" if d==0 else "SAME, PE")
#for i, h in enumerate(back): graphPost[h].Draw("ACL" if i==0 else "C")
stackPost.Draw("HIST")
histPost['total'].Draw("SAME, E2")
stackPost.SetMaximum(stackPost.GetMaximum() * 5.)
stackPost.SetMinimum(max(stackPost.GetMinimum(), 0.01))
histData.Draw("SAME, PE")
leg.Draw()
drawCMS(LUMI, "Preliminary")
drawRegion('XVH' + category, True)
drawAnalysis(category)
c2.GetPad(1).SetLogy()
c2.GetPad(2).SetLogy()
c2.Print("combine/test/" + signalName + "_prepost.png")
c2.Print("combine/test/" + signalName + "_prepost.pdf")
c2.Close()
for cand in cands[cand_type]:
ceff[cand], ccand_perevent[cand] = ({} for _ in range(2))
for iVar, var in enumerate(vars_to_plot[cand_type]):
ceff[cand][var] = TCanvas(f"c{var}{cand}", "", 1200, 400)
ccand_perevent[cand][var] = TCanvas(f"c{var}{cand}_perEvent", "", 1200, 400)
nCutsTested = hvar[cand][var][origins[0]][-1].GetNbinsX()
varTitle = hvar[cand][var][origins[0]][-1].GetXaxis().GetTitle()
ceff[cand][var].Divide(nPtBins, 1)
ccand_perevent[cand][var].Divide(nPtBins, 1)
for iPtBin in range(nPtBins):
# fill legend
for orig in origins:
if iPtBin == 1 and leg_orig.GetNRows() < len(origins):
leg_orig.AddEntry(
hvar[cand][var][orig][iPtBin - 1],
leg_orig_names[orig],
"pl",
)
ptMin = (
hvar_vs_pt[cand][var][origins[0]]
.GetXaxis()
.GetBinLowEdge(iPtBin + 1)
)
ptMax = (
hvar_vs_pt[cand][var][origins[0]]
.GetXaxis()
.GetBinUpEdge(iPtBin + 1)
def hvt(benchmark=['B3', 'A1']):
hxs = {}
hw = {}
gxs = {}
gw = {}
mg = TMultiGraph()
for m in massPoints:
hxs[m] = TH2F("hxs_M%d" % m, ";;", 50, -0.04, 3.96, 100, 0., 2.)
hw[m] = TH2F("hw_M%d" % m, ";;", 50, -0.04, 3.96, 50, 0., 2.)
for m in massPoints:
file = TFile.Open("HVT/scanHVT_M%s.root" % m, "READ")
tree = file.Get("tree")
for entry in range(
tree.GetEntries()): # Fill mass points only if NOT excluded
tree.GetEntry(entry)
gH, gF = tree.gv * tree.ch, tree.g * tree.g * tree.cq / tree.gv
XsBr = tree.CX0 * tree.BRbb * 1000. # in fb
if XsBr < observed[m]: hxs[m].Fill(gH, gF)
if tree.total_widthV0 / float(m) < width: hw[m].Fill(gH, gF)
for b in range(hxs[m].GetNbinsX() * hxs[m].GetNbinsY()):
hxs[m].SetBinContent(b, 1. if hxs[m].GetBinContent(b) > 0. else 0.)
hw[m].SetBinContent(b, 1. if hw[m].GetBinContent(b) > 0. else 0.)
#hxs[m].Smooth(20)
#hw[m].Smooth(20)
gxs[m] = getCurve(hxs[m])
for i, g in enumerate(gxs[m]):
g.SetLineColor(massColors[m])
g.SetFillColor(massColors[m])
g.SetFillStyle(massFill[m]) #(3345 if i>1 else 3354)
g.SetLineWidth(503 * (1 if i < 2 else -1))
mg.Add(g)
#if m==3000:
if m == massPoints[-1]:
gw[m] = getCurve(hw[m])
for i, g in enumerate(gw[m]):
g.SetPoint(0, 0., g.GetY()[0])
g.SetLineWidth(501 * (1 if i < 2 else -1))
g.SetLineColor(920 + 2)
g.SetFillColor(920 + 1)
g.SetFillStyle(3003)
mg.Add(g)
if options.root:
outFile = TFile("plotsLimit/Model.root", "RECREATE")
outFile.cd()
for m in massPoints:
mg[m].Write("X_M%d" % m)
mgW.Write("width")
outFile.Close()
print "Saved histogram in file plotsLimit/Model.root, exiting..."
exit()
### plot ###
c1 = TCanvas("c1", "HVT Exclusion Limits", 800, 600)
c1.cd()
c1.GetPad(0).SetTopMargin(0.06)
c1.GetPad(0).SetRightMargin(0.05)
c1.GetPad(0).SetTicks(1, 1)
mg.Draw("AC")
#mg.GetXaxis().SetTitle("g_{V} c_{H}")
mg.GetXaxis().SetTitle("Higgs and vector boson coupling g_{H}")
mg.GetXaxis().SetRangeUser(-3., 3.)
mg.GetXaxis().SetLabelSize(0.045)
mg.GetXaxis().SetTitleSize(0.045)
mg.GetXaxis().SetTitleOffset(1.)
#mg.GetYaxis().SetTitle("g^{2} c_{F} / g_{V}")
mg.GetYaxis().SetTitle("Fermion coupling g_{F}")
mg.GetYaxis().SetLabelSize(0.045)
mg.GetYaxis().SetTitleSize(0.045)
mg.GetYaxis().SetTitleOffset(1.)
mg.GetYaxis().SetRangeUser(-1.2, 1.2)
mg.GetYaxis().SetNdivisions(505)
# hxs[3500].Draw("CONTZ")
drawCMS(LUMI, "Preliminary", False)
# drawAnalysis("XVH"+category, False)
# latex = TLatex()
# latex.SetNDC()
# latex.SetTextFont(62)
# latex.SetTextSize(0.06)
# latex.DrawLatex(0.10, 0.925, "CMS")
# model B
g_model = {}
for i, b in enumerate(benchmark):
g_model[i] = TGraph(1)
g_model[i].SetTitle(models_name[b])
g_model[i].SetPoint(0, models_point[b][0], models_point[b][1])
g_model[i].SetMarkerStyle(models_style[b])
g_model[i].SetMarkerColor(models_color[b])
g_model[i].SetMarkerSize(1.5)
g_model[i].Draw("PSAME")
# text
latex = TLatex()
latex.SetTextSize(0.045)
latex.SetTextFont(42)
latex.SetTextColor(630)
# for b in benchmark: latex.DrawLatex(models_point[b][0]+0.02, models_point[b][1]+0.02, models_name[b])
latex.SetTextColor(920 + 2)
latex.DrawLatex(-2.8, -0.875,
"#frac{#Gamma_{Z'}}{m_{Z'}} > %.0f%%" % (width * 100, ))
#leg = TLegend(0.68, 0.60, 0.95, 0.94)
leg = TLegend(0.68, 0.34, 0.95, 0.66)
leg.SetBorderSize(1)
leg.SetFillStyle(1001)
leg.SetFillColor(0)
for m in massPoints:
leg.AddEntry(gxs[m][0], "m_{Z'} = %.1f TeV" % (m / 1000.), "fl")
for i, b in enumerate(benchmark):
leg.AddEntry(g_model[i], g_model[i].GetTitle(), "P")
leg.SetY1(leg.GetY2() - leg.GetNRows() * 0.050)
leg.SetMargin(0.35)
leg.Draw()
gxs_ = gxs[massPoints[0]][0].Clone("gxs_")
gxs_.SetLineColor(1)
# gxs_.SetFillColor(1)
latex.SetNDC()
latex.SetTextColor(1)
latex.SetTextSize(0.04)
latex.SetTextFont(52)
latex.DrawLatex(0.15, 0.95, "q#bar{q} #rightarrow Z' #rightarrow b#bar{b}")
c1.Print("plots/model/HVT.png")
c1.Print("plots/model/HVT.pdf")
c1.Print("plots/model/HVT.root")
c1.Print("plots/model/HVT.C")
#g = 0.646879, cH = 0.976246, cF = 1.02433
print "model B = [", 3 * 0.976246, ",", 0.646879 * 0.646879 * 1.02433 / 3, "]"
if not gROOT.IsBatch(): raw_input("Press Enter to continue...")
def plot(var, cut, signal):
sign = signal
### Preliminary Operations ###
treeRead = not 'cutflow' in var
#treeRead = False
channel = cut
if 'inc' in cut:
eventWeightLuminame = 'eventWeightLumi_nobtag'
else:
eventWeightLuminame = 'eventWeightLumi'
if "SB" in cut or "SR" in cut:
channel_name = cut[:-2]
else:
channel_name = cut
if year in ['2016', '2017', '2018']:
NTUPLEDIR = "/work/pbaertsc/heavy_resonance/Ntuples%s/" % (year)
else:
NTUPLEDIR = "/work/pbaertsc/heavy_resonance/"
#showSignal = False if 'SB' in cut or 'TR' in cut else True #'SR' in channel or channel=='qqqq'#or len(channel)==5
if var in [
'dijet_VBF_mass', 'deltaR_VBF', 'deltaR_HVBFjet1',
'deltaR_HVBFjet2'
]:
showSignal = False
else:
showSignal = True
if treeRead:
for k in sorted(selection.keys(), key=len, reverse=True):
if k in cut: cut = cut.replace(k, selection[k])
print "Plotting from", ("tree" if treeRead else
"file"), var, "in", channel, "channel with:"
print " cut :", cut
### Create and fill MC histograms ###
# Create dict
file = {}
tree = {}
hist = {}
### Create and fill MC histograms ###
tree[sign] = TChain("tree")
for j, ss in enumerate(sample[sign]['files']):
tree[sign].Add(NTUPLEDIR + ss + ".root")
if variable[var]['nbins'] > 0:
min_value = variable[var]['min']
max_value = variable[var]['max']
title = variable[var]['title']
hist[sign] = TH1F(
sign,
";" + title + ";Events;" + ('log' if variable[var]['log'] else ''),
variable[var]['nbins'], min_value, max_value)
else:
hist[s] = TH1F(sign, ";" + variable[var]['title'],
len(variable[var]['bins']) - 1,
array('f', variable[var]['bins']))
hist[sign].Sumw2()
cutstring = "%s" % eventWeightLuminame + ("*(" + cut + ")")
tree[sign].Project(sign, var, cutstring)
if not tree[sign].GetTree() == None:
hist[sign].SetOption("%s" % tree[sign].GetTree().GetEntriesFast())
hist[sign].SetFillColor(sample[sign]['fillcolor'])
hist[sign].SetFillStyle(sample[sign]['fillstyle'])
hist[sign].SetLineColor(sample[sign]['linecolor'])
hist[sign].SetLineStyle(sample[sign]['linestyle'])
hist[sign].SetLineWidth(3)
sample[sign]['plot'] = True
# Legend
leg = TLegend(0.6, 0.6, 0.9, 0.9)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
leg.AddEntry(hist[sign], sample[sign]['label'], "fl")
leg.SetY1(0.9 - leg.GetNRows() * 0.05)
# --- Display ---
c1 = TCanvas("c1", hist.values()[0].GetXaxis().GetTitle(), 800, 800)
c1.SetTopMargin(0.06)
c1.SetRightMargin(0.05)
c1.SetTicks(1, 1)
hist[sign].Draw("HIST")
leg.Draw()
drawCMS(LUMI, year, "Preliminary")
drawRegion('XVH' + channel, True)
drawAnalysis(channel)
c1.Update()
if gROOT.IsBatch():
varname = var.replace('.', '_').replace('()', '')
if not os.path.exists(OUTPUTDIR + channel_name):
os.makedirs(OUTPUTDIR + channel_name)
c1.Print("%s" % OUTPUTDIR + channel_name + "/" + varname + sign +
".png")
c1.Print("%s" % OUTPUTDIR + channel_name + "/" + varname + sign +
".pdf")
