def plothist(xtitle,hists,ratio=False,logy=False,norm=False):
# SETTING
outdir = ensuredir("plots/")
fname = outdir+"testPlot"
if ratio:
fname += "_ratio"
if logy:
fname += "_logy"
if norm:
fname += "_norm" # normalize each histogram
rrange = 0.5
width = 0.2 # legend width
position = 'topright' # legend position
header = "Gaussians" # legend header
text = "#mu#tau_{h}" # corner text
grid = True #and False
staterr = True and False # add uncertainty band to first histogram
lstyle = 1 # solid lines
# PLOT
LOG.header(fname)
plot = Plot(xtitle,hists,norm=norm)
plot.draw(ratio=ratio,logy=logy,ratiorange=rrange,lstyle=lstyle,grid=grid,staterr=staterr)
plot.drawlegend(position,header=header,width=width)
plot.drawtext(text)
plot.saveas(fname+".png")
plot.saveas(fname+".pdf")
#plot.saveas(fname+".C")
#plot.saveas(fname+".png",fname+".C")
#plot.saveas(fname,ext=['png','pdf'])
plot.close()
print
def checklatex(texts, tag=""):
"""Check legend entries: colors, titles, ..."""
# https://root.cern.ch/doc/master/classTPaveText.html
LOG.header("checklegend" + tag.replace('_', ' '))
output = ensuredir('plots')
fname = "%s/testLatex%s" % (output, tag)
xdim = 500
ydim = 50 * (len(texts) + 2.5)
print ">>> Canvas: %sx%s (nlines=%d)" % (xdim, ydim, len(texts))
canvas = TCanvas('canvas', 'canvas', xdim, int(ydim))
#pave1 = TPaveText(0.0,0,0.5,1,'ARC') #,'BR')
pave2 = TPaveText(0.04, 0.04, 0.96, 0.96) #'ARC') #,'BR')
#pave1.SetBorderSize(0)
pave2.SetBorderSize(0)
#pave1.SetTextAlign(12)
pave2.SetTextAlign(12)
#pave1.SetTextFont(42)
pave2.SetTextFont(42)
#pave1.SetFillColor(0)
pave2.SetFillColor(0)
#pave1.SetCornerRadius(0.05)
#pave2.SetCornerRadius(0.05)
#pave1.SetMargin(0.12)
#pave1.SetTextSize(tsize)
#pave2.Copy(pave1)
for line in texts:
latex = makelatex(line)
print ">>> %r -> %r" % (line, latex)
#pave1.AddText(line)
pave2.AddText(latex)
#pave1.Draw()
pave2.Draw()
canvas.SaveAs(fname + ".png")
#canvas.SaveAs(fname+".pdf")
canvas.Close()
def plothist(xtitle,hists,ratio=False,logy=False,norm=False,**kwargs):
# SETTING
outdir = ensuredir("plots/")
fname = outdir+"testLegend"
rrange = 0.5
header = "Gaussians" # legend header
text = "#mu#tau_{h}" # corner text
grid = True and False
staterr = True and False # add uncertainty band to first histogram
lstyle = 1 # solid lines
panel = 1 # 1 = main (top) panel, 2 = ratio (bottom) panel
hists = [h.Clone("%s_clone%d"%(h.GetName(),i)) for i, h in enumerate(hists)]
if kwargs.get('pos',False):
fname += "_"+kwargs['pos'].replace(';','')
header = kwargs['pos']
if ratio:
fname += "_ratio"
# PLOT
LOG.header(fname)
plot = Plot(xtitle,hists,norm=norm)
plot.draw(ratio=ratio,logy=logy,ratiorange=rrange,lstyle=lstyle,grid=grid,staterr=staterr)
plot.drawlegend(border=True,header=header,panel=panel,**kwargs)
plot.drawtext(text)
plot.saveas(fname+".png")
#plot.saveas(fname+".pdf")
plot.close()
print
def plotstack(xname,xtitle,datahist,exphists,ratio=False,logy=False):
"""Plot Stack objects for a given data hisogram and list of MC histograms."""
# SETTING
outdir = ensuredir("plots")
fname = "%s/testStack_%s"%(outdir,xname)
if ratio:
fname += "_ratio"
if logy:
fname += "_logy"
rrange = 0.5
text = "#mu#tau_{h} baseline"
grid = True and False
position = 'topright' if logy else 'right'
# PLOT
LOG.header(fname)
plot = Stack(xtitle,datahist,exphists)
plot.draw(ratio=ratio,logy=logy,ratiorange=rrange,grid=grid)
plot.drawlegend(position=position)
plot.drawtext(text)
plot.saveas(fname+".png")
plot.saveas(fname+".pdf")
#plot.saveas(fname+".C")
#plot.saveas(fname+".png",fname+".C")
#plot.saveas(fname,ext=['png','pdf'])
plot.close()
print
def main():
#### Test several initializations of Variable object.
#### Note that key-word arguments starting with 'c' are used for context-dependent attributes
###mvisbins = [0,30,40,50,60,70,75,80,90,100,120,200]
###variables = [
### Variable('m_vis', "m_{vis} [GeV]", 40, 0,200),
### Variable('njets', "Number of jets", 8, 0, 8),
### Variable('m_vis', 40, 0,200),
### Variable('njets', 8, 0, 8),
### Variable('njets', 8, 0, 8, veto='njets'),
### Variable('st', 20, 0,800, title="S_{T}",only="njets",blind="st>600"),
### Variable('m_vis', 40, 0,200, cbins={'njets':mvisbins},blind=(70,90)),
### Variable('m_vis', mvisbins, cbins={"njets[^&]*2":(40,0,200)},),
### Variable('pt_1+pt_2', 20, 0,800, title="S_{T}^{tautau}",ctitle={"njets":"S_{T}"}),
###]
selections = [
Selection(""),
Selection("baseline", "pt_1>50 && pt_2>50"),
Selection("njets>=1", "pt_1>50 && pt_2>50 && njets>=1"),
Selection("1 b tags",
"pt_1>50 && pt_2>50 && njets>=2 && nbtags>=1",
weight="btagweight"),
]
for selection in selections:
LOG.header(selection.name)
print ">>> name='%s', filename='%s', title='%s', cut='%s'" % (color(
selection.name), color(selection.filename), color(
selection.title), color(selection.selection))
print ">>> weight=%r, drawcmd=%r" % (selection.weight,
selection.drawcmd())
sum1 = selection + "dzeta>-40"
print '>>> sum1 = selection + "dzeta>-40"'
print ">>> name=%r, filename=%r, title=%r" % (
sum1.name, sum1.filename, sum1.title)
print ">>> cut=%r" % (sum1.selection)
sum2 = selection + Selection("dzeta", "dzeta>-40")
print '>>> sum2 = selection + Selection("dzeta","dzeta>-40")'
print ">>> name=%r, filename=%r, title=%r" % (
sum2.name, sum2.filename, sum2.title)
print ">>> cut=%r" % (sum2.selection)
def compare(fnames, variables, **kwargs):
# SETTING
tnames = kwargs.get('tree', [])
entries = kwargs.get('entries', []) # entries of files
outdir = kwargs.get('outdir', "compare")
tag = kwargs.get('tag', "")
cut = kwargs.get('cut', "")
norm = kwargs.get('norm', False)
ensuredir(outdir)
if isinstance(tnames, str):
tnames = [tnames] * len(fnames)
else:
while len(tnames) < len(fnames):
tnames.append(tnames[-1])
if norm:
tag += "_norm" # normalize each histogram
# GET FILES & TREES
files = []
trees = []
for fname, tname in zip(fnames, tnames):
file = TFile.Open(fname, 'READ')
tree = file.Get(tname)
files.append(file)
trees.append(tree)
for variable in variables[:]:
if not hasattr(tree, variable.name):
LOG.warning(
"compare: tree %s:%s does not contain branch %r! Skipping..."
% (fname, tname, variable.name))
variables.remove(variable)
while len(entries) < len(trees):
i = len(entries)
entry = "%s %d" % (tnames[i], i + 1)
entries.append(entry)
# PLOT
for variable in variables:
fname = "%s/%s%s.png" % (outdir, variable, tag)
rrange = 0.5
header = "Compare" # legend header
text = "" # corner text
ratio = True #and False
logy = True and False
grid = True #and False
staterr = True #and False # add uncertainty band to first histogram
lstyle = 1 # solid lines
LOG.header(fname)
# GET HISOGRAMS
hists = []
for i, (tree, htitle) in enumerate(zip(trees, entries)):
hname = "%s_%d" % (variable.filename, i)
#hist = variable.gethist(hname,htitle)
#dcmd = variable.drawcmd(hname)
hist = variable.draw(
tree, cut, name=hname,
title=htitle) # create and fill hist from tree
evts = hist.GetEntries()
hist.SetTitle("%s (%d)" % (htitle, evts))
hists.append(hist)
print ">>> %r: entries=%d, integral=%s, mean=%#.6g, s.d.=%#.6g" % (
htitle, evts, hist.Integral(), hist.GetMean(),
hist.GetStdDev())
# DRAW PLOT
plot = Plot(variable, hists, norm=norm)
plot.draw(ratio=True,
logy=logy,
ratiorange=rrange,
lstyle=lstyle,
grid=grid,
staterr=staterr)
plot.drawlegend(header=header)
plot.drawtext(text)
plot.saveas(fname)
plot.close()
print
def compare(fnames, varsets, **kwargs):
# SETTING
tname = kwargs.get('tree', 'Events')
entries = kwargs.get('entries', []) # entries of files
outdir = kwargs.get('outdir', "compare")
tag = kwargs.get('tag', "")
cut = kwargs.get('cut', "")
norm = kwargs.get('norm', False)
ensuredir(outdir)
if norm:
tag += "_norm" # normalize each histogram
tree = TChain(tname)
# GET FILES & TREES
for fname in fnames:
tree.Add(fname)
# PLOT
for varset in varsets:
fname = "%s/%s%s.png" % (outdir, '-'.join(v.filename
for v in varset), tag)
rrange = 0.5
header = "Compare" # legend header
text = "" # corner text
ratio = True #and False
logy = True and False
grid = True #and False
staterr = True #and False # add uncertainty band to first histogram
lstyle = 1 # solid lines
xvar = varset[0]
LOG.header(fname)
# GET HISOGRAMS
hists = []
for i, variable in enumerate(varset):
hname = "%s_%d" % (variable.filename, i)
htitle = variable.name
#hist = variable.gethist(hname,htitle)
#dcmd = variable.drawcmd(hname)
hist = variable.draw(
tree, cut, name=hname,
title=htitle) # create and fill hist from tree
evts = hist.GetEntries()
hist.SetTitle(htitle) #"%s (%d)"%(htitle,evts)
hists.append(hist)
print ">>> %r: entries=%d, integral=%s, mean=%#.6g, s.d.=%#.6g" % (
htitle, evts, hist.Integral(), hist.GetMean(),
hist.GetStdDev())
# DRAW PLOT
plot = Plot(xvar, hists, norm=norm)
plot.draw(ratio=True,
logy=logy,
ratiorange=rrange,
lstyle=lstyle,
grid=grid,
staterr=staterr)
plot.drawlegend(header=header, latex=False)
plot.drawtext(text)
plot.saveas(fname)
plot.close()
print
def main(args):
channels = args.channels
eras = args.eras
parallel = args.parallel
verbosity = args.verbosity
obsset = args.observables
plot = True
analysis = 'ztt_tid' # $PROCESS_$ANALYSIS
tag = ""
fileexp = "$OUTDIR/$ANALYSIS_$OBS_$CHANNEL-$ERA$TAG.inputs.root"
outdir = ensuredir("input")
plotdir = "input/plots/$ERA"
for era in eras:
for channel in channels:
for obs in obsset:
PLOG.header("%s, %s, %s"%(obs,channel,era))
splitbydm = 'mtau' in obs and 'mumu' not in channel
###############
# SAMPLES #
###############
# sample set and their systematic variations
# GET SAMPLESET
join = ['VV','TT','ST']
sname = "$PICODIR/$SAMPLE_$CHANNEL$TAG.root"
sampleset = getsampleset(channel,era,fname=sname,join=join,split=None,table=False,verb=1)
if channel=='mumu':
# RENAME (HTT convention)
sampleset.rename('DY_M50','ZLL')
sampleset.rename('WJ','W')
sampleset.datasample.name = 'data_obs'
# SYSTEMATIC VARIATIONS
varprocs = { # processes to be varied
'Nom': ['ZLL','W','VV','ST','TT','QCD','data_obs'],
}
samplesets = { # sets of samples per variation
'Nom': sampleset, # nominal
}
samplesets['Nom'].printtable(merged=True,split=True)
if verbosity>=2:
samplesets['Nom'].printobjs(file=True)
else:
# SPLIT & RENAME (HTT convention)
GMR = "genmatch_2==5"
GML = "genmatch_2>0 && genmatch_2<5"
GMJ = "genmatch_2==0"
GMF = "genmatch_2<5"
if splitbydm:
idweights = getdmsf(era) # DM-dependent SF, pT > 20 GeV
print idweights
sampleset.split('DY',[('ZTT_DM0', GMR+" && dm_2==0"), ('ZTT_DM1', GMR+" && dm_2==1"),
('ZTT_DM10',GMR+" && dm_2==10"),('ZTT_DM11',GMR+" && dm_2==11"),
('ZL',GML),('ZJ',GMJ)])
sampleset.get('DY',unique=True).replaceweight('idweight_2',idweights[0],verb=2) # replace tau ID SF
else:
idweights = ('idweight_2', 'idweightUp_2', 'idweightDown_2') # pT-dependent SF
sampleset.split('DY',[('ZTT',GMR),('ZL',GML),('ZJ',GMJ),])
sampleset.split('TT',[('TTT',GMR),('TTL',GML),('TTJ',GMJ)])
#sampleset.split('ST',[('STT',GMR),('STJ',GMF),]) # small background
sampleset.rename('WJ','W')
sampleset.datasample.name = 'data_obs'
# SYSTEMATIC VARIATIONS
systs = preparesysts( # prepare systematic variations: syskey, systag, procs
('Nom',"", ['ZTT','ZL','ZJ','W','VV','ST','TTT','TTL','TTJ','QCD','data_obs']), #,'STT','STJ'
('TES',"_shape_tes",['ZTT','TTT']), # tau energy scale
('LTF',"_shape_ltf",['ZL', 'TTL']), # l -> tau energy scale
('JTF',"_shape_jtf",['ZJ', 'TTJ','QCD','W']), # j -> tau energy scale
('tid',"_shape_tid",['ZTT','TTT'], idweights), # replace 'idweight_2' with up/down variations
('zpt',"_shape_dy", ['ZTT','ZL','ZJ'], ('zptweight','(1.1*zptweight-0.1)','(0.9*zptweight+0.1)')), # replace 'zptweight'
ERA=era,CHANNEL=channel) # keys to replace in systag
if splitbydm:
for syskey, syst in systs.iteritems():
if 'ZTT' in syst.procs: # split all ZTT shapes
syst.procs = ['ZTT_DM0','ZTT_DM1','ZTT_DM10','ZTT_DM11']+syst.procs[1:]
samplesets = { # sets of samples per variation
'Nom': sampleset, # nominal
'TESUp': sampleset.shift(systs['TES'].procs,"_TES1p030",systs['TES'].up," +3% TES", split=True,filter=False,share=True),
'TESDown': sampleset.shift(systs['TES'].procs,"_TES0p970",systs['TES'].dn," -3% TES", split=True,filter=False,share=True),
#'TESUp': sampleset.shift(systs['TES'].procs,"_TESUp", systs['TES'].up," +3% TES", split=True,filter=False,share=True),
#'TESDown': sampleset.shift(systs['TES'].procs,"_TESDown",systs['TES'].dn," -3% TES", split=True,filter=False,share=True),
'LTFUp': sampleset.shift(systs['LTF'].procs,"_LTF1p030",systs['LTF'].up," +3% LTF", split=True,filter=False,share=True),
'LTFDown': sampleset.shift(systs['LTF'].procs,"_LTF0p970",systs['LTF'].dn," -3% LTF", split=True,filter=False,share=True),
'JTFUp': sampleset.shift(systs['JTF'].procs,"_JTF1p100",systs['JTF'].up," +10% JTF",split=True,filter=False,share=True),
'JTFDown': sampleset.shift(systs['JTF'].procs,"_JTF0p900",systs['JTF'].dn," -10% JTF",split=True,filter=False,share=True),
}
keys = samplesets.keys() if verbosity>=1 else ['Nom','TESUp','TESDown']
for shift in keys: # print table
if not shift in samplesets: continue
samplesets[shift].printtable(merged=True,split=True)
if verbosity>=2:
samplesets[shift].printobjs(file=True)
###################
# OBSERVABLES #
###################
# observable/variables to be fitted in combine
if channel=='mumu':
observables = [
Var('m_vis', 1, 60, 120, fname='mvis', ymargin=1.6, rrange=0.08),
]
elif splitbydm:
observables = [
Var('dm_2==0?0.1396:m_2', "m_tau", 18, 0, 1.8, fname='mtau'),
]
else:
mvis = Var('m_vis', 30, 50, 200, fname='mvis')
observables = [
#Var('m_vis', 30, 50, 200, fname='mvis'),
Var('m_vis', 38, 10, 200, fname='mvis'), # broad range
#Var('m_vis', 15, 50, 200, tag="_10"), # coarser binning
]
############
# BINS #
############
# selection categories
if channel=='mumu':
baseline = "q_1*q_2<0 && iso_1<0.15 && iso_2<0.15 && !lepton_vetoes && metfilter"
bins = [
Sel('ZMM', baseline),
]
else:
tauwps = ['VVVLoose','VVLoose','VLoose','Loose','Medium','Tight','VTight','VVTight']
tauwpbits = { wp: 2**i for i, wp in enumerate(tauwps)}
tauwps = ['Tight'] # only do Tight
iso_1 = "iso_1<0.15"
iso_2 = "idDecayModeNewDMs_2 && idDeepTau2017v2p1VSjet_2>=$WP && idDeepTau2017v2p1VSe_2>=2 && idDeepTau2017v2p1VSmu_2>=8"
baseline = "q_1*q_2<0 && %s && %s && !lepton_vetoes_notau && metfilter"%(iso_1,iso_2)
zttregion = "%s && mt_1<60 && dzeta>-25 && abs(deta_ll)<1.5"%(baseline)
bins = [
#Sel('baseline', repkey(baseline,WP=16)),
#Sel('zttregion',repkey(zttregion,WP=16)),
]
for wpname in tauwps: # loop over tau ID WPs
wpbit = tauwpbits[wpname]
bins.append(Sel(wpname,repkey(zttregion,WP=wpbit)))
#######################
# DATACARD INPUTS #
#######################
# histogram inputs for the datacards
# https://twiki.cern.ch/twiki/bin/viewauth/CMS/SMTauTau2016
chshort = channel.replace('tau','t').replace('mu','m') # abbreviation of channel
fname = repkey(fileexp,OUTDIR=outdir,ANALYSIS=analysis,CHANNEL=chshort,ERA=era,TAG=tag)
createinputs(fname,samplesets['Nom'],observables,bins,recreate=True)
if channel in ['mutau']:
createinputs(fname,samplesets['TESUp'], observables,bins,systs['TES'].up,filter=systs['TES'].procs)
createinputs(fname,samplesets['TESDown'],observables,bins,systs['TES'].dn,filter=systs['TES'].procs)
createinputs(fname,samplesets['LTFUp'], observables,bins,systs['LTF'].up,filter=systs['LTF'].procs)
createinputs(fname,samplesets['LTFDown'],observables,bins,systs['LTF'].dn,filter=systs['LTF'].procs)
createinputs(fname,samplesets['JTFUp'], observables,bins,systs['JTF'].up,filter=systs['JTF'].procs)
createinputs(fname,samplesets['JTFDown'],observables,bins,systs['JTF'].dn,filter=systs['JTF'].procs)
createinputs(fname,samplesets['Nom'],observables,bins,systs['tid'].up,filter=systs['tid'].procs,replaceweight=systs['tid'].wgtup)
createinputs(fname,samplesets['Nom'],observables,bins,systs['tid'].dn,filter=systs['tid'].procs,replaceweight=systs['tid'].wgtdn)
createinputs(fname,samplesets['Nom'],observables,bins,systs['zpt'].up,filter=systs['zpt'].procs,replaceweight=systs['zpt'].wgtup)
createinputs(fname,samplesets['Nom'],observables,bins,systs['zpt'].dn,filter=systs['zpt'].procs,replaceweight=systs['zpt'].wgtdn)
############
# PLOT #
############
# control plots of the histogram inputs
if plot:
plotdir_ = ensuredir(repkey(plotdir,ERA=era,CHANNEL=channel))
pname = repkey(fileexp,OUTDIR=plotdir_,ANALYSIS=analysis,CHANNEL=chshort+"-$BIN",ERA=era,TAG='$TAG'+tag).replace('.root','.png')
text = "%s: $BIN"%(channel.replace("mu","#mu").replace("tau","#tau_{h}"))
groups = [ ] #(['^TT','ST'],'Top'),]
plotinputs(fname,systs,observables,bins,text=text,
pname=pname,tag=tag,group=groups)
def main():
mvisbins = [0, 30, 40, 50, 60, 70, 75, 80, 90, 100, 120, 200]
# Test several initializations of Variable object.
# Note that key-word arguments starting with 'c' are used for context-dependent attributes
variables = [
Variable('m_vis', "m_{vis} [GeV]", 40, 0, 200),
Variable('njets', "Number of jets", 8, 0, 8),
Variable('m_vis', 40, 0, 200),
Variable('njets', 8, 0, 8),
Variable('njets', 8, 0, 8, veto='njets'),
Variable('st', 20, 0, 800, title="S_{T}", only="njets",
blind="st>600"),
Variable('m_vis',
40,
0,
200,
cbins={'njets': mvisbins},
blind=(70, 90)),
Variable(
'm_vis',
mvisbins,
cbins={"njets[^&]*2": (40, 0, 200)},
),
Variable('pt_1+pt_2',
20,
0,
800,
title="S_{T}^{tautau}",
ctitle={"njets": "S_{T}"}),
]
selections = [
"pt_1>50 && pt_2>50",
"pt_1>50 && pt_2>50 && njets>=1",
"pt_1>50 && pt_2>50 && njets>=2 && nbtags>=1",
"pt_1>50 && pt_2>50 && njets==0",
"pt_1>50 && pt_2>50 && dzeta>-40",
]
for var in variables:
LOG.header(var.name)
print ">>> string=%s, repr=%r" % (var, var)
print ">>> name='%s', filename='%s', title='%s'" % (color(
var.name), color(var.filename), color(var.title))
print ">>> (nbins,xmin,xmax)=(%s,%s,%s), bins=%s" % (
var.nbins, var.xmin, var.xmax, var.bins)
print ">>> hasintbins=%s, hasvariablebins=%s" % (var.hasintbins(),
var.hasvariablebins())
print ">>> cut=%r, blindcuts=%r, blind(50,60)=%r" % (
var.cut, var.blindcuts, var.blind(50, 60))
hist = var.gethist()
print ">>> hist=%s, (nbins,xmin,xmax)=(%s,%s,%s), variable=%s" % (
hist, hist.GetXaxis().GetNbins(), hist.GetXaxis().GetXmin(),
hist.GetXaxis().GetXmax(), hist.GetXaxis().IsVariableBinSize())
gDirectory.Delete(hist.GetName())
for sel in selections: # context-dependent attributes
var.changecontext(sel)
print ">>> context: '%s'" % color(sel, 'grey')
print ">>> plotfor=%s, name='%s', title='%s'" % (
var.plotfor(sel), color(var.name), color(var.title))
print ">>> (nbins,xmin,xmax)=(%s,%s,%s), bins=%s" % (
var.nbins, var.xmin, var.xmax, var.bins)
print