def test_systematics_doublesidedness(self):
if check_sys:
for sys in data.syss:
if "up" in sys:
other = sys.replace("up", "down")
self.assertIn(other, data.syss,
"'%s' sys. uncert. is missing" % other)
elif "down" in sys:
other = sys.replace("down", "up")
self.assertIn(other, data.syss,
"'%s' sys. uncert. is missing" % other)
def test_systematics_doublesidedness(self):
if check_sys:
for sys in data.syss:
if "up" in sys:
other = sys.replace("up", "down")
self.assertIn(other, data.syss,
"'%s' sys. uncert. is missing" % other)
elif "down" in sys:
other = sys.replace("down", "up")
self.assertIn(other, data.syss,
"'%s' sys. uncert. is missing" % other)
def fetch(self):
"""Creates a series of physical system objects using a template and label substitutions.
Returns:
A list pf System objects of the appropriate type and with the appropriate
properties, built by substituting placeholders in a template with the given values
Raises:
ValueError: Raised if the labels and the instance lists have mismatching lengths
"""
super(InputSysTemplate, self).fetch()
template = self.template.fetch()
labels = self.labels.fetch()
lsys = []
for (k, v) in self.extra:
if k == "instance":
ins = v.fetch()
sys = template
if len(labels) != len(ins):
raise ValueError("Labels and instance length mismatch")
for l in xrange(
len(ins)): # string replacement within the template
sys = sys.replace(labels[l], ins[l])
print "Generating system from template: \n", sys
xsys = xml_parse_string(
sys) # parses the string to an XML object
isys = InputSystem()
isys.parse(xsys.fields[0]
[1]) # parses the XML object into an InputSystem
lsys.append(isys.fetch(
)) # fetches the generated System and appends to the list
return lsys
def fetch(self):
"""Creates a series of physical system objects using a template and label substitutions.
Returns:
A list pf System objects of the appropriate type and with the appropriate
properties, built by substituting placeholders in a template with the given values
Raises:
ValueError: Raised if the labels and the instance lists have mismatching lengths
"""
super(InputSysTemplate, self).fetch()
template = self.template.fetch()
labels = self.labels.fetch()
lsys = []
for (k, v) in self.extra:
if k == "instance":
ins = v.fetch()
sys = template
if len(labels) != len(ins):
raise ValueError("Labels and instance length mismatch")
for l in xrange(len(ins)): # string replacement within the template
sys = sys.replace(labels[l], ins[l])
print "Generating system from template: \n", sys
xsys = xml_parse_string(sys) # parses the string to an XML object
isys = InputSystem()
isys.parse(xsys.fields[0][1]) # parses the XML object into an InputSystem
lsys.append(isys.fetch()) # fetches the generated System and appends to the list
return lsys
def getXmlWithDifSys(nounoId, modelList, name, cursor, sys):
doc = minidom.parse("./SystemTree/" + sys)
# get root element
root = doc.documentElement
mainpart = root.getElementsByTagName("model")
for child in mainpart:
if (child.hasAttribute("name")):
if child.getAttribute("name") not in modelList:
child.parentNode.removeChild(child)
else:
child.setAttribute(
"Chinese",
selectChinese(child.getAttribute("name"), cursor)[0])
isdele = True
while isdele:
f = codecs.open("./" + nounoId + ".xml", 'w', 'utf-8')
doc.writexml(f, addindent=' ', newl=' ', encoding='utf-8')
f.close()
doc = minidom.parse("./" + nounoId + ".xml")
root = doc.documentElement
isdele = False
parts = root.getElementsByTagName("part")
#print(len(parts))
for part in parts:
#print(len(part.childNodes))
if len(part.childNodes) == 1:
part.parentNode.removeChild(part)
isdele = True
f = codecs.open("./" + nounoId + "tmp.xml", 'w', 'utf-8')
doc.writexml(f, addindent=' ', newl=' ', encoding='utf-8')
f.close()
sys = sys.replace(".xml", "")
isExists = os.path.exists("./" + sys)
if not isExists:
os.makedirs("./" + sys)
isExists = os.path.exists("./" + sys + "_1")
if not isExists:
os.makedirs("./" + sys + "_1")
mainpart = root.getElementsByTagName("model")
if len(mainpart) != 0:
delblankline("./" + nounoId + "tmp.xml",
"./" + sys + "/" + nounoId + ".xml", name)
# delblankline("./"+nounoId+"tmp.xml","./"+sys+"_1/"+nounoId+"_1.xml",name)
# delblankline("./"+nounoId+"tmp.xml","./"+sys+"_1/"+nounoId+"_2.xml",name)
os.remove("./" + nounoId + "tmp.xml")
os.remove("./" + nounoId + ".xml")
def print_latex_full():
print '\\documentclass[11pt,oneside,a4paper]{article}'
print '\\usepackage{rotating}'
print '\\usepackage[usenames]{color}'
print '\\begin{document}'
for sys in lepsys + tausys + lepES + JES:
print '\\begin{table}'
print '\\centering'
print '\\caption{\\bf{%s}}' % sys.replace('_', '\_')
print '\\begin{tabular}{lrrrrrr}'
print row('', [group.name for group in groups], fmt='latex')
for region in [
'SR_OS', 'SR_SS', 'WCR_OS', 'WCR_SS', 'QCD_OS', 'QCD_SS'
]:
NormStatFull(sys, region)
print '\\end{tabular}'
print '\\end{table}'
print '\\end{document}'
sys_dict[sys].Add(tmp)
elif proc == "Data" or proc == "Fit":
print proc
tmp = plotting.unroll2DHistograms([hists[proc][var]])[0]
tmp.Print("all")
else:
tmp = plotting.unroll2DHistograms([hists[proc][var]])[0]
tot.Add(tmp)
if args.print_all:
tmp.Print("all")
if proc == 'QCD':
qcdHist = tmp
print "Total MC"
tot.Print("all")
print "QCD fraction"
if qcdHist is not None:
qcdFraction = qcdHist.Clone("qcdFraction")
qcdFraction.Divide(tot)
qcdFraction.Print("all")
# Compute the size of each systematic uncertainty,
# averaged across all analysis bins
for sys, hist in sys_dict.iteritems():
if 'Up' in sys:
downHist = sys_dict[sys.replace('Up', 'Down')]
hist.Add(downHist, -1)
hist.Divide(tot)
print "Systematic:", sys.replace('Up', ''),
print hist.Integral() / hist.GetNbinsX()
def book_trees(select = ''):
ordsam = OrderedDict(sorted(samples.items(), key=lambda x: x[1]['order']))
sig_chain = ROOT.TChain('sig_data')
bkg_chain = ROOT.TChain('bkg_data')
for proc in ordsam:
chainName = treename.replace('*',proc)
if ordsam[proc].get("tree","") != "":
chainName = treename.replace('*',ordsam[proc].get("tree",""))
chain = ROOT.TChain(chainName)
chainSysUp = []
chainSysDw = []
if type(samples[proc].get('name')) == type([]):
for sam in samples[proc].get('name',[]):
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chain.Add(f)
if 'signal' in str(samples[proc].get('label','')).lower():
print 'sig -->', f+'/'+treename.replace('*',proc)
sig_chain.Add(f+'/'+treename.replace('*',proc))
elif 'data' not in str(samples[proc].get('label','')).lower():
print 'bkg -->', f+'/'+treename.replace('*',proc)
bkg_chain.Add(f+'/'+treename.replace('*',proc))
if proc != 'Data':
for sys in treesUpSys:
print "debug::(",proc,")", sys, " == ", samples[proc].get('label')
chainUp = ROOT.TChain(sys.replace('*',proc))
for sam in samples[proc].get('name',[]):
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chainUp.Add(f)
chainSysUp.append(chainUp)
for sys in treesDwSys:
chainDw = ROOT.TChain(sys.replace('*',proc))
for sam in samples[proc].get('name',[]):
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chainDw.Add(f)
chainSysDw.append(chainDw)
else:
sam = samples[proc].get('name')
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chain.Add(f)
if 'signal' in str(samples[proc].get('label','')).lower():
print 'sig -->', f+'/'+treename.replace('*',proc)
sig_chain.Add(f+'/'+treename.replace('*',proc))
elif 'data' not in str(samples[proc].get('label','')).lower():
print 'bkg -->', f+'/'+treename.replace('*',proc)
bkg_chain.Add(f+'/'+treename.replace('*',proc))
if samples[proc].get('label') != 'Data':
for sys in treesUpSys:
chainUp = ROOT.TChain(sys.replace('*',proc))
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chainUp.Add(f)
chainSysUp.append(chainUp)
for sys in treesDwSys:
chainDw = ROOT.TChain(sys.replace('*',proc))
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chainDw.Add(f)
chainSysDw.append(chainDw)
# read systematic trees
samples[proc].update({'_root_tree_' : chain})
samples[proc].update({'_root_tree_sysDw_' : chainSysDw})
samples[proc].update({'_root_tree_sysUp_' : chainSysUp})
logger.info('-- sig_chain.GetEntries() = %i' % sig_chain.GetEntries())
logger.info('-- bkg_chain.GetEntries() = %i' % bkg_chain.GetEntries())
return (sig_chain, bkg_chain)
import sys as s, itertools as i
t,u,i2b,o,r=b'T',b'U',lambda i:i.to_bytes(1,'big'),[s.stdout.buffer.write,lambda s:0],range
A,e=s.argv,dict(zip((''.join(x).encode()for x in i.product(*('ACGT',)*4)),map(i2b,r(256))))
with open(A[2],'rb')as I:D,Rr=b''.join(I.read().strip().split()),lambda s:s.replace(t,u)
S,R,M,o,d=D[:-1],D[-1]&4,D[-1]&2,o[::-1]if A[1]=="d"else o,dict(zip(e.values(),e)).get
[o[0](e.get(q.replace(u,t)+(b'A'*(4-len(q)))))for q in(D[i: i+4] for i in r(0,len(D),4))]
o[0](i2b((len( [D[i: i+4] for i in r(0,len(D),4)][-1] )%4)|(4 if u in S else 0)))
o[1](b''.join(Rr(d(i2b(x)))if R else d(x)for x in S)[:-M if M else None]+b'\n')
counter = 0
for mysys in allSys:
print "===>", sys
sys = mysys + "High"
sys2 = sys
if sys.find("photon") >= 0 or sys.find("GeneratorZ") >= 0 or sys.find(
"Kappa") >= 0 or sys.find("QCD") >= 0 or sys.find(
"TopTuneA14") >= 0 or sys.find("TopRadiation") >= 0:
sys2 = "Nom"
if sys.find("QCD") >= 0:
sys2 = "Nom"
histHigh = createGraph(bkg, sys, region, bkg2, sys2, region2)
histLow = createGraph(bkg, sys.replace("High", "Low"), region, bkg2,
sys2.replace("High", "Low"), region2)
if doSym:
newHists = Symmetrize(default, histLow, histHigh)
histLow = newHists[0]
histHigh = newHists[1]
histHigh.SetLineStyle(2)
histHigh.SetLineColor(colors[counter])
histHigh.SetLineWidth(2)
histLow.SetLineStyle(2)
histLow.SetLineColor(colors[counter])
histLow.SetLineWidth(2)
# *** Exit Scipt ***
# ********************
# ********************
import sys
#sys.exit()
# Prepare table
# ----------------------------------------------------------------------
os.system('mkdir -p tablesQCD_ggH')
for proc in sysMap:
table = []
headers = ['Systematic'] + [i + 1 for i in xrange(0, 17)]
for sys in NPnames:
line = [sys.replace('_', '\_')] + [
formatCell(sysMap[proc], cat, sys) for cat in HG.CatLabels[:17]
]
table.append(line)
log = open('tablesQCD_ggH/table_%s1.tex' % proc, 'w+')
print >> log, tabulate.tabulate(table,
headers=headers,
tablefmt='latex')
print ''
print tabulate.tabulate(table, headers=headers)
table = []
headers = ['Systematic'
] + [i + 1 for i in xrange(17,
len(HG.CatLabels) + 1)]
for sys in NPnames:
def chSys(self, event):
sys = self.useSystem.GetValue().lower().strip()
sys = sys.replace("neuronales netz", "net")
sys = sys.replace("hidden markov modell", "markov")
self.system = sys
print self.system
"FitCRv8p2", "FailFitCRv8p2"
]
for region in region_list:
inFile_this.cd()
hist_nominal = inFile_this.Get(region + "__fatJet2MassSD")
outBinName = region.replace("SR",
"").replace("Fail",
"fail").replace(vbdt, "")
hist_nominal.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx])
hists_sys = []
for sys in systs:
hist_Up = inFile_this.Get(region + sys + "Up__fatJet2MassSD")
hist_Up.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" + sys.replace(vbdt, "") + "Up")
hists_sys.append(hist_Up)
hist_Down = inFile_this.Get(region + sys +
"Down__fatJet2MassSD")
hist_Down.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" + sys.replace(vbdt, "") +
"Down")
hists_sys.append(hist_Down)
hist_JMSUp = inFile_this.Get(region + "__fatJet2MassSD_JMSUp")
hist_JMSUp.SetName("histJet2Mass_" + outBinName + "_" + proc[idx] +
"_JMSUp")
hists_sys.append(hist_JMSUp)
hist_JMSDown = inFile_this.Get(region + "__fatJet2MassSD_JMSDown")
hist_JMSDown.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_JMSDown")
htmpnom = copy.deepcopy(ffid.Get(varName).Clone("nom"))
htmpnom.SetLineColor(r.kBlack)
htmpnom.SetLineWidth(2)
htmpnom.GetXaxis().SetTitle(vl.varList[varName]['xaxis'])
for key in ffid.GetListOfKeys():
if key.GetName() == varName or "nom" in key.GetName(
) or "Down" in key.GetName():
continue
sys = key.GetName().replace(varName + "_", "")
if "Up" not in key.GetName() and "Down" not in key.GetName(): continue
htmpup = copy.deepcopy(key.ReadObj().Clone(sys))
htmpdw = copy.deepcopy(
ffid.Get(key.GetName().replace("Up", "Down")).Clone(
sys.replace("Up", "Down")))
htmpup.SetLineColor(r.kBlue)
htmpdw.SetLineColor(r.kRed)
htmpup.SetLineWidth(2)
htmpdw.SetLineWidth(2)
c = r.TCanvas()
htmpnom.Draw()
htmpup.Draw("same")
htmpdw.Draw("same")
c.SaveAs('results/fiduchecks/{var}/{var}_{syst}.png'.format(
var=varName, syst=sys.replace("Up", "")))
del c, htmpup, htmpdw
ffid.Close()
del ffid, htmpnom
def systematics(sys, isShape=False, antiCorr=False):
treeRead = True
file = {}
hist = {}
tree = {}
histUp = {}
histDown = {}
up = {}
down = {}
gUp = TGraph()
gDown = TGraph()
gUp.SetLineWidth(3)
gDown.SetLineWidth(3)
gUp.SetLineColor(632)
gDown.SetLineColor(602)
BTagAK4deepup = False
BTagAK4deepdown = False
BTagAK4deep = False
BTagAK8deepup = False
BTagAK8deepdown = False
BTagAK8deep = False
# g.SetMarkerStyle(20)
# g.SetMarkerColor(418)
# g.SetMarkerSize(1.25)
var = sys
cut = 'nnbbSR'
if var == 'BTagAK4Weight_deep_up':
var = 'X_mass'
BTagAK4deepup = True
elif var == 'BTagAK4Weight_deep_down':
var = 'X_mass'
BTagAK4deepdown = True
elif var == 'BTagAK4Weight_deep':
var = 'X_mass'
BTagAK4deep = True
elif var == 'BTagAK8Weight_deep_up':
var = 'X_mass'
BTagAK8deepup = True
elif var == 'BTagAK8Weight_deep_down':
var = 'X_mass'
BTagAK8deepdown = True
elif var == 'BTagAK8Weight_deep':
var = 'X_mass'
BTagAK8deep = True
upAvg, downAvg, upMin, downMin, upMax, downMax = 0., 0., 2., 2., 0., 0.
for k in sorted(selection.keys(), key=len, reverse=True):
if k in cut:
cut = cut.replace(k, selection[k])
for i, s in enumerate(sign):
if '_MZ' in s: m = int((s.split('_MZ')[1]).split('_MA')[0])
elif '_M' in s: m = int(s.split('_M')[1])
else: m = 0
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):
tree[s].Add(NTUPLEDIR + ss + ".root")
if variable[var]['nbins'] > 0:
min_value = variable[var]['min']
max_value = variable[var]['max']
title = variable[var]['title']
if 'isZtoNN' in cut:
if var == 'MET':
min_value = 200
max_value = 2000
elif var == 'DPhi':
title = "#Delta #varphi (AK8 jet-#slash{E}_{T})"
elif var == 'VH_deltaR':
title = "#Delta R (#slash{E}_{T}, AK8 jet)"
hist[s] = TH1F(
s, ";" + title + ";Events;" +
('log' if variable[var]['log'] else ''),
variable[var]['nbins'], min_value, max_value)
else:
hist[s] = TH1F(s, ";" + variable[var]['title'],
len(variable[var]['bins']) - 1,
array('f', variable[var]['bins']))
hist[s].Sumw2()
cutstring = "(eventWeightLumi)" + ("*(" + cut + ")")
if var == 'LeptonWeightUp':
cutstring = "(eventWeightLumi * LeptonWeightUp/LeptonWeight)" + (
"*(" + cut + ")")
elif var == 'LeptonWeightDown':
cutstring = "(eventWeightLumi * LeptonWeightDown/LeptonWeight)" + (
"*(" + cut + ")")
elif var == 'TriggerWeightUp':
cutstring = "(eventWeightLumi * TriggerWeightUp/TriggerWeight)" + (
"*(" + cut + ")")
elif var == 'TriggerWeightDown':
cutstring = "(eventWeightLumi * TriggerWeightDown/TriggerWeight)" + (
"*(" + cut + ")")
#division by BTagAk4Weight_deep is because the weighted samples are used
elif BTagAK4deepup:
cutstring = "(eventWeightLumi * BTagAK4Weight_deep_up/BTagAK4Weight_deep)" + (
"*(" + cut + ")")
elif BTagAK4deepdown:
cutstring = "(eventWeightLumi * BTagAK4Weight_deep_down/BTagAK4Weight_deep)" + (
"*(" + cut + ")")
elif BTagAK8deep:
cutstring = "(eventWeightLumi * BTagAK8Weight_deep)" + (
"*(" + cut + ")")
elif BTagAK8deepup:
cutstring = "(eventWeightLumi * BTagAK8Weight_deep_up)" + (
"*(" + cut + ")")
elif BTagAK8deepdown:
cutstring = "(eventWeightLumi * BTagAK8Weight_deep_down)" + (
"*(" + cut + ")")
tree[s].Project(s, var, cutstring)
if not tree[s].GetTree() == None:
hist[s].SetOption("%s" % tree[s].GetTree().GetEntriesFast())
"""
for j, ss in enumerate(sample[s]['files']):
file[ss] = TFile(NTUPLEDIR + ss + ".root", "READ")
tmp = file[ss].Get("Sys/"+sys)
if tmp == None: continue
if not s in hist.keys(): hist[s] = tmp
else:
if antiCorr:
for x in range(1, hist[s].GetNbinsX()+1): hist[s].SetBinContent(x, hist[s].GetBinContent(x)+tmp.GetBinContent(tmp.GetNbinsX()+1-x))
else: hist[s].Add( tmp )
if isShape:
tmp = file[ss].Get("Sys/"+sys+"_up")
print "Sys/"+sys+"_up"
if tmp == None: continue
if not s in histUp.keys(): histUp[s] = tmp
else: histUp[s].Add( tmp )
#
tmp = file[ss].Get("Sys/"+sys+"_down")
if tmp == None: continue
if not s in histDown.keys(): histDown[s] = tmp
else: histDown[s].Add( tmp )
if 'accept' in sys:
norm = None
for j, ss in enumerate(sample[s]['files']):
tmp = file[ss].Get("Sys/PDF_scale")
if tmp == None: continue
if norm == None: norm = tmp
else: norm.Add( tmp )
hist[s].Divide(norm)
"""
if (isShape):
shape = TF1("shape", "gaus", 0, 5000)
shapeUp = TF1("shapeUp", "gaus", 0, 5000)
shapeDown = TF1("shapeDown", "gaus", 0, 5000)
hist[s].Fit(shape, "Q0", "")
histUp[s].Fit(shapeUp, "Q0", "")
histDown[s].Fit(shapeDown, "Q0", "")
if 'scale' in sys or 'unc' in sys:
up[s] = histUp[s].GetMean() / hist[s].GetMean()
down[s] = histDown[s].GetMean() / hist[s].GetMean()
# up[s] = shapeUp.GetParameter(1)/shape.GetParameter(1)
# down[s] = shapeDown.GetParameter(1)/shape.GetParameter(1)
elif 'res' in sys:
up[s] = histUp[s].GetRMS() / hist[s].GetRMS()
down[s] = histDown[s].GetRMS() / hist[s].GetRMS()
# up[s] = shapeUp.GetParameter(2)/shape.GetParameter(2)
# down[s] = shapeDown.GetParameter(2)/shape.GetParameter(2)
else:
up[s] = hist[s].GetBinContent(
hist[s].FindBin(+1)) / hist[s].GetBinContent(
hist[s].FindBin(0))
down[s] = hist[s].GetBinContent(
hist[s].FindBin(-1)) / hist[s].GetBinContent(
hist[s].FindBin(0))
gUp.SetPoint(i, m, up[s])
gDown.SetPoint(i, m, down[s])
#if mass < 1000: continue
upAvg += up[s]
downAvg += down[s]
if abs(up[s]) > upMax: upMax = abs(up[s])
if abs(up[s]) < upMin: upMin = abs(up[s])
if abs(down[s]) > downMax: downMax = abs(down[s])
if abs(down[s]) < downMin: downMin = abs(down[s])
upAvg /= len(sign)
downAvg /= len(sign)
print " ---", sys, "--- | up: %.3f, down: %.3f, average: %.3f" % (
upAvg, downAvg, abs(upAvg - 1 + 1. - downAvg) /
2.), "|", "^{%.1f-%.1f}_{%.1f-%.1f}" % (100. * (1. - upMin), 100. *
(1. - upMax), 100. *
(downMin - 1.), 100. *
(downMax - 1.))
c1 = TCanvas("c1", "Signals", 800, 600)
c1.cd()
c1.GetPad(0).SetTicky(2)
gUp.Draw("AL")
gDown.Draw("SAME, L")
gUp.GetYaxis().SetRangeUser(0.65, 1.35)
gUp.GetXaxis().SetTitle("m_{X} (GeV)")
gUp.GetYaxis().SetTitle("Uncertainty")
leg = TLegend(0.5, 0.90 - 0.20, 0.9, 0.90)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
leg.SetHeader(sys.replace('_', ' '))
leg.AddEntry(gUp, "+1 s. d. (%.1f%%)" % (100. * (upAvg - 1.)), "l")
leg.AddEntry(gDown, " -1 s. d. (%.1f%%)" % (100. * (1. - downAvg)), "l")
leg.Draw()
drawCMS(-1, "Simulation", False)
c1.Update()
filename = sys
if sys.startswith('W_mass') or sys.startswith('Z_mass'):
filename += "_" + sign[0][:3]
# c1.Print("plots/Systematics/"+filename+".png")
# c1.Print("plots/Systematics/"+filename+".pdf")
if not gROOT.IsBatch(): raw_input("Press Enter to continue...")
if 'doubleB' in sys or 'subjet' in sys or 'mass' in sys:
print sys + " = {",
for m in [
800, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800,
2000, 2500, 3000, 3500, 4000, 4500
]:
print "%d : [%.3f, %.3f], " % (m, gUp.Eval(m), gDown.Eval(m)),
print "}"
if 'extr' in sys or 'tagging' in sys:
print sys + " = {",
for m in [800, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500]:
print "%d : [%.3f, %.3f], " % (m, gUp.Eval(m), gDown.Eval(m)),
print "}"
if 'QCD_scale' in sys or 'PDF_scale' in sys:
print sys + " = {",
for m in range(800, 4500 + 1, 10):
print "%d : [%.3f, %.3f], " % (m, gUp.Eval(m), gDown.Eval(m)),
print "}"
else:
if vbdt == "v24":
region_list = ["SRv24Bin1", "SRv24Bin2", "SRv24Bin3", "SRv24Bin4", "FailSRv24", "FitCRv24", "FailFitCRv24"]
else:
region_list = ["SRv8p2Bin1", "SRv8p2Bin2", "SRv8p2Bin3", "FailSRv8p2", "FitCRv8p2", "FailFitCRv8p2"]
for region in region_list:
inFile_this.cd()
hist_nominal = inFile_this.Get(region+"__fatJet2MassSD")
outBinName=region.replace("SR", "").replace("Fail", "fail").replace(vbdt, "")
hist_nominal.SetName("histJet2Mass_"+outBinName+"_"+proc[idx])
hists_sys = []
for sys in systs:
hist_Up = inFile_this.Get(region+sys+"Up__fatJet2MassSD")
hist_Up.SetName("histJet2Mass_"+outBinName+"_"+proc[idx]+"_"+sys.replace(vbdt,"")+"Up")
hists_sys.append(hist_Up)
hist_Down = inFile_this.Get(region+sys+"Down__fatJet2MassSD")
hist_Down.SetName("histJet2Mass_"+outBinName+"_"+proc[idx]+"_"+sys.replace(vbdt,"")+"Down")
hists_sys.append(hist_Down)
hist_JMSUp = inFile_this.Get(region+"__fatJet2MassSD_JMSUp")
hist_JMSUp.SetName("histJet2Mass_"+outBinName+"_"+proc[idx]+"_JMSUp")
hists_sys.append(hist_JMSUp)
hist_JMSDown = inFile_this.Get(region+"__fatJet2MassSD_JMSDown")
hist_JMSDown.SetName("histJet2Mass_"+outBinName+"_"+proc[idx]+"_JMSDown")
hists_sys.append(hist_JMSDown)
hist_JMRUp = inFile_this.Get(region+"__fatJet2MassSD_JMRUp")
hist_JMRUp.SetName("histJet2Mass_"+outBinName+"_"+proc[idx]+"_JMRUp")
hists_sys.append(hist_JMRUp)
outBinName = region.replace("SR", "").replace("Fail",
"fail").replace(
vbdt, "")
hist_nominal.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx])
hists_sys = []
for sys in systs:
if idx == 0:
continue
print("hist_Up ", region + sys + "Up" + obs)
if idx == 1:
hist_Up = inFile_this.Get(histname)
hist_Up.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" +
sys.replace(vbdt, "") + "Up")
hist_Down = inFile_this.Get(histname)
hist_Down.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" +
sys.replace(vbdt, "") + "Down")
else:
hist_Up = inFile_this.Get(region + sys + "Up" + obs)
hist_Up.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" +
sys.replace(vbdt, "") + "Up")
hist_Down = inFile_this.Get(region + sys + "Down" +
obs)
hist_Down.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" +
sys.replace(vbdt, "") + "Down")
def sys_naming(sys):
if '_p' in sys:
return 'sys_'+sys.replace('_p','')
return ''
allHist.append(default)
counter=0
for mysys in allSys:
print "===>", sys
sys=mysys+"High"
sys2=sys
if sys.find("photon")>=0 or sys.find("GeneratorZ")>=0 or sys.find("Kappa")>=0 or sys.find("QCD")>=0 or sys.find("TopTuneA14")>=0 or sys.find("TopRadiation")>=0:
sys2="Nom"
if sys.find("QCD")>=0:
sys2="Nom"
histHigh=createGraph(bkg,sys,region,bkg2,sys2,region2)
histLow=createGraph(bkg,sys.replace("High","Low"),region,bkg2,sys2.replace("High","Low"),region2)
if doSym:
newHists=Symmetrize(default,histLow,histHigh)
histLow=newHists[0]
histHigh=newHists[1]
histHigh.SetLineStyle(2)
histHigh.SetLineColor(colors[counter])
histHigh.SetLineWidth(2)
histLow.SetLineStyle(2)
histLow.SetLineColor(colors[counter])
histLow.SetLineWidth(2)
hist_nominal = inFile_this.Get(region + "__fatJet2MassSD")
outBinName = region.replace("SR",
"").replace("Fail",
"fail").replace(vbdt, "")
hist_nominal.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx])
hists_sys = []
for sys in systs:
if idx == 0:
continue
print("hist_Up ", region + sys + "Up__fatJet2MassSD")
if idx == 1:
hist_Up = inFile_this.Get(region + "__fatJet2MassSD")
hist_Up.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" + sys.replace(vbdt, "") +
"Up")
hist_Down = inFile_this.Get(region + "__fatJet2MassSD")
hist_Down.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" + sys.replace(vbdt, "") +
"Down")
else:
hist_Up = inFile_this.Get(region + sys +
"Up__fatJet2MassSD")
hist_Up.SetName("histJet2Mass_" + outBinName + "_" +
proc[idx] + "_" + sys.replace(vbdt, "") +
"Up")
hist_Down = inFile_this.Get(region + sys +
"Down__fatJet2MassSD")
hist_Down.SetName("histJet2Mass_" + outBinName + "_" +
def book_trees(select = ''):
ordsam = OrderedDict(sorted(samples.items(), key=lambda x: x[1]['order']))
sig_chain = ROOT.TChain('sig_data')
bkg_chain = ROOT.TChain('bkg_data')
for proc in ordsam:
chainName = treename.replace('*',proc)
if ordsam[proc].get("tree","") != "":
chainName = treename.replace('*',ordsam[proc].get("tree",""))
chain = ROOT.TChain(chainName)
chainSysUp = []
chainSysDw = []
sysReplace = proc
if samples[proc].get("tree","") != "":
sysReplace = samples[proc].get("tree","")
if type(samples[proc].get('name')) == type([]):
for sam in samples[proc].get('name',[]):
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chain.Add(f)
if 'signal' in str(samples[proc].get('label','')).lower():
print 'sig -->', f+'/'+treename.replace('*',proc)
sig_chain.Add(f+'/'+treename.replace('*',proc))
elif 'data' not in str(samples[proc].get('label','')).lower():
print 'bkg -->', f+'/'+treename.replace('*',proc)
bkg_chain.Add(f+'/'+treename.replace('*',proc))
if proc != 'Data' and samples[proc].get('dosysts',True):
for sys in treesUpSys:
print "debug::(",proc,")", sys, " == ", samples[proc].get('label')
chainUp = ROOT.TChain(sys.replace('*',sysReplace))
for sam in samples[proc].get('name',[]):
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chainUp.Add(f)
chainSysUp.append(chainUp)
for sys in treesDwSys:
chainDw = ROOT.TChain(sys.replace('*',sysReplace))
for sam in samples[proc].get('name',[]):
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chainDw.Add(f)
chainSysDw.append(chainDw)
else:
sam = samples[proc].get('name')
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chain.Add(f)
if 'signal' in str(samples[proc].get('label','')).lower():
print 'sig -->', f+'/'+treename.replace('*',proc)
sig_chain.Add(f+'/'+treename.replace('*',proc))
elif 'data' not in str(samples[proc].get('label','')).lower():
print 'bkg -->', f+'/'+treename.replace('*',proc)
bkg_chain.Add(f+'/'+treename.replace('*',proc))
if samples[proc].get('label') != 'Data':
for sys in treesUpSys:
chainUp = ROOT.TChain(sys.replace('*',sysReplace))
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chainUp.Add(f)
chainSysUp.append(chainUp)
for sys in treesDwSys:
chainDw = ROOT.TChain(sys.replace('*',sysReplace))
for f in glob.glob( sampledir + '/*'+ sam +'*.root'):
chainDw.Add(f)
chainSysDw.append(chainDw)
# read systematic trees
samples[proc].update({'_root_tree_' : chain})
samples[proc].update({'_root_tree_sysDw_' : chainSysDw})
samples[proc].update({'_root_tree_sysUp_' : chainSysUp})
logger.info('-- sig_chain.GetEntries() = {0:d}'.format(sig_chain.GetEntries()))
logger.info('-- bkg_chain.GetEntries() = {0:d}'.format(bkg_chain.GetEntries()))
return (sig_chain, bkg_chain)
io.save_pickle(os.path.join(out_dir, "models"), R_models)
front_sorted = paretto.sort_results_by_accuracy(R_models)
models = [val[0] for val in front_sorted]
print(models)
sys = sb.SystemBuilder(verbose=False)
trigger_train_dataset = os.path.join(data_path, "train_trigger_0")
test_train_dataset = os.path.join(data_path, "test_trigger_0")
# Model 1
for im, m in enumerate(models):
smallClassifier = make.make_classifier("small", m, "trigger")
sys.replace("small", smallClassifier)
small_dict = io.read_pickle(m)
test_images = len(small_dict['test']['gt'])
train_images = len(small_dict['train']['gt'])
# For different thresholds
for th in np.arange(0 + step_th, 1, step_th):
# Data
source = make.make_source(trigger_train_dataset,
test_train_dataset, 2)
data = make.make_data("trigger_data",
train_images,
test_images,
source=source)
