def get_list_of_syst_to_fill(opts):
systematics = ['NOM']
sysOption = opts.syst
excludedSyst = opts.exclude
anySys = sysOption == None
if sysOption == 'fake' or anySys:
systematics += systUtils.fakeSystVariations()
if sysOption == 'object' or anySys:
systematics += systUtils.mcObjectVariations()
if sysOption == 'weight' or anySys:
systematics += systUtils.mcWeightVariations()
if sysOption and sysOption.count(','):
systematics = [
s for s in systUtils.getAllVariations()
if s in sysOption.split(',')
]
elif sysOption in systUtils.getAllVariations():
systematics = [sysOption]
elif not anySys and len(systematics) == 1 and sysOption != 'NOM':
raise ValueError("Invalid syst %s" % str(sysOption))
if excludedSyst:
systematics = [
s for s in systematics
if s not in filterWithRegexp(systematics, excludedSyst)
]
return systematics
def setSyst(self, sys='NOM') :
nominal = 'NOM' # do we have differnt names for nom (mc vs fake)?
self.isObjSys = sys in systUtils.mcObjectVariations()
self.isWeightSys = sys in systUtils.mcWeightVariations()
self.isFakeSys = sys in systUtils.fakeSystVariations()
def nameObjectSys(s) : return s if self.isMc else nominal
def nameWeightSys(s) : return s if self.isMc else nominal
def nameFakeSys(s) : return s if self.isFake else nominal
def identity(s) : return s
sysNameFunc = nameObjectSys if self.isObjSys else nameWeightSys if self.isWeightSys else nameFakeSys if self.isFakeSys else identity
self.syst = sysNameFunc(sys)
return self
def exploreAvailableSystematics(self, verbose=False) :
systs = ['NOM']
if self.isFake :
systs += systUtils.fakeSystVariations()
elif self.isMc :
systs += systUtils.mcObjectVariations()
systs += systUtils.mcWeightVariations()
self.systematics = []
for sys in systs :
self.setSyst(sys)
if os.path.exists(self.filenameHisto) :
self.systematics.append(sys)
if verbose : print "%s : found %d variations : %s"%(self.name, len(self.systematics), str(self.systematics))
def runFill(opts) :
batchMode = opts.batch
inputFakeDir = opts.input_fake
inputGenDir = opts.input_gen
outputDir = opts.output_dir
sysOption = opts.syst
excludedSyst = opts.exclude
verbose = opts.verbose
if verbose : print "filling histos"
mkdirIfNeeded(outputDir)
systematics = ['NOM']
anySys = sysOption==None
if sysOption=='fake' or anySys : systematics += systUtils.fakeSystVariations()
if sysOption=='object' or anySys : systematics += systUtils.mcObjectVariations()
if sysOption=='weight' or anySys : systematics += systUtils.mcWeightVariations()
if sysOption and sysOption.count(',') : systematics = [s for s in systUtils.getAllVariations() if s in sysOption.split(',')]
elif sysOption in systUtils.getAllVariations() : systematics = [sysOption]
elif not anySys and len(systematics)==1 and sysOption!='NOM' : raise ValueError("Invalid syst %s"%str(sysOption))
if excludedSyst : systematics = [s for s in systematics if s not in filterWithRegexp(systematics, excludedSyst)]
if verbose : print "about to loop over these systematics:\n %s"%str(systematics)
for syst in systematics :
if batchMode :
newOptions = " --input-gen %s" % opts.input_gen
newOptions += " --input-fake %s" % opts.input_fake
newOptions += " --output-dir %s" % opts.output_dir
newOptions += " --verbose %s" % opts.verbose
newOptions += " --syst %s" % syst
template = 'batch/templates/check_hft_fill.sh.template'
script = "batch/hft_%s.sh"%syst
scriptFile = open(script, 'w')
scriptFile.write(open(template).read()
.replace('%(opt)s', newOptions)
.replace('%(logfile)s', 'log/hft/fill_'+syst+'.log')
.replace('%(jobname)s', 'fill_'+syst))
scriptFile.close()
cmd = "sbatch %s"%script
if verbose : print cmd
out = getCommandOutput(cmd)
if verbose : print out['stdout']
if out['stderr'] : print out['stderr']
continue
if verbose : print '---- filling ',syst
samplesPerGroup = allSamplesAllGroups()
[s.setSyst(syst) for g, samples in samplesPerGroup.iteritems() for s in samples]
counters, histos = countAndFillHistos(samplesPerGroup=samplesPerGroup, syst=syst, verbose=verbose, outdir=outputDir)
printCounters(counters)
saveHistos(samplesPerGroup, histos, outputDir, verbose)
def get_list_of_syst_to_fill(opts):
systematics = ['NOM']
sysOption = opts.syst
excludedSyst = opts.exclude
anySys = sysOption==None
if sysOption=='fake' or anySys : systematics += systUtils.fakeSystVariations()
if sysOption=='object' or anySys : systematics += systUtils.mcObjectVariations()
if sysOption=='weight' or anySys : systematics += systUtils.mcWeightVariations()
if sysOption and sysOption.count(','):
systematics = [s for s in systUtils.getAllVariations() if s in sysOption.split(',')]
elif sysOption in systUtils.getAllVariations(): systematics = [sysOption]
elif not anySys and len(systematics)==1 and sysOption!='NOM':
raise ValueError("Invalid syst %s"%str(sysOption))
if excludedSyst:
systematics = [s for s in systematics if s not in filterWithRegexp(systematics, excludedSyst)]
return systematics
def runPlot(opts) :
inputDir = opts.input_dir
outputDir = opts.output_dir
verbose = opts.verbose
mkdirIfNeeded(outputDir)
buildTotBkg = systUtils.buildTotBackgroundHisto
buildStat = systUtils.buildStatisticalErrorBand
buildSyst = systUtils.buildSystematicErrorBand
selections = regions_to_plot(opts.include_regions, opts.exclude_regions, opts.regions)
variables = variables_to_plot()
groups = dataset.DatasetGroup.build_groups_from_files_in_dir(opts.samples_dir)
groups.append(first([g for g in groups if g.is_data]).clone_data_as_fake())
if not skip_charge_flip : groups.append(dataset.DatasetGroup.build_qflip_from_simulated_samples(groups))
plot_groups = [systUtils.Group(g.name) for g in groups]
sel_not_specified = len(regions_to_plot())==len(selections)
if sel_not_specified:
selections = guess_available_selections_from_histofiles(inputDir, first(plot_groups), verbose)
systematics_to_use = get_list_of_syst_to_fill(opts)
for group in plot_groups :
group.setCurrentSelection(first(selections))
group.setHistosDir(inputDir).setCurrentSelection(first(selections))
group.exploreAvailableSystematics(verbose)
group.filterAndDropSystematics(systematics_to_use, opts.exclude, verbose)
available_systematics = sorted(list(set([s for g in plot_groups for s in g.systematics])))
systematics = [s for s in systematics_to_use if s in available_systematics]
if verbose :
print "using the following systematics : {0}".format(systematics)
print "missing the following systematics : {0}".format([s for s in systematics_to_use if s not in available_systematics])
fakeSystematics = [s for s in systematics if s in systUtils.fakeSystVariations()]
mcSystematics = [s for s in systematics if s in systUtils.mcObjectVariations() + systUtils.mcWeightVariations()]
mkdirIfNeeded(outputDir)
findByName = systUtils.findByName
simBkgs = [g for g in plot_groups if g.isMcBkg]
data = findByName(plot_groups, 'data')
fake = findByName(plot_groups, 'fake')
signal = findByName(plot_groups, 'signaltaumu')
print 'names_stacked_groups to be improved'
names_stacked_groups = [g.name for g in simBkgs+[fake]]
for sel in selections :
if verbose : print '-- plotting ',sel
for var in variables :
if verbose : print '---- plotting ',var
print_summary_yield = var is 'onebin'
for g in plot_groups :
g.setSystNominal()
g.setCurrentSelection(sel)
nominalHistoData = data.getHistogram(variable=var, selection=sel, cacheIt=True)
nominalHistoSign = signal.getHistogram(variable=var, selection=sel, cacheIt=True)
nominalHistoFakeBkg = fake.getHistogram(variable=var, selection=sel, cacheIt=True)
nominalHistosSimBkg = dict([(g.name, g.getHistogram(variable=var, selection=sel, cacheIt=True))
for g in simBkgs])
nominalHistosBkg = dict([('fake', nominalHistoFakeBkg)] +
[(g, h) for g, h in nominalHistosSimBkg.iteritems()])
nominalHistoTotBkg = buildTotBkg(histoFakeBkg=nominalHistoFakeBkg,
histosSimBkgs=nominalHistosSimBkg)
statErrBand = buildStat(nominalHistoTotBkg)
systErrBand = buildSyst(fake=fake, simBkgs=simBkgs, variable=var, selection=sel,
fakeVariations=fakeSystematics, mcVariations=mcSystematics,
verbose=verbose, printYield=print_summary_yield)
# if print_summary_yield:
# print_stat_syst_yield(fake=fake, variable=var, selection=sel, fakeVariations=fakeSystematics)
plotHistos(histoData=nominalHistoData, histoSignal=nominalHistoSign,
histoTotBkg=nominalHistoTotBkg, histosBkg=nominalHistosBkg,
statErrBand=statErrBand, systErrBand=systErrBand,
stack_order=names_stacked_groups,
topLabel=sel,
canvasName=(sel+'_'+var), outdir=outputDir, options=opts,
printYieldSummary=print_summary_yield)
for group in plot_groups :
group.printVariationsSummary()
def runPlot(opts):
inputDir = opts.input_dir
outputDir = opts.output_dir
verbose = opts.verbose
mkdirIfNeeded(outputDir)
buildTotBkg = systUtils.buildTotBackgroundHisto
buildStat = systUtils.buildStatisticalErrorBand
buildSyst = systUtils.buildSystematicErrorBand
selections = regions_to_plot(opts.include_regions, opts.exclude_regions,
opts.regions)
variables = variables_to_plot()
groups = dataset.DatasetGroup.build_groups_from_files_in_dir(
opts.samples_dir)
groups.append(first([g for g in groups if g.is_data]).clone_data_as_fake())
if not skip_charge_flip:
groups.append(
dataset.DatasetGroup.build_qflip_from_simulated_samples(groups))
plot_groups = [systUtils.Group(g.name) for g in groups]
sel_not_specified = len(regions_to_plot()) == len(selections)
if sel_not_specified:
selections = guess_available_selections_from_histofiles(
inputDir, first(plot_groups), verbose)
systematics_to_use = get_list_of_syst_to_fill(opts)
for group in plot_groups:
group.setCurrentSelection(first(selections))
group.setHistosDir(inputDir).setCurrentSelection(first(selections))
group.exploreAvailableSystematics(verbose)
group.filterAndDropSystematics(systematics_to_use, opts.exclude,
verbose)
available_systematics = sorted(
list(set([s for g in plot_groups for s in g.systematics])))
systematics = [s for s in systematics_to_use if s in available_systematics]
if verbose:
print "using the following systematics : {0}".format(systematics)
print "missing the following systematics : {0}".format(
[s for s in systematics_to_use if s not in available_systematics])
fakeSystematics = [
s for s in systematics if s in systUtils.fakeSystVariations()
]
mcSystematics = [
s for s in systematics if s in systUtils.mcObjectVariations() +
systUtils.mcWeightVariations()
]
mkdirIfNeeded(outputDir)
findByName = systUtils.findByName
simBkgs = [g for g in plot_groups if g.isMcBkg]
data = findByName(plot_groups, 'data')
fake = findByName(plot_groups, 'fake')
signal = findByName(plot_groups, 'signaltaumu')
print 'names_stacked_groups to be improved'
names_stacked_groups = [g.name for g in simBkgs + [fake]]
for sel in selections:
if verbose: print '-- plotting ', sel
for var in variables:
if verbose: print '---- plotting ', var
print_summary_yield = var is 'onebin'
for g in plot_groups:
g.setSystNominal()
g.setCurrentSelection(sel)
nominalHistoData = data.getHistogram(variable=var,
selection=sel,
cacheIt=True)
nominalHistoSign = signal.getHistogram(variable=var,
selection=sel,
cacheIt=True)
nominalHistoFakeBkg = fake.getHistogram(variable=var,
selection=sel,
cacheIt=True)
nominalHistosSimBkg = dict([(g.name,
g.getHistogram(variable=var,
selection=sel,
cacheIt=True))
for g in simBkgs])
nominalHistosBkg = dict(
[('fake', nominalHistoFakeBkg)] +
[(g, h) for g, h in nominalHistosSimBkg.iteritems()])
nominalHistoTotBkg = buildTotBkg(histoFakeBkg=nominalHistoFakeBkg,
histosSimBkgs=nominalHistosSimBkg)
statErrBand = buildStat(nominalHistoTotBkg)
systErrBand = buildSyst(fake=fake,
simBkgs=simBkgs,
variable=var,
selection=sel,
fakeVariations=fakeSystematics,
mcVariations=mcSystematics,
verbose=verbose,
printYield=print_summary_yield)
# if print_summary_yield:
# print_stat_syst_yield(fake=fake, variable=var, selection=sel, fakeVariations=fakeSystematics)
plotHistos(histoData=nominalHistoData,
histoSignal=nominalHistoSign,
histoTotBkg=nominalHistoTotBkg,
histosBkg=nominalHistosBkg,
statErrBand=statErrBand,
systErrBand=systErrBand,
stack_order=names_stacked_groups,
topLabel=sel,
canvasName=(sel + '_' + var),
outdir=outputDir,
options=opts,
printYieldSummary=print_summary_yield)
for group in plot_groups:
group.printVariationsSummary()
def isNeededForSys(self, sys) :
return (sys=='NOM'
or (self.isMc and sys in systUtils.mcWeightVariations())
or (self.isMc and sys in systUtils.mcObjectVariations())
or (self.isFake and sys in systUtils.fakeSystVariations()))
def countAndFillHistos(samplesPerGroup={}, syst='', verbose=False, outdir='./') :
selections = allRegions()
variables = variablesToPlot()
mcGroups, fakeGroups = mcDatasetids().keys(), ['fake']
objVariations, weightVariations, fakeVariations = systUtils.mcObjectVariations(), systUtils.mcWeightVariations(), systUtils.fakeSystVariations()
def groupIsRelevantForSys(g, s) :
isRelevant = (s=='NOM' or (g in mcGroups and s in objVariations+weightVariations) or (g in fakeGroups and s in fakeVariations))
if verbose and not isRelevant : print "skipping %s for %s"%(g, s)
return isRelevant
def dropIrrelevantGroupsForThisSys(groups, sys) : return dict((g, samples) for g, samples in groups.iteritems() if groupIsRelevantForSys(g, syst))
def dropSamplesWithoutTree(samples) : return [s for s in samples if s.hasInputHftTree(msg='Warning! ')]
def dropGroupsWithoutSamples(groups) : return dict((g, samples) for g, samples in groups.iteritems() if len(samples))
samplesPerGroup = dropIrrelevantGroupsForThisSys(samplesPerGroup, syst)
samplesPerGroup = dict((g, dropSamplesWithoutTree(samples)) for g, samples in samplesPerGroup.iteritems())
samplesPerGroup = dropGroupsWithoutSamples(samplesPerGroup)
groups = samplesPerGroup.keys()
counters = bookCounters(groups, selections)
histos = bookHistos(variables, groups, selections)
for group, samplesGroup in samplesPerGroup.iteritems() :
logLine = "---->"
if verbose : print 1*' ',group
histosGroup = histos [group]
countsGroup = counters[group]
for sample in samplesGroup :
if verbose : logLine +=" %s"%sample.name
fillAndCount(histosGroup, countsGroup, sample, blind=False)
if verbose : print logLine
if verbose : print 'done'
return counters, histos
def runPlot(opts) :
inputDir = opts.input_dir
outputDir = opts.output_dir
sysOption = opts.syst
excludedSyst = opts.exclude
verbose = opts.verbose
mkdirIfNeeded(outputDir)
buildTotBkg = systUtils.buildTotBackgroundHisto
buildStat = systUtils.buildStatisticalErrorBand
buildSyst = systUtils.buildSystematicErrorBand
groups = allGroups()
selections = allRegions()
variables = variablesToPlot()
for group in groups :
group.setHistosDir(inputDir)
group.exploreAvailableSystematics(verbose)
group.filterAndDropSystematics(sysOption, excludedSyst, verbose)
mkdirIfNeeded(outputDir)
systematics = ['NOM']
anySys = sysOption==None
if sysOption=='fake' or anySys : systematics += systUtils.fakeSystVariations()
if sysOption=='object' or anySys : systematics += systUtils.mcObjectVariations()
if sysOption=='weight' or anySys : systematics += systUtils.mcWeightVariations()
if sysOption and sysOption.count(',') : systematics = [s for s in systUtils.getAllVariations() if s in sysOption.split(',')]
elif sysOption in systUtils.getAllVariations() : systematics = [sysOption]
if not anySys and len(systematics)==1 and sysOption!='NOM' : raise ValueError("Invalid syst %s"%str(sysOption))
if excludedSyst : systematics = [s for s in systematics if s not in filterWithRegexp(systematics, excludedSyst)]
if verbose : print "using the following systematics : %s"%str(systematics)
fakeSystematics = [s for s in systematics if s in systUtils.fakeSystVariations()]
mcSystematics = [s for s in systematics if s in systUtils.mcObjectVariations() + systUtils.mcWeightVariations()]
simBkgs = [g for g in groups if g.isMcBkg]
data, fake, signal = findByName(groups, 'data'), findByName(groups, 'fake'), findByName(groups, 'signal')
for sel in selections :
if verbose : print '-- plotting ',sel
for var in variables :
if verbose : print '---- plotting ',var
for g in groups : g.setSystNominal()
nominalHistoData = data.getHistogram(variable=var, selection=sel, cacheIt=True)
nominalHistoSign = signal.getHistogram(variable=var, selection=sel, cacheIt=True)
nominalHistoFakeBkg = fake.getHistogram(variable=var, selection=sel, cacheIt=True)
nominalHistosSimBkg = dict([(g.name, g.getHistogram(variable=var, selection=sel, cacheIt=True)) for g in simBkgs])
nominalHistosBkg = dict([('fake', nominalHistoFakeBkg)] + [(g, h) for g, h in nominalHistosSimBkg.iteritems()])
nominalHistoTotBkg = buildTotBkg(histoFakeBkg=nominalHistoFakeBkg, histosSimBkgs=nominalHistosSimBkg)
statErrBand = buildStat(nominalHistoTotBkg)
systErrBand = buildSyst(fake=fake, simBkgs=simBkgs, variable=var, selection=sel,
fakeVariations=fakeSystematics, mcVariations=mcSystematics, verbose=verbose)
plotHistos(histoData=nominalHistoData, histoSignal=nominalHistoSign, histoTotBkg=nominalHistoTotBkg,
histosBkg=nominalHistosBkg,
statErrBand=statErrBand, systErrBand=systErrBand,
canvasName=(sel+'_'+var), outdir=outputDir, verbose=verbose)
for group in groups :
summary = group.variationsSummary()
for selection, summarySel in summary.iteritems() :
colW = str(12)
header = ' '.join([('%'+colW+'s')%colName for colName in ['variation', 'yield', 'delta[%]']])
lineTemplate = '%(sys)'+colW+'s'+'%(counts)'+colW+'s'+'%(delta)'+colW+'s'
print "---- summary of variations for %s ----" % group.name
print "--- %s ---" % selection
print header
print '\n'.join(lineTemplate%{'sys':s,
'counts':(("%.3f"%c) if type(c) is float else (str(c)+str(type(c)))),
'delta' :(("%.3f"%d) if type(d) is float else '--' if d==None else (str(d)+str(type(d)))) }
for s,c,d in summarySel)