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 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 main() :
parser = optparse.OptionParser(usage=usage)
parser.add_option('-b', '--batch', action='store_true', default=False, help='submit to batch (used in fill mode)')
parser.add_option('-f', '--input-fake', help='location hft trees for fake')
parser.add_option('-g', '--input-gen', help='location hft trees for everything else')
parser.add_option('-i', '--input-dir')
parser.add_option('-o', '--output-dir')
parser.add_option('-s', '--syst', help="variations to process (default all). Give a comma-sep list or say 'weight', 'object', or 'fake'")
parser.add_option('-e', '--exclude', help="skip some systematics, example 'EL_FR_.*'")
parser.add_option('-v', '--verbose', action='store_true', default=False)
parser.add_option('-l', '--list-systematics', action='store_true', default=False, help='list what is already in output_dir')
parser.add_option('-L', '--list-all-systematics', action='store_true', default=False, help='list all possible systematics')
(opts, args) = parser.parse_args()
if opts.list_all_systematics :
print "All systematics:\n\t%s"%'\n\t'.join(systUtils.getAllVariations())
return
if opts.list_systematics :
print listExistingSyst(opts.input_dir)
return
inGenSpecified, inDirSpecified = opts.input_gen!=None, opts.input_dir!=None
eitherMode = inGenSpecified != inDirSpecified
if not eitherMode : parser.error("Run either in 'fill' or 'plot' mode")
mode = 'fill' if inGenSpecified else 'plot' if inDirSpecified else None
requiredOptions = (['input_fake', 'input_gen', 'output_dir'] if mode=='fill' else ['input_dir', 'output_dir'])
allOptions = [x.dest for x in parser._get_all_options()[1:]]
def optIsNotSpecified(o) : return not hasattr(opts, o) or getattr(opts,o) is None
if any(optIsNotSpecified(o) for o in requiredOptions) :
parser.error('Missing required option\n'+'\n'.join(["%s : %s"%(o, getattr(opts, o)) for o in requiredOptions]))
if opts.verbose : print '\nUsing the following options:\n'+'\n'.join("%s : %s"%(o, str(getattr(opts, o))) for o in allOptions)
if mode=='fill' : runFill(opts)
elif mode=='plot' : runPlot(opts)
def variationsSummary(self) :
summaries = {} # one summary for each selection
for selection, sysCounts in self.varCounts.iteritems() :
nominalCount = sysCounts['NOM']
summaries[selection] = [(sys, sysCount, (100.0*(sysCount-nominalCount)/nominalCount) if nominalCount else None)
for sys, sysCount in sortedAs(sysCounts, systUtils.getAllVariations())]
return summaries
def main() :
parser = optparse.OptionParser(usage=usage)
parser.add_option('-g', '--group', help='group to be processed (used only in fill mode)')
parser.add_option('--exclude-group', help='exclude group from processing (used only in fill mode)')
parser.add_option('-f', '--input-fake', help='location of fake trees')
parser.add_option('-O', '--input-other', help='location other trees')
parser.add_option('-i', '--input-dir')
parser.add_option('-o', '--output-dir')
parser.add_option('--samples-dir', default='samples/',
help='directory with the list of samples; default ./samples/')
parser.add_option('-s', '--syst', help="variations to process (default all)."
" Give a comma-sep list or say 'weight', 'object', or 'fake'")
parser.add_option('--log-dir', help='directory where the batch logs will be (default log/...)')
parser.add_option('-e', '--exclude', help="skip some systematics, example 'EL_FR_.*'")
parser.add_option('-q', '--queue', default='atlas_all', help="batch queue, default atlas_all")
parser.add_option('-T', '--tight-def', help='on-the-fly tight def, one of defs in fakeUtils.py: fakeu.lepIsTight_std, etc.')
parser.add_option('--regions', default=None, help='comma-separated list of regions to consider')
parser.add_option('--include-regions', default='.*', help='regexp to filter regions')
parser.add_option('--exclude-regions', default=None, help='regext to exclude regions')
# reminder: submit_batch_fill_job_per_group expects argument-less opt to default to False
parser.add_option('--debug', action='store_true')
parser.add_option('--verbose', action='store_true')
parser.add_option('--unblind', action='store_true')
parser.add_option('-b', '--batch', action='store_true', help='submit to batch (used in fill mode)')
parser.add_option('-l', '--list-systematics', action='store_true', help='list what is already in output_dir')
parser.add_option('-L', '--list-all-systematics', action='store_true', help='list all possible systematics')
parser.add_option('--list-all-regions', action='store_true', help='list all possible regions')
parser.add_option('--require-tight-tight', action='store_true', help='fill histos only when both leps are tight')
parser.add_option('--quick-test', action='store_true', help='run a quick test and fill only 1% of the events')
parser.add_option('--disable-cache', action='store_true', help='disable the entry cache')
parser.add_option('--format-aux', action='store_true', help='format plots for paper aux material')
(opts, args) = parser.parse_args()
if opts.list_all_systematics :
print "All systematics:\n\t%s"%'\n\t'.join(systUtils.getAllVariations())
return
if opts.list_systematics :
print listExistingSyst(opts.input_dir)
return
if opts.list_all_regions:
print "All regions:\n\t%s"%'\n\t'.join(sorted(selection_formulas().keys()))
return
inOtherSpecified, inDirSpecified = opts.input_other!=None, opts.input_dir!=None
eitherMode = inOtherSpecified != inDirSpecified
if not eitherMode : parser.error("Run either in 'fill' or 'plot' mode")
mode = 'fill' if inOtherSpecified else 'plot' if inDirSpecified else None
if opts.quick_test : opts.disable_cache = True # don't write bogus entrylists
requiredOptions = (['input_fake', 'input_other', 'output_dir'] if mode=='fill'
else ['input_dir', 'output_dir'])
def optIsNotSpecified(o) : return not hasattr(opts, o) or getattr(opts,o) is None
if any(optIsNotSpecified(o) for o in requiredOptions):
parser.error('Missing required option\n'
+'\n'.join(["%s : %s"%(o, getattr(opts, o)) for o in requiredOptions]))
if opts.verbose : utils.print_running_conditions(parser, opts)
if mode=='fill' : runFill(opts)
elif mode=='plot' : runPlot(opts)
def main():
parser = optparse.OptionParser(usage=usage)
parser.add_option('-g',
'--group',
help='group to be processed (used only in fill mode)')
parser.add_option(
'--exclude-group',
help='exclude group from processing (used only in fill mode)')
parser.add_option('-f', '--input-fake', help='location of fake trees')
parser.add_option('-O', '--input-other', help='location other trees')
parser.add_option('-i', '--input-dir')
parser.add_option('-o', '--output-dir')
parser.add_option(
'--samples-dir',
default='samples/',
help='directory with the list of samples; default ./samples/')
parser.add_option(
'-s',
'--syst',
help="variations to process (default all)."
" Give a comma-sep list or say 'weight', 'object', or 'fake'")
parser.add_option(
'--log-dir',
help='directory where the batch logs will be (default log/...)')
parser.add_option('-e',
'--exclude',
help="skip some systematics, example 'EL_FR_.*'")
parser.add_option('-q',
'--queue',
default='atlas_all',
help="batch queue, default atlas_all")
parser.add_option(
'-T',
'--tight-def',
help=
'on-the-fly tight def, one of defs in fakeUtils.py: fakeu.lepIsTight_std, etc.'
)
parser.add_option('--regions',
default=None,
help='comma-separated list of regions to consider')
parser.add_option('--include-regions',
default='.*',
help='regexp to filter regions')
parser.add_option('--exclude-regions',
default=None,
help='regext to exclude regions')
# reminder: submit_batch_fill_job_per_group expects argument-less opt to default to False
parser.add_option('--debug', action='store_true')
parser.add_option('--verbose', action='store_true')
parser.add_option('--unblind', action='store_true')
parser.add_option('-b',
'--batch',
action='store_true',
help='submit to batch (used in fill mode)')
parser.add_option('-l',
'--list-systematics',
action='store_true',
help='list what is already in output_dir')
parser.add_option('-L',
'--list-all-systematics',
action='store_true',
help='list all possible systematics')
parser.add_option('--list-all-regions',
action='store_true',
help='list all possible regions')
parser.add_option('--require-tight-tight',
action='store_true',
help='fill histos only when both leps are tight')
parser.add_option('--quick-test',
action='store_true',
help='run a quick test and fill only 1% of the events')
parser.add_option('--disable-cache',
action='store_true',
help='disable the entry cache')
parser.add_option('--format-aux',
action='store_true',
help='format plots for paper aux material')
(opts, args) = parser.parse_args()
if opts.list_all_systematics:
print "All systematics:\n\t%s" % '\n\t'.join(
systUtils.getAllVariations())
return
if opts.list_systematics:
print listExistingSyst(opts.input_dir)
return
if opts.list_all_regions:
print "All regions:\n\t%s" % '\n\t'.join(
sorted(selection_formulas().keys()))
return
inOtherSpecified, inDirSpecified = opts.input_other != None, opts.input_dir != None
eitherMode = inOtherSpecified != inDirSpecified
if not eitherMode: parser.error("Run either in 'fill' or 'plot' mode")
mode = 'fill' if inOtherSpecified else 'plot' if inDirSpecified else None
if opts.quick_test:
opts.disable_cache = True # don't write bogus entrylists
requiredOptions = (['input_fake', 'input_other', 'output_dir']
if mode == 'fill' else ['input_dir', 'output_dir'])
def optIsNotSpecified(o):
return not hasattr(opts, o) or getattr(opts, o) is None
if any(optIsNotSpecified(o) for o in requiredOptions):
parser.error('Missing required option\n' + '\n'.join(
["%s : %s" % (o, getattr(opts, o)) for o in requiredOptions]))
if opts.verbose: utils.print_running_conditions(parser, opts)
if mode == 'fill': runFill(opts)
elif mode == 'plot': runPlot(opts)
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)