class puProfile:
def __init__( self, source_sample, cacheDir=os.path.join(cache_directory,"puProfiles") ):
if not os.path.isdir( cacheDir ): os.makedirs( cacheDir )
self.source_sample = source_sample
self.cacheDir = cacheDir
self.initCache( cacheDir )
self.binning = [ 100, 0, 100 ]
self.draw_string = "Pileup_nTrueInt"
def initCache(self, cacheDir):
self.cache = DirDB( os.path.join( cacheDir, 'puProfilesDirDBCache' ))
def uniqueKey( self, *arg ):
'''No dressing required'''
return arg
def cachedTemplate( self, selection, weight = '(1)', save = True, overwrite = False):
key = (selection, weight, self.source_sample.name)
if (self.cache and self.cache.contains(key)) and not overwrite:
result = self.cache.get(key)
logger.info( "Loaded MC PU profile from %s"%(self.cacheDir) )
logger.debug( "Key used: %s result: %r"%(key, result) )
elif self.cache:
logger.info( "Obtain PU profile for %s"%( key, ) )
result = self.makeTemplate( selection = selection, weight = weight)
if result:
result = self.cache.add( key, result, overwrite=save )
logger.info( "Adding PU profile to cache for %s : %r" %( key, result) )
else:
logger.warning( "Couldn't create PU profile to cache for %s : %r" %( key, result) )
else:
result = self.makeTemplate( selection = selection, weight = weight)
return result
def makeTemplate( self, selection, weight='(1)' ):
logger.info( "Make PU profile for sample %s and selection %s and weight %s", self.source_sample.name, selection, weight )
h_source = self.source_sample.get1DHistoFromDraw(self.draw_string, self.binning, selectionString = selection, weightString = weight )
logger.info( "PU histogram contains %s weighted events", h_source.Integral() )
h_source.Scale( 1./h_source.Integral() )
return h_source
max_events = 30000
# Text on the plots
tex = ROOT.TLatex()
tex.SetNDC()
tex.SetTextSize(0.04)
tex.SetTextAlign(11) # align right
# fire up the cache
cache_dir_ = os.path.join(cache_dir, 'fake_cache')
dirDB = DirDB(cache_dir_)
pu_key = (triggerSelection, leptonSelection, jetSelection, args.era,
args.small, "pu")
if dirDB.contains(pu_key) and not args.overwrite:
reweight_histo = dirDB.get(pu_key)
logger.info("Found PU reweight in cache %s", cache_dir_)
else:
logger.info("Didn't find PU reweight histo %r. Obtaining it now.", pu_key)
data_nvtx_histo = data_sample.get1DHistoFromDraw(
"PV_npvsGood", [100 / 5, 0, 100],
selectionString=data_preselectionString,
weightString="weight")
data_nvtx_histo.Scale(1. / data_nvtx_histo.Integral())
mc_histos = [
s.get1DHistoFromDraw("PV_npvsGood", [100 / 5, 0, 100],
selectionString=mc_preselectionString,
weightString="weight*reweightBTag_SF") for s in mc
]
#sample.scale /= sample.normalization
max_events = 30000
# Text on the plots
tex = ROOT.TLatex()
tex.SetNDC()
tex.SetTextSize(0.04)
tex.SetTextAlign(11) # align right
# fire up the cache
cache_dir_ = os.path.join(cache_dir, 'fake_pu_cache')
dirDB = DirDB(cache_dir_)
pu_key = ( triggerSelection, leptonSelection, jetSelection, args.era, args.small)
if dirDB.contains( pu_key ) and not args.overwrite:
reweight_histo = dirDB.get( pu_key )
logger.info( "Found PU reweight in cache %s", cache_dir_ )
else:
logger.info( "Didn't find PU reweight histo %r. Obtaining it now.", pu_key)
data_selectionString = "&&".join([getFilterCut(isData=True, year=year), triggerSelection, leptonSelection, jetSelection])
data_nvtx_histo = data_sample.get1DHistoFromDraw( "PV_npvsGood", [100, 0, 100], selectionString=data_selectionString, weightString = "weight" )
data_nvtx_histo.Scale(1./data_nvtx_histo.Integral())
mc_selectionString = "&&".join([getFilterCut(isData=False, year=year), triggerSelection, leptonSelection, jetSelection])
mc_histos = [ s.get1DHistoFromDraw( "PV_npvsGood", [100, 0, 100], selectionString=mc_selectionString, weightString = "weight*reweightBTag_SF") for s in mc]
mc_histo = mc_histos[0]
for h in mc_histos[1:]:
mc_histo.Add( h )
mc_histo.Scale(1./mc_histo.Integral())
name = "met_mc",
texX = 'p_{T}^{miss} (GeV)', texY = 'Number of Events / 20 GeV' if args.normalizeBinWidth else "Number of Events",
binning = [400/20,0,400],
stack = stack_mc,
# attribute = TreeVariable.fromString('met_pt/F'),
attribute = TreeVariable.fromString( "met_pt_%s/F" % args.variation ) if args.variation in jet_systematics else TreeVariable.fromString('met_pt/F'),
selectionString = selectionModifier(cutInterpreter.cutString(args.selection)) if selectionModifier is not None else None,
weight = mc_weight )
plots.append( met_mc )
############################################################################
# Check DB for existing plots
if args.variation is not None:
key = (args.era, mode, args.variation)
if dirDB.contains(key) and not args.overwrite:
normalisation_mc, normalisation_data, histos = dirDB.get( key )
for i_p, h_s in enumerate(histos):
plots[i_p].histos = h_s
logger.info( "Loaded normalisations and histograms for %s in mode %s from cache.", args.era, mode)
else:
logger.info( "Obtain normalisations and histograms for %s in mode %s.", args.era, mode)
normalization_selection_string = selectionModifier(cutInterpreter.cutString(args.selection))
mc_normalization_weight_string = MC_WEIGHT(variations[args.variation], returntype='string')
normalisation_mc = {s.name :s.scale*s.getYieldFromDraw(selectionString = normalization_selection_string, weightString = mc_normalization_weight_string)['val'] for s in mc}
for s in mc:
print normalisation_mc[s.name]
if args.variation == 'central':
normalisation_data = data_sample.scale*data_sample.getYieldFromDraw( selectionString = normalization_selection_string, weightString = data_weight_string)['val']