def test_vectors( self ):
h_signal = adjust_overflow_to_limit( self.h_signal, x_min, x_max )
h_signal.Scale(1/h_signal.Integral())
h_bkg1 = adjust_overflow_to_limit( self.h_bkg1, x_min, x_max )
h_bkg1.Scale(1/h_bkg1.Integral())
signal = list( h_signal.y() )
bkg1 = list( h_bkg1.y() )
v_from_fit_data = self.fit_data_1.vectors
v_from_single_collection = self.single_collection.vectors()
# v_from_collection = self.collection_1.vectors( 'signal region' )
# v_from_collection_1 = self.collection_1.vectors()['signal region']
self.assertEqual(signal, v_from_fit_data['signal'])
self.assertEqual(bkg1, v_from_fit_data['bkg1'])
self.assertEqual(signal, v_from_single_collection['signal'])
self.assertEqual(bkg1, v_from_single_collection['bkg1'])
def test_adjust_overflow_to_limit( self ):
x_min = 40
x_max = 80
adjusted = adjust_overflow_to_limit(self.h1, x_min, x_max)
# number of events should be unchanged
# the adjusted histogram should have no overflow for this example
self.assertEqual(self.h1.integral( overflow = True ), adjusted.Integral())
# first bin (x_min) should contain all events
# with x <= x_min
x_min_bin = self.h1.FindBin(x_min)
x_max_bin = self.h1.FindBin(x_max)
self.assertEqual(self.h1.integral(0, x_min_bin),
adjusted.GetBinContent(x_min_bin))
# last bin (x_max) should contain all events
# with x >= x_max
self.assertEqual( self.h1.integral( x_max_bin, self.h1.nbins() + 1 ),
adjusted.GetBinContent( x_max_bin ) )
def test_adjust_overflow_to_limit_simple( self ):
x_min = 0
x_max = 95
adjusted = adjust_overflow_to_limit(self.simple, x_min, x_max)
# for entry_1, entry_2 in zip(hist_to_value_error_tuplelist(self.simple), hist_to_value_error_tuplelist(adjusted)):
# print entry_1, entry_2
# print self.simple.integral( overflow = True ), adjusted.integral()
# print self.simple.GetBinContent(1), self.simple.GetBinContent(self.simple.nbins())
# number of events should be unchanged
# the adjusted histogram should have no overflow for this example
self.assertEqual( self.simple.integral( overflow = True ), adjusted.integral() )
# first bin (x_min) should contain all events
# with x <= x_min
x_min_bin = self.simple.FindBin(x_min)
x_max_bin = self.simple.FindBin(x_max)
self.assertEqual(self.simple.integral(0, x_min_bin),
adjusted.GetBinContent(x_min_bin))
# last bin (x_max) should contain all events
# with x >= x_max
self.assertEqual( self.simple.integral( x_max_bin, self.simple.nbins() + 1),
adjusted.GetBinContent( x_max_bin ) )
def test_normalisation( self ):
normalisation = {name:adjust_overflow_to_limit(histogram, x_min, x_max).Integral() for name, histogram in self.histograms_1.iteritems()}
normalisation_from_fit_data = self.fit_data_1.normalisation
normalisation_from_single_collection = self.single_collection.mc_normalisation()
normalisation_from_collection = self.collection_1.mc_normalisation( 'signal region' )
normalisation_from_collection_1 = self.collection_1.mc_normalisation()['signal region']
for sample in normalisation.keys():
self.assertEqual( normalisation[sample], normalisation_from_fit_data[sample] )
self.assertEqual( normalisation[sample], normalisation_from_single_collection[sample] )
self.assertEqual( normalisation[sample], normalisation_from_collection[sample] )
self.assertEqual( normalisation[sample], normalisation_from_collection_1[sample] )
# data normalisation
normalisation = self.h_data.integral( overflow = True )
normalisation_from_fit_data = self.fit_data_1.n_data()
normalisation_from_single_collection = self.single_collection.n_data()
normalisation_from_collection = self.collection_1.n_data( 'signal region' )
normalisation_from_collection_1 = self.collection_1.n_data()['signal region']
self.assertEqual( normalisation, normalisation_from_fit_data )
self.assertEqual( normalisation, normalisation_from_single_collection )
self.assertEqual( normalisation, normalisation_from_collection )
self.assertEqual( normalisation, normalisation_from_collection_1 )
self.assertAlmostEqual(normalisation, self.collection_1.max_n_data(), delta = 1 )
def get_histograms( channel, input_files, variable, met_type, variable_bin,
b_tag_bin, rebin = 1, fit_variable = 'absolute_eta',
scale_factors = None ):
global b_tag_bin_VJets, fit_variables
global electron_control_region, muon_control_region
boundaries = measurement_config.fit_boundaries[fit_variable]
histograms = {}
if not variable in measurement_config.histogram_path_templates.keys():
print 'Fatal Error: unknown variable ', variable
sys.exit()
fit_variable_name = ''
fit_variable_name_data = ''
fit_variable_template = measurement_config.histogram_path_templates[variable]
ht_fill_list, other_fill_list = None, None
if fit_variable == 'absolute_eta':
ht_fill_list = ( analysis_type[channel], variable_bin, channel + '_' + fit_variable )
other_fill_list = ( analysis_type[channel], met_type, variable_bin, channel + '_' + fit_variable )
else:
ht_fill_list = ( analysis_type[channel], variable_bin, fit_variable )
other_fill_list = ( analysis_type[channel], met_type, variable_bin, fit_variable )
if variable == 'HT':
fit_variable_name = fit_variable_template % ht_fill_list
fit_variable_name_data = fit_variable_name
else:
fit_variable_name = fit_variable_template % other_fill_list
if 'JetRes' in met_type:
fit_variable_name_data = fit_variable_name.replace( 'JetResDown', '' )
fit_variable_name_data = fit_variable_name_data.replace( 'JetResUp', '' )
if 'patPFMet' in met_type:
fit_variable_name = fit_variable_name.replace( 'patPFMet', 'PFMET' )
else:
fit_variable_name_data = fit_variable_name
for sample, file_name in input_files.iteritems():
if not file_name:
continue
h_fit_variable = None
if sample == 'data':
h_fit_variable = get_histogram( file_name, fit_variable_name_data, b_tag_bin )
elif sample == 'V+Jets':
# extracting the inclusive V+Jets template across all bins from its specific b-tag bin (>=0 by default) and scaling it to analysis b-tag bin
for var_bin in variable_bins_ROOT[variable]:
temp_variable_name = fit_variable_name.replace(variable_bin, var_bin)
if h_fit_variable == None:
h_fit_variable = get_histogram( file_name, temp_variable_name, b_tag_bin )
else:
h_fit_variable += get_histogram( file_name, temp_variable_name, b_tag_bin )
h_fit_variable_for_scaling = get_histogram(file_name, fit_variable_name, b_tag_bin)
scale = h_fit_variable_for_scaling.integral (overflow = True ) / h_fit_variable.integral( overflow = True )
h_fit_variable.Scale(scale)
else:
h_fit_variable = get_histogram( file_name, fit_variable_name, b_tag_bin )
h_fit_variable.Rebin( rebin )
h_fit_variable = adjust_overflow_to_limit( h_fit_variable, boundaries[0], boundaries[1] )
histograms[sample] = h_fit_variable
h_qcd = get_qcd_histograms( input_files, variable, variable_bin,
channel, fit_variable_name, rebin )
if h_qcd.Integral() < 0.1:
h_qcd.Scale( 0.1/h_qcd.Integral() )
pass
histograms['QCD'] = adjust_overflow_to_limit( h_qcd,
boundaries[0], boundaries[1] )
# normalise histograms
if not measurement_config.luminosity_scale == 1.0:
for sample, histogram in histograms.iteritems():
if sample == 'data':
continue
histogram.Scale( measurement_config.luminosity_scale )
# apply normalisation scale factors for rate-changing systematics
if scale_factors:
for source, factor in scale_factors.iteritems():
if 'luminosity' in source:
for sample, histogram in histograms.iteritems():
if sample == 'data':
continue
histogram.Scale( factor )
for sample, histogram in histograms.iteritems():
if sample in source:
histogram.Scale( factor )
return histograms
def get_histograms( channel, input_files, variable, met_type, variable_bin,
b_tag_bin, rebin = 1, fit_variable = 'absolute_eta',
scale_factors = None ):
global b_tag_bin_VJets, fit_variables
global electron_control_region, muon_control_region
boundaries = measurement_config.fit_boundaries[fit_variable]
histograms = {}
if not variable in measurement_config.histogram_path_templates.keys():
print 'Fatal Error: unknown variable ', variable
sys.exit()
fit_variable_name = ''
fit_variable_name_data = ''
fit_variable_template = measurement_config.histogram_path_templates[variable]
ht_fill_list, other_fill_list = None, None
if fit_variable == 'absolute_eta':
ht_fill_list = ( analysis_type[channel], variable_bin, channel + '_' + fit_variable )
other_fill_list = ( analysis_type[channel], met_type, variable_bin, channel + '_' + fit_variable )
else:
ht_fill_list = ( analysis_type[channel], variable_bin, fit_variable )
other_fill_list = ( analysis_type[channel], met_type, variable_bin, fit_variable )
if variable == 'HT':
fit_variable_name = fit_variable_template % ht_fill_list
fit_variable_name_data = fit_variable_name
else:
fit_variable_name = fit_variable_template % other_fill_list
if 'JetRes' in met_type:
fit_variable_name_data = fit_variable_name.replace( 'JetResDown', '' )
fit_variable_name_data = fit_variable_name_data.replace( 'JetResUp', '' )
if 'patPFMet' in met_type:
fit_variable_name = fit_variable_name.replace( 'patPFMet', 'PFMET' )
else:
fit_variable_name_data = fit_variable_name
for sample, file_name in input_files.iteritems():
if not file_name:
continue
h_fit_variable = None
if sample == 'data':
h_fit_variable = get_histogram( file_name, fit_variable_name_data, b_tag_bin )
elif sample == 'V+Jets':
# extracting the V+Jets template from its specific b-tag bin (>=0 by default) and scaling it to analysis b-tag bin
h_fit_variable = get_histogram( file_name, fit_variable_name, b_tag_bin )
if not '_bl' in fit_variable:
# this procedure is not valid for fit variables requiring at least one b-tag
h_fit_variable_VJets_specific_b_tag_bin = get_histogram( file_name, fit_variable_name, b_tag_bin_VJets )
try:
scale = h_fit_variable.integral( overflow = True ) / h_fit_variable_VJets_specific_b_tag_bin.integral( overflow = True )
h_fit_variable_VJets_specific_b_tag_bin.Scale( scale )
h_fit_variable = h_fit_variable_VJets_specific_b_tag_bin
except:
print 'WARNING: V+Jets template from ' + str( file_name ) + ', histogram ' + fit_variable_name + ' in ' + b_tag_bin_VJets + \
' b-tag bin is empty. Using central bin (' + b_tag_bin + '), integral = ' + str( h_fit_variable.Integral() )
else:
h_fit_variable = get_histogram( file_name, fit_variable_name, b_tag_bin )
h_fit_variable.Rebin( rebin )
# this is not working. M3 has fewer data events than absolute_eta
h_fit_variable = adjust_overflow_to_limit( h_fit_variable, boundaries[0], boundaries[1] )
histograms[sample] = h_fit_variable
h_qcd = get_qcd_histograms( input_files, variable, variable_bin,
channel, fit_variable_name, rebin )
if h_qcd.Integral() < 0.1:
h_qcd.Scale( 0.1/h_qcd.Integral() )
pass
histograms['QCD'] = adjust_overflow_to_limit( h_qcd,
boundaries[0], boundaries[1] )
# normalise histograms
if not measurement_config.luminosity_scale == 1.0:
for sample, histogram in histograms.iteritems():
if sample == 'data':
continue
histogram.Scale( measurement_config.luminosity_scale )
# apply normalisation scale factors for rate-changing systematics
if scale_factors:
for source, factor in scale_factors.iteritems():
if 'luminosity' in source:
for sample, histogram in histograms.iteritems():
if sample == 'data':
continue
histogram.Scale( factor )
for sample, histogram in histograms.iteritems():
if sample in source:
histogram.Scale( factor )
return histograms
