def _process(args):
default_config['jetType'] = args.jet_type
channel = args.channel
default_config['channel'] = channel
year = int(args.year)
default_config['year'] = year
if year in (2017, 2018):
args.weight_file = 'samples/xsec_2017.conf'
if year == 2018:
# FIXME: Need to update JEC when running on NanoAODv5
default_config['jec'] = True
# args.batch = True
basename = os.path.basename(args.outputdir) + '_' + args.jet_type + '_' + channel + '_' + str(year)
args.outputdir = os.path.join(os.path.dirname(args.outputdir), basename, 'data' if args.run_data else 'mc')
args.jobdir = os.path.join('jobs_%s' % basename, 'data' if args.run_data else 'mc')
if args.run_data:
args.datasets = '%s/%s_%d_DATA.yaml' % (args.sample_dir, channel, year)
args.extra_transfer = os.path.expandvars('$CMSSW_BASE/src/PhysicsTools/NanoHRTTools/data/JSON/%s' % golden_json[year])
args.json = golden_json[year]
else:
args.datasets = '%s/%s_%d_MC.yaml' % (args.sample_dir, channel, year)
if args.jet_type == 'ak15':
args.cut = cut_dict_ak15[channel]
else:
args.cut = cut_dict_ak8[channel]
args.imports = [('PhysicsTools.NanoHRTTools.producers.HeavyFlavSFTreeProducer', 'heavyFlavSFTreeFromConfig')]
if not args.run_data:
args.imports.extend([
('PhysicsTools.NanoAODTools.postprocessing.modules.common.puWeightProducer', 'puAutoWeight_2017' if year == 2017 else 'puWeight_%d' % year),
])
# data, or just nominal MC
if args.run_data or not args.run_syst:
cfg = copy.deepcopy(default_config)
if args.run_data:
cfg['data'] = True
run(args, configs={hrt_cfgname: cfg})
return
# MC for syst.
if args.run_syst and not args.run_data:
# nominal w/ PDF/Scale wegihts
logging.info('Start making nominal trees with PDF/scale weights...')
syst_name = 'LHEWeight'
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir), syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
opts.branchsel_out = 'keep_and_drop_output_LHEweights.txt'
run(opts, configs={hrt_cfgname: cfg})
# JES up/down
for variation in ['up', 'down']:
syst_name = 'jes_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jes'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir), syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
# JER up/down
for variation in ['up', 'down']:
syst_name = 'jer_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jer'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir), syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
def _process(args):
default_config['jetType'] = args.jet_type
if args.run_tagger:
default_config['run_tagger'] = True
if args.jet_type == 'ak8':
raise NotImplementedError('No training for ak8')
logging.info('Will run tagger version(s): %s' %
','.join(default_config['tagger_versions']))
if args.run_mass_regression:
default_config['run_mass_regression'] = True
if args.jet_type == 'ak8':
default_config['mass_regression_versions'] = [
'ak8V01a', 'ak8V01b', 'ak8V01c'
]
logging.info('Will run mass regression version(s): %s' %
','.join(default_config['mass_regression_versions']))
year = int(args.year)
channel = args.channel
default_config['year'] = year
default_config['channel'] = channel
if channel in ('qcd', 'photon'):
default_config['sfbdt_threshold'] = args.sfbdt
if year in (2017, 2018):
args.weight_file = 'samples/xsec_2017.conf'
basename = os.path.basename(
args.outputdir) + '_' + args.jet_type + '_' + channel + '_' + str(year)
args.outputdir = os.path.join(os.path.dirname(args.outputdir), basename,
'data' if args.run_data else 'mc')
args.jobdir = os.path.join('jobs_%s' % basename,
'data' if args.run_data else 'mc')
if args.run_data:
args.datasets = '%s/%s_%d_DATA.yaml' % (args.sample_dir, channel, year)
args.extra_transfer = os.path.expandvars(
'$CMSSW_BASE/src/PhysicsTools/NanoHRTTools/data/JSON/%s' %
golden_json[year])
args.json = golden_json[year]
else:
args.datasets = '%s/%s_%d_MC.yaml' % (args.sample_dir, channel, year)
if args.jet_type == 'ak15':
args.cut = cut_dict_ak15[channel]
else:
args.cut = cut_dict_ak8[channel]
args.imports = [
('PhysicsTools.NanoHRTTools.producers.HeavyFlavSFTreeProducer',
'heavyFlavSFTreeFromConfig')
]
if not args.run_data:
args.imports.extend([
('PhysicsTools.NanoAODTools.postprocessing.modules.common.puWeightProducer',
'puAutoWeight_2017' if year == 2017 else 'puWeight_%d' % year),
('PhysicsTools.NanoHRTTools.producers.topPtWeightProducer',
'topPtWeight')
])
# data, or just nominal MC
if args.run_data or not args.run_syst:
cfg = copy.deepcopy(default_config)
if args.run_data:
cfg['jes'] = None
cfg['jer'] = None
cfg['jmr'] = None
cfg['met_unclustered'] = None
run(args, configs={hrt_cfgname: cfg})
return
# MC for syst.
if args.run_syst and not args.run_data:
# nominal w/ PDF/Scale wegihts
logging.info('Start making nominal trees with PDF/scale weights...')
syst_name = 'LHEWeight'
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
opts.branchsel_out = 'keep_and_drop_output_LHEweights.txt'
run(opts, configs={hrt_cfgname: cfg})
# JES up/down
for variation in ['up', 'down']:
syst_name = 'jes_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jes'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
# JER up/down
for variation in ['up', 'down']:
syst_name = 'jer_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jer'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
# MET unclustEn up/down
if args.channel == 'muon':
for variation in ['up', 'down']:
syst_name = 'met_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['met_unclustered'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir),
syst_name)
run(opts, configs={hrt_cfgname: cfg})
def main():
parser = get_arg_parser()
parser.add_argument('--channel',
choices=['muon', 'photon', 'qcd'],
required=True,
help='Channel: muon, photon, qcd'
)
parser.add_argument('--run-syst',
action='store_true', default=False,
help='Run all the systematic trees. Default: %(default)s'
)
parser.add_argument('--run-data',
action='store_true', default=False,
help='Run over data. Default: %(default)s'
)
args = parser.parse_args()
channel = args.channel
default_config['channel'] = channel
if not (args.post or args.add_weight or args.merge):
tar_cmssw()
# args.batch = True
basename = os.path.basename(args.outputdir) + '_' + channel
args.outputdir = os.path.join(os.path.dirname(args.outputdir), basename, 'data' if args.run_data else 'mc')
args.jobdir = os.path.join('jobs_%s' % basename, 'data' if args.run_data else 'mc')
args.datasets = 'samples/%s.conf' % channel
args.cut = cut_dict[channel]
args.imports = [('PhysicsTools.NanoHRTTools.producers.hrtSFTreeProducer', 'hrtSFTreeFromConfig')]
if not args.run_data:
args.imports.extend([
('PhysicsTools.NanoAODTools.postprocessing.modules.common.puWeightProducer', 'puWeight'),
])
if args.channel == 'muon':
args.imports.append(('PhysicsTools.NanoHRTTools.producers.topPtWeightProducer', 'topPtWeight'))
# data, or just nominal MC
if args.run_data or not args.run_syst:
cfg = copy.deepcopy(default_config)
if args.run_data:
cfg['data'] = True
run(args, configs={hrt_cfgname: cfg})
return
# MC for syst.
if args.run_syst:
# nominal w/ PDF/Scale wegihts
logging.info('Start making nominal trees with PDF/scale weights...')
syst_name = 'LHEWeight'
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir), syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
opts.branchsel_out = 'keep_and_drop_output_LHEweights.txt'
# opts.datasets = 'samples/%s_syst.conf' % channel
# if not os.path.exists(opts.datasets):
# opts.datasets = args.datasets
run(opts, configs={hrt_cfgname: cfg})
# JES up/down
for variation in ['up', 'down']:
syst_name = 'jes_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jes'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir), syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
# JER up/down
for variation in ['up', 'down']:
syst_name = 'jer_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jer'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir), syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
# MET unclustEn up/down
for variation in ['up', 'down']:
syst_name = 'met_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['met_unclustered'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir), syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
def _process(args):
args.jet_type = 'ak8'
default_config['jetType'] = args.jet_type
default_config['run_tagger'] = True
default_config['tagger_versions'] = ['V01']
default_config['WRITE_CACHE_FILE'] = False
year = int(args.year)
option = args.option
default_config['year'] = year
default_config['option'] = option
args.weight_file = 'samples/xSections.dat'
basename = os.path.basename(
args.outputdir) + '_' + args.jet_type + '_option' + option + '_' + str(
year)
args.outputdir = os.path.join(os.path.dirname(args.outputdir), basename,
'data' if args.run_data else 'mc')
args.jobdir = os.path.join('jobs_%s' % basename,
'data' if args.run_data else 'mc')
if args.run_signal:
args.outputdir = args.outputdir.replace('mc', 'signal')
args.jobdir = os.path.join('jobs_%s' % basename, 'signal')
sample_str = "hhbbWW"
if args.run_data:
args.datasets = '%s/%s_%d_DATA.yaml' % (args.sample_dir, sample_str,
year)
args.extra_transfer = os.path.expandvars(
'$CMSSW_BASE/src/PhysicsTools/NanoNN/data/JSON/%s' %
golden_json[year])
args.json = golden_json[year]
elif args.run_signal:
args.datasets = '%s/%s_%d_signalMC.yaml' % (args.sample_dir,
sample_str, year)
else:
args.datasets = '%s/%s_%d_MC.yaml' % (args.sample_dir, sample_str,
year)
if samples:
args.select = ','.join(samples[year])
if args.run_signal:
args.imports = [(
'PhysicsTools.NanoAODTools.postprocessing.modules.common.countHistogramsModule',
'countHistogramsProducer')]
args.imports.extend([('PhysicsTools.NanoNN.producers.hhbbWWProducer',
'hhbbWWProducerFromConfig')])
else:
args.imports = [('PhysicsTools.NanoNN.producers.hhbbWWProducer',
'hhbbWWProducerFromConfig')]
args.cut = cut_dict_ak8[str(option)]
if not args.run_data:
args.imports.extend([(
'PhysicsTools.NanoAODTools.postprocessing.modules.common.puWeightProducer',
'puAutoWeight_2017' if year == 2017 else 'puWeight_%d' % year)])
# select branches
args.branchsel_in = None
args.branchsel_out = os.path.expandvars(
'$CMSSW_BASE/src/PhysicsTools/NanoAODTools/scripts/branch_hh4b_output.txt'
)
# data, or just nominal MC
cfg = copy.deepcopy(default_config)
if args.run_data:
cfg['jes'] = None
cfg['jer'] = None
cfg['met_unclustered'] = None
print('run ', args, nn_cfgname)
run(args, configs={nn_cfgname: cfg})
return
def _process(args):
channel = args.channel
default_config['channel'] = channel
year = args.year
default_config['year'] = year
if year in (2017, 2018):
args.weight_file = 'samples/xsec_2017.conf'
if year == 2018:
# FIXME: Need to update JEC when running on NanoAODv5
default_config['jec'] = True
year_dep_cuts = {
'DeepCSV_WP_M': {
2016: 0.6321,
2017: 0.4941,
2018: 0.4184
}[year]
}
# args.batch = True
basename = os.path.basename(
args.outputdir) + '_' + channel + '_' + str(year)
args.outputdir = os.path.join(os.path.dirname(args.outputdir), basename,
'data' if args.run_data else 'mc')
args.jobdir = os.path.join('jobs_%s' % basename,
'data' if args.run_data else 'mc')
if args.run_data:
args.datasets = '%s/%s_%d_DATA.yaml' % (args.sample_dir, channel, year)
args.extra_transfer = os.path.expandvars(
'$CMSSW_BASE/src/PhysicsTools/NanoHRTTools/data/JSON/%s' %
golden_json[year])
args.json = golden_json[year]
else:
args.datasets = '%s/%s_%d_MC.yaml' % (args.sample_dir, channel, year)
args.cut = cut_dict[channel].format(**year_dep_cuts)
args.imports = [('PhysicsTools.NanoHRTTools.producers.hrtSFTreeProducer',
'hrtSFTreeFromConfig')]
if not args.run_data:
args.imports.extend([
('PhysicsTools.NanoAODTools.postprocessing.modules.common.puWeightProducer',
'puAutoWeight_2017' if year == 2017 else 'puWeight_%d' % year),
])
if args.channel == 'muon':
args.imports.append(
('PhysicsTools.NanoHRTTools.producers.topPtWeightProducer',
'topPtWeight')) #FIXME: year dependence?
# data, or just nominal MC
if args.run_data or not args.run_syst:
cfg = copy.deepcopy(default_config)
if args.run_data:
cfg['data'] = True
run(args, configs={hrt_cfgname: cfg})
return
# MC for syst.
if args.run_syst:
# nominal w/ PDF/Scale wegihts
logging.info('Start making nominal trees with PDF/scale weights...')
syst_name = 'LHEWeight'
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
opts.branchsel_out = 'keep_and_drop_output_LHEweights.txt'
run(opts, configs={hrt_cfgname: cfg})
# JES up/down
for variation in ['up', 'down']:
syst_name = 'jes_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jes'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
# JER up/down
for variation in ['up', 'down']:
syst_name = 'jer_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jer'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
run(opts, configs={hrt_cfgname: cfg})
# MET unclustEn up/down
if args.channel == 'muon':
for variation in ['up', 'down']:
syst_name = 'met_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['met_unclustered'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir),
syst_name)
run(opts, configs={hrt_cfgname: cfg})
def _process(args):
args.jet_type = 'ak8'
default_config['jetType'] = args.jet_type
if args.run_mass_regression:
default_config['run_mass_regression'] = True
if args.jet_type == 'ak8':
default_config['mass_regression_versions'] = [
'ak8V01a', 'ak8V01b', 'ak8V01c'
]
logging.info('Will run mass regression version(s): %s' %
','.join(default_config['mass_regression_versions']))
year = int(args.year)
option = args.option
default_config['year'] = year
default_config['option'] = option
args.weight_file = 'samples/xSections.dat'
basename = os.path.basename(
args.outputdir) + '_' + args.jet_type + '_option' + option + '_' + str(
year)
args.outputdir = os.path.join(os.path.dirname(args.outputdir), basename,
'data' if args.run_data else 'mc')
args.jobdir = os.path.join('jobs_%s' % basename,
'data' if args.run_data else 'mc')
if args.run_signal:
args.outputdir = args.outputdir.replace('mc', 'signal')
args.jobdir = os.path.join('jobs_%s' % basename, 'signal')
sample_str = "hh4b"
if option == "10": sample_str = "tt"
if args.run_data:
args.datasets = '%s/%s_%d_DATA.yaml' % (args.sample_dir, sample_str,
year)
args.extra_transfer = os.path.expandvars(
'$CMSSW_BASE/src/PhysicsTools/NanoNN/data/JSON/%s' %
golden_json[year])
args.json = golden_json[year]
elif args.run_signal:
args.datasets = '%s/%s_%d_signalMC.yaml' % (args.sample_dir,
sample_str, year)
else:
args.datasets = '%s/%s_%d_MC.yaml' % (args.sample_dir, sample_str,
year)
if samples:
args.select = ','.join(samples[year])
if args.run_signal:
args.imports = [(
'PhysicsTools.NanoAODTools.postprocessing.modules.common.countHistogramsModule',
'countHistogramsProducer')]
args.imports.extend([('PhysicsTools.NanoNN.producers.hh4bProducer',
'hh4bProducerFromConfig')])
else:
args.imports = [('PhysicsTools.NanoNN.producers.hh4bProducer',
'hh4bProducerFromConfig')]
args.cut = cut_dict_ak8[str(option)]
if not args.run_data:
args.imports.extend([(
'PhysicsTools.NanoAODTools.postprocessing.modules.common.puWeightProducer',
'puAutoWeight_2017' if year == 2017 else 'puWeight_%d' % year)])
# select branches
args.branchsel_in = None
args.branchsel_out = os.path.expandvars(
'$CMSSW_BASE/src/PhysicsTools/NanoAODTools/scripts/branch_hh4b_output.txt'
)
# data, or just nominal MC
if args.run_data or not args.run_syst:
cfg = copy.deepcopy(default_config)
# set all JME to true
cfg['allJME'] = True
if args.run_data:
cfg['allJME'] = False
cfg['jes'] = None
cfg['jer'] = None
cfg['met_unclustered'] = None
print('run ', args, nn_cfgname)
run(args, configs={nn_cfgname: cfg})
return
# MC for syst
if args.run_syst and not args.run_data:
# nominal w/ PDF/Scale weights
'''
logging.info('Start making nominal trees with PDF/scale weights...')
syst_name = 'LHEWeight'
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir), syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
opts.branchsel_out = os.path.expandvars('$CMSSW_BASE/src/PhysicsTools/NanoAODTools/scripts/branch_hh4b_output_LHEweights.txt'
run(opts, configs={nn_cfgname: cfg})
'''
# JES up/down
for variation in ['up', 'down']:
syst_name = 'jes_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jes'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
if args.run_signal:
print('run signal')
opts.outputdir = opts.outputdir + '_signal'
opts.jobdir = opts.jobdir + '_signal'
run(opts, configs={nn_cfgname: cfg})
# JER up/down
for variation in ['up', 'down']:
syst_name = 'jer_%s' % variation
logging.info('Start making %s trees...' % syst_name)
opts = copy.deepcopy(args)
cfg = copy.deepcopy(default_config)
cfg['jer'] = variation
opts.outputdir = os.path.join(os.path.dirname(opts.outputdir),
syst_name)
opts.jobdir = os.path.join(os.path.dirname(opts.jobdir), syst_name)
if args.run_signal:
print('run signal')
opts.outputdir = opts.outputdir + '_signal'
opts.jobdir = opts.jobdir + '_signal'
run(opts, configs={nn_cfgname: cfg})
# MET unclustered up/down
'''
