def get_old_kfac(tag):
if tag.startswith('w'):
f = uproot.open(bucoffea_path('data/sf/theory/merged_kfactors_wjets.root'))
elif tag.startswith('dy'):
f = uproot.open(bucoffea_path('data/sf/theory/merged_kfactors_wjets.root'))
elif tag.startswith('gjets'):
f = uproot.open(bucoffea_path('data/sf/theory/merged_kfactors_gjets.root'))
return f['kfactor_monojet_qcd']
def evaluator_from_config(cfg):
"""Initiates the SF evaluator and populates it with the right values
:param cfg: Configuration
:type cfg: DynaConf object
:return: Ready-to-use SF evaluator
:rtype: coffea.lookup_tools.evaluator
"""
ext = extractor()
for sfname, definition in cfg.SF.items():
if not 'file' in definition:
continue
fpath = bucoffea_path(definition['file'])
if fpath.endswith(".root"):
ext.add_weight_sets(
[f"{sfname} {definition['histogram']} {fpath}"])
ext.add_weight_sets(
[f"{sfname}_error {definition['histogram']}_error {fpath}"])
else:
continue
ext.finalize()
evaluator = ext.make_evaluator()
return evaluator
def load_xs():
"""Function to read per-sample cross sections from file.
:return: Mapping dataset -> cross-section
:rtype: dict
"""
xsfile = bucoffea_path('data/datasets/xs/xs.yml')
with open(xsfile, 'r') as f:
xs_yml = yaml.load(f, Loader=yaml.FullLoader)
# See the documentation in data/datasets/xs/README.md
# for more information on how the XS is chosen.
xs = {}
loading_priority = ['nnnlo', 'nnlo', 'nlo', 'lo', 'gen']
for dataset, xs_dict in xs_yml.items():
if 'use' in xs_dict:
key_to_use = xs_dict['use']
else:
for key in loading_priority:
if key in xs_dict:
key_to_use = key
break
xs[dataset] = xs_dict[key_to_use]
# Data sets that only exist as extensions
# cause problems later on, so we duplicate the XS
# for the base process.
tmp = {}
for k in xs.keys():
base = re.sub('_ext(\d+)', '', k)
if base not in xs.keys():
tmp[base] = xs[k]
xs.update(tmp)
return xs
def btag_weights(bjets, cfg):
# Evaluate weight variations
weight_variations = {}
# Only calculate for DeepCSV
if cfg.BTAG.ALGO != "deepcsv":
weight_variations["central"] = bjets.pt.ones_like()
weight_variations["up"] = bjets.pt.ones_like()
weight_variations["down"] = bjets.pt.ones_like()
return weight_variations
# Heavy lifting done by coffea implementation
bsf = BTagScaleFactor(
filename=bucoffea_path(cfg.SF.DEEPCSV.FILE),
workingpoint=cfg.BTAG.WP.upper(),
methods='comb,comb,incl' # Comb for b and c flavors, incl for light
)
for variation in ["central","up","down"]:
weights = bsf.eval(
systematic=variation,
flavor=bjets.hadflav,
abseta=bjets.abseta,
pt=bjets.pt)
# Cap the weights just in case
weights[np.abs(weights)>5] = 1
weight_variations[variation] = weights
return weight_variations
def pack_repo(path_to_gridpack):
'''Creates a gridpack containing the bucoffea repo'''
if os.path.exists(path_to_gridpack):
raise RuntimeError(f"Gridpack file already exists. Will not overwrite {path_to_gridpack}.")
tar = tarfile.open(path_to_gridpack,'w')
files = get_repo_files()
for f in files:
tar.add(
name=f,
arcname=f.replace(os.path.abspath(bucoffea_path('..')),'bucoffea'),
exclude=lambda x: ('tgz' in x or 'submission' in x)
)
tar.close()
return
def get_repo_files():
'''Returns a list of tracked files in the bucoffea repo'''
import git
repo = git.Repo(bucoffea_path('..'))
to_iterate = [repo.tree()]
to_add = []
while len(to_iterate):
for item in to_iterate.pop():
if item.type == 'tree':
to_iterate.append(item)
elif item.type == 'blob':
to_add.append(item.abspath)
return to_add
def main():
create_tdr_style()
r.gROOT.SetBatch(r.kTRUE)
fname=bucoffea_path('data/sf/pileup/pileup.root')
f = r.TFile(fname)
for year in [2017, 2018]:
c,t1,t2=setup_canvas(want_ratio=True)
hdata = f.Get(f'data{year}_nominal')
hmc = f.Get(f'mc{year}')
hmc.SetLineWidth(4)
hmc.SetFillStyle(4001)
hmc.SetFillColorAlpha(r.kAzure,0.25)
hdata.SetMarkerStyle(20)
hdata.SetMarkerSize(0.7)
ratio = hdata.Clone("ratio")
ratio.Divide(hmc)
s = r.THStack()
s.Add(hmc)
s.Add(hdata,'PE')
t1.cd()
s.Draw('nostack')
s.SetMaximum(.05)
s.GetXaxis().SetRangeUser(0,80)
leg = r.TLegend(0.6,0.7,0.9,0.9)
leg.SetHeader(f'{year}')
leg.AddEntry(hmc,f"MC, #mu = {hmc.GetMean():.1f}")
leg.AddEntry(hdata,f"Data #mu = {hdata.GetMean():.1f}")
leg.Draw()
t2.cd()
ratio.Draw('PE')
apply_style_to_axis(s,is_ratio=False)
apply_style_to_axis(ratio,is_ratio=True,ymin=0.,ymax=2.)
ratio.SetMarkerStyle(20)
s.GetXaxis().SetRangeUser(0,80)
ratio.GetXaxis().SetRangeUser(0,80)
ratio.GetXaxis().SetTitle("True number of PU events")
c.SaveAs(f"pu_weights_{year}.pdf")
def monojet_evaluator(cfg):
"""Initiates the SF evaluator and populates it with the right values
:param cfg: Configuration
:type cfg: DynaConf object
:return: Ready-to-use SF evaluator
:rtype: coffea.lookup_tools.evaluator
"""
ext = extractor()
for sfname, definition in cfg.SF.items():
fpath = bucoffea_path(definition['file'])
ext.add_weight_sets([f"{sfname} {definition['histogram']} {fpath}"])
ext.finalize()
evaluator = ext.make_evaluator()
return evaluator
def get_old_kfac(tag):
'''Given the dataset tag, get the nominal 2D VBF k-factors.'''
f = uproot.open(bucoffea_path('data/sf/theory/2017_gen_v_pt_qcd_sf.root'))
return f[f'2d_{tag}_vbf'].values
