def run_dijet(edges, args):
with open(args.denominator, 'r') as denom_file:
denom = get_denom_dict(denom_file)
with open(args.cross_sections, 'r') as xsec_file:
xsecs = CrossSections(xsec_file, denom)
parts = {}
hist = 0
for ds in args.datasets:
dsid = get_dsid(ds)
print("Current dataset is", dsid)
if not is_dijet(dsid):
continue
if xsecs.datasets[dsid]['denominator'] == 0:
continue
weight = xsecs.get_weight(dsid)
this_dsid = get_hist(ds, edges) * weight
parts[dsid] = np.array(this_dsid)
hist += this_dsid
draw_hist(hist,
edges,
args.out_dir,
parts,
file_name='calculated_dijet.pdf')
save_hist(hist, edges, args.out_dir, 'calculated_jetpt.h5', 'dijet')
def run_higgs_reweighted(edges, args):
hist = 0
parts = {}
out_dir = args.out_dir
with File('%s/calculated_jetpt.h5' % out_dir, 'r') as h5file:
num = h5file['dijet']['hist']
denom = h5file['higgs']['hist']
ratio = np.zeros_like(num)
valid = np.asarray(denom) > 0.0
ratio[valid] = num[valid] / denom[
valid] # this is the histograms ratio
for ds in args.datasets:
dsid = get_dsid(ds)
if not is_dihiggs(dsid):
continue
this_dsid = get_hist_reweighted(ds, edges, ratio)
parts[dsid] = np.array(this_dsid)
hist += this_dsid
draw_hist(hist,
edges,
args.out_dir,
parts,
file_name='calculated_higgs_reweight.pdf')
def run_higgs(edges, args):
hist = 0
parts = {}
for ds in args.datasets:
dsid = get_dsid(ds)
if not is_dihiggs(dsid):
continue
this_dsid = get_hist(ds, edges)
parts[dsid] = np.array(this_dsid)
hist += this_dsid
draw_hist(hist, edges, args.out_dir, parts, file_name='higgs.pdf')
save_hist(hist, edges, args.out_dir, 'jetpt.h5', 'higgs')
def run_wz(edges, args):
hist = 0
parts = {}
for ds in args.datasets:
dsid = get_dsid(ds)
if not is_wz(dsid):
continue
this_dsid = get_hist(ds, edges)
parts[dsid] = np.array(this_dsid)
hist += this_dsid
draw_hist(hist, edges, args.out_dir, parts, file_name='calculated_wz.pdf')
save_hist(hist, edges, args.out_dir, 'calculated_jetpt.h5', 'wz')
def create_wz_reweights(edges, args):
"""
Calculates correct weighting ratio for pt from histogram files. Then applies this to each dataset
"""
out_dir = args.out_dir
with File('%s/jetpt.h5' % out_dir, 'r') as h5file:
num = h5file['dijet']['hist']
denom = h5file['wz']['hist']
ratio = np.zeros_like(num)
valid = np.asarray(denom) > 0.0
ratio[valid] = num[valid] / denom[valid] # This is the ratio I need
for ds in args.datasets:
dsid = get_dsid(ds)
if not is_wz(dsid):
continue
create_reweights_from_ratio(ds, edges, ratio)
def build_summary_dataset(datasets, args):
with open(args.denominator, 'r') as denom_file:
denom = get_denom_dict(denom_file)
with open(args.cross_sections, 'r') as xsec_file:
xsecs = CrossSections(xsec_file, denom)
samples = defaultdict(lambda: defaultdict(list))
for ds_name in datasets:
dsid = get_dsid(ds_name)
if is_dijet(dsid):
if xsecs.datasets[dsid]['denominator'] == 0:
continue
sample, total_entries = sample_slice(ds_name, args.sample_size)
upweight = total_entries / sample['fat_jet'].shape[0]
weight = xsecs.get_weight(dsid) * upweight
sample['fat_jet']['mcEventWeight'] *= weight
for object_name, obj in sample.items():
samples['dijet'][object_name].append(obj)
elif is_dihiggs(dsid):
sample, total_entries = sample_slice(ds_name, args.sample_size)
upweight = total_entries / sample['fat_jet'].shape[0]
pt = sample['fat_jet']['pt']
ratio, edges = get_ratio(args.ratio_dir)
indices = np.digitize(pt, edges) - 1
weight = ratio[indices] * upweight
sample['fat_jet']['mcEventWeight'] = weight
for object_name, obj in sample.items():
samples['higgs'][object_name].append(obj)
else:
print(f'skipping {dsid}!')
continue
# merge things
merged_samples = defaultdict(dict)
for phys_proc_name, obj_dict in samples.items():
for obj_name, samp_list in obj_dict.items():
merged = np.concatenate(samp_list, 0)
merged_samples[phys_proc_name][obj_name] = merged
if args.plots_dir:
draw_hist(merged_samples, ['higgs', 'dijet'], args.plots_dir)
