def save_weights(fp, sample_names, stitching_dict):
for sample_name in sample_names:
sample_dir = fp.mkdir(sample_name)
hist_dir = sample_dir.mkdir(SPLITVAR)
hist_dir.cd()
for bin_count_name in stitching_dict:
bin_array = array.array('f', list(range(max(bin_keys) + 2)))
hist = ROOT.TH1D(bin_count_name, bin_count_name,
len(bin_array) - 1, bin_array)
hist.SetDirectory(hist_dir)
hist.SetTitle(bin_count_name)
hist.SetXTitle(SPLITVAR)
for bin_idx in bin_keys:
hist.SetBinContent(bin_idx + 1,
stitching_dict[bin_count_name][bin_idx])
hist.GetXaxis().SetBinLabel(
bin_idx + 1,
"%d <= %s < %d" % (bin_idx, SPLITVAR, bin_idx + 1))
hist.Write()
def create_histogram(key, title):
if is2D:
histograms[key] = ROOT.TH2D(key, title,
len(bins_x) - 1, bins_x,
len(bins_y) - 1, bins_y)
for bin_idx in range(len(bins_x) - 1):
histograms[key].GetXaxis().SetBinLabel(
bin_idx + 1, '%d <= %s < %d' %
(bins_x[bin_idx], x_var, bins_x[bin_idx + 1]))
for bin_idx in range(len(bins_y) - 1):
histograms[key].GetYaxis().SetBinLabel(
bin_idx + 1, '%d <= %s < %d' %
(bins_y[bin_idx], y_var, bins_y[bin_idx + 1]))
histograms[key].SetXTitle(x_var)
histograms[key].SetYTitle(y_var)
else:
histograms[key] = ROOT.TH1D(key, title, len(bins_x) - 1, bins_x)
for bin_idx in range(len(bins_x) - 1):
histograms[key].GetXaxis().SetBinLabel(
bin_idx + 1, '%d <= %s < %d' %
(bins_x[bin_idx], x_var, bins_x[bin_idx + 1]))
histograms[key].SetXTitle(x_var)
def comp_weights_1(f, samples, samples_to_stitch, split_var, apply_sf=True):
inclusive_samples = samples_to_stitch['inclusive']['samples']
inclusive_binning = samples_to_stitch['inclusive'][split_var]
split_dict = samples_to_stitch['exclusive'][split_var]
split_binning = [sample['value'] for sample in split_dict]
complete_binning = list(
sorted(
list(
map(
float,
set(inclusive_binning) | set(
list(itertools.chain.from_iterable(split_binning)))))))
inclusive_xs = -1
for sample_key, sample_entry in samples.items():
if sample_key == 'sum_events': continue
if sample_entry['process_name_specific'] == inclusive_samples[0]:
inclusive_xs = sample_entry['xsection']
assert (inclusive_xs > 0)
# sum the inclusive nof events
inclusive_nof_events = {}
for sample_key, sample_entry in samples.items():
if sample_key == 'sum_events': continue
if sample_entry['process_name_specific'] in inclusive_samples:
if not inclusive_nof_events:
inclusive_nof_events = copy_nof_events(sample_entry)
else:
nof_events_keys = set(nof_key
for nof_key in sample_entry['nof_events']
if is_valid_event_type(nof_key))
assert (nof_events_keys == set(inclusive_nof_events))
for nof_key, nof_arr in sample_entry['nof_events'].items():
if not is_valid_event_type(nof_key):
continue
assert (len(nof_arr) == len(inclusive_nof_events[nof_key]))
for idx, nof in enumerate(nof_arr):
assert (nof > 0)
inclusive_nof_events[nof_key][idx] += nof
# sum the binned nof events
for binned_sample in split_dict:
nof_events = {}
xs = -1
for sample_key, sample_entry in samples.items():
if sample_key == 'sum_events': continue
if sample_entry['process_name_specific'] in binned_sample[
'samples']:
if not nof_events:
nof_events = copy_nof_events(sample_entry)
inclusive_nof_events_type = set(
event_type
for event_type in inclusive_nof_events.keys()
if is_valid_event_type(event_type))
nof_events_type = set(event_type
for event_type in nof_events.keys()
if is_valid_event_type(event_type))
assert (inclusive_nof_events_type == nof_events_type)
else:
nof_events_keys = set(
nof_key for nof_key in sample_entry['nof_events']
if is_valid_event_type(nof_key))
assert (nof_events_keys == set(nof_events.keys()))
for nof_key, nof_arr in sample_entry['nof_events'].items():
if not is_valid_event_type(nof_key):
continue
assert (len(nof_arr) == len(nof_events[nof_key]))
for idx, nof in enumerate(nof_arr):
assert (nof > 0)
nof_events[nof_key][idx] += nof
if xs < 0:
xs = sample_entry['xsection']
assert (xs > 0)
binned_sample['xsection'] = xs
binned_sample['nof_events'] = nof_events
lumis = {}
for nof_key, nof_arr in binned_sample['nof_events'].items():
lumis[nof_key] = list(
map(lambda nof: nof / binned_sample['xsection'], nof_arr))
binned_sample['lumis'] = lumis
# compute integrated luminosities for the inclusive sample
inclusive_lumis = {}
for nof_key, nof_arr in inclusive_nof_events.items():
inclusive_lumis[nof_key] = list(
map(lambda nof: nof / inclusive_xs, nof_arr))
# decide on the bin indices
idxs_split_sample = []
for binned_sample in split_dict:
binned_idx = complete_binning.index(binned_sample['value'][0]) + 1
binned_sample['idx'] = binned_idx
idxs_split_sample.append(binned_idx)
for inclusive_sample in inclusive_samples:
if inclusive_sample not in [
key.GetName() for key in f.GetListOfKeys()
]:
histogram_dir_root = f.mkdir(inclusive_sample)
else:
histogram_dir_root = f.Get(inclusive_sample)
if split_var not in [
key.GetName() for key in histogram_dir_root.GetListOfKeys()
]:
histogram_dir = histogram_dir_root.mkdir(split_var)
else:
histogram_dir = histogram_dir_root.Get(split_var)
histogram_dir.cd()
for nof_key, lumi_arr in inclusive_lumis.items():
if not is_valid_event_type(nof_key):
continue
for idx, lumi_incl in enumerate(lumi_arr):
histogram_name = '%s_%d' % (nof_key, idx)
binning = array.array('f', complete_binning)
histogram = ROOT.TH1D(histogram_name, histogram_name,
len(binning) - 1, binning)
histogram.SetDirectory(histogram_dir)
histogram.SetXTitle(split_var)
for split_idx in range(1, len(binning)):
if split_idx in idxs_split_sample:
for binned_sample in split_dict:
if split_idx == binned_sample['idx']:
lumi_split = binned_sample['lumis'][nof_key][
idx]
if binning[split_idx] > inclusive_binning[-1] or \
binning[split_idx] < inclusive_binning[0]:
lumi_incl_calc = 0.
else:
lumi_incl_calc = lumi_incl
weight = lumi_incl_calc / (lumi_incl_calc +
lumi_split)
histogram.SetBinContent(split_idx, weight)
else:
histogram.SetBinContent(split_idx, 1.)
histogram.GetXaxis().SetBinLabel(
split_idx, '%.0f <= %s < %.0f' %
(complete_binning[split_idx - 1], split_var,
complete_binning[split_idx]))
histogram.Write()
for binned_sample in split_dict:
for sample_name in binned_sample['samples']:
if sample_name not in [key.GetName() for key in f.GetListOfKeys()]:
histogram_dir_root = f.mkdir(sample_name)
else:
histogram_dir_root = f.Get(sample_name)
if split_var not in [
key.GetName()
for key in histogram_dir_root.GetListOfKeys()
]:
histogram_dir = histogram_dir_root.mkdir(split_var)
else:
histogram_dir = histogram_dir_root.Get(split_var)
histogram_dir.cd()
for nof_key, lumi_arr in inclusive_lumis.items():
if not is_valid_event_type(nof_key):
continue
for idx, lumi_incl in enumerate(lumi_arr):
histogram_name = '%s_%d' % (nof_key, idx)
binning = array.array('f', complete_binning)
histogram = ROOT.TH1D(histogram_name, histogram_name,
len(binning) - 1, binning)
histogram.SetDirectory(histogram_dir)
histogram.SetXTitle(split_var)
for split_idx in range(1, len(binning)):
if split_idx == binned_sample['idx']:
lumi_split = binned_sample['lumis'][nof_key][idx]
if binning[split_idx] > inclusive_binning[-1] or \
binning[split_idx] < inclusive_binning[0]:
weight = 1.
else:
if apply_sf:
weight = lumi_split / (lumi_incl +
lumi_split)
else:
weight = lumi_incl / (lumi_incl +
lumi_split)
assert (weight >= 0.)
histogram.SetBinContent(split_idx, weight)
else:
histogram.SetBinContent(split_idx, 0.)
histogram.GetXaxis().SetBinLabel(
split_idx, '%.0f <= %s < %.0f' %
(complete_binning[split_idx - 1], split_var,
complete_binning[split_idx]))
histogram.Write()
"The most frequent LHE weight is: {:.6e}".format(most_frequent_weight))
with open(output_file, 'w') as output_file_ptr:
output_file_ptr.write('{:.6e}'.format(most_frequent_weight))
logging.info("Wrote output file: {}".format(output_file))
if plot_files:
has_neg_weights = any(weight_freq[0] < 0
for weight_freq in weights_by_frequency)
binning_max = MAX_CUTOFF * most_frequent_weight
binning_min = -MAX_CUTOFF * most_frequent_weight
if not has_neg_weights:
binning_min = max(0., binning_min)
binning = array.array(
'f', list(np.linspace(binning_min, binning_max, PLOT_BINS + 1)))
histogram = ROOT.TH1D(GENWEIGHT_NAME, GENWEIGHT_NAME, PLOT_BINS, binning)
for weight_freq in weights_by_frequency:
histogram.Fill(weight_freq[0], weight_freq[1])
histogram.GetXaxis().SetRange(0, histogram.GetNbinsX() + 1)
for plot_file in plot_files:
title = os.path.splitext(os.path.basename(plot_file))[0]
histogram.SetTitle(title)
canvas = ROOT.TCanvas()
canvas.SetCanvasSize(1000, 800)
canvas.SetLogy(True)
histogram.Draw('hist')
canvas.SaveAs(plot_file)
canvas.Close()
del canvas
for sample_idx, sample_name in enumerate(results.keys()):
logging.info("Creating plots for sample {} ({}/{})".format(
sample_name, sample_idx + 1, nof_samples))
plot_output(pdf)
out_fptr = ROOT.TFile.Open(output, 'recreate')
for sample_key in results:
logging.info("Saving histograms for process {}".format(sample_key))
sample_dir = out_fptr.mkdir(sample_key)
sample_dir.cd()
assert ('central' in results[sample_key])
for sys_key in results[sample_key]:
if central_only and sys_key != 'central':
continue
if not sys_key or sys_key.startswith('envelope'):
continue
ratios = results[sample_key][sys_key]['ratio']['count']
ratio_errs = results[sample_key][sys_key]['ratio']['err']
nbins = len(ratios)
xbins = array.array('f', range(nbins))
histogram = ROOT.TH1D(sys_key, sys_key, len(xbins) - 1, xbins)
histogram.SetDirectory(sample_dir)
histogram.SetXTitle('# preselected jets')
histogram.SetTitle('{} ({})'.format(sample_key, sys_key))
for bin_idx in range(nbins):
histogram.SetBinContent(bin_idx + 1, ratios[bin_idx])
histogram.SetBinError(bin_idx + 1, ratio_errs[bin_idx])
histogram.Write()
del histogram
out_fptr.Close()