def run(self):
wrp = next(rebin_chhadiso(
gen.gen_sum(
[
gen.fs_filter_active_sort_load({
"analyzer" : sb_anzlrs,
"is_data" : True
})
]
)
))
# multiply with weight
if do_dist_reweighting:
wrp = gen.op.prod((
settings.post_proc_dict["TemplateFitToolChHadIsoSbBkgInputBkgWeight"],
wrp,
))
wrp.lumi = settings.data_lumi_sum()
self.result = [wrp]
gen.consume_n_count(
gen.save(
gen.canvas((self.result,)),
lambda c: self.plot_output_dir + c.name
)
)
def write_tabular(self):
table = [
r"\begin{tabular}{l c c } \\",
r"\hline",
r"\hline",
r"Sample & fake-rate (MC) & fake-rate (MC truth) \\",
r"\hline",
]
data_lumi_sum = settings.data_lumi_sum()
for smp in sorted(settings.active_samples + ["TTPoHe","TTMadG","TTMCNLO"]):
smp = settings.samples[smp]
if smp.is_data:
continue
norm = data_lumi_sum / smp.lumi
tight = smp.log_event_counts["FullTightIDCount,"] * norm
real = smp.log_event_counts["realFullTightIDCount,"] * norm
shilp = smp.log_event_counts["ShilpiSumDirect,"] * norm
table.append(
"\;\;\;"
+ smp.name.replace("_", r"\_")
+ r" & $%.1f\pm%.1f$ & $%.1f\pm%.1f$ \\" % (
shilp,
0.3 * shilp, # 30% on fr count
tight - real,
((tight - real) * norm)**.5 / norm, # MC stat error
)
)
table += (
r"\hline",
r"\hline",
r"\end{tabular}",
)
with open(self.plot_output_dir + "shilpi_tabular.tex", "w") as f:
f.writelines(map(lambda l: l + "\n", table))
def get_shilp_counts(self, c):
data_lumi_sum = settings.data_lumi_sum()
for smp in settings.mc_samples().itervalues():
norm = data_lumi_sum / smp.lumi
tight = smp.log_event_counts[self.tight_cnt] * norm
real = smp.log_event_counts["realFullTightIDCount,"] * norm
c.tight += tight
c.shilp += tight - real
def run(self):
wrp = next(rebin_chhadiso(
gen.gen_sum(
[gen.fs_filter_active_sort_load({
"analyzer" : "TemplateRandConereal",
"is_data" : True
})]
)
))
# normalize to mc expectation
integral_real = next(
gen.gen_integral(
gen.gen_norm_to_data_lumi(
gen.filter(
settings.post_proc_dict["TemplateStacks"],
{"analyzer": "TemplateRandConereal"}
)
)
)
)
print integral_real
wrp = gen.op.prod((
gen.op.norm_to_integral(wrp),
integral_real
))
# multiply with weight
if do_dist_reweighting:
wrp = gen.op.prod((
settings.post_proc_dict["TemplateFitToolRandConeIsoInputSigWeight"],
wrp,
))
wrp.lumi = settings.data_lumi_sum()
self.result = [wrp]
gen.consume_n_count(
gen.save(
gen.canvas((self.result,)),
lambda c: self.plot_output_dir + c.name
)
)
def run(self):
self.configure()
if not self.n_sig_ttgam_wrp:
self.message("WARNING Did not find result in post_proc_dict. Skipping...")
return
# store results in wrapper
r = copy.deepcopy(self.n_sig_ttgam_wrp)
r.name = self.name
self.result = r
# prepare mc counts
class counts(object): pass
c = counts()
c.sig_pre = 0.
c.sig_fid = 0.
c.sig_post = 0.
c.bkg_pre = 0.
c.bkg_post = 0.
c.tt_pre = 0.
c.tt_post = 0
for smp in settings.mc_samples().itervalues():
legend = smp.legend
if legend == "t#bar{t}#gamma (Signal)":
c.sig_pre += smp.log_event_counts[self.get_sig_count_name(self.pre_count_name)] / smp.lumi
c.sig_fid += smp.log_event_counts[self.get_sig_count_name(self.fid_count_name)] / smp.lumi
c.sig_post += smp.log_event_counts[self.get_sig_count_name(self.post_count_name)] / smp.lumi
else:
c.bkg_pre += smp.log_event_counts[self.pre_count_name] / smp.lumi
c.bkg_post += smp.log_event_counts[self.post_count_name] / smp.lumi
if legend == "t#bar{t} inclusive":
c.tt_pre += smp.log_event_counts[self.pre_count_name] / smp.lumi
c.tt_post += smp.log_event_counts[self.post_count_name] / smp.lumi
data_lumi_sum = settings.data_lumi_sum()
for k in c.__dict__.keys():
c.__dict__[k] *= data_lumi_sum
# prepare data counts
c.data_pre = 0.
c.data_post = 0.
for smp in settings.data_samples().itervalues():
c.data_pre += smp.log_event_counts[self.pre_count_name]
c.data_post += smp.log_event_counts[self.post_count_name]
# selection performance
r.eff_gamma = c.sig_post / c.sig_fid
r.eff_gamma_fid = c.sig_fid / c.sig_pre
r.pur_tt = (c.tt_pre + c.sig_pre) / (c.bkg_pre + c.sig_pre)
r.N_presel_data = c.data_pre
r.N_sel_data = c.data_post
r.StoB_gamma = c.sig_post / c.bkg_post
r.StoB_presel = c.tt_pre / (c.bkg_pre - c.tt_pre)
# background-substracted number of ttgamma signal events
# r.n_sig_ttgam = self.n_sig_ttgam_wrp.n_sig_ttgam
# R_fid
R_fid_denom = r.eff_gamma * r.N_presel_data * r.pur_tt
r.R_fid = r.n_sig_ttgam / R_fid_denom
r.R_fid_err_stat= r.n_sig_ttgam_err / R_fid_denom
# R
R_denom = r.eff_gamma_fid * r.eff_gamma * r.N_presel_data * r.pur_tt
r.R = r.n_sig_ttgam / R_denom
r.R_err_stat = r.n_sig_ttgam_err / R_denom
# xsec
r.xsec = r.R * settings.ttbar_xsec_cms
r.xsec_err_stat = r.xsec * r.R_err_stat / r.R
self.message(str(r))
def get_shilp_counts(self, c):
data_lumi_sum = settings.data_lumi_sum()
for smp in settings.mc_samples().itervalues():
norm = data_lumi_sum / smp.lumi
c.tight += smp.log_event_counts[self.tight_cnt] * norm
c.shilp += smp.log_event_counts[self.shilp_cnt] * norm
