def ccsn_limits(sn_type):
base = "analyses/ccsn/stasik_2017/calculate_sensitivity/"
path = base + sn_type + "/real_unblind/"
savepath = limit_output_path(path)
print("Loading limits from", savepath)
with open(savepath, "r") as f:
results = Pickle.load(f)
return results
def limit_sens(mh_name, pdf_type):
base = f'analyses/ccsn/stasik2017/calculate_sensitivity/{mh_name}/{pdf_type}/'
sub_base = base.split(os.sep)
for i, _ in enumerate(sub_base):
p = sub_base[0]
for d in range(1, i):
p += f'{os.sep}{sub_base[d]}'
p = limits_dir + p
if not os.path.isdir(p):
logging.debug(f'making directory {p}')
os.mkdir(p)
return limit_output_path(base)
def limit_sens(base):
# base = f'{raw_output_dir}/calculate_sensitivity/{mh_name}/{pdf_type}/'
sub_base = base.split(os.sep)
for i, _ in enumerate(sub_base):
p = sub_base[0]
for d in range(1, i):
p += f"{os.sep}{sub_base[d]}"
p = limits_dir + p
if not os.path.isdir(p):
logging.debug(f"making directory {p}")
os.mkdir(p)
return limit_output_path(base)
def __init__(self, unblind_dict, seed=None, scan_2d=False, **kwargs):
self.unblind_dict = unblind_dict
unblind_dict["unblind_bool"] = True
unblind_dict["mock_unblind_bool"] = mock_unblind
unblind_dict["inj_dict"] = {}
if np.logical_and(not mock_unblind, not disable_warning):
self.check_unblind()
if mock_unblind is False:
self.mock_unblind = False
else:
self.mock_unblind = True
ParentMiminisationHandler.__init__(self, unblind_dict)
try:
if self.mock_unblind:
self.limit_path = limit_output_path(
self.unblind_dict["name"] + "mock_unblind/")
self.unblind_res_path = unblinding_output_path(
self.unblind_dict["name"] + "mock_unblind/")
else:
self.limit_path = limit_output_path(
self.unblind_dict["name"] + "real_unblind/")
self.unblind_res_path = unblinding_output_path(
self.unblind_dict["name"] + "real_unblind/")
except KeyError:
self.limit_path = np.nan
self.unblind_res_path = np.nan
if self.name != " /":
if self.mock_unblind:
self.name += "mock_unblind/"
else:
self.name += "real_unblind/"
self.plot_dir = plot_output_dir(self.name)
try:
os.makedirs(os.path.dirname(self.unblind_res_path))
except OSError:
pass
# Avoid redoing unblinding
if not reunblind:
try:
logger.info(
"Not re-unblinding, loading results from {0}".format(
self.unblind_res_path))
with open(self.unblind_res_path, "rb") as f:
self.res_dict = pickle.load(f)
except FileNotFoundError:
logger.warning(
"No pickle files containing unblinding results found. "
"Re-unblinding now.")
# Otherwise unblind
if not hasattr(self, "res_dict"):
logger.info("Unblinding catalogue")
# Minimise likelihood and produce likelihood scans
self.res_dict = self.simulate_and_run(0, seed)
logger.info(self.res_dict)
# Quantify the TS value significance
self.ts = np.array([self.res_dict["TS"]])[0]
self.ts_type = unblind_dict.get('ts_type', 'Standard')
self.sigma = np.nan
logger.info("Test Statistic of: {0}".format(self.ts))
ub_res_dict = {
"Parameters": self.res_dict['Parameters'],
"TS": self.res_dict['TS'],
"Flag": self.res_dict['Flag']
}
logger.info("Saving unblinding results to {0}".format(
self.unblind_res_path))
with open(self.unblind_res_path, "wb") as f:
pickle.dump(ub_res_dict, f)
try:
path = self.unblind_dict["background_ts"]
self.pickle_dir = name_pickle_output_dir(path)
self.output_file = self.plot_dir + "TS.pdf"
self.compare_to_background_TS()
except KeyError:
logger.warning("No Background TS Distribution specified. "
"Cannot assess significance of TS value.")
if full_plots:
self.calculate_upper_limits()
if isinstance(self, FlareMinimisationHandler):
self.neutrino_lightcurve()
else:
self.scan_likelihood(scan_2d=scan_2d)
def __init__(self, unblind_dict, seed=None, scan_2d=False, **kwargs):
self.unblind_dict = unblind_dict
unblind_dict["unblind_bool"] = True
unblind_dict["mock_unblind_bool"] = mock_unblind
unblind_dict["inj_dict"] = {}
# specifies whether all available background distributions should be merged
# to improve statistics (see loop over subdirs in self.compare_to_background_TS()
self.use_merged_background_ts_distributions = kwargs.get(
"merge_background_ts_distr", True)
self.background_median = np.nan
# gives the medians for the separate background distributions
# this is used when background TS distributions are not merged
self.background_median_gamma = dict()
if np.logical_and(not mock_unblind, not disable_warning):
self.check_unblind()
if mock_unblind is False:
self.mock_unblind = False
else:
self.mock_unblind = True
ParentMiminisationHandler.__init__(self, unblind_dict)
try:
if self.mock_unblind:
self.limit_path = limit_output_path(
self.unblind_dict["name"] + "mock_unblind/")
self.unblind_res_path = unblinding_output_path(
self.unblind_dict["name"] + "mock_unblind/")
else:
self.limit_path = limit_output_path(
self.unblind_dict["name"] + "real_unblind/")
self.unblind_res_path = unblinding_output_path(
self.unblind_dict["name"] + "real_unblind/")
except KeyError:
self.limit_path = np.nan
self.unblind_res_path = np.nan
if self.name != " /":
if self.mock_unblind:
self.name += "mock_unblind/"
else:
self.name += "real_unblind/"
self.plot_dir = plot_output_dir(self.name)
try:
os.makedirs(os.path.dirname(self.unblind_res_path))
except OSError:
pass
# Avoid redoing unblinding
if not reunblind:
try:
logger.info(
"Not re-unblinding, loading results from {0}".format(
self.unblind_res_path))
with open(self.unblind_res_path, "rb") as f:
self.res_dict = pickle.load(f)
except FileNotFoundError:
logger.warning(
"No pickle files containing unblinding results found. "
"Re-unblinding now.")
# Otherwise unblind
if not hasattr(self, "res_dict"):
logger.info("Unblinding catalogue")
# Minimise likelihood and produce likelihood scans
self.res_dict = self.simulate_and_run(0, seed)
logger.info(self.res_dict)
# Quantify the TS value significance
self.bkg_ts_array = None
self.ts = np.array([self.res_dict["TS"]])[0]
self.sigma = np.nan
# the raw pre-trial p-value will be calculated using only the TS distributions without fitting them
# This is only valid if you ran enough background trials!
self.raw_pre_trial_pvalue = np.nan
self.ts_type = get_ts_fit_type(unblind_dict)
logger.info("Test Statistic of: {0}".format(self.ts))
ub_res_dict = {
"Parameters": self.res_dict["Parameters"],
"TS": self.res_dict["TS"],
"Flag": self.res_dict["Flag"],
}
logger.info("Saving unblinding results to {0}".format(
self.unblind_res_path))
with open(self.unblind_res_path, "wb") as f:
pickle.dump(ub_res_dict, f)
try:
path = self.unblind_dict["background_ts"]
self.pickle_dir = name_pickle_output_dir(path)
self.output_file = self.plot_dir + "TS.pdf"
self.compare_to_background_TS()
except KeyError as e:
logger.warning(f"No Background TS Distribution specified. "
f"Cannot assess significance of TS value: {e}!")
if full_plots:
self.calculate_upper_limits()
if isinstance(self, FlareMinimisationHandler):
self.neutrino_lightcurve()
else:
self.scan_likelihood(scan_2d=scan_2d)