take a marginalized likelihood file name in the Data init, simplify some logic...

take a marginalized likelihood file name in the Data init, simplify some logic involving when to comput quantities needed for likelihood ranking, check for Nans in FAR column before sending gracedb alerts

gstlal-ugly/python/inspiral.py

@@ -295,7 +295,7 @@ def gen_likelihood_control_doc(far, instruments, name = u"gstlal_inspiral_likeli
 
 
 class Data(object):
-	def __init__(self, filename, process_params, instruments, seg, coincidence_threshold, FAR, injection_filename = None, time_slide_file = None, comment = None, tmp_path = None, assign_likelihoods = False, likelihood_snapshot_interval = None, thinca_interval = 50.0, gracedb_far_threshold = None, likelihood_file = None, gracedb_group = "Test", gracedb_type = "LowMass", verbose = False):
+	def __init__(self, filename, process_params, instruments, seg, coincidence_threshold, FAR, marginalized_likelihood_file = None, injection_filename = None, time_slide_file = None, comment = None, tmp_path = None, assign_likelihoods = False, likelihood_snapshot_interval = None, thinca_interval = 50.0, gracedb_far_threshold = None, likelihood_file = None, gracedb_group = "Test", gracedb_type = "LowMass", verbose = False):
 		#
 		# initialize
 		#
@@ -313,6 +313,7 @@ class Data(object):
 		self.verbose = verbose
 		# True to enable likelihood assignment
 		self.assign_likelihoods = assign_likelihoods
+		self.marginalized_likelihood_file = marginalized_likelihood_file
 		# Set to None to disable period snapshots, otherwise set to seconds
 		self.likelihood_snapshot_interval = likelihood_snapshot_interval
 		# Set to 1.0 to disable background data decay
@@ -458,29 +459,30 @@ class Data(object):
 				event.event_id = self.sngl_inspiral_table.get_next_id()
 
 			# update likelihood snapshot if needed
-			if self.assign_likelihoods and (self.likelihood_snapshot_timestamp is None or (self.likelihood_snapshot_interval is not None and buf_timestamp - self.likelihood_snapshot_timestamp >= self.likelihood_snapshot_interval)):
+			if (self.likelihood_snapshot_timestamp is None or (self.likelihood_snapshot_interval is not None and buf_timestamp - self.likelihood_snapshot_timestamp >= self.likelihood_snapshot_interval)):
 				self.likelihood_snapshot_timestamp = buf_timestamp
-
-				# smooth the distribution_stats
-				self.far.smooth_distribution_stats()
-				# update stream thinca's likelihood data
-				self.stream_thinca.set_likelihood_data(self.far.distribution_stats.smoothed_distributions, self.far.distribution_stats.likelihood_params_func)
-				# FIXME don't hard code
-				remap = {frozenset(["H1", "H2", "L1"]) : frozenset(["H1", "L1"]), frozenset(["H1", "H2", "V1"]) : frozenset(["H1", "V1"]), frozenset(["H1", "H2", "L1", "V1"]) : frozenset(["H1", "L1", "V1"])}
-
-				# generate the background likelihood distributions
-				self.far.compute_joint_instrument_background(remap, verbose = self.verbose)
-				self.ranking_data = far.RankingData(self.far)
-				self.ranking_data.compute_joint_cdfs()
+				if self.assign_likelihoods:
+					# smooth the distribution_stats
+					self.far.smooth_distribution_stats(verbose = self.verbose)
+					# update stream thinca's likelihood data
+					self.stream_thinca.set_likelihood_data(self.far.distribution_stats.smoothed_distributions, self.far.distribution_stats.likelihood_params_func)
+
+					# Read in the the background likelihood distributions that should have been updated asynchronously
+					# FIXME don't screw with startTable methods this way
+					# Hack to read XML after dbtables import
+					orig_start_table = ligolw.LIGOLWContentHandler.startTable
+					ligolw.LIGOLWContentHandler.startTable = lsctables.startTable
+					self.ranking_data, procid = far.RankingData.from_xml(utils.load_filename(self.marginalized_likelihood_file, verbose = self.verbose))
+					ligolw.LIGOLWContentHandler.startTable = orig_start_table
+					self.ranking_data.compute_joint_cdfs()
+				else:
+					self.ranking_data = None
 
 				# write the new distribution stats to disk
 				utils.write_filename(gen_likelihood_control_doc(self.far, self.instruments), self.likelihood_file, gz = (self.likelihood_file or "stdout").endswith(".gz"), verbose = False, trap_signals = None)
 
 			# run stream thinca
-			if self.assign_likelihoods:
-				noncoinc_sngls = self.stream_thinca.add_events(events, buf_timestamp, FAP = self.ranking_data)
-			else:
-				noncoinc_sngls = self.stream_thinca.add_events(events, buf_timestamp)
+			noncoinc_sngls = self.stream_thinca.add_events(events, buf_timestamp, FAP = self.ranking_data)
 
 			# update the parameter distribution data.  only
 			# update from sngls that weren't used in coincs
@@ -553,7 +555,7 @@ class Data(object):
 				# do we keep this event?
 				#
 
-				if false_alarm_rate > self.gracedb_far_threshold:
+				if false_alarm_rate > self.gracedb_far_threshold or numpy.isnan(false_alarm_rate):
 					continue
 
 				#