def _print_likelihood(evaluators):
tot_likeli = 0.
dict_likeli = {}
for analysis, v in evaluators.iteritems():
ana_likeli =0.
AdvPrint.set_cout_file(Info.files['output_evaluation_likelihood'][analysis], True)
AdvPrint.mute()
AdvPrint.cout("SR o b db s ds likeli")
for sr, ev in v.iteritems():
AdvPrint.cout(sr+" "
+str(float(ev.obs))+" "
+str(float(ev.bkg))+" "
+str(float(ev.bkg_err))+" "
+str(ev.resultCollector.signal_normevents)+" "
+str(ev.resultCollector.signal_err_tot)+" "
+str(ev.likelihood))
ana_likeli += ev.likelihood
AdvPrint.format_columnated_file(Info.files['output_evaluation_likelihood'][analysis])
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
dict_likeli[analysis] = ana_likeli
tot_likeli += ana_likeli
AdvPrint.set_cout_file(Info.files['likelihood'], True)
AdvPrint.mute()
AdvPrint.cout("Analysis -2lnL")
for a in dict_likeli:
AdvPrint.cout(a+" "+str(dict_likeli[a]))
AdvPrint.cout("\nTotal: "+str(tot_likeli))
AdvPrint.format_columnated_file(Info.files['likelihood'])
def get_resultCollectors(self):
""" Gathers results from all events"""
# setup resultCollector object
resultCollectors_pr = dict()
for analysis in Info.analyses:
resultCollectors_pr[analysis] = dict()
signal_regions = Info.get_analysis_parameters(analysis)["signal_regions"]
for sr in signal_regions:
resultCollectors_pr[analysis][sr] = ResultCollector(self.name, analysis, sr)
# loop over all associated events and average results in all resultCollectors
for event in self.eventsList:
resultCollectors_ev = event.get_resultCollectors()
for analysis in resultCollectors_pr:
for sr in resultCollectors_pr[analysis]:
resultCollectors_pr[analysis][sr].add_and_average(resultCollectors_ev[analysis][sr])
# Write process file, if wanted
if Info.parameters["ProcessResultFileColumns"] != []:
AdvPrint.mute()
AdvPrint.set_cout_file(self.result_output_file, True)
for col in Info.parameters["ProcessResultFileColumns"]:
AdvPrint.cout(col+" ", "nlb")
AdvPrint.cout("")
for a in sorted(resultCollectors_pr.keys()):
for sr in sorted(resultCollectors_pr[a].keys()):
AdvPrint.cout(resultCollectors_pr[a][sr].line_from_data(Info.parameters["ProcessResultFileColumns"]))
AdvPrint.format_columnated_file(self.result_output_file)
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
return resultCollectors_pr
def get_resultCollectors(self):
""" Gathers results from all events"""
# setup resultCollector object
resultCollectors_pr = dict()
for analysis in Info.analyses:
resultCollectors_pr[analysis] = dict()
signal_regions = Info.get_analysis_parameters(
analysis)["signal_regions"]
for sr in signal_regions:
resultCollectors_pr[analysis][sr] = ResultCollector(
self.name, analysis, sr)
# loop over all associated events and average results in all resultCollectors
for event in self.eventsList:
resultCollectors_ev = event.get_resultCollectors()
for analysis in resultCollectors_pr:
for sr in resultCollectors_pr[analysis]:
resultCollectors_pr[analysis][sr].add_and_average(
resultCollectors_ev[analysis][sr])
# Write process file, if wanted
if Info.parameters["ProcessResultFileColumns"] != []:
AdvPrint.mute()
AdvPrint.set_cout_file(self.result_output_file, True)
for col in Info.parameters["ProcessResultFileColumns"]:
AdvPrint.cout(col + " ", "nlb")
AdvPrint.cout("")
for a in sorted(resultCollectors_pr.keys()):
for sr in sorted(resultCollectors_pr[a].keys()):
AdvPrint.cout(resultCollectors_pr[a][sr].line_from_data(
Info.parameters["ProcessResultFileColumns"]))
AdvPrint.format_columnated_file(self.result_output_file)
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
return resultCollectors_pr
def get_resultCollectors(self):
resultCollectors = dict(
) # list of all collectors of all analyses and all signal regions
resultCollector = ResultCollector(
self.identifier, "",
"") # base collector object which we will just edit and copy
for analysis in Info.analyses:
# check if results file exists
if not os.path.isfile(self.analysis_signal_files[analysis]):
AdvPrint.cerr_exit(
"\t events::get_resultCollector() \n"
"\t Required analysis result file does not exist: \n "
"\t\t" + self.analysis_signal_files[analysis] + "\n"
"\t It is very likely that something went wrong in the delphes and/or the analysis step. \n"
"\t Please check \n "
"\t \t " + Info.files['delphes_log'] + " \n "
"\t \t " + Info.files['analysis_log'] + "* \n "
"\t for error messages and, should you not be able to fix them yourself, contact the authors under \n"
"\t \t [email protected]")
# setup resultCollector object
resultCollector.analysis = analysis
resultCollectors[analysis] = dict()
signal_regions = Info.get_analysis_parameters(
analysis)["signal_regions"]
# Read result file
f = open(self.analysis_signal_files[analysis], "r")
for line in f:
# Ignore empty or commented lines
line = line.rstrip()
if line == "" or line[0] == "#":
continue
# Read file:
line = AdvPrint.remove_extra_spaces(line)
tokens = [t for t in line.split(" ") if t != ""]
# First, read information on total events number
if tokens[0] == "MCEvents:":
resultCollector.total_mcevents = float(tokens[1])
elif tokens[0] == " SumOfWeights:":
resultCollector.total_sumofweights = float(tokens[1])
elif tokens[0] == " SumOfWeights2:":
resultCollector.total_sumofweights2 = float(tokens[1])
elif tokens[0] == " NormEvents:":
resultCollector.total_normevents = float(tokens[1])
elif tokens[0] == "XSect:":
xsect = float(tokens[1].split(" ")[0])
elif tokens[0] == " Error:":
xsecterr = float(tokens[1].split(" ")[0])
else:
# SR Sum_W Sum_W2 Acc N_Norm
for sr in signal_regions:
if tokens[0].startswith(sr):
resultCollector.sr = sr
# Read number of events
resultCollector.signal_sumofweights = float(
tokens[1])
resultCollector.signal_sumofweights2 = float(
tokens[2])
resultCollector.signal_normevents = float(
tokens[4])
# Calculate errors
if resultCollector.signal_sumofweights > 0:
resultCollector.signal_err_stat = resultCollector.signal_normevents * sqrt(
resultCollector.signal_sumofweights2
) / resultCollector.signal_sumofweights
resultCollector.signal_err_sys = resultCollector.signal_normevents * xsecterr / xsect
resultCollector.signal_err_tot = sqrt(
resultCollector.signal_err_stat**2 +
resultCollector.signal_err_sys**2)
else:
resultCollector.signal_err_stat = 0
resultCollector.signal_err_sys = 0
resultCollector.signal_err_tot = 0
# put copy of resultCollector in collector dict
resultCollectors[analysis][sr] = deepcopy(
resultCollector)
f.close()
# Write events file, if wanted
if Info.parameters["EventResultFileColumns"] != []:
AdvPrint.mute()
AdvPrint.set_cout_file(self.result_output_file, True)
for col in Info.parameters["EventResultFileColumns"]:
AdvPrint.cout(col + " ", "nlb")
AdvPrint.cout("")
for a in sorted(resultCollectors.keys()):
for sr in sorted(resultCollectors[a].keys()):
AdvPrint.cout(resultCollectors[a][sr].line_from_data(
Info.parameters["EventResultFileColumns"]))
AdvPrint.format_columnated_file(self.result_output_file)
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
return resultCollectors
def runFritz(self):
self.prepareFritz()
""" Runs Fritz """
from events import MG5Events
for event in self.eventsList:
if event.processed:
continue
fritz_command = Info.files["fritz_bin"]+" "+event.configFile
result = subprocess.Popen(fritz_command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
maxlen = 0
try:
for line in iter(result.stdout.readline, b''):
# Print to logfile. If it does not start with the Fritz::Global prefix, it comes from MG5 and should be redirected
AdvPrint.mute()
if not line.startswith("|~| ") and isinstance(event, MG5Events):
AdvPrint.set_cout_file(os.path.join(Info.paths["output_mg5"], "mg5amcatnlo_"+event.identifier+".log"))
elif "PYTHIA Rndm::dumpState" in line:
AdvPrint.set_cout_file(os.path.join(Info.paths["output_pythia"], "pythia_"+event.identifier+".log"))
else:
line = line.replace("|~| ", "")
AdvPrint.set_cout_file(os.path.join(Info.paths["output_fritz"], "fritz_"+event.identifier+".log"))
AdvPrint.cout(line.rstrip())
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
# We should not exceed the terminal terminal_width:
terminal_width = AdvPrint.get_terminal_width()
print_line = " |-> "+str(line.strip())
print_line.replace("\t", " ")
while "\r" in print_line:
print_line = print_line[print_line.index("\r"):]
len_of_print_line = len(print_line)
maxlen = max(maxlen, len(print_line))
# As we print line by line in the same terminal row, we have to add spaces if the curr line is shorter than a line before
fill_spaces = ""
if len(print_line) < maxlen:
fill_spaces = " "*(maxlen-len(print_line))
# if line is too long, make it shorter by appropriate amoung
if len(print_line+fill_spaces) >= terminal_width and len(print_line) <= terminal_width:
fill_spaces = " "*(terminal_width - len(print_line)-1)
elif len(print_line) > terminal_width:
fill_spaces = ""
print_line = print_line[:terminal_width-4]+"..."
AdvPrint.cout("\r"+print_line+fill_spaces+"\x1b[0m\r", "nlb")
except KeyboardInterrupt:
AdvPrint.cout("Caught Keyboard Signal. Aborting Fritz")
result.send_signal(signal.SIGTERM)
for line in iter(result.stderr.readline, b''):
AdvPrint.unmute()
AdvPrint.set_cout_file(Info.files['fritz_log'])
# remove nasty ROOT6-CLING warnings from on-screen output
if "cling::AutoloadingVisitor::InsertIntoAutoloadingState:" in line:
AdvPrint.mute()
elif "Missing FileEntry for ExRootAnalysis" in line:
AdvPrint.mute()
elif "requested to autoload type" in line:
AdvPrint.mute()
AdvPrint.cout(line.rstrip()+"")
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
AdvPrint.cout("")
# Abort if there was an error
result.wait()
if result.returncode != 0:
AdvPrint.cerr_exit("Fritz returned with error. Check logfiles in result folder for more information!")
# Remove all empty analysisstdout files
for f in [x for x in os.listdir(Info.paths['output_analysis']) if x.startswith("analysisstdout")]:
if os.stat(os.path.join(Info.paths['output_analysis'], f)).st_size == 0:
os.remove(os.path.join(Info.paths['output_analysis'], f))
# Associate result files to event
for a in Info.analyses:
event.analysis_signal_files[a] = os.path.join(Info.paths['output_analysis'], event.identifier+'_'+a+'_signal.dat')
if os.path.isfile(event.analysis_signal_files[a]):
AdvPrint.format_columnated_file(event.analysis_signal_files[a])
event.analysis_cutflow_files[a] = os.path.join(Info.paths['output_analysis'], event.identifier+'_'+a+'_cutflow.dat')
if os.path.isfile(event.analysis_cutflow_files[a]):
AdvPrint.format_columnated_file(event.analysis_cutflow_files[a])
# finish
event.processed = True
def evaluate(self):
""" Performs statistical evaluation of the result """
AdvPrint.cout("Evaluating Results")
resultCollectors = self.get_resultCollectors()
# evaluate all results
evaluators = dict()
for analysis in resultCollectors:
evaluators[analysis] = dict()
# only process those results and those signal regions that are given in the reference file
for analysis in Info.analyses:
signal_regions = Info.get_analysis_parameters(analysis)["signal_regions"]
for sr in signal_regions:
evaluator = Evaluator(resultCollectors[analysis][sr])
# Calculate everything that should be calculated
# TODO: Beware analyses with unknown background
evaluator.calc_efficiencies()
evaluator.calc_r_values()
if Info.flags["likelihood"]:
evaluator.calc_likelihood()
if Info.flags["fullcls"]:
evaluator.calc_cls_values()
if Info.flags["zsig"]:
evaluator.calc_zsig()
evaluators[analysis][sr] = evaluator
if Info.parameters["bestcls"] != 0:
AdvPrint.cout("Calculating CLs for the "+str(Info.parameters["bestcls"])+" most sensitive signal regions!")
best_evaluators = find_strongest_evaluators(evaluators, Info.parameters["bestcls"])
# if "bestcls" is 1, find_strongest_evaluators does not return a list but just the single best
if Info.parameters["bestcls"] == 1:
best_evaluators = [best_evaluators]
for ev in best_evaluators:
ev.calc_cls_values()
# find best result
best_evaluator_per_analysis = dict()
for analysis in evaluators:
# Find bes of all SRs in analysis
best_evaluator_per_analysis[analysis] = find_strongest_evaluators(evaluators[analysis], 1)
best_evaluator = find_strongest_evaluators(best_evaluator_per_analysis, 1)
AdvPrint.set_cout_file(Info.files['output_totalresults'], True)
AdvPrint.mute()
for col in Info.parameters["TotalEvaluationFileColumns"]:
AdvPrint.cout(col+" ", "nlb")
AdvPrint.cout("")
for a in sorted(evaluators.keys()):
for sr in sorted(evaluators[a].keys()):
AdvPrint.cout(evaluators[a][sr].line_from_data(Info.parameters["TotalEvaluationFileColumns"]))
AdvPrint.format_columnated_file(Info.files['output_totalresults'])
AdvPrint.set_cout_file(Info.files['output_bestsignalregions'], True)
AdvPrint.mute()
for col in Info.parameters["BestPerAnalysisEvaluationFileColumns"]:
AdvPrint.cout(col+" ", "nlb")
AdvPrint.cout("")
# print analyses in alphabetic order
for a in sorted(best_evaluator_per_analysis.keys()):
AdvPrint.cout(best_evaluator_per_analysis[a].line_from_data(Info.parameters["BestPerAnalysisEvaluationFileColumns"]))
AdvPrint.format_columnated_file(Info.files['output_bestsignalregions'])
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
best_evaluator.check_warnings()
best_evaluator.print_result()
if Info.flags['zsig']:
_print_zsig(evaluators)
if Info.flags['likelihood']:
_print_likelihood(evaluators)
def runFritz(self):
self.prepareFritz()
""" Runs Fritz """
from events import MG5Events
for event in self.eventsList:
if event.processed:
continue
fritz_command = Info.files["fritz_bin"] + " " + event.configFile
result = subprocess.Popen(fritz_command,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
maxlen = 0
try:
for line in iter(result.stdout.readline, b''):
# Print to logfile. If it does not start with the Fritz::Global prefix, it comes from MG5 and should be redirected
AdvPrint.mute()
if not line.startswith("|~| ") and isinstance(
event, MG5Events):
AdvPrint.set_cout_file(
os.path.join(
Info.paths["output_mg5"],
"mg5amcatnlo_" + event.identifier + ".log"))
elif "PYTHIA Rndm::dumpState" in line:
AdvPrint.set_cout_file(
os.path.join(Info.paths["output_pythia"],
"pythia_" + event.identifier +
".log"))
else:
line = line.replace("|~| ", "")
AdvPrint.set_cout_file(
os.path.join(Info.paths["output_fritz"],
"fritz_" + event.identifier + ".log"))
AdvPrint.cout(line.rstrip())
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
# We should not exceed the terminal terminal_width:
terminal_width = AdvPrint.get_terminal_width()
print_line = " |-> " + str(line.strip())
print_line.replace("\t", " ")
while "\r" in print_line:
print_line = print_line[print_line.index("\r"):]
len_of_print_line = len(print_line)
maxlen = max(maxlen, len(print_line))
# As we print line by line in the same terminal row, we have to add spaces if the curr line is shorter than a line before
fill_spaces = ""
if len(print_line) < maxlen:
fill_spaces = " " * (maxlen - len(print_line))
# if line is too long, make it shorter by appropriate amoung
if len(print_line + fill_spaces) >= terminal_width and len(
print_line) <= terminal_width:
fill_spaces = " " * (terminal_width - len(print_line) -
1)
elif len(print_line) > terminal_width:
fill_spaces = ""
print_line = print_line[:terminal_width - 4] + "..."
AdvPrint.cout(
"\r" + print_line + fill_spaces + "\x1b[0m\r", "nlb")
except KeyboardInterrupt:
AdvPrint.cout("Caught Keyboard Signal. Aborting Fritz")
result.send_signal(signal.SIGTERM)
for line in iter(result.stderr.readline, b''):
AdvPrint.unmute()
AdvPrint.set_cout_file(Info.files['fritz_log'])
# remove nasty ROOT6-CLING warnings from on-screen output
if "cling::AutoloadingVisitor::InsertIntoAutoloadingState:" in line:
AdvPrint.mute()
elif "Missing FileEntry for ExRootAnalysis" in line:
AdvPrint.mute()
elif "requested to autoload type" in line:
AdvPrint.mute()
AdvPrint.cout(line.rstrip() + "")
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
AdvPrint.cout("")
# Abort if there was an error
result.wait()
if result.returncode != 0:
AdvPrint.cerr_exit(
"Fritz returned with error. Check logfiles in result folder for more information!"
)
# Remove all empty analysisstdout files
for f in [
x for x in os.listdir(Info.paths['output_analysis'])
if x.startswith("analysisstdout")
]:
if os.stat(os.path.join(Info.paths['output_analysis'],
f)).st_size == 0:
os.remove(os.path.join(Info.paths['output_analysis'], f))
# Associate result files to event
for a in Info.analyses:
event.analysis_signal_files[a] = os.path.join(
Info.paths['output_analysis'],
event.identifier + '_' + a + '_signal.dat')
if os.path.isfile(event.analysis_signal_files[a]):
AdvPrint.format_columnated_file(
event.analysis_signal_files[a])
event.analysis_cutflow_files[a] = os.path.join(
Info.paths['output_analysis'],
event.identifier + '_' + a + '_cutflow.dat')
if os.path.isfile(event.analysis_cutflow_files[a]):
AdvPrint.format_columnated_file(
event.analysis_cutflow_files[a])
# finish
event.processed = True
def get_resultCollectors(self):
resultCollectors = dict() # list of all collectors of all analyses and all signal regions
resultCollector = ResultCollector(self.identifier, "", "") # base collector object which we will just edit and copy
for analysis in Info.analyses:
# check if results file exists
if not os.path.isfile(self.analysis_signal_files[analysis]):
AdvPrint.cerr_exit("\t events::get_resultCollector() \n"
"\t Required analysis result file does not exist: \n "
"\t\t"+self.analysis_signal_files[analysis]+"\n"
"\t It is very likely that something went wrong in the delphes and/or the analysis step. \n"
"\t Please check \n "
"\t \t "+Info.files['delphes_log']+" \n "
"\t \t "+Info.files['analysis_log']+"* \n "
"\t for error messages and, should you not be able to fix them yourself, contact the authors under \n"
"\t \t [email protected]")
# setup resultCollector object
resultCollector.analysis = analysis
resultCollectors[analysis] = dict()
signal_regions = Info.get_analysis_parameters(analysis)["signal_regions"]
# Read result file
f = open(self.analysis_signal_files[analysis], "r")
for line in f:
# Ignore empty or commented lines
line = line.rstrip()
if line == "" or line[0] == "#":
continue
# Read file:
line = AdvPrint.remove_extra_spaces(line)
tokens = [t for t in line.split(" ") if t != ""]
# First, read information on total events number
if tokens[0] == "MCEvents:":
resultCollector.total_mcevents = float(tokens[1])
elif tokens[0] == " SumOfWeights:":
resultCollector.total_sumofweights = float(tokens[1])
elif tokens[0] == " SumOfWeights2:":
resultCollector.total_sumofweights2 = float(tokens[1])
elif tokens[0] == " NormEvents:":
resultCollector.total_normevents = float(tokens[1])
elif tokens[0] == "XSect:":
xsect = float(tokens[1].split(" ")[0])
elif tokens[0] == " Error:":
xsecterr = float(tokens[1].split(" ")[0])
else:
# SR Sum_W Sum_W2 Acc N_Norm
for sr in signal_regions:
if tokens[0].startswith(sr):
resultCollector.sr = sr
# Read number of events
resultCollector.signal_sumofweights = float(tokens[1])
resultCollector.signal_sumofweights2 = float(tokens[2])
resultCollector.signal_normevents = float(tokens[4])
# Calculate errors
if resultCollector.signal_sumofweights > 0:
resultCollector.signal_err_stat = resultCollector.signal_normevents*sqrt(resultCollector.signal_sumofweights2)/resultCollector.signal_sumofweights
resultCollector.signal_err_sys = resultCollector.signal_normevents*xsecterr/xsect
resultCollector.signal_err_tot = sqrt(resultCollector.signal_err_stat**2+resultCollector.signal_err_sys**2)
else:
resultCollector.signal_err_stat = 0
resultCollector.signal_err_sys = 0
resultCollector.signal_err_tot = 0
# put copy of resultCollector in collector dict
resultCollectors[analysis][sr] = deepcopy(resultCollector)
f.close()
# Write events file, if wanted
if Info.parameters["EventResultFileColumns"] != []:
AdvPrint.mute()
AdvPrint.set_cout_file(self.result_output_file, True)
for col in Info.parameters["EventResultFileColumns"]:
AdvPrint.cout(col+" ", "nlb")
AdvPrint.cout("")
for a in sorted(resultCollectors.keys()):
for sr in sorted(resultCollectors[a].keys()):
AdvPrint.cout(resultCollectors[a][sr].line_from_data(Info.parameters["EventResultFileColumns"]))
AdvPrint.format_columnated_file(self.result_output_file)
AdvPrint.set_cout_file("#None")
AdvPrint.unmute()
return resultCollectors