def runAsymptoticLimit(mass, cardname, year):
if '-' in year:
split_year = year.split('-')
else:
split_year = [year]
for y in split_year:
datacard = getDataCard(mass, cardname, y)
combineSets(mass, y)
combineYears(mass, cardname)
datacard = getDataCard(mass, cardname, year)
output_folder = datacard.replace('dataCards', 'output').rsplit('/', 1)[0] +'/asymptotic/'+cardname
print 'Running Combine for mass', str(mass), 'GeV'
if args.blind:
runCombineCommand('combine -M AsymptoticLimits '+datacard+ ' --run blind')
else:
runCombineCommand('combine -M AsymptoticLimits '+datacard)
makeDirIfNeeded(output_folder+'/x')
os.system('scp '+os.path.expandvars('$CMSSW_BASE/src/HNL/Stat/data/output/tmp/*root')+ ' ' +output_folder+'/.')
os.system('rm '+os.path.expandvars('$CMSSW_BASE/src/HNL/Stat/data/output/tmp/*root'))
print 'Finished running asymptotic limits for mass', str(mass), 'GeV'
return
def printSelections(in_file_path, out_file_path):
in_file = TFile(in_file_path)
key_names = [k[0] for k in rootFileContent(in_file, '/', starting_dir = 'cutflow')]
makeDirIfNeeded(out_file_path)
out_file = open(out_file_path, 'w')
for k in key_names:
if 'Total' in k: continue
out_file.write(k.split('/')[-1] + '\n')
return
def getOutputName(region):
output_name = os.path.join(getOutputBase(region), sample.output)
if args.isChild:
output_name += '/tmp_' + sample.output
output_name += '/' + sample.name + '_events_' + subjobAppendix + '.root'
makeDirIfNeeded(output_name)
return output_name
def combineYears(mass, cardname):
if len(args.year) == 1:
return
else:
datacard_massbase = os.path.join(datacards_base('-'.join(args.year)), 'HNL-'+args.flavor+'-m'+str(mass), 'shapes')
datacard_path = lambda year : os.path.join(datacards_base(year), 'HNL-'+args.flavor+'-m'+str(mass), 'shapes', cardname+'.txt')
categories = [datacard_path(y) for y in args.year if os.path.isfile(datacard_path(year))]
makeDirIfNeeded(datacard_massbase+'/'+cardname+'.txt')
runCombineCommand('combineCards.py '+' '.join(categories)+' > '+datacard_massbase+'/'+cardname+'.txt')
def write(self, path, append=False):
append_string = 'recreate'
if append and isValidRootFile(path): append_string = 'update'
makeDirIfNeeded(path)
output_file = ROOT.TFile(path, append_string)
output_file.mkdir(self.name)
output_file.cd(self.name)
self.hist.Write()
output_file.Close()
def mergeSmallFile(merge_file):
path, name = merge_file.rsplit('/', 1)
if not args.isTest:
os.system('hadd -f -v ' + path + '/' + name.split('_', 1)[1] +
'.root ' + merge_file + '/*root')
os.system('rm -r -f ' + merge_file)
else:
makeDirIfNeeded(path + '/testArea/' + name.split('_', 1)[1] + '.root')
os.system('hadd -f -v ' + path + '/testArea/' + name.split('_', 1)[1] +
'.root ' + merge_file + '/*root')
def getOutputName():
output_name = getOutputBase()
output_name += '/' + sample.output
if args.isChild:
output_name += '/tmp_' + sample.output
output_name += '/' + sample.name + '_events_' + subjobAppendix + '.root'
makeDirIfNeeded(output_name)
return output_name
def write(self, append=False):
append_string = 'recreate'
if append and isValidRootFile(self.path): append_string = 'update'
makeDirIfNeeded(self.path)
output_file = TFile(self.path, append_string)
output_file.mkdir(self.name)
output_file.cd(self.name)
self.eff_numerator.getHist().Write()
self.eff_denominator.getHist().Write()
self.misid_numerator.getHist().Write()
self.misid_denominator.getHist().Write()
output_file.Close()
def makeRunScript(script, arguments, i, condor_base):
original_script = os.path.expandvars(
os.path.join('$CMSSW_BASE', 'src', 'HNL', 'Tools', 'scripts',
'runOnCondor.sh'))
new_script_name = os.path.realpath(
os.path.join(condor_base, 'runscripts',
str(i) + '.sh'))
makeDirIfNeeded(new_script_name)
new_script = open(new_script_name, 'w')
for line in open(original_script, 'r'):
new_line = line.replace('$1', arguments[0])
new_line = new_line.replace('$2', arguments[1])
new_script.write(new_line)
new_script.close()
return new_script_name
def getOutputName(prompt_str):
if not args.isTest:
output_name = os.path.join(os.getcwd(), 'data',
__file__.split('.')[0], args.selection,
args.region, sample.output, prompt_str)
else:
output_name = os.path.join(os.getcwd(), 'data', 'testArea',
__file__.split('.')[0], args.selection,
args.region, sample.output, prompt_str)
if args.isChild:
output_name += '/tmp_' + sample.output
output_name += '/' + sample.name + '_events_' + subjobAppendix + '.root'
makeDirIfNeeded(output_name)
return output_name
def mergeLargeFile(merge_file):
path, name = merge_file.rsplit('/', 1)
sub_files = sorted(glob.glob(merge_file + '/*'))
split_list = []
tmp_list = []
if args.isTest:
makeDirIfNeeded(
os.path.join(path, 'testArea', name, 'tmp_batches', 'x'))
else:
makeDirIfNeeded(os.path.join(path, name, 'tmp_batches', 'x'))
for i, file_list in enumerate(sub_files):
if i != 0 and (i % 20 == 0 or i == len(sub_files) - 1):
tmp_list.append(file_list)
split_list.append([x for x in tmp_list])
tmp_list = []
else:
tmp_list.append(file_list)
for j, file_list in enumerate(split_list):
if not args.isTest:
os.system('hadd -f -v ' + path + '/' + name +
'/tmp_batches/batch_' + str(j) + '.root ' +
' '.join(file_list))
for f in file_list:
os.system('rm ' + f)
else:
os.system('hadd -f -v ' + path + '/testArea/' + name +
'/tmp_batches/batch_' + str(j) + '.root ' +
' '.join(file_list))
if args.isTest:
os.system('hadd -f -v ' + path + '/testArea/' + name.split('_', 1)[1] +
'.root ' + path + '/testArea/' + name + '/tmp_batches/*root')
else:
os.system('hadd -f -v ' + path + '/' + name.split('_', 1)[1] +
'.root ' + path + '/' + name + '/tmp_batches/*root')
os.system('rm -rf ' + path + '/' + name)
def write(self, append=False, name=None, subdirs=None):
append_string = 'recreate'
if append and isValidRootFile(self.path): append_string = 'update'
makeDirIfNeeded(self.path)
output_file = ROOT.TFile(self.path, append_string)
if subdirs is None:
output_file.mkdir(self.name)
output_file.cd(self.name)
else:
nomo = ''
for d in subdirs:
nomo += d + '/'
output_file.mkdir(nomo)
output_file.cd(nomo)
if name is not None:
self.efficiency_num.getHist().Write(name + '_num')
self.efficiency_denom.getHist().Write(name + '_denom')
self.getEfficiency().Write(name + '_efficiency')
else:
self.efficiency_num.getHist().Write()
self.efficiency_denom.getHist().Write()
self.getEfficiency().Write()
output_file.Close()
def write(self, append = False, name=None, is_test=None):
append_string = 'recreate'
if append and isValidRootFile(self.path): append_string = 'update'
if is_test is None:
path_to_use = self.path
else:
split_path = self.path.split('/')
index_to_use = split_path.index('testArea')+1
path_to_use = os.path.expandvars("$HOME/Testing/Latest/"+'/'.join(split_path[index_to_use:-1])+'/'+is_test+'/'+split_path[-1])
makeDirIfNeeded(path_to_use)
output_file = ROOT.TFile(path_to_use, append_string)
if self.subdirs is None:
pass
# output_file.mkdir(self.name)
# output_file.cd(self.name)
else:
nomo = ''
for d in self.subdirs:
nomo += d + '/'
output_file.mkdir(nomo)
output_file.cd(nomo)
if name is not None:
self.efficiency_num.getHist().Write(name+'_num')
self.efficiency_denom.getHist().Write(name+'_denom')
# self.getEfficiency().Write(name+'_efficiency')
else:
self.efficiency_num.getHist().Write()
self.efficiency_denom.getHist().Write()
# self.getEfficiency().Write()
output_file.Close()
if is_test is not None:
self.write(append=append, name=name, is_test=None)
def savePlot(self, destination, message=None):
makeDirIfNeeded(destination)
destination_components = destination.split('/')
cleaned_components = [
x for x in destination_components if not isTimeStampFormat(x)
]
try:
index_for_php = cleaned_components.index('src')
except:
index_for_php = None
if index_for_php:
php_destination = '/user/lwezenbe/public_html/'
php_destination += '/'.join([
comp for comp in cleaned_components[index_for_php + 1:]
if (comp != 'data' and comp != 'Results')
])
makeDirIfNeeded(php_destination)
os.system('cp -rf $CMSSW_BASE/src/HNL/Tools/php/index.php ' +
php_destination.rsplit('/', 1)[0] + '/index.php')
cmssw_version = os.path.expandvars('$CMSSW_BASE').rsplit('/',
1)[-1]
central_destination = '/user/lwezenbe/private/Backup/' + cmssw_version + '/'
central_destination += '/'.join(
[comp for comp in destination_components[index_for_php + 1:]])
makeDirIfNeeded(central_destination)
self.canvas.SaveAs(destination + ".pdf")
self.canvas.SaveAs(destination + ".png")
self.canvas.SaveAs(destination + ".root")
#Clean out the php directory you want to write to if it is already filled, otherwise things go wrong with updating the file on the website
#os.system("rm "+php_destination.rsplit('/')[0]+"/*")
if index_for_php:
self.canvas.SaveAs(php_destination + ".pdf")
self.canvas.SaveAs(php_destination + ".png")
self.canvas.SaveAs(php_destination + ".root")
#
# Save to a central, local backup that you can return to also after you switched CMSSW
#
if index_for_php:
self.canvas.SaveAs(central_destination + ".pdf")
self.canvas.SaveAs(central_destination + ".png")
self.canvas.SaveAs(central_destination + ".root")
if message is not None:
pt.writeMessage(destination.rsplit('/', 1)[0], message)
def clearLogs(base):
last_base = base.replace('Latest', 'Previous')
makeDirIfNeeded(last_base + '/x')
os.system('rm -r ' + last_base + '/*')
os.system('scp -r ' + base + '/* ' + last_base + '/.')
os.system('rm -r ' + base)
cutter = Cutter(chain=chain)
#
# Create new reduced tree (except if it already exists and overwrite option is not used)
#
output_base = os.path.expandvars(
os.path.join('/user/$USER/public/ntuples/HNL')
) if not args.isTest else os.path.expandvars(
os.path.join('$CMSSW_BASE', 'src', 'HNL', 'TMVA', 'data', 'testArea'))
signal_str = 'Signal' if chain.is_signal else 'Background'
output_name = os.path.join(
output_base, 'TMVA', str(args.year),
args.region + '-' + args.selection, signal_str, 'tmp_' + sample.output,
sample.name + '_' + sample.output + '_' + str(args.subJob) + '.root')
makeDirIfNeeded(output_name)
# if not args.isTest and isValidRootFile(output_name):
# log.info('Finished: valid outputfile already exists')
# exit(0)
output_file = ROOT.TFile(output_name, "RECREATE")
#
# Switch off unused branches and create outputTree
#
output_tree = {}
for prompt_str in ['prompt', 'nonprompt']:
output_tree[prompt_str] = {}
for c in SUPER_CATEGORIES.keys():
output_tree[prompt_str][c] = ROOT.TTree('trainingtree',
var['HNLmass'] = (lambda c : c.HNLmass, np.array(mass_range), ('m_{N} [GeV]', 'Efficiency'))
from HNL.Tools.efficiency import Efficiency
from HNL.EventSelection.eventCategorization import returnCategoryTriggers
from HNL.Triggers.triggerSelection import applyCustomTriggers
eff = {}
for c in categories:
for v in {k for k in var.keys()}:
for t in category_map[c]:
if args.separateTriggers is None or args.separateTriggers == 'full':
name = str(c)+ '_' +v + '_' + t
if v != 'HNLmass':
eff[(c, v, t)] = Efficiency(name, var[v][0], var[v][2], getOutputName(v), var[v][1], subdirs=['efficiency_'+str(c), v, 'allTriggers'])
else:
eff[(c, v, t)] = Efficiency(name, var[v][0], var[v][2], getOutputName(v), var[v][1], subdirs=['efficiency_'+str(c), v, 'allTriggers'])
makeDirIfNeeded(getOutputName(v))
else:
for i, trigger in enumerate(category_triggers(chain, c)):
name = str(c)+ '_' +v + '_' + t +'_'+str(i)
if v != 'HNLmass':
eff[(c, v, t, i)] = Efficiency(name, var[v][0], var[v][2], getOutputName(v), var[v][1], subdirs=['efficiency_'+str(c), v, str(returnCategoryTriggerNames(c)[i])])
else:
eff[(c, v, t, i)] = Efficiency(name, var[v][0], var[v][2], getOutputName(v), var[v][1], subdirs=['efficiency_'+str(c), v, str(returnCategoryTriggerNames(c)[i])])
makeDirIfNeeded(getOutputName(v))
#
# Set event range
#
if args.isTest:
max_events = 20000
event_range = xrange(max_events) if max_events < len(sample.getEventRange(args.subJob)) else sample.getEventRange(args.subJob)
input_files = glob.glob(os.getcwd() + '/data/calcTriggerEff/*/' +
args.separateTriggers + '/*.root')
sample_names = {in_file_name.split('/')[-3] for in_file_name in input_files}
f_names = {f.split('/')[-1].split('.')[0] for f in input_files}
#Prepare output dir
timestamp = time.strftime("%Y%m%d_%H%M%S")
for f_name in f_names:
if args.isTest:
out_dir = os.getcwd(
) + '/data/testArea/Results/' + args.separateTriggers + '/' + f_name + '/' + timestamp
else:
out_dir = os.getcwd(
) + '/data/Results/' + args.separateTriggers + '/' + f_name + '/' + timestamp
makeDirIfNeeded(out_dir + '/x')
# def isUniqueElement(l, item):
# list_of_truths = [e == item for e in l]
# return not any(list_of_truths)
import ROOT
from HNL.Tools.helpers import rootFileContent, getObjFromFile
from HNL.Plotting.plot import Plot
from HNL.Plotting.plottingTools import extraTextFormat
from HNL.EventSelection.eventCategorization import CATEGORIES, CATEGORY_TEX_NAMES, returnCategoryTriggerNames, TRIGGER_CATEGORIES
from HNL.Tools.efficiency import Efficiency
# #TODO: Can this be made general?
# def makeNameCompList(name_list):
output_name += '/tmp_' + sample.output
print 'Getting things ready to start the event loop'
#Initialize histograms
from HNL.Tools.ROC import ROC
from HNL.Tools.efficiency import Efficiency
list_of_roc = []
list_of_var_hist = {'efficiency': {}, 'fakerate': {}}
for algo in algos:
algo_wp = algos[algo]
tot_name = output_name + '/' + sample.name + '_' + algo + '-ROC-' + str(
args.flavor) + '_' + args.subJob + '.root'
makeDirIfNeeded(tot_name)
list_of_roc.append(ROC(algo, tot_name, working_points=algo_wp))
#Efficiency histograms for variables
var_hist = {
'pt': [
lambda c: c.pt_l,
np.arange(0., 210., 10.),
('p_T^{#tau}(offline) [GeV]', 'Efficiency')
],
'eta': [
lambda c: c.eta_l,
np.arange(-2.5, 3., 0.5),
('#eta^{#tau}(offline) [GeV]', 'Efficiency')
]
}
else:
os.system(command + ' > ' +
out_file_name(signal).split('.root')[0] + '.txt')
exit(0)
#
# List the input variables here
#
from HNL.TMVA.mvaVariables import getVariableList
print 'Processing', args.signalname
if not args.skipTraining:
input_var = getVariableList(args.signalname)
makeDirIfNeeded(out_file_name(args.signalname))
out_file = ROOT.TFile(out_file_name(args.signalname), 'RECREATE')
factory = ROOT.TMVA.Factory("factory", out_file, "")
if args.writeInTree:
if args.signalname.split('_')[-1] in ['e', 'mu']:
name = 'NoTau/trainingtree'
else:
name = 'Total/trainingtree'
in_tree = ih.getTree(args.signalname, name=name).CloneTree()
in_tree.Write('backuptree')
loader = ih.getLoader(args.signalname, input_var)
preselection = ROOT.TCut("")
# options = ""
#Merges subfiles if needed
if args.batchSystem != 'HTCondor':
merge_files = glob.glob('/storage_mnt/storage/user/' +
os.path.expandvars('$USER') +
'/public/ntuples/HNL/' + skim_selection_string +
'/' + args.year + '/' + args.skimName + '/tmp*')
else:
merge_files = glob.glob(pnfs_base + '/tmp*')
merge_files = sorted(merge_files)
if args.batchSystem != 'HTCondor':
os.system('gfal-mkdir ' + pnfs_base)
os.system('gfal-mkdir ' + pnfs_backup_base)
else:
makeDirIfNeeded(pnfs_base)
makeDirIfNeeded(pnfs_backup_base)
for mf in merge_files:
if 'Data' in mf: continue
path, name = mf.rsplit('/', 1)
new_name = name.split('_', 1)[1] + '.root'
if not args.isTest:
if args.batchSystem != 'HTCondor':
os.system('gfal-rm ' + pnfs_backup_base + '/' + new_name)
os.system('gfal-copy ' + pnfs_base + '/' + new_name + ' ' +
pnfs_backup_base + '/' + new_name)
os.system('gfal-rm ' + pnfs_base + '/' + new_name)
else:
if isValidRootFile(pnfs_base + '/' + new_name):
def makeDataCard(bin_name,
flavor,
year,
obs_yield,
sig_name,
bkgr_names,
sig_yield=None,
bkgr_yields=None,
shapes=False,
coupling_sq=1e-4):
if not shapes and len(bkgr_yields) != len(bkgr_names):
raise RuntimeError(
"length of background yields and names is inconsistent")
if not shapes:
out_name = os.path.join(os.path.expandvars('$CMSSW_BASE'),
'src', 'HNL', 'Stat', 'data', 'dataCards',
str(year), flavor, sig_name, 'cutAndCount',
bin_name + '.txt')
else:
out_name = os.path.join(os.path.expandvars('$CMSSW_BASE'),
'src', 'HNL', 'Stat', 'data', 'dataCards',
str(year), flavor, sig_name, 'shapes',
bin_name + '.txt')
makeDirIfNeeded(out_name)
out_file = open(out_name, 'w')
out_file.write('# coupling squared = ' + str(coupling_sq) + ' \n \n')
out_file.write('imax 1 number of bins \n')
out_file.write('jmax * number of processes\n')
out_file.write('kmax * \n')
out_file.write('-' * 400 + '\n')
if shapes:
shapes_path = os.path.join(os.path.expandvars('$CMSSW_BASE'),
'src', 'HNL', 'Stat', 'data', 'shapes',
str(year), flavor, sig_name,
bin_name + '.shapes.root')
out_file.write('shapes * * \t' + shapes_path +
' $PROCESS $PROCESS_SYSTEMATIC')
out_file.write('-' * 400 + '\n')
out_file.write('bin ' + bin_name + ' \n')
if shapes:
out_file.write('observation -1 \n')
# out_file.write('observation '+str(obs_yield)+ ' \n')
else:
out_file.write('observation ' + str(obs_yield) + ' \n')
out_file.write('-' * 400 + '\n')
out_file.write(tab(['bin', ''] + [bin_name] * (len(bkgr_names) + 1)))
out_file.write(tab(['process', ''] + bkgr_names + [sig_name]))
out_file.write(
tab(['process', ''] + [str(i)
for i in xrange(1,
len(bkgr_names) + 1)] + ['0']))
if shapes:
out_file.write(tab(['rate', ''] + ['-1'] * (len(bkgr_names) + 1)))
else:
out_file.write(
tab(['rate', ''] +
['%4.6f' % by if by >= 0 else 0.0
for by in bkgr_yields] + ['%4.6f' % sig_yield]))
out_file.write('-' * 400 + '\n')
# For now no systematics, just lumi as example
out_file.write(tab(['lumi_13TeV', 'lnN'] + [1.025] *
(len(bkgr_names) + 1)))
#autoMCstats
if shapes:
out_file.write('* autoMCStats 3 1 1')
out_file.close()
import glob, os
from HNL.Tools.helpers import makeDirIfNeeded, progress
list_of_datafiles = [
'SingleMuon', 'SingleElectron', 'DoubleMuon', 'DoubleEG', 'MuonEG'
]
import argparse
argParser = argparse.ArgumentParser(description="Argument parser")
argParser.add_argument('--year', action='store', default='2016')
argParser.add_argument('--skim', action='store', default='Reco')
args = argParser.parse_args()
output_folder = '/storage_mnt/storage/user/lwezenbe/public/ntuples/HNL/OldAnalysis/' + args.year + '/' + args.skim + '/tmp_DataFiltered'
makeDirIfNeeded(output_folder)
event_information_set = set()
# for f_name in list_of_datafiles:
# file_list = glob.glob('/storage_mnt/storage/user/lwezenbe/public/ntuples/HNL/OldAnalysis/'+args.year+ '/' + args.skim +'/tmp_'+f_name+'/*.root')
# for i, sub_f_name in enumerate(file_list):
# print sub_f_name
# f = ROOT.TFile(sub_f_name)
# c = f.Get('blackJackAndHookers/blackJackAndHookersTree')
# try:
# c.GetEntry()
# except:
# continue
# output_file = ROOT.TFile(output_folder +'/'+ sub_f_name.split('/')[-1], 'recreate')
print 'Chain initialized'
#Set output dir
#Since number of subjobs was set to be 1 (HNL samples are small), this name was chosen since no overlap possible
#If this changes, this needs to be changed as well
from HNL.Tools.helpers import makeDirIfNeeded
if not args.isTest:
output_name = os.path.join(os.getcwd(), 'data',
os.path.basename(__file__).split('.')[0],
sample.output)
else:
output_name = os.path.join(os.getcwd(), 'data', 'testArea',
os.path.basename(__file__).split('.')[0],
sample.output)
output_name += '/' + sample.name + '.root'
makeDirIfNeeded(output_name)
print 'Getting things ready to start the event loop'
#Initialize histograms
pt_hist = [
ROOT.TH1D('leading_pt', sample.name, 100, 0, 100),
ROOT.TH1D('subleading_pt', sample.name, 100, 0, 100),
ROOT.TH1D('trailing_pt', sample.name, 100, 0, 100)
]
if args.plotSoftTau and 'tau' in sample.name:
pt_hist.append(ROOT.TH1D('softest_tau', sample.name, 100, 0, 100))
if args.plotSoftTau and 'tau' in sample.name:
pt_hist.append(ROOT.TH1D('subleading_tau', sample.name, 100, 0, 100))
#Determine if testrun so it doesn't need to calculate the number of events in the getEventRange
default='baseline',
type=str,
help='What region do you want to select for?',
choices=['baseline', 'lowMassSR', 'highMassSR'])
argParser.add_argument(
'--selection',
action='store',
default='default',
help='Select the type of selection for objects',
choices=['leptonMVAtop', 'AN2017014', 'default', 'Luka', 'TTT'])
args = argParser.parse_args()
ih = InputHandler(args.year, args.region, args.selection)
#use getTree and scramble background
for signal in ih.signal_names:
if signal.split('_')[-1] in ['e', 'mu']:
name = 'NoTau/trainingtree'
else:
name = 'Total/trainingtree'
signal_tree = ih.getTree(signal, name=name, signal_only=True).CloneTree()
bkgr_tree = ih.getTree(signal, name=name, bkgr_only=True).CloneTree()
fname = '/storage_mnt/storage/user/lwezenbe/public/ntuples/HNL/TMVA/' + str(
args.year
) + '/' + args.region + '-' + args.selection + '/ForExternalTraining/' + signal + '.root'
makeDirIfNeeded(fname)
out_file = ROOT.TFile(fname, 'recreate')
signal_tree.Write('signaltree')
bkgr_tree.Write('backgroundtree')
out_file.Close()
if __name__ == "__main__":
iso_wp = ['vvvloose', 'vvloose', 'vloose', 'loose', 'medium', 'tight', 'vtight', 'vvtight']
ele_wp = ['vvloose', 'vloose', 'loose', 'medium', 'tight', 'vtight', 'vvtight']
mu_wp = ['vloose', 'loose', 'medium', 'tight']
import numpy as np
import os
from HNL.Tools.helpers import makeDirIfNeeded
from HNL.Tools.logger import getLogger, closeLogger
log = getLogger('INFO')
pt_range = np.arange(30, 70, 10)
out_name_tex = os.path.join(os.path.expandvars('$CMSSW_BASE'), 'src', 'HNL', 'Weights', 'data', 'Results', __file__.split('.')[0], 'deeptauVSjets.txt')
makeDirIfNeeded(out_name_tex)
# out_file_tex = open(out_name_tex, 'w')
# out_file_tex.write('\\begin{table}[] \n')
column_str = '|'.join(['c' for _ in pt_range])
# out_file_tex.write("\\begin{tabular}{|c|"+column_str+"|} \n")
# out_file_tex.write("\hline \n")
# out_file_tex.write('& '+'&'.join(['$p_T = $'+str(m)+ ' GeV' for m in pt_range])+' \\\\ \n')
# out_file_tex.write("\hline \n")
for iso in iso_wp:
tausftool = TauSF(2016, 'deeptauVSjets', iso, ele_wp[0], mu_wp[0])
# out_file_tex.write(iso)
for pt in pt_range:
sf = tausftool.sftool_iso.getSFvsPT(pt)
up = tausftool.sftool_iso.getSFvsPT(pt, unc='Up') - tausftool.sftool_iso.getSFvsPT(pt)
down = tausftool.sftool_iso.getSFvsPT(pt) - tausftool.sftool_iso.getSFvsPT(pt, unc='Down')
#
# Save histograms
#
if args.isTest: continue
subjobAppendix = '_subJob' + args.subJob if args.subJob else ''
output_name = os.path.join(os.getcwd(), 'data', 'plotTau',
reco_or_gen_str, sample.output)
if args.isChild:
output_name += '/tmp_' + sample.output + '/' + sample.name + '_'
else:
output_name += '/'
makeDirIfNeeded(output_name + 'variables' + subjobAppendix + '.root')
output_file = TFile(
output_name + 'variables' + subjobAppendix + '.root', 'RECREATE')
for v in var.keys():
output_file.mkdir(v)
output_file.cd(v)
list_of_hist[sample.output][v].getHist().Write()
output_file.Close()
#If the option to not run over the events again is made, load in the already created histograms here
else:
import glob
hist_list = glob.glob(os.getcwd() + '/data/plotTau/' + reco_or_gen_str +
import os
skipDirs = ['documents', 'log', '.git', 'virenv', 'utils']
skipFiles = []
extensions_to_copy = ['root', 'pdf', 'png']
file_paths = []
base_path = os.path.join(os.path.expandvars('$CMSSW_BASE'), 'src', 'HNL')
for root, dirs, files in os.walk(base_path, topdown=True):
dirs[:] = [d for d in dirs if d not in skipDirs]
root_files = [
os.path.join(root, f) for f in files
if f.split('.')[-1] in extensions_to_copy and f not in skipFiles
]
if len(root_files) != 0:
for ext in extensions_to_copy:
file_paths.append(os.path.join(root, '*.' + ext))
cmssw_version = os.path.expandvars('$CMSSW_BASE').rsplit('/', 1)[-1]
central_destination = makePathTimeStamped('/user/lwezenbe/private/Backup/' +
cmssw_version + '/AllOutput')
for rf in file_paths:
try:
index_for_backup = rf.split('/').index('src')
except:
index_for_backup = None
backup_path = '/'.join([central_destination] +
rf.split('/')[index_for_backup + 1:-1])
makeDirIfNeeded(backup_path + '/x')
os.system('scp ' + rf + ' ' + backup_path + '/.')
default='HTCondor',
help='choose batchsystem',
choices=['local', 'HTCondor', 'Cream02'])
args = argParser.parse_args()
if args.batchSystem == 'Cream02':
input_folder = '/storage_mnt/storage/user/lwezenbe/public/ntuples/HNL/' + args.skimSelection + '/' + args.year + '/' + args.skim + '/tmp_Data/'
output_folder = '/storage_mnt/storage/user/lwezenbe/public/ntuples/HNL/' + args.skimSelection + '/' + args.year + '/' + args.skim + '/tmp_DataFiltered'
else:
pnfs_base = os.path.join('/pnfs/iihe/cms/store/user',
os.path.expandvars('$USER'), 'skimmedTuples/HNL',
args.skimSelection, args.year, args.skim)
input_folder = pnfs_base + '/tmp_Data'
output_folder = pnfs_base + '/tmp_DataFiltered'
makeDirIfNeeded(output_folder + '/x')
event_information_set = set()
file_list = glob.glob(input_folder + '/*.root')
for i, sub_f_name in enumerate(file_list):
progress(i, len(file_list))
f = ROOT.TFile(sub_f_name)
c = f.Get('blackJackAndHookers/blackJackAndHookersTree')
try:
c.GetEntry()
except:
continue
output_file = ROOT.TFile(output_folder + '/' + sub_f_name.split('/')[-1],
shape_hist[sample_name] = ROOT.TH1D(
sample_name, sample_name, n_search_regions, 0.5,
n_search_regions + 0.5)
for sr in xrange(1, n_search_regions + 1):
shape_hist[sample_name].SetBinContent(
sr, list_of_values['bkgr'][sample_name][ac][sr])
shape_hist[sample_name].SetBinError(
sr, list_of_errors['bkgr'][sample_name][ac][sr])
for sample_name in list_of_values['signal'].keys():
out_path = os.path.join(os.path.expandvars('$CMSSW_BASE'),
'src', 'HNL',
'Stat', 'data', 'shapes',
str(args.year), args.flavor,
sample_name, ac + '.shapes.root')
makeDirIfNeeded(out_path)
out_shape_file = ROOT.TFile(out_path, 'recreate')
n_search_regions = srm[
args.region].getNumberOfSearchRegions()
shape_hist[sample_name] = ROOT.TH1D(
sample_name, sample_name, n_search_regions, 0.5,
n_search_regions + 0.5)
for sr in xrange(1, n_search_regions + 1):
shape_hist[sample_name].SetBinContent(
sr, list_of_values['signal'][sample_name][ac][sr])
shape_hist[sample_name].SetBinError(
sr, list_of_errors['signal'][sample_name][ac][sr])
shape_hist[sample_name].Write(sample_name)
bkgr_names = []
# for bkgr_sample_name in list_of_values['bkgr'].keys()+['data_obs']:
for bkgr_sample_name in background_collection + [
