def haddFastsimFiles(fileLists, smsName, inDir, outDir, OneDScan=False, dryRun=True):
"""
fileLists: dictionary in which keys are (mGluino, mLSP) pairs
and values are lists of ntuple files for the corresponding mass point
smsName: name of signal model
outDir: directory where output files should go
Runs hadd on the files for each signal mass point and puts the output in
the specified directory
"""
for pair in fileLists:
print "Signal:",smsName,pair
if not OneDScan:
if (os.path.isfile(outDir+'/'+smsName+'_'+pair[0]+'_'+pair[1]+'.root')):
print "output file exists already. skip"
else:
exec_me('hadd -f '+outDir+'/'+smsName+'_'+pair[0]+'_'+pair[1]+'.root '
+' '.join([inDir+'/'+f+' ' for f in fileLists[pair]]), dryRun)
else:
if (os.path.isfile(outDir+'/'+smsName+'_'+pair+'.root')):
print "output file exists already. skip"
else:
exec_me('hadd -f '+outDir+'/'+smsName+'_'+pair+'.root '
+' '.join([inDir+'/'+f+' ' for f in fileLists[pair]]), dryRun)
signalFilename = dirToUse + '/' + modelName + '.root'
#to modify branching ratios in T1ttbb sample
brString = ""
if 'T1x' in args.model:
xBR = float(args.model[args.model.find('x') +
1:args.model.find('y')].replace('p', '.'))
yBR = float(args.model[args.model.find('y') + 1:].replace(
'p', '.'))
brString = '--xBR %.2f --yBR %.2f' % (xBR, yBR)
signalFilename = dirToUse + '/SMS-T1ttbb_' + str(
args.mGluino) + '_' + str(args.mLSP) + '.root'
exec_me(
'python python/SMSTemplates.py --merge-bins -c %s -d %s --lumi %d --box %s %s %s %s --max-events %d'
% (config, outDir, LUMI, curBox,
((args.noSys) * ('--no-signal-sys')), signalFilename, brString,
args.maxSignalEvents), False)
#load SMS template histograms
signalHistFilename = '%s/%s_lumi-%.3f_0-3btag_%s.root' % (
outDir, modelName, LUMI * 1.0 / 1000, curBox)
signalHists = macro.importHists(signalHistFilename)
#update with correct names
for x, h in signalHists.items():
h.SetName(h.GetName().replace(curBox + '_' + args.model,
modelName))
signalHists[h.GetName()] = signalHists.pop(x)
#apply reduced efficiency method to correct for the presence of signal in the control regions
if args.reducedEff:
macro.doDeltaBForReducedEfficiencyMethod(backgroundHists,
if __name__ == '__main__':
boxes = ['MultiJet']
fits = ['Sideband']
weights = ['unweighted']
configs = ['config/run2_20151108_Preapproval_2b3b.config']
dryRun = False
lumi = 17000
btag = '0-3btag'
dateString = str(datetime.date.today()).replace("-","_")
dateString = '2015_11_17'
outDir = "fits_%s"%dateString
exec_me('mkdir -p %s'%outDir,dryRun)
for box in boxes:
for cfg in configs:
backgroundDsName = 'Datasets/RazorInclusive_SMCocktail_weighted_lumi-%.3f_%s_%s.root'%(lumi/1000.,btag,box)
#exec_me('python python/DustinTuple2RooDataSet.py -c %s -b %s -d Datasets/ Backgrounds/*.root -l %f -w'%(cfg,box,lumi),dryRun)
if 'unweighted' in weights:
backgroundDsName = 'Datasets/RazorInclusive_SMCocktail_unweighted_lumi-%.3f_%s_%s.root'%(lumi/1000.,btag,box)
#exec_me('python python/RooDataSet2UnweightedDataset.py -c %s -b %s -d Datasets/ Datasets/RazorInclusive_SMCocktail_weighted_lumi-%.3f_%s_%s.root'%(cfg,box,lumi/1000.,btag,box),dryRun)
for weight in weights:
outDir = "fits_%s/%s_%iifb/"%(dateString,weight,lumi/1000.)
exec_me('mkdir -p %s'%outDir,dryRun)
outDir = "fits_%s/%s_%iifb/%s"%(dateString,weight,lumi/1000.,box)
exec_me('mkdir -p %s'%outDir,dryRun)
exec_me('hadd -f '+outDir+'/'+smsName+'_'+pair[0]+'_'+pair[1]+'.root '
+' '.join([inDir+'/'+f+' ' for f in fileLists[pair]]), dryRun)
else:
if (os.path.isfile(outDir+'/'+smsName+'_'+pair+'.root')):
print "output file exists already. skip"
else:
exec_me('hadd -f '+outDir+'/'+smsName+'_'+pair+'.root '
+' '.join([inDir+'/'+f+' ' for f in fileLists[pair]]), dryRun)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument("inDir", help="input path")
parser.add_argument("-s", "--smsName", required=True, help="SMS name, e.g. T1bbbb")
parser.add_argument("--dryRun", action="store_true")
parser.add_argument("--OneDScan", action="store_true")
args = parser.parse_args()
inDir=args.inDir
smsName=args.smsName
dryRun=args.dryRun
OneDScan = args.OneDScan
fileLists = makeFileLists(inDir, smsName, OneDScan)
#output directory
outDir = inDir+'/combined'
exec_me('mkdir '+outDir, dryRun)
haddFastsimFiles(fileLists, smsName, inDir, outDir, OneDScan, dryRun=dryRun)
boxes = ['TTBarSingleLepton','WSingleLepton']
#fits = ['Sideband']
#fits = ['Full']
fits = ['Sideband','Full']
configs = ['config/controlsample.config']
lumi = 23.8
dryRun=False
for box in boxes:
for cfg in configs:
exec_me('python python/CRTuple2RooDataSet.py -c %s -b %s -d ControlSampleFits/ ControlSampleFits/SingleMuonAndElectron_Run2015B-GOLDEN.root -l %f'%(cfg,box,lumi),dryRun)
#exec_me('python python/CRTuple2RooDataSet.py -c %s -b %s -d ControlSampleFits/ ControlSampleFits/RunTwoRazorControlRegions_OneLeptonFull_Run2015B_GoodLumiDCS_NoDuplicates.root -l %f'%(cfg,box,lumi),dryRun)
for fit in fits:
if box in ['TTBarSingleLepton']:
btag = '1btag'
else:
btag = '0btag'
lumiString = '%.4f'%(lumi/1000)
#lumiString = lumiString.replace('.','p')
fitString = ''
if fit=='Sideband':
fitString = '--fit-region LowMR,LowRsq'
dirToUse = NOPILEUPWEIGHTS_SIGNAL_DIR
if 'T1' in args.model or 'T5' in args.model:
modelName = 'SMS-'+args.model+'_'+str(args.mGluino)+'_'+str(args.mLSP)
else:
modelName = 'SMS-'+args.model+'_'+str(args.mStop)+'_'+str(args.mLSP)
signalFilename=dirToUse+'/'+modelName+'.root'
#to modify branching ratios in T1ttbb sample
brString = ""
if 'T1x' in args.model:
xBR = float(args.model[args.model.find('x')+1:args.model.find('y')].replace('p','.'))
yBR = float(args.model[args.model.find('y')+1:].replace('p','.'))
brString = '--xBR %.2f --yBR %.2f'%(xBR,yBR)
signalFilename = dirToUse+'/SMS-T1ttbb_'+str(args.mGluino)+'_'+str(args.mLSP)+'.root'
exec_me('python python/SMSTemplates.py --merge-bins -c %s -d %s --lumi %d --box %s %s %s %s --max-events %d'%(config, outDir, LUMI, curBox, ((args.noSys)*('--no-signal-sys')), signalFilename, brString, args.maxSignalEvents), False)
#load SMS template histograms
signalHistFilename = '%s/%s_lumi-%.3f_0-3btag_%s.root'%(outDir,modelName,LUMI*1.0/1000,curBox)
signalHists = macro.importHists(signalHistFilename)
#update with correct names
for x,h in signalHists.items():
h.SetName(h.GetName().replace(curBox+'_'+args.model,modelName))
signalHists[h.GetName()] = signalHists.pop(x)
#apply reduced efficiency method to correct for the presence of signal in the control regions
if args.reducedEff:
macro.doDeltaBForReducedEfficiencyMethod(backgroundHists, signalHists, contamHists, sfHists, unrollBins=unrollBins, debugLevel=debugLevel)
#combine signal and background dictionaries
hists = backgroundHists.copy()
hists.update(signalHists)
args.mStop) + '_' + str(args.mLSP)
signalFilename = dirToUse + '/' + modelName + '.root'
brString = ""
if 'T1x' in args.model:
xBR = float(args.model[args.model.find('x') +
1:args.model.find('y')].replace('p', '.'))
yBR = float(args.model[args.model.find('y') + 1:].replace(
'p', '.'))
brString = '--xBR %.2f --yBR %.2f' % (xBR, yBR)
signalFilename = dirToUse + '/SMS-T1ttbb_' + str(
args.mGluino) + '_' + str(args.mLSP) + '.root'
signalHistFilename = '%s/%s_lumi-%.3f_0-3btag_%s.root' % (
outDir, modelName, LUMI * 1.0 / 1000, curBox)
if not glob.glob(signalHistFilename):
exec_me(
'python python/SMSTemplates.py --merge-bins -c %s -d %s --lumi %d --box %s %s %s --no-signal-sys'
% (config, outDir, LUMI, curBox, signalFilename, brString),
False)
#load SMS template histograms
signalHists = macro.importHists(signalHistFilename)
#update with correct names
for x, h in signalHists.items():
h.SetName(h.GetName().replace(curBox + '_' + args.model,
modelName))
signalHists[h.GetName()] = signalHists.pop(x)
signalHist = signalHists[modelName]
#make combined unrolled histograms for background
makeRazorBinEvidencePlots(curBox,
samples=samples,
inDir=BACKGROUND_DIR,
outDir=outDir,
int(splitF[-1])
mGluino = splitF[-1]
except ValueError:
print "Cannot parse gluino mass from", f, ": skipping"
continue
if mGluino not in fileLists:
fileLists[mGluino] = []
#add this file to appropriate list
fileLists[mGluino].append(f)
print "Adding", f, "to list of files for model point", mGluino
#output directory
outDir = inDir + '/combined'
exec_me('mkdir ' + outDir, dryRun)
#hadd the files for each signal mass point
for pair in fileLists:
print "Signal:", smsName, pair
if not OneDScan:
if (os.path.isfile(outDir + '/SMS-' + smsName + '_' + pair[0] +
'_' + pair[1] + '.root')):
print "output file exists already. skip"
else:
exec_me(
'hadd -f ' + outDir + '/SMS-' + smsName + '_' + pair[0] +
'_' + pair[1] + '.root ' +
' '.join([inDir + '/' + f + ' '
for f in fileLists[pair]]), dryRun)
else:
mGluino = splitF[-2]
except ValueError:
print "Cannot parse gluino mass from",f,": skipping"
continue
try:
int(splitF[-1])
mLSP = splitF[-1]
except ValueError:
print "Cannot parse LSP mass from",f,": skipping"
continue
pair = (mGluino, mLSP)
#add to dictionary if not present
if pair not in fileLists:
fileLists[pair] = []
#add this file to appropriate list
fileLists[pair].append(f)
print "Adding",f,"to list of files for model point",pair
#output directory
outDir = inDir+'/combined'
exec_me('mkdir '+outDir, dryRun)
#hadd the files for each signal mass point
for pair in fileLists:
print "Signal:",smsName,pair
exec_me('hadd -f '+outDir+'/SMS-'+smsName+'_'+pair[0]+'_'+pair[1]+'.root '+' '.join([inDir+'/'+f+' ' for f in fileLists[pair]]), dryRun)
boxes = ['TTBarSingleLepton', 'WSingleLepton']
#fits = ['Sideband']
#fits = ['Full']
fits = ['Sideband', 'Full']
configs = ['config/controlsample.config']
lumi = 23.8
dryRun = False
for box in boxes:
for cfg in configs:
exec_me(
'python python/CRTuple2RooDataSet.py -c %s -b %s -d ControlSampleFits/ ControlSampleFits/SingleMuonAndElectron_Run2015B-GOLDEN.root -l %f'
% (cfg, box, lumi), dryRun)
#exec_me('python python/CRTuple2RooDataSet.py -c %s -b %s -d ControlSampleFits/ ControlSampleFits/RunTwoRazorControlRegions_OneLeptonFull_Run2015B_GoodLumiDCS_NoDuplicates.root -l %f'%(cfg,box,lumi),dryRun)
for fit in fits:
if box in ['TTBarSingleLepton']:
btag = '1btag'
else:
btag = '0btag'
lumiString = '%.4f' % (lumi / 1000)
#lumiString = lumiString.replace('.','p')
fitString = ''
if fit == 'Sideband':
fitString = '--fit-region LowMR,LowRsq'
#call SMS template maker
if 'T1' in args.model or 'T5' in args.model:
modelName = 'SMS-'+args.model+'_'+str(args.mGluino)+'_'+str(args.mLSP)
else:
modelName = 'SMS-'+args.model+'_'+str(args.mStop)+'_'+str(args.mLSP)
signalFilename=dirToUse+'/'+modelName+'.root'
brString = ""
if 'T1x' in args.model:
xBR = float(args.model[args.model.find('x')+1:args.model.find('y')].replace('p','.'))
yBR = float(args.model[args.model.find('y')+1:].replace('p','.'))
brString = '--xBR %.2f --yBR %.2f'%(xBR,yBR)
signalFilename = dirToUse+'/SMS-T1ttbb_'+str(args.mGluino)+'_'+str(args.mLSP)+'.root'
signalHistFilename = '%s/%s_lumi-%.3f_0-3btag_%s.root'%(outDir,modelName,LUMI*1.0/1000,curBox)
if not glob.glob(signalHistFilename):
exec_me('python python/SMSTemplates.py --merge-bins -c %s -d %s --lumi %d --box %s %s %s --no-signal-sys'%(config, outDir, LUMI, curBox, signalFilename, brString), False)
#load SMS template histograms
signalHists = macro.importHists(signalHistFilename)
#update with correct names
for x,h in signalHists.items():
h.SetName(h.GetName().replace(curBox+'_'+args.model,modelName))
signalHists[h.GetName()] = signalHists.pop(x)
signalHist = signalHists[modelName]
#make combined unrolled histograms for background
makeRazorBinEvidencePlots(curBox, samples=samples, inDir=BACKGROUND_DIR, outDir=outDir,
signalHist=signalHist, unrollBins=unrollBins, debugLevel=debugLevel, zmin=1e-3)
#draw signal and background in unrolled format
if args.model == 'T2tt':
signalString = 'pp #rightarrow #tilde{t}#tilde{t}, #mu = 1.0, #tilde{t} #rightarrow t#tilde{#chi}^{0}_{1}'
elif args.model == 'T1bbbb':
btag = '0-3btag'
if box in ['DiJet', 'LeptonJet']:
btag = '0-2btag'
fitString = '--fit-region Full'
plotString = '--plot-region Full'
if fit == 'Sideband':
fitString = '--fit-region LowMR,LowRsq'
plotString = '--plot-region LowMR,LowRsq'
#exec_me('python python/DustinTuple2RooDataSet.py -b %s -c %s -d Datasets/ Run2016BCD/%s --data -l %i'% (box, config, dataset[box], lumi ), dryRun )
outDir = 'fits_%s/Run2016BCD/%s/%s/' % (dateString, box, fit)
roodataset = dataset[box].replace(
'.root', '_lumi-%.3f_%s_%s.root' % (lumi / 1000., btag, box))
#exec_me('python python/BinnedFit.py -b %s -c %s -d %s Datasets/%s --data -l %i %s %s'% (box, config, outDir, roodataset, lumi, fitString, plotString), dryRun )
#exec_me('python python/PlotFit.py -b %s -c %s -d %s -i %s/BinnedFitResults_%s.root --data -l %i'% (box, config, outDir, outDir, box, lumi), dryRun )
varyNString = ''
#if fit=='Sideband':
# varyNString = '--vary-n'
#exec_me('python python/RunToys.py -c %s -d %s -b %s -l %i -i %s/BinnedFitResults_%s.root %s --no-stat -t 10000 %s' %(config,outDir,box,lumi,outDir,box,varyNString,fitString), dryRun )
#exec_me('python python/RunToys.py -c %s -d %s -b %s -l %i -i %s/BinnedFitResults_%s.root %s -t 10000 %s' %(config,outDir,box,lumi,outDir,box,varyNString,fitString), dryRun )
exec_me(
'python python/PlotFit.py -b %s -c %s -d %s -i %s/BinnedFitResults_%s.root --data -l %i -t %s/toys_Bayes_%s.root -s %s/toys_Bayes_noStat_%s.root %s --no-stat'
% (box, config, outDir, outDir, box, lumi, outDir, box, outDir,
box, fitString), dryRun)
#exec_me('python python/PlotFit.py -b %s -c %s -d %s -i %s/BinnedFitResults_%s.root --data -l %i -t %s/toys_Bayes_%s.root -s %s/toys_Bayes_noStat_%s.root %s --no-stat --print-errors %s'% (box, config, outDir, outDir, box, lumi, outDir, box, outDir,box,fitString,plotString), dryRun )
#exec_me('python python/RunToys.py -c %s -d %s -b %s -l %i -i %s/BinnedFitResults_%s.root --freq -t 200 -s 0 %s' %(config,outDir,box,lumi,outDir,box,fitString), dryRun )
#exec_me('python python/PlotGOF.py -c %s -d %s -b %s -l %f -t %s/toys_Freq_s0_%s.root %s --data' %(config,outDir,box,lumi,outDir,box,fitString),dryRun)
# get combine parameters
if args.signif:
combineMethod = 'ProfileLikelihood'
combineFlags = '--signif -t -1 --toysFreq'
else:
combineMethod = 'Asymptotic'
combineFlags = ''
if args.saveWorkspace:
combineFlags += ' --saveWorkspace'
# run combine
if len(boxList) == 1:
cardName = getCardName(modelName, boxList[0], outDir)
combineName = 'MADD_'+boxList[0]+'_'+modelName
exec_me('combine -M %s %s -n %s %s'%(
combineMethod,cardName,combineName,combineFlags), False)
exec_me('mv higgsCombine%s.%s.mH120.root %s/'%(
combineName,combineMethod,outDir),False)
elif len(boxList) > 1:
cardNames = []
combineName = 'MADD_'+('_'.join(boxList))+'_'+modelName
for curBox in boxList:
cardName = getCardName(modelName, curBox, outDir)
cardNames.append(os.path.basename(cardName))
combinedCardName = ('RazorInclusiveMADD_%s_%s.txt'%(
modelName, '_'.join(boxList)))
exec_me('cd '+outDir+'; combineCards.py '+(' '.join(cardNames))
+' > '+combinedCardName+'; cd ..', False)
exec_me('combine -M '+combineMethod+' '+outDir+'/'
+combinedCardName+' -n '+combineName+' '+combineFlags,
False)
lumi = 3000
for box in boxes:
for cfg in configs:
if box in ['MuEle','EleEle','MuMu']:
btag = '0-1btag'
elif '1-3btag' in cfg:
btag = '1-3btag'
else:
btag = '0-3btag'
backgroundDsName = 'Datasets/RazorInclusive_SMCocktail_weighted_lumi-%.1f_%s_%s.root'%(lumi/1000.,btag,box)
if not glob.glob(backgroundDsName):
exec_me('python python/DustinTuple2RooDataSet.py -c %s -b %s -d Datasets/ -w -q Backgrounds/*.root'%(cfg,box),False)
backgroundDsName = 'Datasets/RazorInclusive_SMCocktail_unweighted_lumi-%.1f_%s_%s.root'%(lumi/1000.,btag,box)
if not glob.glob(backgroundDsName):
exec_me('python python/RooDataSet2UnweightedDataset.py -c %s -b %s -d Datasets/ Datasets/RazorInclusive_SMCocktail_weighted_lumi-%.1f_%s_%s.root'%(cfg,box,lumi/1000.,btag,box),False)
exec_me('mkdir -p fits_07102015/weighted/%s_%s/' %(box,btag),False)
exec_me('mkdir -p fits_07102015/unweighted/%s_%s/' %(box,btag),False)
# --fit-region LowMR,LowRsq
#exec_me('python python/Fit.py -c %s -d fits_06212015/unbinned/%s_%s/ --fit-region LowMR,LowRsq -b %s Datasets/RazorInclusive_SMCocktail_unweighted_lumi-%.1f_%s_%s.root' %(cfg,box,btag,box,lumi/1000,btag,box),False)
#exec_me('python python/BinnedFit.py -c %s -d fits_07102015/weighted/%s_%s/ -b %s Datasets/RazorInclusive_SMCocktail_weighted_lumi-%.1f_%s_%s.root' %(cfg,box,btag,box,lumi/1000,btag,box),False)
#exec_me('python python/RunToys.py -c %s -d fits_07102015/weighted/%s_%s/ -b %s -i fits_07102015/weighted/%s_%s/BinnedFitResults_%s.root -t 10000' %(cfg,box,btag,box,box,btag,box),False)
exec_me('python python/PlotFit.py -c %s -d fits_07102015/weighted/%s_%s/ -b %s -i fits_07102015/weighted/%s_%s/BinnedFitResults_%s.root -t fits_07102015/weighted/%s_%s/toys_Bayes_%s.root --print-errors' %(cfg,box,btag,box,box,btag,box,box,btag,box),False)
#exec_me('python python/BinnedFit.py -c %s -d fits_07102015/unweighted/%s_%s/ -b %s Datasets/RazorInclusive_SMCocktail_unweighted_lumi-%.1f_%s_%s.root' %(cfg,box,btag,box,lumi/1000,btag,box),False)
#exec_me('python python/RunToys.py -c %s -d fits_07102015/unweighted/%s_%s/ -b %s -i fits_07102015/unweighted/%s_%s/BinnedFitResults_%s.root -t 10000' %(cfg,box,btag,box,box,btag,box),False)
if __name__ == '__main__':
boxes = ['MultiJet','MuMultiJet','EleMultiJet','LooseLeptonMultiJet','DiJet']
fits = ['Sideband','Full']
configs = ['config/run2.config']
lumi = 3000
dryRun = False
btag = '0-3btag'
for box in boxes:
for cfg in configs:
backgroundDsName = 'Datasets/RazorInclusive_SMCocktail_weighted_lumi-%.1f_%s_%s.root'%(lumi/1000.,btag,box)
exec_me('python python/DustinTuple2RooDataSet.py -c %s -b %s -d Datasets/ -w -q Backgrounds/*.root'%(cfg,box),dryRun)
backgroundDsName = 'Datasets/RazorInclusive_SMCocktail_unweighted_lumi-%.1f_%s_%s.root'%(lumi/1000.,btag,box)
exec_me('python python/RooDataSet2UnweightedDataset.py -c %s -b %s -d Datasets/ Datasets/RazorInclusive_SMCocktail_weighted_lumi-%.1f_%s_%s.root'%(cfg,box,lumi/1000.,btag,box),dryRun)
lumiString = '%.1f'%(lumi/1000)
lumiString = lumiString.replace('.','p')
for weight in ["weighted","unweighted"]:
backgroundDsName = 'Datasets/RazorInclusive_SMCocktail_%s_lumi-%.1f_%s_%s.root'%(weight,lumi/1000.,btag,box)
for fit in fits:
fitString = ''
if fit=='Sideband':
fitString = '--fit-region LowMR,LowRsq'
dateString = str(datetime.date.today())
