signalFastSim = [
SMS_T2tt_dM_10to80_genHT_160_genMET_80_mWMin_0p1
]
samples = signalFullSim + signalFastSim
from CMGTools.TTHAnalysis.setup.Efficiencies import *
dataDir = "$CMSSW_BASE/src/CMGTools/TTHAnalysis/data"
#Define splitting
for comp in signalFullSim:
comp.isMC = True
comp.isData = False
comp.isFastSim = False
comp.splitFactor = 250 # if comp.name in [ "WJets", "DY3JetsM50", "DY4JetsM50","W1Jets","W2Jets","W3Jets","W4Jets","TTJetsHad" ] else 100
for comp in signalFastSim:
comp.isMC = True
comp.isData = False
comp.isFastSim = True
comp.splitFactor = 250 # if comp.name in [ "WJets", "DY3JetsM50", "DY4JetsM50","W1Jets","W2Jets","W3Jets","W4Jets","TTJetsHad" ] else 100
if __name__ == "__main__":
from CMGTools.RootTools.samples.tools import runMain
runMain(samples, localobjs=locals())
import sys
if "test" in sys.argv:
from CMGTools.RootTools.samples.ComponentCreator import testSamples
testSamples(samples)
mcSamples_Asymptotic25ns = TTs + SingleTop + VJets + DYJetsM50HT + WJetsToLNuHT + ZJetsToNuNuHT + GJetsHT + QCDPt + DiBosons + TTV + Higgs + TriBosons + Rares + QCD_ElX + QCD_MuX
mcSamples_Asymptotic50ns = [ TTJets_50ns, TTJets_LO_50ns, TT_pow_50ns, WJetsToLNu_50ns, DYJetsToLL_M10to50_50ns, DYJetsToLL_M50_50ns, GJets_Pt15to6000_50ns] + DYJetsM50HT_50ns + QCDPt_50ns + SingleTop_50ns + DiBosons_50ns + QCD_MuX_50ns
mcSamples = RelVals740 + mcSamples_Asymptotic25ns + mcSamples_Asymptotic50ns
samples = mcSamples
### ---------------------------------------------------------------------
from CMGTools.TTHAnalysis.setup.Efficiencies import *
dataDir = "$CMSSW_BASE/src/CMGTools/TTHAnalysis/data"
#Define splitting
for comp in mcSamples:
comp.isMC = True
comp.isData = False
comp.splitFactor = 250 # if comp.name in [ "WJets", "DY3JetsM50", "DY4JetsM50","W1Jets","W2Jets","W3Jets","W4Jets","TTJetsHad" ] else 100
comp.puFileMC=dataDir+"/puProfile_Summer12_53X.root"
comp.puFileData=dataDir+"/puProfile_Data12.root"
comp.efficiency = eff2012
if __name__ == "__main__":
import sys
if "test" in sys.argv:
from CMGTools.RootTools.samples.ComponentCreator import testSamples
testSamples(samples)
mcSamples_monojet_Asymptotic25ns = DYJetsToNuNu_MJ + VJets_MJ + Top_MJ + DiBosons_MJ + QCDHT
mcSamples_monojet_Asymptotic25ns_signals = DM_Scalars + DM_Pseudoscalars + DM_Vectors + DM_Axials
### ----------------------------- summary ----------------------------------------
mcSamples_monojet = mcSamples_monojet_Asymptotic25ns + mcSamples_monojet_Asymptotic25ns_signals
from CMGTools.TTHAnalysis.setup.Efficiencies import *
dataDir = "$CMSSW_BASE/src/CMGTools/TTHAnalysis/data"
#Define splitting
for comp in mcSamples:
comp.isMC = True
comp.isData = False
comp.splitFactor = 250 # if comp.name in [ "WJets", "DY3JetsM50", "DY4JetsM50","W1Jets","W2Jets","W3Jets","W4Jets","TTJetsHad" ] else 100
comp.puFileMC=dataDir+"/puProfile_Summer12_53X.root"
comp.puFileData=dataDir+"/puProfile_Data12.root"
comp.efficiency = eff2012
# for comp in dataSamples:
# comp.splitFactor = 1000
# comp.isMC = False
# comp.isData = True
if __name__ == "__main__":
import sys
if "test" in sys.argv:
from CMGTools.RootTools.samples.ComponentCreator import testSamples
testSamples(mcSamples)
def runMain(samples, args=None, localobjs=None):
if args == None: args = sys.argv
selsamples = _filterSamples(samples, args)
if "help" in args or "--help" in args or "-h" in args:
print """
python samplefile.py test [--AAA] [samples] :
tries accessing the first file of each sample
option -AAA: allow AAA as fallback
python samplefile.py locality [samples] :
check the locality of the samples
python samplefile.py refresh [samples] [ --pretend ] [ --suspicious ]:
forces a refresh of the cache
option --pretend: print the list of samples to refresh, instead of actually refreshing them
option --suspicious: selects for refresh the samples that look bogus (zero files, or zero events for official CMS datasets)
python samplefile.py list [samples]:
python samplefile.py summary [samples]:
two equivalent commands that prints a list of samples, with number of files, events, equivalent luminosity, etc
python samplefile.py genXSecAna [samples] [ --pretend ] [ --verbose ] [ --AAA ]:
check the cross sections using genXSecAna on one of the files
python samplefile.py checkdecl:
check that all samples are declared in the samples list
"""
if "test" in args:
from CMGTools.RootTools.samples.ComponentCreator import testSamples
testSamples(selsamples, allowAAA=("--AAA" in args))
if "locality" in args:
import re
from CMGTools.Production.localityChecker import LocalityChecker
tier2Checker = LocalityChecker("T2_CH_CERN", datasets="/*/*/MINIAOD*")
for comp in selsamples:
if len(comp.files) == 0:
print '\033[34mE: Empty component: ' + comp.name + '\033[0m'
continue
if not hasattr(comp, 'dataset'): continue
if not re.match("/[^/]+/[^/]+/MINIAOD(SIM)?", comp.dataset):
continue
if "/store/" not in comp.files[0]: continue
if re.search("/store/(group|user|cmst3)/", comp.files[0]): continue
if not tier2Checker.available(comp.dataset):
print "\033[1;31mN: Dataset %s (%s) is not available on T2_CH_CERN\033[0m" % (
comp.name, comp.dataset)
else:
print "Y: Dataset %s (%s) is available on T2_CH_CERN" % (
comp.name, comp.dataset)
if "refresh" in args:
from CMGTools.Production.cacheChecker import CacheChecker
checker = CacheChecker()
for d in selsamples:
if "--suspicious" in args:
if len(d.files) > 0:
if "/store/mc " in d.files[0] or "/store/data" in d.files[
0]:
if getattr(d, 'dataset_entries', -1) > 0:
continue
else:
continue
print "Checking ", d.name, " aka ", d.dataset
if "--pretend" in args: continue
checker.checkComp(d, verbose=True)
if "check_versions" in args:
from CMGTools.Production.datasetVersionChecker import DatasetVersionChecker
checker = DatasetVersionChecker()
for d in selsamples:
if "--pretend" in args:
print "Would check ", d.name, " aka ", d.dataset
else:
print "Checking", d.name, " ",
checker.checkComp(d, verbose=True)
if "list" in args or "summary" in args:
from CMGTools.RootTools.samples.configTools import printSummary, mergeExtensions
if "--merge-extensions" in args or "--mex" in args:
selsamples = mergeExtensions(selsamples, verbose=True)[0]
printSummary(selsamples)
if "genXSecAna" in args:
import subprocess, re
if "--fetch" in args or not os.path.exists(
"%s/src/genXSecAna.py" % os.environ['CMSSW_BASE']):
print "Retrieving genXSecAna.py"
os.system(
"wget -O " + os.environ['CMSSW_BASE'] +
"/src/genXSecAna.py https://raw.githubusercontent.com/syuvivida/generator/master/cross_section/runJob/ana.py"
)
for d in selsamples:
if not hasattr(d, 'xSection'):
print "Skipping %s which has no cross section" % d.name
continue
if "--pretend" in args:
print "Would check ", d.name, " aka ", d.dataset
continue
if "--AAA" in args:
from CMGTools.Production.changeComponentAccessMode import convertComponent
convertComponent(d, "root://cms-xrd-global.cern.ch/%s")
else: # use LFNs
d.files = [
re.sub(".*(/store/.*)(?:\\?.*)?", "\\1", f)
for f in d.files[:]
]
print "Sample %s: XS(sample file) = %g pb, ... " % (d.name,
d.xSection),
if len(d.files) == 0:
print "\n\033[01;31m ERROR: no files in sample, so cannot run the analyzer \033[00m"
continue
if "--verbose" in args:
print "\n ".join([
"cmsRun", os.environ['CMSSW_BASE'] + "/src/genXSecAna.py",
"inputFiles=%s" % d.files[0], "maxEvents=-1"
])
xsecAnaOut = subprocess.check_output([
"cmsRun", os.environ['CMSSW_BASE'] + "/src/genXSecAna.py",
"inputFiles=%s" % d.files[0], "maxEvents=-1"
],
stderr=subprocess.STDOUT)
if "--verbose" in args:
for l in xsecAnaOut.split("\n"):
print "\t>> " + l
m = re.search(
r"After filter: final cross section = (\S+) \+- (\S+) pb",
xsecAnaOut)
if m and float(m.group(1)) == 0:
m = re.search(
r"Before matching: total cross section = (\S+) \+- (\S+) pb",
xsecAnaOut)
m1 = re.search(
r"After matching: total cross section = (\S+) \+- (\S+) pb",
xsecAnaOut)
if m1 and m and float(m1.group(1)) < 0 and float(
m.group(1)) > 0 and abs(
float(m1.group(1)) / float(m.group(1)) + 1) < 1e-2:
print "\033[01;33m [after filter Xsec is zero, using before filter one] \033[00m"
else:
m = None
if not m or float(m.group(1)) <= 0:
print "\n\033[01;31m ERROR: could not find After filter cross section in the output, or it's zero. \033[00m"
continue
xs, xserr = float(m.group(1)), float(m.group(2))
kfactor = d.xSection / xs
if abs(xs - d.xSection) < min(3 * xserr, 1e-2 * xs):
(col, stat) = '\033[01;36m', "OK"
elif 0.8 < kfactor and kfactor < 1.4:
(col, stat) = '\033[01;36m', "OK?"
elif 0.5 < kfactor and kfactor < 2.0:
(col, stat) = '\033[01;33m', "WARNING"
else:
(col, stat) = '\033[01;31m', "ERROR"
m = re.search(
r"After filter: final fraction of events with negative weights = (\S+) \+- (\S+)",
xsecAnaOut)
if m:
fnegv, fnegerr = (float(m.group(1)), float(m.group(2)))
fneg = " f(negw): %.3f +- %.3f " % (fnegv, fnegerr)
if getattr(d, 'fracNegWeights', None) != None:
if abs(d.fracNegWeights - fnegv) < 0.02:
fneg += "(%.3f in sample file, \033[01;36mOK\033[00m)" % d.fracNegWeights
else:
fneg += "(%.3f in sample file, \033[01;33mWARNING\033[00m)" % d.fracNegWeights
else:
fneg = ""
print "XS(genAnalyzer) = %g +/- %g pb : %s kFactor = %g %s\033[00m%s" % (
xs, xserr, col, kfactor, stat, fneg)
if "checkdecl" in args:
if localobjs == None:
raise RuntimeError(
"you have to runMain(samples,localobjs=locals())")
import PhysicsTools.HeppyCore.framework.config as cfg
ok = 0
for name, obj in localobjs.iteritems():
if name == "comp": continue # local variable used in loops
if isinstance(obj, cfg.Component):
if obj not in samples:
print "\tERROR: component %s is not added to the samples list " % name
elif obj.name != name:
print "\tERROR: component %s has inconsistent name %s " % (
name, obj.name)
else:
ok += 1
print "\tINFO: %d correctly declared components" % ok
DoubleEG_ZElectron_ReReco_files = [ f.strip() for f in open("%s/src/CMGTools/MonoXAnalysis/python/samples/ZElectron_ReReco_PS2016.txt" % os.environ['CMSSW_BASE'], "r") ]
DoubleEG_ZElectron_ReReco = kreator.makePrivateDataComponent('DoubleEG_ZElectron_ReReco','/store/group/dpg_ecal/comm_ecal/localreco/data2016_zskim_multifits/miniaod/run2016B_rereco/',DoubleEG_ZElectron_ReReco_files, dcsjson )
DoubleEG_ZElectron_std_files = [ f.strip() for f in open("%s/src/CMGTools/MonoXAnalysis/python/samples/ZElectron_ReReco_std.txt" % os.environ['CMSSW_BASE'], "r") ]
DoubleEG_ZElectron_std = kreator.makePrivateDataComponent('DoubleEG_ZElectron_std','/store/group/dpg_ecal/comm_ecal/localreco/data2016_zskim_multifits/miniaod/run2016B_std/',DoubleEG_ZElectron_std_files, dcsjson )
PrivateSamplesData = [DoubleEG_ZElectron_std, DoubleEG_ZElectron_ReReco]
from CMGTools.TTHAnalysis.setup.Efficiencies import *
dataDir = "$CMSSW_BASE/src/CMGTools/TTHAnalysis/data"
#Define splitting
for comp in mcSamples:
comp.isMC = True
comp.isData = False
comp.splitFactor = 250 # if comp.name in [ "WJets", "DY3JetsM50", "DY4JetsM50","W1Jets","W2Jets","W3Jets","W4Jets","TTJetsHad" ] else 100
comp.puFileMC=dataDir+"/puProfile_Summer12_53X.root"
comp.puFileData=dataDir+"/puProfile_Data12.root"
comp.efficiency = eff2012
# for comp in dataSamples:
# comp.splitFactor = 1000
# comp.isMC = False
# comp.isData = True
if __name__ == "__main__":
import sys
if "test" in sys.argv:
from CMGTools.RootTools.samples.ComponentCreator import testSamples
testSamples(mcSamples)
##Higgsino
## ------------------------------------------------------
#Higgsino_300_120=kreator.makeMCComponentFromLocal("Higgsino_300_120","/higgsino_300_120_v5/","/afs/cern.ch/work/b/bschneid/public/higgsino-points/%s",".*root")
#Higgsino_300_80=kreator.makeMCComponentFromLocal("Higgsino_300_80","/higgsino_300_80_v5/","/afs/cern.ch/work/b/bschneid/public/higgsino-points/%s",".*root")
#Higgsino_500_120=kreator.makeMCComponentFromLocal("Higgsino_500_120","higgsino_500_120_v5","/afs/cern.ch/work/b/bschneid/public/higgsino-points/%s",".*root")
#Higgsino_500_80=kreator.makeMCComponentFromLocal("Higgsino_500_80","higgsino_500_80_v5","/afs/cern.ch/work/b/bschneid/public/higgsino-points/%s",".*root")
Higgsino = [
] #Higgsino_300_120, Higgsino_300_80, Higgsino_500_120, Higgsino_500_80]
## mcSamplesPriv
## ------------------------------------------------------
mcSamplesPriv = SqGltttt + T1ttbb + T1ttbbWW + T1tttt + T2tt + T2ttDeg + T2cc + T5qqqqVV + T5qqqqVVDeg + T5ttttDeg + T6ttWW + T6qqWW + Higgsino
from CMGTools.TTHAnalysis.setup.Efficiencies import *
dataDir = "$CMSSW_BASE/src/CMGTools/TTHAnalysis/data"
#Define splitting
for comp in mcSamplesPriv:
comp.isMC = True
comp.isData = False
comp.splitFactor = 250 # if comp.name in [ "WJets", "DY3JetsM50", "DY4JetsM50","W1Jets","W2Jets","W3Jets","W4Jets","TTJetsHad" ] else 100
comp.puFileMC = dataDir + "/puProfile_Summer12_53X.root"
comp.puFileData = dataDir + "/puProfile_Data12.root"
comp.efficiency = eff2012
if __name__ == "__main__":
import sys
if "test" in sys.argv:
from CMGTools.RootTools.samples.ComponentCreator import testSamples
testSamples(mcSamplesPriv)
def runMain(samples, args=None):
if args == None: args = sys.argv
selsamples = _filterSamples(samples, args)
if "help" in args or "--help" in args or "-h" in args:
print """
python samplefile.py test [--AAA] [samples] :
tries accessing the first file of each sample
option -AAA: allow AAA as fallback
python samplefile.py locality [samples] :
check the locality of the samples
python samplefile.py refresh [samples] [ --pretend ] [ --suspicious ]:
forces a refresh of the cache
option --pretend: print the list of samples to refresh, instead of actually refreshing them
option --suspicious: selects for refresh the samples that look bogus (zero files, or zero events for official CMS datasets)
python samplefile.py list [samples]:
python samplefile.py summary [samples]:
two equivalent commands that prints a list of samples, with number of files, events, equivalent luminosity, etc
"""
if "test" in args:
from CMGTools.RootTools.samples.ComponentCreator import testSamples
testSamples(selsamples, allowAAA=("--AAA" in args))
if "locality" in args:
import re
from CMGTools.Production.localityChecker import LocalityChecker
tier2Checker = LocalityChecker("T2_CH_CERN", datasets="/*/*/MINIAOD*")
for comp in selsamples:
if len(comp.files) == 0:
print '\033[34mE: Empty component: ' + comp.name + '\033[0m'
continue
if not hasattr(comp, 'dataset'): continue
if not re.match("/[^/]+/[^/]+/MINIAOD(SIM)?", comp.dataset):
continue
if "/store/" not in comp.files[0]: continue
if re.search("/store/(group|user|cmst3)/", comp.files[0]): continue
if not tier2Checker.available(comp.dataset):
print "\033[1;31mN: Dataset %s (%s) is not available on T2_CH_CERN\033[0m" % (
comp.name, comp.dataset)
else:
print "Y: Dataset %s (%s) is available on T2_CH_CERN" % (
comp.name, comp.dataset)
if "refresh" in args:
from CMGTools.Production.cacheChecker import CacheChecker
checker = CacheChecker()
for d in selsamples:
if "--suspicious" in args:
if len(d.files) > 0:
if "/store/mc " in d.files[0] or "/store/data" in d.files[
0]:
if getattr(d, 'dataset_entries', -1) > 0:
continue
else:
continue
print "Checking ", d.name, " aka ", d.dataset
if "--pretend" in args: continue
checker.checkComp(d, verbose=True)
if "check_versions" in args:
from CMGTools.Production.datasetVersionChecker import DatasetVersionChecker
checker = DatasetVersionChecker()
for d in selsamples:
if "--pretend" in args:
print "Would check ", d.name, " aka ", d.dataset
else:
print "Checking", d.name, " ",
checker.checkComp(d, verbose=True)
if "list" in args or "summary" in args:
from CMGTools.HToZZ4L.tools.configTools import printSummary
dataSamples = samples
printSummary(selsamples)
## ------------------------------------------------------
## note: cross section for q~ q~ from https://twiki.cern.ch/twiki/bin/view/LHCPhysics/SUSYCrossSections13TeVsquarkantisquark (i.e. gluinos and stops decoupled)
##T6qqWW_mSq950_mCh325_mChi300 = kreator.makeMCComponentFromEOS('T6qqWW_mSq950_mCh325_mChi300', '/SMS_T6qqWW_mSq950_mChi325_mLSP300/', '/store/cmst3/group/susy/alobanov/MC/PHYS14/PU20_25ns/%s', '.*root', 0.0898112)
T6qqWW = [ ]
## mcSamplesPriv
## ------------------------------------------------------
mcSamplesPriv = SqGltttt + T1ttbb + T1ttbbWW + T1tttt + T2tt + T2ttDeg + T5qqqqVV + T5qqqqVVDeg + T5ttttDeg + T6ttWW + T6qqWW
from CMGTools.TTHAnalysis.setup.Efficiencies import *
dataDir = "$CMSSW_BASE/src/CMGTools/TTHAnalysis/data"
#Define splitting
for comp in mcSamplesPriv:
comp.isMC = True
comp.isData = False
comp.splitFactor = 250 # if comp.name in [ "WJets", "DY3JetsM50", "DY4JetsM50","W1Jets","W2Jets","W3Jets","W4Jets","TTJetsHad" ] else 100
comp.puFileMC=dataDir+"/puProfile_Summer12_53X.root"
comp.puFileData=dataDir+"/puProfile_Data12.root"
comp.efficiency = eff2012
if __name__ == "__main__":
import sys
if "test" in sys.argv:
from CMGTools.RootTools.samples.ComponentCreator import testSamples
testSamples(mcSamplesPriv)
