Skip to content

Commit

Permalink
Merge pull request #16 from cmstas/scale-wgts
Browse files Browse the repository at this point in the history 
Scale wgts and systematics (muR/muF and ISR/FSR)
  • Loading branch information
@kmohrman
kmohrman authored Jan 25, 2024
2 parents bde7a11 + ff0b8d6 commit 8084e60
Show file tree
Hide file tree
Showing 140 changed files with 6,815 additions and 5,552 deletions.
12 changes: 12 additions & 0 deletions analysis/wwz/make_sample_lst_json.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -333,6 +333,16 @@
},
}

# CI example
# Note, if using this to remake the json for the CI, should replace the file in the "files" with just output_1.root (no path) since assumes will be downloaded locally
ci_dict = {
"UL17_WWZJetsTo4L2Nu_forCI" : {
"path" : "/cmsuf/data/store/user/kmohrman/test/ci_for_wwz",
"histAxisName": "UL17_WWZJetsTo4L2Nu",
"xsecName": "WWZ4l",
},
}


############################ Convenience function ############################

Expand All @@ -353,6 +363,7 @@ def make_jsons_for_dict_of_samples(samples_dict,prefix,year,out_dir,on_das=False
xsec_name = xsec_name,
hist_axis_name = hist_axis_name,
on_das = on_das,
include_lhe_wgts_arr = True,
)
out_name = sample_name+".json"
if not os.path.exists(out_name):
Expand All @@ -375,6 +386,7 @@ def main():

# A simple example
#make_jsons_for_dict_of_samples(test_wwz_dict, "/ceph/cms/","2017",".",on_das=False) # An example
make_jsons_for_dict_of_samples(ci_dict, "","2017","../../input_samples/sample_jsons/test_samples/",on_das=False) # For CI json

# Specify output paths
jsons_path = "../../input_samples/sample_jsons/"
Expand Down
35 changes: 9 additions & 26 deletions analysis/wwz/run_wwz4l.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -15,19 +15,6 @@

LST_OF_KNOWN_EXECUTORS = ["futures","work_queue","iterative"]

WGT_VAR_LST = [
"nSumOfWeights_ISRUp",
"nSumOfWeights_ISRDown",
"nSumOfWeights_FSRUp",
"nSumOfWeights_FSRDown",
"nSumOfWeights_renormUp",
"nSumOfWeights_renormDown",
"nSumOfWeights_factUp",
"nSumOfWeights_factDown",
"nSumOfWeights_renormfactUp",
"nSumOfWeights_renormfactDown",
]

if __name__ == '__main__':

parser = argparse.ArgumentParser(description='You can customize your run')
Expand Down Expand Up @@ -100,9 +87,9 @@
port.append(port[0])

# Figure out which hists to include
if args.hist_list == ["ana"]:
# Here we hardcode a list of hists used for the analysis
hist_lst = ["njets","lj0pt","ptz"]
if args.hist_list == ["few"]:
# Here we hardcode a reduced list of a few hists
hist_lst = ["j0pt", "njets", "nbtagsl", "nleps", "met", "l0pt"]
elif args.hist_list == ["cr"]:
# Here we hardcode a list of hists used for the CRs
hist_lst = ["lj0pt", "ptz", "met", "ljptsum", "l0pt", "l0eta", "l1pt", "l1eta", "j0pt", "j0eta", "njets", "nbtagsl", "invmass"]
Expand Down Expand Up @@ -177,13 +164,10 @@ def LoadJsonToSampleName(jsonFile, prefix):
samplesdict[sname]['nGenEvents'] = int(samplesdict[sname]['nGenEvents'])
samplesdict[sname]['nSumOfWeights'] = float(samplesdict[sname]['nSumOfWeights'])
if not samplesdict[sname]["isData"]:
for wgt_var in WGT_VAR_LST:
# Check that MC samples have all needed weight sums (only needed if doing systs)
if do_systs:
if (wgt_var not in samplesdict[sname]):
raise Exception(f"Missing weight variation \"{wgt_var}\".")
else:
samplesdict[sname][wgt_var] = float(samplesdict[sname][wgt_var])
# Check that MC samples have all needed weight sums (only needed if doing systs)
if do_systs:
if ("nSumOfLheWeights" not in samplesdict[sname]):
raise Exception(f"Sample is missing scale variations: {sname}")
# Print file info
print('>> '+sname)
print(' - isData? : %s' %('YES' if samplesdict[sname]['isData'] else 'NO'))
Expand All @@ -196,9 +180,8 @@ def LoadJsonToSampleName(jsonFile, prefix):
print(' - nGenEvents : %i' %samplesdict[sname]['nGenEvents'])
print(' - SumWeights : %i' %samplesdict[sname]['nSumOfWeights'])
if not samplesdict[sname]["isData"]:
for wgt_var in WGT_VAR_LST:
if wgt_var in samplesdict[sname]:
print(f' - {wgt_var}: {samplesdict[sname][wgt_var]}')
if "nSumOfLheWeights" in samplesdict[sname]:
print(f' - nSumOfLheWeights : {samplesdict[sname]["nSumOfLheWeights"]}')
print(' - Prefix : %s' %samplesdict[sname]['redirector'])
print(' - nFiles : %i' %len(samplesdict[sname]['files']))
for fname in samplesdict[sname]['files']: print(' %s'%fname)
Expand Down
73 changes: 43 additions & 30 deletions analysis/wwz/wwz4l.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -145,8 +145,6 @@ def columns(self):
# Main function: run on a given dataset
def process(self, events):

TMPdosys = 0 # Temporary standin flag for now (eventualy just use self._do_systematics)

# Dataset parameters
dataset = events.metadata["dataset"]

Expand All @@ -158,30 +156,32 @@ def process(self, events):

# Get up down weights from input dict
if (self._do_systematics and not isData):
# Otherwise we have an NLO xsec, so for these systs we will have e.g. xsec_NLO*(N_pass_up/N_gen_up)
lhe_sow = self._samples[dataset]["nSumOfLheWeights"]
# This assumes we have an NLO xsec, so for these systs we will have e.g. xsec_NLO*(N_pass_up/N_gen_up)
# Thus these systs should only affect acceptance and effeciency and shape
# The uncty on xsec comes from NLO and is applied as a rate uncty in the text datacard
sow_ISRUp = self._samples[dataset]["nSumOfWeights_ISRUp" ]
sow_ISRDown = self._samples[dataset]["nSumOfWeights_ISRDown" ]
sow_FSRUp = self._samples[dataset]["nSumOfWeights_FSRUp" ]
sow_FSRDown = self._samples[dataset]["nSumOfWeights_FSRDown" ]
sow_renormUp = self._samples[dataset]["nSumOfWeights_renormUp" ]
sow_renormDown = self._samples[dataset]["nSumOfWeights_renormDown" ]
sow_factUp = self._samples[dataset]["nSumOfWeights_factUp" ]
sow_factDown = self._samples[dataset]["nSumOfWeights_factDown" ]
sow_renormfactUp = self._samples[dataset]["nSumOfWeights_renormfactUp" ]
sow_renormfactDown = self._samples[dataset]["nSumOfWeights_renormfactDown" ]
if lhe_sow == []:
sow_renormDown = sow
sow_factDown = sow
sow_factUp = sow
sow_renormUp = sow
elif len(lhe_sow) == 9:
sow_renormDown = lhe_sow[1]
sow_factDown = lhe_sow[3]
sow_factUp = lhe_sow[5]
sow_renormUp = lhe_sow[7]
elif len(lhe_sow) == 8:
sow_renormDown = lhe_sow[1]
sow_factDown = lhe_sow[3]
sow_factUp = lhe_sow[4]
sow_renormUp = lhe_sow[6]
else: raise Exception("ERROR: Unknown LHE weights length {len(lhe_sow)}")
else:
sow_ISRUp = -1
sow_ISRDown = -1
sow_FSRUp = -1
sow_FSRDown = -1
sow_renormUp = -1
sow_renormDown = -1
sow_factUp = -1
sow_factDown = -1
sow_renormfactUp = -1
sow_renormfactDown = -1


datasets = ["SingleMuon", "SingleElectron", "EGamma", "MuonEG", "DoubleMuon", "DoubleElectron", "DoubleEG"]
for d in datasets:
Expand Down Expand Up @@ -286,28 +286,41 @@ def process(self, events):
weights_obj_base.add("norm",(xsec/sow)*genw*lumi*sm_wgt)


# Set up the list of systematics that are handled via event weight variations
wgt_correction_syst_lst = [
"btagSFlight_correlated", "btagSFbc_correlated", f"btagSFlight_uncorrelated_{year}", f"btagSFbc_uncorrelated_{year}",
"lepSF_elec", "lepSF_muon", "PreFiring", "PU"
]
wgt_correction_syst_lst = append_up_down_to_sys_base(wgt_correction_syst_lst)

if not isData:
# Scale weights
cor_tc.AttachPSWeights(events)
cor_tc.AttachScaleWeights(events)
# FSR/ISR weights
# For now only consider variations in the numerator
weights_obj_base.add('ISR', events.nom, events.ISRUp, events.ISRDown)
weights_obj_base.add('FSR', events.nom, events.FSRUp, events.FSRDown)
# Renorm/fact scale
weights_obj_base.add('renorm', events.nom, events.renormUp*(sow/sow_renormUp), events.renormDown*(sow/sow_renormDown))
weights_obj_base.add('fact', events.nom, events.factUp*(sow/sow_factUp), events.factDown*(sow/sow_factDown))

# Misc other experimental SFs and systs
weights_obj_base.add('PreFiring', events.L1PreFiringWeight.Nom, events.L1PreFiringWeight.Up, events.L1PreFiringWeight.Dn)
weights_obj_base.add('PU', cor_tc.GetPUSF((events.Pileup.nTrueInt), year), cor_tc.GetPUSF(events.Pileup.nTrueInt, year, 'up'), cor_tc.GetPUSF(events.Pileup.nTrueInt, year, 'down'))

# Lepton SFs and systs
weights_obj_base.add("lepSF_muon", events.sf_4l_muon, copy.deepcopy(events.sf_4l_hi_muon), copy.deepcopy(events.sf_4l_lo_muon))
weights_obj_base.add("lepSF_elec", events.sf_4l_elec, copy.deepcopy(events.sf_4l_hi_elec), copy.deepcopy(events.sf_4l_lo_elec))


# Set up the list of systematics that are handled via event weight variations
wgt_correction_syst_lst = [
"btagSFlight_correlated", "btagSFbc_correlated", f"btagSFlight_uncorrelated_{year}", f"btagSFbc_uncorrelated_{year}",
"lepSF_elec", "lepSF_muon", "PreFiring", "PU",
"renorm", "fact", "ISR", "FSR",
]
wgt_correction_syst_lst = append_up_down_to_sys_base(wgt_correction_syst_lst)


######### The rest of the processor is inside this loop over systs that affect object kinematics ###########

obj_correction_systs = [] # Will have e.g. jes etc

# If we're doing systematics and this isn't data, we will loop over the obj correction syst lst list
if TMPdosys and not isData: obj_corr_syst_var_list = ["nominal"] + obj_correction_systs
if self._do_systematics and not isData: obj_corr_syst_var_list = ["nominal"] + obj_correction_systs
# Otherwise loop juse once, for nominal
else: obj_corr_syst_var_list = ['nominal']

Expand Down Expand Up @@ -391,7 +404,7 @@ def process(self, events):
weights_obj_base_for_kinematic_syst.add("btagSF", wgt_btag_nom)

# Put the btagging up and down weight variations into the weights object
if TMPdosys:
if self._do_systematics:
for btag_sys in ["correlated", "uncorrelated"]:
year_tag = f"_{year}"
if btag_sys == "correlated": year_tag = ""
Expand Down Expand Up @@ -767,7 +780,7 @@ def process(self, events):
# Set up the list of weight fluctuations to loop over
# For now the syst do not depend on the category, so we can figure this out outside of the filling loop
wgt_var_lst = ["nominal"]
if TMPdosys:
if self._do_systematics:
if not isData:
if (obj_corr_syst_var != "nominal"):
# In this case, we are dealing with systs that change the kinematics of the objs (e.g. JES)
Expand Down
37 changes: 24 additions & 13 deletions input_samples/sample_jsons/test_samples/UL17_WWZJetsTo4L2Nu_forCI.json
View file
Original file line number Diff line number Diff line change
@@ -1,15 +1,26 @@
{
"xsec": 0.002067,
"year": "2017",
"treeName": "Events",
"histAxisName": "UL17_WWZJetsTo4L2Nu",
"options": "",
"WCnames": [],
"files": [
"output_1.root"
],
"nEvents": 54868,
"nGenEvents": 109760,
"nSumOfWeights": 227.15914198199997,
"isData": false
"xsec": 0.002067,
"year": "2017",
"treeName": "Events",
"histAxisName": "UL17_WWZJetsTo4L2Nu",
"options": "",
"WCnames": [],
"files": [
"output_1.root"
],
"nEvents": 54868,
"nGenEvents": 109760,
"nSumOfWeights": 227.15914198199997,
"isData": false,
"path": "/cmsuf/data/store/user/kmohrman/test/ci_for_wwz",
"nSumOfLheWeights": [
240.5168487561945,
236.51634455287,
233.27592728014238,
230.91358444751938,
224.05776881139798,
222.95829859531798,
219.4090775656835,
216.425230687383
]
}
Loading

0 comments on commit 8084e60

Please sign in to comment.