## * rescale MC
## * maximum runtime of 1h
## * maximum memory per cpu of 2GB
samples = [
{
"ID": 2129,
# Either specify ID *OR* DB name
#"db_name": "ST_t-channel_4f_leptonDecays_13TeV-amcatnlo_MiniAODv2_v1.0.0+7415_TTAnalysis_12d3865",
"files_per_job": 10,
# Optional: specify a path to a "json skeleton" that will be filled and used for this sample (see below)
#"json_skeleton": "myJson.json",
}
]
mySub = slurmSubmitter(samples, "<path to plotter / skimmer exe>", "test_slurm/", rescale=True, runtime="60", memory=2000)
## Create test_slurm directory and subdirs
mySub.setupDirs()
## Write command and data files in the slurm directory
mySub.createFiles()
## Actually submit the jobs
## It is recommended to do a dry-run first without submitting to slurm
#mySub.submit()
## Example Json skeleton: the field "#DB_NAME#" will be filled by slurmSubmitter.
## You can specify anything else you want that will be passed to createHistoWithtMultiDraw
## for this particular sample.
def create_slurm(samples, output, executable):
## Create Slurm submitter to handle job creating
#mySub = slurmSubmitter(samples, "%s/build/" % output + executable, "DUMMY", output + "/", rescale=True)
mySub = slurmSubmitter(samples,
"%s/build/" % output + executable,
output + "/",
rescale=True)
## Create test_slurm directory and subdirs
mySub.setupDirs()
splitTT = False
splitDY = False
def get_node(db_name):
split_name = db_name.split("_")
node = None
for i, it in enumerate(split_name):
if it == "node":
node = split_name[i + 1]
break
if node is None:
raise Exception(
"Could not extract node from DB name {}".format(db_name))
return node
def get_node_id(node):
if node == "SM": return "-1"
elif node == "box": return "0"
else: return node
## Modify the input samples to add sample cuts and stuff
for sample in mySub.sampleCfg[:]:
# TTbar final state splitting
if splitTT and 'TT_TuneCUETP8M2T4_13TeV-powheg-pythia8' in sample[
"db_name"]:
# Fully leptonic
#tt_fl_sample = copy.deepcopy(sample)
#newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
#tt_fl_sample["db_name"] = sample["db_name"].replace("TT_Tune", "TT_FL_Tune")
#newJson["sample_cut"] = "(hh_gen_ttbar_decay_type >= 4 && hh_gen_ttbar_decay_type <= 10 && hh_gen_ttbar_decay_type != 7)"
#tt_fl_sample["json_skeleton"][tt_fl_sample["db_name"]] = newJson
#tt_fl_sample["json_skeleton"].pop(sample["db_name"])
#mySub.sampleCfg.append(tt_fl_sample)
## Semi leptonic
#tt_sl_sample = copy.deepcopy(sample)
#newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
#tt_sl_sample["db_name"] = sample["db_name"].replace("TT_Tune", "TT_SL_Tune")
#newJson["sample_cut"] = "(hh_gen_ttbar_decay_type == 2 || hh_gen_ttbar_decay_type == 3 || hh_gen_ttbar_decay_type == 7)"
#tt_sl_sample["json_skeleton"][tt_sl_sample["db_name"]] = newJson
#tt_sl_sample["json_skeleton"].pop(sample["db_name"])
#mySub.sampleCfg.append(tt_sl_sample)
## Fully hadronic
#tt_fh_sample = copy.deepcopy(sample)
#newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
#tt_fh_sample["db_name"] = sample["db_name"].replace("TT_Tune", "TT_FH_Tune")
#newJson["sample_cut"] = "(hh_gen_ttbar_decay_type == 1)"
#tt_fh_sample["json_skeleton"][tt_fh_sample["db_name"]] = newJson
#tt_fh_sample["json_skeleton"].pop(sample["db_name"])
#mySub.sampleCfg.append(tt_fh_sample)
# Not fully leptonic
tt_other_sample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
tt_other_sample["db_name"] = sample["db_name"].replace(
"TT_Tune", "TT_Other_Tune")
newJson[
"sample_cut"] = "(hh_gen_ttbar_decay_type <= 3 || hh_gen_ttbar_decay_type == 7)"
tt_other_sample["json_skeleton"][
tt_other_sample["db_name"]] = newJson
tt_other_sample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(tt_other_sample)
mySub.sampleCfg.remove(sample)
if splitDY and 'DYJetsToLL_' in sample["db_name"]:
# Z + jj' ; j = b/c, j' = b/c
dy_bb_sample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
dy_bb_sample["db_name"] = sample["db_name"].replace(
"DYJetsToLL", "DYBBOrCCToLL")
newJson[
"sample_cut"] = "(hh_llmetjj_HWWleptons_nobtag_cmva.gen_bb || hh_llmetjj_HWWleptons_nobtag_cmva.gen_cc)"
dy_bb_sample["json_skeleton"][dy_bb_sample["db_name"]] = newJson
dy_bb_sample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(dy_bb_sample)
# Z + jj' ; j = b/c, j' = l
dy_bx_sample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
dy_bx_sample["db_name"] = sample["db_name"].replace(
"DYJetsToLL", "DYBXOrCXToLL")
newJson[
"sample_cut"] = "(hh_llmetjj_HWWleptons_nobtag_cmva.gen_bl || hh_llmetjj_HWWleptons_nobtag_cmva.gen_cl || hh_llmetjj_HWWleptons_nobtag_cmva.gen_bc)"
dy_bx_sample["json_skeleton"][dy_bx_sample["db_name"]] = newJson
dy_bx_sample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(dy_bx_sample)
# Z + jj' ; j = l, j' = l
dy_xx_sample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
dy_xx_sample["db_name"] = sample["db_name"].replace(
"DYJetsToLL", "DYXXToLL")
newJson[
"sample_cut"] = "(hh_llmetjj_HWWleptons_nobtag_cmva.gen_ll)"
dy_xx_sample["json_skeleton"][dy_xx_sample["db_name"]] = newJson
dy_xx_sample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(dy_xx_sample)
# Merging with HT binned sample: add cut on inclusive one
if 'DYJetsToLL_M-5to50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8' in sample[
"db_name"] or 'DYJetsToLL_M-50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8' in sample[
"db_name"]:
sample["json_skeleton"][
sample["db_name"]]["sample_cut"] = "event_ht < 100"
#if 'WJetsToLNu_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8' in sample["db_name"]:
# sample["json_skeleton"][sample["db_name"]]["sample_cut"] = "event_ht < 100"
## Write command and data files in the slurm directory
mySub.createFiles()
# Actually submit the jobs
# It is recommended to do a dry-run first without submitting to slurm
if args.submit:
mySub.submit()
def create_slurm(samples, output, executable):
## Create slurm submitter to handle job creating
mySub = slurmSubmitter(samples, "%s/build/" % output + executable, output + "/", rescale=True, memory=2000, runtime="360")
## Create test_slurm directory and subdirs
mySub.setupDirs()
splitTT = True
splitDY = False
def get_node(db_name):
split_name = db_name.split("_")
node = None
for i, it in enumerate(split_name):
if it == "node":
node = split_name[i+1]
break
if node is None:
raise Exception("Could not extract node from DB name {}".format(db_name))
return node
def get_node_id(node):
if node == "SM": return "-1"
elif node == "box": return "0"
else: return node
## Modify the input samples to add sample cuts and stuff
for sample in mySub.sampleCfg[:]:
# TTbar final state splitting
if splitTT and 'TT_TuneCUETP8M2T4_13TeV-powheg-pythia8' in sample["db_name"]:
# Fully leptonic
#tt_fl_sample = copy.deepcopy(sample)
#newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
#tt_fl_sample["db_name"] = sample["db_name"].replace("TT_Tune", "TT_FL_Tune")
#newJson["sample_cut"] = "(hh_gen_ttbar_decay_type >= 4 && hh_gen_ttbar_decay_type <= 10 && hh_gen_ttbar_decay_type != 7)"
#tt_fl_sample["json_skeleton"][tt_fl_sample["db_name"]] = newJson
#tt_fl_sample["json_skeleton"].pop(sample["db_name"])
#mySub.sampleCfg.append(tt_fl_sample)
## Semi leptonic
#tt_sl_sample = copy.deepcopy(sample)
#newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
#tt_sl_sample["db_name"] = sample["db_name"].replace("TT_Tune", "TT_SL_Tune")
#newJson["sample_cut"] = "(hh_gen_ttbar_decay_type == 2 || hh_gen_ttbar_decay_type == 3 || hh_gen_ttbar_decay_type == 7)"
#tt_sl_sample["json_skeleton"][tt_sl_sample["db_name"]] = newJson
#tt_sl_sample["json_skeleton"].pop(sample["db_name"])
#mySub.sampleCfg.append(tt_sl_sample)
## Fully hadronic
#tt_fh_sample = copy.deepcopy(sample)
#newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
#tt_fh_sample["db_name"] = sample["db_name"].replace("TT_Tune", "TT_FH_Tune")
#newJson["sample_cut"] = "(hh_gen_ttbar_decay_type == 1)"
#tt_fh_sample["json_skeleton"][tt_fh_sample["db_name"]] = newJson
#tt_fh_sample["json_skeleton"].pop(sample["db_name"])
#mySub.sampleCfg.append(tt_fh_sample)
# Not fully leptonic
tt_other_sample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
tt_other_sample["db_name"] = sample["db_name"].replace("TT_Tune", "TT_Other_Tune")
newJson["sample_cut"] = "(hh_gen_ttbar_decay_type <= 3 || hh_gen_ttbar_decay_type == 7)"
tt_other_sample["json_skeleton"][tt_other_sample["db_name"]] = newJson
tt_other_sample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(tt_other_sample)
mySub.sampleCfg.remove(sample)
if splitDY and 'DYJetsToLL_' in sample["db_name"]:
# Z + jj' ; j = b/c, j' = b/c
dy_bb_sample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
dy_bb_sample["db_name"] = sample["db_name"].replace("DYJetsToLL", "DYBBOrCCToLL")
newJson["sample_cut"] = "(hh_llmetjj_HWWleptons_nobtag_cmva.gen_bb || hh_llmetjj_HWWleptons_nobtag_cmva.gen_cc)"
dy_bb_sample["json_skeleton"][dy_bb_sample["db_name"]] = newJson
dy_bb_sample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(dy_bb_sample)
# Z + jj' ; j = b/c, j' = l
dy_bx_sample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
dy_bx_sample["db_name"] = sample["db_name"].replace("DYJetsToLL", "DYBXOrCXToLL")
newJson["sample_cut"] = "(hh_llmetjj_HWWleptons_nobtag_cmva.gen_bl || hh_llmetjj_HWWleptons_nobtag_cmva.gen_cl || hh_llmetjj_HWWleptons_nobtag_cmva.gen_bc)"
dy_bx_sample["json_skeleton"][dy_bx_sample["db_name"]] = newJson
dy_bx_sample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(dy_bx_sample)
# Z + jj' ; j = l, j' = l
dy_xx_sample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
dy_xx_sample["db_name"] = sample["db_name"].replace("DYJetsToLL", "DYXXToLL")
newJson["sample_cut"] = "(hh_llmetjj_HWWleptons_nobtag_cmva.gen_ll)"
dy_xx_sample["json_skeleton"][dy_xx_sample["db_name"]] = newJson
dy_xx_sample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(dy_xx_sample)
# Merging with HT binned sample: add cut on inclusive one
# if 'DYJetsToLL_M-5to50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8' in sample["db_name"] or 'DYJetsToLL_M-50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8' in sample["db_name"]:
# sample["json_skeleton"][sample["db_name"]]["sample_cut"] = "event_ht < 100"
#if 'WJetsToLNu_TuneCUETP8M1_13TeV-amcatnloFXFX-pythia8' in sample["db_name"]:
# sample["json_skeleton"][sample["db_name"]]["sample_cut"] = "event_ht < 100"
# Benchmark to training grid reweighting (ME-based)
if "node" in sample["db_name"]:
base = get_node(sample["db_name"])
for grid_point in nonresonant_signal_grid:
kl = "{:.2f}".format(grid_point[0])
kt = "{:.2f}".format(grid_point[1])
weight_args = [get_node_id(base), grid_point[0], grid_point[1], number_of_bases]
newSample = copy.deepcopy(sample)
newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
point_str = "base_" + base + "_point_" + kl + "_" + kt
point_str = point_str.replace(".", "p").replace("-", "m")
newSample["db_name"] = sample["db_name"].replace("node_" + base, point_str)
newJson["sample-weight"] = "training_grid"
newJson["sample-weight-args"] = weight_args
newSample["json_skeleton"][newSample["db_name"]] = newJson
newSample["json_skeleton"].pop(sample["db_name"])
mySub.sampleCfg.append(newSample)
mySub.sampleCfg.remove(sample)
## Cluster to MV reweighting (ME-based)
#operators_MV = ["OtG", "Otphi", "O6", "OH"]
#if "node" in sample["db_name"]:
# for base, base_name in enumerate(rwgt_base):
# for i, op1 in enumerate(operators_MV):
# newSample = copy.deepcopy(sample)
# newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
#
# newSample["db_name"] = sample["db_name"].replace("node_" + base_name, "SM_" + op1)
# newJson["sample-weight"] = "base_" + base_name + "_SM_" + op1
#
# newSample["json_skeleton"][newSample["db_name"]] = newJson
# newSample["json_skeleton"].pop(sample["db_name"])
#
# mySub.sampleCfg.append(newSample)
#
# for j, op2 in enumerate(operators_MV):
# if i < j: continue
#
# newSample = copy.deepcopy(sample)
# newJson = copy.deepcopy(sample["json_skeleton"][sample["db_name"]])
#
# newSample["db_name"] = sample["db_name"].replace("node_" + base_name, op1 + "_" + op2)
# newJson["sample-weight"] = "base_" + base_name + "_" + op1 + "_" + op2
#
# newSample["json_skeleton"][newSample["db_name"]] = newJson
# newSample["json_skeleton"].pop(sample["db_name"])
#
# mySub.sampleCfg.append(newSample)
## Write command and data files in the slurm directory
mySub.createFiles()
# Actually submit the jobs
# It is recommended to do a dry-run first without submitting to slurm
if args.submit:
mySub.submit()