Turned isVBFsearch boolean into vbf-search argument for skimmer.

rkansal47 · Aug 21, 2023 · 2005c9b · 2005c9b
1 parent 7e1b2be
commit 2005c9b
Show file tree

Hide file tree

Showing 6 changed files with 26 additions and 6 deletions.
diff --git a/src/HHbbVV/processors/bbVVSkimmer.py b/src/HHbbVV/processors/bbVVSkimmer.py
@@ -118,7 +118,7 @@ class bbVVSkimmer(processor.ProcessorABC):
         min_branches.append(f"DijetMass_{shift}")
 
     def __init__(
-        self, xsecs={}, save_ak15=False, save_systematics=True, inference=True, save_all=True
+        self, xsecs={}, save_ak15=False, save_systematics=True, inference=True, save_all=True,vbf_search = False
     ):
         super(bbVVSkimmer, self).__init__()
 
@@ -133,6 +133,9 @@ def __init__(
 
         # save all branches or only necessary ones
         self._save_all = save_all
+
+        # search for VBF production events
+        self._vbf_search = vbf_search
 
         # for tagger model and preprocessing dict
         self.tagger_resources_path = (
@@ -142,7 +145,7 @@ def __init__(
         self._accumulator = processor.dict_accumulator({})
 
         logger.info(
-            f"Running skimmer with inference {self._inference} and systematics {self._systematics} and save all {self._save_all}"
+            f"Running skimmer with inference {self._inference} and systematics {self._systematics} and save all {self._save_all} and VBF search {self._vbf_search}"
         )
 
     def to_pandas(self, events: Dict[str, np.array]):
@@ -188,7 +191,6 @@ def process(self, events: ak.Array):
 
         isData = "JetHT" in dataset
         isQCD = "QCD" in dataset
-        isVBFSearch = True
         isSignal = (
             "GluGluToHHTobbVV" in dataset
             or "XToYHTo2W2BTo4Q2B" in dataset
@@ -295,7 +297,7 @@ def process(self, events: ak.Array):
                 }
 
         # VBF ak4 Jet vars (pt, eta, phi, M, nGoodJets)
-        if isVBFSearch:
+        if self._vbf_search:
             ak4JetVars = {
                 **self.getVBFVars(events, ak8FatJetVars, bb_mask, isGen="VBF_HHTobbVV" in dataset)
             }
@@ -781,7 +783,6 @@ def getVBFVars(
             ak4_jet_mask
             & electron_muon_overlap_mask
             & fatjet_overlap_mask
-            & (np.abs(jets.eta) > 1.5)
         )
         vbfJets = jets[vbfJets_mask]
 

diff --git a/src/condor/submit.py b/src/condor/submit.py
@@ -108,6 +108,7 @@ def main(args):
                     "eosoutroot": f"{eosoutput_dir}/root/nano_skim_{j}.root",
                     "save_ak15": "--save-ak15" if args.save_ak15 else "--no-save-ak15",
                     "save_all": "--save-all" if args.save_all else "--no-save-all",
+                    "save_all": "--vbf-search" if args.vbf_search else "--no-vbf-search",
                     "save_systematics": "--save-systematics"
                     if args.save_systematics
                     else "--no-save-systematics",

diff --git a/src/condor/submit.templ.sh b/src/condor/submit.templ.sh
@@ -9,7 +9,7 @@ mkdir outfiles
 
 # run code
 # pip install --user onnxruntime
-python -u -W ignore $script --year $year --starti $starti --endi $endi --samples $sample --subsamples $subsample --processor $processor --maxchunks $maxchunks --chunksize $chunksize --label $label --njets $njets ${save_ak15} ${save_systematics} ${inference} ${save_all}
+python -u -W ignore $script --year $year --starti $starti --endi $endi --samples $sample --subsamples $subsample --processor $processor --maxchunks $maxchunks --chunksize $chunksize --label $label --njets $njets ${save_ak15} ${save_systematics} ${inference} ${save_all} ${vbf_search}
 
 #move output to eos
 xrdcp -f outfiles/* $eosoutpkl

diff --git a/src/condor/submit_configs/skimmer_inputs_23_07_31.yaml b/src/condor/submit_configs/skimmer_inputs_23_07_31.yaml
@@ -11,5 +11,19 @@
             "TTbarBoosted": { "files_per_job": 20 },
             "WJetsToQQ": { "files_per_job": 20 },
             "ZJetsToQQ": { "files_per_job": 20 },
+            "HH": { "subsamples": [
+                            "GluGluToHHTobbVV_node_cHHH0",
+                            "GluGluToHHTobbVV_node_cHHH1",
+                            "GluGluToHHTobbVV_node_cHHH2p45",
+                            "GluGluToHHTobbVV_node_cHHH5",
+                            "VBF_HHTobbVV_CV_1_C2V_1_C3_1",
+                            "VBF_HHTobbVV_CV_0_5_C2V_1_C3_1",
+                            "VBF_HHTobbVV_CV_1_C2V_0_C3_1",
+                            "VBF_HHTobbVV_CV_1_C2V_1_C3_1",
+                            "VBF_HHTobbVV_CV_1_C2V_2_C3_1",
+                            "VBF_HHTobbVV_CV_1_5_C2V_1_C3_1",
+                            "VBF_HHTobbVV_CV_1_C2V_1_C3_0",
+                            "VBF_HHTobbVV_CV_1_C2V_1_C3_2",
+                        ], "files_per_job": 20, "chunksize": 10000 },
         },
 }
diff --git a/src/run.py b/src/run.py
@@ -138,6 +138,7 @@ def main(args):
         args.save_systematics,
         args.inference,
         args.save_all,
+        args.vbf_search,
     )
 
     if len(args.files):

diff --git a/src/run_utils.py b/src/run_utils.py
@@ -113,6 +113,7 @@ def get_processor(
     save_systematics: bool = None,
     inference: bool = None,
     save_all: bool = None,
+    vbf_search: bool = None,
 ):
     # define processor
     if processor == "trigger":
@@ -128,6 +129,7 @@ def get_processor(
             save_systematics=save_systematics,
             inference=inference,
             save_all=save_all,
+            vbf_search=vbf_search,
         )
     elif processor == "input":
         from HHbbVV.processors import TaggerInputSkimmer
@@ -180,3 +182,4 @@ def parse_common_args(parser):
     add_bool_arg(parser, "save-systematics", default=False, help="save systematic variations")
     add_bool_arg(parser, "save-all", default=True, help="save all branches")
     add_bool_arg(parser, "inference", default=True, help="run inference for ak8 jets")
+    add_bool_arg(parser, "vbf-search", default=False, help="run selections for VBF production search")