Improve cut diagnostics print outs, clean up dipho preselection

Preselection/data/samples.json

@@ -1,5 +1,6 @@
 {
     "HH_ggTauTau" : { 
+        "resonant" : true,
         "fpo" : 1,
         "process_id" : -1,
         "2016" : {
@@ -17,6 +18,7 @@
         } 
     },
     "Data" : {
+        "resonant" : false,
         "fpo" : 10,
         "process_id" : 0,
         "2016" : {
@@ -33,6 +35,7 @@
         }
     },
     "ZGamma" : {
+        "resonant" : false,
         "fpo" : 1,
         "process_id" : 2,
         "2016" : {
@@ -49,6 +52,7 @@
         }
     },
     "DiPhoton" : {
+        "resonant" : false,
         "fpo" : 5,
         "process_id" : 3,
         "2016" : {
@@ -65,6 +69,7 @@
         }
     },
     "WGamma" : {
+        "resonant" : false,
         "fpo" : 1,
         "process_id" : 4,
         "2016" : {
@@ -81,6 +86,7 @@
         }
     },
     "TTbar" : {
+        "resonant" : false,
         "fpo" : 5,
         "process_id" : 5,
         "2016" : {
@@ -97,6 +103,7 @@
         }
     },
     "TTGamma" : {
+        "resonant" : false,
         "fpo" : 1,
         "process_id" : 6,
         "2016" : {
@@ -113,6 +120,7 @@
         }
     },
     "TTGG" : {
+        "resonant" : false,
         "fpo" : 1,
         "process_id" : 7,
         "2016" : {
@@ -129,6 +137,7 @@
         }
     },
     "GJets_HT40To100" : {
+        "resonant" : false,
         "fpo" : 5,
         "process_id" : 8,
         "2016" : {
@@ -145,6 +154,7 @@
         }
     },
     "GJets_HT-100To200" : {
+        "resonant" : false,
         "fpo" : 5,
         "process_id" : 8,
         "2016" : {
@@ -161,6 +171,7 @@
         }
     },
     "GJets_HT-200To400" : {
+        "resonant" : false,
         "fpo" : 5,
         "process_id" : 8,
         "2016" : {
@@ -177,6 +188,7 @@
         }
     },
     "GJets_HT-400To600" : {
+        "resonant" : false,
         "fpo" : 5,
         "process_id" : 8,
         "2016" : {
@@ -193,6 +205,7 @@
         }
     },
     "GJets_HT-600ToInf" : {
+        "resonant" : false,
         "fpo" : 5,
         "process_id" : 8,
         "2016" : {
@@ -209,6 +222,7 @@
         }
     },
     "VH" : {
+        "resonant" : true,
         "fpo" : 1,
         "process_id" : 9,
         "2016" : {

Preselection/data/samples_and_scale1fb.json

@@ -33,7 +33,8 @@
             ]
         },
         "fpo": 10,
-        "process_id": 0
+        "process_id": 0,
+        "resonant": false
     },
     "DiPhoton": {
         "2016": {
@@ -70,7 +71,8 @@
             ]
         },
         "fpo": 5,
-        "process_id": 3
+        "process_id": 3,
+        "resonant": false
     },
     "GJets_HT-100To200": {
         "2016": {
@@ -108,7 +110,8 @@
             ]
         },
         "fpo": 5,
-        "process_id": 8
+        "process_id": 8,
+        "resonant": false
     },
     "GJets_HT-200To400": {
         "2016": {
@@ -146,7 +149,8 @@
             ]
         },
         "fpo": 5,
-        "process_id": 8
+        "process_id": 8,
+        "resonant": false
     },
     "GJets_HT-400To600": {
         "2016": {
@@ -184,7 +188,8 @@
             ]
         },
         "fpo": 5,
-        "process_id": 8
+        "process_id": 8,
+        "resonant": false
     },
     "GJets_HT-600ToInf": {
         "2016": {
@@ -222,7 +227,8 @@
             ]
         },
         "fpo": 5,
-        "process_id": 8
+        "process_id": 8,
+        "resonant": false
     },
     "GJets_HT40To100": {
         "2016": {
@@ -260,7 +266,8 @@
             ]
         },
         "fpo": 5,
-        "process_id": 8
+        "process_id": 8,
+        "resonant": false
     },
     "HH_ggTauTau": {
         "2016": {
@@ -298,7 +305,8 @@
             ]
         },
         "fpo": 1,
-        "process_id": -1
+        "process_id": -1,
+        "resonant": true
     },
     "TTGG": {
         "2016": {
@@ -335,7 +343,8 @@
             ]
         },
         "fpo": 1,
-        "process_id": 7
+        "process_id": 7,
+        "resonant": false
     },
     "TTGamma": {
         "2016": {
@@ -374,7 +383,8 @@
             ]
         },
         "fpo": 1,
-        "process_id": 6
+        "process_id": 6,
+        "resonant": false
     },
     "TTbar": {
         "2016": {
@@ -411,7 +421,8 @@
             ]
         },
         "fpo": 5,
-        "process_id": 5
+        "process_id": 5,
+        "resonant": false
     },
     "VH": {
         "2016": {
@@ -448,7 +459,8 @@
             ]
         },
         "fpo": 1,
-        "process_id": 9
+        "process_id": 9,
+        "resonant": true
     },
     "WGamma": {
         "2016": {
@@ -487,7 +499,8 @@
             ]
         },
         "fpo": 1,
-        "process_id": 4
+        "process_id": 4,
+        "resonant": false
     },
     "ZGamma": {
         "2016": {
@@ -526,6 +539,7 @@
             ]
         },
         "fpo": 1,
-        "process_id": 2
+        "process_id": 2,
+        "resonant": false
     }
 }

Preselection/helpers/loop_helper.py

@@ -26,6 +26,9 @@ def __init__(self, **kwargs):
         self.options = kwargs.get("options")
         self.systematics = kwargs.get("systematics")
         self.years = kwargs.get("years").split(",")
+        self.select_samples = kwargs.get("select_samples")
+        if self.select_samples != "all":
+            self.select_samples = self.select_samples.split(",")
 
         self.output_tag = kwargs.get("output_tag")
         self.output_dir = kwargs.get("output_dir")
@@ -98,6 +101,10 @@ def prepare_jobs(self):
         self.jobs_manager = []
 
         for sample, info in self.samples_dict.items():
+            if self.select_samples != "all":
+                if sample not in self.select_samples:
+                    continue
+
             if self.debug > 0:
                 print("[LoopHelper] Running over sample: %s" % sample)
                 print("[LoopHelper] details: ", info)
@@ -117,7 +124,8 @@ def prepare_jobs(self):
                     "process_id" : info["process_id"],
                     "year" : year,
                     "scale1fb" : 1 if sample == "Data" else year_info["metadata"]["scale1fb"],
-                    "lumi" : lumi_map[year]
+                    "lumi" : lumi_map[year],
+                    "resonant" : info["resonant"]
                 }
 
                 file_splits = self.chunks(files, info["fpo"])
@@ -207,21 +215,21 @@ def write_summary(self):
     ### Physics: selections, etc ###
     ################################
 
-    def select_events(self, events):
+    def select_events(self, events, metadata):
         # Dipho preselection
-        events = photon_selections.diphoton_preselection(events, self.debug)
+        events = photon_selections.diphoton_preselection(events, metadata["resonant"], self.debug)
         events.Photon = events.Photon[photon_selections.select_photons(events, self.debug)]
 
         if self.selections == "HHggTauTau_InclusivePresel":
             events = analysis_selections.ggTauTau_inclusive_preselection(events, self.debug)
-            events.Electron = events.Electron[lepton_selections.select_electrons(events, self.debug)]
-            events.Muon = events.Muon[lepton_selections.select_muons(events, self.debug)]
-            events.Tau = events.Tau[tau_selections.select_taus(events, self.debug)]
+            events.Electron = events.Electron[lepton_selections.select_electrons(events, events.Photon, self.debug)]
+            events.Muon = events.Muon[lepton_selections.select_muons(events, events.Photon, self.debug)]
+            events.Tau = events.Tau[tau_selections.select_taus(events, events.Photon, events.Muon, events.Electron, self.debug)]
 
         elif self.selections == "ttH_LeptonicPresel":
             events = analysis_selections.tth_leptonic_preselection(events, self.debug)
-            events.Electron = events.Electron[lepton_selections.select_electrons(events, self.debug)]
-            events.Muon = events.Muon[lepton_selections.select_muons(events, self.debug)]
+            events.Electron = events.Electron[lepton_selections.select_electrons(events, events.Photon, self.debug)]
+            events.Muon = events.Muon[lepton_selections.select_muons(events, events.Photon, self.debug)]
 
         return events
 
@@ -243,6 +251,10 @@ def loop_sample(self, job):
         files = job["files"]
         output = job["output"]
 
+        selection_metadata = {
+            "resonant" : info["resonant"]
+        }
+
         if self.debug > 0:
             print("[LoopHelper] Running job with parameters", job)
 
@@ -262,7 +274,7 @@ def loop_sample(self, job):
             if events is None:
                 self.outputs.pop(output)
                 return
-            events = self.select_events(events)
+            events = self.select_events(events, selection_metadata)
 
             events["process_id"] = numpy.ones(len(events)) * process_id
             if data:

Preselection/loop.py

@@ -29,6 +29,12 @@
     type = str,
     default = "2016,2017,2018"
 )
+parser.add_argument(
+    "--select_samples",
+    help = "csv list of samples to run over (should be a subset of samples in args.samples)",
+    type = str,
+    default = "all"
+)
 
 # --options points to a json file containing options for looping
 # this could include things like additional scaling of bkg samples,

Preselection/scripts/scale1fb.py

@@ -89,6 +89,8 @@ def calculate_metadata(files, xs, debug):
     for sample, info in output.items():
         if sample in original_output.keys():
             for year, year_info in original_output[sample].items():
+                if "201" not in year:
+                    continue
                 if "metadata" in year_info.keys():
                     if "scale1fb" in year_info["metadata"].keys():
                         output[sample][year] = original_output[sample][year]

Preselection/selections/analysis_selections.py

@@ -9,24 +9,37 @@
 import selections.photon_selections as photon_selections
 
 def ggTauTau_inclusive_preselection(events, debug):
-    cut_diagnostics = utils.CutDiagnostics(n_events_initial = len(events), debug = debug, cut_set = "[analysis_selections.py : ggTauTau_inclusive_preselection]")
+    cut_diagnostics = utils.CutDiagnostics(events = events, debug = debug, cut_set = "[analysis_selections.py : ggTauTau_inclusive_preselection]")
 
     # Get number of electrons, muons, taus
-    n_electrons = awkward.num(events.Electron[lepton_selections.select_electrons(events, debug)])
-    n_muons = awkward.num(events.Muon[lepton_selections.select_muons(events, debug)])
+    electron_selection = lepton_selections.select_electrons(events, events.Photon, debug)
+    muon_selection = lepton_selections.select_muons(events, events.Photon, debug)
+    tau_selection = tau_selections.select_taus(events, events.Photon, events.Muon[muon_selection], events.Electron[electron_selection], debug) 
+
+    n_electrons = awkward.num(events.Electron[electron_selection])
+    n_muons = awkward.num(events.Muon[muon_selection])
+    n_taus = awkward.num(events.Tau[tau_selection])
+
+    # Require >= 1 lep/tau
+    n_leptons_and_taus = n_electrons + n_muons + n_taus
+    lep_tau_cut = n_leptons_and_taus >= 1
 
-    n_taus = awkward.num(events.Tau[tau_selections.select_taus(events, debug)])
+    # Require OS leptons/taus for events with 2 leptons/taus
+    sum_charge = awkward.sum(events.Electron[electron_selection].charge, axis=1) + awkward.sum(events.Muon[muon_selection].charge, axis=1) + awkward.sum(events.Tau[tau_selection].charge, axis=1)
+    charge_cut = sum_charge == 0
+    n_lep_cut = n_leptons_and_taus == 2
+    not_two_leptons = n_leptons_and_taus != 2
+    os_cut = (n_lep_cut & charge_cut) | not_two_leptons # only require 2 OS leptons if there are ==2 leptons in the event
 
-    n_leptons_and_taus = n_electrons + n_muons + n_taus
+    all_cuts = lep_tau_cut & os_cut
+    cut_diagnostics.add_cuts([lep_tau_cut, os_cut, all_cuts], ["N_leptons + N_taus >= 1", "OS dileptons", "all"])
 
-    lep_tau_cut = n_leptons_and_taus >= 1
-    events = events[lep_tau_cut]
-    cut_diagnostics.add_cut(len(events), cut_name = "leptons and taus >= 1 cut")
+    events = events[all_cuts]
 
     return events
 
 def tth_leptonic_preselection(events, debug):
-    cut_diagnostics = utils.CutDiagnostics(n_events_initial = len(events), debug = debug, cut_set = "[analysis_selections.py : tth_leptonic_preselection]")
+    cut_diagnostics = utils.CutDiagnostics(events = events, debug = debug, cut_set = "[analysis_selections.py : tth_leptonic_preselection]")
 
     # Get number of electrons, muons
     n_electrons = awkward.num(events.Electron[lepton_selections.select_electrons(events, debug)])