DL selection cleanup

uhh-cms · Aug 7, 2024 · c778967 · c778967
1 parent feeeb1d
commit c778967
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 13 deletions.
diff --git a/hbw/selection/dl_remastered.py b/hbw/selection/dl_remastered.py
@@ -107,7 +107,7 @@ def dl_lepton_selection(
 
     dilepton = ak.pad_none(lepton, 2)
     dilepton = dilepton[:, 0] + dilepton[:, 1]
-    events = set_ak_column(events, "mll", dilepton.mass)
+    events = set_ak_column(events, "mll", dilepton.mass, value_type=np.float32)
     lepton_results.steps["DiLeptonMass81"] = ak.fill_none(dilepton.mass <= m_z.nominal - 10, False)
     # lepton channel masks
     lepton_results.steps["Lep_mm"] = mm_mask = (

diff --git a/hbw/selection/jet.py b/hbw/selection/jet.py
@@ -58,8 +58,8 @@ def jet_selection(
         (events.Jet.jetId >= 2)  # 1: loose, 2: tight, 4: isolated, 6: tight+isolated
     )
 
-    electron = events.Electron[lepton_results.objects.Electron.FakeableElectron]
-    muon = events.Muon[lepton_results.objects.Muon.FakeableMuon]
+    electron = events.Electron[lepton_results.objects.Electron.LooseElectron]
+    muon = events.Muon[lepton_results.objects.Muon.LooseMuon]
 
     jet_mask = (
         (events.Jet.pt >= self.jet_pt) &
@@ -293,9 +293,9 @@ def sl_boosted_jet_selection(
 
     # get separation info between FatJets and AK4 Jets
     dr_fatjet_ak4 = events.FatJet.metric_table(ak4_jets)
-    events = set_ak_column(events, "FatJet.n_subjets", ak.sum(dr_fatjet_ak4 < 0.8, axis=2))
-    events = set_ak_column(events, "FatJet.n_separated_jets", ak.sum(dr_fatjet_ak4 > 1.2, axis=2))
-    events = set_ak_column(events, "FatJet.max_dr_ak4", ak.max(dr_fatjet_ak4, axis=2))
+    events = set_ak_column(events, "cutflow.FatJet.n_subjets", ak.sum(dr_fatjet_ak4 < 0.8, axis=2))
+    events = set_ak_column(events, "cutflow.FatJet.n_separated_jets", ak.sum(dr_fatjet_ak4 > 1.2, axis=2))
+    events = set_ak_column(events, "cutflow.FatJet.max_dr_ak4", ak.max(dr_fatjet_ak4, axis=2))
 
     # baseline fatjet selection
     fatjet_mask = (
@@ -400,11 +400,7 @@ def sl_boosted_jet_selection_init(self: Selector) -> None:
 
     # add produced variables to *produces* only when requested
     if self.config_inst.x("do_cutflow_features", False):
-        self.produces |= {"cutflow.n_fatjet", "cutflow.n_hbbjet"} | four_vec(
-            {"FatJet", "HbbJet"},
-            {"n_subjets", "n_separated_jets", "max_dr_ak4"},
-            skip_defaults=True,
-        )
+        self.produces |= {"cutflow.{n_fatjet,n_hbbjet}, cutflow.FatJet.{n_subjets,n_separated_jets,max_dr_ak4"}
 
 
 # boosted selection for the DL channel (only one parameter needs to be changed)

diff --git a/hbw/selection/lepton.py b/hbw/selection/lepton.py
@@ -235,7 +235,7 @@ def add_nlep_variables(config: od.Config) -> None:
         "n_fakeable_muon",
         "n_tight_electron",
         "n_tight_muon",
-        "n_veto_tau"
+        "n_veto_tau",
     ):
         config.add_variable(
             name=f"cf.{var}",

diff --git a/hbw/tasks/inspection.py b/hbw/tasks/inspection.py
@@ -32,12 +32,16 @@ class DumpAnalysisSummary(
         description="Keys of interest to be printed in the summary",
     )
 
+    @property
+    def keys_repr(self):
+        return "_".join(sorted(self.keys_of_interest))
+
     def requires(self):
         return {}
 
     def output(self):
         output = {
-            "dataset_summary": self.target("dataset_summary.txt"),
+            "dataset_summary": self.target(f"dataset_summary_{self.keys_repr}.txt"),
         }
         return output
 
@@ -52,6 +56,7 @@ def write_dataset_summary(self, outp):
                 "n_events": "Number of events",
                 "n_files": "Number of files",
                 "das_keys": "DAS keys",
+                "rucio": "Rucio DAS keys",
                 "process": "Process name",
                 "xsec": "Cross section [pb]",
                 "xsec_unc": "Cross section +- unc [pb]",
@@ -66,6 +71,7 @@ def write_dataset_summary(self, outp):
                         "n_events": dataset.n_events,
                         "n_files": dataset.n_files,
                         "das_keys": dataset.get_info("nominal").keys[0],
+                        "rucio": "cms:" + dataset.get_info("nominal").keys[0],
                         "process": dataset.processes.get_first().name,
                         "xsec": round_sig(xsec.nominal, 4) if xsec else "0",
                         "xsec_unc": xsec.str("pdg", combine_uncs="all") if xsec else "0",