Merge pull request #61 from uhh-cms/stitching_debuggin_objects

Add unstitched tags and Datasets for debugging purpose

hbt/config/configs_hbt.py

@@ -7,6 +7,7 @@
 from __future__ import annotations
 
 import os
+import re
 import itertools
 import functools
 
@@ -345,7 +346,6 @@ def if_not_era(*, values: list[str | None] | None = None, **kwargs) -> list[str]
 
         # add the dataset
         dataset = cfg.add_dataset(campaign.get_dataset(dataset_name))
-
         # add tags to datasets
         if dataset.name.startswith("data_e_"):
             dataset.add_tag({"etau", "emu_from_e", "ee"})
@@ -359,8 +359,12 @@ def if_not_era(*, values: list[str | None] | None = None, **kwargs) -> list[str]
             dataset.add_tag({"has_top", "single_top", "st"})
         if dataset.name.startswith("dy_"):
             dataset.add_tag("dy")
+        if re.match(r"^dy_m50toinf_\dj_(|pt.+_)amcatnlo$", dataset.name):
+            dataset.add_tag("dy_stitched")
         if dataset.name.startswith("w_lnu_"):
             dataset.add_tag("w_lnu")
+        if re.match(r"^w_lnu_\dj_(|pt.+_)amcatnlo$", dataset.name):
+            dataset.add_tag("w_lnu_stitched")
         # datasets that are known to have no lhe info at all
         if law.util.multi_match(dataset.name, [
             r"^(ww|wz|zz)_.*pythia$",
@@ -510,10 +514,29 @@ def if_not_era(*, values: list[str | None] | None = None, **kwargs) -> list[str]
             dataset.name for dataset in cfg.datasets
             if dataset.is_mc and not dataset.has_tag("signal")
         ]),
+        "backgrounds_unstitched": (backgrounds_unstitched := [
+            dataset.name for dataset in cfg.datasets
+            if (
+                dataset.is_mc and
+                not dataset.has_tag("signal") and
+                not dataset.has_tag({"dy_stitched", "w_lnu_stitched"}, mode=any)
+            )
+        ]),
         "sm_ggf": (sm_ggf_group := ["hh_ggf_hbb_htt_kl1_kt1_powheg", *backgrounds]),
-        "sm": (sm_group := ["hh_ggf_hbb_htt_kl1_kt1_powheg", "hh_vbf_hbb_htt_kv1_k2v1_kl1_madgraph", *backgrounds]),
+        "sm": (sm_group := [
+            "hh_ggf_hbb_htt_kl1_kt1_powheg",
+            "hh_vbf_hbb_htt_kv1_k2v1_kl1_madgraph",
+            *backgrounds,
+        ],
+        ),
+        "sm_unstitched": (sm_group_unstitched := [
+            "hh_ggf_hbb_htt_kl1_kt1_powheg",
+            "hh_vbf_hbb_htt_kv1_k2v1_kl1_madgraph",
+            *backgrounds_unstitched,
+        ]),
         "sm_ggf_data": data_group + sm_ggf_group,
         "sm_data": data_group + sm_group,
+        "sm_data_unstitched": data_group + sm_group_unstitched,
         "dy": [dataset.name for dataset in cfg.datasets if dataset.has_tag("dy")],
         "w_lnu": [dataset.name for dataset in cfg.datasets if dataset.has_tag("w_lnu")],
     }

hbt/config/variables.py

@@ -41,14 +41,6 @@ def add_variables(config: od.Config) -> None:
         x_title="Luminosity block",
         discrete_x=True,
     )
-    add_variable(
-        config,
-        name="n_jet",
-        expression="n_jet",
-        binning=(11, -0.5, 10.5),
-        x_title="Number of jets",
-        discrete_x=True,
-    )
     add_variable(
         config,
         name="n_hhbtag",
@@ -640,6 +632,15 @@ def build_hh(events, which=None):
         x_title=r"Subleading muon $\phi$",
     )
 
+    add_variable(
+        config,
+        name="njets",
+        expression=lambda events: ak.num(events.Jet["pt"], axis=1),
+        aux={"inputs": {"Jet.pt"}},
+        binning=(11, -0.5, 10.5),
+        x_title=r"Number of jets",
+    )
+
     for proc in ["hh", "tt", "dy"]:
         # outputs of the resonant pDNN at SM-like mass and spin values
         add_variable(

hbt/weight/default.py

@@ -60,6 +60,12 @@ def default_init(self: WeightProducer) -> None:
         }
 
 
+normalization_inclusive = default.derive(
+    "normalization_inclusive",
+    cls_dict={"drop_weights": {"normalization_weight"}},
+)
+
+
 normalization_only = default.derive(
     "normalization_only",
     cls_dict={"keep_weights": {"normalization_weight"}},