add memory configuration and load gen columns only when required

hbw/production/weights.py

@@ -27,7 +27,7 @@
 from hbw.production.normalized_weights import normalized_weight_factory
 from hbw.production.normalized_btag import normalized_btag_weights
 from hbw.production.dataset_normalization import dataset_normalization_weight
-from hbw.util import has_tag
+from hbw.util import has_tag, IF_DY, IF_TOP
 
 
 np = maybe_import("numpy")
@@ -39,8 +39,8 @@
 
 
 @producer(
-    uses={gen_parton_top, gen_v_boson, pu_weight},
-    produces={gen_parton_top, gen_v_boson, pu_weight},
+    uses={IF_TOP(gen_parton_top), IF_DY(gen_v_boson), pu_weight},
+    produces={IF_TOP(gen_parton_top), IF_DY(gen_v_boson), pu_weight},
     mc_only=True,
 )
 def event_weights_to_normalize(self: Producer, events: ak.Array, results: SelectionResult, **kwargs) -> ak.Array:

hbw/util.py

@@ -663,3 +663,27 @@ def IF_MC(self: ArrayFunction.DeferredColumn, func: ArrayFunction) -> Any | set[
         return self.get()
 
     return self.get() if func.dataset_inst.is_mc else None
+
+
+@deferred_column
+def IF_DY(self: ArrayFunction.DeferredColumn, func: ArrayFunction) -> Any | set[Any]:
+    if getattr(func, "dataset_inst", None) is None:
+        return self.get()
+
+    return self.get() if func.dataset_inst.has_tag("is_v_jets") else None
+
+
+@deferred_column
+def IF_TOP(self: ArrayFunction.DeferredColumn, func: ArrayFunction) -> Any | set[Any]:
+    if getattr(func, "dataset_inst", None) is None:
+        return self.get()
+
+    return self.get() if func.dataset_inst.has_tag("has_top") else None
+
+
+@deferred_column
+def IF_TT(self: ArrayFunction.DeferredColumn, func: ArrayFunction) -> Any | set[Any]:
+    if getattr(func, "dataset_inst", None) is None:
+        return self.get()
+
+    return self.get() if func.dataset_inst.has_tag("is_ttbar") else None

law.cfg

@@ -126,6 +126,18 @@ lfn_sources: local_desy_dcache, wlcg_fs_desy_store, wlcg_fs_infn_redirector, wlc
 ; c22post__cf.CalibrateEvents__nomin*: htcondor_memory=5GB
 ; cf.MLTraining: htcondor_memory=10GB, htcondor_gpus=1
 
+# use cluster default for signals
+# cf.SelectEvents__hh*: dummy=0
+# cf.ReduceEvents__hh*: dummy=0
+# DY and W needs more memory due to gen_v_boson Producer
+cf.SelectEvents__dy*: htcondor_memory=3GB
+cf.ReduceEvents__dy*: htcondor_memory=3GB
+cf.SelectEvents__w_lnu: htcondor_memory=3GB
+cf.ReduceEvents__w_lnu: htcondor_memory=3GB
+# default resources for all other datasets
+cf.SelectEvents: htcondor_memory=2GB
+cf.ReduceEvents: htcondor_memory=2GB
+
 
 [luigi_cf.DummyTask]
 # To set defaults on a per-task basis