Merge pull request #57 from uhh-cms/dev

v0.2

hbw/config/categories.py

@@ -80,6 +80,9 @@ def kwargs_fn(root_cats):
     kwargs = {
         "id": sum([c.id for c in root_cats.values()]),
         "label": ", ".join([c.name for c in root_cats.values()]),
+        "aux": {
+            "root_cats": {key: value.name for key, value in root_cats.items()},
+        },
     }
     return kwargs
 

hbw/config/config_run2.py

@@ -359,7 +359,7 @@ def make_jme_filename(jme_aux, sample_type, name, era=None):
         return f"{jme_aux.source}/{jme_full_version}/{jme_full_version}_{name}_{jme_aux.jet_type}.txt"
 
     # external files
-    json_mirror = "/afs/cern.ch/user/m/mrieger/public/mirrors/jsonpog-integration-dfd90038"
+    json_mirror = "/afs/cern.ch/user/m/mrieger/public/mirrors/jsonpog-integration-9ea86c4c"
     cfg.x.external_files = DotDict.wrap({
         # jet energy correction
         "jet_jerc": (f"{json_mirror}/POG/JME/{year}{corr_postfix}_UL/jet_jerc.json.gz", "v1"),

hbw/config/datasets.py

@@ -46,7 +46,7 @@ def add_hbw_datasets(config: od.Config, campaign: od.Campaign):
         "st_twchannel_t_powheg",
         "st_twchannel_tbar_powheg",
         "st_schannel_lep_amcatnlo",
-        # "st_schannel_had_amcatnlo",  # NOTE: this dataset produces some weird errors, so skip it for now
+        "st_schannel_had_amcatnlo",
         # WJets
         "w_lnu_ht70To100_madgraph",
         "w_lnu_ht100To200_madgraph",

hbw/config/defaults_and_groups.py

@@ -202,7 +202,7 @@ def set_config_defaults_and_groups(config_inst):
         "default": ["Lepton", "VetoLepton", "Jet", "Bjet", "Trigger"],
         "thesis": ["Lepton", "Muon", "Jet", "Trigger", "Bjet"],  # reproduce master thesis cuts for checks
         "test": ["Lepton", "Jet", "Bjet"],
-        "dilep": ["Jet", "Bjet", "Lepton", "Trigger", "TriggerAndLep"],
+        "dilep": ["Jet", "Bjet", "Lepton", "Trigger"],
     }
 
     # plotting settings groups
@@ -257,12 +257,16 @@ def set_config_defaults_and_groups(config_inst):
         "SR": ("cat_1e_ggHH_kl_1_kt_1_sl_hbbhww", "cat_1mu_ggHH_kl_1_kt_1_sl_hbbhww"),
         "vbfSR": ("cat_1e_qqHH_CV_1_C2V_1_kl_1_sl_hbbhww", "cat_1mu_qqHH_CV_1_C2V_1_kl_1_sl_hbbhww"),
         "BR": ("cat_1e_tt", "cat_1e_st", "cat_1e_v_lep", "cat_1mu_tt", "cat_1mu_st", "cat_1mu_v_lep"),
+        "SR_dl": ("cat_2e_ggHH_kl_5_kt_1_dl_hbbhww", "cat_2mu_ggHH_kl_5_kt_1_dl_hbbhww"),
+        "BR_dl": ("cat_2e_t_bkg", "cat_2e_v_lep", "cat_2mu_t_bkg", "cat_2mu_v_lep"),
     }
 
     config_inst.x.default_bins_per_category = {
         "SR": 10,
         "vbfSR": 5,
         "BR": 3,
+        # "SR_dl": 10,
+        # "BR_dl": 3,
         # "cat_1e_ggHH_kl_1_kt_1_sl_hbbhww": 10,
         # "cat_1e_tt": 3,
         # "cat_1e_st": 3,
@@ -277,6 +281,8 @@ def set_config_defaults_and_groups(config_inst):
         "SR": ("ggHH_kl_1_kt_1_sl_hbbhww", "qqHH_CV_1_C2V_1_kl_1_sl_hbbhww"),
         "vbfSR": ("ggHH_kl_1_kt_1_sl_hbbhww", "qqHH_CV_1_C2V_1_kl_1_sl_hbbhww"),
         "BR": lambda proc_name: "hbbhww" not in proc_name,
+        # "SR_dl": ("ggHH_kl_5_kt_1_dl_hbbhww",),
+        # "BR_dl": lambda proc_name: "hbbhww" not in proc_name,
         # "cat_1e_ggHH_kl_1_kt_1_sl_hbbhww": ("ggHH_kl_1_kt_1_sl_hbbhww", "qqHH_CV_1_C2V_1_kl_1_sl_hbbhww"),
         # "cat_1e_tt": lambda proc_name: "hbbhww" not in proc_name,
         # "cat_1e_st": lambda proc_name: "hbbhww" not in proc_name,

hbw/config/dl/variables.py

@@ -11,9 +11,11 @@
 np = maybe_import("numpy")
 ak = maybe_import("awkward")
 
-from columnflow.columnar_util import EMPTY_FLOAT  # noqa
+from hbw.config.styling import default_var_binning, default_var_unit
+from hbw.util import call_once_on_config
 
 
+@call_once_on_config()
 def add_dl_variables(config: od.Config) -> None:
     # bjet features
     config.add_variable(
@@ -99,3 +101,156 @@ def add_dl_variables(config: od.Config) -> None:
         binning=(40, 0, 3),
         x_title=r"$ \Delta \phi(ll,jj)$",
     )
+
+
+@call_once_on_config()
+def add_dl_ml_variables(config: od.Config) -> None:
+    """
+    Adds ML input variables to a *config*.
+    """
+
+    # reconstructed variables
+    config.add_variable(
+        name="mli_ht",
+        expression="mli_ht",
+        binning=(40, 0, 1200),
+        unit="GeV",
+        x_title="HT",
+    )
+    config.add_variable(
+        name="mli_n_jet",
+        expression="mli_n_jet",
+        binning=(11, -0.5, 10.5),
+        x_title="Number of jets",
+    )
+    config.add_variable(
+        name="mli_n_deepjet",
+        expression="mli_n_deepjet",
+        binning=(11, -0.5, 10.5),
+        x_title="Number of b-tagged jets (deepjet medium WP)",
+    )
+    config.add_variable(
+        name="mli_deepjetsum",
+        expression="mli_deepjetsum",
+        binning=(40, 0, 4),
+        x_title="sum of deepjet scores",
+    )
+    config.add_variable(
+        name="mli_b_deepjetsum",
+        expression="mli_b_deepjetsum",
+        binning=(40, 0, 4),
+        x_title="sum of bjet deepjet scores",
+    )
+    config.add_variable(
+        name="mli_dr_bb",
+        expression="mli_dr_bb",
+        binning=(40, 0, 8),
+        x_title=r"$\Delta R(b,b)$",
+    )
+    config.add_variable(
+        name="mli_dphi_bb",
+        expression="mli_dphi_bb",
+        binning=(40, 0, 3.2),
+        x_title=r"$\Delta\Phi(b,b)$",
+    )
+    config.add_variable(
+        name="mli_mbb",
+        expression="mli_mbb",
+        binning=(40, 0, 400),
+        unit="GeV",
+        x_title=r"m(b,b)",
+    )
+    config.add_variable(
+        name="mli_mindr_lb",
+        expression="mli_mindr_lb",
+        binning=(40, 0, 8),
+        x_title=r"min $\Delta R(l,b)$",
+    )
+    config.add_variable(
+        name="mli_dphi_bb_nu",
+        expression="mli_dphi_bb_nu",
+        binning=(40, 0, 3.2),
+        x_title=r"$\Delta\Phi(bb,\nu)$",
+    )
+    config.add_variable(
+        name="mli_dr_bb_l",
+        expression="mli_dr_bb_l",
+        binning=(40, 0, 8),
+        x_title=r"$\Delta R(bb,l)$",
+    )
+    config.add_variable(
+        name="mli_mll",
+        expression="mli_mll",
+        binning=(40, 0, 80),
+        x_title=r"$m_{ll}$",
+    )
+    config.add_variable(
+        name="mli_dr_ll",
+        expression="mli_dr_ll",
+        binning=(40, 0, 8),
+        x_title=r"$\Delta R(ll)$",
+    )
+    config.add_variable(
+        name="mli__min_dr_llbb",
+        expression="mli_min_dr_llbb",
+        binning=(40, 0, 8),
+        x_title=r"$\Delta R(bb,ll)$",
+    )
+    config.add_variable(
+        name="mli_bb_pt",
+        expression="mli_bb_pt",
+        binning=(40, 0, 500),
+        unit="GeV",
+        x_title=r"$bb_p_T$",
+    )
+    config.add_variable(
+        name="mli_mllMET",
+        expression="mli_mllMET",
+        binning=(40, 0, 200),
+        x_title=r"$m_{llMET}$",
+    )
+    config.add_variable(
+        name="mli_dr_bb_llMET",
+        expression="mli_dr_bb_llMET",
+        binning=(40, 0, 8),
+        x_title=r"$\Delta R(bb,llMET)$",
+    )
+    config.add_variable(
+        name="mli_dphi_bb_llMET",
+        expression="mli_dphi_bb_llMET",
+        binning=(40, 0, 8),
+        x_title=r"$\Delta \phi(bb,llMET)$",
+    )
+    config.add_variable(
+        name="mli_mbbllMET",
+        expression="mli_mbbllMET",
+        binning=(40, 0, 500),
+        unit="GeV",
+        x_title=r"$m_{bbllMET}$",
+    )
+    config.add_variable(
+        name="mli_dphi_ll",
+        expression="mli_dphi_ll",
+        binning=(40, 0, 8),
+        unit="GeV",
+        x_title=r"$\Delta \phi_{ll}$",
+    )
+    config.add_variable(
+        name="mli_ll_pt",
+        expression="mli_ll_pt",
+        binning=(40, 0, 200),
+        unit="GeV",
+        x_title=r"$ll p_T$",
+    )
+
+    for obj in ["b1", "b2", "lep", "lep2", "met"]:
+        for var in ["pt", "eta"]:
+            if var == "eta" and obj == "met":
+                continue
+            config.add_variable(
+                name=f"mli_{obj}_{var}",
+                expression=f"mli_{obj}_{var}",
+                binning=default_var_binning[var],
+                unit=default_var_unit.get(var, var),
+                x_title="{obj} {var}".format(obj=obj, var=var),
+            )

hbw/config/ml_variables.py

@@ -201,6 +201,12 @@ def add_ml_variables(config: od.Config) -> None:
         log_x=True,
         x_title=r"$S_{min}$",
     )
+    config.add_variable(
+        name="mli_mindr_jj",
+        expression="mli_mindr_jj",
+        binning=(40, 0, 8),
+        x_title=r"min $\Delta R(j,j)$",
+    )
     config.add_variable(
         name="mli_vbf_deta",
         expression="mli_vbf_deta",
@@ -221,6 +227,16 @@ def add_ml_variables(config: od.Config) -> None:
         x_title="existence of at least two vbf jets = 1, else 0",
     )
 
+    for obj in ["b1", "b2", "j1", "j2"]:
+        for var in ["btagDeepFlavB"]:
+            config.add_variable(
+                name=f"mli_{obj}_{var}",
+                expression=f"mli_{obj}_{var}",
+                binning=default_var_binning[var],
+                unit=default_var_unit.get(var, var),
+                x_title="{obj} {var}".format(obj=obj, var=var),
+            )
+
     for obj in ["b1", "b2", "j1", "j2", "lep", "met"]:
         for var in ["pt", "eta"]:
             config.add_variable(

hbw/config/processes.py

@@ -84,3 +84,30 @@ def add_hbw_processes(config: od.Config, campaign: od.Campaign):
     )
     v_lep.add_process(config.get_process("w_lnu"))
     v_lep.add_process(config.get_process("dy_lep"))
+
+    # Custom t_bkg process for ML Training, combining tt+st
+    t_bkg = config.add_process(
+        name="t_bkg",
+        id=97842611,  # random number
+        xsecs={13: config.get_process("tt").get_xsec(13) + config.get_process("st").get_xsec(13)},
+        label="tt + st",
+    )
+    t_bkg.add_process(config.get_process("tt"))
+    t_bkg.add_process(config.get_process("st"))
+
+    if config.has_tag("is_dl") and config.has_tag("is_nonresonant"):
+        # Custom signal  process for ML Training, combining multiple kl signal samples
+        signal_processes = [
+            config.get_process(f"ggHH_kl_{kl}_kt_1_dl_hbbhww")
+            for kl in [0, 1, "2p45"]
+        ]
+        sig = config.add_process(
+            name="sig",
+            id=75835213,  # random number
+            xsecs={
+                13: sum([proc.get_xsec(13) for proc in signal_processes]),
+            },
+            label="signal",
+        )
+        for proc in signal_processes:
+            sig.add_process(proc)

hbw/config/styling.py

@@ -24,6 +24,7 @@
     "v_lep": "#4daf4a",  # green
     "higgs": "#984ea3",  # purple
     "st": "#ff7f00",  # orange
+    "t_bkg": "#e41a1c",  # orange
     "dy_lep": "#ffff33",  # yellow
     "ttV": "#a65628",  # brown
     "VV": "#f781bf",  # pink
@@ -64,6 +65,7 @@
     "w_lnu": "W",
     "dy_lep": "DY",
     "v_lep": "W+DY",
+    "t_bkg": "tt+st",
 }