Improve yuma config

changelog.md

@@ -1,7 +1,8 @@
 # Changelog
 
 ### [Latest]
-- Autoinstall deps and steamline yuma config[!236](https://github.com/umami-hep/puma/pull/236)
+- More improvements to Yuma configuration [!237](https://github.com/umami-hep/puma/pull/237)
+- Autoinstall deps and steamline yuma config [!236](https://github.com/umami-hep/puma/pull/236)
 
 
 ### [v0.3.2] (2024/02/13)

docs/source/examples/yuma.md

@@ -15,8 +15,7 @@ Additional arguments can be included:
 
 ## taggers.yaml
 
-The ```taggers.yaml``` file contains required information for taggers we wish to load. A section 'tagger_defaults' can be used to include default values, such as f_c, f_b, or cuts. For all taggers that are loaded, these values will be used by default, unless overwritten by the tagger.
-
+The ```taggers.yaml``` file contains required information for taggers we wish to load.
 Under 'taggers' are the defined taggers. Each should be assigned a tag, which is used as the 'name' of the tagger and its settings. Within in tagger, arguments that can be parsed to the Tagger high-level class can be used.
 
 ## plot_cfg.yaml

examples/plt_cfg.yaml

@@ -1,22 +1,24 @@
 plot_dir: /home/output_dir
 timestamp: False
 
+sample_path: dummy/path/1
+
 results_config:
+  sample: ttbar
+  atlas_first_tag: Simulation Internal
   atlas_second_tag: $\sqrt{s} = 13.6$ TeV, MC23
-  remove_nan: True # for plotting eff profiles for variables with NaNs (e.g, Lxy)
+  atlas_third_tag: $t\overline{t}$ 20 < $p_T$ < 250 GeV, $|\eta| < 2.5$
   num_jets: 100_000
-
-taggers_config: !include taggers.yaml
-taggers: [dummy1, dummy2, dummy3]
-
-sample:
-  sample: ttbar
+  remove_nan: True # for plotting eff profiles for variables with NaNs (e.g, Lxy)
   global_cuts:
-    - "pt > 20000"
-    - "pt < 250000"
+    - "pt > 20_000"
+    - "pt < 250_000"
     - "eta < 2.5"
     - "eta > -2.5"
-  tag: $t\overline{t}$ 20 < $p_T$ < 250 GeV, $|\eta| < 2.5$
+    - "n_truth_promptLepton == 0"
+
+taggers_config: !include taggers.yaml
+taggers: [dummy1, dummy2, dummy3]
 
 roc_plots:
   -  args:

examples/taggers.yaml

@@ -1,20 +1,14 @@
-tagger_defaults:
+dummy1: # Uses default sample path
+  label: Dummy default
+  f_c: 0.2
+  f_b: 0.05
+  reference: True
+dummy2: # Uses default sample path
+  label: Dummy $f_c$ = 0.2
+  f_c: 0.2
+  f_b: 0.05
+dummy3: # Uses updated sample path
+  sample_path: /dummy/path
+  label: Dummy 3
+  f_c: 0.2
   f_b: 0.05
-  f_c: 0.1
-  sample_path: dummy/path/1
-  cuts:
-    - "n_truth_promptLepton == 0"
-
-taggers:
-  # Uses default sample path, and default f_c
-  dummy1:
-    label: Dummy default
-    reference: True
-  # Uses default sample path, and updated f_c
-  dummy2:
-    f_c: 0.2
-    label: Dummy $f_c$ = 0.2
-  # Uses updated sample path, and default f_c
-  dummy3:
-    sample_path: /dummy/path
-    label: Dummy 3

puma/hlplots/configs.py

@@ -1,6 +1,6 @@
 from __future__ import annotations
 
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from datetime import datetime
 from pathlib import Path
 
@@ -17,23 +17,18 @@
     config_path: Path
     plot_dir: Path
 
-    taggers_config: dict
-    sample: dict
-
     results_config: dict[str, dict[str, str]]
+    taggers_config: dict
     taggers: list[str] | list[Tagger] | None = None
-    timestamp: bool = True
 
-    roc_plots: dict[str, dict] = None
-    fracscan_plots: dict[str, dict] = None
-    disc_plots: dict[str, dict] = None
-    prob_plots: dict[str, dict] = None
-    eff_vs_var_plots: dict[str, dict] = None
-
-    signal: str = None
+    timestamp: bool = True
+    sample_path: Path = None
 
-    results: Results = None
-    default_second_atlas_tag: str = None
+    roc_plots: dict[str, dict] = field(default_factory=dict)
+    fracscan_plots: dict[str, dict] = field(default_factory=dict)
+    disc_plots: dict[str, dict] = field(default_factory=dict)
+    prob_plots: dict[str, dict] = field(default_factory=dict)
+    eff_vs_var_plots: dict[str, dict] = field(default_factory=dict)
 
     def __post_init__(self):
         # Define a plot directory based on the plot config file name, and a date time
@@ -43,27 +38,8 @@
             plot_dir_name += "_" + date_time_file
         self.plot_dir_final = Path(self.plot_dir) / plot_dir_name
 
-        tagger_defaults = self.taggers_config.get("tagger_defaults", {})
-        taggers = self.taggers_config.get("taggers", {})
-
-        if self.taggers is None:
-            self.taggers = list(taggers.keys())
-
-        self.taggers = {
-            k: {
-                **tagger_defaults,
-                **t,
-                "yaml_name": k,
-            }
-            for k, t in taggers.items()
-            if k in self.taggers
-        }
-
-        self.roc_plots = self.roc_plots or {}
-        self.fracscan_plots = self.fracscan_plots or {}
-        self.disc_plots = self.disc_plots or {}
-        self.prob_plots = self.prob_plots or {}
-        self.eff_vs_var_plots = self.eff_vs_var_plots or {}
+        for k, kwargs in self.taggers_config.items():
+            kwargs["yaml_name"] = k
 
     @classmethod
     def load_config(cls, path: Path) -> PlotConfig:
@@ -81,24 +57,26 @@
         file, and adds them.
         """
         kwargs = self.results_config
-        tag = kwargs.get("atlas_second_tag", "")
-        kwargs["atlas_second_tag"] = tag + "\n" + self.sample.pop("tag", "")
-        kwargs.update(self.sample)
         kwargs["signal"] = self.signal
         kwargs["perf_vars"] = list(
             {plot["args"].get("perf_var", "pt") for plot in self.eff_vs_var_plots}
         )
 
-        # Store default tag incase other plots need to temporarily modify it
-        self.default_second_atlas_tag = kwargs["atlas_second_tag"]
-
         # Instantiate the results object
         results = Results(**kwargs)
 
+        # Store default tag incase other plots need to temporarily modify it
+        results.default_atlas_second_tag = results.atlas_second_tag
+
         good_colours = get_good_colours()
         col_idx = 0
         # Add taggers to results, then bulk load
-        for key, t in self.taggers.items():
+        for key, t in self.taggers_config.items():
+            # if the a sample is not defined for the tagger, use the default sample
+            if not self.sample_path and not t.get("sample_path", None):
+                raise ValueError(f"No sample path defined for tagger {key}")
+            if self.sample_path and not t.get("sample_path", None):
+                t["sample_path"] = self.sample_path
             # Allows automatic selection of tagger name in eval files
             t["name"] = get_tagger_name(
                 t.get("name", None), t["sample_path"], key, results.flavours

puma/hlplots/results.py

@@ -33,6 +33,7 @@ class Results:
     backgrounds: list = field(init=False)
     atlas_first_tag: str = "Simulation Internal"
     atlas_second_tag: str = None
+    atlas_third_tag: str = None
     taggers: dict = field(default_factory=dict)
     perf_vars: str | tuple | list = "pt"
     output_dir: str | Path = "."
@@ -47,6 +48,8 @@ def __post_init__(self):
             self.output_dir = Path(self.output_dir)
         if isinstance(self.perf_vars, str):
             self.perf_vars = [self.perf_vars]
+        if self.atlas_second_tag is not None and self.atlas_third_tag is not None:
+            self.atlas_second_tag = f"{self.atlas_second_tag}\n{self.atlas_third_tag}"
 
         self.plot_funcs = {
             "probs": self.plot_probs,
@@ -82,7 +85,7 @@ def flavours(self):
         """
         return self.backgrounds + [self.signal]
 
-    def add(self, tagger):
+    def add(self, tagger: Tagger):
         """Add tagger to class.
 
         Parameters
@@ -127,7 +130,6 @@ def add_taggers_from_file(
         """Load one or more taggers from a common file, and adds them to this
         results class
 
-
         Parameters
         ----------
             @self.load_taggers_from_file

puma/hlplots/tagger.py

@@ -36,8 +36,6 @@ class Tagger:
     )
     disc_cut: float = None
     working_point: float = None
-    f_c: float = None
-    f_b: float = None
 
     # Used only by YUMA
     yaml_name: str = None

puma/hlplots/yuma.py

@@ -54,10 +54,11 @@
             plt_cfg.results, eff_vs_var
         )
         plot_kwargs = get_plot_kwargs(eff_vs_var, suffix=[inc_str, perf_var])
+        # move this logic into the results class
         if not (bins := eff_vs_var["args"].get("bins", None)):
-            if plt_cfg.sample["sample"] == "ttbar":
+            if plt_cfg.results.sample == "ttbar":
                 bins = [20, 30, 40, 60, 85, 110, 140, 175, 250]
-            elif plt_cfg.sample["sample"] == "zprime":
+            elif plt_cfg.results.sample == "zprime":
                 bins = [250, 500, 750, 1000, 1500, 2000, 3000, 4000, 5500]
             else:
                 raise ValueError(
@@ -129,7 +130,7 @@
         info_str = f"$f_{frac_flav}$ scan" if frac_flav != "tau" else "$f_{\\tau}$ scan"
         # info_str += f" {round(efficiency*100)}% {plt_cfg.results.signal.label} WP"
         plt_cfg.results.atlas_second_tag = (
-            plt_cfg.default_second_atlas_tag + "\n" + info_str
+            plt_cfg.results.default_atlas_second_tag + "\n" + info_str
         )
 
         eff_str = str(round(efficiency * 100, 3)).replace(".", "p")
@@ -146,7 +147,7 @@
         plt_cfg.results.plot_fraction_scans(efficiency=efficiency, **plot_kwargs)
         plt_cfg.results.taggers = all_taggers
         plt_cfg.results.backgrounds = tmp_backgrounds
-        plt_cfg.results.atlas_second_tag = plt_cfg.default_second_atlas_tag
+        plt_cfg.results.atlas_second_tag = plt_cfg.results.default_atlas_second_tag
 
 
 def make_roc_plots(plt_cfg):

puma/tests/test_yuma.py

@@ -52,8 +52,9 @@ def testGetIncludeTaggers(self):
 
         with tempfile.TemporaryDirectory() as tmp_file:
             fpath1, file = get_mock_file(fname=(Path(tmp_file) / "file1.h5").as_posix())
-            taggers["tagger_defaults"]["sample_path"] = fpath1
-            taggers["taggers"]["dummy3"]["sample_path"] = fpath1
+            taggers["dummy1"]["sample_path"] = fpath1
+            taggers["dummy2"]["sample_path"] = fpath1
+            taggers["dummy3"]["sample_path"] = fpath1
             updated_plt_cfg = Path(tmp_file) / "plt_cfg.yaml"
 
             plt_cfg["roc_plots"][0]["reference"] = "dummyNot"
@@ -108,9 +109,9 @@ def testAllPlots(self):
         with tempfile.TemporaryDirectory() as tmp_file:
             fpath1, file = get_mock_file(fname=(Path(tmp_file) / "file1.h5").as_posix())
             fpath2, file = get_mock_file(fname=(Path(tmp_file) / "file2.h5").as_posix())
-
-            taggers["tagger_defaults"]["sample_path"] = fpath1
-            taggers["taggers"]["dummy3"]["sample_path"] = fpath2
+            taggers["dummy1"]["sample_path"] = fpath1
+            taggers["dummy2"]["sample_path"] = fpath1
+            taggers["dummy3"]["sample_path"] = fpath2
             updated_plt_cfg = Path(tmp_file) / "plt_cfg.yaml"
             plt_cfg["plot_dir"] = tmp_file + "/plots"
             plt_cfg["taggers_config"] = taggers
@@ -161,8 +162,9 @@ def testNoPlots(self):
             fpath1, file = get_mock_file(fname=(Path(tmp_file) / "file1.h5").as_posix())
             fpath2, file = get_mock_file(fname=(Path(tmp_file) / "file2.h5").as_posix())
 
-            taggers["tagger_defaults"]["sample_path"] = fpath1
-            taggers["taggers"]["dummy3"]["sample_path"] = fpath2
+            taggers["dummy1"]["sample_path"] = fpath1
+            taggers["dummy2"]["sample_path"] = fpath1
+            taggers["dummy3"]["sample_path"] = fpath2
 
             updated_plt_cfg = Path(tmp_file) / "plt_cfg.yaml"
             plt_cfg["taggers_config"] = taggers