Merge pull request #8 from MadAnalysis/improve_cutflow

Improvement on code structure

docs/releases/changelog-dev.md

@@ -8,6 +8,12 @@
   * Improve the logger output
     ([#7](https://github.com/MadAnalysis/ma5_expert/pull/7)).
 
+  * Improve CLs computer.
+    ([#8](https://github.com/MadAnalysis/ma5_expert/pull/8))
+
+  * Improved cutflow structure.
+    ([#8](https://github.com/MadAnalysis/ma5_expert/pull/8))
+
 ## Bug fixes
  * Make exceptions accessible through `ma5.system` module
     ([#7](https://github.com/MadAnalysis/ma5_expert/pull/7)).

src/ma5_expert/backend/ma5_backend.py

@@ -2,13 +2,40 @@
 import os, sys
 
 from ma5_expert.system.exceptions import MadAnalysisPath
-from typing import Optional, Text
+from typing import Optional, Text, Union
 from enum import Enum, auto
 
 
 class PADType(Enum):
-    PAD = auto()
-    PADForSFS = auto()
+    PAD = "PAD"
+    PADForSFS = "PADForSFS"
+
+    def __str__(self):
+        return self.value
+
+    def __repr__(self):
+        return self.value
+
+
+class ExpectationAssumption(Enum):
+    APRIORI = "apriori"
+    APOSTERIORI = "aposteriori"
+
+    def __str__(self):
+        return self.value
+
+    def __repr__(self):
+        return self.value
+
+    @classmethod
+    def get(cls, value):
+        if isinstance(value, ExpectationAssumption):
+            return value
+        elif isinstance(value, str):
+            if value.lower() == str(ExpectationAssumption.APRIORI):
+                return ExpectationAssumption.APRIORI
+            else:
+                return ExpectationAssumption.APOSTERIORI
 
 
 @dataclass
@@ -72,14 +99,21 @@ def __post_init__(self):
         if not self.ma5_main.session_info.has_padsfs:
             self.ma5_main.session_info.has_padsfs = self.enforce_padforsfs
 
-    def get_run_recast(self, sample_path: Text, padtype: PADType):
+    def get_run_recast(
+        self,
+        sample_path: Text,
+        padtype: PADType,
+        expectation_assumption: Union[ExpectationAssumption, Text] = ExpectationAssumption.APRIORI,
+    ):
         """
         Get run recast class
 
         Parameters
         ----------
         sample_path: Text
             PAth of the MadAnalysis workspace
+        expectation_assumption: ExpectationAssumption
+            assumption on expectation
 
         Returns
         -------
@@ -88,6 +122,10 @@ def get_run_recast(self, sample_path: Text, padtype: PADType):
         from logging import DEBUG
         from madanalysis.misc.run_recast import RunRecast
 
+        self.ma5_main.recasting.expectation_assumption = str(
+            ExpectationAssumption.get(expectation_assumption)
+        )
+
         run_recast = RunRecast(self.ma5_main, sample_path)
         if padtype == PADType.PAD:
             run_recast.pad = os.path.join(self.madanalysis_path, "tools/PAD")

src/ma5_expert/cutflow/cut.py

@@ -1,13 +1,12 @@
 from math import sqrt
-
-from ma5_expert.system.exceptions import InvalidInput
-
-from typing import List, Any, Union, Tuple, Text, Dict, Sequence, Optional
+from dataclasses import dataclass, field
+from typing import Any, Text, Optional
 import logging
 
 log = logging.getLogger("ma5_expert")
 
 
+@dataclass
 class Cut:
     """
     Cut object
@@ -34,59 +33,15 @@ class Cut:
         luminosity [fb^-1]
     """
 
-    def __init__(
-        self,
-        name: Optional[Text] = "__unknown_cut__",
-        Nentries: Optional[int] = None,
-        sumw: Optional[float] = None,
-        sumw2: Optional[float] = None,
-        previous_cut: Optional[Any] = None,
-        initial_cut: Optional[Any] = None,
-        xsec: Optional[float] = None,
-        Nevents: Optional[float] = None,
-        lumi: Optional[float] = None,
-    ):
-
-        self.id = name  # Name of the cut
-        self.Nentries = Nentries if Nentries is not None else 0  # Number of MC events
-        self._sumW = sumw  # sum of weights
-        self._sumW2 = sumw2  # sum of square of the weights
-        self._initial_cut = initial_cut
-        self._previous_cut = previous_cut
-        self._lumi = lumi
-        self._xsection = xsec
-
-        if Nevents is not None:
-            self._Nevents = Nevents
-
-    @property
-    def sumW(self):
-        """
-        Sum of weights
-        """
-        return self._sumW
-
-    @property
-    def xsec(self) -> float:
-        """
-        Cross section [pb]
-        """
-        if hasattr(self, "_xsection"):
-            return self._xsection
-        else:
-            return -1
-
-    @xsec.setter
-    def xsec(self, val: float) -> None:
-        self._xsection = xsec
-        if hasattr(self, "_Nevents"):
-            delattr(self, "_Nevents")
-
-    @property
-    def lumi(self):
-        if self._lumi is not None:
-            return self._lumi
-        return -1
+    name: Optional[Text] = "__unknown_cut__"
+    Nentries: Optional[int] = None
+    sumW: Optional[float] = None
+    sumW2: Optional[float] = None
+    _previous_cut: Optional[Any] = field(default=None, repr=False)
+    _initial_cut: Optional[Any] = field(default=None, repr=False)
+    xsec: Optional[float] = None
+    _Nevents: Optional[float] = field(default=None, repr=False)
+    lumi: float = -1.0
 
     @property
     def eff(self):
@@ -98,7 +53,7 @@ def eff(self):
                 try:
                     return self.sumW / self._initial_cut.sumW
                 except ZeroDivisionError as err:
-                    return 0.
+                    return 0.0
             else:
                 try:
                     return float(self.Nevents) / self._initial_cut.Nevents
@@ -149,7 +104,7 @@ def mc_rel_eff(self):
             try:
                 return self.Nentries / self._previous_cut.Nentries
             except ZeroDivisionError as err:
-                log.warning("Previous entry has no MC event")
+                log.warning(f"Previous entry (cut named {self._previous_cut.name}) has no MC event")
                 return 0.0
 
         return -1
@@ -159,21 +114,19 @@ def mc_unc(self) -> float:
         """
         Monte Carlo uncertainty
         """
-        if self.Nentries > 0 and self._lumi is not None:
-            return self.Nevents * sqrt(
-                self.eff * (1.0 - self.eff) / float(self.Nentries)
-            )
+        if self.Nentries > 0 and self.lumi > 0:
+            return self.Nevents * sqrt(self.eff * (1.0 - self.eff) / float(self.Nentries))
 
         return 0.0
 
     @property
     def Nevents(self) -> float:
-        if hasattr(self, "_Nevents"):
+        if self._Nevents is not None:
             return self._Nevents
         else:
             if self.lumi >= 0.0:
                 if self.xsec >= 0.0:
-                    return self.xsec * self.eff * 1000.0 * self._lumi
+                    return self.xsec * self.eff * 1000.0 * self.lumi
                 else:
                     return self.eff * self._initial_cut.Nevents
             else:
@@ -183,7 +136,7 @@ def Nevents(self) -> float:
     def __repr__(self):
         nentries = self.Nentries if self.Nentries is not None else -1
         txt = (
-            f"  * {self.id} : \n"
+            f"  * {self.name} : \n"
             + f"     - Number of Entries    : {nentries:.0f}\n"
             + f"     - Number of Events     : {self.Nevents:.3f} ± {self.mc_unc:.3f}(ΔMC)\n"
             + f"     - Cut & Rel Efficiency : {self.eff:.3f}, {self.rel_eff:.3f}\n"

src/ma5_expert/cutflow/objects.py

@@ -53,7 +53,7 @@ def xsec(self):
     @xsec.setter
     def xsec(self, val: float):
         for cut in self:
-            cut.xsec = xsec
+            cut.xsec = val
 
     @property
     def lumi(self):
@@ -62,7 +62,7 @@ def lumi(self):
     @lumi.setter
     def lumi(self, val: float):
         for cut in self:
-            cut._lumi = lumi
+            cut._lumi = val
 
     @property
     def CutNames(self):
@@ -78,11 +78,11 @@ def items(self):
         return ((ix, cut) for ix, cut in enumerate(self._data))
 
     def keys(self):
-        return (cut.id for cut in self._data)
+        return (cut.name for cut in self._data)
 
     def getCut(self, id):
         for cut in self:
-            if cut.id == id:
+            if cut.name == id:
                 return cut
 
         return None

src/ma5_expert/cutflow/reader.py

@@ -11,7 +11,7 @@
 log = logging.getLogger("ma5_expert")
 
 
-class Collection(object):
+class Collection:
     def __init__(self, cutflow_path="", saf_file=False, **kwargs):
         """
 
@@ -83,12 +83,12 @@ def _readCollection(
                         name="Initial",
                         Nentries=int(cutflow[i].split()[0])
                         + int(cutflow[i].split()[1]),
-                        sumw=float(cutflow[i + 1].split()[0])
+                        sumW=float(cutflow[i + 1].split()[0])
                         + float(cutflow[i + 1].split()[1]),
-                        sumw2=float(cutflow[i + 2].split()[0])
+                        sumW2=float(cutflow[i + 2].split()[0])
                         + float(cutflow[i + 2].split()[1]),
                         xsec=xsec,
-                        Nevents=nevents,
+                        _Nevents=nevents,
                         lumi=self.lumi,
                     )
                     currentSR.addCut(current_cut)
@@ -99,13 +99,13 @@ def _readCollection(
                         name=cutflow[i].split('"')[1],
                         Nentries=int(cutflow[i + 1].split()[0])
                         + int(cutflow[i + 1].split()[1]),
-                        sumw=float(cutflow[i + 2].split()[0])
+                        sumW=float(cutflow[i + 2].split()[0])
                         + float(cutflow[i + 2].split()[1]),
-                        sumw2=float(cutflow[i + 3].split()[0])
+                        sumW2=float(cutflow[i + 3].split()[0])
                         + float(cutflow[i + 3].split()[1]),
                         xsec=xsec,
-                        previous_cut=currentSR[-1],
-                        initial_cut=currentSR[0],
+                        _previous_cut=currentSR[-1],
+                        _initial_cut=currentSR[0],
                         lumi=self.lumi,
                     )
                     currentSR.addCut(current_cut)
@@ -196,5 +196,5 @@ def get_alive(self):
     def regiondata(self):
         regdat = {}
         for k, i in self.items():
-            regdat[k] = i.regiondata[i.id]
+            regdat[k] = i.regiondata[i.name]
         return regdat

src/ma5_expert/cutflow/table.py

@@ -159,7 +159,7 @@ def write_comparison_table(self, *args, **kwargs):
             txt += "\\\ \\hline\n"
             # write cutflow
             for cutID, cut in self.ref_sample[SR].items():
-                name = cut.id
+                name = cut.name
                 if "$" not in name:
                     name = name.replace("_", " ")
                 txt += "      " + name.ljust(40, " ") + "& "
@@ -183,7 +183,7 @@ def write_comparison_table(self, *args, **kwargs):
                     if raw:
                         txt += tmp.format(
                             scientific_LaTeX(cut.Nentries, sty=event_style),
-                            cut.raw_rel_eff,
+                            cut.mc_rel_eff,
                         )
                     else:
                         if not (MCunc and cut.Nentries > 0):
@@ -222,7 +222,7 @@ def write_comparison_table(self, *args, **kwargs):
                         if raw:
                             txt += tmp.format(
                                 scientific_LaTeX(smp[cutID].Nentries, sty=event_style),
-                                smp[cutID].raw_rel_eff,
+                                smp[cutID].mc_rel_eff,
                             )
                         else:
                             if not (MCunc and smp[cutID].Nentries > 0):
@@ -252,12 +252,15 @@ def write_comparison_table(self, *args, **kwargs):
                             + " "
                         )
                         if raw:
-                            rel_eff = abs(
-                                1 - (smp[cutID].raw_rel_eff / cut.raw_rel_eff)
-                            )
+                            try:
+                                rel_eff = abs(
+                                    1 - (smp[cutID].mc_rel_eff / cut.mc_rel_eff)
+                                )
+                            except ZeroDivisionError as err:
+                                rel_eff = -1
                             txt += tmp.format(
                                 scientific_LaTeX(smp[cutID].Nentries, sty=event_style),
-                                smp[cutID].raw_rel_eff,
+                                smp[cutID].mc_rel_eff,
                                 rel_eff * 100.0,
                             )
                         else:
@@ -420,7 +423,7 @@ def write_signal_comparison_table(self, *args, **kwargs):
                     txt += "\\\ \\hline\n"
             # write cutflow
             for cutID, cut in self.ref_sample[SR].items():
-                name = cut.id
+                name = cut.name
                 if "$" not in name:
                     name = name.replace("_", " ")
                 txt += "      " + name.ljust(40, " ") + "& "
@@ -578,12 +581,7 @@ def WriteMake(self, file, make=True):
                 + "clean:\n"
                 + "\trm -f *.aux *.log *.out *.toc *.blg *.dvi *.t1 *.1 *.mp *spl *.lol *Notes.bib\n"
             )
-            if make:
-                try:
-                    file.close()
-                    os.system("make")
-                except:
-                    print("Compilation failed.")
+
         else:
             raise ValueError("Can not find " + file.name)