From d33a96c8474fe85ccb425fc4d10d24416738db03 Mon Sep 17 00:00:00 2001
From: Sebastien Wertz <sebastien.wertz@cern.ch>
Date: Tue, 16 Aug 2022 20:21:41 +0200
Subject: [PATCH] add plotter to retrieve normalization effect of theory
 systematics at ttB level

---
 python/getTtBNormalization.py | 88 +++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)
 create mode 100644 python/getTtBNormalization.py

diff --git a/python/getTtBNormalization.py b/python/getTtBNormalization.py
new file mode 100644
index 0000000..a5f5732
--- /dev/null
+++ b/python/getTtBNormalization.py
@@ -0,0 +1,88 @@
+import os
+
+import logging
+logger = logging.getLogger("ttbb plotter")
+
+from bamboo.plots import Plot
+from bamboo.plots import EquidistantBinning as EqBin
+from bamboo import treefunctions as op
+
+import utils
+import HistogramTools as HT
+from baseTtbbPlotter import baseTtbbPlotter
+
+class getTtBNormalization(baseTtbbPlotter):
+    """"""
+    def __init__(self, args):
+        super().__init__(args)
+
+    def customizeAnalysisCfg(self, analysisCfg):
+        self._loadSampleLists(analysisCfg)
+        samples = analysisCfg["samples"]
+
+        # only work with the ttB contribution
+        for smp in list(samples.keys()):
+            if samples[smp].get("subprocess") != "ttB":
+                samples.pop(smp)
+
+        super().customizeAnalysisCfg(analysisCfg)
+
+    def definePlots(s, t, noSel, sample=None, sampleCfg=None):
+        plots = []
+        plots.append(Plot.make1D("yields", op.c_float(0.), noSel, EqBin(1, 0, 1)))
+        return plots
+
+    def postProcess(self, taskList, config=None, workdir=None, resultsdir=None, **kwargs):
+        super().postProcess(taskList, 
+                            config=config,
+                            workdir=workdir, 
+                            resultsdir=resultsdir,
+                            **kwargs)
+
+        def _find_hists(base_name, ufhandle):
+            return map(lambda k: k.split(";")[0],  # careful: uproot keys include cycle number ';1'
+                        filter(lambda k: k.startswith(base_name), ufhandle))
+
+        def _getSyst(name):
+            spl = name.split("__")
+            if len(spl) == 1:
+                return ""
+            else:
+                return spl[1]
+
+        import uproot
+        hist_list = []
+        for task in taskList:
+            if "syst" in task.config:
+                continue
+            signal_tag = task.config["signal_tag"]
+            xs = task.config["cross-section"]
+            outfile = os.path.join(resultsdir, task.outputFile)
+            tf = HT.openFileAndGet(outfile)
+            nomWeights = utils.getSumw(tf, task.config, self.readCounters)
+            tf.Close()
+            with uproot.open(outfile) as ufhandle:
+                hists = _find_hists("yields", ufhandle)
+                for h in hists:
+                    systVar = _getSyst(h)
+                    h = ufhandle[h].values()#.to_hist()
+                    sumw = xs * h.sum() / nomWeights
+                    hist_list.append({"signal_tag": signal_tag, "syst": systVar, "sumw": sumw})
+
+        import pandas as pd
+        df = pd.DataFrame(hist_list)
+        results = {}
+        for signal_tag, sig_df in df.groupby("signal_tag"):
+            nom_df = sig_df[sig_df.syst == ""]
+            var_df = sig_df[sig_df.syst != ""]
+            nom_sumw = nom_df["sumw"].sum()
+            results[signal_tag] = {}
+            for systVar, syst_df in var_df.groupby("syst"):
+                ratio = syst_df["sumw"].sum() / nom_sumw
+                systVar = systVar.replace("up", "Up").replace("down", "Down")
+                results[signal_tag][systVar] = float(ratio)
+
+        import yaml
+        with open(os.path.join(resultsdir, "ttB_systematics.yaml"), 'w') as f:
+            yaml.dump(results, f, default_flow_style=False, indent=4)
+
-- 
GitLab