Merge branch 'main' of https://github.com/rkansal47/HHbbVV into main

inference_scans/README.md

@@ -16,12 +16,12 @@ mkdir ~/work/HH
 
 2. Setup inference as follows
   - Set up the environment in a new folder with a clean environment (no `cmsenv`, ETC.)
-  - Use `some_name` as the name of the environment.
+  - `v1` is the name of the environment here.
 
 ```bash
 git clone ssh://[email protected]:7999/hh/tools/inference.git
 cd inference
-source setup.sh c1
+source setup.sh v1
 ```
 
 TODO: update this with better datacard dirs and setting up card repos.
@@ -32,6 +32,7 @@ export DHI_DATACARDS_RUN2="/afs/cern.ch/user/r/rkansal/work/hh/datacards_run2"
 export DHI_DATA="/afs/cern.ch/user/r/rkansal/work/hh/inference/data"
 export DHI_STORE="/eos/user/r/rkansal/bbVV/inference"
 export Cbbww4q="bbww_hadronic/v1"
+export Cbbww4qInject="$Cbbww4q<$DHI_DATACARDS_RUN2/$Cbbww4q/inject.json"
 export VERSION="dev"
 export UNBLINDED="False"
 ```

inference_scans/nuisance_renames.py

@@ -0,0 +1,11 @@
+def rename_nuisance(nuisance: str):
+    if nuisance.startswith("CMS_bbWW_hadronic_"):
+        nuisance = nuisance.split("CMS_bbWW_hadronic_")[1]
+
+    if nuisance.startswith("tf_dataResidual_bbFatJetParticleNetMass_"):
+        return "TF_" + nuisance.split("tf_dataResidual_bbFatJetParticleNetMass_")[1]
+
+    if nuisance.startswith("lp_sf"):
+        return "Lund plane SF"
+
+    return nuisance

inference_scans/run_impacts.sh

@@ -0,0 +1,22 @@
+#!/usr/bin/env bash
+
+law run PlotPullsAndImpacts \
+    --version "$VERSION" \
+    --datacards "$Cbbww4qInject" \
+    --file-types "pdf,png" \
+    --pois r \
+    --PullsAndImpacts-workflow "htcondor" \
+    --PullsAndImpacts-tasks-per-job 10 \
+    --PullsAndImpacts-custom-args="--rMin -40 --rMax 40" \
+    --parameters-per-page 40 \
+    --order-by-impact \
+    --labels "nuisance_renames.py" \
+    --skip-parameters "*dataResidual_Bin*" \
+    --campaign run2 \
+    --page 0 \
+    --pull-range 3 \
+    --unblinded $UNBLINDED \
+    --remove-output 0,a,y
+    # --print-command 2 \
+    # --use-snapshot True \  # for (fit to data) after unblinding
+    # --Snapshot-workflow "local" \

inference_scans/run_upper_limits_point_r.sh

@@ -3,7 +3,7 @@
 # export DHI_CMS_POSTFIX="Supplementary"
 law run PlotUpperLimitsAtPoint \
     --version "$VERSION" \
-    --multi-datacards "$Cbbww4q" \
+    --multi-datacards "$Cbbww4qInject" \
     --datacard-names "bbVV" \
     --file-types "pdf,png" \
     --pois r \
@@ -16,5 +16,5 @@ law run PlotUpperLimitsAtPoint \
     --h-lines 1 \
     --save-hep-data True \
     --remove-output 0,a,y
-    # --use-snapshot True \
+    # --use-snapshot True \  # for (fit to data) after unblinding
     # --Snapshot-workflow "local" \

inference_scans/run_upper_limits_r_1d_c2v.sh

@@ -3,7 +3,7 @@
 # export DHI_CMS_POSTFIX="Supplementary"
 law run PlotUpperLimits \
     --version "$VERSION" \
-    --datacards "$Cbbww4q" \
+    --datacards "$Cbbww4qInject" \
     --xsec fb \
     --pois r \
     --scan-parameters C2V,-1,3,10 \
@@ -19,6 +19,6 @@ law run PlotUpperLimits \
     --remove-output 0,a,y \
     --save-hep-data True \
     --frozen-groups signal_norm_xsbr
-    # --use-snapshot True \  # for after unblinding
+    # --use-snapshot True \  # for (fit to data) after unblinding
     # --Snapshot-workflow "local" \
     # --scan-parameters kl,-3,5,8 \

inference_scans/run_upper_limits_r_1d_kl.sh

@@ -3,10 +3,10 @@
 # export DHI_CMS_POSTFIX="Supplementary"
 law run PlotUpperLimits \
     --version "$VERSION" \
-    --datacards "$Cbbww4q" \
+    --datacards "$Cbbww4qInject" \
     --xsec fb \
     --pois r \
-    --scan-parameters kl,-30,-12,7:kl,-7,-5,3:kl:kl,-2,7,19:kl,10,30,6 \
+    --scan-parameters kl,-30,-7,4:kl,-6,7,14:kl,15,30,4 \
     --UpperLimits-workflow "htcondor" \
     --UpperLimits-tasks-per-job 1 \
     --file-types "png,pdf" \
@@ -19,5 +19,6 @@ law run PlotUpperLimits \
     --remove-output 0,a,y \
     --save-hep-data True \
     --frozen-groups signal_norm_xsbr
-    # --use-snapshot True \  # for after unblinding
+    # --scan-parameters kl,-30,-12,7:kl,-7,-5,3:kl,-2,7,19:kl,10,30,6 \
+    # --use-snapshot True \  # for (fit to data) after unblinding
     # --Snapshot-workflow "local" \

paper/latex_tables.ipynb

@@ -124,7 +124,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -315,7 +315,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.8"
+   "version": "3.10.11"
   },
   "orig_nbformat": 4,
   "vscode": {

paper/samples/MC_bg.yaml

@@ -34,7 +34,7 @@ samples:
     datasets:
       TTToHadronic         : /TTToHadronic_TuneCP5_13TeV-powheg-pythia8/RunIISummer20UL17MiniAODv2-106X_mc2017_realistic_v9-v2/MINIAODSIM
       TTToSemiLeptonic     : /TTToSemiLeptonic_TuneCP5_13TeV-powheg-pythia8/RunIISummer20UL17MiniAODv2-106X_mc2017_realistic_v9-v1/MINIAODSIM
-      TTToSemiLeptonic_ext1: /TTToSemiLeptonic_TuneCP5_13TeV-powheg-pythia8/RunIISummer20UL17MiniAODv2-PUForMUOVal_106X_mc2017_realistic_v9_ext1-v2/MINIAODSIM
+      # TTToSemiLeptonic_ext1: /TTToSemiLeptonic_TuneCP5_13TeV-powheg-pythia8/RunIISummer20UL17MiniAODv2-PUForMUOVal_106X_mc2017_realistic_v9_ext1-v2/MINIAODSIM
 
   SingleTop:
     datasets: 

paper/tables/datasets_mcbg.tex

@@ -34,8 +34,7 @@
  & ST\_tW\_antitop\_5f\_inclusiveDecays\_TuneCP5\_13TeV-powheg-pythia8 & 32.51 \\
  & ST\_tW\_top\_5f\_inclusiveDecays\_TuneCP5\_13TeV-powheg-pythia8 & 32.45 \\
 \hline
-\multirow{3}{*}{TTbar} & TTToHadronic\_TuneCP5\_13TeV-powheg-pythia8 & 303.85 \\
- & TTToSemiLeptonic\_TuneCP5\_13TeV-powheg-pythia8 & 306.14 \\
+\multirow{2}{*}{TTbar} & TTToHadronic\_TuneCP5\_13TeV-powheg-pythia8 & 303.85 \\
  & TTToSemiLeptonic\_TuneCP5\_13TeV-powheg-pythia8 & 306.14 \\
 \hline
 \multirow{6}{*}{V+Jets} & WJetsToQQ\_HT-400to600\_TuneCP5\_13TeV-madgraphMLM-pythia8 & 277.00 \\

src/HHbbVV/postprocessing/InferenceAnalysis.ipynb

@@ -57,8 +57,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# plot_dir = f\"{MAIN_DIR}/plots/TaggerAnalysis/23Apr28\"\n",
-    "plot_dir = f\"{MAIN_DIR}/plots/BDT/23May12\"\n",
+    "plot_dir = f\"{MAIN_DIR}/plots/TaggerAnalysis/23Aug28\"\n",
+    "# plot_dir = f\"{MAIN_DIR}/plots/BDT/23May12\"\n",
     "os.system(f\"mkdir -p {plot_dir}\")\n",
     "\n",
     "samples_dir = f\"{MAIN_DIR}/../data/skimmer/Feb24\"\n",
@@ -109,10 +109,11 @@
     "\n",
     "for mX, mY in res_mps:\n",
     "    res_samples[\n",
-    "        f\"X[{mX}]->H(bb)Y[{mY}](VV)\"\n",
+    "        f\"X[{mX}]->H(bb)Y[{mY}](WW)\"\n",
     "    ] = f\"NMSSM_XToYH_MX{mX}_MY{mY}_HTo2bYTo2W_hadronicDecay\"\n",
     "\n",
-    "res_sig_keys = list(res_samples.keys())"
+    "res_sig_keys = list(res_samples.keys())\n",
+    "nonres_sig_keys = [\"HHbbVV\"]"
    ]
   },
   {
@@ -368,12 +369,12 @@
     "        \"colour\": \"orange\",\n",
     "    },\n",
     "    \"thvv4q\": {\n",
-    "        \"title\": \"ParT MD THWW\",\n",
+    "        \"title\": \"ParT MD THVV\",\n",
     "        \"score_label\": \"ak8FatJetParTMD_THWW4q\",\n",
     "        \"colour\": \"green\",\n",
     "    },\n",
     "    \"thvv4qt\": {\n",
-    "        \"title\": \"ParT MD THWW\",\n",
+    "        \"title\": \"ParT MD THVV\",\n",
     "        \"score_label\": \"ak8FatJetParTMD_THWWvsT\",\n",
     "        \"colour\": \"green\",\n",
     "    },\n",
@@ -455,6 +456,7 @@
     "                np.zeros(int(np.ceil(np.sum(np.concatenate(bg_cuts)) / bg_skip))),\n",
     "            ]\n",
     "        )\n",
+    "        # print(y_true[np.sum(sig_cut):])\n",
     "\n",
     "        weights = np.concatenate(\n",
     "            [events_dict[sig_key][\"weight\"][sig_cut]]\n",
@@ -463,6 +465,7 @@
     "                for bg_key, bg_cut in zip(bg_keys, bg_cuts)\n",
     "            ],\n",
     "        )\n",
+    "        # print(weights[np.sum(sig_cut):])\n",
     "\n",
     "        for t, pvars in plot_vars.items():\n",
     "            score_label = pvars[\"score_label\"]\n",
@@ -473,6 +476,7 @@
     "                    for bg_key, bg_cut in zip(bg_keys, bg_cuts)\n",
     "                ],\n",
     "            )\n",
+    "            # print(scores[np.sum(sig_cut):])\n",
     "            fpr, tpr, thresholds = roc_curve(y_true, scores, sample_weight=weights)\n",
     "            rocs[cutstr][sig_key][t] = {\n",
     "                \"fpr\": fpr,\n",
@@ -522,7 +526,7 @@
     "]\n",
     "\n",
     "sig_splits = [\n",
-    "    [\"HHbbVV\"] + [f\"X[{mX}]->H(bb)Y[{mY}](VV)\" for (mX, mY) in mps] for mps in sig_split_points\n",
+    "    [\"HHbbVV\"] + [f\"X[{mX}]->H(bb)Y[{mY}](WW)\" for (mX, mY) in mps] for mps in sig_split_points\n",
     "]"
    ]
   },
@@ -563,7 +567,7 @@
     "    #     \"colour\": \"green\",\n",
     "    # },\n",
     "    \"thvv4qt\": {\n",
-    "        \"title\": \"ParT MD THWW\",\n",
+    "        \"title\": \"ParT MD THVV\",\n",
     "        \"score_label\": \"ak8FatJetParTMD_THWWvsT\",\n",
     "        \"colour\": \"green\",\n",
     "    },\n",
@@ -588,7 +592,7 @@
     "                    color=c,\n",
     "                )\n",
     "                for th in plot_thresholds:\n",
-    "                    idx = find_nearest(thresholds, th)\n",
+    "                    idx = find_nearest(roc[\"thresholds\"], th)\n",
     "                    pths[th][0].append(roc[\"tpr\"][idx])\n",
     "                    pths[th][1].append(roc[\"fpr\"][idx])\n",
     "\n",
@@ -605,7 +609,7 @@
     "                    *pths[th],\n",
     "                    marker=\"o\",\n",
     "                    s=40,\n",
-    "                    label=f\"THWW > {th:.2f}\",\n",
+    "                    label=f\"THVV > {th:.2f}\",\n",
     "                    color=th_colours[k],\n",
     "                    zorder=100,\n",
     "                )\n",

src/HHbbVV/postprocessing/hh_vars.py

@@ -178,7 +178,7 @@
 ]
 
 for mX, mY in res_mps:
-    res_samples[f"X[{mX}]->H(bb)Y[{mY}](VV)"] = f"NMSSM_XToYHTo2W2BTo4Q2B_MX-{mX}_MY-{mY}"
+    res_samples[f"X[{mX}]->H(bb)Y[{mY}](WW)"] = f"NMSSM_XToYHTo2W2BTo4Q2B_MX-{mX}_MY-{mY}"
 
 res_sig_keys = list(res_samples.keys())
 

src/HHbbVV/processors/corrections.py

@@ -185,6 +185,8 @@ def add_systs(systlist, qcdcorr, ewkcorr, vpt):
 def add_ps_weight(weights, ps_weights):
     """
     Parton Shower Weights (FSR and ISR)
+    "Default" variation: https://twiki.cern.ch/twiki/bin/view/CMS/HowToPDF#Which_set_of_weights_to_use
+    i.e. scaling ISR up and down
     """
 
     nweights = len(weights.weight())