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

rkansal47 · Jun 30, 2023 · 48cbcbb · 48cbcbb
2 parents af59b1e + b774bcd
commit 48cbcbb
Show file tree

Hide file tree

Showing 34 changed files with 2,855 additions and 437 deletions.
diff --git a/.gitignore b/.gitignore
@@ -61,6 +61,8 @@ src/HHbbVV/postprocessing/outs
 !paper
 !paper/**
 
+!inference_scans
+!inference_scans/**
 
 !.gitignore
 !.github

diff --git a/AnalysisPlan.md b/AnalysisPlan.md
@@ -121,8 +121,9 @@
   - [x] Trim features
   - [x] Check sculpting
 - [x] Scan Txbb, BDT WPs
-- [ ] Run over all kL and k_2V
-  - [ ] Re-run VBF with gen selection
+- [x] Run over all kL and k_2V
+  - [x] Re-run VBF with gen selection
+- [x] HH inference
 - [ ] Theory uncertainties
 
 

diff --git a/inference_scans/README.md b/inference_scans/README.md
@@ -0,0 +1,59 @@
+# HH Inference scans
+
+For HH limits, kl/c2v scans, etc. Adapted from https://github.com/LPC-HH/inference_scans.
+
+- [HH Inference scans](#hh-inference-scans)
+  - [Setup](#setup)
+
+
+## Setup
+
+1.  Make a fresh area on lxplus, for example:
+
+```bash
+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.
+
+```bash
+git clone ssh://git@gitlab.cern.ch:7999/hh/tools/inference.git
+cd inference
+source setup.sh c1
+```
+
+TODO: update this with better datacard dirs and setting up card repos.
+
+3. Add the following to the `.setups/v1.sh` file 
+```bash
+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 VERSION="dev"
+export UNBLINDED="False"
+```
+
+4. Checkout this repo outside of `inference`
+
+```bash
+cd ..
+git clone https://github.com/rkansal47/HHbbVV.git
+```
+
+
+5. Finally, from the `inference_scans` directory in `screen` or `tmux`, run the desired script:
+
+```bash
+cd inference
+source setup.sh v1 # required before running a scan
+cd ../HHbbVV/inference_scans/
+source run_likelihood_scan_2d_kl_c2v.sh
+``` 
+
+Note! Need to make persistent screens to log off lxplus while running: 
+https://hsf-training.github.io/analysis-essentials/shell-extras/persistent-screen.html
+
+Also keep track of the specific lxplus node.
diff --git a/inference_scans/run_upper_limits_point_r.sh b/inference_scans/run_upper_limits_point_r.sh
@@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+
+# export DHI_CMS_POSTFIX="Supplementary"
+law run PlotUpperLimitsAtPoint \
+    --version "$VERSION" \
+    --multi-datacards "$Cbbww4q" \
+    --datacard-names "bbVV" \
+    --file-types "pdf,png" \
+    --pois r \
+    --show-parameters kl,kt,C2V,CV \
+    --UpperLimits-workflow "htcondor" \
+    --UpperLimits-tasks-per-job 1 \
+    --x-log \
+    --campaign run2 \
+    --unblinded $UNBLINDED \
+    --h-lines 1 \
+    --save-hep-data True \
+    --remove-output 0,a,y
+    # --use-snapshot True \
+    # --Snapshot-workflow "local" \
diff --git a/inference_scans/run_upper_limits_r_1d_c2v.sh b/inference_scans/run_upper_limits_r_1d_c2v.sh
@@ -0,0 +1,24 @@
+#!/usr/bin/env bash
+
+# export DHI_CMS_POSTFIX="Supplementary"
+law run PlotUpperLimits \
+    --version "$VERSION" \
+    --datacards "$Cbbww4q" \
+    --xsec fb \
+    --pois r \
+    --scan-parameters C2V,-1,3,10 \
+    --UpperLimits-workflow "htcondor" \
+    --UpperLimits-tasks-per-job 1 \
+    --file-types "png,pdf" \
+    --campaign run2 \
+    --unblinded $UNBLINDED \
+    --y-log \
+    --show-parameters "kt,C2V,CV" \
+    --br bbww \
+    --save-ranges \
+    --remove-output 0,a,y \
+    --save-hep-data True \
+    --frozen-groups signal_norm_xsbr
+    # --use-snapshot True \  # for after unblinding
+    # --Snapshot-workflow "local" \
+    # --scan-parameters kl,-3,5,8 \
diff --git a/inference_scans/run_upper_limits_r_1d_kl.sh b/inference_scans/run_upper_limits_r_1d_kl.sh
@@ -0,0 +1,23 @@
+#!/usr/bin/env bash
+
+# export DHI_CMS_POSTFIX="Supplementary"
+law run PlotUpperLimits \
+    --version "$VERSION" \
+    --datacards "$Cbbww4q" \
+    --xsec fb \
+    --pois r \
+    --scan-parameters kl,-30,-12,7:kl,-7,-5,3:kl:kl,-2,7,19:kl,10,30,6 \
+    --UpperLimits-workflow "htcondor" \
+    --UpperLimits-tasks-per-job 1 \
+    --file-types "png,pdf" \
+    --campaign run2 \
+    --unblinded $UNBLINDED \
+    --y-log \
+    --show-parameters "kt,C2V,CV" \
+    --br bbww \
+    --save-ranges \
+    --remove-output 0,a,y \
+    --save-hep-data True \
+    --frozen-groups signal_norm_xsbr
+    # --use-snapshot True \  # for after unblinding
+    # --Snapshot-workflow "local" \
diff --git a/src/HHbbVV/combine/run_blinded.sh b/src/HHbbVV/combine/run_blinded.sh
@@ -104,6 +104,7 @@ cards_dir="./"
 ws=${cards_dir}/combined
 wsm=${ws}_withmasks
 wsm_snapshot=higgsCombineSnapshot.MultiDimFit.mH125
+CMS_PARAMS_LABEL="CMS_bbWW_hadronic"
 
 outsdir=${cards_dir}/outs
 mkdir -p $outsdir
@@ -123,8 +124,8 @@ if [ $resonant = 0 ]; then
     freezeparamsblinded=""
     for bin in {5..9}
     do
-        setparamsblinded+="CMS_bbWW_boosted_ggf_qcdparam_msdbin${bin}=0,"
-        freezeparamsblinded+="CMS_bbWW_boosted_ggf_qcdparam_msdbin${bin},"
+        setparamsblinded+="${CMS_PARAMS_LABEL}_tf_dataResidual_Bin${bin}=0,"
+        freezeparamsblinded+="${CMS_PARAMS_LABEL}_tf_dataResidual_Bin${bin},"
     done
 
     # remove last comma

diff --git a/src/HHbbVV/scale_factors/top_analysis.ipynb → src/HHbbVV/postprocessing/TopAnalysis.ipynb b/src/HHbbVV/scale_factors/top_analysis.ipynb → src/HHbbVV/postprocessing/TopAnalysis.ipynb
@@ -13,12 +13,12 @@
     "import pandas as pd\n",
     "\n",
     "# from pandas.core.common import SettingWithCopyWarning\n",
-    "from sample_labels import sig_key, data_key\n",
+    "from hh_vars import data_key\n",
     "\n",
     "# ignore these because they don't seem to apply\n",
     "# warnings.simplefilter(action=\"ignore\", category=SettingWithCopyWarning)\n",
     "\n",
-    "from PyPDF2 import PdfFileMerger\n",
+    "from PyPDF2 import PdfMerger\n",
     "\n",
     "from copy import deepcopy\n",
     "\n",
@@ -49,7 +49,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "plot_dir = \"../../../plots/ttsfs/23Feb1\"\n",
+    "plot_dir = \"../../../plots/ttsfs/23Jun22\"\n",
     "\n",
     "import os\n",
     "\n",
@@ -75,7 +75,7 @@
     "top_matched_key = \"TT Top Matched\"\n",
     "\n",
     "# data_dir = \"../../../../data/ttsfs/Dec22/\"\n",
-    "data_dir = \"../../../../data/ttsfs/Jan31/\"\n",
+    "data_dir = \"../../../../data/ttsfs/23Jun9/\"\n",
     "year = \"2018\"\n",
     "\n",
     "# filters = [(\"('ak8FatJetPt', '0')\", \">=\", 500)]\n",
@@ -90,37 +90,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "data_dir = \"../../../../data/ttsfs/Jan17/\"\n",
     "sample = \"TTToSemiLeptonic\"\n",
-    "\n",
-    "# utils.get_cutflow(f\"{data_dir}/{year}/TTToSemiLeptonic/pickles\", year, \"TTToSemiLeptonic\")\n",
-    "utils.get_cutflow(f\"{data_dir}/{year}/{sample}/pickles\", year, sample)\n",
-    "\n",
-    "# import pickle\n",
-    "\n",
-    "# for i in range(68):\n",
-    "#     with open(f\"{data_dir}/{year}/{sample}/pickles/out_{i}.pkl\", \"rb\") as file:\n",
-    "#         out_dict = pickle.load(file)[year][sample]['cutflow']\n",
-    "#         print(out_dict[\"all\"])\n",
-    "# utils.get_cutflow(f\"{data_dir}/{year}/{sample}/pickles\", year, sample)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "data_dir = \"../../../../data/ttsfs/Jan19/\"\n",
-    "\n",
-    "# utils.get_cutflow(f\"{data_dir}/{year}/TTToSemiLeptonic/pickles\", year, \"TTToSemiLeptonic\")\n",
-    "utils.get_cutflow(f\"{data_dir}/{year}/{sample}/pickles\", year, sample)\n",
-    "\n",
-    "# import pickle\n",
-    "\n",
-    "# with open(f\"{data_dir}/{year}/{sample}/pickles/out_1.pkl\", \"rb\") as file:\n",
-    "#     out_dict = pickle.load(file)[year][sample]['cutflow']\n",
-    "#     print(out_dict)"
+    "utils.get_cutflow(f\"{data_dir}/{year}/{sample}/pickles\", year, sample)"
    ]
   },
   {
@@ -233,16 +204,36 @@
     "    \"TT Unmatched\",\n",
     "    \"TT W Matched\",\n",
     "    top_matched_key,\n",
-    "]"
+    "]\n",
+    "\n",
+    "bg_colours = {\n",
+    "    \"QCD\": \"lightblue\",\n",
+    "    \"Single Top\": \"darkblue\",\n",
+    "    \"TT Unmatched\": \"darkgreen\",\n",
+    "    \"TT W Matched\": \"green\",\n",
+    "    \"TT Top Matched\": \"orange\",\n",
+    "    \"W+Jets\": \"darkred\",\n",
+    "    \"Diboson\": \"red\",\n",
+    "}"
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "metadata": {},
    "source": [
     "### Pre plots"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "events_dict.keys()"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -251,16 +242,17 @@
    "source": [
     "# {var: (bins, label)}\n",
     "plot_vars = {\n",
-    "    \"ak8FatJetMass\": ([20, 125, 225], r\"$m_{SD}$ (GeV)\"),\n",
-    "    \"ak8FatJetPt\": ([20, 500, 1200], r\"$p_T$ (GeV)\"),\n",
-    "    \"ak8FatJetnPFCands\": ([20, 0, 120], r\"# of PF Candidates\"),\n",
-    "    \"ak8FatJetParticleNet_Th4q\": ([20, 0.6, 1], r\"ParticleNet $T_{H4q}$ Non-MD\"),\n",
-    "    \"ak8FatJetParTMD_THWW4q\": ([20, 0.6, 1], r\"ParT $T_{HWW4q}$ MD\"),\n",
-    "    \"tau21\": ([20, 0.04, 0.8], r\"$\\tau_{21}$\"),\n",
-    "    \"tau32\": ([20, 0.2, 1], r\"$\\tau_{32}$\"),\n",
-    "    \"tau43\": ([20, 0.42, 1], r\"$\\tau_{43}$\"),\n",
-    "    \"tau42\": ([20, 0, 1], r\"$\\tau_{42}$\"),\n",
-    "    \"tau41\": ([20, 0, 1], r\"$\\tau_{41}$\"),\n",
+    "    \"ak8FatJetMass\": ([30, 50, 200], r\"$m_{SD}$ (GeV)\"),\n",
+    "    \"ak8FatJetParticleNetMass\": ([30, 50, 200], r\"$m_{reg}$ (GeV)\"),\n",
+    "    # \"ak8FatJetPt\": ([20, 500, 1200], r\"$p_T$ (GeV)\"),\n",
+    "    # \"ak8FatJetnPFCands\": ([20, 0, 120], r\"# of PF Candidates\"),\n",
+    "    # \"ak8FatJetParticleNet_Th4q\": ([20, 0.6, 1], r\"ParticleNet $T_{H4q}$ Non-MD\"),\n",
+    "    # \"ak8FatJetParTMD_THWW4q\": ([20, 0.6, 1], r\"ParT $T_{HWW4q}$ MD\"),\n",
+    "    # \"tau21\": ([20, 0.04, 0.8], r\"$\\tau_{21}$\"),\n",
+    "    # \"tau32\": ([20, 0.2, 1], r\"$\\tau_{32}$\"),\n",
+    "    # \"tau43\": ([20, 0.42, 1], r\"$\\tau_{43}$\"),\n",
+    "    # \"tau42\": ([20, 0, 1], r\"$\\tau_{42}$\"),\n",
+    "    # \"tau41\": ([20, 0, 1], r\"$\\tau_{41}$\"),\n",
     "}\n",
     "\n",
     "pre_hists = {}\n",
@@ -271,24 +263,29 @@
     "            events_dict, var, bins, label, weight_key=\"weight\"\n",
     "        )\n",
     "\n",
-    "# merger_pre_plots = PdfFileMerger()\n",
+    "merger_pre_plots = PdfMerger()\n",
     "\n",
-    "# for var, var_hist in pre_hists.items():\n",
-    "#     name = f\"{plot_dir}/pre_{var}.pdf\"\n",
-    "#     plotting.ratioLinePlot(\n",
-    "#         var_hist,\n",
-    "#         plot_samples,\n",
-    "#         year,\n",
-    "#         bg_err=None,\n",
-    "#         name=name,\n",
-    "#     )\n",
-    "#     merger_pre_plots.append(name)\n",
+    "for var, var_hist in pre_hists.items():\n",
+    "    name = f\"{plot_dir}/pre_{var}.pdf\"\n",
+    "    plotting.ratioHistPlot(\n",
+    "        var_hist,\n",
+    "        year,\n",
+    "        [],\n",
+    "        plot_samples,\n",
+    "        # bg_err=None,\n",
+    "        name=name,\n",
+    "        bg_colours=bg_colours,\n",
+    "        bg_order=plot_samples,\n",
+    "        ratio_ylims=[0.6, 1.3],\n",
+    "    )\n",
+    "    merger_pre_plots.append(name)\n",
     "\n",
-    "# merger_pre_plots.write(f\"{plot_dir}/PrePlots.pdf\")\n",
-    "# merger_pre_plots.close()"
+    "merger_pre_plots.write(f\"{plot_dir}/PrePlots.pdf\")\n",
+    "merger_pre_plots.close()"
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "metadata": {},
    "source": [
@@ -367,6 +364,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "metadata": {},
    "source": [