Run 3 Muon Pveto Workflow

This repository contains the scripts used to calculate the muon background veto probability (Pveto) for the disappearing tracks analysis using Run 3 ntuples.

The workflow is:

1. Generate EOS filelists
2. Submit HTCondor jobs
3. Run the Pveto analysis over ntuples
4. Produce per-job ROOT + JSON outputs
5. Merge JSON outputs
6. Produce LaTeX tables for the Analysis Note

---

Main Scripts

File	Purpose
MuonBackground_v2_table16_pveto_json_pairfix_taujet.py	Main Pveto analysis
MergePvetoJsonAndMakeTables_v2.py	Merge JSON outputs and make LaTeX tables
make_muon_filelists.py	Generate EOS filelists
run_job.sh	Condor worker job
submit_dataset.sh	Helper script for Condor submission

---

Environment Setup

Inside CMSSW:

cmsrel CMSSW_15_0_10
cd CMSSW_15_0_10/src
cmsenv
voms-proxy-init --voms cms --valid=168:00
cd DisappTrks_v2/BkgdEstimation/scripts

Create the local python dependency directory:

mkdir python_env

python3 -m pip install --target python_env \
    awkward \
    uproot \
    vector \
    numpy \
    fsspec-xrootd

Test imports:

PYTHONPATH=$PWD/python_env:$PYTHONPATH python3 -c \
"import awkward, uproot, vector, numpy, fsspec_xrootd; print('imports OK')"

---

Step 1: Generate Filelists

Generate all filelists:

python3 make_muon_filelists.py --year all

Or a specific dataset:

python3 make_muon_filelists.py --year 2023_D

This creates:

filelists/filelist_2023_C_Muons.txt
filelists/filelist_2023_D_Muons.txt
filelists/filelist_2024_Muons.txt
filelists/filelist_2025_Muons.txt

---

Step 2: Prepare Condor Directories

Create log directories:

mkdir -p logs/2023_C
mkdir -p logs/2023_D
mkdir -p logs/2024
mkdir -p logs/2025

Output directories are created automatically by run_job.sh.

---

Step 3: Configure JDL Files

Each dataset has its own JDL:

submit_2023_C.jdl
submit_2023_D.jdl
submit_2024.jdl
submit_2025.jdl

The important line is:

arguments = $(ProcId) 2023_D

which specifies the dataset name.

The JDL should also contain:

should_transfer_files = YES
when_to_transfer_output = ON_EXIT

transfer_output_files = analysis_output
transfer_input_files = \
    MuonBackground_v2_table16_pveto_json_pairfix_taujet.py,\
    helper_functions.py,\
    filelists,\
    python_env

Queue size is supplied automatically by the helper script.

---

Step 4: Submit Jobs

Use the helper script:

./submit_dataset.sh 2023_D

This script:

1. Reads the correct filelist
2. Counts the number of files
3. Computes the number of Condor jobs
4. Submits the jobs automatically

---

Step 5: Monitor Jobs

Check job status:

condor_q

Check held jobs:

condor_q -hold

Remove jobs:

condor_rm <clusterid>

Release held jobs:

condor_release -name <schedd> <clusterid>

---

Step 6: Output Files

Per-job outputs are written to:

analysis_output/<dataset>/

Example:

analysis_output/2023_D/

Each job produces:

Muon_2023_D_Pveto_<job>.root
Muon_2023_D_Pveto_<job>.json

---

Step 7: Merge JSON Outputs

After all jobs finish, merge the JSON outputs:

python3 MergePvetoJsonAndMakeTables_v2.py \
    --input-dir analysis_output/2023_D \
    --output-dir merged_output/2023_D

This produces:

• merged Pveto counts
• cutflow LaTeX tables
• Pveto LaTeX tables

---

Step 8: Use the LaTeX Tables

The merge script produces .tex files suitable for direct inclusion into the Analysis Note.

Example:

\input{pveto_2023D_muon_v2.tex}

---

Physics Notes

Pair-Level ΔR Fix

The muon veto must be applied at the pair level:

pair_dr > 0.15

This is implemented in:

passes_muon_veto

inside the analysis script.

---

Pveto Definition

The estimate uses:

$$P_{\mathrm{veto}} = \frac{ (N_{\mathrm{OS}}^{\mathrm{veto}} - N_{\mathrm{SS}}^{\mathrm{veto}}) }{ (N_{\mathrm{OS}}^{\mathrm{den}} - N_{\mathrm{SS}}^{\mathrm{den}}) }$$

Negative Pveto values are clipped to zero following the analysis note convention.

---

Tau Cleaning

Tau cleaning uses DeepTau IDs with:

• VSjet = Needs to be decided/done
• VSe = Needs to be decided/done
• VSmu = Needs to be decide/done

along with:

• decayModeFindingNewDMs
• rejection of decay modes 5 and 6

---

Recommended .gitignore

python_env/
logs/
analysis_output/
merged_output/

*.root
*.json
*.out
*.err
*.log

filelists/filelist_*.txt

---

Typical Workflow Example

Generate filelists:

python3 make_muon_filelists.py --year 2023_D

Submit jobs:

bash submit_dataset.sh 2023_D

Merge outputs:

python3 MergePvetoJsonAndMakeTables_v2.py \
    --input-dir analysis_output/2023_D \
    --output-dir merged_output/2023_D

Include table in AN:

\input{pveto_2023D_muon_v2.tex}