HH->WWbb Run-3 analysis

This repository uses the bamboo analysis framework, you can install it via the instructions here: https://bamboo-hep.readthedocs.io/en/latest/install.html#fresh-install

then install CMSJMECalculators and correctionlib

git clone https://gitlab.cern.ch/cp3-cms/CMSJMECalculators.git
pip install ./CMSJMECalculators
pip install correctionlib

Finally, clone this repository in the parent directory containing the bamboo installation:

git clone ssh://[email protected]:7999/aguzel/HHtoWWbb_Run3.git && cd HHtoWWbb_Run3

Execute these each time you start from a clean shell on lxplus or any other machine with cvmfs:

source /cvmfs/sft.cern.ch/lcg/views/LCG_105/x86_64-el9-gcc11-opt/setup.sh
source (path to your bamboo installation)/bamboovenv/bin/activate
export PYTHONPATH="${PYTHONPATH}:${PWD}/python/"

and the followings before submitting jobs to the batch system (HTCondor, Slurm, Dask and Spark are supported) or running on files stored on the grid:

voms-proxy-init --voms cms -rfc --valid 192:00 
export X509_USER_PROXY=$(voms-proxy-info -path)

if you encounter problems with accessing files when using batch, the following lines may solve your problem

voms-proxy-init --voms cms -rfc --valid 192:00  --out ~/private/gridproxy/x509
export X509_USER_PROXY=$HOME/private/gridproxy/x509

Then cutflow study of the analysis is executed via the following command line using batch (you can pass --maxFiles 1 to use only 1 file from each sample for a quick test):

bambooRun -m python/cutflowAnalysis.py config/<2022 or 2023>_v12.yml -o ./outputDir/ --envConfig config/cern.ini  --distributed driver

Instead of passing --envConfig config/cern.ini everytime, you can copy the content of that file to ~/.config/bamboorc.

using the skims in the root files that are in the results directory of bamboo output ./outputDir/, you can perform machine learning applications. Once you get the output from your machine learning (preferably in onnx format), you can execute

bambooRun -m python/mvaEvaluator.py config/<2022 or 2023>_v12.yml -o ./outputDir/ --envConfig config/cern.ini  --distributed driver --mvaModel <path to your ML model in onnx format>

where you get the ML model evaluated on the analysis.

To produce sync skim for syncronization exercises, you can run the following command:

bambooRun -m python/syncSkimmer.py config/2022_v12_sync.yml -o output/syncTest --sync

If you change something in the analysis workflow which you think may have an effect on the dilepton trigger efficienct in the analysis, re-produce dilepton trigger scalefactors with the following command:

bambooRun -m python/trigger_eff.py:TriggerEff config/<2022 or 2023>_v12.yml -o ./outputDir/

To convert the trigger efficiency histograms to JSON format, run:

python scripts/trg_eff_to_json.py --bamboo_output ./outputDir/

or if you think that a similar effect may be present in the b-tagging scalefactors, re-produce them with:

bambooRun -m python/btagReweighting.py config/<2022 or 2023>_v12.yml -o ./outputDir/

and copy the produced b-tagging scale factor files from the output directory to data/ folder.