Summary

This in-development analyzer and associated plotting scripts are for developing tau and b-jet Level-1 Triggers for the CMS Experiment's Run 3 and Phase 2.

Installation

Installation instructions are in checkout.sh.

B-tagging

Running the analyzer

To run the Analyzer:

cd $CMSSW_BASE/src/L1Trigger/phase2L1BTagAnalyzer/test/
cmsenv
voms-proxy-init --rfc --voms cms  # provide GRID password
cmsRun test-Analyzer.py #this will produce an analyzer.root

Alternatively, you can have it run as a background process so it conntinues after logging out, and redirect the terminal output to a logRun file:

nohup cmsRun test-Analyzer.py > logRun &

The .root file is written to the same directory (usually called analyzer_X.root). Rename it and save it to a folder (e.g. outputs/) and update a README in the folder that describes which miniAOD/FEVT files were used to generate it.

Plotting reconstructed jet properties

To plot histograms of the reco-level jet pT, Eta, Phi, and (up to 4) track impact parameters, with b-jets in red and light-flavor jets in blue, use plotRecoJetProperties.C macro with command line arguments to make a .pdf.

cd $CMSSW_BASE/src/L1Trigger/phase2L1BTagAnalyzer/btag_exercise/plot_jet_features/
root -l "plotRecoJetProperties.C(\"path/to/input/file.root\", \"outputname.pdf\")"	

Training ROOT TMVA classifiers

To train TMVA classifiers (BDTs, ANNs, etc.) using the reconstructed jet properties, use the ROOT macro TMVAAnalysis.C. First navigate to the directory:

cd $CMSSW_BASE/src/L1Trigger/phase2L1BTagAnalyzer/btag_exercise/mva

1. Update the input file path in the "Load data" section of TMVAAnalysis.C.
2. Run the macro:

root -l -b -q TMVAAnalysis.C

• Add > logRun to the end to pipe the terminal output to logRun.

3. This will generate an output file "TMVA_output.root" in the same directory, whose contents can be viewed and plotted with the interactive TMVA GUI.

root -l -e 'TMVA::TMVAGui("TMVA_output<keyword>.root")'

Applying the TMVA classifiers to other files

To apply the standalone classifiers to other .ROOT samples, set the input and output file paths in applyWeightFile.C and run:

root -l -b -q applyWeightFile.C

Tau tagging

Workflow

This repository does not include the tau analyzer, so the workflow for training TMVA classifiers on tau ttrees is:

1. In tau_exercise, set input and output file paths in TMVAAnalysis_tau.C. Run the macro to generate an output file (replace X with whatever identifier the input file has):

root -l -b -q TMVAAnalysis_tau.C > files/trainingOutputs/TMVA_taus_logX 

Create and view desired plots in the interactive TMVA GUI:

root -l -e 'TMVA::TMVAGui("out_[keyword].root")

The output .eps files in dataset/plots will only be generated/ updated by creating them afresh in the interactive TMV GUI, which can be laggy. Check the Last Edited timestamp to avoid mistaking old plots for new ones.

2. In /applyWeightFiles, set the input tree path and output file name.

root -l -b -q applyWeightFile_taus.C