J/psi-jets Analysis

(created on April 11, 2025. Aliases aimed to my pc)

Repository for J/psi-jets analysis within ALICE's Run 3. In general terms, this framework is meant for:

• Prepare ALICE's data for analysis via data skimming
• Reconstruct J/psi mesons from dielectron decays.
• Find J/psi jets using anti-kt clustering algorithm
• Calculate the momentum-fraction distribution of J/psi in jets and other related observables
• Separate prompt from non-prompt J/psis
• Apply similar methods to Monte Carlo (MC) datasets. With this,
  • Calculate reconstruction efficiencies
  • Calculate bayesian-unfolding corrections
• Apply those corrections on measured distributions
• Compare the results with theoretical models

Below, a detailed tutorial on running things locally.

Contents

• Preparing the Framework
• Calculation of Momentum Fraction
• Monte Carlo
• Efficiency Calculation
• Separation of Prompt and Non-Prompt J/psi

Preparing the Framework

The usage of this repository generally demands ALICE's O2 analysis framework

1. Install O2Physics (see O2 Documentation)

2. Recomendations:

   • add this to your .bashrc in order to enter ALICE environment with alice:

   O2PhysicsBranch='O2Physics/latest'
   alice() {
   cd ~/alice/
   alienv enter $O2PhysicsBranch
   }
   

   • Add bashrc_alienv to alice/ folder
   • Add Jpsi-Jets-Analysis/ repository to alice/

3. Enter O2Physics environment:

   • Using alias: alice & source bashrc_alienv
   • Or alienv enter O2Physics/latest

Calculation of Momentum Fraction

This is mainly done by jpsiFragmentationFunction.cxx task.

1. The input AODs must be downloaded and prepared
2. Compile my task:
   • Can be done with AliBuild or with Ninja
   • To compile with Ninja (maybe direnv allow is needed!):
     • alias: ninjajpsi
     • or cd ~/alice/sw/BUILD/O2Physics-latest/O2Physics/ & ninja PWGJE/Tasks/install -j 3
3. Run my task:

   • To run, a json file must be provided containing:

     • the AOD file as input or a txt file containing a list of AODs paths (in this case the string must start with "@")

       • Example: @CompleteAO2DFilesSubJobs_24Gb_LHC24am_pass1.txt

     • Parameters for the jet finder, such as:

       • Vertex z cut

       • event selection

       • p_T, eta, phi

       • Jetalgorithm

     • Parameters for JPsi Task, such as: p_T bounds (ex.: 5, 7, 15 and 35 GeV)

   • If a json file doesn't exist yet, run without it and it will be automatically generated

   • A script was created to run this, which can be executed as following:

   • cd into ~/alice/Jpsi-Jets-Analysis/JpsiWorkDir or cd $JpsiDir (alias)

   • Run the script providing the json file:

     • Make it executable: chmod +x runJPsiFragmentation.sh

     • Execute it: ./runJPsiFragmentation.sh dpl-config.json

     • (Or o2-analysis-je-jet-jpsi-fragmentation -b --configuration json:<JSON> | o2-analysis-je-jet-finder-dielectron-data-charged -b --configuration json://<JSON>)

   • A AnalysisResults.root should've been created with jet spectra and z-vs-mass histograms for each pT range

4. Run the fitter:
   • cd ~/alice/Jpsi-Jets-Analysis/JpsiFitter
   • a json file must me passed with
     • Input root file
     • Histogram to be fitted
     • Signal function
     • Bkg function
     • etc.
   • Run tutorial.py (or similar ones), which runs DQFitter for every x-projection of the histo:
     • python tutorial.py configFit_z_Xi.json --run_fit_projections
   • A root file should've been created in output/, containing:
     • A copy of the input histogram and inclusive mass distribution
     • Results of the fits for every x projection
     • For each fit range, a histogram compiling these results

Monte Carlo

Analysis done in ~/alice/Jpsi-Jets-Analysis/JpsiWorkDir/MC

• Given J/psi Monte Carlo datasets, Skim them using TableMakerMC_withAssoc. Usually using Hyperloop.
  • Provide reco and gen level MCSignals to it. Ex.: eFromJpsi,eFromPromptJpsi,eFromNonpromptJpsi,Jpsi,nonPromptJpsi,promptJpsi,allBeautyHadrons,Bplus,protonPrimary, everythingFromEverythingFromBeauty
  • Output:
    • AO2D.root with reduced tables for reco-level tracks (mainly electrons) and gen-level (particles matching provided MCSignals or matching reco tracks)
    • AnalysisResults.root with general statistics
• run dqEfficiency_withAssoc on reduced AO2D (done in ~/alice/Jpsi-Jets-Analysis/JpsiWorkDir/MC/DQEfficiency)
  • analysis on pairs of electrons
  • For same-event-pairing, provide gen and reco MCsignals, such as"cfgBarrelMCGenSignals": "promptJpsi,nonPromptJpsi" and"cfgBarrelMCRecSignals": "eePrimaryFromPromptJPsi,eePrimaryFromNonPromptJPsi"
  • Important: in order to produce dielectronAll table, one must:
    • Produce the reduced dataset containing ReducedTracksBarrelInfo
    • Enable processBarrelOnlyWithCollSkimmed, cfgFlatTables and fgUseKFVertexing
    • Add kalman-filter histograms to cfgAddSEPHistogram
  • Output:
    • AnalysisResults.root with histograms for dielectron which passes each MCSignal
    • AnalysisResults_Trees.root with dielectron trees (tables). Can be dielectronsAll, for example

Efficiency Calculation

Done by the matching between MC-truth-level and MC-reconstructed-level J/psis

1. Create a Python virtual environment inside ALIEnv (alice/Hipe4MLenv, for example)
2. Install Hipe4ML in it
3. Enter ALIEnv: alice & source bashrc_alienv
4. Enter Hipe4ML venv: source Hipe4MLev/bin/activate
5. Efficiency calculated in Jpsi-Jets-Analysis/efficiencyAndML/efficienciesJpsi.ipynb
   • Open with code and select Hipe4MLenv Python kernel

Separation of Prompt and Non-Prompt J/psi

This is done using Boosted Decision Trees

1. Produce a MC J/psi tree (such as one containing dielectronall produced using dqEfficiency)
2. Enter Hipe4ML virtual environment such as described in Efficiency section
3. run promptSeparation.ipynb
4. TODO (25-11-14): With the model trained and converted to ONNX, do the inference on data using MLResponse framework