EGamma Regression Trainer

The egamma regression training has following steps:

1. Get the input sample for the training

   1. The regression training needs two input root file. One corresponds to the ideal Intercalibration Constants (IC) and another for the real IC.

   Note

   For run - 3, the raw energy includes both ECAL and PS energy combined. For run - 2 regression, we didn’t include the PS energy in the training.

2. Train the regression and check its performance.

3. Re-validate the training by re-running the HLT step with the updated .db file.

4. Upload the .db file having the training information in the Global Tag (GT).

Contents

• 1. Get HLT Configuration file
  • 1.1. Setup
  • 1.2. Get hltConfig file
  • 1.3. Crab Submit
• 2. Reduced Ntuples
  • 2.1. Step -1: Setup
  • 2.2. Step - 2: Get Flat ROOT File
• 3. Regression
  • 3.1. Some general information
  • 3.2. Regression setup
  • 3.3. Verify new regression by re-running hlt
  • 3.4. Re-Run HLT with updated training information
• 4. HLT Config for the CMSSW feature comparison
• 5. References
  • 5.1. Regression references
  • 5.2. Others
• 6. List of Pull Requests related to Run - 3
  • 6.1. Main PRs
  • 6.2. Other related PRs
• 7. List of talks
  • 7.1. Run -3 related talks
• 8. Default README
  • 8.1. E/gamma Energy Regression Trainer
  • 8.2. setup instructions
  • 8.3. running the regression training
  • 8.4. example workflow
  • 8.5. build system