Coffea by Example

The following pages are rendered jupyter notebooks that provide an overview and example usage of Coffea features. Each notebook builds on the previous one so it is recommended to go through them in order.

• Reading data with coffea NanoEvents
• Applying corrections to columnar data
  • Coffea lookup_tools
    • Opening a root file and using it as a lookup table
    • Building and using your own correction from a histogram
  • CMS high-level tools
    • Applying energy scale transformations with jetmet_tools
    • Applying CMS b-tagging corrections with btag_tools
  • Using correctionlib
• Coffea Processors
  • Getting fancy
• Running inference tools
  • Why these wrapper tools are needed
  • Example using ParticleNet-like jet variable calculation using PyTorch (GNN)
  • Example using ParticleNet-like jet variable calculation using PyTorch (DNN)
  • Comments about generalizing to other ML tools
  • Additional comments on common prepare_awkward patterns
    • Casting multiple fields a collection to be separate axis
    • Fixing collection dimensions
• Dataset discovery tools
  • Using Rucio utils directly
    • Dataset replicas
    • Filtering sites
  • Using the DataDiscoveryCLI
    • Filtering sites
    • Save the replicas metadata
  • DataDiscoveryCLI from shell
  • Preprocess the fileset with dask
• PackedSelection in Coffea 2023