AGC Wishlist (discussed at PyHEP.dev)

[pfackeldey]

Data

• 2016 Open Data NanoAOD

Systematics

• object-based:
  • JEC
• weight-based:
  • e.g. b-tagging weight variations
• modelling variations:
  • e.g. different $m_t$

Data-driven Background Estimation

• Fake/ABCD estimation

ML

• S vs B for e.g. ttH measurement

Sensitivity Optimization

• Binning/ML Optimization

Technical

• coffea 2024
• workflow system (law)
• XGBoost

[alexander-held]

Some more context can be found in the summary report from the workshop: https://arxiv.org/abs/2410.02112. We will presumably tag AGC v2.0 corresponding to the current setup, move to adopt coffea 2024 in this repository and then tag v2.1, and then evolve the analysis task as described here towards an eventual tag v3.0. This follows the current approach: major version for analysis task changes, minor versions for implementation changes.

[alexander-held]

Discussed in AGC meeting: switch to RNTuple (or make RNTuple an option at least).