Following the jet calibration example discussed at pyhep.dev, there are a few different ways forward in the direction of extending functionality. Some of these can be pursued in parallel.

The pipeline function contains them all:

def pipeline(jets, a):
    """analysis pipeline: calculate mean of dijet masses"""
    return np.mean(get_mass(correct_jets(jets, a)))

• replace correct_jets by more complex calibration operations, e.g. with a differentiable correctionlib
• replace get_mass by a more complicated physics analysis, e.g. involving combinations of objects and sorting (the AGC top mass reconstruction is a good candidate)
• replace np.mean by statistical model construction + inference

In addition to this:

• use nanoevents -> #2

Gathering related issues to easily be able to track everything in one place.

• scikit-hep/awkward#2591
• CoffeaTeam/coffea#874
• scikit-hep/awkward#2603 (partner issue to coffea one above)
• scikit-hep/awkward#2637
• scikit-hep/awkward#2638

meta items:

• CoffeaTeam/coffea#899