Skip to content

Create test implementations of new commondata format

Created by: Zaharid

To make things in #1416 (closed) more concrete we agreed to try and produce test implementations of the (still fuzzy) current version of the idea for a couple of datasets.

The datasets should be one DIS and one hadronic and one of them should involve multiple (COMPOUND) fktables.

These should have:

  • A metadata file with the informations in the "runcards" folder, PLOTTING file, COMPOUND file and others.
  • A kinematics file with bin edges and central value in some format.
  • A data file with experimental central value and uncertainties (probably in a separate file).

The uncertainties should support a way of specifying correlated systematics across datasets. Probably also some way to deal with additive vs multiplicative.

Regarding formats, I would think for this a text format of some sort is more adequate (easier to read and exchange around) and also one that trivially round trips with pandas (ie better than csv), so maybe something like json.