Treatment of electron jets

[samvanstroud]

In most samples we write n_truth_promptLepton which can discriminate between mislabelled electron-jets and light-jets. In the GN1 plotting code (which we will stop using in favour of this package) we require all jets to have n_truth_promptLepton == 0.

We might consider (optionally) doing the the same here to avoid inflating light rejection at high pT (when electron jets become a significant fraction of light jets).

More info here: https://its.cern.ch/jira/browse/AFT-593

[joschkabirk]

Hi @samvanstroud,

great to see that you will make use of this package!

I think, however, that this should not be included in puma itself, but rather in the GNN-specific plotting scripts that are used in the end?

But let's see what @afroch, @manuguth and @philippgadow think.

[samvanstroud]

I guess we are really having the same discussion as in #122. But I would note that removing the electron jets should be done when evaluating performance for all taggers, and not just GN1. Since there are several "suggested ways of doing things" in terms of these selections and e.g. discriminant calculations, it makes sense to me to have this package make these things as easy as possible for the user. But let's continue the discussion in #122 :)

[samvanstroud]

Some progress: #165 added this selection in the high level example docs https://umami-hep.github.io/puma/main/examples/high_level_api.html#initialising-the-taggers