Fix datasets with jets

[cschwan]

In Madgraph5_aMC@NLO v3.3.1 the cutting routine was changed, so that pjet isn't calculated in the place where we use it anymore. We either have to 1) run the jet algorithm ourselves or 2) move jet-related custom cuts from passcuts_user to passcuts_jets in cuts.f.

[alecandido]

I've not a perfect understanding, but to me seems like option 1) is sufficiently more complicated than 2) (do we need to run an extra step in the runner? do we need to embed the fortran/python code?).

Option 2) looks like something I can do with sed in the cuts_code folder (plus moving the place where code is injected, maybe...).
Probably I'm really not understanding something...

[cschwan]

I like option 2 better than 1, because it's faster as you said, but it requires some non-trivial adjustments. We need to essentially sort our custom cuts according to jet cut/non-jet cuts, and then write them into different functions in cuts.f.

[alecandido]

We can simply split them in two folders (the natural way to create scopes on filesystem).

We can split by hand, they are not that many (and we should check we are choosing them correctly).
However, just to point it out, I believe there is simple rule to decide if it's a jet cut or not: "if it contains at least a j in the file name it's a jet, otherwise it is not".

[cschwan]

@alecandido I'm a bit unhappy about the fact that Madgraph5 changed this; a possible problem that I see is that there are cuts that concern jets and leptons at the same time, meaning that we can't separate them, and I wonder if it isn't a better choice to rewrite the cutting routine slightly so that we don't have to change our cuts.

[alecandido]

I'm not an expert of mg5 internals, but if you feel like of course you can have a try, and I can just review it :)

[alecandido]

@cschwan I'm not sure if this is 100% related to the MadGraph pinecards, or I should transfer to pinefarm. But maybe you know.

[cschwan]

This concerns the code that patches/runs Madgraph, so I suppose it should go to pinefarm.

[cschwan]

This concerns the implementation of patches/cut in Madgraph, so it should be a pinefarm issue.