Remove photon and top from LHAPDF grid if they are not active about eko

[alecandido]

In principle, you could always erase from it.

We never finished turning genpdf into a full-fledged toolbox, but you can check if stripping distributions is already supported there, and otherwise add it yourself (it should be pretty simple)

https://github.com/NNPDF/eko/tree/master/src/ekobox/genpdf
https://eko.readthedocs.io/en/latest/code/genpdf.html?highlight=genpdf

[giacomomagni]

Actually, looking at the code the easiest fix, that will solve your problem would be to change the pids list passed here:

https://github.com/NNPDF/nnpdf/blob/e470c7e73490f3d4725dde6813aba2a2de018b69/n3fit/src/evolven3fit_new/evolve.py#L184-L186

So we can drop 22 if there is no QED and top according to MaxNfPdf.
If you agree, I suggest to move the issue there and I can draft a PR.

[Radonirinaunimi]

Thanks both!

Actually, looking at the code the easiest fix, that will solve your problem would be to change the pids list passed here:

https://github.com/NNPDF/nnpdf/blob/e470c7e73490f3d4725dde6813aba2a2de018b69/n3fit/src/evolven3fit_new/evolve.py#L184-L186

So we can drop 22 if there is no QED and top according to MaxNfPdf. If you agree, I suggest to move the issue there and I can draft a PR.

This is indeed the best place to apply the modification. I am not sure though what people think about this. If @scarlehoff thinks that this is a reasonable, then if you @giacomomagni could open a PR to address this that'd be great (otherwise I can also do so).

[scarlehoff]

Yes, LHAPDF actually just puts 0 when it doesn't find a flavour so it should be equivalent.

However

The reason is that some random MC programs will crash just because of this

which MC programs are these? Wouldn't this be a bug as well to be reported to them? Or is it expected behavior? (we can fix it in our side since it wouldn't change anything ofc but still)

[Radonirinaunimi]

Closing this as it was indeed a limitation in the MC codes (which for now has to be fixed by the user).