Evolution operator in exported PDF grids is incorrect for scale variation theories
Created by: Zaharid
According to the theory covmat paper, the PDF evolution should only depend on whatever we call factorization scale multiplier, but not on the renormalization scale multiplier. By default, APFEL uses a different convention where the evolution depends on the ratio of the two multipliers: The only information from the multipliers theories it looks at is here
which in turn is used (after several indirection) as the kren parameter, e.g. here
This is called by the code that computes the evolution operator needed to produce the LHAPDF grid at the end of the fit (e.g. here):
so this is using the "old" convention that has nothing to do with what is discussed in the paper.
It turns out that this might not affect APFLECOMB (so the evolution operator in the FKtables) because there is some undocumented flag in APFEL that is set here
and has the effect to undo all the options in ways that I do not understand very well (@lucarottoli @scarrazza would you please double check that this does the right thing in the end? I tried to convince myself that it is correct and I couldn't) though a series of if conditions littered throughout the APFEL code base. In any case the evolution in the FKtables is different from that in the PDF grids.