How to create PDF sets with flavor-number variations
The default PDF fits are generated assuming a
nf=5 flavor scheme. It is
possible to use a different flavour scheme for the evolution, given an existing
fit at the initial scale.
After performing a fit, the
varflavours script can be
used to create a PDF evolved with a different maximum number of flavors.
Instead of running
evolven3fit to evolve the fit with the theory used in
the predictions, one would call varflavours as follows
varflavors fit_folder max_replicas THEORYID --new_name=[name of output fit]
max replicas are the maximum number of replicas evolved by
THEORYID is again the theoryID of the theory used to do
the DGLAP evolution, and
new_name is an optional argument setting the new
name for the fit.
To finalize the PDF set, postfit should be run in the usual manner.
Note that the resulting PDFs will not correspond to the best PDF in that flavour scheme (for which we would need to do an actual fit yielding different initial scale PDFs).
Under the hood, the
varflavors script uses the
evolven3fit which allows for the possibility to use a different theory
during evolution than was used during the fit itself.
As such, one can perform a fit using a theory corresponding to the
flavor scheme, and then perform the evolution by running
evolven3fit --theory_id=[THEORYID] [configuration folder] [number of replicas]
THEORYID corresponds to a theory with a flavor number scheme that
differs from the
nf=5 flavor scheme used during the fit.
In case it is desired to create a PDF with flavor-number differing from the
fit, while also keeping the original
nf=5 version, one can use
vp-pdfrename to create two PDF sets with different
names that are based on the same input fit.
If we do this, the values for
MZ in the
need to be changed to take the
Qref values from the theory
used to do the evolution.