I think the closure data (level 1 - central values generated from vp-setupfit) already gets loaded as expected, provided that vp-rebuilddata has been ran but I want to add some tests to double check whether this is the case.
In particular consistency between c++ and python with the central values and the covmats.
I would then be interested what happens if you load a cut commondata file and then apply with_cuts, judging by the function if the cuts match then there is no issue. In this way we can easily get vp-setupfit to dump the cut data as a pd dataframe and then we don't need rebuilddata anymore.. I think that some caching of the closure central values is useful because it ensures backwards compatibility with older fits, but of course we could also generate the level 1 shift with each replica - I think this is unneccessarily inefficient though
I think the closure data (level 1 - central values generated from vp-setupfit) already gets loaded as expected, provided that
vp-rebuilddatahas been ran but I want to add some tests to double check whether this is the case.In particular consistency between c++ and python with the central values and the covmats.
I would then be interested what happens if you load a cut commondata file and then apply
with_cuts, judging by the function if the cuts match then there is no issue. In this way we can easily get vp-setupfit to dump the cut data as a pd dataframe and then we don't need rebuilddata anymore.. I think that some caching of the closure central values is useful because it ensures backwards compatibility with older fits, but of course we could also generate the level 1 shift with each replica - I think this is unneccessarily inefficient though