Access to pineapplgrids?

[LucianHL]

Hi, when running a fit with a theoryID locally for the first time the relevant fastkernal files are downloaded from here.

I was just wondering if there was a way to make the pineapplgrid files on which these are based available (or if these are available, to point me to where)? This would just be useful from the point of view of checking with different evolution variants that are not by default available.

[scarlehoff]

Hi @LucianHL we are planning to make them public but there are a few logistical issues to solve.

One of them is that not all of them were produced with final versions of public codes (so we would need to curate which one we can make public without having explicit permission to do so).
And the other is storage. Having the fastkernel files public is already a bit of a problem, we are in the process of moving them somewhere else.

That said

1. Some of them are just from Plougshare.

2. All DIS grids (and some of the others) can be regenerated with:

https://github.com/NNPDF/pinecards
https://github.com/NNPDF/pinefarm

3. The ideal goal is not to make the grids public ourselves but rather have the runcard in pinecard and the grids in plougshare (e.g., most of the DY pineappl grids that we will use in 4.1 are already in plougshare https://inspirehep.net/literature/2871928 ; those differ from 4.0 in the choice of scale).

4. I think we can privately share most of the 3.1/4.0 grids you might need, but I'll cc @enocera since he might know whether some are subjected to embargoes.

(I'll transfer this issue to https://github.com/NNPDF/pinefarm )

[LucianHL]

Ok great thanks a lot for the info - I will take a look at using pinefarm soon. Indeed the 4.0 baseline is what I am interested in, so let's see what @enocera says!

[enocera]

Dear @LucianHL if your focus is NNPDF4.0, I don't think that we have reasons not to make the grids public. We generated all of them ourselves, using public pieces of code. Note that these grids are accurate to NLO. When generating FK tables from grids, at NNLO, one also needs K-factors, but these are already distributed, with the consensus of the authors who produced them, through https://nnpdf.web.cern.ch/nnpdf/tables/. This is going to change in the future, in the sense that NNLO-accurate grids will be produced without the need of relying on K-factors.

[scarlehoff]

I will take a look at using pinefarm soon. Indeed the 4.0 baseline is what I am interested in

just wanted to add, if the 4.0 baseline (i..e, no N3LO/QED/MHOU) is what you are looking at, those were not generated with pinefarm yet.

[LucianHL]

Hi @enocera ok great thank you, and yes understood. @scarlehoff if it is e.g. possible to also upload the pineapplgrids to

https://nnpdf.web.cern.ch/nnpdf/tables/

that would be brilliant, or else pass them on in some other way.

[LucianHL]

I will take a look at using pinefarm soon. Indeed the 4.0 baseline is what I am interested in

just wanted to add, if the 4.0 baseline (i..e, no N3LO/QED/MHOU) is what you are looking at, those were not generated with pinefarm yet.

ok thanks understood