Validphys with pineappl rebased to master

[scarlehoff]

I've decided to not push on top of #1529 after the rebase just in case

But there's nothing new with respect to #1529 I just want to have both active at the same time.

[github-actions]

Greetings from your nice fit 🤖 !
I have good news for you, I just finished my tasks:

• Fit Name: NNBOT-b49f9eb66-2022-07-12
• Fit Report: https://vp.nnpdf.science/A-YkvCVmTd2VDqtc3lRynw==
• Fit Data: https://data.nnpdf.science/fits/NNBOT-b49f9eb66-2022-07-12.tar.gz

Check the report carefully, and please buy me a ☕ , or better, a GPU 😉!

[scarlehoff]

@andreab1997 I've addressed the comments. In order to ensure there are no other rebase problems hidden, could you run a theory-uncertainties fit and then a report with the previous one you ran in order to check the results are unchanged?

Edit: I've pushed the commits to the wrong branch >:(

[andreab1997]

@andreab1997 I've addressed the comments. In order to ensure there are no other rebase problems hidden, could you run a theory-uncertainties fit and then a report with the previous one you ran in order to check the results are unchanged?

Edit: I've pushed the commits to the wrong branch >:(

Yes sure, I will run it in a moment

[andreab1997]

I approved but I am still running the fit. The moment I have the report I will post it here @scarlehoff

[andreab1997]

@andreab1997 I've addressed the comments. In order to ensure there are no other rebase problems hidden, could you run a theory-uncertainties fit and then a report with the previous one you ran in order to check the results are unchanged?

Edit: I've pushed the commits to the wrong branch >:(

The fit just finished. It seems that the files .fitinfo, .params, .preproc and .sumrules are not there. Is this wanted or an error? In both cases postfit can't run without that files @scarlehoff

[scarlehoff]

This is not an error but postfit should run! Maybe you have an old installation of postfit?

[andreab1997]

This is not an error but postfit should run! Maybe you have an old installation of postfit?

yes, it is possbile. Let me try

[andreab1997]

Ok you were right. This is the link of the fit https://data.nnpdf.science/fits/220325-ab-nnpdf31like_wthcovmat_fullcovmat_test.tar.gz.
I am running now the comparison.

[andreab1997]

Here is the comparison report.
https://vp.nnpdf.science/0-syr4ezTpqoYO8odcRoLA==
I believe that they are actually equivalent :) @scarlehoff

[scarlehoff]

Looks great! Thanks!

[andreab1997]

Looks great! Thanks!

I am also trying the new theories now. I you want I can post here the results before merging

[andreab1997]

And if is not a problem, I would add the new dictionaries to

nnpdf/validphys2/src/validphys/scalevariations/scalevariationtheoryids.yaml

Line 2 in 17952b4

scale_variations_for:

in this PR

[andreab1997]

Just as a proof of concept these are the reports of the thcovmat plots with 3pt prescription for scheme-B and scheme C
https://vp.nnpdf.science/T-LvbshRQ3CskNCRgm0xXg==
https://vp.nnpdf.science/6RmobzooSsOGycj3-eGx_g==

[andreab1997]

And just for reference I add here some other reports
https://vp.nnpdf.science/fbn88G-dRPqULc7rpKsgcg==
https://vp.nnpdf.science/S6brmMCPRjaCR_AAK4kJ4g==

[scarlehoff]

Since this has been tested quit a bit by now, I'll merge it soon if no bugs are found

[scarlehoff]

@andreab1997 are the changes in the two python files intentional? (they look like they are from black but one of them changes a lot of lines so just to be sure)

[andreab1997]

@andreab1997 are the changes in the two python files intentional? (they look like they are from black but one of them changes a lot of lines so just to be sure)

Yes, they are from black: I was debugging stuffs so I decided to also run black. The only real change is in scalevariationstheoryids.yaml.

[scarlehoff]

Ok, then. I will merge this after as soon as I manage to finish a fit (Galileo is a bit full so it might take a few days) since now there's you but also @giacomomagni who will need the stuff in this branch to proceed. At this point it should clearly be in master.

(and let's freeze it in its current state, otherwise once I manage to get the fit through I will have to restart it!)

[scarlehoff]

Comparison between theory 400 (in its current form, so positivity / integrability is from theory 200) and NNPDF4.0 https://vp.nnpdf.science/be1bDitCQl2ndF4x2cd9pA==/

This together with the fit bot and the reports that @andreab1997 has already showed makes clear that there are no bugs (at least no bugs that would affect a fit or its comparison report).

So I'll merge this now. @andreab1997 @giacomomagni @goord when doing any fits that would require pineappl (or any features from this branch) please use master to ensure that we are testing master (and that bugs in master are caught!). This also closes #1541

C++ is only left in the closure test generation and common data loading (prioritizing that now from my side!)