Provide an unique truth for all theory cards

[andreab1997]

Here we address #2019.

• Create a dataclass collecting all the possible parameters that can appear in a theory card, removing the ones that are not used anymore
• Add utilities function to read a theory which can be also used by pineko
• Provide a way to read and/or search in all theories

cc @scarlehoff

[andreab1997]

At the moment if you see # it means that I am not sure if the parameter is still used or not. I am adding now comments and questions for all the problematic ones

[scarlehoff]

Thanks @andreab1997, let's have next week as the target to merge this

With respect to the missing items:

Add utilities function to read a theory which can be also used by pineko

I think this can be decoupled

Provide a way to read and/or search in all theories

I'll add this to #2037 (but I think it can also be decoupled)

[andreab1997]

Thanks @andreab1997, let's have next week as the target to merge this

With respect to the missing items:

Add utilities function to read a theory which can be also used by pineko

I think this can be decoupled

Provide a way to read and/or search in all theories

I'll add this to #2037 (but I think it can also be decoupled)

Ok agreed. Do you want me to do something before merging it?

[scarlehoff]

Some comments.

[RoyStegeman]

Since I believe this is almost final, don't forget to update (the theory_params.csv file in) the docs

[RoyStegeman]

One thing I just realised is that by having defaults we make it even harder to search within the theorydb based on theoryparameters. Of course I'll concede that since they are the defaults it's probably not particularly useful to use it as a filter

[scarlehoff]

Having defaults doesn't mean we should not fill in a value. The main goal of the defaults is for the old runcards not to fail and allow us to remove some keys going forward.

So, even if some keys are added with a default we can decide that new theories must have a value defined (while leaving old theories alone). For now in #2062 Q0 is mandatory again.

[RoyStegeman]

But all old theories have an explicit entry for e.g. alphas (and all other values that are given defaults in this PR), so why would those not be made mandatory now. There is no need to give defaults for older theories

[scarlehoff]

But all old theories have an explicit entry for e.g. alphas (and all other values that are given defaults in this PR), so why would those not be made mandatory now. There is no need to give defaults for older theories.

The idea is that some of the values in the old theories were either redundant or not needed. But maybe we were overeager and there's some that have been moved to optional that shouldn't be.

(if so, please mention which ones are those and we can make them mandatory again!)

[RoyStegeman]

wrt to this PR I would not add defaults for MaxNfPdf, Q0, Qref, alphaqed, HQ

[github-actions]

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

• Fit Name: NNBOT-d251495cb-2024-04-30
• Fit Report wrt master: https://vp.nnpdf.science/P78tP5e6QVSOdN9pv7y9kg==
• Fit Report wrt latest stable reference: https://vp.nnpdf.science/UrYku_tBR-Sh2G1zU3obGQ==
• Fit Data: https://data.nnpdf.science/fits/NNBOT-d251495cb-2024-04-30.tar.gz

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

[scarlehoff]

If nobody is explicitly against this, I'll merge this PR on Thursday.

@RoyStegeman I'd appreciate a second look at the 41000000.yml file since I've made a few changes.

[RoyStegeman]

@RoyStegeman I'd appreciate a second look at the 41000000.yml file since I've made a few changes

As you know I'm happy with explicitly defining parameters in the runcard instead of relying on the defaults, so for me this is not a huge problem, but IterEv and HQ are set to the default values so they can also be removed.

[scarlehoff]

I also prefer having explicit values even if defaults are defined.

[RoyStegeman]

Good, other than that 41000000.yml seems fine