Compute F1

[toonhasenack]

We added a case in the yadism/src/esf/exs.py file where one would like to calculate F1. Dependencies of this F1 have also been added.

[Radonirinaunimi]

621d32f should not be part of this PR. Could you please remove it?

Edit: fixed now!

[felixhekhorn]

@Radonirinaunimi actually, is this consistent? When you request the observable g1 from yadism you get $2xg_1$ as inherited from LeProHQ, right? - see also #222 ... shouldn't we do here the same? i.e. you request F1 but you get $2xF_1$? it is a bit confusing, I know, but if we do the same thing in polarized and unpolarized we release some of the tension, because all that matters is $g_1/F_1 = (2xg_1)/(2xF_1)$

[giacomomagni]

it is a bit confusing, I know, but if we do the same thing in polarized and unpolarized we release some of the tension, because all that matters is g1/F1=(2xg1)/(2xF1)

Good point, maybe the best would be the other way round ie remove the 2x pre-factor also from g1, no ?

[felixhekhorn]

Good point, maybe the best would be the other way round ie remove the 2x pre-factor also from g1, no ?

this is of course an equally good option - the reason why I chose $2xg_1$ in my thesis is that with this normalization $F_2$ and $2xg_1$ have the analogous parton model expression, see e.g. PDG

(same holds for $xF_3$ ... )

[Radonirinaunimi]

You are absolutely right @felixhekhorn! The initial idea was to account for the extra $2x$ in the $g_i$ later and keep using $F_1$. In hindsight, since the $F_1$ is mainly relevant in the $g_1/F_1$ context, we should provide here $2xF_1$ instead.

    elif kind == "2xF1":
        return np.array([1, -1, 0])

[Radonirinaunimi]

I think this would be the easiest option (taking also into account your physics' argument), so if you agree @felixhekhorn, @giacomomagni we can correct this PR according to that.

[giacomomagni]

    elif kind == "2xF1":
        return np.array([1, -1, 0])

But let's keep calling it F1, otherwise it's even more inconsistent with the rest (right now both g1 and F3 are multiplied by x).
We should edit the documentation and spell this out properly there.

[Radonirinaunimi]

    elif kind == "2xF1":
        return np.array([1, -1, 0])

But let's keep calling it F1, otherwise it's even more inconsistent with the rest (right now both g1 and F3 are multiplied by x). We should edit the documentation and spell this out properly there.

Ok, agreed. It's not ideal but yes we should make this explicit not only in the documentation but also in the various metadata when computing predictions.

[toonhasenack]

Okay, if everyone agrees I will generate a new pull request

[felixhekhorn]

But let's keep calling it F1, otherwise it's even more inconsistent with the rest (right now both g1 and F3 are multiplied by x). We should edit the documentation and spell this out properly there.

I perfectly agree this should be proper documented

just to be explicit: what I believe are equal player are $F_2, F_L, xF_3, 2xF_1$ for unpolarized structure functions and $2xg_1, g_4, g_L, 2xg_5$ for the polarized case (see PDG above)

[Radonirinaunimi]

But let's keep calling it F1, otherwise it's even more inconsistent with the rest (right now both g1 and F3 are multiplied by x). We should edit the documentation and spell this out properly there.

I perfectly agree this should be proper documented

just to be explicit: what I believe are equal player are F2,FL,xF3,2xF1 for unpolarized structure functions and 2xg1,g4,gL,2xg5 for the polarized case (see PDG above)

Perfect!

Okay, if everyone agrees I will generate a new pull request

So, @toonhasenack you could now proceed this way.