[Clascomment] OPT-IN: Cross sections for the exclusive photon electroproduction on the proton and Generalized Parton Distributions

[Hyon-Suk JO]

Hi Sebastian,

Thanks for your comments.

The revised version of the paper can be found here:
https://www.jlab.org/Hall-B/shifts/admin/paper_reviews/2015/dvcs_cross_sections_prl_draft_v4-5413120-2015-03-31-v4.pdf

> Fig. 4): There is a lot of information in these figures, and it might be possible to make it easier accessible and clearer with a few changes. For one, it's not clear that one really needs 3 different t values in each row - by choosing only two, the plots can be made bigger which would be a big improvement. Also, the caption should contain at least a list of the curves/shades drawn - it would become a little longer that way but would make the figure much more understandable (instead of having to hunt in the text for the explanations). Ideally, the figure would fill (most of) one column, but I understand you are pressed for space (is PRL still THAT restrictive? I thought I recently saw some indication that they allow a little more space now...)

We have included the description of the curves and bands
in the caption of Fig. 4. However, we insist on keeping the
three t-bins. Besides the frustration of showing the results
of only six of our (Q2, xB, t) bins out of 110, the
t-dependence of the observables contains physics information,
noticeable by eye. For instance, the amplitude of the
beam-polarized cross-section difference rises first between
the first and second t-bins (maybe not really noticeable by
the data but by the theory curves) and then drops between
the second and third t-bins. This "rise-and-drop" feature
was also observed in the previously published CLAS DVCS BSA
(Ref. [13]) and Hall A cross sections (Ref. [12]). Obviously,
removing one t-bin would make this feature vanish. We don't
comment particularly on this behavior in the text because
the interpretation and the link to the CFFs is not direct
(the CFFs drop monotonously as a function of t as can be
seen in Fig. 5 because the DVCS+BH amplitude contains some
extra t-dependence) but we nevertheless think that it should
be made visible on the figure.

> In the discussion of that figure, you state "Figure 4 shows that the predictions of standard GPD models like VGG, GK, and KM10a, whose compatibility is remarkable in spite of their different approaches, are in good agreement with our unpolarized cross-section data." Also, in the abstract it says "Several models of Generalized Parton Distributions (GPDs) describe the data well", as well as in line 324-7 (flip the order of "the data" and "well" in that sentence). One could argue that the difference between Chi2/dof = 1.9 and 1.5 is very significant, so maybe it would be fair to say KM10a does the best job. This is even more obvious once you include the polarized data. In fact, as Fig. 5 also shows, VGG does not describe the data at low x very well.

Regarding the comparison of the models, the standard GPD
models indeed describe the data well in the six bins that
we show but the agreement is not equally good in all our
110 (Q2, xB, t) bins. We are giving average chi2 values to
give a general indication of how the models compare to the
data but in fact, the chi2 varies for each model as a
function of the bins/kinematics. Thus, one should not
conclude that, among the standard GPD models (VGG, KMS,
KM10a), one model is doing a better job than another as the
situation can differ in each bin. One should also note that
none of the four models discussed have been tuned to these
new data and that each of them can certainly adjust some
parameters and achieve a better agreement. The main point
of the comparison of the models with the data is the fact
that the data appear to be incompatible with the KM10 model
that includes the very strong H_tilde contribution. Because
of that, we might have oversimplified one of the
conclusions, saying that the other three models "describe
the data well". In order to take into account your remark,
we therefore add on lines 231-232 the sentence: "Note that
none of these four models has been tuned to our data.".
We also add the reference [33] as a footnote: "The chi2
values, integrated over all the bins, give a general
indication but the level of agreement or disagreement
between each GPD model's predictions and the data varies
as a function of the kinematics.".

Cheers,
Hyon-Suk