[Clascomment] OPT-IN: Inclusive Electron Scattering in the Resonance Region off a Hydrogen Target with CLAS12

[Stepan Stepanyan]

Hello Valerii et al., Nice paper, congratulations!
Here are comments:

- line 26, strictly speaking, this is not true. The trigger defined in (1) is one of three triggers used for the RG-A Fall18 inbending dataset. It is true, though, that this is the trigger that the presented analysis of electron scattering used.
- line 38, looking at Fig.2, it is clear that there are no points close to Q2=12 GeV2. It bearly gets close to 11. 

- Why do you need to subsection label “A”, Section III has no subsections. 
- the cut on sampling fraction in Table I and stated in the text on line 68 are different, +/-5\sigma vs. >3.5\sigma, why?
- since you refer to [46] for momentum corrections, it will be fair to refer to Pierre’s analysis for PCal and ECin sampling fraction cut. He implemented it first.
- is there a reason to discuss momentum corrections in the middle of electron ID cuts? It will be better to move the paragraph that starts on line 48 to the end of the section (if nowhere else it fits) 

- line 79, again, as in momentum correction reference, there should be a reference to BG merging CLAS12 Note 2020-005
- what is the meaning of Fig.6? It would have been informative to have MC point(s) with BG merging to show that the procedure worked. Otherwise, Fig.6 seems unnecessary.   

Line numbering restarts for every page, starting from page 6, so below are line numbers according to that change. 
- page 7, line 14, replace “generated” with “incident”
- page 7, line 32, what is the exact definition of Y!, what means yield, is it not the number of events? 
- page 8, Eq.(5) does not take into account tracking efficiency differences for full and empty target data. It may be a small effect, but it will be good to mention
- page 9, Eq.(6), in the studies for BG merging, it was found that the efficiency slope for electrons is different from negatively charged tracks. So, using such correction and not running MC at 55 nA introduces uncertainty. 
- page 9, line 21, 4\pi is irrelevant for this study, especially in providing a wide kinematic range over W and Q2. It is more appropriate to say, “the nearly 2\pi azimuthal acceptance …”
- page 12, Eq.(16), what is \sigma_{center}, is this one of 231 bins close to the center of (Q2, W) bin? 
- Fig.17, what are the statistical errors on these points?
- Fig.18 and 19, is there a good explanation of cross-section changes after each iteration? For example, after the first iteration, data and EG are well aligned in the first Q2 bin (Fig.18). While the second and third iterations increase in cross-section 
- Fig.20 and 21, what are the uncertainties 

- page 15, line 6, why do you need “for the electron”? Maybe “for the electron scattered …”?
- page 15, line 9, normalization to L/6, it is clear why it is done. But it is a wrong way to explain it. I would argue that you should calculate the cross-section in each sector in a way that you can compare with MC or other data, and then the final product will be an average of 6 independent measurements.   
- page 15, line 15, normal distribution for statistical uncertainties applies to the number of events, is Y=N (see the question on it above).
- page 15, line 32, re-definition of symbol Y, now it is not a number events. Assigning the number of events to symbol Y, naming it yield, and redefining what Y (yield) means is confusing 
- page 16, line 6, what is the average number of events in each (Q2,W) bin? 
- page 16, line 14, in Section V.C, it was stated that the efficiency correction uses the Iterative Bayes deconvolution method. The definition of stat errors presented here is true for bin-by-bin correction, Eq.(10)
- page 16, line 18, another definition of Y, now it includes all the corrections
- page 17, line 24, looking at Fig.3, it does seem that the number of events changes by more than 1% from normal to tight cuts. How procedure was applied? Did the empty target subtraction was done for each cut?  
- page 18, line 6, it claims 1% on average for bin-by-bin and Bayesian difference. In the analysis note, the statement on page 63, line 18 claims a few percent difference between Bayesian and bin-by-bin methods, which is true according to the figures in the note. Q: Why was bin-by-bin not used as a primary for cross-section calculation and Bayesian as a secondary for the check? After all, statistical errors are caused by using estimates based on the bin-by-bin method.
- page 18, line 7, how about the difference between tracking efficiency for full and empty targets? See the comment above. 
- page 18, lines 39 to 40, the statement “each sector has azimuthal coverage varying from 50% of 2\pi at 5 …” is incorrect, it should be “50% of 2\pi/6 ...“
- page 19, line 8, I strongly disagree with the statement that DIS is the main source of the \pi- contamination. Events triggered by \pi- and reconstructed as an inclusive electron scattering are from \pi-s produced by electrons scattered at ~0 degrees, almost photoproduction. CLASDIS is not reliable for the inclusive \pi- rate.  
- page 19, line 65, besides the target length uncertainty, there is also target density uncertainty - how about that?
- page 21, line 37, does not seem like the reference to “nearly 4\pi “ a proper statement for this inclusive electron measurements. Why not say “large electron scattering angle acceptance”?