Mandy: This is mostly about English style-- but there are just a few comments and questions inside it! Line 28 no comma after 'momenta' >>> done Line 29 'scale' should be 'scales'\ >>> done Line 34 say either 'different' or 'various' but not both >>> different Line 46 don't need the word 'further' and it sounds clumsy >>> ok Line 52 'given above' sounds better than 'given before' >>> right Line 55/57 re-arrange sentence ' Therefore the charm and beauty contributions can be disentangled by using observables directly sensitive to th lifetime of the decaying heavy flavoured hadrons' then it reads smoothly >>> done Line 58 suggest a colon after 'heavy flavoured hadron' >>> done Line 60 suggest a semi colon before 'the number of tracks' and again before 'the invariant mass; Otherwise this last sentence of the paragraph is difficult to follow. >>> done Line 68 'B mesons' >>> done Line 73 suggest that we don't need the words 'under consideration' >>> removed Line 76 'and the large statistical correlations' >>> changed Line 82 something looks odd in the spacing of 'data set' >>> dataset Line 98 the flow of the sentence does not need the word 'predictions' before 'using ABMP16' >>> removed Line 102 I think this says the FONLL-C scheme extended by low-x resummation is used, but FONLL-C without low-x resummation is also used, so after the bracket it needs to say 'both with and without low-x resummation'. >>> text changed Line 105/6 I think it would flow better as 'from HERA to make and NLO determination of the running charm and beauty quark masses, as defined in the QCD Lagrangian in the modified minimum-subtraction scheme.' >>> changed according to Peter Tuoel's proposal Line 111 'frameworks for' >>> changed Line 113/4/5 to improve the flow 'in section 4 and in section 5 they are compared.....and VFNS.' >>> done Line 117 'dependence' >>> thanks Line 118 'Finally, the conclusions are given in Section 7' >>> changed Section 2 'combined analyses' Line 126 cut the word 'only' >> done Line 147 'in[26] using scale dependent (running) heavy quark masses' >> o.k. Line 158 the phrase 'in the MSbar scheme' is redundant >> left-over removed Line 168 don't need the word 'also' since you began the sentence with 'In addition' >> o.k. Section 3 Line 213/4 'The different forms of the convolution integral for \sigma** and \sigma** necessitate the consideration of different sets of theory parameters' >> changed Line 216 '....limits to estimate the....' >> o.k. Line 231 'kept fixed' >> o.k. Just one physics question here - the conditions for the HERAPDF1.0 are mostly the same as those for the present analysis, which seems appropriate- but this is not so for the renormalisation and factorisation scales... you don't comment on the extent to which this matters/or doesn't matter? I don't quite see how a corresponding PDf can be used if the settings are not quite the same? >>>>>>>> This not HERAPDF1.0 which is VFNS. This HERAPDF1.0 FF which was used in our previous charm paper and the treatment of scales is consistent <<<<<<<<<<<<< Line 237 comma after 'assumptions' and >> no, the comma is after sigam_vis Line 239 comma after 'tagging' makes the sentence easier to read. >> ja Line 242 'in addition to those needed for \sigma^{th}_{red}' >> added Line 273 'to any order provided these calculations include uncertainties for potenetial deviations from the 'true' result' Section 4 Line 321 ' are reduced significantly - by up to factors of two or more' >> o.k. Line 329 'and these reductions are independent of X_{Bj} and Q^2' >> o.k. Line 337 'and reaches about 15% at small x_{Bj}...' >> o.k. Line 342 'in precision of about 20%..' >> fine Line 343 'reaches about 30% in the range..' >> o.k. Table 2 caption: typo in 'uncertainties' in the last line >> corrected in tab. 2 and 3 Table 4 caption: 'in units of \sigma after the first iteration of the combination' >> o.k. A question the Tables give total systematic uncertainties. Shouldn't we tell the reader that the full correlations are also available, and where to get them at this point in the text?- at least in a footnote, or refer to where we do tell them? >>>>> yes, is included Figures 4 and 5 captions: I think the phrase 'For better visibility' is better than 'For presentation purposes' >> fine Section 5 Line 362/3- don't we also compare to the NNPDF calculation without the low-x resummation? The text reads as if we only compare to the low-s resummation version. >>> we compare to NNPDF calculations referred to as NNPDF3.1sx at NNLO and NNLO+NLLX. In their paper 'NNPDF3.1sx' is used to denote both NNLO and NNLO+NLLX vatriants. Line 372-I suppose you may be asked to justify why the uncertainties of the HERAPDF FF that you consider are only the experimental and not also the ALL model/param and why you include scale uncertainty but not say alphas uncertainty?? >> We compare now to the full uncertainties - footnote is removed. Fig 14 caption: 'similar size as those shown for the FONLL calculations' (no need for the word 'plain' >> o.k. Line 429 'agree well with the previous measurement given that the theoretical calculations show tension in describing the underlying process' >>>>> no, this has a different meaning. This sentence says: Since the measurement is an estimate of 'reality', the measurement comes closer to reality with improved precision. If the theory does not describe reality, then the data will stay away from theory because it approaches reality. Therefore the chi2 will increase quadratically with 1/improvement of the measurement. What is wrong here is 'tension'. We use 'tension' for the consistency between theory and data, not 'reality/truth' <<<<<<<< Table 5 caption: says the \chi^2 and d.o.f are given, but the d.o.f is not given what is given is the number of data points and the p-values. Also in the final sentence of the caption after the comma say 'reducing the number of data points to 47' >> corrected Section 6 Line 439 'The theory description of the x_{Bj} dependence of the reduced cross section of charm production is also investigated' >o.k. Line 473 no need for the word 'independently' >o.k. Line 486 'noticeably' is misspelt > corrected Line 499 'to the inclusive data only fixing...' leave out the word 'with' >o.k. Line 515 'typically a few MeV' >o.k. Line 517 'from all other variations of the parameters' >o.k. Line 534 'The sensitivity to..' >o.k. Line 535 'demonstrate that' (there are two parts to the subject so the verb must be the plural) >right Line 536 'from inclusive HERA data alone' -the word 'data ' is missing...and I thik you mean 'are not sensitive in this framework' athough it is also tru that they are not sensible! >> well, I wanted to say sensible: It does not make sense to try it. - changed to 'not reasonable' Line 540 'by this fit' rather than 'by the fit' makes it clearer >> o.k. Line 544 'especially that of the gluon and because the description of the data is similar...' >o.k. Fig 15: caption 'obtained from the present QCD fit' >> has changed Fig 16: caption 'determined by the present fit' >> has changed Fig 17: same as for Fig 16 >> has changed Line 551 'which are more precise than the heavy flavour measurements in the kinematic region of overlap' >> o.k. Section 6.3 Just a comment while reading paragraph 566/579- it seems to me we have now referred to 'the present fit' or 'the fit of this analysis' many times and it is getting a bit clumsy. Why don't we give 'the fit of this analysis' a name we can refer to it by? Like HERAPDF FFmcmb. >> currently it's HERAPDF-HQMASS Line 581 drop the word 'with' and 'reqiting different values of the minimum x_{Bj' don't repeat the word 'values' >> done Line 582 comma before 'in the range' >> ok Line 589/590 put commas around the phrase 'shown in....at the scale..GeV^2' >> ok Fig 20 caption: 'from the present fit'-unless we give it a name >> HERAPDF-HQMASS Fig 22 caption: 'from the present QCD fits..' the caption mentions full lines for the yellow fit and says nothing about the graphics of the blue fit-doesn't this need to be clarified for black and white? >> this has to be checked - for B/W we need a lighter and a darker colour. Paragraph beginning Line 593 at this point the reader is getting really confused as to which fit is which. If we don't name our fits then we will need to say something like (note the re-arranged order) : Line 594/6 'obtained from the present fit to the heavy flavour data and the full inclusive data set and from the alternative fit in which inclusive data are subject to the cut x_{Bj} > 0.01, to the reference cross-sections of HERAPDf2.0 FF3A' Yes I think you need to specify what the reference cross section is again since the reader has so MANY fits to keep in mind. >> rewritten in the spirit of your proposal. Line 597 'imposed on'...'are rising more strongly' (adverb not adjective) >>ok Line 599 'an x_{Bj} cut' >>ok Line 602 ' three PDF sets discussed in the last paragraph' >>ok Line 603 drop 'also'...Figure 24 rather than Figures >>ok Line 608 'predict' since calculations is plural >>ok Line 612 'it does not seem possible to resolve the ~3\sigma tension in describing..' (no need to say 'in theory' when the sentence starts with 'In the theoretical framework..' >>ok Line 613 'from HERA using this simple approach of changing..' >>ok Line 616 'than that observed at NLO' >>ok Line 617/8 'some tensions in the theoretical description of the inclusive DIS data' >>ok Line 620 'However, a dedicated investigation shows ''' you may be asked to say a bit more about this? Like what did you actually do? >>this is something we have to discuss. In the determination of the PDFs with x_min cut also the masses were fitted and they were found to be consistent within uncertainties. Fig 24 caption: 'from the present fits' >>changed Section 7 Line 647 cut the word 'mainly' - it is said in the text, it sounds clumsy in the conclusion >>ok Line 648 I would also cut 'especially the' here, it reads better, detail can be in the text. >>ok That's it Mandy ++++++++++++++++++++++++++++++++++++++++ Sergey: Dear Karin, I have only few minor remarks to the new draft. 1. Now we use three different "jargon names" for the same theory calculation: -- low-x resummation (e.g. in Intro, line 102) -- log x resummation (l. 432) -- log 1/x resummation (l. 634, Conclusions) Showld it be unified? >> should be consistently: low-x 2. Figures 2,3 look redundant to me. All necessary information is present in Fig.4-6 anyway. >> removed 3. Cosmetics: - I suggest to move column "N_b" immediately after the column "N_c" in table 1. >> done - l. 100: O(as)^3 -> O(as^3) (move power 3 inside the brackets) >>> done - l. 163: here you use \cal{O} for the "order of", while before that (lines 95-108) and after (l. 102) - plain O is used. >>> corrected - Figures displaying PDFs (fig. 15, 22) do not have global "H1 and ZEUS" title. Is that by purpose? To show that we have nothing to do with that? )) >> added - Concerning Fig.20 we already discussed this during T0 meeting: (label ccbar, bbar for sigmas; less number of x-grid lines; etc.) >> changed Sergey +++++++++++++++++++++= Joerg: Dear editors of the paper on combined charm and bottom, congratulations that you could finish this complex analysis. I appreciate very much that this final combination was possible leading to improved HERA results. I wonder why the parameters of the main fit are not given. >> table added (Appendix) The discussion with Fig. 24 I find a bit overdoing. That a quite different gluon distribution does not describe the NC inclusive any more can be expected. >> It's a bit more complicated because there is more than just the gluon in DIS Several points are unclear to me and I hope they can be phrased more clearly. I have many questions and suggestions. At several places I find some confusion of the content of figures, main text and captions. Going through the draft in sequence: 1 Introduction ------------------ 29 scale --> scales >>> done 54, 63 not very clear what the "fragmentation fractions" mean in this context. >>> in 54: the lifetime is different for the different D mesons (ctau=120 mu for the D^0, 150 mu fuer D_s and 380 mu for the D^+/-). For the B mesons the differences are smaller(450-490 mu). The statement is about the average, therefore you have to account for the probability that a given meson is produced (fragmentation fraction) Sentence changed to: The fragmentation fraction weighted average of the proper lifetime of $B$ mesons is about a factor of $2$ to $3$ that of $D$ mesons. in 63: the probabilities to have a leptonic decay differ significantly for the D mesons:( D+:16%, D0,D_S:6.5%)<<<<< 84 meaning a bit unclear due to the length: "suitable for comparison" may be intended to refer to the "consistent dataset", but it reads at first like referring to the cross-correlations, in spite of the comma. 95-103 The acronyms RTOPT, FONLL-C have some meaning, different from acronyms of names like ABKM09. The meaning should be given as you do e.g. for FFNS, VFNS, NLO, NNLO. >>> RTOPT is similar AB...: it's one of the versions of Robert Thorne and it stated that it is VFNS NLO and NNLO 311 suggest frameworks ??? are ---> framework ??? is >>> there different frameworks considered and described - I think the plural is correct 2 Open heavy flavor production in DIS --------------------------------------- 119 + 2 suggest: analyses combined ---> combined analyses (more easy to understand) >>> sentence rephrased equation 1: I find alpha^2(Q^2) difficult to read, at first sight somewhat misleading. Suggest (alpha(Q^2))^2 >> changed 126 Ref[42] gives a calculation for F_L. The way it is quoted here, one expects a solid experimental result. Suggest: In charm production it is expected to reach ??? Or to make it weaker in some other way. >> changed 158 there is an extra "in the MS bar scheme." in this line. >> removed 163 O(m_Q^2) has unusual O. >> changed 168 NLL: "next-to-leading log" somehow clear, especially with next line, but better explained already together with FONLL in line 101. >> has been added to the introduction 3 Combination ??? ------------------ 188 may be somewhat confusing. Not very clear whether data set 8 of present table 1 is superseded, but still listed? Data set 9 refers to present table 1, data set 8 not? >> text changed 204 HVQDIS is a name not a scheme, but I still suggest to write The program for heavy quark production in DIS HVQDIS [43] ??? >> changed 250 fro ---> from >> ok. 253, 254 the difference to 244-252 should be more obvious. It took me quite some time to notice, that now we talk about fractions. May be bring only fractions in bolt or include these lines in the upper bullet. >>>> we think it's clear as is. the discussion on the function concerns our modelling while the uncertainties on the fraction is taken from elsewhere<<< 270-274 somewhat difficult to read. Maybe it helps if in the sentence the theoretical ratio of eq. (3) appears, to make clear that the mentioned cross sections have nothing to do with experiment. Suggest e.g.: While the central values of sigma_red^th(x_Bj,Q^2)/sigma^th_vis,bin (see eq. 3) are obtained ??? Suggest also to write "deviations from the unknown "true" QCD result" to help understanding. >> text changed 287 which compreiseS ---> which comprise >> o.k. eq.4 the delta_i,e,statmu^i,e and delta_i,e,uncorrm^i look like deltas. I wondered some time, why the deltas are the relative uncertainties and not the sigmas. Suggest to write (mu^i,e \cdot delta_i,e,stat)^2 and (m^i \cdot delta_i,e,uncorr)^2. I find this much more easy to grasp. >>> changed 292,293 I have the impression that this sentence is not correct or difficult to understand. What refers "they change" to? What is changing? The uncertainties? The cross sections? Why changing? To say "they are assumed to be proportional to the expected central values" seems more clear to me. Is this what is meant? >> changed to: ... multiplicative nature, i.e. the systematic uncertainties are proportional to the expected cross section values. 4 Combined cross sections ---------------------------- 319 I have problems to understand. I take the second row in table 4. I guess these are mu_r, mu_f variations. Nominal 1 sigma corresponds to factor 2 variation. Following the description of the caption of Table 4: 0.82 sigma = 0.82 * 2 = 1.64. If I shift by that: 2-1.64 = 0.36 Thus the effective (fitted) upward varied scale is mu_r,nominal (1 + 0.36). Looking to the downward variation: 0.5 * 0.82 = 0.41. A shift by that: 0.5 - 0.41 = 0.09 Then the new variation would be from (1-0.09) mu_r,nominal to (1 + 0.36) mu_r,nominal. I am sure this is not what you do. Is it rather that you use as the new uncertainty (1-0.41) mu_r,nominal to (1.64) mu_r,nominal? But then 0.82 is not a shift, rather a scale factor. And what is the reduction factor? The reduction of the uncertainty range? In the last case I get 0.8, not 0.45. So what is exactly done? For the time being I assume that the text should be more clear or explicit (here or in table 4 caption). >>>Let's consider a source of a correlated systematic uncertainty. It may change (shift) the cross section by a certain amount. Before the combination we don't know if it has an effect on the measurement at all. We publish the cross section without applying a correction due to this source and we estimate/assign an uncertainty on the measurement due to this source. Now, in the combination the different measurements have different sensitivities on this source. Since we assume that all measurements measure the same cross section, we may get information by how much the cross section has to be shifted in terms of the assigned uncertainty of this systematic source. This is the shift we quantify in sigma of the original uncertainty of this source. At the same time it may happen, that due to the combination the uncertainty of this source is better known. This is the reduction. Take an example: we measure 1.00 and the uncertainty due to the source is 0.05. in the combination we find we should apply a shift due source of say +0.6 sigma this means the cross section is now 1.00+0.6*0.05. Our knowledge on this source may also have improved say by 0.8 sigma this means: Initially we had 1.00 +-0.05(corr) this changes to 1.03+-0.04(corr). So even if we have only one measurement at a given grid point we may have learnt by the combination that the cross section should not be at 1.00 but at 1.03 and the uncertainty is reduced from 0.05 to 0.04. This is why also cross section for which only one measurement exist may change after conbination. We tried to improve on the text <<<<<< 321 "reduced to factors 2 or larger": suggest "reduced by factors 1/2 or smaller" >> native English speakers prefer: by up to factors of two or more 5 Comparison with theory predictions ------------------------------------------ 397 see points to Fig. 10 (11), 12 (13) below. 412 prefer here: which aims FOR >> o.k. 420 To the data? >> 'to compare to' is fine (To my understanding 'with' would put both at equal levels, but the reference is data - will check with native English speakers. 421 "p-values" is a bit jargon like. fit probabilities? 429 "if"? Logic? Meant like "as"? >>>>>no, 'if' is correct.This sentence says: Since the measurement is an estimate of 'reality', the measurement comes closer to reality with improved precision. If the theory does not describe reality, then the data will stay away from theory because it approaches reality. Therefore the chi2 will increase quadratically with 1/improvement of the measurement. What is wrong here is 'tension'. We use 'tension' for the consistency between theory and data, not 'reality/truth' <<<<< 6 QCD analysis ---------------- I wonder why the PDF parameters of the central fit are not given. The paper is long and explicit and presents the detailed equation (6), but not the results. >> added, Tab. A.3 of the appendix 457 The chi^2 definition is a central issue of the analysis and should be given fully within the present paper explaining the difference with respect to eq. (4). >>>> these are two completely different quantities eq. (4) is the definition of chi2 used for the combination. Here it's the chi2 of the QCD fit data. The EB decided not to add more information here, because it can be found in ref. [40] <<< 466 Naive question: where are the non-valence, non-anti u(x), d(x), s(x)? >>> the total xu(x) = xu_v(x)+xubar(x) and similar for the total xD(x) a sentence is added <<< 473,474 Procedure is unclear, because you include the parameters "independently" and only "one at a time". How can the procedure then arrive at more than one of these parameters? >> "independently" is removed. Text changed: The parameters in equation~(\ref{eq:genericPDF}) are selected by first fitting with all $D$ and $E$ parameters set to zero, and then including them one at a time in the fit. The improvement in the $\chi^2$ of the fit is monitored. If $\chi^2$ improves significantly, the parameter is added and the procedure is repeated until no further improvement is observed. <<< 479 what is the basis of this alpha_s variation? A reference? 514 suggest: effects on the model uncertaintieS ---> effects on the model uncertainty (the somewhat difficult sentence more easily to understand) >>ok 527 a bit disappointing that we just state that we have done something and claim agreement but give not the result of the study. >> they were given in the meetings, but it was decided, not to give values, because this puts too much weight on it. 536-578 I do not understand: The uncertainties ??? are covered? What I see is that the result 1.8 +0.14-0.13 is quite far away from 1.29 (even more from PDG value). This indicates some inconsistency in the fit of inclusive data only. So the reader regrets not to get other uncertainties beyond "fit". >>> This was an exercise to see what happen if we use the inclusive data only and we conclude in line 536 that it is not sensible to perform a fit to the masses with inclusive data only. The default PDF parameterisation yields 1.8 GeV when fitting the inclusive data only. However, when setting E_u_v=0 this changes to 1.45 and even more dramatically for mb (inclusive DIS + default PDF par.: 8.45, with E_u_V=0: 4.00) This is why we say it does not make sense to use inclusive DIS only. We don't give more than the fit error because the values are not a result of the paper. The result is: Don't do it! Because we see a large effect in this study we included in the default fit scenario (heavy flavour + inclusive) an extra variation setting E_y_v=0, which is domination parameterisation uncertainties. The sentence 536-538 is misleading. It is changed. The large effect on the fitted masses observed here, when setting $E_{u_v} = 0$, motivates the $E_{u_v}$ variation in the \pdfhq~fit. <<<<< 545 one could also add that the fitted quark masses are not far from the ones previously used. >>> this was already said in lines 517-520. 549-552 A bit difficult to understand: "which are more precise"? More precise than charm and beauty data or more precise in this kinematic region? >>> changed to: There are only marginal differences between the two calculations because of the dominance of the high precision inclusive data. You mention then the "few per cent" differences, but in spite of "dominance of inclusive" data there are sizable differences in the region of large x in Fig. 19, where there are no bottom (and charm) data. A comment? >>>> You cannot compare x_Bj of inclusive data with that of heavy flavour. They are testing different partonic x. For beauty you can see this at small Q2, where the shaded band stops. At Q2=2.5, x_bj=0.02 corresponds to x=1. There we also don't have inclusive data. The deviations you see between the different calculations (beyond the few per cent) are in the x-range where we don't data (neither HQ nor inclusive) <<< 573 Unclear what this average for a given x,Q2 point means. You average over the acceptance of the detector contributing to the cross section? Fig. 20 and caption tell nothing about averaging. >>>>>>> the paragraph is rewritten also explaining why the average is needed and what it is<<<<<<<<<<< 577,578 it may still be consistent with being independent of Q2, but Q2 > 200 does not support this and looks different. >>>>> When trying to fit a constant value you get: chi2=89.1 ndf=51 ->probability 0.0007 ->3.4 sigma this is why we conclude that is deviates from the reference. the behaviour at high q2 does not follow the trend at small x, but why should it be a monotonic deviation. the point at x=0.01 and q2>200 fits well to the other measurements in this x-range. <<<< Footnote 7 "the data have been used in the fit": this is always the case. Suggest "the data have been used in the fit to adjust theoretical uncertainties. Therefore the theory???" >>ok 587 if this shallow minimum of c+b only is worth mentioning, then we can not say that DIS+c+b is increasing in this region, it is rather falling again. >> changed to: while the $\chi^2/{\rm d.o.f.}$ for the inclusive plus heavy flavour data sample slightly larger than that obtained without cut in \xbj. 588 it would be convenient to have the degrees of freedom together with the 91 here available. >> is unchanged because there is no cut on the heavy flavour data, given in line 562 but added again 602-608 I find Fig. 24 a bit overdoing. It is fairly obvious that a much steeper gluon distribution as shown in Fig.22 has also influence on the inclusive NC cross section. I suggest to skip this plot. >> the EB decided to keep the figure 613 it reads a bit naive that one may expect to describe the inclusive NC data using a gluon distribution which is clearly outside the uncertainty band of NC gluon fits. >> it depends on what can be compensated by the other partons. 620, 621 Some indication what in this "dedicated investigation" is done? Now it reads a bit strange. The mass measurement is considered by the authors as important, they show some problems with theory and then just tell that these are unimportant for the main result (at least the abstract gives the impression that this is the main result). >>>>> The fits with x_Bj cut have been performed with free masses. for x>0.01 we get mc=1.40+-0.06 GeV and mb=4.1+-0.1 GeV. this is within uncertainties for beauty and about 2 sigma for charm. we have this 3 sigma tension. it is difficult to quantify something <<<<<<<<<<<<< 647 reads a bit naive, see 613. >>> not really. The change in the gluon could have been small enough to get agreement, however, we need such a steep gluon for charm which is disagreeing with what is needed for the inclusive data. Table 1: why N_b not put directly right of Nc? caption: Tagging not mentioned. It should also get some explanations, especially VTX. >>OK Tables 2,3: caption: I am afraid, also Q2 and x must be mentioned >> normally we don't do it Table 4: second row: theory, scales. I guess these are mu_r, mu_f. Should be indicated for clarity. Caption: reduction of what? Suggest: reduction of correlated uncertainties. Also main text is short here. See also comment to line 319. >>OK Table 5 : Suggest to add in first column to "HERA 2012 c" the reference [36] (to make understanding more easy). But the publication is 2013. >> Reference added Prefer in first column "Present" instead of "New". changed: charm[36], charm, beauty caption: fit probability should be mentioned. ("p-value" as in text is a bit jargon like. at the end: reduces --> reduced (or no "is") >> p-value introduced in the text now Fig.2: the data points are rather large, so the inner error bars and even the outer ones are mostly invisible. Suggest either smaller dots or different symbols e.g. x. >> figure removed - (old figure 4) it does not make sense to use smaller symbols - the data are too precise - this is the message of this kind of plots. A blowup is shown in (old) figure 6 for q2=32 Fig. 10 and 12: legend and colours consistant? Comparing the figures, I think that in Fig. 12 the colour choice for appr. should be the same as in Fig.10. Is the labeling correct?. I expected to see the drop of appr. NNLO at large x also in Fig. 12(?). Similar for Fig. 11 and 13. >> figs. 10 and 12 compare different things (theories) - no need that the coding is the same. Why should approx. NNLO drop at large x relative to the reference?<<<<<< Fig. 12 caption: last word: calculationS ---> calculation Only one is presented with uncertainties. >>OK Fig. 13 caption: similar size AS those last word: calculationS ---> calculation >>OK Fig. 15: The legend DIS+c+b with yellow fork suggests that also the yellow curve is a band. Should be removed. Legend DIS only is too close to mu_f^2 This x Fitter Logo needed for license reasons? Can it be removed? >>OK caption: I see no broken lines. The blue error band is given according the figure for DIS only, but the caption tells that inclusive DIS has no error band. >>corrected "experimental/fit" looks like either or, or respectively. What is actually meant? Figs. 16, 17 solid and dashed the other way round than in caption. Also for error band plot, legend and caption not consistent. >>OK Figs. 18,19 Which uncertainty do we want to show? From the new fit or of the reference? >> we show consistently the uncertainties for the reference calculations - the uncertainties are almost identical - it would be too confusing changing it form figure to figure. Fig. 20: the caption tells nothing on averaging x in contrast to main text >> changed Fig. 22: "full lines"? Both cases have full lines. Fig. 24: Better: Combined reduced NC cross sections .. but the abbreviation NC is not yet introduced. The dashed dotted line within the red band is hardly visible. See also comment at 602-608 >> NC now introduced in the text - fewer panels now to improve visibly of curves Have a nice time in advent season, Joerg ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Brian General comment. I find this paper very difficult to understand. It has too much detail and the important results are obscured. I have not attempted to make issues of hyphenation and punctuation consistent throughout but have tried to correct the most obvious inconsistencies where they cannot be ignored. The introduction is far too long and detailed. Much of the text belongs in the experimental method not the introduction. >>>> The experimental methods are now described in section 3 line 21: no hyphen in ???deep inelastic??? >> removed 28: no comma after ???momenta??? >> removed 29: ???hard scales??? not ???hard scale??? >> changed 32,33: no hyphen before ???scheme??? >> removed 34: either ???different??? or ???various??? but not both >> different 36: ???longevity??? is the wrong word in this context. ???lifetime??? is probably clear enough. >> changed 47: ???that??? not ???which??? >> changed 48: ??????is small, so that the mesons?????? >> changed 54: I don???t understand what is meant by ???fragmentation fractions??? here - should it be ???fragmentation functions???? >>>>>>>> fragmentation fraction is correct. It is the probability that a heavy quark fragments to a specific heavy meson. Especially the proper lifetime of the D-mesons are quite different ranging from 120mu for D0 to 380mu for D+. The lifetime of B mesons ranges from 450 to 490 mu. For giving an average factor one has to account for the probabilities that a specific hadron is created, this is given by the fragmentation fraction. sentence changed to: The fragmentation fraction weighted average of the proper lifetime of $B$ mesons is about a factor of $2$ to $3$ that of $D$ mesons. <<<<<<<< 55: "Therefore the charm and beauty contributions can be disentangled by using observables directly sensitive to the lifetime of the decaying heavy flavoured hadrons??? >> ok, changed 58/59: "of the particle with lifetime information w.r.t. the flight direction??? doesn???t make sense >>>>> well, that's what it is. The impact parameter of a particle produced in the decay of a heavy flavoured hadron is defined w.r.t. the flight direction of the decaying heavy hadron. Technically the flight direction is either approximated by the jet direction (H1) or by the direction of the momentum sum of the particles attached to a reconstructed secondary vertex (ZEUS if I'm not mistaken) <<<<< 60: neither does " the number of tracks with lifetime information??? - is what is being tried here to give pairs of variables that can be used? If so, then why does the last one, the invariant mass, not have a second variable? >>>> Well, if you read the paragraph carefully, it says that lifetime information is used and the separation is improved by something in addition: ptrel, # of tracks with lifetime information (H1) and the invariant mass calculated from the charged particles attached to the secondary vertex (ZEUS) (it's an enumeration) <<<< 62: no comma >> removed 65: ???B mesons??? >> done 66: comma after ???mass??? >>done 67: delete ???being??? >> done 72: comma after ???sample???; ???phase space limitations??? is a vey strange concept - replace with ???losses??? or ???inefficiency" >> losses 74: ???Fully inclusive or semi-inclusive lepton analyses, which are sensitive to both charm and beauty production, profit from larger???.polar angle. They are however affected by a worse signal-to-background ratio and???.??? >> thanks, changed 78: comma after ???paper??? - ???a simultaneous determination??? - a simultaneous combination is meaningless >> comma added >>>>>>> concerning combination or determination: in any of the VTX or lepton analyses from H1 or ZEUS a simultaneous determination of the charm and beauty cross sections is done - now we combine the data and this is done simultaneously for charm and beauty (as we did for NC and CC and high-y NC in the inclusive paper). You could also combine the charm data and beauty data separately. In this case we would not account for the correlations between charm and beauty inherent to the experimental methods. This we did not do. I thing combination is the right term to be used. <<<<<< 80: comma after ???[36]??? >> done 81: comma after ???result??? >> done 84: delete ???suitable for???.??? >> o.k. - and sentence split into two sentences 86: ???The procedure use is based on that described in ?????? >> o.k. "The procedure used is..." 90: ???..this procedure leads to a significant reduction of systematic uncertainties.??? >>>>> also the statistical uncertainties are significantly reduced. <<<<<<<< 98-100: ???respectively??? cannot be used like this. Replace with: "In addition, QCD calculations in the RTOPT VFNS at NLO [32] and NNLO are compared with the data. The NLO calculations are at O(??^2_s ) for PDFs and massive parts of the coefficient functions, O(??_s) for massless parts of the coefficient functions; the NNLO calculations split identically but are one order of a_s higher.??? Why is there no reference for the NNLO calculation? If it is part of [32], move the reference to the end of the sentence. >> changed accordingly - [32] moved after NNLO 110: comma after ???2??? >> done 111: ???framework??? not ???frameworks??? >>>>> it is not a single framework but different frameworks <<<<< 112: Section 3, not 3.1 >> o.k. 115: comma after ???6" >> done 116: ???charm and beauty quarks??? >> done 118: ???Section 7 contains the conclusions.??? >> sentence changed 120: I don???t understand what ???Open??? is supposed to imply in this context - it is just confusing. We don???t mention quarkonia, for obvious reasons. Delete. >> fine 119.1??? ???deep inelastic??? >> ok 119.2: ???the analyses combined??? makes no sense. Replace by ???the combined analysis??? >> changed 119.3: ??????.exchange dominates??? >> ok 124: comma after ???addressed??? >> ok 129: comma after ???QCD??? >> sentence removed 131: ???are realised??? is not correct. Probably ???can be used??? is what is meant >> ok 136: a correct theoretical treatment is always mandatory. Replace with ???a careful theoretical??? >> ok 139: comma after ???paper??? >> ok 140: ???the??? before ???full???- delete the comma before ???and???. Comma after ???scheme" >> ok 147: comma after ???[26]??? >> ok 148: comma after ???paper??? >> ok 150: delete ???respectively??? - I have no idea what it means in this context. >> ok 157: comma before ???heavy??? >> ok 158: delete ???in the MSbar scheme??? >> done 161: what is meant here by ???approximate NNLO??? ? >>>> massive coefficient functions are known only approximately at NNLO The sentence starting at 160 is completely impenetrable. Rewrite. >> sentence rewritten 163: comma after ???schemes??? >> ok 164: comma before ???with??? >> ok 165: comma after ???calculations??? >> ok 178: don???t use ???different??? and ???individual??? - it sounds like we want to mean different things - use either ???different??? twice of ???individual??? twice. >> individual 180: "from both??? not ???both from??? >> o.k. 186: ???datasets??? >changed 187: ???are included for the first time in this analysis??? >> o.k. 187: no comma after ???note" >> o.k. 188: ???dataset??? - and throughout decide whether it is ???data set??? or dataset??? and use whichever consistently! I prefer ???dataset??? >> changed to dataset consistently 192: comma after ???measurements??? - this sentence is however unintelligible. Rewrite. >> rewritten 195: ???In the case of inclusive D meson cross sections, small ??? >> o.k. 197: delete ???the removal of??? - and in 199 >> o.k. 203.1: ??????.. in the full phase space?????? >>o.k. 205: replace ???in terms of??? by ??? as a function of??? >>o.k. 208: ???in pQCD, \sigma etc etc??? >>o.k. 211: comma after \sigma^h_vis >>o.k. 212: ???In the case of???programme, non-perturbative... ??? >>o.k. 223: comma after ???constant??? and after ???chosen" >>o.k. 226: ???No heavy flavour measurements were included in the determination of these PDF sets??? >>o.k. 230: comma after ???PDFs??? >>o.k. 231: ???..was kept fixed at?????? >>o.k. 233: I don???t understand what this sentence means >> rewritten 235: ??????[40]; the differences are found to be smaller than the cross-section uncertainties??? >>o.k. 247: comma after ???system" >>o.k. 249: what does ???have been transported??? mean? >> adapted 250: "Transverse fragmentation is modelled by assigning to the charmed hadron a transverse momentum kT with respect to the direction of the charmed quark?????? >> changed 259: no hyphen >>o.k. 263: ???PYTHIA??? >>o.k. 263: Why is this bullet in past tense while the others are present? In my view ALL these bullets and indeed all the text in this section should be past tense - they describe what we did. >> changed 272: I have no idea what the sentence starting ???The resulting reduced???.??? is supposed to mean. 278: comma after ???[36]??? >> o.k. 284. This sentence is impossibly complicated. Here is an attempt to split it up, but I am not sure I understand precisely what it is trying to say: "The three sums run over the input data sets e listed in table 1. The (xBj,Q^2) grid points i for which the measured cross sections ??^i,e are combined with the cross sections m^i. The sources j of the shifts b_j are in units of standard deviations of the correlated uncertainties, which are obtained from the correlated systematic uncertainties and the statistical correlation between the charm and beauty cross section measurements.??? >>>> sentence changed (not taking your proposal, because it does not describe what we were trying to say <<<<< 290: "The components of the vector m are the combined cross sections m_i, while those of the vector b are the shifts b_j.??? >> changed 292/3: ???In the present analysis, the correlated and uncorrelated systematic uncertainties vary proportionally to the expected central values." >>>> changed to: In the present analysis, the correlated and uncorrelated systematic uncertainties are predominantly of multiplicative nature, i.e. they are proportional to the expected cross sections $m^i$. <<<<< 296: comma after ???table 1??? >> o.k. 297: comma after ???necessary??? >> o.k. 305/6: commas before ???together" and after ???uncertainties??? >> o.k. 308: delete ???respectively??? >o.k. 309: comma after ???combination" >>o.k. 1st paragraph of section 4- don???t repeat information - delete either the first mention of conservative estimates of the uncertainties of delete the last sentence of the paragraph. >> first "conservative.." removed 318: comma after ???4??? and after ???listed" >> o.k. - sentence changed 321: ??????.reduced by factors of 2 or more.??? >> changed according to Mandy's proposal 323: comma after the close bracket >> o.k. 326: ???combination; some are further significantly reduced due to the inclusion of new precise data [19-21].??? >> o.k. 328: comma after ???observed??? >> sentence changed according to Mandy's proposal: Only small reductions in the range of $10\%$ are observed and these reductions are independent of $\xbj$ and $Q^2$. 329: what is independent? >> sentence changed according to Mandy's proposal: Only small reductions in the range of $10\%$ are observed and these reductions are independent of $\xbj$ and $Q^2$. 334: comma after ???6??? >> o.k. 347: comma after ???sections??? >> o.k. 348: ???Predictions using??? not ???predictions of??? >> o.k. 351: ???In the case of VNFS, recent calculations???.??? >> o.k. 357: comma after ???9??? >>o.k. 372: comma after ???set??? >>o.k. 377: comma after ???precise??? >> o.k. 378: comma after ???considered??? >>o.k. 380: ???For beauty production (figure 9), the predictions?????? >>o.k. 385: comma after ???GeV^2??? >>o.k. 386: comma after ???region??? >>o.k. 387: comma after ???x_bj" >>o.k. 388: comma after second ???GeV^2??? >>o.k. 389: comma after second ???x_bj??? >>o.k. 391: comma after ???11??? >>o.k. 392: comma after ???HERA??? >>o.k. 393: comma after ???uncertainties??? >>o.k. 395: comma after ???12??? >>o.k. 397: comma after ???10??? >>o.k. 399: delete ???within their uncertainties??? >>o.k. 400: ??????differences, about 10% smaller than the reference???.??? >>o.k. 405: ???In figure 13, the same ratios discussed in the preceding paragraph are shown for beauty production." >>o.k. 406: comma after ???HERA??? Paragraph beginning at line 408: What is actually being used here, and what is shown on Fig. 14? The key to Fig. 14 shows "NNPDF31sx_nnlo_as_0118 FONLL-C??? and "NNPDF31sx_nnlonllx_as_0118 FONLL-C??? but the text says "(NLLO+NLLsx) and without (NNLO) low-x resummation??? This doesn???t seem consistent. Should it perhaps say ???NLLO and (NNLO+NLLsx) without (NNLO) low-x resummation???? That at least is consistent with the figure caption and matches my recollection of what Ball et al. actually did. Anyway, I don???t understand what we say about this figure. ???However, the predictions lie significantly below the data in most of the phase space.??? That is true for the dashed purple line, presumably the leading log one, but not for the solid purple band, which is generally speaking a better fit to the data than either the dashed or the HERAPDF band - at least up to 32 GeV^2, after which there is little to choose between them. So a) I don???t know what models the text is actually commenting on and b) irrespective of that, what we have written is wrong. 413: ???The charm data from the previous combination have already been used for the determination of the NNPDF3.1sx PDFs.??? >> o.k. 417. I don???t understand the sentence ???Overall, the description is not improved?????? It surely is from Fig. 14 alone - does ???Overall??? mean looking at other variables that we don???t discuss? 425: comma after ???combination??? >> o.k. 426: comma after ???4??? >> o.k. 428: I can???t understand the sentence starting "The observed changes?????? >>>>> This sentence says: Since the measurement is an estimate of 'reality', the measurement comes closer to reality with improved precision. If the theory does not describe reality, then the data will stay away from theory because it approaches reality. Therefore the chi2 will increase quadratically with 1/improvement of the measurement. What is wrong here is 'tension'. We use 'tension' for the consistency between theory and data, not 'reality/truth' <<<<< 430: comma after ???considered??? >> o.k. 432: comma after ???sections??? >> o.k. 434: comma after ???cases??? >> o.k. 439: ???The theory description of ???.production is also investigated." >>ok 442: comma after ???program??? >>ok 452: ???usually??? rather than ???commonly??? >>ok 453: comma after ???applicable??? >>ok 454: comma after ???data??? >>ok Why are we showing Figs. 18 & 19 anyway? As far as I can see, the data in there and the HERAPDF fit is identical to Figs 13 and 14 . The only addition is Figs 18 and 19 is the purple dashed line labelled ???NLO fit??? in the legend, but this is referred to neither in this paragraph nor in the figure caption. What in fact is the chisq between the HERAPDF fit and this data - it looks pretty terrible - yet we make no comment about it - nor did we for Figs 13,14? And finally, why is the order of Figs 18, 19 reversed compared to 13, 14? >> 18,19 are charm and beauty compared to the fit result performed in this paper, 10/11 shows the comparison to FFNS (HERAPDF2.0 FF3A, ABMxy), 12/13 shows the comparison to VFNS, 14 is NNPDF, which exists only for charm The figures are referred to in line 548, the caption is a bit misleading, now the fit is termed HERAPDF-HQMASS the order is not reversed 459.2: comma before ???the density??? >>ok 460: "where x is the momentum fraction transferred to the struck parton in the infinite momentum frame of the incoming proton.??? >>ok 475: delete the comma >>ok 476: comma after ???[40]??? >>ok 486: insert ???that??? before ???with???. >>ok 486: ???..parameter; changing this parameter noticeably affects the mass determination.??? >>ok 487: ???furthermore??? doesn???t make sense here - perhaps you mean ???In addition???? >>ok 488: ???/... uncertainty is determined at each x_Bj value from the maximum differences ?????? >>ok 489: delete the sentence "This uncertainty???.??? - this is obvious and adding this statement just makes the reader wonder why it is there. >>ok 494: comma after ???data??? >>ok 498: comma after ???15??? >>ok 498: I am having to guess what is meant here - I think it should be ??? Also shown are the PDFs whose experimental uncertainties arise from a fit to the inclusive data only, with the heavy quark masses fixed to their PDG values [51]. No significant differences between the two PDFs are observed.??? >> first sentence is not what was intended to say. - Sentence rephrased. 501: ???functions??? doesn???t make sense. Do you mean ???regions???? >> no, rephrased 502: ??????density, a slight enhancement compared to that determined from the inclusive data only can be observed around x~~ 10^-3 when including the heavy flavour data in the fit.??? >> rephrased Actually, just delete from 501: "When comparing?????? to 505: ??????uncertainties???. If it isn???t significant we shouldn???t be commenting on it. >> no, this is already pointing to the region where we expect charm to contribute, which is discussed in the next subsection - sentence starting with "However" removed 506: comma after the first ???data??? >>ok 508: comma after ???analysis??? >>ok Footnote 5 - this has to be a proper sentence : ???This did not include scale???.??? >> changed 513: comma after ???mass??? - ???also yields??? not ???yields also??? - comma after ???contribution??? >>ok 515: delete ???of" >>ok 524: delete ???given??? - comma after ???uncertainties??? >>ok 529: comma after ???case??? >>ok 530: comma after ???parameters??? >>ok 531: comma after the second ???GeV??? >>ok 533: comma after ???0??? >>ok 534: ???to??? not ???on??? >>ok 536: ??????from inclusive HERA data alone?????? >>ok 537: unclear what ???covered??? means here >>ok 539: what ???PDF set???? >> the PDF has a name now 539: This sentence is too complicated - there are 3 separate ???combined???s involved. >> rephrased 543: ?????? can be observed. This is to be expected because of the similarities of the PDFs, especially of the gluon. The description of the data???.??? >>ok 546: Another impossibly complicated sentence! ???Figure 18 show the ratio of data and predictions from the results of this analysis as well as the ratio of data to predictions based on the fixed HERAPDF2.0 FF3A PDF set, for charm quarks. Figure 19 shows the same ratios for beauty. ??? >> sentence changed 549: delete the rest of this paragraph after ???calculations???. >>ok 556: This section seems to me to have the wrong title - almost all of its content is about various fits to all the DIS data and c+b as a function of x_Bj. I don???t understand the thrust of the discussion either. It seems to me that Fig 14 indicates that the purple band of FONLL-C is an improved description giving a reasonable overall chisq for the charm data - and presumably from the inclusive data, from which it was mostly derived. It is fine to examine why our HERAPDF formalism doesn???t work well, but we leave the impression that QCD fits are failing here - while it seems as if at least one approach, FONLL-C, gives a reasonable description. >> indeed, FONlL-C does describe the x-shape better a low Q2 but is off in normalisation and/or has a different shape in Q2. In table 5 you can see, that the chi2 is very poor for both versions of FONLL-C overall these calculations give the worst description of the data. The title is misleading and has changed 559: delete "with the fitted parameters and the PDF parametrisation chosen.??? >>ok 560: comma after ???[36]??? >>ok 563: the font seems to have changed briefly here >>corrected 563: no hyphen >>ok 564: ??????paper. All calculations???/??? >>ok 567: "The contribution to charm production at HERA arising from light flavours amounts to five to eight per cent ???' >>ok 570: ???accessible by" is wrong but I don???t know if you mean ???accessible to??? or ???accessible from??? >> sentence rephrased 575: ???...the beauty data is limited to a higher x range, 0.004??? 0.1 because of???."' >>ok 576: comma after ???data??? >>ok In fact, I can???t tell anything of the sort from Fig. 20 - what "steeper slope??? is referred to in line 577?? Figure 20 is very difficult to interpret. >>>> When you look at figure 18 you see a clear difference in slope between data and calculations for Q2=12 GeV2. Then people start to argue: well everything else is consistent. The message is partially spoiled because x_Bj is quite different from x for heavy flavours. When going from x_Bj to x you see in figure 20 that the slopes in the region 0.0005>ok 578: " Due to the larger experimental uncertainties, no conclusion can be drawn for the beauty data.??? >>ok Footnote 7: ???...p-value given here do not???.??? >>ok 580: comma after ???function" >>ok 581: delete ???with??? >>ok 587: delete ???the??? >>ok 592: Delete ???within experimental uncertainties.??? >>ok 593: "In figure 23, a comparison is presented of the ratios of the combined reduced charm cross section, ?? cc red and the cross-section predictions obtained from the fit to the heavy-flavour data and the inclusive data fulfilling xBj ??? 0.01 to the reference cross sections. The predictions from the fit to the heavy flavour data and the full inclusive data set are also shown. As expected, the charm cross sections??? inclusive data rise more strongly towards ???" >> rephrased 598: ???In general, the ..??? >>ok 599: "A similar study for beauty was also made but no significant differences were observed.??? Delete the rest of the paragraph. >>ok 603: ???In figure 24, these predictions are compared to the inclusive reduced cross sections ?????? >>ok 605/6: ???... obtained in this analysis by the fit to the combined heavy flavour and inclusive data agree with the inclusive measurement.??? What? Is there some distinction here between ???inclusive data??? and ???inclusive measurement???? Is this too different sets of inclusive data? Please clarify! >> no, rephrased 608: ???predict??? not ???predicts???. ???Larger??? - than what? >>ok 609: Delete comma >>ok. 610: "within the framework for PDFs applied by excluding the low-xBj inclusive data in the fit.??? What does this mean? What framework for PDFs? And what is being applied? >>>> We don't have the slightest clue what the physics behind PDFs is. We simply do some ad hoc assumption and choose a more or less clever parameterisation of our lack of understanding. We have chosen the form detailed in section 6.1. This is our framework which is then applied in the calculations/fits 612: ???3 sigma tension in theory??? Does this mean that the tension is theoretical not real? I can have a guess at what this is supposed to mean: " In the theoretical framework used in this analysis, it seems impossible using only variation in the gluon density to resolve the ??? 3 ?? tension between the fits to the inclusive and charm data.??? However, it does seem possible for FONNL! >>> rewritten - No, FONNL is 5 and 7 sigma off for the two schemes 614. Delete the rest of the paragraph starting from ???However???.??? and replace with ???As shown in section 5, this tension between the charm and inclusive data is unlikely to be resolved at NNLO, which gives a worse fit to the charm data. >>> to be clear: there is no tension between the data. Within uncertainties the data describe 'reality' The tension is in theory in describing both samples simultaneously. And this is true for all approaches considered in this paper. This doe not mean that it is a failure of NLO/NNLO QCD. This may be perfectly o.k. It may be related to the way we use QCD with parameterising our ignorance by PDFs. In order to avoid the impression, that we claim QCD is wrong (at 3 sigma), we use the term 'theoretical framework' <<<<<< However, the quark mass measurements are not significantly affected by ???.??? By what? The current text implies that we have investigated the ???tensions??? but it is unclear how or what we have investigated. Please clarify. >>>>> The fits with x_Bj cut have been performed with free masses. for x>0.01 we get mc=1.40+-0.06 GeV and mb=4.1+-0.1 GeV. this is within uncertainties for beauty and about 2 sigma for charm. we have this 3 sigma tension. it is difficult to quantify something <<<<<<<<<<<<< 623: no hyphen >> ok 624: Now that we are actually summarising what we HAVE written about what we HAVE done, to use the present tense is incomprehensible. ??????experiments have been combined???. The beauty cross sections have been combined???.??? 629: ???..combined data have been compared?????? Hyphen between ???leading??? and ???order??? 630: ???the charm data. The beauty data, which have larger experimental uncertainties, are well described by the QCD predictions.??? >>ok 632: ?????? The next-to-leading-order calculations in the fixed-flavour-number scheme >>ok We cannot end this very important paper with an inconclusive discussion about disagreements between inclusive and charm QCD fits at NLO and NNLO. I serious doubt whether this whole subject should be included at all since our discussion of it in the main body of the paper seems to me to be confusing and inconclusive. It detracts from the important results of the paper. If it remains, then the order of the final two paragraphs must be swapped so that we end with the determination of the running masses. >> the EB decided not to change the order in the conclusions 636: ???HERA are analysed in next-to-leading-order QCD in the fixed-flavour-number scheme ??? >>ok 641: "The QCD analysis reveals some tensions in describing both the inclusive and the charm HERA DIS data in the same fit.??? >>rephrased 643: delete ???the theoretical framework of??? - ???A study in which inclusive data with x_bj < 0.01 were excluded from the fit was carried out.??? 645: ??????could be achieved in this way. However, the resulting PDFs fail to describe the inclusive data in the excluded xBj region, a situation that is not improved at higher orders in QCD.??? Delete the rest of the paragraph from ???This points???." >>> Sentence starting with "this points..." is removed. Last sentence is kept, because all calculations fail at least at the level of 3 sigma, not only NNLO. Reference [32] is missing authors' names >>OK Reference [44] - there should be an ???and: before the third author???s name in both references >>OK Reference [59] Sjoestrand???s name is incorrectly spelled. >>OK Caption to Table 1: ???..For each dataset, the Q^2?????? >>OK Caption to Table 4: add comma after ???For each source??? and after ???simultaneously???. Delete ???to the data set number??? >>OK Caption to Table 5: ???is reduced??? not ???is reduces??? >>OK Caption to Fig. 4, 5. Shifted by what amount in x_bj? Caption to Fig. 6. Shifted by what in x_bj? ???Towards larger values??? makes no sense - delete. >> it's important to know, that they are always shifted to larger values - otherwise, there are a few input data points where it may be not obvious where belong to. The captions until Fig 5 had commas around the \sigmas. Make the captions of Fig 7 and the rest consistent and add commas. >>OK Captions to Figs. 12, 13: ???They are of similar size to those presented for?????? >>OK Caption to Fig. 14 - the text is completely unintelligible unless one is also looking at the main text. This is not how figure captions should be written. The models shown should be spelled out, not using acronyms at least the first time they are used, and if necessary a reference should be given. >> this is not different from other figures - text rewritten. Caption to Fig. 15 - there aren???t any dashed lines - just diagonally shaded.I don???t understand the last sentence - so what uncertainties are shown? And how? >> figure/caption changed - uncertainties are experimental (as the caption says) Captions to Figs. 18 & 19 - what is the purple dashed line labelled ???NLO fit??? and why is it not referred to either in the caption or the main text? >> it from the fit performed in the paper, which is termed HERAPDF-HQMASS now - it is in the caption now Figure 20 is not a proper figure - it should be labelled a) and b) and it should be stated in the axis labels that the top one is charm and the bottom one beauty. The right hand labels should be indicated to be Q^2. The caption should be changed accordingly. >>changed Caption to Fig. 21 - again, the caption can only be understood in conjunction with the main text - we need to spell out the details, such as what ???the fit??? is. >> try to improve on the text Figure 22. Please label on the figure that there are u_v, d_v etc - otherwise they are useless for people giving talks. >> sorry, they ARE labeled (y-axis labels) the same way as in other H1+ZEUS publications when showing them in separate panels. Figure 24 is too small. We make ourselves look foolish by pretending we believe people can see the difference between dashed and dashed-dotted lines when in fact they can???t see the lines at all. Splitting it into (at least) two parts is essential. We could leave the current figure with just the data and red and blue features as it is quite a useful visualisation - then have additional split plots containing these dot-dashed etc lines. >> has changed - only few selected Q2 bins are chosen now. Cheers, Brian ---------------------------------------------------------------------------- Brian Foster Alexander von Humboldt Professor University of Hamburg/DESY Gruppe FLA Notkestrasse 85 22607 Hamburg Germany Tel: +49 40 89983201 ++++++++++++++++++++++++++++++++++++++++++++++++ Dietrich: Dera collegues, its really an interesting paper, well presented and should be published soon. Here a few commenst and corrections 1.line 17/ 18: ..(fit) ...(mod) ...(par): puzzling -> eliminate, explanation in formula (7) on page 14 sufficient >> decided to keep breakdown in the abstract bu use (exp/fit) (model) and (parameterisation) 2.line 100 : O(alpha**3) >> corrected 3.line 144 massive coefficient function - slang 4. lines 223/224 reference for values 5. line 250 fro -> from >> done 6. line 252: reference for the value of k_T >> added 7. line 268: corresponding clustering algorithm: not clear which >>> changed to "the same clustering algorithm as used on detector level <<<<< 8. line 348: FFNS [24 -31] and the VFNS [32-35] >>> references added 9. line 372.. set theory uncertainties are given .. not visible in fig8 and fig 9 >>> if you look closely they are visible as shaded bands in fig. 8+9, best visible at Q2=5 GeV2 10. line 634: .. do not improve the overall dscription, with and without the inclusion of log(1/x) resummation: Where was this shown? >>> last paragraph of section 5.2 and figure 14 of draft 1 (12 of 2nd draft) 11. ref 18: Measurement of .... missing >>OK - titles will be removed in the final version 12. ref [25}: The 3,4,5 - flavor NNLO ... 13. ref [26]: Running Heavy Quark Masses in DIS 14. ref[28]: Phys. Rev. D96(2017)014011 >>added 15 ref[29]:On the value of heavy flavor distributins at high energy colliders 16 ref [32] R.S. Thorne ... Phys. Rev. D86(2012) 074017 >>thanks 17. ref [34] Impact of Heavy Quark Masses on parton distributions at LHC phenomenology Nucl. Phys. b855... unbiased global.... 18. ref [36] Aaron -> H. Abramovicz >>oups 19. ref [41] A. Behring et al Phys. Rev D92 (2015)11405 >>OK 20. ref [46,48,49,50,52] titles of publications missing (be consistent!) 21. ref [59] High energy event generation with PYTHIA 6.1 hep-ph/0010017 >>OK 22 ref [60] BELLE title and hep-ex BABAR Phys. Rev. D67( 2003) 031101 hep-ex/0208018 >>OK 23 fig 7: fig caption "HERA 2012" -> [36] >>OK 24 fig 15: shading for DIS+c+b hardly visible; true? >colours have changed 25 fig 20: " brown colors" hardly separable >>OK 2 fig 24:dashed lines-> not visible >> only a few larger panels now Regards Dietrich +++++++++ Stefan Dear Sasha and Karin, congratulations for releasing this nice paper draft. I only have minor comments to draft V1 as listed below. I am looking forward to a timely publication. Best regards, Stefan General: ======== I would suggest to reduce the number of figures. My proposal is as follows: - remove figure 2 and 3. In the text, remove the second part of line 330 (after "are") and the first part of line 331 (up to "are"). >>OK - remove figure 16 and 17. In the text, remove end of line 541 (" in figures ...") and the rest of this paragraph. [this may require some small further adjustments in the new paragraph] >> no Of couse we should keep the removed figures as extra material Lines 593-601: as others mentioned as well, for this paragraph it may enhance clarity to have name tags for the different fits which are compared. The tags will have to be used consistently in the figures. >> done Figures: ======== Figure 1: remove the "statistics" box. Add a legend which says something like this: Gaussian fit: mu=0.03+/-0.05 sigma=0.77+/-0.03 >>OK Figure 15: change blue hatched style to filled area (or reverse hatching direction) Color: can we use a color code for the DIS+c+b fit which is consistent with figure 18/19 Consistent colors with fig 21? >>OK Figure 20: remove the grid. I like the horizontal line at unity but the other dotted lines are too much for my taste For the x-axis, I would prefer the label: (maybe:) try to use the same y-axis range for charm and beauty >>ok Figure 21: choose Colors consistent with figure 15,22 The horizontal axis is labelled x_min but in the text (line 587) it is named x_Bj,min (same comment holds for figure caption) >>OK Figure 22: choose consistent colors with fig 21 choose consistent color of DIS+c+b with fig 18/19 >>OK small text comments: ==================== 73: add comma: ... limitations, because ... >> In German I would have put a comma --- native English did not ask for it 92-93: too many times "reduced" "reduces" proposal: fullstop after measurements and remove: "and thereby ... further" >> done 140-141: "... at all scales..." I do not understand, should this read "at all orders"??? >> it means: at any mu2 (or Q2) - changed to: always treated as massive 188-189: I find it hard to read without confusing dataset 8 of [36] with the present dataset 8. Proposal: put the details to the end, all in a bracket: ... data set 9 supercedes a data set of the previos combination (data set 8 in table 1 of [36]). >> sentence changed 212: ... "the corresponding fragmentation functions" 250: (typo): "... originating from ..." 346-347: I think it is not recessary to explain that we do this "Before" we do someting else. -> remove line 346, start with: "The combined heavy flavor data are compared ..." >> removed 348: replace: "pre-existing" by: "various" >> o.k. 371: remove the statement "The theory predictions are obtained without fitting the data." (this is obvious in this section) >> o.k. 448: "if not" -> "unless" >>ok 455: ... are above 3.5 GeV^2 for all these measurements. >>ok 467: remove "the": ... determined by QCD sum rules". (otherwsie we will have to be more specific about which sum rules we use, but it is too much for this paper) >> I don't think that we need to be more explicit 468-469: The parameter C'_g=25 is fixed [64]. (this avoids the odd sequence of number and [ref]: ... 25 [64].) >>ok 486: "... parameter while the change ..." >> rephrased 502: ... around x~3x10^-3 ... (This is what I get from figure 20) >> it's 1.5x10^-3 in figure 15 as well as in figure 20 -> 2 514/515: ".. typically of a few MeV" -> "typically a few MeV in size" >> has changed page 14, footnote 5: (use a complete sentence) The previous charm mass result did not ... >> done 556: maybe try this to make x_Bj bold: \begin{boldmath} \section{ ... } \label{...} \end{boldmath} 563: "equivalent to" has a different font? >> corrected 563,589,612: please unify 2.9\sigma without space 1.8 \sigma with space 3 \sigma with space (I prefer to have no space or a small space only) >>done 609: remove comma: ... study shows that..." >>done 634: ... description, neither with nor without ... >> 647-650: ... by changing only the PDFs of the proton. The alternative next-to-leading order or next-to-next-to leading order QCD calculations considered are not able to provide a better description of the combined heavy flavour data either. >>ok ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Ewald Dear all, The current analysis is based on considerably more charm data than in [36] and, because of the beauty decays to charm, also on more beauty data. Published charm and beauty cross sections are combined separately. Are the possible correlations between charm and beauty cross sections somewhere quantified and taken into account in the systematics considered in this paper? How large are those? How do our published charm cross sections in [36] compare to the current charm cross sections? >>> yes, the charm and beauty cross sections are combined simultaneously. The correlation between charm and beauty are taken into account in the combination. They amount to about 50-70 %. They are taken from the publications of the input data. When necessary, corrections are applied using HVQDIS (e.g. charm and beauty cross sections hav different binning in case of H1). The comparison to the published input data is shown in fig. 4(charm) and 5(beauty) of draft 1. The improvement w.r.t. the previous combination is about 20% on average. <<<< Table5 and caption do not fit together: the values of "d.o.f." mentioned in the caption, are missing in the table! And for "HERA 2012 c" the proper reference should be included. >>> d.o.f. -> number of data points (no fit), ref. included I have spend some time to study figure 18 and the discussion of it in section 6.3 and in the conclusions. In figure 18, we see that the description by NLO FFNS is rather poor in some Q^2 bins, either in shape or in magnitude or in both. The corresponding b-results in Fig.19 look somewhat better. This might be partially explained by the larger uncertainty of the data points, however I think we cannot rule out that this is not only a question of the difference event statistics. I suggest to discuss this observation in more detail in the text! >>> There are reasons for the better description description: 1. the beauty data is less precise; 2. beauty production probes larger x of the gluon. The EB is against putting more emphasis on this. In lines 557/558 is noted that the dependence on x_JB is steeper than the predicted flat distributions. I think this is not the whole truth. We see that the magnitude value is also significantly different in at least two Q^2 bins. >> A different Normalisation is not really evident. The fact, that most of the points for q2=2.5 are higher, is related to the parton x range probed here. (see fig.20) In line 561, a "partial chi^2-value" is quoted which is not explaind at all. In case that this number is an overall estimate for all Q^2-bins together, it makes no sense. Moreover I cannot see that to quantify an overall deviation of 2.9 sigma without ideas about the origin delivers any meaningful information. Would it make sense to give chi^2 for each Q^2 range sepaparately? >> The partial chi2 odf the heavy flavour data is their contribution to the total chi2 of the fit to the inclusive+HQ data (including all correlations) It does not make sense to give a chi2 in bins of q2, because the data points are corelated across different q2 bins. There is another observation which worries me. Looking into [36] at figures 6 to 8, the data are well described by various predictions. In lines 559/560 is written that in [36] we see a similar behaviour, whereas have the impression that the new data are changed significantly, e.g. for Q^2=2.5 GeV^2. >> There are two effects. The data improved by about 20%. The agreement between the new and old combination is shown in figure 7 of draft 1. The second effect is that the predictions have changed. This is shown in fig 10. Therefore you get the impression that the old data agree better with theory. (see also discussion on tab. 5) In lines 609/610 and 642-645 is explained that the inclusive data with x_BJ<0.01 are causing the disagreements. I wonder if it makes sense to present also results without the inclusive data. >>>>>>>>>> The point is that HF data are directly sensitive to the gluon while the inclusive data are only indirectly sensitive to the gluon via scaling violations. However, the precision of the inclusive data is so high, that they dominate the determination of the gluon PDF. The influence can be reduced by removing the low-x inclusive data. This way the charm data can fix the gluon density and the high x inclusive data are fixing the other PDFs. If you remove the inclusive data you cannot perform a PDF fit, because there is no constraint on the quark PDFs. You could perform a determination of the gluon PDF from the charm data only, but this is a completely different analysis. <<<<<<<<<<<<< In lines 565/566 is said "show some tensions desribing the combined data". "Some tensions" is also mentioned as a conclusion (in line 641) The "tension" should be somewhat quantified and the relation to the inclusive data should be discussed in the text. >>>> the tension is quantified in the text as 2.9 sigma. Not having been close to this analysis, from going through the paper draft only, I have the general impression that important problems with the problems with the current results, e.g. the difference to previously published charm data, are not discussed in a convincing manner. >>>>> As shown above there is no problem with our previous publication and we think we have clearly discussed this in the paper <<<<<< Best regards, Ewald ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Boris Dear All, Here is a collection of detected misprints: l.100, replace )^3 by ^3) >> done l.481, replace 2.5 by 2.5 GeV^2 >> done l. befor 512, Eq.7, 2???nd line, ???0.033 is not the same as in Abstract and Conclusions >>>>>>> fixed l. 531, Is it really the fit gives mb(mb)=8.45 GeV ? >> yes -and even more problematic: depending on the choice of the PDF parameterisation you can get almost any number you wish -> it does not make sense to attempt to determine the HQ masses from inclusive eta only. l.665, 668, 671: replace ," by ", >>done l.673, the article title is not complete >>done l. 678, 680, 683, 685, 690, 693, 696, 699: replace ," by ", >> done l.701, the article title is not complete >>done l.717, 720, 722, 724, 731: replace ," by ", >>done l.793, replace ". by ", l>>done .807, replace .," by ", l>>done . 827, replace HERA, by HERA", l>>done .832, replace ," by ", \>>done l.834, replace ," by ", >>done p. 45, in Fig 15 insert: more space between lines with an inserted text to avoid the text overlap >>ok p. 52, Fig 22, the same as p.45 >>ok p. 54, Fig 24, second line from top. One need to avoid overlap ???FF3A??? and ???????????????NLO fit???. Add a shift to left >>ok With best regards, Boris +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Katja Dear All, I only have very few comments to the new draft, most of my comments have been answered before. Cheers, Katja - general: will we provide a table with the detailed breakdown of the uncertainties as a data file? if yes, should we mention that somewhere? - l. 41: mesons samples -> meson samples >> done - l. 54: I think the ???on average??? fits better earlier in the sentence, e.g. The proper lifetime of B mesons is on average ??? >>>> Sentence changed: The fragmentation fraction weighted average of the proper lifetime of $B$ mesons is about a factor of $2$ to $3$ that of $D$ mesons. <<<< - l. 101: I think you cannot compare to a scheme, but only to predictions or calculations in a scheme >> right -> changed: A comparison is also made to predictions of the FONLL-C ... - l. 112: section 3.1 -> section 3 >> done - l. 116: measurementS >> done - l. 218,232: use the same order of mu_r and mu_f in all the equations, so swap them in "mu_f=mu_r=sqrt(Q^2+4m_Q^2)??? >> done - eq. 4, text below it, and tables 2 and 3: to me the treatment of the correlated stat. unc. is not fully clear. In l. 267 it says that the correlated unc. comprise systematic and statistical components. But then in the next sentence I read that the gamma are correlated syst., the delta_stat statistical and the delta_uncorr the uncor. syst. unc. Which one of the three contains the statistical correlations? And in tables 2 and 3 we use a slightly different notation with delta_stat, delta_uncor and delta_cor >>> b_j are the shifts correlated uncertainties of source j of the combined cross section, gamma_j^i,j and delta_i,e,uncorr are the systematic uncertainties of the input measurements, In the tables 2,3 the uncertainties are the uncertainties of the combined cross sections. <<<<< - l. 328: I???m a bit surprised that this effect of ???cross-calibration??? between charm and beauty is so small (or maybe I misunderstand the sentence). I assumed that the significantly smaller uncertainties on the charm cross section due to the combination would lead to a sizeable (more than 10%) reduction of the beauty uncertainties since the VTX measurement is one of the most precise beauty measurements, and for this the (anti-) correlation is large. >>>>> Sasha did some estimate, what to expect: It's hard to come to a certain estimate of the effect of cross-calibration between charm and beauty taking into account all effects; possible reasons: (1) small number of beauty measurements than charm (in fact only two precise vertex H1 and ZEUS measurements) (2) even for these precise measurement, the number of measured beauty points is smaller than charm, such that one beauty bin is spread over a few charm bins. This can be seen e.g. page 10 of https://indico.desy.de/getFile.py/access?contribId=0&resId=0&materialId=slides&confId=16627 In all our paired charm-beauty measurements we have about 60-65% statistical anticorrelation. In the very beginning I did some rough estimates and concluded that based on the precision of the ZEUS vertex measurement, if there would be only stat. uncertainties and unlimited precision for charm, one should expect ~25% improvement for beauty (page 13 of https://indico.desy.de/getFile.py/access?contribId=2&resId=0&materialId=slides&confId=16056). Now with all systematic and extrapolation uncertainties and limited precision for combined charm, we have ~10% improvement, it is not dramatically worser. <<<<<<<<<<<< - l. 360: remove ???program??? >> o.k. - l. 401ff: many ???calculation???s -> use ???prediction??? instead for one or two >> tried to improve - l. 480: the _variation of the_ strangeness fraction >>ok - l. 496: The ratio \chi2 -> The ratio \chi^2 (exponent!) >>ok - l. 498: as discussed in the last meeting: I think the differences between the new fit to the inclusive data only, fixing the heavy quark masses, to HERAPDF2.0 FF3A are rather small (running mass vs. pole mass, ???), and we use HERAPDF2.0 FF3A in the previous section as a reference, so I think it makes sense to point out these differences here. >>>>> I think, it it worth pointing to the differences here because you can see, that where the central gluon PDF deviates ist just where you see the large difference between charm data and predictions (fig.20) <<<< - section 6.2: I think it would make some parts easier to read if we give names/abbreviations to the variants of the fit, instead of having to write a long statement describing the fit every time >> done - l. 563: ???equivalent to??? is in a different font >> changed - table 5: abm11_3n_nlo is the only PDF not written with capital letters >> changed - it should be ABMP16_3_nlo - caption of table 5: remove one ???the??? in line 4; reduces -> reduced in line 5 >>ok ++++++++++++++++++++++++++++++ Jan&Nelly Dear Karin, dear Stefan, I just sent the comments below to the address given in Stefans mail some 10 days ago. I get no indication that this list actually works, so "vorsichtshalber" I send it also to you. Sorry if you get it twice... Cheers, Jan Dear Editors and Referees of the paper "Combination and QCD Analysis of Beauty and Charm Production Cross Section Measurements in Deep-Inelastic ep Scattering at HERA" We are happy to congratulate you to this paper, and to the completion of this analysis! We have read the present draft from 23.11.2017 and find that you have given a very consise description of the analysis and its results, as presented in several meetings in the last couple of years. We have no questions to the first part of the paper, the cross section measurements. However, in the second part of the paper, the QCD analysis with the expressed focus on the determination of the running quark masses, a question arises: We are aware that the ZEUS collaboration already published a "prerunner" of this paper, namely DESY 14-083 (also quoted as ref. 21 in the present paper), arXiv.1405.6915 and we are also aware that the result on the charm mass presented there was subsequently critized by Richard Ball, in arXiv.1612.03790 If we understand correctly, the criticism points to an incorrect or incomplete treatment of pole quark mass vs. running quark mass, leading to unjustified small errors on the obtained running mass. We now wonder if this criticism is valid also for the present paper. Indeed, we do miss a introduction/discussion in the present paper, in which the roles of pole and running quark masses in the used Monte Carlo simulations and in the used fit procedures are detailed. We think that such an introduction/discussion would be very valuable for the general reader and that such an introduction/discussion is also appropriate in a paper which focuses on the c and b running masses determination. >>> this was answered by Achim<<<<<<<<<<<< Maybe this would also give increased weight to the physics message of this paper, beyond the cross section measurements and agreement or disagreement with selected models, which now constitute the sole physics message? >>> we measured the heavy quark masses Independent of this criticism, we also think that the message of the paper would gain from the addition in the Conclusions, of a paragraph which points to the possible improvements in the LHC cross section predictions, due to these new c and b running mass measurements. >>>>> EB decided, not to enlarge the content of the paper. We made the point in the previous publication. we could not add much to it. <<<<<<<<<<<<<<<<<< A small point, which was heavily discussed in the last H1/ZEUS meeting: We support the suggestion to quantify the deviation, which is remarked on in lines 642-3, Conclusions. The paper is very well written, and we found only very few mistakes in spelling, or choice of words. One general remark is of course that the paper has a tendency to formulate very long sentences, using "German grammar, verb at the very far end", which makes the reading sometimes very tedious. This tendency is worsened by the lack of commas in many sentences, leading to ambivalence in the actual meaning of the text. Thus, we hope that an English native will have a serious go with the paper text, making it more smooth and fluent, before the final publication! We make a few suggestions in this direction below. >>> sorry, it's difficult to get rid of what you learnt at school :) we tried to improve here We wish all success in the further publication procedure, and congratulate again to the paper and the tremendous work behind it! Our detailed text comments follow below. Best greetings, Jan and Nelly Line 5: in the Title is used "deep inelastic", as is also written in two places in the Abstract (lines 9,15). In the paper body text however, "deep-inelastic" is used everywhere. --> Consistent usage? 14: Perturbative QCD predictions are compared to the combined data. The latter are used together with the combined inclusive deep- inelastic scattering... (We reverse the order in the comparison: Theory is always compared to the data, never the other way around) >> changed 26: "the mass of the heavy quark involved" >> ok , changed to: ...proton and the heavy quark mass. 28: "momenta, of the" --> "momenta of the" >> done 29: "several hard scales" (plural !) >> done 35: Why no charge indication on "D", but indicating charges on D* ? >>>> because it's both D^0 and D^\pm (Table 1) 45: "significantly suppressed further" --> "further significantly suppressed" >> "further" is removed 51-52: "Although the first two reasons given above for the suppression of beauty production relative to charm production also hold in this case," >> changed 54: "is on average about a factor of 2 to 3 larger than that of D mesons, when taking..." >> sentence changed 55-57: "Therefore, using... flavoured hadrons, the charm..." >> sentence changed 67: "are on average harder" --> "have on average higher momenta" >> changed 68: "relative to the production cross section" --> "relative to the observed $c$-induced fraction" >> o.k. 73: "limitations because" --> "limitations, because" >> o.k. 74: "inclusive or lepton" --> "inclusive and lepton" >> right 94: "by the data" "by the new results" >> fine 108: "have not yet been fully" --> "are not fully" >> sentence removed - we do not say what was NOT done 113: Better: "... charm cross sections are given in section 4. In section 4 theoretical calculations based... VFNS are compared to these cross sections." >>>> sentence changed according to Brian's proposal 116: "charm and beauty quark" --> "charm and beauty quarks" >>>> have to check with a native English speaker !!!! 118: "are presented" --> "is presented" >>> changed measurement -> measurements - are is o.k. "dependance" --> "dependence" >> o.k. 119+2: "analyses combined" --> "combined analyses" >> 'combined' removed 125: "reaches up to" --> "reaches" >> o.k. 129: "occur for" --> "are involved in" >> sentence removed 131: "are realised" --> "are used", or "are applied" >> changed 139: "used for" --> "used in" >> 'for' is correct here 168: "NLL" --> "next-to-leading-log (NLL)" >> now defined in the introduction 169: "next-to-leading-log" --> "NLL" >> done. 175: "interaction point" --> (better) "interaction region" >> o.k. 208: "in general sigma can be" --> "sigma can in general be" >> " in general" removed 237-239: "assumptions on ... tagging have to be made." --> "assumptions have to be made on ... tagging. >> o.k. 264: "programme" --> "program" (cf. computer program, government programme) >> this is also what I learnt at school, but in the last H1ZEUS final reading i learnt that it is always "programme" in British Englisch 281: "directly taken" --> "taken directly" >> done 299: "(dataset 1) using" --> "(dataset 1), using" >>> done 336: "and below" --> "and lower" >>> native English speaker were happy with "below" - it's below 5 % 340: "data of" --> "results of" >> o.k. 342: "previous measurement." --> "previous measurements." >> ja 377: "show a somewhat steeper xBj dependence than" --> "show an xBj dependence somewhat steeper than" >> o.k. 408: "show some tension in general" --> "in general show some tension" >> o.k. 486: "noticabely" --> "noticeably >>OK 491: "fit, model, and" --> "fit, model and " >>OK 517: "parameterisation" --> "parametrisation" Note: You are almost everywhere using the spelling "parametrisation". This is OK, as would also be the spelling "parameterisation", which now occurs in a few places. We suggest to make a search in the latex-source, and use only one of these spellings. >> changed to parameterisation everywhere - you can see from the spelling that the text with parametrisation was not written by me - I tried to get rid of all 'z's but forgot about this 535: "demonstrates" --> "demonstrate" (plural actor!) >>OK 544: "PDFs especially of the gluon and the" --> "PDFs, in particular the gluon PDF. The" >>OK 570: "to see the ranges of x accessible by" --> "to determine which ranges of x are accessible" >> text has changed 575: "0.1 the" --> "0.1, the" >>OK 577: "evident showing" --> "evident, showing" >>OK footnote 7: "does not" --> "do not" "chi2 or p-value" --> "chi2 and p-value" >>OK 581: "performed with requiring different values of the minimum xBj values" --> "performed, varying the values of the minimum xBj" >>OK 584: "as function" --> "as a function" >>OK 586: "0.04 while" --> "0.04, while" >>OK 597: "imposed to" --> "imposed on" >>OK 597: "rising stronger towards small xBj" --> "rising stronger for smaller xBj-values" >> rephrased according to Brian 602: "calculated for inclusive DIS also." --> "calculated also for inclusive DIS." >> 'also' removed 603: "figures 24" --> "figure 24" >>OK 609: "shows, that" --> "shows that" >>OK 610: "excluding ... in" --> "excluding ... from" >> sentence changed Ref. 44: "S.Alekhin, J.Bluemlein and S.Moch" "I.Bierenbaum, J.Bluemlein and S.Klein" >>OK Ref. 59: "Sj\"ostrand et al." >>OK Ref. 61: "G.Curci, W.Furmanski and R.Petronzio" "S.Moch, J.A.M.Vermaseren and A.Vogt" "A.Vogt, S.Moch and J.A.M.Vermaseren" >>OK Tables 2,3 captions: "obtained by" --> "obtained through" "uncetrainties" --> "uncertainties" >>OK Table 4: "luminosity" --> "integrated luminosity" 6 times >>OK Table 4, caption: "extracted...sections simultaneously a" --> "simultaneously extracted ... sections a" >>OK Table 5, caption: "the the" --> "the" "is reduces" --> "is reduced" or "reduces" >>OK Figure 12, caption "uncertianties" --> "uncertainties" >>OK Figure 13, caption" "They are of similar size than those" --> "These are in size similar to those" >>OK Figure 14, caption: "They are of similar size as those" --> "These are in size similar to those" >>OK Figure 20, caption: "data as" --> "data, as" "for the different" --> "for different" We also suggest to use labels a) and b) in this figure, instead of "upper and lower panel". Who knows what orientation this figure will have in the final publication? >>OK Figure 21, caption: "data (triangles) only" --> "data only (triangles)" >> has changed Figure 23, caption: "shaded band" --> "shaded bands" (like in fig.22) >>OK Figure 23,24 captions: "(full line)" --> "(full lines)" >>OK +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Max Dear Colleagues congratulations to this combination and analysis, a monument to the strong c,b and QCD efforts of our Collaborations, many thanks. Below please find some comments on the paper draft 1.0 (Nov23) with apologies for being somewhat late. Best regards, Max p0 l12 is the photon virtuality Q or Q2 or -Q2? I would term Q2 once as what it is the negative 4-momentum transfer^2. at high Q2 it may also be the Z virtuality, and we measure to 2000 GeV2 l17/18 I would term the 'fit' error source 'exp' because there sits all the experimental uncertainty, even if we use a fit to determine these for mc,mb >> changed p1 I would move 4 paragraphs, l38-l77 from the introduction to data samples 3.1. it is a long, qualitative discussion, important, but in my view not for the general introduction. >>>>> moved to section 3 <<<<<<<<<<<<< p2 l99-102 I would delete the statements in parentheses (O(alphas2.. respectively) and (O(alphas3... functions) as they are very technical and not important at this point. you cite [32] for NLO but then talk abut NNLO, is there no/a ref to that? >>>>> the EB decided to keep it. Some sentences have been changed such that some parentheses could be removed. <<<<< p3 l108 delete 'ongoing' > done eq 1 and elsewhere, probably too late, I know, but I would use 'x' instead of 'x_Bj' as in DIS x_Bj is known to be x, there is no x_F to confuse the notation >>>>> we keep x and x_Bj because 'x' is the partonic-x which is also in DIS only in LO identical to x_Bj and we are dealing with NLO and NNLO PDFs in this paper. <<<<< eq 1 and eq 2 I would write alpha, not alpha(Q2) but in l123 then write elm coupling alpha=alpha(Q). the finestructure constant running is less dramatic than the one of alphas as we all know. >>>> the running of a_em is in the per cent region and therefore relevant given the precision of the data <<<<< p4 l158 delete .in the MSbar scheme >> done l163 there is a Landau O before in a different style, l99ff also I would write Q2=O(mQ2) because the O already implies \simeq >> changed p5 l188 data set --> dataset >> done p10 l366 and figures: I would delete ABM09, it is past and brings nothing to the discussion, even if by any reason it is higher at Q2 2.5 for cc, to have ABM11 and 16 should suffice. the fact that it has not used HERA HQ data is fine, but it neither used DIS data or LHC, so a comparison cannot be leading to any real conclusion. >>> no. we think it is important to include ABM09. This is what we used in our previous analysis. There the agreement between theory and data is better than it is now. This is not only because of the higher precision of our data but also because ABM09 was better in describing x_Bj dependence. <<<< p12 eq 6 and discussion: do we know what happens to the x dependence if A'_g is set to zero, and the gluon is not allowed to 'disappear'? the steepness of Fcc,bb will have to do with xg and this negative term we obtained from RT has improved the chi2 but it puts also epWZ16 to an extreme xg prediction and should perhaps be questioned in the context of the discussion that is following in our paper, see also remarks below >>>>>>>> this has been tried - but it does not describe the steepness of the charm x-sect, because charm mainly contributes in the range of 0.0005> thanks p14 l529-538 I would delete that exercise because if you only fit Fcc and Fbb (which is what I conclude is described here) you must fail, and you do. >>>>> no, the first sentence says: 'Fits to the combined inclusive data only are also tried.' It's just the opposite from what you concluded. The conclusion is: The inclusive data are not able to fix the masses <<<<<<<<<< p15+16 6.3 I find there is not enough motivation, from fig 21 (chi^2 vs xmin) to go on for a study with x>0.01 for the DIS data. that is really extreme. it is obvious that such a fit cannot describe the data at x=10^-4, it has no handle. Moreover, you even see that the full DIS+c,b chi2 is improving! when going to lower xmin. I thus would stop the paper at fig21 and state that one observes a trend that the x dependence is somewhat! steeper in Fcc than the fit wants it to be. give it more freedom and it follows, but do not cut all the HERA NC+CC data out. of course, if the only constraint at small x is Fcc (and bb) then you reproduce them better, fig 23, but that is almost a trivial statement: you let the inclusive cross sections go to whereever and fit the low x HQ data. this in my view is not illustrating or telling us anything really. we observe certain tension between Fcc,bb and inclusive DIS, which we could not resolve in our framework, that is an interesting result, not the x<0.01 toy fit study. >>>>>>>> We disagree. Charm production is directly 'proportional' to the gluon, while the inclusive data can fix the gluon only indirectly via scaling violations. The charm data have reached a precision which allows us to fix the gluon from the charm data only. This is what is attempted here. The cut at x_Bj=0.01 does not remove CC data. The lever arm in x is sufficient to fix the quark distributions. The rest is done by DGLAP and the sum rules. In the inclusive paper we have found a 3 sigma tension between theory and data. This kept the analysers busy for more than a year, but we couldn't find a reason for it. The current analysis shows another 3 sigma tension in theory in describing simultaneously the inclusive data and the charm data. The study shows, that both data samples prefer different gluon shapes. As far as I remember, we had also in the inclusive analysis a problem with the calculations describing the turn-over at very small x. We think, this study adds valid input for further theory developments. <<<<<<<<<<<<<<<<<<< p17 perhasp call it a Summary rather than Conclusions >>OK l630 i would list the QCD predictions here which were used >> no, too much for the Summary l640 perhaps one needs here a line to say what HERAPDF2.0 FF3A is >> no, not needed l641 'some tension' needs to be quantified here, is it a valid fit or is not? if indeed you accepted a bit of my reservation against the high weight given to the x>0.01 approach, the conclusion would basically be that in a joint NC+CC + c,b fit the x dependence of Fcc cannot be reproduced well, neither in NLO nor NNLO. >> is quantified: 2.9 sigma Acknowledgements: perhaps one thanks SA, JB, SM? >> no, there is no reason for it p19 [28] is: Published in Phys.Rev. D96 (2017) no.1, 014011 >> thanks p22 [59] Sjoestrand >>OK p30 are the nr of points 52,47 really the same in the 2012 set and now? >>yes what about stating that the inclusion of the PDF uncertainties has a negligible effect on the chi2 and delete the rather repetetive last column? >>>>>> done <<<<<<< p52 the gluon is too sensitive to the radical 0.01 cut. it does one good thing, it indicates that Fcc wants it to rise. since the inclusive DIS data seem not to want that, Fcc^thy is less steep than the data. therefore i would think that A'_g=0 may be an interesting case study. one may argue that xg has to be positive and exclude the negative term, perhaps this helps and it is a less radical cure or case study than the 0.01 thing. >>>> the gluon is fixed by charm which is directly constraining the gluon because it's 95% BGF while the inclusive data constrain the gluon only indirectly via scaling violations. The fits with A'_g have also been tried. In this case the description is not better - the gluon is not flexible enough. What happens when cutting at 0.01 The A'_g becomes negative! such that the gluon rises even stronger because of the additional flexibility.<<<<<<<<<<<<<<<<< a very nice result and very important paper, thank you! ------------------------------------------------------------- Prof. Max Klein University of Liverpool Department of Physics L69 7ZE UK tel: +44 (0) 151 794 3353 CERN, 1211 Geneva 23 Switzerland tel: +41 (0) 22 767 1319 ------------------------------------------------------------- ++++++++++++++++++++++++++++++++++++ Oleg Dear Analysis Team, Congratulations with performing the cross section combination and QCD analysis. The paper draft is nicely detailed. Please find my proposed textual comments below. Sorry for sending them after the deadline. I hope they could still be considered. With best regards, Oleg L28: Probably, "$p_{T}$, " was meant to be inserted between "momenta, " and "of the outgoing quarks"? Otherwise the comma seems to be not needed. >> removed L29: "several hard scale" -> "several hard scales". >> changed L34: It seems "different various" needs to be replaced with only one of the words. >> different L100: "O(\alpha_s)^3" -> "O(\alpha_s^3)". >> done L102: "low-x" -> "low-$x$". >> done L112: "section 3.1" -> "section 3". >> done L117: "dependance" -> "dependence" (?) >> done L118: Is it a study of the x_Bj-dependence of the measurement, or of the x_Bj-dependence of the cross section? >> cross section L150: ", respectively" seems to be redundant. >> removed L156: Is it "programme" or "program"? Different versions are used throughout the draft. >> programme (Britsh) L158: Please remove ". in the MS-bar scheme". >> done L178: I was for a moment confused reading the sentence. Could the number of data sets entering the combination be also quoted? E.g. "The 13 data sets included in the combination are listed in table 1 and correspond to 209 individual charm and 57 beauty cross section measurements." >>>"and correspond to 209 individual charm and 57 beauty cross section measurements" removed L192: For consistency, it is better to use Roman "red" in the notation for the reduced cross section. >> done L193: Same as above for "vis,bin". >done L206: Capital "E" in "Eq" is used for denoting the equation in this line, but small "e" is used in the rest of the draft. It would be good to consistently use the same style (HERAPDF 2.0 paper was using "Eq.", "Tab.", "Fig.", "Section"). >> Last time it was ZEUS-style, this time it's H1-style :) -> changed to: equation (3) L231: "kept fix" -> "kept fixed". >> done L231: There seem to be a whitespace between "GeV" and "." Please remove it. >> done L231: "factorisation were" -> "factorisation scales were". >> done L250: "fro" -> "from". >> done L263: Since the {\sc } environment seems to be used for xFitter and HERAFitter, would it make sense also to use it for other programs, like Pythia, OPENQCDRAD, QCDNUM, and HVQDIS? L264: Move "MC" closer to "Monte Carlo". >> done L344: "( data" -> "(data". >> ok. L359: It might sound better (and more fair to our theory colleagues) to replace "with" -> "within". >> o.k. L360: Probably "program" can be safely removed. It is already called framework. >> removed The two sentences in L358-360 and L361-363 might be combined into one sentence, and kept in the place of the first one. E.g.: "The theory predictions are obtained within the open-source QCD fit framework for PDF determination {\sc xFitter} [45] (version 1.2.0), which uses the program {\sc OPENQCDRAD} [44] for the calculation of reduced cross sections." >> changed accordingly L453: The comma between "applicable" and "the" would make the reading a bit easier. >>OK L496: The "(d.o.f.)" has already been introduced earlier in the text. >>OK L507: Is the word "form" used in the meaning of "shape"? >>yes L511: Full stop is missing in the end of the sentence. The text of the footnote should probably start from a capital letter. >>OK L519: Is "both masses" denoting the values of charm quark mass determined in this and previous [36] papers? Or is it about charm and beauty masses? If it's the latter, I'd suggest to replace "both" -> "both charm and beauty". >> Rearranged sentence Footnote 7: "does not" -> "do not". >>OK L563: The text "equivalent to" is displayed in a different font on my system (Adobe Acrobat Reader on Windows 10), compared to the rest of the text. >>OK L585,586,587: Please use Roman font for "min", as in L582. >>OK L603: "figures" -> "figure". >>OK L660: Can the symbols "0" and "+-" go as superscripts for "D"? >> all titles will be removed in the final version L661: The whitespace in "e p" could be removed. L808: To be consistent, finish with a semicolon instead of full stop. >>OK Fig. 15 caption: "Q_0" -> "Q^{2}_{0}". >>ok Fig. 15: Maybe it's my screen only, but both PDF sets look as continuous lines, while the caption says that one of them is dashed. >> you were right - the coding has changed now Fig. 15: Caption says that the uncertainties for the fit to inclusive data only are not shown. But they are shown on the plot (and the opposite is for DIS+c+b data). Is the legend correct? >> was a mistake Fig. 20: Are the data points in this figure distinguishable from each other when printed in grayscale? >> has different markers now Fig. 22: In the caption, "(full lines)" could be removed (all lines look full). >> caption has changed Fig. 23,24: In the caption, use Roman font for "min" in the notation for $x_{\mathrm{Bj,min}}$. >> caption has changed Thanks a lot! ++++++++++++++++++++++++++++++ Peter Comments to ???Combination and QCD Analysis of Beauty and Charm Production Cross Section Measurements in Deep Inelastic ep Scattering at HERA??????Peter Tru??l General: Within the heavy flavour working group active during the pre-data phase of H1, which I was asked to coordinate, we started off with the notion that HERA was among other things a ???charm factory???. This illusion quickly disappeared when we realized that in lack of suitable triggers the heavy quarks would escape nearly unnoticed. It is therefore gratifying that in the end, 25 years later, the final analysis of the relevant cross sections measured by both collaborations has been finished and is ready for publication. Many thanks to all involved in the preparation of this final section of the long journey, among them some of members of my group in the early phases. My minor comments only concern the text. It seems that the different chapters have been written by different persons with the consequence that there appear some repetitions which need to be weeded out. The title of the paper could be shortened to ???QCD Analysis of Beauty and Charm Production Data from Deep Inelastic ep Scattering at HERA??? Abstract (shorten somewhat, e.g. like): Open beauty and charm production cross sections in deep inelastic ep scattering measured at HERA by the H1 and ZEUS collaborations are combined. The data cover a kinematic range of photon virtuality ??????. The combination method accounts for correlations of the statistical and systematic uncertainties among the different data sets. The data are compared to perturbative QCD predictions and also used together with inclusive deep inelastic scattering data from HERA in a next-to-leading order QCD analysis. ???. Throughout the text: It is not necessary to repeat ???heavy quark???, ???charm and beauty??? within a section several times, if it is clear from the beginning that nothing else is being discussed, first examples line 25 and 34 below. >>> will be removed where ever possible Semi-leptonic -> semileptonic Both ???program??? and ???programme??? are used in the text, decide on one them >> programme should be used Introduction: L 25: The cross section therefore depends strongly on the gluon distribution in the proton and the mass of the heavy quarks involved. >> sentence changed L 28: ??? the transverse momenta of the outgoing quarks and the virtuality, Q2, of the exchanged photon. The presence of several hard scales ??? >> sentence changed L 34: At HERA different various flavour tagging methods have been applied. >> done L 53: The proper lifetime of B mesons is about a factor of two to three larger than that of D mesons on average >> sentence changed L 91: The simultaneous combination of charm and beauty cross section measurements reduces the correlations between them and hence also the uncertainties. >> fine L 98: In addition QCD calculations in the RTOPT VFNS <- what does this mean >>>> RTOPT is an acronym like ABMblabla and VFNS is defined in the last sentence of the first paragraph.<<<< L 104: The new data are subjected to a QCD analysis together with inclusive DIS cross section data from HERA [40] allowing for running charm and beauty quark masses in NLO, as defined in the QCD Lagrangian in the modified minimum subtraction (MS) scheme. >>> o.k. ~ taken L 110: The paper is organized as follows. In section 2 the reduced heavy flavour cross section is defined and the theoretical framework is briefly introduced. The data samples and the combination method are presented in section 3.1. The resulting reduced cross sections are presented in section 4 and compared with theoretical calculations based on existing PDF sets at NLO and at NNLO in the FFNS and VFNS in section 5. In section 6 the NLO QCD analysis is described and the measurement of the running masses of the charm and beauty quark in the MS scheme at NLO is presented. This section also contains a study of the xBj-dependance of the cross section. Finally, the paper is concluded in section 7. >>> combined with other proposals Section 2: Electro-weak -> electroweak, per cent -> percent >>> 'per cent' is correct L 119: In the kinematic range explored by the analysis of the data presented here the virtuality of the exchanged boson is small, i.e. Q2 ???M2Z, such that virtual photon exchange dominates. >>> sentence changed L 120: ??? where y denotes the lepton inelasticity. (Bjorken has been defined before) >>> done L129-131: cut, already in introduction >>> first sentence removed L 158: drop ???in the MS scheme.??? >> done L 160-162 ff: RTOPT ? (authors of ref. 32 ?), FONLL_C ? This paragraph can only be understood by specialists. >>>>> try to rephrase it: Predictions from different variants of the VFNS are also compared to the data. The expectations from the NLO and approximate NNLO RTOPT~\cite{rt_opt} implementation as used for HERAPDF2.0 \cite{HERAPDF20} are confronted with both the charm and beauty cross sections while the FONNL-C calculations~\cite{fonllb_and_c,nnlonllx} are compared to the charm data only. <<<<< Section 3: L 170 ??? 171: ??? high-resolution vertex detectors [48,49]. (the references to the vertex detectors suffice, the names are irrelevant) >> done L 179: drop ???and correspond to 209 individual charm and 57 different beauty cross section measurements???, appears again in L 306-307 >> done L 181: ??? includes measurements using different tagging methods: (we know by that we deal with charm and beauty) >> right :) -> removed L 184: ???. muons from semileptonic decays >> o.k. L 204-206: ???. theoretical predictions for ??? and ???. in the NLO FFNS scheme. Only their shape in function of the kinematic variables is relevant for the corrections, while their normalisation cancels in Eq. (3). >> o.k. L 231: ??? was fixed at .. >>o.k. L 233: For all parameter sets the corresponding PDF set is used. P 7 footnote 2: While ??? -> Since ??? >> o.k. L 279: The results are converted to a centre-of-mass energy ???s = 318 GeV. >> sentence changed L 281: The combination is based on the ???.. procedure [37] used previously [36,38-40] >> o.k. L 302: ???. yields a significant reduction of the overall uncertainties of the combined data, as detailed in the next section. >> o.k. Section 4: L 310: drop ???and a conservative estimate of the uncertainties of the individual measurements.??? comes again in L 314 >> dropped Section 5: L 346 ff: Shorten to ???Before performing a dedicated QCD analysis of the data they are compared with calculations using pre-existing PDF sets. Predictions in the FFN and the VFN schemes are considered focusing on results using HERAPDF2.0 PDF sets.??? >> 1st sentence removed. 2nd sentence taken. In the following ???combined??? could be cut everywhere no uncombined data are considered anyway, maybe even ???combined reduced charm (beauty) cross section??? could be replaced by ???charm (beauty) data??? or ???results??? to make the chapters shorter and more readable. I guess for combinations such as ???theory predictions??? and the likes ???theoretical??? would be better. >> o.k. Footnote 4: The calculated cross sections ??? were provided by the authors. >> o.k. Section 5.3 Summary of the comparison to theoretical predictions >> o.k. L 421 ??? 430 a rather clumsy explanation of a simple fact, why not just write: The table also includes a comparison to the combined charm data published previously [36]. The apparent poorer agreement of the new data compared to the previous results can be traced to the increased precision of the new data. >>> no, it's more than this. The measurement is an estimate of 'reality'. With increasing statistics data would come closer to 'reality' (and stays at chi2/ndf=1 w.r.t. 'reality', assuming Gaussian errors). If 'reality' and theory would coincide, the measurement would approach theory in the same way and the chi2/ndf w.r.t. theory would stay constant at 1. The fact that it increases shows that theory deviates from 'reality'. This is not trivial, but maybe not easy to express. <<<<<< Section 6: L 494: In the QCD fit the running heavy quark masses are fitted simultaneously with the PDF parameters. The fit yields a total X2 = 1435 for 1208 degrees of freedom (d.o.f.). The ratio X2/d.o.f. = 1.19 is similar to the values obtained in the analysis of the HERA inclusive data [40]. >> ok. L 512: The model uncertainties are dominated by those arising from the scale variations. >>OK L 514: ??? while the other sources lead to uncertainties of typically a few MeV ??? >>OK L 515: ???. Is set to zero, the other contributions are negligible. >>OK L 539 ff: The NLO FFNS predictions based on the PDF set and the running beauty and charm quark masses determined by the fit are compared to the data in figures 16 and 17, respectively. >> sentence has changed L 546 ff: In order to better visualize the differences of the present to the latter analysis the ratios of data to predictions are shown in figures 18 and 19 for charm and beauty, respectively. >>> sentence has changed Section 7: L 527: The charm cross sections presented in this paper are significantly more precise than those previously published. The data are compared ???. >>OK References: A few inconsistencies 1 ??? 23: reorganize in chronological order ? >> corrected 18, 26, 28, 32, 35, 66 details are missing >> Titles will be removed in the final version. 61 ??? 65 the information following doi: should be scratched Figures: Except for the theoretical curves 8 and 9 are identical to 2 and 3, hence one could omit the latter; if captions contain identical sentences a reference to the first occurrence may suffice. >> Fig. 2,3 removed??? +++++++++++++++++++++++++++++++++++++++++++++++ Daniel Hi Sasha, Karin, et al. please excuse the late sending of the comments. Please ignore my comments, if those are considered as 'too late'. The analysis is in a very well shape and the results are also very well presented. Congratulations ! I have only minor general remarks, but a number of smaller corrections and improvements of consistency. Thanks for this impressive work. It is really a great paper. Cheers, Daniel General comments --------------- The introduction appears to be too long. It may be appropriate to have a subsection "3.0 discussion of data sets", or "measurement techniques" and mention in the introdction only, that there have been different measurement techniques employed, which is then benefitial for the combination (l38-l77). >>>> experimental part of the introduction moved to section 3 -> E.g. l71: This paragraph has no references and cannot be understood as it is. -> E.g. l95-l103. All these details have not to be discussed in the introduction >> decided to keep in the introduction \times -> \cdot or '' >> changed Often, the subscript or superscripts are not in roman fonts x_{{\rm Bj},min} -> x_{\rm Bj,min} etc... The sections 6.2 + 6.3 appear like, that you were lazy to write the paper in a more compact form. >>>> THANK YOU FOR BEING SO POLITE !!!!!!! Indeed, I've better things to do than writing this paper - PLEASE FEEL FREE TO TAKE OVER !!!!!!! Section 6.2 should be shortened. It is difficult to get the relevant information. Section 6.3 should be sharpened as well. A table of the tests performed would maybe help. All appearances of \alpha, have a larger font. Very strange. datasets -> data sets (multiple times) Title --------- Maybe add the mass determination, as this is also more prominent in the abstract than the QCD analysis: Combination of ... and determination of charm and beuaty quark masses Abstract --------- L9 _all_ measurements ? >> not all L9 'neutral current' is missing l13 \times -> '' or \times -> '\cdot' >> \times changed to \cdot L14 vice-versa: Perturbative QCD corrections are compared to data >> right l15. New sentence for 'together with combined inclusive...' >> reformulated l16 It should be mentioned, how these masses are obtained >> decided not to do it in the abstract l16 beauty and charm -> Order consistently throughout the paper >> will try l21 ^^^ L17,18 (fit) is not a reasonable uncertainty. I suppose, this is just a linear error propagation of exp. uncertainties. Thus, it should be called (exp). Only in case, it is NOT just an error propagation of exp. uncertainties, one may consider other terminologies. -> In case, it is (fit), it would be interesting to have a split-up into exp and 'fit' uncertainties -> In l500, and 501 it is correctly named, 'experimental' uncertainties. >> after a long discussion in the EB we changed it to exp/fit L17,18 (fit) (mod) (par) are not defined >>> (exp/fit), (model) and (parameterisation) L21. electron-proton -> lepton-proton ?? >>> it's e^+/e^- but not muon or neutrion -> keep electron\footnote{..} L21. It may be appropriate to add a footnote, that "beauty" denotes the "bottom" quark. >> not needed L24. Did the measurements showed that boson-gluon fusion is dominating, or rather the calculations ?? L21-l37. I propose to have one paragraph for the data, and one paragraph for the theory introduction >>>>> most of the experimental discussion is moved to section 3, the theory discussion stays <<<<<<<<<<< L42. is suppressed by about a factor of 1/4 (this does not hold for NNLO) >>>> this is about the coupling of the photon which does not depend on the order in QCD L46. "... is significantly suppressed further". -> "... is significicantly suppressed further for the accessible kinematic ranges at HERA." >>> sentence changed - no need to add the kinematic range of HERA here. L49 "often escape detection": a bit colloquial: maybe somethine along: "are outside the acceptance of the HERA detectors" ?? >> changed to: remain undetected. l54 of the D mesons~\cite{add reference} >> added L58 p_T^{rel} -> P_{\rm T}^{\rm rel} >> done -> large or small 'p' ?? l59 w.r.t. -> with respect to >> done l66. that of the D meson~\cite{add reference} >> done l82 in DIS -> in NC DIS >> I think it is sufficient that we said once that it is NC - it's obvious that CC does not play a role in this range of Q2 l84 cross-correlations -> colloquial -> including the resulting correlations of systemtic uncertainties >>> sentence has changed l84. remove 'suitable for comparison..." (this is obvious) >>> removed l86-l94. This paragraph appears not to belong to the introduction, or can be significantly shortened. >>> We think it is needed here, but has changed L95-l103. For the introduction it is sufficient, that different PDFs are studied in NLO and aNNLO, and different heavy-flavor schemes are studied. >>> It was decided to leave, apart from some rephrasing as is. l107. An NNLO ... -> this sentence does not belong to the introduction >>>> right, in general we do not say what we do not do l111. What doe mean: 'briefly introduced' as there is already a 1.5-page introduction of these data in the introduction >>> theory is briefly introduced l117 \xbj-dependance -> \xbj dependence ('hyphen' and typo) >>> corrected L118. Finally, ... (remove this sentence), or add. "this introduction ends herewith." >> sentence kept, but finally removed L110-L118, It would be convenient to have click-able cross-references. Section 2 -------------- L119+1 'neutural-current deep-inelastic ep scattering' -> NC DIS (as already introdcued earlier (if NC is introduced in l21)) >> left unchanged L119+2 'the virtuality of the exchanged boson is small' -> \Qsq is small >> changed l119. is dominating?! dominating over what?! >> dominate is fine -> Is the data probably corrected for gZ and ZZ exchange? l119+4, l25. particles are mainly abrevieated with small letters: u,d,c,b,t,g,gamma, etc... (but weak bosons Z,W,H) >> im heavy flavour physics 'Q' is normally used when the statement/process applies to both charm and beauty (or even top) l119+6: F_L -> F_{\rm L} >> done eq1. There, the nomenclature of using 'Q' for heavy-quarks is ambigious with 'Q2' (\alpha(Q^2) * F_2^{QQ}) >> '\rm Q' = heavy quark , $ Q^2$ = $-q^2$ l121 'heavy QQbar pair' -> of a heavy quark pair (qqbar) >> done l122 electro-weak -> electroweak >> done eq1. I think, there is some problem with this definition: alpha_em(Q) can be taken out of the structure functions only + if gZ and ZZ exhange are excluded, or + alpha_em(0) if data are corrected for running alpha_em(Q) This is because the gZ and ZZ terms are proportional to kappa_z, with kappa_z~alpha_em(0), but not to alpha_em(Q), because alpha_em(Q) is a purely QED correction. It is sufficient, to mention explicitly that not gammaZ and ZZ contributions are not considered in this paper. (see also l199) >>> please READ the second sentence of this section which clearly states that photon exchange is dominant. l 132 At photon virtualities not -> At \Qsq not >> At values of $Q^2$ l138 'a correct theoretical treatment of the h f masses is mandatory' -> 'a correct treatment of the h f masses in the calculations is mandatory' >> careful treatment l139. FFNS~\cite{reference needed to review or original article}, or it has to be introduced first, and then abbreviated. >> not needed - given in the introduction l149. \mur^2=\muf^2=\Qsq+4\mqsq (avoids the sqrt) l158. corrupted sentence. >> removed l159. this wass already said. >> no L163 \mathcal{O} or \mathscr{O} or 'O' -> elsewhere (O(asmz)) is used >> corrected l163 it should be muf^2 here, right? What is actaully done?! >> no l164 \Qsq -> \mufsq (right?) >> no l173 past or present? >> here it should past tense - the detectors are not existing anymore l180. In HERAPDF2.0, I think we have used 'HERA I' and 'HERA II', i.e. withough '-'. >> o.k. l180 (1992--2000) -> (in the years: 1992-2000) >> fine l180. A remark on l176 should be made, that CST and MVD were only available at HERA II >>> Text has changed; CST was also available at HERA I l189 30~\% -> 30\,\% >> o.k. l189. I think, one wants to say here, that these data are statistically correlated. (which is said now only indirectly) >> ... based on a subset of only about ... l191. Use consistently: 'of reference~\cite{xyx}' or 'of~\cite{xyz}' The latter was used before rather often. >> changed 192. _{red} -> _{\rm red} >> done l193 _{\rm vis,bin} (as in eq3) >> done l194 eta not defined >>> slightly rewritten and footnote added to define eta l195 In case of inclusive D meson cross sections~\cite{add papers} >>> references are given 11 lines above l205. Well. for a single point, there is no 'normalisation' or 'shape'. What is meant here? >>> changed to: Since the ratio in equation~(\ref{formel_ftc}) describes the extrapolation from the visible phase space in $p_T$ and $\eta$ of the heavy flavour tag to the full phase space, only the shape of the cross section predictions in $p_T$ and $\eta$ is relevant for the corrections, while theory uncertainties related to normalisation cancel. l206 Eq -> eq >> equation l207. 'Uncertainties are correspondingly reduced.' -> Uncertainties are scaled accrodingly. >> removed l209-l212. This should to be reformulated. (what does 'however' refer to?, why the convolutions is mentione two-times, although it is irrelevant for the purpose of this paragraph, etc...) >>>> No. It says that for the reduced cross section you need to consider only two terms in the convolution: PDF \otimes ME. For the visible cross section you need to consider three terms: PDF \otimes ME \otimes FF <<<< l213. Why there are different forms of convolution integrals? >>> because you have to detect the tag therefore in sigma_vis you have to include the fragmentation function (as is said in the text) <<< l220 use consistently 'c quark' ('b quark') or 'charm and beauty quark' >> try to be consistent l220,223. References would be good to have here l225. These variants are strictly-speaking no variants of HERAPDF1.0, since they are published within ref [39]. It should be said, that PDFs are determined following the HERAPDF1.0 approach, using HERA I data. Or simpler: add ref[36] after HERAPDF1.0 [36,39] >> [36,39] l229, 231. replace renorm. and fact... simply by mu_r and mu_f >> Enorm.. scale, since a sentence should not begin with a symbol. L232. Q denotes the heavy quark, right? >> no, sqrt{-q2} or do you mean the Q in m_{\rm Q} then yes L 245 c-quark -> charm quark (or c quark) >> charm quark eq4,l288 stat -> \rm stat, uncorr -> \rm uncorr >> done L294. I don't understand this sentence. >> changed to: In the present analysis, the correlated and uncorrelated systematic uncertainties are predominantly of multiplicative nature, i.e. they are proportional to the expected cross sections $m^i$. (this is what you can see in eqn. 4) 'multiplicative nature' stay for the moment, I guess Achim likes it much :) L303 (l305). It would be interesting to quote the formula, how the error breakdown into stat. and uncorr. uncertaitnies of the results are obtained. >> decided to be not more explicit l310. remove 'and a conservative estimate...' (repetitive to l314) >> removed l323 a unique (a 'junique') >> native English speakers were fine with 'an' l336,338,388l481,l487, (and often elsewhere) consistently: 'xy < Q2 < 123 GeV2' (as in abstract) or 'xy GeV2 < Q2< 1234 GeV2' >> o.k. l354 HERAPDF 2.0 -> HERAPDF2.0 >> o.k. l357 remove MSbar running mass >> no, there are different schemes - this is what is used l359-l375. Wasn't this already said? >> yes, in section 3 - but it doesn't harm to repeat it here l410 Replace footnote 4 by reference [A. BCD, private communication] or [Calculation provided by...] >> done l421 p-values: $p$-values or $p$ values. >> o.k. l446: Shorten: mur2 and muf2 are set to Q2+4mQ2 (or remove this sentence, as already said). >> a bit shortened l455 above 3.5 GeV2 -> above Q2min. Or: "..data since always \muf2 > Q2min'. >> no L458. The question arises, why the log-term is not included in the combination. >>> because it is iterative, and at each iteration uncertainties are rescaled properly - no need for another correction to accountfor rescaling of uncertainties L459+2 mu_{\rm f,0} -> mu_{f,0} (as muf elsewhere) >> o.k. eq5. This 'generic form' misses the prime part, right? >> yes, sentence added l477. What are 'fit' uncertainties. A reference should be added, if there is some specific definition. If this is only a linear error propagation of stat. and syst. uncertainties as defined above, then this should be denoted as 'exp'. >> changed to experimental as in our inclusive combination paper l478. Model uncertainties (mod) l482 'parameterisation uncertianty' (par) >> will come after eq.7 l485. '... is only 5 units worse... ' this should maybe go to the results section. >> we think, it can stay here, because here we discuss what has be varied. later we only say what are the dominant uncertainties. L490. The total PDF uncertainty -> Isn't it just 'the total uncertainty' ?? >> yes 'PDF' can be removed here l 496 'd.o.f.' -> n_{d.o.f.} In natural science one commonly uses a single letter for numerical numbers. l496 \chi2 -> \chi^2 >> o.k. l496 is similar -> is of similar size >>OK l499. What are 'experimental' uncertainties here !? >> fit changed to experimental l505. 'experimental uncertainties' :) >>:) l517. ' The running charm quark mass determined here agrees' -> The value of mc(mc) agrees...' >> OK l521. A cross check of what? >>of the fit procedure l521-528. This can be shortened or droped. E.g.: "The uncertainties (which ones?) are found to be consistent with an alternative error propagation using a Monte Carlo method~\cite{xFitter}." l529. Improve these sentences. L529-538. This can be shortened. What is the message of this paragraph? Everybody expects, that when using data, which is not sensitive to those parameters, then the results are not good. >>>>>>> this is not clear - global fitters did extract the masses in the past from inclusive data only. We think, the message is important<<<<<<< L543. Split sentence into two. "... . This is to be expected..." >>OK l546-l555. These lines can be shortened, and discussed together with the previous paragraph (as no additional information is added, but only differently displayed). >> is much shorter now. l556 \boldmath \xbj l561. 'partial chi2 is not defined.' -> It maybe good, to calculate the chisq as a 'full' chisq for these data here. l563 'equivalent to' is written in a different font. Very strange... >>corrected l572. Fig 20 does not show but 'x' >>> changed l572 needs to be defined here. How is calculated ? -> The values are defined as..., and calculated using HVQDIS. (reference is not needed here) >> text has changed , geometric mean l575. This sentence is kind of obvious. >> I think, it is worth noticing that for a given x_Bj charm probes a different range in x than inclusive and beauty probes even a higher range. This is not obvious to everybody. I think if we show a figure we have also to discuss what we see. l566-l579. Does this study mean, that the PDF fit does not correctly 'shape' the gluon?? >> this means that the shape of the gluon we get from DIS is different from what charm would need. l627. 'have significantly reduced uncertainties', compared to what? Compared to individal data sets, of course. Compared to all data sets? well, this is what is presented. >> compared to the individual data sets. l629. Split sentence into two: "...predictions. The charm data..." >> split into two. The calculations are found to be in fair agreement with data. l633. 'provide the best description...... do not improve'. Kind of obvious. >>removed l634. In the text, it is logx resummation, not log1/x -> consistency. >> removed Table 2. Drop the column 'bin' or rename: 'data point number' Table 2. header x-> x_Bj Table 2. The precision of sigma_red should meet the precision of the uncertainties (4 digits?) >>OK Table 4. How the 'reduction factor' is defined? Table 4 caption: datasets - data sets >> we use now 'dataset' consistently Table 5. Dataset -> Data set Table 5. The data set names are not defined. Table 5 does not fit the text-width -> Maybe the last column can be droped, and just mentioned that the PDF uncertainty does not significantly reduce chi2. (although, I really admire this study!!!) >> now charm~[36], charm and beauty; last column dropped and sentence added to the text Table 5 stat, uncorr, cor tot -> {\rm ...} Fig 1 'H1 and ZEUS' labels are missing >> for the pull distributions we normally don't have 'H1 and ZEUS' Fig 1. The numerical values in the box need explanation, or a dedicated selection should be presented. Fig 1. The pull distribution of what ? >>> figure changed and text added - of the input data to the combined cross sections Fig 1. It appears, that there are too many uncertainties with pull=0. Are those understood and included in table 4? >> it's not related to the correlated uncertainties - >>>>>>>>>>>>>>>>>the probability(Poisson, N>=24, expectation=14.5==fit) is about 1.5% - This you have to multiply with the number of bins. Then you know that such a deviation should happen about once in this plot <<<<<<<<<<<<<<<<<<<< Fig 4 and others: having Q2 pads going from top-to-bottom, would allow to zoom into the relvant x-region (as it is done for fig6) Fig 15. 'H1 and ZEUS' labels are missing >>added Fig 20. 'Q^2' and 'data' sould be added to the legend 'charm' and 'beauty' as well Fig 20 should be labelled 'H1 and ZEUS' Fig 20. Is it x or at the x-axis? >> new plot in new draft - this was initially meant only for internal use for discussion Fig21. x-axis: x2/N_dat, but in caption: x2/d.o.f. >> consistent now Fig21. 'when including in the fit only' mayb better: -> 'when using only for the fit' Fig21. x_min -> x_{\rm min}