Comment on reading version 1.6. and actions =========================================== --- Peter Bussy -------------------------------------------------------- This is a very impressive piece of work. Inevitably, the outside reader has a few comments on the text..... Abstract 11: This paper presents the final..... -> was rephrased 13: This needs easing a bit. You don't make it very clear which PDfs were used where. PDFs have been determined from a fit to the data. They are used together with.....to determine..... [Distinguishes better between the past work and the present.] -> some rephrasing done 15: Omit "The". PDF sets for HERAPDF2Jets -> no NNLO were then determined from fits...... [But this still requires a lot of prior knowledge.] -> I hope it is okay now 16: I think you had better say 0.1155 (rounded) and... -> long discussion ended in do nothing 18 What kind of inclusive data? -> specified 20: These are hardly predictions since the data were used to calculate them. You might get away with this better by writing: "As a consistency check, ..." A further improvement would be to replace "predictions" by "calculations" -> These are predictions based on PDFs - a lot of other stuff goes in Intro 27-28. Maybe merge the two paragraphs. -> done 34: I think it is necessary to be explict about what "jet data" means. Inclusive single-jet? Or what? -> relegated to the section on data 35: Not originally possible at the time.... -> feasible 36ff: I would spell it out more clearly. "First, the jet cross section data....PDF. Since this is correlated with....a simultaneous fit to .... was now performed to evaluate alphasMz. A new fit to the PDFs was finally performed with alphazsZ fixed to this value. The PDFs were also determined..... In this way, the resulting PDF uncertainties are uncorrelated with that of alphaSMZ." [Is the last sentence actually needed here?] -> This suggestion was followed with some rephrasing This paragraph was not in the original darft and was requested by EB after some comments from outside readers. 43: Suggest: ...for the PDG18 value, alphas= etc.[ref] -> it is no the cobentional value 52: Not quite clear. Do experimental data sets involve mass parameters? Presumably, but this isn't very obvious. Maybe: ...in evaluating the two experimental data sets. [Or else I have misunderstood, in which case please clarify!] -> The truth is that H1 and ZEUS experts could not agree on how to do it. Now we don't say at al why it is difficult 54: Now you say that the HERA inclusive date are a single and consistent data set. This makes it all the more needful to clarify the previous paragraph. -> It has nothing to do with the inclusive data set. 55: Omit "Furthermore" -> done 57: Omit "In addition" -> After strong feelings about it in the EB, all this is now gone. Probably downgrading the importance of the paper on purpose. 60: I think the final sentence could go. The previous sentence is strong enough. -> we kept it -- the paper has to have some purpose Data 71: Make it clearer in the text whether or not there is overlap between the old and new data. I think it would help if Table 1 had a new first section (use a horizontal line) to include the original data sets in the list . -> There is now only one citation for new data and text was rephrased, citations reordered. Table was not changed. 74: were applicable only to... Or omit "only"as well -> was rephrased 77: ...perturbative series in the NNLO calcuation, and... -> the perturbative series was eliminated 76: Could omit "All" -> done 79: I.e. the b mass is treated as zero in the calculation? Perhaps say ... larger than the physical b-quark mass,... -> was kept in reading after some discussion 87: were used only..... quark mass parameters... There seems to be an inconguity between zero and finite quark masses in different parts of the calculation. No doubt this can't be helped, but it might be slightly better to replace "needed" by "required" in l88: (Of course this is all spelt out in 3.3) -> yes -- a bit involved -- some rephrasing done used required in post-reading draft QCD analysis 94: Why not 2.0? Maybe add a phrase to justify 1.9 -> tradition -- as low as possible, but above charm mass 97: I'm usually a stickler for GB spelling, but perhaps we just have to use "program" for a computer program. (Well. at least you're not calling it an app!! -> programme was forced upon me -- changed happily to program 105: The present analysis.... [snappier] -> done treatment -> calculation -> done 106: approximation to what? -> to reality, what else? 3.1 133: Omit actually. Perhaps used -> required, which is more explicit. -> done 3.2 157: the central fit This is the definitive fit, right? -> eliminated central 171: A little too terse - I suggest adding a few definite articles. -> done Moved this to the end of the discussions on uncertainties. 180: Another hyphen after "one"? or maybe just "1-$\sigma" or just "The uncertainties on..." -> rephrased 186, 188 Myself, I don't understand the term "fixed point" since it seems that the quantity varies and is not fixed after all. Do you mean "initial value"? -> rephrased 265: arose within -? was present in [Good, the questions seems solved now.] -> was kept [as the question is still asked] there isevidence for highr twis up to 100 GeV^2 277: Can we omit the somewhat vague "very" -> Mandy insted on it during the reading 300: I think we should add something like "rounded to the nearest 0.0005" -> there is now a sentence about correspondence... 303: omit "both," or at least the comma. -> done 304: You don't need to say "together with their uncertainties". This is automatic. -> done 314: ...local minimum, which is required ... omit "unknown amount of". We know this by now. -> done 335: The ranges of variation of Mc.... -> done in post-reading draft 358: Needs a bit of simplification. ... NNLO fit the data on jet production well, demonstrating the consistency of the inclusive and the jet-production data sets that were used in the present NNLO fit. [as in the conclusion] [To a limited extent you are comparing the jet data to itself, and so the statement here should not be too strong.] -> was rephrased. 382 Better to omit "very", which is subjective. -> rephrased -- excellent -- also sunjective, of course... 385 ? ... and is one of the legacies... --> done --- Brian Foster --------------------------------------------------------- Title: “Impact of jet-production data on the next-to-next-to-leading-order…” -> done Abstract: “…2015. The final stage is presented, a next-to-next-to-leading-order (NNLO) analysis…” -> done Line 14: comma before “with the result” -> done 16: “… with separate fits using two fixed values of alpha_s(..), 0.1155 and 0.118, since the latter…” -> done 19: “…the consistency of these PDFs…” -> done Do we need the last sentence? Although it is of course more complicated than this, but the non-expert reader will find it strange that we say that the input to the fit is compatible with its output. -> sentence gone Introduction -------------- 26: Don’t we normally say “Quantum Chromodynamics”? -> done 33: “next-to-next-leading-order (NNLO)…” -> done 39: “…jet cross-section data…” -> done 45: “The calculation of jet cross sections…” -> done 50: “However, no attempt was made to include the heavy-quark data explicitly in the pQCD fits including jet data, since the mass parameters were treated differently in the two data sets.” -> replaced by a full-stop. 56: Start a new paragraph instead of the current sentence “In addition…”: “The HERA-only PDF fits presented here serve another purpose. The PDF fits to LHC data might be biased by any physics Beyond the Standard Model whose effects have so far escaped detection, thereby reducing the sensitivity of searches for BSM. The HERAPDF2.0 ensemble of PDFs provides a benchmark to which PDFs including data from LHC colliders may be compared. This could reveal BSM effects or the need for an extension of the QCD analyses for some processes.” -> the paragraph was butchered. Some of the suggestion survived 65: delete “in a coherent way” -> done 67: “…ZEUS. These data were also used for the present analysis at NNLO.” -> rephrased ZEUS, which were ... 69: “…were added in addition to the present NNLO analysis. -> rephrased These data reach to lower Q^2, where Q^2…” -> done 72: Delete “A summary…Table 1” and -> the tow sentences were exchanged and the now last sentence rephrased. add the next sentence to the end of the previous paragraph. 77: Surely we do not need “of the theoretical predictions”? -> yes, we do 79: “..which is necessary because the NNLO calculations assume massless partons.” -> no 80: I don't understand the last part of this sentence. -> New text from TG 82: Delete “The resulting reduction of data points is detailed in Table 1” -> no 84. Add at end “Details of the data used are summarised in Table 1.” -> no 85: I still don't understand how this paragraph relates to the statement starting on line 50, Anyway, we could say “The inclusive charm data [29], which were included in the analysis at NLO [2], and the heavy-quark data [28] were only used to optimise the mass parameter values or charm, Mc, and beauty, Mb, which are needed as input to the RTOPT [27] NNLO approach adopted to fit the inclusive data.” -> is happy now. -- left as is 92: comma before “with” and before “were” -> done Surely we should repeat the chi^2 definition here rather than force people to go to a reference to find this key information? -> surely not -- 1 page of technicalities 100: Delete “as previously…[2]” -> rephrased Delete “i.e….uncertainties.” -> done 109: change to “fast grid-interpolation codes” -> done 120: what are we saying is responsible for the improvement – NNLO vs NLO or the different scale? Clarify. -> a plural now indicates that it is a combination 122: “scale variations were used…” -> no , they re not always used 148: delete “the determination of” -> done and sentnce adjusted 151: “14-parameter fit..” -> done 153: “..inclusive data [2]” -> done Delete “The jet data did not require extra parameters” -> done 155: delete everything after “region” -> sentence was eliminated 157: What do we mean by “a central fit” – do we mean “the central fit(s)”? -> rephrased 158: “by using fits with modified…” -> done 167: I don't understand – if adding D_u_v (?) is significant, why wasn;t it retained in the central fit? -> because the results are not liked and the fits do not like to converge 168: “The maximal deviation of the fit at each x value was considered an uncertainty, forming an envelope around the central fit.” -> used 180: “one-standard-deviation” -> done Footnote 4 “..values was observed..” -> done Footnote 5: “..heavy-quark data” -> done 184-186: “As a first iteration at NNLO (NLO), the mass parameter values used for HERAPDF2.0 NNLO (NLO) were used as fixed points, so that Mc was varied with fixed Mb =4.5GeV (4.5GeV) and Mb was varied with fixed Mc = 1:43GeV (1.47GeV).” -> used 189: “to within 0.1%... were obtained.” -> used 191: Can we say “compatible with” rather than “quite close”? -> done 194: I don't understand what we mean by “the technical limit of the fitting procedure” -> explanation added 195: “The model uncertainty on the heavy-flavour mass parameters would normally have been obtained by varying Mc by its one-standard-deviation uncertainty.” -> partially done -- it would not be correct like this . 198: “... scheme, in order that charm …” -> done 202: “…the upper uncertainty value of Mc.” -> rephrased 203..”…and, again, the resulting…was symmetrised.” -> done 206: Do we really need this last paragraph? It only says what we didn't do. -> taken out At some point, it was considered absolutely necessary to justify not using other heavy-flavour schemes. 212: “as determined in the original publications, were…” -> done for 213: replace “while for technical reasons” with “;” and add a footnote after “increased”: -> done Footnote{This increase was done for technical reasons that have no bearing on the current analysis.” -> done and shortened as the second half is already in the text. 219: No new paragraph -> done Don't we want to say why we have done something different. “In order to reduce the importance of the hadronization corrections, a different procedure was employed here.” Or something? -> no -- we just improve it... 229: Delete “Inclusive…QED vertex, and, thus,” which I don't understand and is anyway irrelevant. -> Peter has asked for it but did not insist ==> out 236: “approximate proxy” is ghastly – delete “approximate” -> done Replace “due to the unknown influence of” by “from” -> no -- it would not be correct 239-240: “maximum” not “maximal” -> done Replace “of the results as the de facto….uncertainties.” with “, which were observed at…” -> were assigned as the scale uncertainties 245: delete “the result of” -> done 251: I feel the need for some more explanation here – why have we taken 100% correlation here when we assumed 50-50 before? Particularly as at the end of section 3 we discuss 50-50. -> trijet and heavy-quark data are given as excuse In reality, it was a decision by TG after a long fight between H1 and ZEUS experts 287: Do we really need the last sentence of the paragraph? It is obscure speculation that I for one don't understand. -> At some point, it was very much wanted by some dear "colleagues" Now it is gone. 290: Do we understand why it would appear from these values that adding the ZEUS data makes no difference to the errors. This is, to say the least, surprising and would be much more worth a comment that at the end of the previous paragraph. -> There is reduction and at this point nobody wants to say anything. 299: “ analysis (see Appendix A).” -> no 303: delete the comma -> deleted also both 306: delete “as” -> done 307: delete “the uncertainties” -> rephrased 308: This is a horribly complicated paragraph which readers have to wade through only to be told that it is irrelevant. Replace with “A quantification of the influence of higher orders by varying the renormalisation and factorisation scales is questionable in fits in which the value of alpha_s (M_Z^2) is kept constant*. Nevertheless, a cross-check with scale variations as described in Section 4.1was made. The impact on the resulting PDFs was found to be negligible compared to the other uncertainties presented in Fig. 3.” The footnote could be {*Any variation of the renormalisation scale effectively amounts to a modification of the value of _..., since the compensation with the explicit scale-dependent terms in the NLO and NNLO coefficients is incomplete. If a fit is performed with a fixed value of …, it might thus not reach a local minimum. However, such a local minimum is required to estimate the unknown amount of influence of higher orders by varying the scales.}, although personally I think the footnote could be omitted. -> it stays The EB decided upon it long ago 312: misspelling “modification” -> done 315: delete “amount of” -> done 328: replace sentence starting “However” with “However, the HERAPDF2.0Jets NNLO analysis results in a significant reduction of the uncertainties on the gluon PDFs, as shown in Fig. 7 at the scale of _f = 10 GeV2 and in Fig. 8 at the scale of _f = …” -> done 333: insert comma after “important” -> done 334: replace the comma with “for HERAPDF2.0Jets NNLO” after “10^-3” -> done 335: delete both commas -> done 336: delete comma. Replace “but this had only little effect” with “, with minimal effect” -> rephrased 337: insert comma before “as” -> done 339: “HERAPDF2.0” -> done 340: delete comma -> done 345: “Other selected ratio plots..” -> done 348: “Each figure a) gives a direct comparison of the cross sections, while b) shows the respective ratios.” -> done and further rephrased 354: “cross-section calculation” -> done 355: insert “of each jet” after “momentum” -> the two jets 358: “..clearly give a very good fit to the jet-production data, showing that using both the inclusive data and jet production data as input to the NNLO QCD fit gives fully consistent results.” -> rephrased 367: delete “it” -> done 371: comma before “which” -> done 374: Surely we need to say where they are available, probably in a footnote. -> is in Appendix A 374: Replace the sentence starting with “Comparisons” with “Comparisons between the PDFs of HERAPDF2.0Jets NNLO obtained for the two values of …, as well as comparisons to HERAPDF2.0 NNLO, for which jet data were not used as input to the fit, show they are consistent over the whole kinematic range.” -> no sentence was seen as too involved. 379: Delete “on balance,” -> rephrased 380: replace sentences starting “Predictions” with “Predictions based on the PDFs of HERAPDF2.0Jets NNLO give an excellent description of the jet production data used as input.” -> done 384: don't we mean “self-consistent”? or maybe we mean “is fully consistent with the framework of pQCD”? -> used self-consistent 387: Where did we get this first sentence from – it is clearly false? Replace with “We are grateful to the HERA machine group, whose outstanding efforts have made the analysis presented in this paper possible.” -> was modified 414: “arxiv” missing before the reference -> fixed 418: “J. Currie et al.” -> stays up to 4 -> fixed arXiv 423: “D. Britzger et al.” -> fixed 425: Reference needs to be rewritten in the proper format -> there is only a URL... but I did the rest [never got a working reference from Britzger/applfast] 427: the arxiv reference looks wrong -> changed to hep-ph/ 444: “A.D. Martin et al.” -> stays up to 4 Caption to Fig. 1 : “The difference between the fit chisq and the minimum chisq for both HERAPDF2.0Jets NNLO fits with fixed _s(M2Z) = 0.1155 and NLO fits and fixed _s(M_Z^2) = 0.118, as indicated in the legend. For a) and b), M_b was fixed at 4.2 GeV; for c) and d), M_c was fixed at 1.41 GeV.” I got as far as this and then I realised that the caption is hopelessly confused. We say e.g. that b) has M_b = 4.2 GeV AND 4.3 GeV! It needs to be rewritten along the lines of what I started to suggest above with self-consistent values. --> caption was optimised for space --> rewritten for clarity as longer text Fig 2 – split into 2 – One is just a), the other b) and c) --> done – then we can expand b) so that we can show all the points. --> was decided against because ths scale is to be kept as for inclusive only Fig. 11 – I don’t understand why the top plot doesn't have ta line for he inclusive data alone. --> green line sits at =1 -- was added for avoiding misunderstandings --> black line became blue Fig. 12 – the caption for a) needs to say that the error bars are smaller than the symbols. --> done Figs 12 - 20. The captions shouldn't start off with saying the data is H1 or ZEUS. We are not presenting H1 or ZEUS data here but our results in comparison to H1, ZEUS. Say “Differential jet cross- section predictions dsigma… based on HERAPDF2.0Jets NNLO in bins of Q2 between 5 and 100 GeV 2 compared to H1 data. The bands represent…” --> all rewritten according to reading based on your suggestion I haven’t looked at the Appendices. --> you missed the best.... --- Achim Geiser --------------------------------------------------------- Thanks a lot for bringing this paper to reading level. I read it and will make only one remark: The paper is very technical and we are strongly underselling its value by focusing mainly on the PDF aspects (see title), on which the impact is minor, and only very little on the strong impact on the improved determination of alphas, i.e. its actual physics result. The paper does not include a single plot or table illustrating the alphas improvement with respect to previous results and with respect to the rest of the world, nor a corresponding discussion. So I am rather disappointed how this paper has evolved and wish you a happy reading. -> as the paper follows the EB decisions, no action was taken. --- Thomas Gehrmann ------------------------------------------------------ Data Section: In addition, for each Q2 interval, the six data points with the lowest $\ptav$ were excluded from the ZEUS data set. Due to the kinematical cuts applied for the selection of di-jet events, the Born-level two-jet contribution vanishes in these bins. Consequently, the NNLO theory predictions for di-jet production amount only to NLO accuracy in these bins. Likewise, the tri-jet data [14], which were used …. were excluded as no NNLO theory predictions were available for them. --> implemented --- Max Klein ------------------------------------------------------------ admittedly almost too late, I looked at the paper you read tomorrow, which is very interesting and surely reflects much work, also on the theory side. I have 3 comments: a) I am very happy to see that Q2min has no influence on alphas; b) you compare with the H1 alphas value and it looks pretty much the same, one wonders what is new, which is not meant provocative; c) I believe the conclusion from fig 7 is not about a significant improvement, hardly much is different, but about the inclusive data being precise enough to determine the Pdfs (i think one would like to see this comparison made for all pdfs you fit, my guess is there is no change of uncertainties.?) isn’t the in my view strong result then that one may continue to use Herapdf2.0, for what it’s worth, but based on the H1 result and this result now can choose alphas as about 0.15, Not .18, which we already found in the H1and Bcdms NLO inclusive analysis, worth citing in support of your low value! I recall that W van Neerven asked us to keep Qcd analyses of inclusive and jet data separate because may be xg was different. The fact that you get the same means sth one also could stress. thank you for this work, take maybe time for the physics analysis discussion, it now reads a bit like a fit report. best wishes , Max -> as the paper follows the EB decisions, no action was taken. --- Florian Lorkowski ---------------------------------------------------- In line 339, it says "HERAPDf2.0". The "f" should be capital. --> done In the caption of figure 2, panel a) and b) are described as "for the inclusive data". This should read "for the inclusive and jet data". --> Fig.11 The black curve in figure 11a serves the same function as the blue curve in 11b-d. It should be made blue as well. --> For completeness, one could also think about adding the green line to the first panel. It would lie exactly at 1 and help to judge the reduction of total uncertainties --> alternatively, one could draw a dashed black line). Also, there is still room on the page to make the figure bigger in the vertical direction. In its current form, it is a bit hard to read. --> very little The caption of figure 20 is hard to read. [now Fig.21] If it were written similar to figure 11, it would be easier. The new phrasing could be along the lines of: "Ratios of uncertainties relative to the total uncertainties of HERAPDF2.0 NNLO αs(M2Z) = 0.118 a) experimental b) experimental plus model c) experimental plus parameterisation uncertainties for HERAPDF2.0Jets NNLO αs(M2Z) = 0.1155 and HERAPDF2.0 NNLO αs(M2Z) = 0.118, at the scale μf2 = 10 GeV2." The same applies to figures 21, 22 and 23. --> implemented Figure 21 and 23 say "mu_f^2 = M_Z^2 GeV^2". The "GeV^2" should be removed. --> now Figs 22 and 24 --> yes --- Klaus Rabbatz -------------------------------------------------------- Comments on physics description: ================================ Introduction: - l44: Please do NOT quote 0.118 as "the PDG value", because it's not. Suggest to add "(rounded to three digits)". This is BTW also true for the PDG2020. -> conventional value - l51/2: Could you be more specific what is meant by "different treatment"? ZM versus GM? -> we now give no reason - l56ff: IMHO the statement on a potential bias of fits using LHC data can be made in a similar way for HERA data, too. Imagine you have BSM effects on HERA data that escaped so far detection(!), but which are responsible to make our "wrong" SM predictions compatible with the data ... then we'd have them, but couldn't notice anything so far. Suggest to rephrase without speculating about biases at the LHC. -> after a long discussion on energy scale etc the whole thing was rephrased Data: - l73: Suggest to use "jets were chosen to use the kT ...", because for some datasets there are alternatives. -> we do not discuss alternatives - l78: The numbers 10% and 24% suddenly appear here. Isn't there a reference or some other motivation? -> only the 10% are now mentioned -- rephrasing does not include motivation [10% was believed a standard number] - l81/2ff and footnote 2: These statements have not the right twist to the problem. The omitted ZEUS dijet data unfortunately are simply defined such that with the existing code NNLO predictions are not available for them in contrast to other phase space regions where the lowest order process is not suppressed. As for the footnote: The "NNLO term" is NOT "the second non-zero term"; the NNLO term for these bins has not yet been calculated since one order higher in alpha_s. If emphasising that for some data NNLO simply is not available, then all three exclusions, ZEUS points, trijet, inclusive charm, can be explained in one go. -> a new text was adopted QCD analysis: - l101: "fit uncertainties" means experimental & hadronisation uncertainties. For the comparison of two codes that supposedly use the same data and same fit technology, these uncertainties are irrelevant. Both codes should reproduce the same results AND uncertainties within their NUMERICAL precision. Please rephrase. -> done - l115: Please explain which uncertainties are meant here. The numerical uncertainties included in the NNLO predictions? Or something else like scales? -> included - l213: Can't "technical reasons" be better explained? -> no - l224: The procedure is changed and one observes a reduction in uncertainty. But that does not justify the change. Maybe better emphasize that the 100% correlation of the offset method was maybe exaggerated? -> was done HERADPDF results: - paragraphs ll 279-296: First of all, the H1 publication [37] also has a combined PDF+as fit as discussed in the second paragraph. This should be mentioned earlier to avoid the impression that there are two different H1 publications that are discussed. Secondly, to mention the Kassabov-Forte paper in connection to the other fit of H1 or the one in the APPLfast paper is not quite right, because the fitting methodology is different from what is critisised by Kassabov-Forte. There is no simple parabola vs. alpha_s with fixed input PDFs used without account for uncertainties. This is explained in detail in the H1 paper. So claiming a bias of unknown magnitude referring to Kassabov-Forte is disputable. Moreover, the alpha_s used in the evolution is, consistently with the hard process, the fitted alpha_s parameter. -> the editor would have been much happier if the people involved would have agreed much earlier what they want. There was a long section on this which was written on request of H1 and eliminated on request of H1. I can only hope that the lines left after the reading are the final version. Text comments: ============== Abstract: - l13: pQCD is undefined, suggest to spell out in abstract -> done - l18/9: The sentence "The consistency ... data." doesn't really tell much. Suggest to drop or rephrase, since the last sentence really explains, why all this is consistent. -> rephrased Data: - l70: Move the "squared" after "transfer" -> moved before four- - l74: "applicable" is the wrong word; probably you mean "applied". -> rephrased - l76: Suggest to rephrase: "... had to be excluded in order to ensure the convergence ..." --> "... had to be excluded to delimit the impact of higher-order corrections of the perturbative series and hence the NNLO scale uncertainties to below ..." -> rephrased without convergence QCD analysis: - l103: RTOPT is described here slightly differently compared to ll 46/7. Can this be unified? -> done VFNS RTOPT - l184: Looks strange. Can't the two footnotes be unified into one? -> actually not -- they adress different things HERADPDF results: - l239: Essentially this is again the offset method, just for the list of scale variations. -> slightly rephrased - footnote7: This is cumbersome. When writing variation of scale factors to (mur/mu0, muf/mu0) one can simply list (0.5,0,5), (1,1), ... instead of all the mu's inside parentheses. -> better too clear than asking for previous knowledge - l339: Typo HERAPDf -> fixed --- Mark Sutton --------------------------------------------------- As GeV is a unit it should not be in "italic" but in "roman". However, the captions were written in italic. After a long a discussion, the editor was asked to decide herself. The editor went back to the caption style in the 2015 paper. Now the captions are in "Roman" anyway. Problem solved. -------------------------------------------------------------------