Actions on draft pap10 for draft pap15 --> 3.9.21 ======================================= A) Handling of decisons made at EB-210824 ========================================= Figures: -- Figures 7/8/9/10 stay --> done nothing -- ZZ: new plot wanted: -- total uncertainties and ratios of all 4 PDFs -- we looked at a first version of this at 10 GeV^2: == no extra information obtained ==> --> internal material for 10GeV^2 style of Fig.4 for NNLO and JetsNNLO ratio plot with total uncertainties Action: KK produced the individual plots =====> IA creates a figure out of it for additional internal material +++++ to be done Content: -- SS:line 305-311: I have great difficulties to understand that text. My proposal would be to remove this, or to say something more simple. Maybe like this: ... questionable. For jet cross section predictions, a variation of alpha_s is similar to a scale variation, and HERAPDF2.0Jets is made available for two different choices of fixed alpha_s. Nevertheless, a cross check ... ==> EB2 Decision: Keep it in and Iris and Thomas work on it =====> modified text after one iteration -- comparison with past results; lines 278 -- 291 the red stuff goes. =====> done Should the rest stay or just be replaced with their is broad agreement with previous results? ==> EB2 Decision: keep as it as far as numbers are concerns and add sentences on consistency. Include NNLOJet result from their erratum. =====> updated numbers =====> consistency --> see Stefan Schmidt -- MS: no line numbers between 34 and 35 ==> EB2 decison: Keep paragraph -- rephrase to make it clearer -- discuss with AMCS, BF, KK & MS =====> modified text after one iteration -- TG: line 47 It is only a standard assumption ==> EB2 decision: l.46 In addition --> even if this assumption is valid. == out new text: In addition --directly--> might be biased ... =====> done -- BF: line 40 explicite inclusions of heavy quark data sets line 78 to 82 sounds like we excluded data -- has to be avoided. =====> sent first attempt to correct to BF 1.9.21 ======================================================================== B) Actions on comments sent via Email by EB members Mandy: line 108/109 although the uncertainties provided on the NNLOjet predictions are correlated the words 'fully correlated' are problematic because we were told to use these 1/2 correlated and 1/2 uncorrelated =====> corrected line 169 why do we need the question mark after 'using the standard procedure', we did use the standard procedure =====> fixed citation line 214 again the recommended procedure for these hadronisation uncertainties was also 1/2 correlated and 1/2 uncorrelated =====> done Line 271 I would prefer to say 'not VERY sensitive to the details of the gluon parametrisation' =====> done Line 272-277, I thought we had been given a look at the erratum for the H1 results [37]--I have included this comparison in my draft presentation. Line 283, the H1 new comparable scale uncertainty is \pm0.0038 rather than \pm 0.0042, but they have full results. =====> fixed with update information Line 289, h1 also have a new comparable result for this fit which is 0.1147\pm 0.0011(exp)\pm0.002(mod)\pm0.0003(param)\pm 0.0023(scale) =====> fixed with update information Lines 303/314 I find this tortured, I hope it will not be needed! =====> EB2 decision to keep and rephrase Lines 331/339 I dont agree with this. If you look at the Internal Extra Material Figure 24 and compare the green uncertainties, which use the jet data but with the old Mc,Mb ranges and old double counting of Mc,Q2o uncertainties, to the blue uncertainties, which use the jet data with the new ranges/procedure, you will see that the new procedure does give us the improvement at low-x, but at high-x it must be the jets and this can be seen in experimental, model and param (although mostly in param). =====> Mandys text will be rephrased Line 369 typo should be 'decrease to 0.0022' =====> done old answers: Thank you, that's why we now have an EB meeting. It is what I undertood. > 2. line 169 why do we need the question mark after 'using the standard > procedure', we did use the standard procedure As asked by EB, I added a citation, which after some bibtex problems did not work and i did not notice -- Sorry > 3. line 214 again the recommended procedure for these hadronisation > uncertainties was also 1/2 correlated and 1/2 uncorrelated Thank you, that's why we now have an EB meeting. It is what I undertood. > 4. Line 271 I would prefer to say 'not VERY sensitive to the details of > the gluon parametrisation' No problem > 5. Line 272-277, I thought we had been given a look at the erratum for > the H1 results [37]--I have included this comparison in my draft > presentation. see above -- these lines are not to stay > 6. Line 283, the H1 new comparable scale uncertainty is \pm0.0038 > rather than \pm 0.0042, but they have full results. see above > 7. Line 289, h1 also have a new comparable result for this fit which is > 0.1147\pm 0.0011(exp)\pm0.002(mod)\pm0.0003(param)\pm 0.0023(scale) see above > 8. Lines 303/314 I find this tortured, I hope it will not be needed! These lines were added by decision of the EB and provided by Thomas. I do not want to rediscuss EB decisions! > 9. Lines 331/339 I dont agree with this. If you look at the Internal > Extra Material Figure 24 and compare the green uncertainties, which > use the jet data but with the old Mc,Mb ranges and old double > counting of Mc,Q2o uncertainties, to the blue uncertainties, which > use the jet data with the new ranges/procedure, you will see that > the new procedure does give us the improvement at low-x, but at > high-x it must be the jets and this can be seen in experimental, > model and param ( although mostly in param). This is basically your text; and it says basically what you wrote here? --> will be rephrased for the reading. > 10. Line 369 typo should be 'decrease to 0.0022' Thank you ======================================================================== Stefan Schmidt this is to let you know about the status of the H1 alpha_s fit erratum, reference [37] of the paper. As Iris sugested, we can discuss in the EB what to do. Here are just the facts, so we can save some time on Tuesday. - the H1 Erratum has been accepted by EPJC. We did update the paper on arXiv:1709:07251 we already sent the corrected proofs of the erratum to EPJC, but it seems it is not yet published, so we do not have a journal reference. It can perhaps be referenced in [37] as "Erratum accepted by EPJC" and then we put the arXiv number at the very end. =====> see updates as provided for bibtex - the updated numbers are: alpha_s fit with fixed PDF and mu>2*mb: updated scale uncertainty is 0.0039 (line 283 of the draft) =====> updated alpha_s+PDF fit: 0.1147+/-0.0011_exp+/-0.0002_model+/-0.0003_par+/-0.0023_scale (line 289 of the draft) =====> updated We also have an erratum for the jet cross section paper ref [13] in a similar state. That erratum for [13] is only relevant for the bibliography; the data did not change. =====> see updates as provided for bibtex All the best, Stefan - - - - - - - - - - - - - - - - - - - - - - Matthew Wing I believe the errata are at: https://link.springer.com/article/10.1140/epjc/s10052-021-09370-8 https://link.springer.com/article/10.1140/epjc/s10052-021-09394-0 They were published online by EPJC yesterday. - - - - - - - - - - - - - - - - - -- - - - - - - @article{Currie:2017tpe, author = "Currie, James and Gehrmann, Thomas and Huss, Alexander and Niehues, Jan", title = "{NNLO QCD corrections to jet production in deep inelastic scattering}", eprint = "1703.05977", archivePrefix = "arXiv", primaryClass = "hep-ph", doi = "10.1007/JHEP07(2017)018", journal = "JHEP", volume = "07", pages = "018", year = "2017", note = "[Erratum: JHEP 12, 042 (2020)]" } =====> implemented @article{H1:2017bml, author = "Andreev, V. and others", collaboration = "H1", title = "{Determination of the strong coupling constant $\alpha_s(m_Z)$ in next-to-next-to-leading order QCD using H1 jet cross section measurements}", eprint = "1709.07251", archivePrefix = "arXiv", primaryClass = "hep-ex", reportNumber = "DESY-17-137, DESY17-137", doi = "10.1140/epjc/s10052-017-5314-7", journal = "Eur. Phys. J. C", volume = "77", number = "11", pages = "791", year = "2017", note = "[Erratum: Eur. Phys. J. C 81, 738 (2021)]" } =====> implemented @article{H1:2016goa, author = "Andreev, Vladimir and others", collaboration = "H1", title = "{Measurement of Jet Production Cross Sections in Deep-inelastic ep Scattering at HERA}", eprint = "1611.03421", archivePrefix = "arXiv", primaryClass = "hep-ex", reportNumber = "DESY-16-200", doi = "10.1140/epjc/s10052-017-4717-9", journal = "Eur. Phys. J. C", volume = "77", number = "4", pages = "215", year = "2017", note = "[Erratum: Eur. Phys. J. C 81, 739 (2021)]" } =====> implemented ---------------------------------------------------------------------- Hi Iris, dear EB members, thank you Iris for all the editoral work. I found a few places in the text which perhaps can be improved a bit. None of my suggestions is such that it is necessary to discuss them in the meeting today, I think they can be handled by the editor. Anyway, I am sending my ideas below. All the best, Stefan ------------------------------------------ line 47/48: "It is only a standard assumption ... hadron-hadron interaction processes" to my taste this sounds a bit too strong, because there are a number of processes where factorisation is valid in hadron-hadron. my proposal: to "invert" the statement, something similar to this: , whereas in hadron-hadron collisions factorisation is established only for certain processes such as Drell-Dan production. Alternative: perhaps add the words "in general" at the end of the original statement: ... hadron-hadron processes in general. =====> solved by EB decision line 278-285: I think that here some interpretation is missing: why are the scale uncertainties when using fixed PDFs larger than in the PDF+alpha_s fit? Now we simply list the numbers but do not interpret them. We could add something similar to this: "The somewhat reduced scale uncertainty in the present fit is attributed to the extra sensitivity of the inclusive DIS data to alpha_s, originating from the PDF evolution." I do not really understand this sentence. The EB said something about consistency. I think that the text on scale uncertainties and local minima might be relevant here. ======> I tried my best to accomodate all wishes on this paragraph. line 305: I think we should insert here the words "for the jet cross section predictions alone": ... in the fit for the jet cross section predictions alone becomes questionable. ======> paragraph was reworked with TG line 305-311: I have great difficulties to understand that text. My proposal would be to remove this, or to say something more simple. Maybe like this: ... questionable. For jet cross section predictions, a variation of alpha_s is similar to a scale variation, and HERAPDF2.0Jets is made available for two different choices of fixed alpha_s. Nevertheless, a cross check ... ======> see new paragraph developed with TG ==================================================================== Mark ----- On 23.08.21 16:53, sutt wrote: > > > Hi, I have had a look through the draft and had a few points, and several small changes for the text. However, > I think that most of these small changes could be address be email at the "reading" stage. > > However there were a where I thought some discussion might be useful for my own clarification as much as > anything, hence my writing to this shorter list as I did not wna to bother everyone else if it turned out that > there was no issue. > > First some minor points which I rink are none-the-less important. Firstly, on line 349, it says "provided by applfast", really this should state "caclulated by NNLOJET". APPLfast is just the mechanism to provide the provide the results of the calculation, so could that be changed ? This could not be changed before EB2 as the draft was out. =====> done Also, for lines 272 - 291, naturally some of that is a request, but my own feeling is that we should not dwell on the previous calculations ie we do not need to quote the values and I feel that a statement such as "broadly in agreement with previous determintions [37,28]." would be ok. The text as written is not strictly correct, since while the APPLfast paper did have an alphas fit using only fixed PDF's, the "H1 paper" actually *did* include a PDF fit, although this was not the main result, but a cross check that was not clearly reported in the "H1 paper"" or at least the actual resulting PDF was not discussed more that the PDF parameters reported in a table, This was only using a ZM-VFN ======> EB2 decision was to keep it. Therefore, if saying "cannot be compared because they used a fixed PDF" then you would need to qualifiy this as - I paraphrase, this is not a verbatim suggestion - ******* The APPLfast measurement [37] and the primary measurement from H1 [38] used fixed PDFs and so are not comparable > and then just not mention the 'PDF fit" alphas measurement at all, or mention that the value with a "fit of the PDF parameters was performed only as a cross check", but Daniel will likely take extreme exception to statements that the PDF was not fitted, so the best approach is to not mention it or to state "and is also consistent with the cross check from the H1 analysis, where asimultaneous fit of the PDF was performed, albeit with a ZM-VFN scheme." Or something like that. I don;t think we need to provide the actual alphas values. The H1 people might complain because they want to take all the credit for that result, but of course, as they are authors of *this* paper also, then these is no need to quote the result in full, and indeed it would not usually be the case, and in fact, in that paper, they did not quote, or compare with any previous values of alphas, even from their earlier determinations, so the precedent *not* to actually quote the earlier values was set by H1 itself. In the EB, I will suggest that these previous value be removed, since it reads more like a review, rather than a fit paper, and I don't feel such a review has a place in this paper. This text came from Mandy on request of H1. I adjusted it to the treatment of scale uncertainties as were decided by Thomas for this analysis. During EB1, this was agreed upon. EB2 upheld this decision -- =====> I implemented some modifications For my own information I was somewhat puzzled by the (unnumberd) paragraph in the introduction that the proceedure is to first simulataneously fir alphas and the PDF, then to fix alphas to the value obtained by that fit, and fit again. This makes little sense to me, since clearly if you fix a parameter to that value obtained in a fit where that parameter was free, then by definition, fitting the same remaining parameters must give the same overall best fit with the same parameters. What would be different, would be the uncertainties, since with the parameter fixed, then the correlation of that parameter with the others would be gone, so that proceedure would only be used as an alternative estimate of the uncertainties, and not to the actual fit itself. In this case, it should be presented as such, ie "An additional estimate of the unceratinties fixes alphas at the values from the fit and ..." Would this be correct, or would I be missing something ? Of course, if you were to *add* additional parameters to the fit with free alphas and then estimate the parameterisation uncertainty, you would get a better fit, so if this fixed alphas were also used in the param. uncertainty, then the statement about being the same fit would not apply, but then these changes could not be used an as uncertainty on the alphas value or the paremeter uncertainty of the free alphas fit, just as param uncertainties on the fixed alphas fit. I think a sentence added in the introduction, and in the section where this is discussed might be a good idea. I think this paragraph is not needed at all. However, during EB1, it was decided that I write this paragraph -- came from an H1 request. EB2 decision: keep the paragraph and rewrite ======> done I still wonder about this still being the "2.0" fit, as it makes we wonder what would be needed to become a "2.1" fit, or whether it would be a mode directly to a "3.0", in which case what is ".0" actually for ? Although I won't labour the point any more than I have already - don;t worry it is not my intention o drag this up again in any meetings, but it was just brought to mind by the discussion about completeing the HERAPDF2.0 "ensemble", although wouldn't "family" be more usual ? Orginially, there was a plan which included also some refitting at NLO with and without HQ data and others. Indeed, the refits lead to 2.5 and finally to a 3.0. However, the experts could not agree on any procedure to be adequate. Discussions were endless as we didn't have grids anyhow. In the end, the program was reduced to the NNLO analysis as presented now. Another endless discussion on the number started to focus on the idea that 2.5 would be "not final" and 3.0 would indicate superiority which was not applicable because there had not been an NNLO Jets fit. Therefore, we agreed finally that it is a member of the HERAPDF2.0 family. Yes, I agree it is a family and it was also described as one before EB1. but I was told to take family out and replace it by set. But we call a set of PDFs what comes out of one fit. Thus, I invented ensemble. EB2 decision: keep ensemble > ========================================================================== After EB ==Mandy=================================================================== The unnumbered paragraph between 34 and 35, maybe it would be easier to understand without the sentence on avoiding bias by not using fixed pdfs. That could go elsewhere- see later near line 278/285 =====> done =====> yes it was rephrased until Mark, Brian and you were happy. Line 108. I have already mentioned that uncertainties are applied as 1/2 correlated and 1/2 uncorrelated. Maybe in this sentence if you drop the world 'fully' and then the words 'were taken into account' covers it? Well I plan to use the phrase as we had it for scale uncs. "50% correlated and 50% uncorrelated between processes and bins" This is correct? ---Mandy: -----OKAY fine =====> done Line 172 when you say 'to facilitate the pQCd analysis at NLO published previoulsy' it sounds as though we are going to go back and re-analyse that. I thought the reason was more that pQCD analyses at NLO maybe done at the LHC using the xFitter framework which makes use of our results and procedures. Maybe one does not actually want to say that so something like 'to facilitate future NLO pQCD analyses'? =====> rephrased Line 177 saying 'at NLO alphas=0.118 was used' makes it sound as if we are doing an NLO analysis here, so this would be the place to say 'in the published NLO pqcd analysis [28] the values alphas=0.118 was used' =====> well, it was not published in [28] and I do not want to write that we forgot that. -- rephrased Lines 212-218 I may have said this before. The hadronisation uncertainties are actually applied 1/2 correlated and 1/2 uncorrelated not fully correlated. I don't know if we want to give such detail. why not? ---Mandy-------OKAY then it is same as what you said above-------- =====> done Line 271 'is not very sensitive to the details' because it is sensitive to the gluon at some level. =====> done Lines 278/285 Here after line 279 is the place to say that fixed PDfs could lead to biases in the value of alphas extracted and to refer to reference [24]. =====> done Line 283 the new H1 value to replace the red values is \pm 0.0038 =====> done Line 285 the new NNLOjet value is \pm 0.0033 =====> done And carrying on into the next paragraph the new H1 result to replace the red values on line 289 is 0.1147 \om 0.0011(exp,NP,PDF) \pm 0.0002(model) \pm 0.0003(param) \pm 0.0023(scale) =====> done You could mention that it is only a ZMVFN scheme used if you like but I dont think we need to. The EB didn't discuss it. However, it was also noted by Marc. =====> put in Lines 303/314 I yunderstood what Thomas meant much better on the second time I read it. I hope you can sort something out with him. =====> rephrased I know that lines 330/339 are not blue, but if they come from me I would like to improve them. Commas around 'compared to HERAPDF2.0NNLO' then after 'were reduced,' but this had only a very small effect. The major effect came from the need to symmetrise the variations of \mu_f0^2 and m_c^2 as discussed in Section 3.3. This removed a double counting of sources of uncertainty that had been present in the orginal HERAPDf2.0 procedure.' =====> implemented and adjusted to rest of text. Line 343/344 to be honest I don't think the extra plots in Appendix B really add anything. EB2 decision: keep Line 369 typo its 0.0022 not 0.0021 =====> done Line 383 typo on/one =====> done Acknowledgements. Please can we thank the Leverhulme Foundation. It is where my money is currently coming from. =====> done Line 479 in internal material. The \mu cut is >10 not 13.5 here, the analysis was repeated for the new cut/grids Similarly line 490. =====> done ====Matthew=================================================== - l.10. “… order (NNLO) analysis …”. =====> done - l.31. “next-to-next-to-leading order (NNLO)" =====> done - l.169. Question mark in citation. =====> fixed - l.332. “but but” -> “but”. =====> fixed - l.383. “on” -> “one”. =====> fixed - Table 1, caption, l.2. “its” -> “fits”. =====> fixed - Figure 1, caption, end of l.2. “… 0.1155 and b) and d) the …”. =====> fixed ===Mark===========> pdf file ===Brian==========> docx file