Date: Fri, 5 Aug 2005 16:52:44 +0900 From: Masashi YOKOYAMA To: scifimapaper@neutrino.kek.jp Subject: Re: MA paper draft Dear Rik and all, In general, the draft is well written, but still needs more improvement before the circulation to the collaboration. My comments for the first half (before analysis) follow: ********** Abstract: - "stat." and "syst." are swapped in the presentation of result! I. Introduction - In the first paragraph, the motivation of this measurement is not clear for me. Probably you mention two things -- study of nucleon/nucleus structure, and neutrino oscillation experiments, but no clear statement is given what kind of new information we can (want to) obtain with neutrino-nucleus scattering for what purpose. - In the second paragraph, "K2K" is used without any definition before. II. Cross section - Don't we need some introductory sentence here? Or, some overview of the paper structure at the end of introduction section? A. QE - For Eq.1, no explanation is given the correspondence b/w nu/anti-nu and -/+ sign. Actually, since we present result with only neutrino, I don't think we need to include the expression for anti-neutrino. (The text actually says " ... for neutrino quasi-elastic scatteing".) - F_A is used (in the second sentence below Eq.3) before its definition. - The statement "Previous studies show that this approximation is reasonable [4-6], given the statistical and systematic limitations of neutrino scattering experiments." sounds for me negatively stating the poor measurement capability of neutrino experiments. Why can't we say just the dipole approximation is found reasonable with past experiments (within measurement uncertainty, of course)? II B. - The words "Monte Carlo" means the name of city in Monaco. I was told when I was writing my thesis that we always need to say "Monte Carlo (MC) simulation", "MC prediction", etc. I think "NEUT Monte Carlo [7]" -> "NEUT Monte Carlo (MC) simulation library [7]" is better. - For DIS, if I understand correctly, we take nuclear structure function from GRV94 and use original code and PYTHIA/JetSet to generate hadrons. The current description may be confusing. - "This analysis takes the coherent pion cross section to be zero following [11]" => "... the charged-current coherent pion cross section to be zero ..." Because we do not say anything about the neutral current. II C. - The last sentence, "The above nuclear effects are discussed quantitatively in the results section and in Fig. 7." looks odd for me -- discuss in a figure? I would suggest removing "and in Fig. 7". III. - Don't we need subsection title for the experimental setup? - The first sentence looks too long. Can you separate it into two sentences? - Second paragraph: "A magnetic horn focuses positively charged pions ..." There was a comment for the coherent pion paper, that the horns focus not only pions but other charged particles (kaons, ..), thus this expression is not accurately correct. BTW, we use two magnetic horns, or a magnetic horn system, not a magnetic horn. - Third paragraph: we call scintillator hodoscopes surrounding SciFi "veto" counters, but it is not necessarily used as veto. Actually, we REQUIRE a hit in "veto" counter to be accepted as candidate event! I think we need better name here, although we are so familiar with the name "veto". Also, description of "veto" counter is scattering in this paragraph and sixth par. - Fourth paragraph: "There are a total of twenty 240 cm wide tracking layers," You mean 240cm x 240cm? - Fifth par: How many CCDs do you use in total? - Seventh par: "we require that it start in the SciFi fiducial volume ..." The fiducial volume is not defined yet. III A. - Third para: You quote 60% QE purity for both 2-track QE and 1-track samples, and 15% for 2-track non-QE. These are all different from what we quote in the last oscillation paper. What makes this difference? - In table I, "Q^2" should read "Q^2_rec". III B. - Actually, I don't like the word "Basic distributions" used in a paper, since I think it is kind of jargon inside K2K. (what does "basic" mean?) The section title may not be appropriate, because we also mention Q^2_rec and the discrepancy here. Since "Q^2_rec" is actually defined in the next section, we may need to reorder some part here (and the next section). - First sentence: "... along with the Monte Carlo prediction ..." What is M_A (for QE) used here in MC simulation? IV. A - p.9 "The kinematics of the muon, the longest track in our events, are sufficient to .." We identify and assume the longest track as muon, but nevertheless we must not mix up "muons" and "muon candidates". => "The kinematics of the muon are sufficient to .." - "E_{\nu rec}" looks odd, "E_\nu^{rec}" or "E^\nu_{rec}" ? - I believe what we estimate here is E_nu and Q^2, and estimated values are denoted with "_rec". Therefore, in the first sentence we should use E_nu and Q^2. - p. 10, Second para, "The resolution for theta_mu is about 1 degree, but there is a tail to this distribution ..." Can we be more quantitatively here? I cannot estimate the effect of this tail to the resolution with current description. How about adding figures for E_mu, theta_mu, and E_nu resolutions? IV. B. - First para. of this section: Q^2 -> Q^2_rec (two occurrences) - Just below Eq.7: "the reweighting function R accounts for the effects of the nucleus." I believe more description is needed. How it is estimated? How much effect on the measurement? - p.12 l.2 "... and giving a fit for the QE axial vector form factor." => "... and give ..." ? - p.12 l.8 "The quasi-elastic cross section must still be modified .." I don't think the the detector acceptance and resolution modify cross section. - In this section, sometimes "q^2" and "q^2_rec" are used. It is not a mistake (at least for q^2), but may be better to use consistent expression. V. - "We fit a large ensemble of E_nu and Q^2 distributions:" -> "We fit a large ensemble of E_nu^rec and Q^2_rec distributions:" - end of p.12, "computed separately and use a MC sample ..." -> "computed separately using a MC sample ..." ? - p.13, l.2 "... as is the non-QE/QE ratio and proton rescattering." We never mention proton rescattering effect is included in the fit. We need to describe what sysmteatic parameters are included in this analysis and how we do that. - p.13, l.3 "The result of the combined fit is MA = 1.16 +/-0.12 GeV." I thought the error of fitting itself was much smaller. Is this error includes the systematic error? Also, what about numbers given in this and the next page as fit results? - p.13, l.4, "The Q^2 distributions for the data and the best fit M_A are ..." -> "The Q^2_rec distributions for the data and MC simulation with the best fit M_A are ..." V. A. - p.14, just below Fig.5: "In Fig. 6, when no cut is applied ..." No explanation of Fig.6 is given! We should write like "In Fig.6, .... is shown. When no cut is applied ..." Also, what is the purple vertical line around Q^2_rec minimum = 0.2 in Fig.6? V. B. - In general, I feel the current draft lacks description of how each systematic uncertainty on M_A is estimated. - In the caption of Table II, "Errors smaller than 0.03 are not included in the total." sounds odd. It is negligible compared to other uncertainties anyway, what is the purpose of this sentence? I would suggest to remove it. Energy spectrum: - I was confused by inconsistent statements in the manuscript, in some part it reads our measurement is not affected by neutrino energy spectrum because we are fitting the Q^2 shape only, while the relative flux and normalization is a dominant uncertainty. - As I mentioned above, I cannot figure out how the uncertainty was estimated. - I also cannot understand the description in the first paragraph of page 16. Momentum scale: - This part is better written than other part. However, I am not yet convinced we can really constrain this parameter from the neutrino data. Many systematic effects that may affect the momentum scale may not be considered yet. For example, spectrum shape in higher energy region should have correlation with MRD energy scale. In the spectrum analysis, this correlation gives relatively small uncertainty to neutrino energy spectrum, but for M_A, small change of the momentum scale causes large uncertainty, and I am not sure if we really understand it with currently quoted precision (1%). (In addition, there may be effects from neutrino interaction uncertainty other than QE.) - Other comments: what is "energy binning effect" referred in the first para.? "spectrum fit analysis" in the third line from bottom, p.16, is not clear for non-experts. nonQE/QE ratio: - when you calculated nonQE/QE(=1.38), how did you take into account the reweighting factors, e.g. elimination of CC coherent pion? "our default neutrino interaction MC" means NEUT after coherent and DIS corrections? non-QE BG: - "We also consider the case where charged-current coherent pion events are produced which increases ..." -> " ... produced according to Rein and Sehgal[12] with modifications following Marteau[13, 14], which increases ..." nuclear effects: - The caption of Fig.7 needs improvement. For example, no description is given what are two lines in the left figure. What is k_f used to make this figure? Why the line in the right figure is broken like this? - You mention Pauli Blocking is "one possible contribution to the discrepancy at low Q2 described earlier." Because our statement is that CC coherent pion was most contributing to the discrepancy, and already in this paper we assume CC zero coherent pion, the above statement should mean possible contribution in addition to dominant CC coherent contribution. - BTW, I have noticed you quote ~20% discrepancy b/w data and MC in low Q^2 region. Is this value in the case CC coherent pion is removed? As we use no CC coherent pion as the default model in this paper, we should quote consistent number. - p.19, 6th line from the bottom, "The resulting uncertainty in MA is +/- 0.01, ..." I could not understand how this number is derived. - The next line, " ... and the comparison of the Fermi gas model ..." "... while the comparison of .... " ?? ***** End! ***** Date: Mon, 8 Aug 2005 12:57:37 +0200 (CEST) From: Sofia Andringa To: scifimapaper@neutrino.kek.jp Cc: gran@u.washington.edu Subject: Re: MA paper draft Abstract: - As Yokoyama-san points out, there is a change between stat. and syst. errors. Also, in the result section MA=1.16+-0.12 GeV/c2 [should be GeV/c2 all the times]. Is 0.12 the total (including statistical error), or should they be always presented separately? I. Introduction: - I also agree that an overview of the paper would be useful and could serve to introduce K2K, etc. Also there could be a paragraph explaining the importance of the QE description for the neutrino experiments, and saying MA can only be measured in these experiments. II. Cross-sections: A. QE - There is a wrong sign in (s-u), should be 4ME+q2-m2, according to the references (if I am not mistaken). - gA=1.267(3) should be given a name, and reference for its big precision (PDG98 from beta-decay). It seems most of the old results, in table III, used gA=1.23. This changes the relative importance of the axial part of the cross-section, much as the vector form-factors do. Although it is a change of 3% and not 10%, the effect is linear and so can be expected to be stronger. I think it should be commented together with the vector form factors. As a note, according to a fast study by Xavi, the 3% change translates directly to the MA result. B. Other - "single pion events from the model of" should be "single pion production modeled according to", etc. C. Nuclear effects - Fig. 7 should not be introduced before the results section, where it is included. III. Experiment: - Is it possible to divide better the 1st part (detector description only) and first steps of analysis/data selection? - The description of the veto system and its use is scattered, first with beam description and then after veto description (i would rather leave it only in second part). - The description for the Lead Glass is absent - although we "require tracks to pass through" it in the analysis, it is not clear what information is used. Also the name for "Lead Glass" should be kept consistent (do not say "a lead glass detector" in III.A). - You give the purity of the MRD matching, but not the efficiency, and before its description. Also, the 3.5 GeV/c maximum on muon momentum should come together with the minimum values of p(mu). A. Data samples - What does it mean "For one track events, the proton or pion is absent or below threshold"?? If it exists and is absent is because it is below threshold, no? Maybe better "the requirement of 3 layers (...), and events with protons or pions below these thresholds will be classified as one-track events". - Table I "The columns that include (...) are used for MA", is strange. Better, "Only events with reconstructed Q2 (eq. 6) are used in the MA analysis, the corresponding numbers are shown in separate columns". B. Pmu and Thmu (-> III.A.?) - I am not sure if the Coherent pion contribution should be shown. The absence of coherent pion was already introduced in II.B and since there is already a paper (hopefully already published when this one is submitted), i think it might be discussed only in the systematics. Thus, I would rather introduce the two figures (saying they are based on the results of the [17 and 11] for the K2K beam energy spectrum and some aspects of the interaction models) as the basic ingredients for the calculation of Enu and Q2 in the next section. We can then comment as in paragraph 2, that the low Th is related to the low Q2, and that region is more affected by the nuclear effects. IV. Analysis A. see above. - "Th is the angle [to the neutrino direction] determined from the hits" B. Fit procedure - "some systematic error parameters" (line 2 of parag. 2). Which? For example the proton rescattering is now not discussed at all until the results section! - For me it would be simpler to describe it all in a different order. Namely, to introduce the several parameters which need to be fitted, and which constraints are there for those, and then how to construct the expectations for each bin. Something like: "We want to extract MA from the Q2 data distributions. For this we bin the data in Q2rec and in Erec, and perform a likelihood... The expectation depends on MA and some systematic parameters, accounting for uncertainties in the neutrino flux, the non-QE to QE ratio and the proton rescattering cross-section. The neutrino flux in K2K... The non-QE/QE ... The proton rescattering... The expectation for each bin is given by the MC expectation for nonQE (reweighted by the nonQE/QE and the F(E)) plus an estimate for QE which depends on MA. The QE expectation is calculated from the cross-section is computed as a function of MA,Etrue,Q2true and transformed by R in order to include nuclear effects; a matrix M, taken from MC, introduces the effects of detector acceptance and resolution, together with those of proton rescattering." For me it is more clear, but is just a suggestion. Or else, the present text can be made more explicit, by identifying better the rec/true quantities, and the R, M, and the systematic parameters: - The expectation for each reconstructed E,Q2 bin is computed as: n_i(E,q2)=A.[].F(E), integrated over all true E,q2 values. The nonQE reconstructed values come directly from MC, while for QE there is a migration matrix M... - Also when discussing the F(E): "Five free parameters F(E) reweight the predicted flux for different energy regions, accounting for its uncertainty". And "the non-QE must be reweighted only by the non-QE/QE ratio and the flux in each energy region". (To separate the true energy regions where the flux aplies and which, in principle, do not be the same as the 5 Enrec bins). The description of F(E) appears in two different places... - The motivation for fitting A and F(E) should be more explicit: say that MA affects strongly the total cross-section (as a function of energy) but we do not want to (or can not) depend on that information. This is also why we need to bin the data in Erec. V. Results - I miss a table (table 2 in old draft) with all the fitted values. In particular we want to check if the results for the flux are within the 20% uncertainties which are referred, and what is the value for proton rescattering... - Also when looking at QE/nonQE, it would be interesting to see if the effect is just the compensation of the global change in QE cross-section (from MA=1.1 to the new value) or there is extra differential information from the fit. Ie, does the QE purity of the samples remains the same after the fit, or does it change? - Can you explain me again which are the free parameters, i do not get the 230 degrees of freedom you mention! - I agree the A should be common to both data samples, and if (as you say below) the error would be reduced with statistics, this might be a start! In this case it would somehow indicate cross-section*flux*fiducial volume (can be useful as an internal K2K information). Is that possible? On the other hand, both A and B can reflect changes in the total QE cross-section, and the results on A will have a more clear meaning after knowing what happens with B. - (Text:) "The parameters for the flux, nonQE, proton rescattering are the same for both data-sets." ("subsamples" was used for the different selections). - Figure 4: The upper part of the plot for 1 track, K2K-IIa is cut. Also there should be a line indicating the fitted region. And a comment about the nuclear effects at low energy. A. Consistency - Maybe the first consistency check should be the Q2 cut variation: as it justifies the use of the dipole approximation! Again, starting with the larger Q2, then commenting on low Q2 and nuclear effects, and then coherent pions (which I would leave to the systematics section). - Figure: What is the vertical line? And there are no errors? Also the first point (Q2>0) is hardly seen. - For the different energies: You refer the large migration effects from bin to bin. In principle this affects mostly the background (for the QE the energy bins are much larger than the resolution): should this be commented? -Figure: Can the vertical scale be unified with that of Q2? - Then the two data sets. There should be a comment here about what could be different between both. Also, chi2 for each value should be given (and maybe partial chi2 for the different samples somewhere?), and for the low Q2 sample, where it should be worse! B. Systematics - It is not clear if the statistical error is to be considered separately from the systematic as in the abstract, or is included in the total "systematic" error. 1. Flux - The use of the flux fit has a different problem, apart from different data sets (this could be replaced by SciBar+1KT flux fit, if necessary), which is that it is done with MA=1.1 (and old form factors?) and thus the correlation with MA might be present: in principle this correlation is small, is it true? 2. Muon momentum scale - This section is much more clear now! What does it mean "energy binning effect" in the end of the first paragraph? Do you mean in some cases the event changes Erec bin, but mostly just the (tighter) Q2rec bin? 3. nonQE ratio - Comments above 4. nonQE shape - I would concentrate all discussion about coherent here. V. C and D - References should be added for the electron scattering experiments and the updated parameterizations. Also for the "discrepancy between the polarization transfer measurement and the Rosenbluth separation method". - The parameterizations used now should be referred before. And only then the old ones for comparison. The fit value shouldn't be referred to as "rising", since the effective parameter changes meaning... - How is the table ordered?? Maybe it would be useful to add the dates so that one can follow the differences in parameters with time. - As I said above, I think the change in FA(q2=0)=gA is most relevant because it measures the relative importance of the axial part of the cross-section. Maybe a test with the old value should also be reported? - Finally, a comment should be made about the compatibility with the old results, once they are presented. ----------------- Chris Walter gave me comments over the phone. One sigifnicant comment is that the fact that nuclear final state interactions, such as proton rescattering, is not zero is something that was not considered in the neutrino interaction models, and does not clearly appear in journals like Phys. Rev. D. We should take some care to record this feature of the interaction models in a publication like this.