Dear Gerco, I give my comments to your comments with a *: Best regards, Tjeerd General: It is not fully understand why the analysis is split in 7 * explained and 8 TeV. In particular concerning the choice of the binning and multivariate optimisation. L2: "An observation of ...". That is better than "Any observation ". * or "Observation of ..." * is preferred over "Any observation of ...". You actually say later that a branching ration of 10^-40 is still within the SM (w/ massive nu). * Make a remark at the relevant line. Here I do not understand why you mention this. L14: Write "\taum" instead of "tau". To be consistent with the rest * or write "tau lepton" as in Ref.[6]. of the text. L13-25: This paragraph contains a lot of "LHCb". By the end the reader surely knows who did the measurement. Suggest to remove the occurrences in L18 and L21. * It is needed in L18 and still OK in L21. Remove comment. L18: Add comma after CL. * "Remove" I suppose. L22: remove hyphen in cross section. Although the EB guidelines may give another suggestion, this word is not hyphenised in any dictionary * hyphened (included those suggested by the EB). L25: Add "at 7 TeV." at the end of this sentence. * after "cross section". L29-30: Include term "blind analysis" : "To avoid potential bias a blind analysis is performed by initially excluding mu-mu+mu- candidates ...." * the suggestion that "blind" means "exclusion of mu-mu+mu- candidates" is not correct. Remove this comment. L31+94: "expected mass resolution". This sounds a bit like we are * may read as if we are still trying to understand our detector. But in reality, we should talk about the mass resolution of our detector (which of course depends on the momenta of the final state particles and the Q-value). Not about mass resolution of a specific decay. Suggest "intrinsic mass resolution" or similar. * I think "expected mass resolution" is OK, but it could be supported by "from simulations" or "from other reconstructed masses". L61: "In THE simulation ..." Which one is that? This is the first mentioning of a simulation. Add a senctence to explain the purpose of * sentence the simulation. * Can we make a suggestion for this sentence? It will be rather general I am afraid and not so useful. * Ask to add "event" in "In the event simulation". L69: there is no such thing as a "b hadron". Either a "b-quark" or a "hadron containing a b-quark". Also in L.71. * Dordell uses "b-flavoured hadron", but I think that is ugly. * SLAC used B hadron, but that seems to be old fashioned. LHCb now uses b hadron, although B meson is still OK. * The hyphen is there in the adjective "b-hadron lifetime". L86: "The decay Ds- -> eta ...". Start a new paragraph before this sentence. * indicate "why". L93: "The signal region ...". Start a new paragraph before this sentence. * indicate "why". L112: "An ensemble selected ... is used." This is too much information in one sentence. Better to explain a little bit more. Ensemble selection is new to me. If you keep using it like this, please add a hyphen "ensemble-selected". * "classifier" nor "MatrixNet" is mentioned in Ref.[18]. A better reference would be the CERN report of V. Doneva (2013). * I read about "a classifier ensemble" not "an ensemble classifier" in [17]. * I agree this is full of jargon and suggest to replace by: "A new multivariate technique MatrixMet is used [Doneva]". L127+134: You say that the PDF is calibrated. But which PDF? This is not mentioned before. Don't you mean that the classifiers are calibrated such that they return a probability or likelihood? In that case the mentioning of likelihood in L136 also makes sense (now it does not). * Just ask what "probability distribution function" is used. L139: ".... the signal BF assumption." Explain, lingo. * Just ask what "signal BF assumption" means. * What you write is even worse than the line you comment on. L136 & Fig. 1: Why the uneven bin widths, rather than equal-width ones? Optimization of the separation power is claimed, but do these binwidth have any significant effect? * bin width * This is probably explained in the analysis note. * I think this text is OK, as it describes what is done. L152: why meson. Shouldn't it be heavy hadron instead, to not exclude baryons? * Ask how large the contribution of b baryons is. L154: "There IS a large number of ...". Not are. Number is singular! * Just suggest to replace "are" by "is". We do not need to lecture them. L160: "... has a mass distribution CONSISTENT with ...": as the process is different, it cannot be "consistent". Perhaps "... has a SIMILAR mass distribution ..." * Just suggest to replace "consistent with" by "similar to". L165-169: "... for each bin ... using an extended, unbinned ...." -> unfortunate choice of words (too many different "bins"), consider rephrasing. * I do not see anything wrong here. L166: "for each bin in ..., ... and MASS" a fit is made to the spectra outside the signal windows. This is contradictory. Isn't the (shape of) the mass-spectrum outside the signal region fitted using an unbinned likelihood for each 2D bin in MPID and M3body? * I agree that "background" and "signal" relates to the "reconstructed mass distribution". It should not be mixed up in this way with classifiers M3body and MPID. I do not understand your comment in detail. L169: "the SMALL differences" -> "the differences". BTW, differences in WHAT? So even better "the differences in the estimated number of background events in the signal region are taken as systematic uncertainties." Perhaps even including a statement somewhere that this difference is used to estimate the a priori unknown shape of the background. * Also give a suggestion for the sentence which they can take, modify of * ignore. L183 "BR(Ds->Phi(K+K-)pi) is taken from the BaBar amplitude analysis [25], which considers only the Phi->K+K- resonant part of the Ds decay." * You do not suggest to remove "-", as that is covered with footnote 1. Is the second half of the sentence indeed needed? We would skip the second half of the sentence, as the first half of the sentence already implies that this refers to the resonant Phi->K+K- part. * I agree to suggest to remove this second half. L184-186: This sounds like KKpi coming via phi is a bad thing, while explicitly normalising to phi(mumu)pi. So the phi is what is needed. What is precisely meant? Please explain. * I do not understand your comment. Remove comment if not necessary. L184 "This is motivated by the negligible contribution of non-resonant Ds->mumupi events seen in the data." * I thing they mean "This procedure". Q1: What does "This" refer to? If you refer to the calculation of * Why "Q1:"? BR(Ds->Phi(mumu)pi), than we propose a rephrase: "The estimate of BR(Ds->mumupi) using BR(Ds->Phi(K+K-)pi) is more precise than the direct measurement of BR(Ds->mumupi)." L185: Could you explain how the non-resonant Ds->mumupi events would affect BR(Ds->Phi(mumu)pi) ? * I do not understand this comment. L206: is the effect of eliminating the lowest likelihood bins taken into account (a la L200-201)? L207: "The YIELD of .. candidates in THE data, Ncal, ... IS determined ..." (not plural, add the). * I suggest "The yields of ... candidates in the different data samples". L208: "The VARIATION in the YIELD if ... IS considered ..." (singular) L209: variation of the widths of the Gaussian components: by how much are these widths varied and what is this variation motivated by? * As the systematic uncertainty is not given, the change in widths is not relevant to ask. L213-216: The text does not make clear whether the data of the individual bins in M3Body and MPID are used individually or combined (excluding the lowest likelihood bins). If so, some measure to substantiate the claim that "no significant evidence for an excess is observed" could/should be given. E.g. for 7TeV the bin * 7 TeV [0.40-0.45,0.46-0.54] expects 2.89+-0.63 and sees 6 and [0.45-0.54,0.54-0.65] expects 2.83+-0.63 and sees 8! Of the 25 measurements, several seems rather high, which in the combination are offset by a fair number of low or very low number of observed counts. For 8TeV there is indeed little to worry. Please comment on the * 8 TeV consistency of "no excess" per bin. L217: "estimates". Not estimations. L230: "In summary, the previous LHCb limits ..." * No need for "previous" here. L231: Remove "3.0 fb-1 of". Jargon. * Replace "the full 3.0 fb-1 of data" by "all data" and add "corresponding to 3.0 fb-1 of collisions" at the end of the sentence. L232: "No evidence for any signal is found" -> "No evidence for the LFV decay tau->3mu is found." * lepton flavour violating decay tau-->mu-mu+mu- L232: "The present limits supersede" -> There is only one, unless the 90% and 95% CL. limits are counted as two ... Replace by "The presented limits" * measured limits. L233: "B-factories". B in italic. * and no hyphen. L233: allow -> give L233: "... should allow improved ..." -> "... improve the constraints placed on ...." L234: Remove "to be placed" * Suggest a new sentence with the remarks above included. Fig. 6/supplementary material: What is the color coding?