Dear proponents, Congratulations on this analysis and its useful results. Please find some comments from the Nikhef group below. Best regards, Jeroen on behalf of the Nikhef group -------- general: -------- * There are some differences in conventions in the parameterization of the angular distributions between this analysis and the closely related B0->mumuK*0 and B_s0->J/psiKK analyses (transversity instead of helicity angles, |A_para|^2 instead of |A_0|^2, delta_S-delta_0 instead of delta_S-delta_perp). Probably it is too much work to change these conventions now, but we feel it would be good to make this analysis consistent with others in the future. * It is assumed that B0->J/psiK*0 is a flavour specific decay (Equation 1) and that the decay time and angular acceptances factorize (lines 130-142). This makes the decay time and angular distributions factorize and therefore we don't see a reason to include the decay time in the model that is used to extract the angular observables. If this is correct, we would like to propose to remove the discussion of the decay time model and the decay time plot from the paper. It is not needed and was a source of confusion for us. It might be possible to gain in sensitivity by including the decay time in the model that is used to calculate sWeights. Was this considered? * The angular definitions in Figure 1 are inconsistent with the combinations of amplitudes and angular functions in Table 1 and the results for delta_para-delta_0 and delta_perp-delta_0 close to pi. Are you using the "standard" ("angle calculator") LHCb definitions for decay angles? In that case the y-axis of the transversity system points in the direction of the kaon in the K-pi plane, not in the direction of the pion (see for example LHCb-ANA-2012-067, K+ for B, K- for Bbar). * How was the binned representation of the acceptance function in Figure 4 obtained? The caption of the figure says that these are projections of a histogram, but this seems odd, since the acceptance is not a PDF (please change the caption to match the description in lines 140-141). Is this binned representation used in the analysis? * L. 100: The analysis note claims there are 1.3 B candidates per event. We now know that is incorrect but this should be fixed. How many multiple candidates do you have? Is the 3.5% background from candidates with mis-reconstructed tracks related to multiple candidates, i.e. what is the nature of this background? * Related to multiple candidates: How do you deal with K-pi swaps? Including this type of candidate in the fit of the decay angle distributions would give a dilution in the k=4,6,8 terms in Table 1. This may have an effect on the values you measure for (the phases of) the transversity amplitudes. See LHCb-PAPER-2013-019 and LHCb-ANA-2012-051 for how K-pi swaps are handled in the B0->mumuK*0 analysis. * Effects from backgrounds with mis-identified particles, like B_s0->J/psiKK, were found to be negligible for this analysis. Could you comment on how these backgrounds were treated? Distributions of the J/psiKpi mass and the decay angles are potentially correlated for these backgrounds because of the mis-identification. * How was determined that the effect from B_s0->J/psiK*0bar background is negligible? If it was included in the J/psiKpi mass fit, why is this fit not used for the nominal result? Maybe it is possible to remove this background by reducing the upper mass side band? Perhaps you can include a plot of the J/psiKpi mass distribution on a logarithmic scale to show how large the effect is in the mass. * From the J/psiKpi mass parameters in Table 2 we conclude that it was assumed that the background mass model is the same for all Kpi mass bins. Is this a correct assumption? Are the signal fractions also assumed to be the same? Are sWeights calculated for the four Kpi mass bins separately? * The J/psiKpi mass model systematics are evaluated by replacing the double Gaussian model by a single Gaussian (Crystal Ball) model for the signal. We assume the motivation for the double Gaussian model is that a single Gaussian doesn't fit the mass distribution. Isn't the uncertainty in the J/psiKpi mass model overestimated by comparing to a model that clearly does not describe the data? * You include a systematic to account for the uncertainty in the combinatorial background description. We don't understand where this uncertainty comes from, since you do a fit on background-subtracted data with a model that does not describe background. The background subtraction itself could be a source of systematic uncertainty. One of the assumptions is that the J/psiKpi mass and decay angle distributions factorize, which is not necessarily true. Also, the statistical uncertainties in the J/psiKpi mass model are not automatically propagated. Were these sources considered? See LHCb-ANA-2012-067 for the background systematics that were considered in the B_s0->J/psiKK analysis. * What was the motivation to vary the C_SP factors up to one to evaluate the systematic uncertainty associated to the Kpi mass model? What happens when you vary them down? Maybe it would be better to determine the variation you would expect in these factors by comparing different mass models for JpsiK* and the S-wave? * Do you understand the behaviour of the "angular acceptance" and "further resonances" systematics for the S-wave parameters in different Kpi mass bins? Are the very different values a result of the variation in the S-wave phases from bin to bin? Maybe this is related to the behaviour of the S-wave event yield? * L. 243-244: "The reweighting to include the S-wave in the MC has not been performed." What distributions were used in the simulation? Is this also how BaBar, Belle and CDF evaluated their efficiency function? Otherwise it is probably better to use our best estimate of the efficiency by including the S-wave in the simulation before reweighting to match the real data. ------------- presentation: ------------- * Why are you only showing results for the magnitudes of the amplitudes in the abstract and not for their phases? * There is an admixture of K* and K*(892) which you use interchangeably, until you encounter K*(1410) and others. We suggest you write once that K* always stands for the 892 except where specified. And then remove all "(892)". * L. 14-17 + conclusion: It is not entirely clear to us what the implications of these results are for factorization and final state interactions. Does the presense of final state interactions already imply that naive factorization does not hold? What do the measured values of |A_para|^2 and |A_perp|^2 imply? * L. 22: "Measuring direct" -> "Direct" (not much to do with whether it's measured or not). But do you need this sentence? That you need 2 diagrams with different phases is a trivial statement found in any good textbook. That one of those has to be NP is not correct as you could generate CPV in the interference of tree and penguin. Is this what Ref [6] discusses? * L. 30-32: We think this statement is a bit too strong. Surely B0->J/psiK*0 and B_s0->J/psiphi are related, but we don't think this holds for _any_ physics beyond the Standard Model. We're also not sure if any CP violation in the decay would affect time-dependent CP violation and phi_s (reference [10] does not predict this). * Fig. 3: The difference between the red and the blue curve is not visible when printed in greyscale. * Tab. 2: Is it necessary to include the J/psiKpi mass fit result in the paper? The B mass is nine standard deviations from the PDG value. Of course we know why it is, but it might be confusing for the reader and give a wrong signal. * L. 143-152: This paragraph on S-wave is oddly placed. It may not be clear to the reader which "ambiguity" you refer to. Why not put it after line 53? * Fig. 4: Labels and caption should say this is simulation. The y-axis labels are not entirely correct. The acceptance function is not a PDF, so epsilon(cospsi), epsilon(costheta) and epsilon(phi) have no meaning. In LHCb-PAPER-2013-002 the labels say "Scaled acceptance integral". Are the "projections" really normalized to unity? * Fig. 5: Is this a background-subtracted distribution? If it is, can you mention it is? If it's not, can you make it background subtracted? * Fig. 6: Can you extend the curves all the way to the edges of the plots? For some of them it is hard to extrapolate by eye. Can you use more bins for the angle plots? Maybe the same number as for the acceptance plots in Figure 4 (although a factor 3 more would perhaps be even better)? * Tab. 7: Are these 30896 signal candidates or the result of a fit, which then should have an error? Maybe also mention the signal yield in the text? The labels could be made clearer. Why not put B->J/psiK* instead of |A_i|^2 in the column label? * Tab. 8: Is this relevant for this paper? That could be additional material. -------- textual: -------- * abstract: Remove "P-wave". You only discuss B0->J/psiK*0 here and did not mention any other resonant components yet. * L. 2-3: Say that you're looking at K*0->K+pi- and K*0bar->K-pi+ * L. 3-6: The _decay_ can be decomposed in terms of three transversity _states_, which have amplitudes A_0, A_para and A_perp * L. 8-9: The S-wave is not only non-resonant. * L. 10-11, 34, 53, Tab. 1, ... : Please consistently use an order for the transversity amplitudes (suggest 0, para, perp, S). * L. 15: factorization <--> L. 133: factorises * L. 15: A_0 = 0.5: Shouldn't this be F_L = |A_0|^2 / (|A_0|^2 + |A_para|^2 + |A_perp|^2) = 0.5? * L. 16: Is the unit "rad" needed here? * L. 18 and 41: Different spellings of flavour. It should be flavor for PRD. * L. 18, 39, 42, 50, 76, 77, 92, 94, Eq. 1, Fig. 3, ... : You often use "K*0" to refer to all the resonant components of the Kpi system. Please check where it should be "K*0" and where "Kpi". * L. 32: space after phi_s * L. 35: You can't really refer to cos(psi) and cos(theta) as "angles". Maybe "angular observables" instead of "transversity angles"? * L. 35, Fig. 2, Tab. 1, Fig. 4, Fig. 6: Please consistently use an order for the angles (suggest psi, theta, phi). * L. 41-42: "at the production vertex" --> "at production time" * Eq. 1 and Eq. 3: Differential operator "d" in math and text mode (suggest text mode) * L. 70: "For the simulation" --> "For simulation" (simulation has not been discussed before) * L. 101: remove the "-" in decay-time (and throughout the rest of the paper) * L. 110: Theses physics -> These physics * L. 111: determined by -> determined from * L. 121: Already say at this point to what the fit is performed. * Eq. 2: ln -> \ln * L. 148: measuring the sign -> measure the sign * L. 160: "measurements" --> "measurement" (one theta_SP, one delta_S per bin) * L. 177: "K*" --> "K*0bar" * L. 190: add comma after "To account for this" * L. 201, 203, 204, 209: "m_Kpi" --> "m(Kpi)" * L. 204: anglular -> angular * L. 220: space between "pi," and "indicating" * L. 236: Different "bar" on B0 than in the rest of the paper? * Caption Fig. 6: "interfernece" --> "interference" * Citations: Use the template. That will get you Sjöstrand spelled correctly. * Ref [25] hep-ex/0408104: You miss the psi. And this is a CONF which has been replaced by http://arxiv.org/abs/hep-ex/0504030 published in PRL. Luckily the numbers are the same.