Hi All, you will find attached a first draft of our reply to the referees. Please have a look and let us know how they can be improved. Thanks! Concezio and Roberto --- Re: LH11701 Measurements of partial branching fractions for $\bar{B} \rightarrow X_u l \bar{\nu}$ and determination of $|V_{ub}|$ by B. Aubert, M. Bona, D. Boutigny, Y. Karyotakis, J.P. Lees, et al. Dr. R. Sacco Stanford Linear Accelerator Center Mail Stop 35 2575 Sand Hill Road Menlo Park, CA 94025 Dear Dr. Sacco, The above manuscript has been reviewed by our referees. We ask you to consider the enclosed comments from the reports. While we cannot make a definite commitment, the probable course of action if you choose to resubmit is indicated below. ( ) Acceptance, if the editors can judge that all or most of the criticism has been met. (X) Return to the previous referee(s) for review if available. ( ) Submittal to new referee(s) for review. With any resubmittal, please include a summary of changes made and a brief response to all recommendations and criticisms. Yours sincerely, Robert Garisto Associate Editor Physical Review Letters Email: [log in to unmask] Fax: 631-591-4141 http://prl.aps.org/ ---------------------------------------------------------------------- Report of Referee A -- LH11701/Aubert ---------------------------------------------------------------------- The precise determination of |V_{ub}| is crucial to testing the CKM sector of the standard model. The length of the side of the unitary triangle opposite the angle beta is proportional to the ratio |V_ub| / |V_cb|, making its determination a high priority. Given the rapid theoretical and experimental progress in this area, it is very interesting to see how the determination of |V_{ub}| develops. In this context, the paper uses 4.3 times the data previously used by BaBar for the |V_{ub}| determination and significantly improves on the precision of the earlier measurement. For the substantial advance in its subfield and implications, I regard this work suitable for publication in PRL. The letter is well written and clear. Please consider the following, minor corrections: R: we have taken into account all the following remarks. Title: V_{ub} --> |V_{ub}| Abstract, L7: (and everywhere else) replace "sys" with "syst" p7, L4: "....measure branching fractions for such decays." please add reference 20 (PRL92_071802) at the end of this sentence, as it places this paper in the context of the previous measurement. p7, L9: "..we present measurements of partial branching fractions.." --> "..we present a measurement of partial branching fractions..." p7, L22: change "fb^1" into roman: {\rm fb} p8, FIG1 caption: "(points with statistical error)". One can't see on the PRL size whether they have a statistical error, perhaps use "(full circles)" to refer to them ? p8, L5: "..estimated on Monte Carlo (MC) as.." --> "..estimated using Monte Carlo (MC) simulation and is defined as... p8, L11: I would perhaps use a capital letter for "threshold function [15]" --> "Threshold function [15]" p8, L15: "p*_l >1 GeV/c" what does the * stand for? Do you label a particle's momentum with p* instead of simple p? p9, L2: use a long dash, with "--", between 0.473 and 0.523 p10, L6: "..and $N^{out}_u$ refers.." --> "..and N^{\rm out}_u$ refers..." that is, don't italicize "out" p10, L19: "We estimate the error due to the signal..." --> "We estimate the uncertainty due to the signal..." p10, L19: "The signal modeling uncertainties..." -> "The uncertainty on the signal modeling are due..." p10, L21: "We also calculate the errors associated with the uncertainties in the non-perturbative..." -> "We also calculate the uncertainties due to the non-perturbative..." Here and in several other places later on (e.g. in caption of Tab I) , you should change "error" with "uncertainty". By systematic "error" one means a constant shift to the central value of the measurement. By systematic "uncertainty" it is intended a distribution of the error with a certain width, which is the quoted number. So the +/- numbers in Tab I should be referred to as "uncertainties" and not "errors". p11, Tab I, caption: "Summary of the measurements of the fitted numbers of events..." --> "Summary of the fitted number of events..." p11, Tab I: Place a label for the M_X, P_+ and M_X,q^2 column. For example "Method", or "Cuts" p11, Tab I: Last column, add a "\times" before "10^{-3}" p11, Tab 2, caption: Add ":" after "..kinematic cuts, from" in order to brake the long sentence. p11, Tab2: the formatting of the first row is not easily readable. Remove the vertical space between "Shape" and "function" and on all the other two-word column headings. p11, Tab2: add "\pm" in front of numbers in the table p12, L2: "...reducing the relative error by..." --> "..reducing the relative uncertainty by..." p12, L3: Add "between the two measurements" after "...highly correlated". ---------------------------------------------------------------------- Report of Referee B -- LH11701/Aubert ---------------------------------------------------------------------- This Letter provides inclusive partial branching fractions for charmless semileptonic B meson decay in kinematic regions in which the dominant semileptonic decay to charm is suppressed. From these regions, seven different values of Vub are extracted using several different theoretical calculations of the corresponding partial widths. This paper summarizes the work for three challenging measurements. The measurements themselves are key to improving our knowledge of Vub, as they represent the kinematic regions that hold best theoretical promise for control of the nonperturbative effects. It's crucial to have a unified analysis such as this so that results from these different region can be compared with clear control of the experimental systematic uncertainties so that we can learn how well we can understand the theoretical uncertainties associated with Vub. Improved extraction of Vub is crucial to our understanding of the flavor sector, and particularly to CP violation within that sector. Therefore the paper is of interest both to the high energy physics community -- experimental and theoretical -- and to the broader physics community. I should note that the analysis itself is quite complex, and that much work has clearly gone into the measurements. However, with the terseness forced by the PRL length, it's difficult to judge the analysis and the completeness. The basic techniques are certainly sound -- but the devil for an analysis like this is in control of the background and exploration of how well that control is understood. I strongly encourage the authors to submit a companion PRD in which the details of the analysis are fully described. R: indeed, we plan to provide a full description of the techniques we employ in a PRD. The plan is to include further studies on a timescale of a few months. Overall the writing is clear. I recommend publication after the items below have been considered or addressed. 1) Since this is intended for PRL, the introduction and conclusion must be reworked to be more accessible to a general audience. For the introduction, this should be straightforward: the context alluded to in the first sentence could be expanded. The second sentence should be clarified (B->pipi is also proportional to |Vub|^2 and doesn't have a pesky neutrino -- sharpen the argument). The jargon (eg, X_u, X_c) will be familiar to the specialist but are undefined and, in a related fashion, the basic mass difference explanation for why X_clnu is suppressed is not given. Etc... R: we rewrote the introduction accordingly. More serious, though, is the conclusion. Table 1 presents 7 different values of Vub, for which the uncertainties are highly correlated. To me that implies that for any given comparison, the largest th'y error dominates the level of uncertainty. Comparing results for Mx vs P+ vs Mx,q2, there are differences that are over two "standard deviations" apart (or however one should interpret those theory errors). The situation is not even acknowledged in the paper. At the very least some statement *must* be made. Preferably, some more interesting questions should be addressed, which would truly make the analysis high quality for PRL: What value of Vub should the reader take away (for comparison to which value in the PDG, for example)? What are potential pitfalls in the different regions that may bias on region or another (in theory or exp't)? Could the pattern of differences tell us anything? In short, what can we learn from these 7 numbers either directly or relative to other measurements? If there's nothing that we can do immediately to learn something from these numbers, then shouldn't the theory errors at the very least be inflated? Also in the conclusion, but more for spirit of clarification, are the Vub's being compared to the inclusive/exclusive average in the PDG Vub+Vcb review? Are these results appreciably correlated with that average given the correlation with the previous Mx analysis? Perhaps more direct comparison with other uncorrelated experimental determinations with similar kinematic regions, and of "the" Vub from this paper with exclusive measurements, would be more beneficial to the general reader. The shortcomings of the conclusion for PRL are the sole reason for not, at this point, recommending "The paper should be published in PRL after minor revisions are made, without further review." R: we have reworked the conclusions in order to evaluate the compatibility of our different determinations of Vub. We have done that, in particular, in the BLNP framework after determining the statistical and systematic correlations between the experimental determination of the partial branching fractions. We see an agreement between the Mx and Mx-q^2 analyses, while the Pplus measurement differs at 2.5 sigma level (this disagreement is also seen by Belle). We also state that BGE and BLNP give consistent results and that the values we measure are in good agreement with other inclusive $V_{ub}$ determinations, and compatible, although systematically larger, with measurements from charmless exclusive semileptonic decays. 2) The definition of P+ refers to a jet direction -- can this direction be clarified? Is it the jet axis calculated from the hadronic X system? R: It is indeed calculated from the hadronic system, as its definition suggests. We decided to drop the reference to the jet direction though, and just give the definition of P+, along with a reference to theoretical papers. 3) For the Breco purity selection, does "Breco decay" simply refer rejection of final states with a particular particle content (ie, a particular decay mode) such as "5 charged pi's + 2 pi0's", or is it more finely structured: "5 charged pi's + 2 pi0's in particular kinematic regions"? As written, it sounds like the latter imposed on an event-by-event basis. Have I misunderstood, and it's really just a final state selection? R: it is not a rejection, rather a selection of decays of the other B in the event into a specific state. The number of final states that we fully reconstruct is rather large (of the order of 1000); for some combinations of particles we may have specific invariant mass requirements. However, there is not enough space inn this paper to embark in a detailed description of our selection; we plan to include a thorough description the upcoming PRD. 4) Does "photon energy loss" in the m_ES fit description refer to initial state radiation, final state radiation or both? The phrase "caused by photon energy loss" refers only to the m_ES tail, correct? For either ISR or FSR, does the systematic parameterization ansatz include uncertainty in the level of radiation (or is it not important on the few % scale)? R: "photon energy loss" refers to energy loss in the detector material. 5) The paper mentions Xclnu decays entering because of undetected KL's. Do D semileptonic decays also pose a problem since these will also give an additional undetected particle? The charged leptons tend to be rather soft, so depending on the lepton id criteria might not get vetoed... If these are important, have the D s.l. branching fractions and the B->D momentum spectrum been considered in the systematics? R: The requirements on the minimum lepton momentum (1 GeV) and the lepton charge reduce the background from D s.l. decays to a level of a few percent. We take into account the residual contribution by varying the D s.l. branching ratios in our assessment of systematic uncertainties. 6) Some idea of the effectiveness of the D*lnu veto based on m_nu^2 would be useful to include -- both rejection factor and signal efficiency. The shape in signal that one would expect is not so decay. For clarity in the paper, I suggest changing the variable name to avoid confusion with the p_nu introduced in the previous paragraph, whose mass this variable is not... R: the requirement m_veto<-3 reduces the D*lnu background by about 36% while keeping more than 90% of signal events. 7) Minor typo: In the paragraph and sentence beginning "To extract the distribution in the variables...", I think "subsequently separating" should be "subsequently separate" R: We changed the sentence. 8) Somewhat more detail on the spectral fits is warranted, since these are a key component of the method. For example, presumably N_u and N_u^out are normalizations for two different fit components corresponding to decays within and without the final kinematic region at the generator level. How are these treated in the fit? Are they tied together by the particular theoretical model used to evaluate the efficiency? Or, do the float independently? Nice to know, since that couples to how one needs to consider the models in the systematics analysis. R: N_u is, in fact, the number of data events in the signal region after all cuts. N_u^out is an estimate of the contribution to N_u of signal events reconstructed in the signal kinematic region but coming from outside the said region. In the fit, the MC shapes of signal and N_U^out components are forced to be the same, according to the theoretical model. 9) I couldn't tell how Nsl and BGsl are determined. Are these obtained from fits to the Breco mES distribution where an additional lepton has been required? If I haven't simply missed something, the method should be stated in the paper. R: the number of background events is determined from MC simulation. The corresponding systematic uncertainty due to our knowledge of the background composition has been taken into account. 10) Minor typo: Nsl= ... and BGsl=... are the measured *numbers of* semileptonic events and... "numbers of" is missing. R: We changed the sentence. 11) In systematics, at what level does modeling of hadronic showers (beyond KL's) in the calculation of the X system matter? Eg., fluctuations or splashback could result in showers that are relatively isolated from tracks in the calorimeter, so will bias the X calculation and therefore any efficiency or spectral shape for Mx and P+. R: modeling of hadronic showers, including fluctuations and splashback is taken into account in the GEANT 4 MC simulation. The associated systematic uncertainties are included in what we call "detector systematics". The combined effect due to the reconstruction of neutral clusters in the calorimeter actually dominates the total detector systematic uncertainty, ranging from 1.4% for Mx, to 2.9% for Pplus, to 2.5% for Mx-q2. 12) In the systematics table, does {\cal B}(D) refer to D branching fractions? Can the paper clarify how these branching fractions, or what subset, is varied? R: we varied the branching fractions within their experimental errors around the central values. We have rewritten the description in the paper. 13) There's no mention of final state radiation, which I would assume to be sizable for electrons. What effect does this radiation have on efficiency and background smearing, and how well is it known? What effect does radiation have on generator level quantities (spectral shape biases), and at the 1-2% level of systematics considered in the paper, do the kinematic regions with radiation map well enough onto the regions the theorists have calculated rates for without radiation? This question can be both absolute in nature or refer simply to how the MC is treated -- for example, one could get an inappropriate generator level q2 boundary if by looking at the generator level (p_e + p_nu)^2 after radiation. R: in our analysis, FSR is simulated using PHOTOS. We checked that the effect of FSR is negligible by studying kinematic variables in MC samples produced with and without PHOTOS. 14) The paper notes specifically notes that the mix of charged and neutral B pairs is not 50-50. Is there any systematic effect associated with this in terms of the mass spectrum of the hadronic system being, in reality, somewhat different, etc., particularly when average B lifetimes are used in the end? From figure 2 and the roughly 60/40 B+B-/B0B0bar split, it looks like there should be enough statistics to get Vub independently from these two samples (and the tagging, of course, allows the separation into those subsamples). Are the Vub results from the two subsamples consistent, and do the patterns in the different kinematic regions manifest themselves in the same way? R: we checked that PBR obtained separating the samples by the charge of the reconstructed B meson are consistent with each other within the uncertainties and indeed the patterns in the different kinematic regions manifest themselves in the same way. A measurement of PBR separately for charged and neutral Bs is important to determine weak annihilation effects that are currently taken into account in the theoretical models. A more detailed study of experimental cross-feed between charged and neutral Bs is required and will be included in the upcoming PRD.