{"id":"9f1a4db5-82c1-49e4-a6af-658681a0cf0a","arxiv_id":"2412.13084","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"New subleading-power corrections to B to K* form factors from SCET sum rules yield SM branching fractions BR(Bbar0 to Kbar*0 nu nubar)=8.09(96)e-6 and F_L=0.44(4).","lead":"This paper computes improved Standard Model predictions for how B mesons transition into K* mesons, adding previously neglected subleading corrections. The updated form factors are used to predict the rare decay B to K* plus neutrino-antineutrino pairs, a channel Belle II will measure precisely.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The combined BCL fit has chi2/dof=40.1/23, so the claimed consistency between LCSR and lattice inputs is not demonstrated; the quoted branching fractions inherit this unresolved tension.","rationale":"The strongest claim has two parts: a technical calculation of NLP corrections, and a numerical claim of improved precision with specific branching fractions. The technical part (Sec. 3) appears internally consistent and cross-checks against B -> D* in Ref. [38]; I do not find a showstopper there. The numerical claim, however, rests on the BCL fit in Sec. 4.2. The paper reports chi2_min/dof = 40.1/23, a value indicating significant tension, yet interprets the fit as confirming consistency between LCSR and lattice inputs. The fit is the bridge from the low-q2 sum rules to the full kinematic range used in the branching fractions. A poor chi2 means either the uncertainties are underestimated or the two datasets are incompatible; in either case the fitted z-coefficients and the quoted BRs are not reliable at the claimed precision. Table 2 further shows that the new LCSR form factors at q2=0 carry 50-70% relative uncertainties, which is hard to square with the word 'precision'. The reader's selected weakest point, the B-meson LCDA model, is real and is explicitly flagged in the paper (Sec. 4.1 and Fig. 7); however, it is a modeled input with a stated range, whereas the chi2/dof is an internal, unaddressed red flag. I therefore recommend keeping the CONDITIONAL verdict but shifting the emphasis to the fit quality. A dataset-exclusion test would settle whether the tension actually biases the headline numbers.","tokens_in":35046,"tokens_out":4555,"duration_ms":44045,"concrete_test":"Re-run the BCL fit described in Sec. 4.2 in three variants: (i) lattice points only (q2 = 12, 14, 16 GeV^2), (ii) LCSR points only (q2 = -4, 0, 4 GeV^2), and (iii) the full combined dataset, then compare the resulting BR(B0 -> K*0 nu nu) and F_L. If the central values in variants (i) or (ii) differ from the combined fit by more than the quoted uncertainties (0.96e-6 and 0.04), the combined-fit result is not robust and the 'precision' claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central numerical output, BR(B->K* nu nu) = 8.09(96)e-6 and F_L = 0.44(4), is produced by the BCL z-expansion fit described in Sec. 4.2. The fit yields chi2_min/dof = 40.1/23, which the paper reports without discussion and despite stating that the fit 'confirm[s] the consistency of the two complementary methods'. A chi2/dof of 1.74 with 23 dof is a poor fit; if the tension comes from the LCSR pseudo-data disagreeing with lattice QCD in the overlap region, the fitted z-coefficients and the q2-shape of all seven form factors are biased. The large relative uncertainties at q2=0 (e.g. V=0.20(14) and A0=0.066(38) from Table 2) also undercut the 'improve the precision' framing, since Ref. [27] quotes ~10% uncertainties. Because the branching fractions integrate these form factors over the full kinematic range, an unaddressed dataset tension directly propagates into the headline numbers.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs light-cone sum rules for the seven B→K* form factors using vacuum-to-B-meson correlation functions, and adds tree-level subleading-power corrections from three sources: higher-twist two- and three-particle B-meson light-cone distribution amplitudes, the power-suppressed terms in the hard-collinear propagator expansion, the subleading HQET weak current, and twist-five/six four-particle contributions. These NLP corrections are combined with the existing NLL leading-power results and are then fitted together with lattice QCD data in the small-recoil region using a BCL z-parametrization. The resulting form factors are used to predict the branching fractions BR(bar B^0 -> bar K*0 nu_l bar nu_l) = 8.09(96) x 10^-6, BR(bar B^+ -> bar K*+ nu_l bar nu_l) = 9.95(1.05) x 10^-6, and the longitudinal polarization fraction F_L = 0.44(4). The paper also discusses the impact of the inverse moment λ_B and provides a comparison with the light-meson-sum-rule results of Ref. [27].","tokens_in":35248,"tokens_out":7192,"duration_ms":70317,"significance":"If the calculation is correct, the analytic NLP results are a valuable and nontrivial extension of the SCET sum-rule program from B→D* to B→K*. The comparison in Sec. 3.3 with the earlier B→D* calculation is a useful cross-check, and the finding that the NLP corrections amount to roughly 30% of the leading-power results is important for precision flavor physics. The derivation is not circular: the branching fractions are predictions, not fitted inputs, and the calculation uses external inputs such as LCDA moments and lattice data. However, the numerical evidence for the central claims is not fully established. The combined BCL fit has a poor χ²/dof, and the q²=0 LCSR results in Table 2 carry much larger fractional uncertainties than the corresponding light-meson LCSR results, so the 'improved precision' claim requires a more careful quantitative comparison. The strong sensitivity to the model-dependent B-meson LCDA parameters, including the unresolved discrepancy between λ_B ≈ 300 MeV preferred by light-meson sum rules and the lattice value λ_B = 389(35) MeV, also needs to be confronted directly.","major_comments":[{"comment":"The combined fit yields χ²_min/dof = 40.1/23, corresponding to p ≈ 0.015. This is a poor fit for 23 degrees of freedom and contradicts the later statement in Sec. 5 that the fit 'confirm[s] the consistency of the two complementary methods'. Because the BCL coefficients extracted from this fit directly determine the form factors and the headline branching fractions, the paper must diagnose the source of the tension (for example, which lattice or LCSR points contribute most to χ²), report the p-value, and either enlarge the covariance matrices or use a more robust combination scheme. As written, the quoted uncertainties on the final observables likely underestimate the systematic tension between the LCSR and lattice inputs.","section":"Sec. 4.2, Eq. (77)"},{"comment":"The abstract's claim that the paper 'improve[s] the precision of theoretical predictions' is not supported by the q²=0 LCSR entries in Table 2: for example V = 0.20(14), A0 = 0.066(38), and A1 = 0.19(13) have relative errors of roughly 60–70%, whereas the corresponding light-meson LCSR results from Ref. [27] are V = 0.29(3), A0 = 0.118(16), and A1 = 0.306(33). If the precision claim is intended for the final full-range BCL fit, the paper should present a direct comparison of the BCL form-factor uncertainties with previous determinations, and should soften the wording for the large-recoil LCSR inputs. Otherwise the central 'improved precision' claim is not demonstrated.","section":"Table 2 and Sec. 4.2"},{"comment":"The final numerical predictions are strongly sensitive to λ_B and to the three-parameter model for the B-meson LCDAs, whose validity the authors themselves restrict to small momenta in Appendix D. The paper notes that light-meson LCSR fits prefer λ_B ≈ 300 MeV while the recent lattice value is λ_B = 389(35) MeV, and the final predictions nevertheless use λ_B = 350(150) MeV. Since λ_B enters the convolutions that determine the NLP corrections as well as the leading-power results, the authors should quote the shifts in BR and F_L obtained with the lattice-motivated λ_B = 389(35) MeV, and should state explicitly whether the unresolved λ_B discrepancy is included in the quoted uncertainties or treated as an additional model-dependence caveat.","section":"Sec. 4.1, Fig. 7, Appendix D"}],"minor_comments":[{"comment":"There is a typo in the first paragraph: 'In addtion' should be 'In addition'.","section":"Sec. 5"},{"comment":"The text writes 'K´allen function'; the correct spelling is 'Källén function'.","section":"Sec. 4.3"},{"comment":"The entries for {σ̂1, σ̂2} appear as three separate rows with values {0.7, 6.0}, {0.0, π²/6}, and {-0.7, -6.0}; the table should clarify which is the central value and which values are the boundaries of the uncertainty range.","section":"Table 1"},{"comment":"The fit coefficients and correlation matrix are said to be provided as supplemental material on the arXiv page; for a journal submission, this material should be included as an ancillary file or an appendix so that the fit results are available to referees and readers.","section":"Sec. 4.2"}],"recommendation":"major_revision","confidential_remarks":"This is a technically solid SCET sum-rule paper with a genuinely new NLP calculation and useful cross-checks. My main reservation concerns the numerical section: the BCL fit quality is poor and is not critically discussed, and the Table 2 results do not by themselves demonstrate improved precision. These issues are addressable in revision, so I do not recommend rejection, but the current wording of the precision and consistency claims needs to be substantiated or softened."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a serious, detailed SCET sum rule calculation. The genuinely new content is the twist-5/6 four-particle contributions and the complete tree-level subleading-power set for B->K*, while the leading-power NLL and the two/three-particle higher-twist corrections were already in Ref. [31]. The cross-check against the B->D* calculation in Sec. 3.3 is a good sanity check, and the appendices provide enough detail to reproduce the results.\n\nThe paper is honest about the dominant uncertainty: the B-meson LCDA model, especially lambda_B, and it compares the conventional 350(150) MeV with the lattice 389(35) MeV. The 30% NLP correction is a useful result.\n\nThe main soft spot is the combined BCL fit. Chi2/dof=40.1/23 is reported without comment, yet the summary claims the fit confirms consistency between LCSR and lattice. A chi2/dof of 1.74 for 23 dof is not a good fit; it points to underestimated uncertainties or a real tension between the two datasets. The paper should either improve the fit, enlarge the uncertainties, or at least discuss the discrepancy. This matters because the z-expansion coefficients and the q2 shape carry the tension into the branching fractions.\n\nMinor issue: the four-particle factorization approximation has no assigned systematic, though it appears numerically small (3-7% of NLP). The q^2=0 values have large uncertainties (e.g., V=0.20(14)), which undercuts the 'precision' framing, but this is consistent with the input uncertainties. The citation pattern is appropriate, with prior SCET work and lattice results properly acknowledged.\n\nNone of this destroys the central derivation. The analytic work looks solid, the power counting is checked, and the results are reproducible. The paper deserves a serious referee, but the fit issue needs to be addressed honestly. I would send it to peer review with a requirement to fix or discuss the fit and to propagate the fit quality into the final uncertainties.","headline":"Genuinely new subleading-power SCET calculation for B->K*, but the headline branching fractions sit on an underexplained poor-quality fit.","tokens_in":35881,"tokens_out":3319,"would_cite":true,"duration_ms":31582,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper extends the SCET light-cone sum-rule calculation of B→K* form factors to next-to-leading power, finding 30% corrections that sharpen Standard Model predictions for B→K*νν̅ decays.","keywords":["B→K* form factors","light-cone sum rules","soft-collinear effective theory","subleading-power corrections","B-meson light-cone distribution amplitudes","B→K*νν̅ decay","branching fraction","longitudinal polarization fraction"],"falsifier":"A lattice-QCD determination of the inverse moment $\\lambda_B$ with uncertainty below about $50\\,\\mathrm{MeV}$ that falls outside the $350$--$500\\,\\mathrm{MeV}$ window, combined with a Belle II measurement of the $B\\to K^*\\nu\\bar\\nu$ branching fraction at roughly 10\\% total uncertainty, would settle whether the 30\\% NLP correction is correctly normalized.","tokens_in":34758,"feed_emoji":"📐","tokens_out":12122,"duration_ms":100452,"temperature":0.7,"pith_summary":"This paper extends the SCET light-cone sum-rule (LCSR) calculation of the seven $B\\to K^*$ form factors to next-to-leading power (NLP), adding tree-level corrections from the hard-collinear propagator, the subleading effective weak current, and higher-twist $B$-meson distribution amplitudes. It finds that the NLP terms are about 30\\% of the leading-power result, comparable in size to the next-to-leading-logarithmic corrections, so they cannot be neglected. Combining the improved low-$q^2$ LCSR predictions with high-$q^2$ lattice QCD data through the BCL parametrization yields form factors across the full kinematic range. The resulting Standard Model predictions are $\\mathrm{BR}(\\bar B^0\\to \\bar K^{*0}\\nu_\\ell\\bar\\nu_\\ell)=8.09(96)\\times 10^{-6}$, $\\mathrm{BR}(\\bar B^+\\to \\bar K^{*+}\\nu_\\ell\\bar\\nu_\\ell)=9.95(1.05)\\times 10^{-6}$, and the longitudinal polarization fraction $F_L=0.44(4)$, giving sharp targets for Belle II.","feed_headline":"Subleading corrections shift B→K* form factors by 30%","feed_subtitle":"New power corrections are as big as the NLL terms, tightening SM rare-decay rates to ~10%.","key_machinery":"The machinery is the vacuum-to-$B$-meson correlation function in soft-collinear effective theory, expanded beyond leading power. The new subleading-power terms come from three sources: the heavy-quark expansion of the hard-collinear propagator, the subleading effective current $\\bar{q}\\Gamma[i\\,\\slashed{D}_{\\perp}/(2m_b)]h_v$, and higher-twist (including twist-five/six four-particle) $B$-meson light-cone distribution amplitudes. These are converted into sum rules using dispersion relations, quark-hadron duality, and Borel transformation. The numerical evaluation uses a three-parameter model for the $B$-meson LCDAs fixed by the inverse moment $\\lambda_B=350\\pm150\\,\\mathrm{MeV}$ and the inverse-logarithmic moments $\\hat\\sigma_1,\\hat\\sigma_2$.","core_discovery":"The central discovery is that the complete set of next-to-leading-power contributions at tree level changes the $B\\to K^*$ form factors by about 30\\% relative to the leading-power results, an effect as large as the NLL resummation itself. The dominant NLP source is the two-particle higher-twist $B$-meson light-cone distribution amplitudes, while the twist-5/6 four-particle effects contribute only 3--7\\% of the NLP total. The paper then shows that a combined fit of these improved LCSR form factors with lattice QCD results gives the Standard Model predictions $\\mathrm{BR}(\\bar B^0\\to \\bar K^{*0}\\nu_\\ell\\bar\\nu_\\ell)=8.09(96)\\times10^{-6}$, $\\mathrm{BR}(\\bar B^+\\to \\bar K^{*+}\\nu_\\ell\\bar\\nu_\\ell)=9.95(1.05)\\times10^{-6}$, and $F_L=0.44(4)$, with uncertainties dominated by the hadronic form factors.","pith_inferences":["Since the tree-level NLP effect is already 30\\% of the leading power, the next order (NLO in $\\alpha_s$ at NLP) could shift the branching fractions by several percent, so a full NLO-NLP calculation may be needed before claiming the quoted ~10\\% precision as a Standard Model benchmark.","The same SCET sum-rule machinery should apply to other $B\\to V$ transitions ($\\rho$, $\\omega$, $\\phi$); if similar 30\\% NLP effects appear, ratios such as $B\\to K^*$ to $B\\to\\rho$ form factors could be cleaner observables than individual form factors for new-physics searches.","The paper leaves out power-suppressed corrections from the interpolating currents and one-loop spectator-quark mass effects; given that the included NLP terms already shift the form factors by 30\\%, these omissions could be numerically relevant at the few-percent level.","The tension between $\\lambda_B\\approx 300\\,\\mathrm{MeV}$ preferred by light-meson sum rules and the lattice value of $389(35)\\,\\mathrm{MeV}$, noted in the paper, points to a genuine puzzle in the $B$-meson LCDA; higher-precision lattice determinations or a measurement of the $B\\to K^*\\nu\\bar\\nu$ $q^2$ distribution could indicate which side underestimates power corrections."],"forward_implications":["The 30% NLP correction means previous leading-power-only LCSR predictions for $B\\to K^*$ observables are shifted by an amount comparable to the NLL effects, so future extractions from $b\\to s$ transitions should adopt the updated form factors.","The combined LCSR+lattice fit gives $B\\to K^*$ form factors with smaller uncertainties over the full $q^2$ range than lattice-only fits, sharpening the interpretation of Belle II data on $B\\to K^*\\nu\\bar\\nu$.","The Standard Model predictions $\\mathrm{BR}(\\bar B^0\\to \\bar K^{*0}\\nu_\\ell\\bar\\nu_\\ell)=8.09(96)\\times10^{-6}$ and $F_L=0.44(4)$ are concrete, testable targets for the upcoming Belle II measurements.","The paper isolates the dominant source of the NLP corrections as the two-particle higher-twist $B$-meson LCDAs, identifying where better non-perturbative input would most reduce the uncertainty."],"supporting_citations":[{"why":"Supplies the leading-power NLL sum rules and effective B-meson LCDAs to which the present NLP corrections are added.","marker":"[31]"},{"why":"Provides the three-parameter model for the B-meson LCDAs and the central values of λ_B, σ̂1, σ̂2 used in the numerics.","marker":"[54]"},{"why":"Provides the high-q2 lattice QCD form factors used in the combined BCL fit.","marker":"[13]"},{"why":"Second lattice QCD data set for the same form factors at high q2, combined with [13] in the fit.","marker":"[14]"},{"why":"The analogous subleading-power calculation for B→D* against which the hard-collinear and current NLP results are checked.","marker":"[38]"},{"why":"Lattice determination of λ_B=389(35) MeV used as the alternative inverse-moment input for the form-factor comparison.","marker":"[70]"},{"why":"Light-meson-LCDA sum-rule form factors and branching fractions used as comparison in Tables 2 and 4.","marker":"[27]"},{"why":"Earlier calculation of two- and three-particle higher-twist subleading-power corrections that the present HT-NLP terms build on.","marker":"[29]"}],"fun_headline_variants":["B→K* form factors shift 30% from subleading power","Higher-twist B-meson effects drive 30% B→K* form factor change","SM rare-decay rates for B→K*νν now predicted to 8–10%","Longitudinal K* polarization predicted as F_L=0.44(4)"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the assumed model for the $B$-meson light-cone distribution amplitude and its inverse-moment parameter $\\lambda_B$, which is only fixed to $350\\pm150\\,\\mathrm{MeV}$; the quoted form factors and branching fractions are highly sensitive to this parameter.","fun_headline_variants_meta":{"raw":{"variants":["B→K* form factors shift 30% from subleading power","Higher-twist B-meson effects drive 30% B→K* form factor change","SM rare-decay rates for B→K*νν now predicted to 8–10%","Longitudinal K* polarization predicted as F_L=0.44(4)"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000408,"raw_usage":{"total_tokens":2212,"prompt_tokens":1133,"completion_tokens":1079,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":749,"completion_tokens_details":{"reasoning_tokens":989}},"tokens_in":749,"tokens_out":1079,"duration_ms":10288,"temperature":1.0,"reasoning_tokens":989,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:26:51.755472+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A lattice-QCD determination of the inverse moment $\\lambda_B$ with uncertainty below about $50\\,\\mathrm{MeV}$ that falls outside the $350$--$500\\,\\mathrm{MeV}$ window, combined with a Belle II measurement of the $B\\to K^*\\nu\\bar\\nu$ branching fraction at roughly 10\\% total uncertainty, would settle whether the 30\\% NLP correction is correctly normalized.","supporting_citations":[],"review_version":1}