{"id":"f5b28f3b-7ea9-4886-9590-28ef7c1880d2","arxiv_id":"hep-ph/0308039","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"QCD factorization at next-to-leading order is extended to the full set of 96 B to PP and B to PV decay channels, with branching-ratio and CP-asymmetry predictions and a quantified error budget.","lead":"Beneke and Neubert compute, at next-to-leading order in QCD, the amplitudes for all 96 two-body B decays into two pseudoscalars or a pseudoscalar plus a vector meson, and compare the resulting branching ratios and CP asymmetries with 2003 B-factory data. It became the standard reference for QCD factorization predictions in this sector.","discovery_kind":"extension","skeptic_critique":{"model":"selfhosted/deepseek-v4.1-flash","headline":"The sharpest structural prediction — distinct penguin magnitudes in PP, PV, and VP — is tested precisely where the non-factorizable annihilation correction is largest relative to the penguin, so the discriminating comparison is confounded by the least-controlled ingredient.","rationale":"I agree with the reader that the phenomenological layer, not the perturbative calculation, is where the paper is soft, and I agree that the power-correction modeling is the locus. I would sharpen the point. The reader's weakest_assumption centers on the universality of X_A and on the ρ_A ≥ 2 exclusion of Section 5.4. That exclusion is a secondary claim, and the paper is explicit that it is contingent on universality (it relaxes the assumption in scenario S4). The more load-bearing issue is upstream: the PP/PV/VP discrimination, which is the paper's headline structural prediction, is tested in exactly the channels where β_3^c is largest relative to the penguin. In PP the annihilation correction is a ~10% effect and Eq. (75) is a genuine parameter-free success; in PV and VP it is a factor-of-several effect with an arbitrary phase, so the predicted suppression of α̂_4^c(πK̄*) cannot be cleanly confirmed or refuted. The paper is honest about this — Section 9 says the reduction 'is not as large as predicted by theory' — and the Note added in proof reports that the Belle π⁻K̄*⁰ rate fell to (8.5 ± 1.3)×10⁻⁶, which the authors say would remove one of the major discrepancies. That is a point in the paper's favor: the PV sector is empirically unsettled and the framework is falsifiable there. I also note the acknowledged limitations the reader cites: F_2 = 0 with its uncertainty excluded from the quoted errors (Section 3), the 'educated guess' status of ρ_A ≤ 1 (Section 5.4), and scenario S4 being tuned to the data and then called favored. These are real but acknowledged, and per the review rules I do not present them as new findings. Credit where due: the NLO kernels of Section 2.4 and Appendix B are written down and internally consistent, the amplitude catalogue of Appendix A is complete, the scheme-dependence check around Eq. (37) for vector mesons is correct, and Eq. (75) is a genuine prediction that survives without adjustment. The concern I raise does not overturn the reader's CONDITIONAL verdict; it supports it and locates the soft spot more precisely. Hence UNCHANGED.","tokens_in":74244,"tokens_out":7476,"duration_ms":139681,"concrete_test":"Recompute the ratio |α̂_4^c(πK̄*)/α̂_4^c(πK̄)| of Eq. (76) with β_3^c set to zero in both channels, i.e. using only the factorized coefficients of Eqs. (31)–(48). If the pure-factorization ratio is near the quoted 0.32 while the data require 0.58 ± 0.07, determine the value of β_3^c(πK̄*) needed to close the gap and check whether it lies inside the ρ_A ≤ 1 band of Eq. (63) with a natural phase, or requires ρ_A > 1 or a fine-tuned φ_A. If the latter, the PV suppression prediction is not testable within the stated error model and the claim of distinct PP/PV/VP penguin magnitudes reduces to a statement about an unconstrained parameter. Repeat for Eq. (79) as a cross-check.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's most distinctive structural claim is that the penguin coefficient α̂_4^c takes markedly different values in PP, PV, and VP final states, a pattern that follows from the sign and size of the r_χ a_6 term in Eq. (31) together with N_i(M_2)=0 for i=6,8 when M_2 is a vector (Eq. 36). The cleanest quantitative form of this claim is Eq. (76), |α̂_4^c(πK̄*)/α̂_4^c(πK̄)| = 0.58 ± 0.07 measured against 0.32 predicted. The problem is that the annihilation correction β_3^c, which is the least controlled quantity in the analysis, is small in PP (|β_3^c/α_4^c| = 0.09 +0.32−0.09 in Section 5.1.1) but comparable to or larger than the penguin in PV (0.16 +0.89−0.14) and VP (0.18 +0.83−0.18). The predicted PV suppression is therefore not a clean test: the gap between 0.32 and 0.58 can be closed by dialing β_3^c anywhere in a range spanning roughly a factor of five, with an arbitrary phase φ_A in Eq. (63). The same entanglement appears in Eq. (79), where the theory value 1.33 +0.43−0.36 is compared with 0.82 ± 0.10. Section 9's statement that 'the reduction is not as large as predicted by theory' is thus a statement about the central value of an uncalculable correction, not a test of factorization. This does not undermine the PP sector, where Eq. (75) gives 0.093 against 0.103 ± 0.008 ± 0.023 with β_3^c contributing only 0.085→0.093, nor the R_00 < 1 argument of Section 5.2.2, which is genuinely robust. The concern lands specifically on the PP/PV/VP discrimination.","agreement_with_reader":"partial"},"referee_report":{"model":"selfhosted/deepseek-v4.1-flash","summary":"The manuscript presents a complete next-to-leading-order QCD factorization calculation of the 96 two-body decay amplitudes of B−, B̄0, and B̄s into PP and PV final states. The technical core is the completion of the hard-scattering kernels, including the new vertex kernel h(x) of Eq. (38) and the twist-3 vector-meson convolutions collected in Appendix B, together with a unified flavor-operator notation whose master formula, Eq. (119), generates every amplitude from a single expression. The phenomenological part compares these predictions with world averages, using a four-class error budget (CKM, hadronic 1, hadronic 2, power) and four named scenarios S1–S4, and discusses in detail the magnitudes of the penguin coefficients in PP, PV, and VP topologies, the πρ system including the five time-dependent asymmetries, and the mixing-induced asymmetries in φKS and η′KS.","tokens_in":74696,"tokens_out":4254,"duration_ms":100885,"significance":"If the framework holds, this is the reference calculation for the field: it fixes the NLO kernels and the amplitude bookkeeping that subsequent B→PP and B→PV analyses build on, and it supplies predictions for a large set of channels that were unmeasured at the time of writing. The paper is unusually explicit about its own error taxonomy, separating CKM, hadronic, Gegenbauer, and power-correction uncertainties and exhibiting correlations through the four scenarios rather than quoting a single number per mode. Several results are genuinely falsifiable and are stated as such: the Standard Model prediction R00 < 1 in Eq. (87), the near-unity ratio RL in Eq. (91), and the few-percent differences SφKS − SJ/ψKS and Sη′KS − SJ/ψKS in Eq. (83). The technical completion of the vector-meson kernels and the master formula are concrete, reusable contributions independent of the phenomenological conclusions.","major_comments":[{"comment":"The PP/PV/VP discrimination is presented as a test of factorization, but the discriminating ratios are precisely where the least-controlled ingredient is largest. In §5.1.1 the paper itself quotes |β3c/α4c| = 0.09 +0.32−0.09 for PP but 0.16 +0.89−0.14 for PV and 0.18 +0.83−0.18 for VP, so the predicted suppression |α̂4c(πK̄*)/α̂4c(πK̄)| = 0.32 can be moved toward the measured 0.58 ± 0.07 by dialing β3c within its stated range with an arbitrary phase φA. The same applies to Eq. (79), where 1.33 +0.43−0.36 is compared with 0.82 ± 0.10. The authors acknowledge this in §9 (the prediction 'may be in agreement with the measured branching fractions within their errors'), but the intermediate text and the abstract's framing of 'the magnitudes of the penguin amplitudes in PP and PV final states' read more strongly than the calculation supports. I would ask that the text at Eqs. (76) and (79) state explicitly that these ratios are not clean tests of the PP/PV/VP pattern because the annihilation correction is comparable to the penguin in the PV and VP cases, and that the abstract and §9 be calibrated accordingly.","section":"§5.1.1, Eqs. (76) and (79)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The exclusion of ρA ≥ 2 in Table 7 is derived under the assumption that XA is universal across PP, PV, and VP, and the authors themselves relax that assumption only in scenario S4, where the fitted phases differ substantially by topology (φA = −55°, −20°, −70°). The sentence 'values ρA ≥ 2 can already be excluded, at least if universal annihilation is a reasonable approximation' is appropriately hedged, but the table itself does not display the conditional nature of the bound. A short non-universal scan, or an explicit statement in the table caption that the entries assume a single XA for all three topologies, would make the scope of the bound unambiguous. As written, a reader could take Table 7 as a model-independent statement about the size of weak annihilation.","section":"§5.4, Table 7","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The default F2 = 0 for the singlet contribution to the B→η(′) form factor is stated in §3 to introduce 'an additional theoretical uncertainty not taken into account in the error ranges given below', and the text notes that setting F2 = 0.1 raises Br(B−→π−η′) to about 7×10−6 and Br(B−→ρ−η′) to 16×10−6. The η(′) tables (Tables 14 and 16) nevertheless quote four-column error budgets that do not include this variation, so the quoted errors understate the true spread for those modes by a potentially large factor. I would ask that the η(′) tables either include an F2 variation in the error columns or carry an explicit caption flag pointing to the §3 caveat, so that the reader does not compare those errors on the same footing as the ππ and πK entries.","section":"§3 and Tables 14, 16","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The experimental average for B−→π−K̄*0 in Table 2 is 13.0 ± 3.0 with S = 1.4, but the Note added in proof reports a Belle value of (8.5 ± 1.3)×10−6 that would substantially change this average. A footnote in Table 2 pointing to the note would prevent the table from being read as current.","section":"Table 2","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The parameterization XA = (1 + ρA e^{iφA}) ln(mB/Λh) with ρA ≤ 1 is described as assigning a 100% uncertainty to the default value. It would help to state once, in the equation's immediate vicinity, that ρA ≤ 1 is a modeling choice rather than a derived bound, since the rest of §5.4 leans on it.","section":"Eq. (63)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The additional electroweak penguin contribution for neutral vector mesons is evaluated at ν = µ, and the text notes that the ν dependence cancels against the leading-order decay amplitude to order α. A sentence quantifying the residual scale dependence at the quoted order would make the approximation's size explicit.","section":"Eq. (53)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The error breakdown for 2df cosθf is quoted with asymmetric errors whose ordering does not obviously match the four-class convention used in Tables 2–3. A parenthetical mapping the four numbers to the CKM, hadronic 1, hadronic 2, and power classes would remove the ambiguity.","section":"Eq. (82)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The discussion of the CP and sign conventions for the π∓ρ± amplitudes, and in particular the statement that the ratio of form factors and decay constants in ρπ−ρ+ is negative, is important for the interpretation of Eq. (105) but is confined to a long footnote. Promoting the essential part of that argument to the main text would help readers who take the sign of Pρπ/Tρπ from Table 12 at face value.","section":"§6.3, footnote 5","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The experimental averages inflate the combined error by S = √(χ²/(N−1)) when S > 1, and the text states that correlated errors are ignored. Since several of the averages in Tables 25–28 combine BaBar, Belle, and CLEO results with common systematics, a sentence on the expected size of the neglected correlations would help the reader judge the averages used in the comparison sections.","section":"Appendix C","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"minor_revision","confidential_remarks":"This is a long, careful, and technically complete paper that will be a standard reference; the requested revisions are about calibrating the strength of specific claims and about error bookkeeping in the η(′) tables, not about the correctness of the calculation. The 'Guide to reading the paper' is a helpful concession to the manuscript's length. Several tables will need updating as the experimental averages move, and the Note added in proof already flags two such cases; the journal may wish to encourage the authors to fold those updates into the tables rather than leaving them in a note."},"author_rebuttal":null,"desk_editor":{"model":"selfhosted/deepseek-v4.1-flash","letter":"Beneke and Neubert is the reference you keep on the shelf if you work on charmless hadronic B decays. The question is whether the phenomenology layer deserves the same trust as the perturbative one, and mostly it does not — though less of the paper is affected than the stress-test note implies. What is actually new is bounded and real: the twist-3 vector-meson convolutions in Appendix B (Eqs. 135–139), the neutral-vector electroweak penguin of Eq. (53), the singlet annihilation coefficient b^S_3, and a complete NLO amplitude catalogue recast in a flavor-operator notation with the master trace formula of Eq. (119). That last piece is the practical contribution — it is what people lifted into code. The error budget is unusually explicit for 2003: four separated classes, four named scenarios. The soft spot is the annihilation sector. X_A is a model with rho_A <= 1, which Section 5.4 itself calls an educated guess. The stress-test concern lands: the PP/PV/VP discrimination sits exactly where beta_3^c is least controlled. In PP, |beta_3^c/alpha_4^c| is about 0.09, so Eq. (75)'s 0.093 against 0.103 +/- 0.008 +/- 0.023 is a genuine test. In PV and VP the ratio is 0.16 +0.89-0.14 and 0.18 +0.83-0.18 — a factor-of-five range with a free phase. So Eq. (76)'s 0.32 against 0.58 +/- 0.07, and Eq. (79)'s 1.33 against 0.82, are not clean tests of the r_chi a_6 mechanism; they test a central value of an uncalculable correction. Section 9's line that the reduction is not as large as predicted should be read that way. The Section 5.4 exclusion of rho_A >= 2 also rests on universality, which S4 relaxes. Two smaller things. S4 is tuned to the data and then called favored, sitting in the same table as the predictions; the authors are open about it, but it is not a prediction. And F_2 = 0 with the resulting uncertainty explicitly excluded from the quoted errors (Section 3) means the eta-prime rows carry error bars missing a known systematic — they flag it, the tables do not. Credit where due: the Note added in proof is the most honest passage in the paper, reporting the Belle pi- K*0 drop themselves and saying it would remove a major discrepancy. For anyone doing B to PP/PV phenomenology this is required reading and required citation. Send it out.","headline":"The NLO kernels and amplitude catalogue are the durable contribution; the PP/PV/VP penguin discrimination is confounded by the annihilation correction it is supposed to test.","tokens_in":75417,"tokens_out":1860,"would_cite":true,"duration_ms":35796,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":[],"pacs":["13.25.Hw","12.38.Bx","11.30.Er","14.40.Nd"],"model":"selfhosted/deepseek-v4.1-flash","headline":"QCD factorization predicts 96 B-meson decay rates and a topology-dependent penguin amplitude.","keywords":["QCD factorization","B-meson decays","penguin amplitudes","CP violation","weak annihilation","light-cone distribution amplitudes","CKM angle gamma","B_s decays"],"falsifier":"A measurement of the semileptonic spectrum in $Bto pi lnu$ near $q^2=0$, combined with $B^-to pi^-pi^0$, would determine $|alpha_1(pipi)+alpha_2(pipi)|$ without form-factor uncertainty; a value incompatible with the factorization prediction $1.17$ would falsify the tree sector, while a pure-annihilation mode such as $bar B^0to K^-K^+$ observed at a rate requiring a universal $X_A$ outside the $rho_Ale1$ range would falsify the annihilation model.","tokens_in":75512,"feed_emoji":"⚛️","tokens_out":5264,"duration_ms":105543,"temperature":0.7,"pith_summary":"The paper asks whether QCD factorization — the claim that in the heavy-quark limit a two-body $B$ decay amplitude factorizes into a calculable hard kernel times form factors and light-cone distribution amplitudes — can be carried through consistently for every charmless final state made of two pseudoscalars or one pseudoscalar and one vector meson. The authors complete the next-to-leading-order hard-scattering kernels and assemble a single flavor-operator master formula that generates all 96 $B^-$, $B^0$, and $B_s$ decay amplitudes. They then add model estimates for the two non-factorizable power corrections, chirally enhanced spectator scattering and weak annihilation, and compare the resulting branching ratios and CP asymmetries with the available data. A sympathetic reader should care because a working calculation of this kind turns a large set of hadronic $B$ decays into quantitative tests of the Standard Model and into inputs for the CKM angle $gamma$, while also exposing where the data already strain the picture.","feed_headline":"QCD factorization predicts 96 B-meson decay rates","feed_subtitle":"The same calculation ties three penguin amplitudes to distinct final-state types and bounds annihilation.","key_machinery":"The load-bearing device is a transition operator written as a sum of flavor operators, with coefficients $alpha_i(M_1M_2)$ and $beta_i(M_1M_2)$, rather than the older $a_i$ notation. The $alpha_i$ contain the leading-power physics — vertex corrections, penguin contractions, and hard spectator scattering — and the $beta_i$ contain the power-suppressed weak annihilation terms. A single trace formula over flavor matrices, Eq. (119), then generates every decay amplitude, so the 96 channels are not 96 independent calculations but one calculation evaluated on different flavor labels. The two endpoint-divergent integrals that factorization cannot control are replaced by complex parameters $X_H$ and $X_A$, with $X_A=(1+rho_A e^{iphi_A})ln(m_B/Lambda_h)$ and $rho_Ale1$, and the variation of these parameters is what the paper calls its power-correction error.","core_discovery":"The central claim is that QCD factorization, at leading power in $Lambda_{rm QCD}/m_b$ and next-to-leading order in $alpha_s$, gives a coherent description of the complete set of 96 $Bto PP$ and $Bto PV$ decay channels once the known power corrections are included in the stated error model. The sharpest quantitative statement is that the dominant penguin coefficient $hatalpha_4$ is correctly predicted in the $PP$ channel without adjusting parameters: the measured ratio in Eq. (75) is $0.103pm0.008pm0.023$ against a predicted $0.093$. The paper further claims that the same coefficient is genuinely smaller in $PV$ and $VP$ final states, by roughly a factor of three, for reasons specific to factorization — the vanishing of the scalar penguin contribution when the vector meson is the second meson — and that the data on $pi K^*$ are consistent with a suppressed penguin amplitude, though less suppressed than the default prediction. It also reports several observables that are hard to fit, especially the large $bar B^0to pi^0bar K^0$ rate, the difference $S_{phi K_S}-S_{J/psi K_S}$, and the direct CP asymmetries in $pi^-pi^+$ and $pi^-rho^+$, and argues that these are not easily explained by hadronic rescattering within the factorization framework.","pith_inferences":["The topology-dependent penguin pattern is a cleaner test of factorization than any single branching ratio, because it survives form-factor and CKM uncertainties; a global fit treating $hatalpha_4(PP)$, $hatalpha_4(PV)$, and $hatalpha_4(VP)$ as three independent parameters would turn the paper's prediction into a falsifiable three-point relation.","If $X_A$ is not universal, the paper's power-correction error bars are underestimated in channels where annihilation competes with a suppressed penguin, especially $PV$ modes; future data on $Bto rho K$ and $B_sto K K^*$ could measure the spread.","The same master-formula organization could be extended to two-vector final states, which the paper explicitly leaves out, and to $D$-meson decays where the heavy-quark limit is less secure but the flavor-operator bookkeeping still applies.","The anomalies the paper flags are not all independent: $R_{00}$ and $R_L$ both point to an enhanced electroweak penguin, so a future global analysis should test whether one new-physics coefficient can fit both without spoiling the $phi K_S$ constraint."],"forward_implications":["The complete tables give target branching ratios for dozens of unobserved modes, including $B_s$ decays that will become accessible at hadron machines.","If the $PP$/$PV$/$VP$ penguin pattern holds, a measured penguin rate in one topology calibrates the others and reduces the form-factor uncertainty in extracting $gamma$.","The time-dependent asymmetry $S$ in $Bto pirho$ gives $gamma=(72pm11)^circ$ or $(151pm10)^circ$, a constraint competitive with the $pipi$ channel.","The Standard Model predictions $S_{phi K_S}-S_{J/psi K_S}=0.025pm0.012pm0.010$ and $R_{00}<1$ turn the observed deviations into quantitative tests for new physics in electroweak penguins.","The bound $rho_A<2$ from $Kphi$ and related modes says weak annihilation cannot be arbitrarily large, which constrains any future nonperturbative treatment."],"supporting_citations":[{"why":"Introduces the QCD factorization formula for two-body $B$ decays and establishes the leading-power hard-scattering picture the paper extends.","marker":"[8]"},{"why":"Supplies the previous next-to-leading-order $pipi$ and $pi K$ analysis and the $X_A$/$X_H$ model for power corrections that this paper generalizes.","marker":"[10]"},{"why":"Provides the treatment of $eta$ and $eta'$ mesons, including their two-gluon content and singlet annihilation, used in Section 7.","marker":"[18]"},{"why":"Gives the Wandzura-Wilczek relations that reduce the vector-meson twist-3 amplitudes to $Phi_v$, the input for the new $PV$ kernels.","marker":"[33]"},{"why":"Defines the Feldmann-Kroll-Stech mixing scheme adopted for $eta$ and $eta'$.","marker":"[29]"},{"why":"Supplies the isospin relation behind $R_L=1$ up to quadratic corrections in small amplitude ratios.","marker":"[55, 56]"},{"why":"Independent analysis reaching the same conclusion that $pi K$ data may require a non-standard electroweak penguin.","marker":"[53]"},{"why":"Estimates the error from treating the $rho$ width as negligible in the quasi-two-body $pirho$ analysis.","marker":"[61]"}],"fun_headline_variants":["QCD factorization covers all 96 B to PP and PV decays","Penguin coefficient predicted without tuning in B to PP decays","Factorization predicts a threefold smaller penguin in PV final states","96 B-meson decays tested against next-to-leading QCD factorization","QCD factorization fits 96 B decays but misses several CP observables"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The load-bearing premise is that the two non-factorizable power corrections can be compressed into the complex parameters $X_H$ and $X_A$, with the same $X_A$ used for $PP$, $PV$, and $VP$ final states and with $rho_Ale1$ as an educated guess; the paper's exclusion of $rho_Age2$ depends on that universality, which the authors relax only in scenario S4.","fun_headline_variants_meta":{"raw":{"variants":["QCD factorization covers all 96 B to PP and PV decays","Penguin coefficient predicted without tuning in B to PP decays","Factorization predicts a threefold smaller penguin in PV final states","96 B-meson decays tested against next-to-leading QCD factorization","QCD factorization fits 96 B decays but misses several CP observables"]},"model":"selfhosted/deepseek-v4.1-flash","effort":"low","attempts":1,"cost_usd":0.0,"raw_usage":{"total_tokens":1668,"prompt_tokens":1013,"completion_tokens":655,"prompt_tokens_details":null,"completion_tokens_details":{"reasoning_tokens":579}},"tokens_in":1013,"tokens_out":655,"duration_ms":15372,"image_pages":[],"temperature":1.0,"reasoning_budget":50,"reasoning_tokens":579,"transport_provider":"selfhosted","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-10-29T17:05:21.256555+00:00","model_set":{"reader":"selfhosted/deepseek-v4.1-flash"},"falsifier":"A measurement of the semileptonic spectrum in $Bto pi lnu$ near $q^2=0$, combined with $B^-to pi^-pi^0$, would determine $|alpha_1(pipi)+alpha_2(pipi)|$ without form-factor uncertainty; a value incompatible with the factorization prediction $1.17$ would falsify the tree sector, while a pure-annihilation mode such as $bar B^0to K^-K^+$ observed at a rate requiring a universal $X_A$ outside the $rho_Ale1$ range would falsify the annihilation model.","supporting_citations":[{"cited_title":"QCD Factorization for $B\\to\\pi\\pi$ Decays: Strong Phases and CP Violation in the Heavy Quark Limit","cited_arxiv_id":"hep-ph/9905312","evidence_quote":"Introduces the QCD factorization formula for two-body $B$ decays and establishes the leading-power hard-scattering picture the paper extends."},{"cited_title":"QCD Factorization in B -> pi K, pi pi Decays and Extraction of Wolfenstein Parameters","cited_arxiv_id":"hep-ph/0104110","evidence_quote":"Supplies the previous next-to-leading-order $pipi$ and $pi K$ analysis and the $X_A$/$X_H$ model for power corrections that this paper generalizes."},{"cited_title":"Flavor-Singlet B-Decay Amplitudes in QCD Factorization","cited_arxiv_id":"hep-ph/0210085","evidence_quote":"Provides the treatment of $eta$ and $eta'$ mesons, including their two-gluon content and singlet annihilation, used in Section 7."},{"cited_title":"Higher Twist Distribution Amplitudes of Vector Mesons in QCD: Formalism and Twist Three Distributions","cited_arxiv_id":"hep-ph/9802299","evidence_quote":"Gives the Wandzura-Wilczek relations that reduce the vector-meson twist-3 amplitudes to $Phi_v$, the input for the new $PV$ kernels."},{"cited_title":"A Possibility of large electro-weak penguin contribution in B -> K pi modes","cited_arxiv_id":"hep-ph/0306147","evidence_quote":"Independent analysis reaching the same conclusion that $pi K$ data may require a non-standard electroweak penguin."}],"review_version":1}