{"id":"5b3b788a-a719-44ba-a315-a3c3cefeefff","arxiv_id":"2502.08427","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Including twist-5 and twist-6 B_s-meson distribution amplitudes enhances the charm-loop non-local form factors in B_s to phi l+l- by about an order of magnitude, producing a larger but still Standard-Model-consistent correction to C9.","lead":"This paper recalculates the 'charm-loop' contribution to a rare particle decay, B_s to phi plus muon and antimuon, including higher-order corrections to the B_s meson's internal structure. The new corrections make the charm-loop effect about ten times larger than earlier estimates, which may change how deviations from the Standard Model are interpreted in this decay.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Twist-5 LCDA normalization (λ2_E, λ2_H) fully controls the claimed order-of-magnitude enhancement; the paper's own Fig. 2 and Eq. (43) make the claim contingent on one poorly known input combination without presenting a significance statement.","rationale":"The reader's weakest_assumption correctly targets the Exponential-Model twist-5 LCDA and the poorly determined λ2_E,H. My independent check agrees: Eq. (43) and Table 3 show the enhancement is a linear-in-λ2_E,λ2_H effect with a sign and magnitude set by the twist-5 combination; competing determinations shift it by orders of magnitude; and the paper's own uncertainty (≈61%) makes the 'order of magnitude' statement formally sub-1σ. I therefore agree with the conditional verdict and with the request for a significance statement. I add one sharper technical point: the twist-5 dominance itself is a consequence of the Exponential Model's specific ω1,ω2 dependence (ψ̃5 ∝ ω1 e^{-(ω1+ω2)/ω0}), and the cancellation pattern among Lorentz structures at O((v·x)^{-1}) in Table 3 — ψ̃5 almost exactly cancels X_A, ψ_V, etc. — could rearrange if a different model shape were used. Thus the strongest_claim is not internally inconsistent, but it is input-dominated rather than conclusion-robust.","tokens_in":33266,"tokens_out":1769,"duration_ms":16265,"concrete_test":"Recompute the non-local form factors with the twist-5/6 input modeled by the Local Duality Model (mentioned in §4.1.2 as giving a similar pattern) plus the alternative λ2_E,H values from Regions II and III, and quote the ratio êV_λ(twist5+twist6)/êV_λ(twist4-only) with full propagated uncertainties. If the ratio at q²=−1 GeV² is compatible with 1 within 1σ for any published input set, the order-of-magnitude claim is not robust.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The claim that twist-5/twist-6 three-particle LCDAs enhance the non-local form factors by an order of magnitude rests almost entirely on the Exponential-Model twist-5 LCDA ψ̃5 — in Table 3, the twist-5 contribution −143.32×10⁻⁸ to êV_⊥ at q²=−1 GeV² is dominated by −139.47×10⁻⁸ from ψ̃5, which the paper flags as proportional to |λ2_E + λ2_H|. Eq. (43) shows êV_⊥(−1) = λ2_H[(50.416−50.416R)_3 + (−8.247−24.772R)_4 + (−178.114−121.170R)_5 + (21.952−21.952R)_6]×10⁻⁷ with R=λ2_E/λ2_H. At the central values λ2_E=0.03, λ2_H=0.06 GeV² this gives the large negative twist-5 term; but the published 1σ ranges (0.01–0.05, 0.03–0.09) and especially the competing determinations in Regions II–V (e.g. λ2_E=0.11±0.06, λ2_H=0.18±0.07; λ2_E=0.01±0.01, λ2_H=0.15±0.05) change the linear combination substantially, and Fig. 2 shows order-of-magnitude variation across those regions. The large quoted error on êV_⊥ (Table 2: −13.364±8.171) already implies the enhancement is not established at the 1σ level relative to the twist-4-only value (0.302±2.049): the difference is 1.7σ. Thus the central claim is a model-selection-and-parameter-normalization artifact unless a significance statement is provided.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper extends the LCSR calculation of non-local charm-loop form factors for B_s → φ ℓ+ℓ− by including twist-5 and twist-6 three-particle B_s-meson LCDAs. Using the Exponential Model for the LCDAs, the authors find that the twist-5 contributions break the cancellations seen in the earlier twist-4 analysis of Ref. [12], leading to an approximately order-of-magnitude increase in the non-local form factors. The results are continued to the physical region using double-subtracted dispersion relations with resonance inputs from B_s → φ V data and fitted continuum parameters. The paper also updates local form factors with higher-twist three-particle contributions and studies the impact on C9 and angular observables.","tokens_in":33648,"tokens_out":9071,"duration_ms":83720,"significance":"If the enhancement survived scrutiny, it would have important implications for b → s ℓℓ phenomenology, as the charm-loop correction to C9 would be large and polarization-dependent. The paper is thorough in providing breakup tables (Table 3, Appendix D), explicit sensitivity expressions (Eq. 43), and a detailed account of the LCSR framework, and it reproduces the twist-4 limit of Ref. [12] (up to input differences). However, the headline enhancement is driven by a single twist-5 LCDA ψ̃5 whose normalization λ2_H is poorly constrained, and the paper's own Fig. 2 and sensitivity analysis show that alternative determinations can change the prediction by orders of magnitude. Hence the central quantitative claim is not yet established at the 1σ level.","major_comments":[{"comment":"The claimed order-of-magnitude enhancement is driven by the twist-5 term in Eq. (42), which for eV⊥(−1) equals −14.334×10⁻⁷ and is dominated by the ψ̃5 contribution (−139.47×10⁻⁸ of the −143.32×10⁻⁸ twist-5 total in Table 3). In the Exponential Model of Eq. (70), ψ̃5 is proportional to λ2_H, so the enhancement is controlled by one poorly known input combination. The difference between the twist-4-only value eV⊥(−1)=0.302±2.049 and the full result −13.364±8.171 (Table 2) is only about 1.6σ once uncertainties are added in quadrature. The paper should provide a quantitative significance statement for the enhancement, and should present central predictions under the alternative λ2_E,H determinations (Regions II–V of Fig. 2) to show whether the order-of-magnitude effect persists. As it stands, the headline claim is not statistically robust.","section":"Sec. 4.1.1, Eq. (42), Table 3"},{"comment":"The double-subtracted dispersion relations in Eqs. (38)–(39) contain the subtraction constants Hλ(q0²) and dHλ/dq² at q0² = −1 GeV², but Table 8 lists only the phases φ0_V and continuum parameters aλ, bλ. The manuscript does not state how the subtraction constants and their derivatives are obtained: if they are taken from the LCSR predictions at q0² = −1, the derivative is not directly available from the sum rule and requires an additional modeling assumption; if they are fitted, they should appear as fit parameters with uncertainties. This information is needed to validate the extrapolation used for the C9 predictions in Fig. 3.","section":"Sec. 3, Eqs. (38)–(39), Sec. 4.2.2, Table 8"},{"comment":"Eq. (43) and Fig. 2 show that eVλ is linear in λ2_H and R = λ2_E/λ2_H, and the paper acknowledges that along Regions IV and V 'such estimates can even vary by several orders of magnitude'. The uncertainties quoted in Table 2 (e.g., eV⊥(−1) = −13.364±8.171) reflect only the Region I 1σ ranges of Table 1 and do not include the spread across the alternative determinations of λ2_E,H presented in Regions II–V. The paper should either restrict the analysis to a well-justified range of λ2_E,H or propagate the full spread into the quoted errors; otherwise the central values are not representative of the current state of knowledge.","section":"Sec. 4.1.2, Eq. (43), Fig. 2"}],"minor_comments":[{"comment":"The comparison with Ref. [12] for eV0 at q² = −1 GeV² shows a difference of about 2.5σ between the central values (0.101±0.065 vs −0.15±0.08), which is more than the 'slight difference' described in the text; the authors should clarify the input choices that produce this shift.","section":"Sec. 4.1.1, Table 2"},{"comment":"The fitted phases φ0_V are quoted without uncertainties, although they are fit parameters; the authors should report their errors or explain why they are fixed.","section":"Sec. 4.2.2, Table 8"},{"comment":"There are numerous typographical artifacts, e.g., 'Ge V2' instead of 'GeV²' in Sec. 4.1.1 and elsewhere, and inconsistent spacing in 'Bs → ϕ¯ℓℓ'; the manuscript would benefit from a careful proofreading pass.","section":"Throughout"},{"comment":"The paper notes that anomalous thresholds from multiparticle states are ignored and a simple linear continuum model is used (Eq. (40)); a brief comment on the expected size of this approximation would be useful, especially given the recent literature cited as Refs. [26,27].","section":"Sec. 3, Eq. (40)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a hep-ph journal and the LCSR calculation is a useful, transparent contribution. The main reservation is the robustness of the headline enhancement, which rests on a poorly constrained input (λ2_H) and is not significant at the 1σ level. The authors are, however, candid about the sensitivity, and the issues appear addressable in a revision that provides significance statements and propagates the full input spread."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does something real: it extends the established LCSR calculation of non-local charm-loop form factors for B_s → φℓℓ from twist-4 (Gubernari–van Dyk–Virto) to include twist-5 and twist-6 three-particle B_s-meson LCDAs. The twist-4 limit reproduces Ref. [12], the breakup tables and ancillary files are genuinely useful, and the paper is transparent about its inputs and fit definitions. That is solid workmanship.\n\nThe problem is the headline. The claimed order-of-magnitude enhancement is driven almost entirely by the twist-5 LCDA ψ̃5, whose normalization is proportional to λ2_E + λ2_H. At the central values (0.03, 0.06 GeV²) this gives the large negative contribution, but the quoted uncertainties already tell the story: Table 2 gives eV⊥(−1) = −13.364 ± 8.171 versus the twist-4-only value 0.302 ± 2.049. The shift is about 1.7σ. The paper’s own Fig. 2 shows that across the allowed regions for λ2_E and λ2_H—including alternative determinations from the literature—the form factors vary by orders of magnitude. So the central claim is conditional on a poorly known input combination, and the paper does not provide a significance statement. The abstract’s “enhances by approximately an order of magnitude” is an overstatement without that caveat.\n\nI want to be fair: the paper is not hiding the sensitivity. Section 4.1.2 is explicit about it, and the summary says the corrections are highly sensitive to λ2_E and λ2_H. But the framing in the abstract and introduction leads with the enhancement, and the phenomenological discussion (Fig. 4) treats it as a real effect rather than an illustrative possibility. The dispersion-relation continuation adds further model dependence through the linear continuum ansatz and fitted phases; the paper mentions anomalous cuts but does not delve into them. For the local form factors, the twist-5/6 impact is correctly found to be below 1%.\n\nThe calculation itself deserves a serious referee. It is a legitimate extension, the algebra is internally consistent, and the sensitivity analysis is informative. What is missing is a significance statement and a discussion of model dependence—for example, showing the Local Duality Model results rather than just mentioning them. A revision that reframes the central claim as “the effect is potentially large but currently uncertain, and depends strongly on λ2_E + λ2_H” would be honest and still useful.\n\nSend it to referees, but expect heavy revision on the interpretation. For my own work, I would cite it as the first twist-5/6 LCSR calculation, not as a reliable prediction of the charm-loop effect.","headline":"A clean, well-documented LCSR extension to twist-5/6, but the claimed order-of-magnitude enhancement is a one-parameter effect that is not statistically robust once the λ2_E, λ2_H uncertainties are taken into account.","tokens_in":34327,"tokens_out":2222,"would_cite":true,"duration_ms":25309,"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":"Adding twist-5 and twist-6 $B_s$-meson distribution amplitudes enlarges the non-local charm-loop form factors for $B_s\\to\\phi\\bar{\\ell}\\ell$ by about an order of magnitude, with a twist-5 term dominating the shift.","keywords":["B_s -> phi l+l- decay","charm-loop effect","light-cone sum rules","B-meson light-cone distribution amplitudes","higher-twist effects","non-local form factors","Wilson coefficient C9","rare b -> s l+l- transitions"],"falsifier":"Compute the same non-local form factor with an independent determination of the twist-5 normalization, for example from a lattice QCD calculation of the relevant three-particle correlation function or from alternative QCD sum rules for $\\lambda_2^E$ and $\\lambda_2^H$, and check whether $\\tilde V_\\perp(-1\\,\\mathrm{GeV}^2)$ stays near $-13\\times 10^{-7}$ rather than returning to the old value near $0.3\\times 10^{-7}$. A complementary check is to fit the low-$q^2$ $B_s\\to\\phi\\mu^+\\mu^-$ angular data with and without the enhanced charm loop and see which reproduces the measured $S_3$ and $S_7$ bins.","tokens_in":32967,"feed_emoji":"","tokens_out":11422,"duration_ms":97167,"temperature":0.7,"pith_summary":"The paper tries to establish that higher-twist (twist-5 and twist-6) three-particle $B_s$-meson light-cone distribution amplitudes cannot be dropped when estimating the non-local charm-loop contribution to $B_s\\to\\phi\\bar{\\ell}\\ell$. Including them in the light-cone sum rule changes the non-local form factors by roughly an order of magnitude relative to the previous twist-4 truncation, because the new twist-5 term breaks a cancellation that had suppressed the twist-3 and twist-4 contributions. This matters for rare $b\\to s\\ell\\ell$ decays, where the charm loop is one of the largest theory uncertainties and where measured observables have shown deviations from Standard Model expectations. The paper's corrected charm loop translates into a $q^2$-dependent shift of the effective Wilson coefficient $C_9$ that is larger than the factorizable-only prediction but still consistent with the Standard Model within uncertainties.","feed_headline":"Charm-loop effect in B_s -> phi l+l- jumps an order of magnitude","feed_subtitle":"Twist-5 and twist-6 B_s LCDAs enlarge the predicted C9 shift for Bs -> phi l+l-.","key_machinery":"The load-bearing object is the set of three-particle $B_s$-meson light-cone distribution amplitudes (LCDAs) of definite collinear twist, in particular the twist-5 LCDA $\\tilde\\phi_5$ in the Exponential Model, whose normalization is set by $\\lambda_2^E+\\lambda_2^H$. Twist labels the degree of suppression in the light-cone expansion, and these functions parametrize the $B_s$-to-vacuum matrix element of the non-local quark-antiquark-gluon operator produced when the charm loop emits a soft gluon. The LCSR in Eq. (28) converts them into the non-local form factors $\\tilde V_i$. The second engine is the double-subtracted hadronic dispersion relation in $q^2$, with the $\\phi$, $J/\\psi$, and $\\psi(2S)$ poles plus a fitted continuum, which carries the spacelike LCSR predictions into the decay region. The numerical cancellation among Lorentz-structure contributions of the same twist is what makes the twist-5 term decisive.","core_discovery":"On its own terms, the paper claims that the non-local form factors $\\tilde V_\\perp$, $\\tilde V_\\parallel$, and $\\tilde V_0$ receive a twist-5-dominated contribution that earlier truncations missed. At the benchmark $q^2=-1\\,\\mathrm{GeV}^2$, the perpendicular combination is $10^7 \\tilde V_\\perp(-1)=1.536|_{\\mathrm{twist}\\,3}-1.235|_{\\mathrm{twist}\\,4}-14.334|_{\\mathrm{twist}\\,5}+0.668|_{\\mathrm{twist}\\,6}$, moving the central value from about $0.3$ (the twist-3 plus twist-4 part) to about $-13.4$ in these units. Within the Exponential Model's complete set of eight three-particle LCDAs, the twist-5 function $\\tilde\\phi_5$ alone contributes $-139.47\\times 10^{-8}$ of the total $-143.32\\times 10^{-8}$ for $\\tilde V_\\perp$ at that point. Continuing these LCSR results into the physical region with a double-subtracted dispersion relation that includes the $\\phi$, $J/\\psi$, and $\\psi(2S)$ resonances, the paper obtains a polarization-dependent, $q^2$-dependent correction $\\Delta C_{9,\\lambda}(q^2)$ that is positive over most of the physical region and larger than the Standard Model prediction without non-factorizable charm loops, yet compatible with it within uncertainties.","pith_inferences":["If this enhancement survives scrutiny, global fits of $b\\to s\\ell^+\\ell^-$ data that were tuned to the older, smaller charm-loop estimates may need to be redone; the enlarged effect could absorb part of the apparent tension with the Standard Model or shift the preferred window for new physics.","The dominance of $\\tilde\\phi_5$ suggests that truncating the twist expansion at an even order (twist-4) is structurally unsafe; a direct test would be to repeat the computation with the alternative Local Duality model of the LCDAs, which the paper mentions but does not present, and check whether the order-of-magnitude enhancement survives the model choice.","A lattice QCD determination of the moments $\\lambda_2^E$ and $\\lambda_2^H$ would be a sharp test, since the paper's sensitivity study shows the prediction can vary by orders of magnitude across the currently allowed parameter regions."],"forward_implications":["The effective $C_9$ shift becomes polarization-dependent and $q^2$-dependent, so $B_s\\to\\phi\\mu^+\\mu^-$ angular observables such as $S_3$ and $S_7$ change most in the low-to-intermediate $q^2$ bins, where current data already show some tension.","Precision on $\\lambda_2^E$ and $\\lambda_2^H$ becomes a bottleneck: across the five allowed parameter regions the non-local form factors vary by up to orders of magnitude, so better determinations of these two constants directly translate into sharper rare-decay predictions.","The local $B_s\\to\\phi$ form factors are affected at the sub-percent level by the higher-twist three-particle LCDAs, so the larger charm-loop effect is not coming from rescaled local inputs.","Because a similar cancellation pattern holds for $B\\to K^*\\bar\\ell\\ell$, the same twist-5 and twist-6 input likely shifts that mode's non-local estimates as well, though with slightly smaller magnitude."],"supporting_citations":[{"why":"provides the previous LCSR estimate of the non-local matrix elements with twist-4 LCDAs that this paper's twist-5/6 result is compared against","marker":"[12]"},{"why":"introduced the light-cone OPE treatment of the charm-loop non-local matrix element that the present calculation extends","marker":"[8]"},{"why":"supplies the complete set of higher-twist B-meson LCDAs in HQET and the equations-of-motion relations used for the twist-5 and twist-6 input","marker":"[14]"},{"why":"gives the NLO matching of Delta C7 and Delta C9 used in the factorizable part of the non-local amplitude and in the dispersion relation","marker":"[23]"},{"why":"provides the LCSR formalism for B->P,V local form factors from B-meson LCDAs that is updated here with twist-5 and twist-6 three-particle terms","marker":"[24]"},{"why":"supplies the QCD sum rule values of lambda2_E and lambda2_H (Region I) that drive the central twist-5 contribution","marker":"[36]"},{"why":"gives the Grozin-Neubert relation and the exponential-model parametrization underlying the sensitivity regions used for lambda2_E and lambda2_H","marker":"[37]"}],"fun_headline_variants":["Twist-5 terms boost Bs->phi l+l- form factors tenfold","Higher-twist LCDAs enlarge C9 shift in Bs->phi l+l-","Twist-5 dominates non-local Bs->phi l+l- form factors","Charm-loop effect upsized by tenfold in Bs->phi l+l-"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole order-of-magnitude result rests on one modeling assumption: the exponential shape and normalization of the twist-5 three-particle distribution amplitude of the $B_s$ meson, whose strength is set by $\\lambda_2^E+\\lambda_2^H$; if that model is wrong, the enhancement could vanish.","fun_headline_variants_meta":{"raw":{"variants":["Twist-5 terms boost Bs->phi l+l- form factors tenfold","Higher-twist LCDAs enlarge C9 shift in Bs->phi l+l-","Twist-5 dominates non-local Bs->phi l+l- form factors","Charm-loop effect upsized by tenfold in Bs->phi l+l-"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000976,"raw_usage":{"total_tokens":4247,"prompt_tokens":1144,"completion_tokens":3103,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":760,"completion_tokens_details":{"reasoning_tokens":3013}},"tokens_in":760,"tokens_out":3103,"duration_ms":24446,"temperature":1.0,"reasoning_tokens":3013,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T05:08:21.412541+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the same non-local form factor with an independent determination of the twist-5 normalization, for example from a lattice QCD calculation of the relevant three-particle correlation function or from alternative QCD sum rules for $\\lambda_2^E$ and $\\lambda_2^H$, and check whether $\\tilde V_\\perp(-1\\,\\mathrm{GeV}^2)$ stays near $-13\\times 10^{-7}$ rather than returning to the old value near $0.3\\times 10^{-7}$. A complementary check is to fit the low-$q^2$ $B_s\\to\\phi\\mu^+\\mu^-$ angular data with and without the enhanced charm loop and see which reproduces the measured $S_3$ and $S_7$ bins.","supporting_citations":[],"review_version":1}