{"id":"2fdd4802-5d5e-4f8c-9bdd-d08ef8dfcbae","arxiv_id":"2411.14422","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Causality and unitarity force J≥4 resonances to appear below Λcaus approximately sqrt(MP Fπ/κg) in any EFT with a U(1)-gravitational anomaly.","lead":"This paper uses causality and unitarity of graviton scattering to derive a new universal cutoff in theories with a chiral-gravitational anomaly: states with spin 4 or higher must appear below roughly the square root of the Planck mass times the pseudo-Goldstone decay constant. It shows dynamical gravity is essential for the bound and derives consequences for axion models and holographic 5D gauge theories.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central bound Eq. (38) rests on positivity of the smeared kernels, Eqs. (30)-(31), for all intermediate J and masses; the paper supplies only finite-J numerical spot checks and a J→∞ proof, so a counterexample in the unproven region would invalidate the main result.","rationale":"Read in good faith, the paper's central result is the universal cutoff Λcaus ∼ sqrt(MP Fπ/κg), derived from Eq. (38) via smeared dispersion relations. The most load-bearing step is not the existence of the dispersion relations themselves, but the positivity inequalities (30)-(31), because they are what convert the smeared sum rules into an upper bound. If the chosen functions A and B fail to satisfy those inequalities for some intermediate spin or mass, the comparison in Eq. (32) can have either sign and no bound on κg follows. The paper's own evidence is a figure for J ≲ 14 at threshold masses, plus a large-J asymptotic argument in Appendix A; the asymptotic argument relies on f(M)=0 and f'(0)>0, which is sufficient at J→∞ but says nothing about the intermediate-J region where the u-channel integrand is sign-changing. This is an internal-completeness gap rather than a disagreement with external consensus. The tree-level spectral assumption and the MIR dependence are real limitations, but they are acknowledged and do not affect the parametric scale; the positivity gap is the one place where the proof could fail silently. I therefore agree with the reader's weakest-assumption identification and see no reason to move the CONDITIONAL verdict: the suggested numerical audit would either close the gap or force a repair of the derivation.","tokens_in":24192,"tokens_out":17288,"duration_ms":174948,"concrete_test":"Evaluate the integrals Au(m,J) − Bu(m,J) in Eq. (29) using the smearing functions A,B of Eq. (35) with |t|max = Mu^2, on a grid m/Mu ∈ [1, 1000] (log-spaced, including m = Mu) and J ∈ {4,5,...,200}, with high-precision quadrature and the explicit Wigner d^J_44. Focus first on m = Mu, J = 4, where the integrand changes sign at |t| = 0.115 Mu^2. Publish the code and resulting table, or supply an analytic monotonicity argument covering all intermediate J. Any negative value means Eq. (38) does not follow as stated; a clean nonnegative result would close the main loophole.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The bound Eq. (38) is obtained by termwise comparison of the smeared k=2 and k=3 sum rules, which requires the inequalities (30) and (31), As,u ≥ Bs,u, to hold for every exchanged state, i.e. for all masses above threshold and all allowed J. For the smearing functions (35), the s-channel condition reduces essentially to positivity of As, but the u-channel integrand at m = Mu is proportional to A(t) d^J_44 [0.115 Mu^2 − |t|] / (Mu^2 − |t|)^4, whose sign changes at |t| = 0.115 Mu^2. Positivity of Au − Bu is therefore a delicate weighted average of a sign-changing integrand against an oscillating Wigner d-function, not a consequence of pointwise positivity. The paper checks this numerically for the lowest mass and moderate J, and proves only the J→∞ limit in Appendix A; no complete proof or code is provided for intermediate J. A single state with J ≥ 4 and m ≥ Mu for which Au < Bu invalidates Eq. (32) and, with it, the central upper bound on M_u and Λcaus. This is the most load-bearing link in the argument: all phenomenological claims in Section 7 inherit it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies causality and unitarity constraints on 2→2 amplitudes involving gravitons and a pseudo-Goldstone η with a U(1)-gravitational anomaly coupling κg η h+ h-. It first shows that with gravitons as non-dynamical sources the anomaly coefficient cannot be bounded (Sec. 5). It then uses smeared dispersion relations for h+h+→h+h+ and ηh+→ηh+ scattering with dynamical gravitons to derive an upper bound, Eq. (38), on M_u, the mass of the lightest J≥4 state coupled to two gravitons, and a similar bound, Eq. (48), for J≥2 states in η-graviton scattering. Parametrically the cutoff is Λ_caus ∼ √(M_P F_π/κ_g) (Eq. 50). Applications are discussed for axion EFTs, large-N gauge theories, and 5D holographic models, where the bound becomes a consistency condition on 5D parameters.","tokens_in":24421,"tokens_out":14986,"duration_ms":148242,"significance":"If correct, this is a significant result: it turns the chiral-gravitational anomaly into a quantitative lower bound on the mass scale of higher-spin states and gives a causality cutoff below the Planck scale for axion-like EFTs, with concrete, falsifiable parametric predictions. The paper is explicit about the mass scales involved and makes a clear distinction between the non-dynamical and dynamical graviton cases. The main proof strategy is standard (smeared dispersion relations), and Appendix A provides a useful large-J asymptotic analysis. However, the central claim inherits several positivity assumptions that are not fully demonstrated; the result should be viewed as a promising derivation whose rigor depends on closing this gap.","major_comments":[{"comment":"The central inequality Eq. (32), and hence the main bound Eq. (38), requires As≥Bs and Au≥Bu for every exchanged state, i.e. for all allowed J and for all masses above threshold. The paper states that these inequalities have been \"checked explicitly ... for general mi and Ji\", but it then says that for larger Ji the values are \"difficult to get numerically\", and Appendix A proves only the J→∞ limit for the case m=M. Since the u-channel integrand is a sign-changing weighted average of d^J_4,4 against the smearing function, positivity is not a consequence of pointwise positivity; a counterexample at intermediate J or for m>M_u would invalidate Eq. (32) and with it the main bound. A complete analytic proof or a reproducible numerical verification (code, high-resolution scans, or explicit tables) for all intermediate J and all m≥M_s,u is needed to support the central claim.","section":"Sec. 6.1, Eqs. (29)-(31), Fig. 4"},{"comment":"The bounds Eq. (48) and Eq. (49) depend on positivity of Cs,u and on the coupled inequalities Ds,u ≤ √(Cs,u As,u) for all masses and allowed J. The text says these are \"simple to verify\" for the functions in Eq. (108), but no proof or numerical evidence is presented. Because these inequalities again involve oscillating Wigner d-functions with |t|max equal to the threshold mass, the same gap as in Eqs. (30)-(31) occurs. The authors should provide the verification or explicitly list these as assumptions of the derivation.","section":"Sec. 6.2 and Appendix C, Eqs. (44), (104)-(105)"},{"comment":"The bound contains the IR regulator M_IR through ln(M_u/M_IR), and the paper assumes that M_IR can be chosen close to M_u so that the logarithm is O(1). This choice is not derived from any physical input, and the coefficient 7 in Eq. (38) depends on it. If the paper claims a \"rigorously and precisely derived\" bound (Introduction), the status of M_IR should be clarified: either state the bound as M_u^2|κg|/(M_P F_π) ≲ 7√ln(M_u/M_IR) with M_IR an open parameter, or explain how M_IR is fixed. The parametric scale Eq. (50) is unaffected, but the precise statement of the bound is not.","section":"Sec. 6.1, Eqs. (37)-(38) and Eq. (50)"}],"minor_comments":[{"comment":"The figure lacks axis labels and panel labels; please specify which panel corresponds to As/Bs and which to Au/Bu, and whether the plotted quantity is A−B or A and B separately.","section":"Fig. 4"},{"comment":"The notation M_{++--} is used for the elastic h+h+→h+h+ amplitude in Sec. 6, while the same kind of helicity label appears for the inelastic process in Sec. 5; please define the all-incoming helicity labeling once and use it consistently throughout.","section":"Secs. 5-6"},{"comment":"The heading \"Attemping to bound κg without dynamical gravitons\" contains a typo: it should be \"Attempting\".","section":"Section 5 heading"},{"comment":"The phrase \"both integrants have the same high-energy behaviour\" should read \"both integrands\".","section":"Sec. 6.2"},{"comment":"The parametric formula Eq. (1) omits the logarithmic factor that appears in Eqs. (38) and (48); please state explicitly that Eq. (1) is the parametric form obtained by taking the logarithm to be O(1).","section":"Eqs. (1), (38), (48)"},{"comment":"The argument that the middle integral vanishes because a and b are zeros of the Wigner d-function needs more detail; please specify how a and b scale with J and why the endpoint contributions dominate uniformly in that limit.","section":"Appendix A, after Eq. (89)"}],"recommendation":"major_revision","confidential_remarks":"The paper is well within scope for a hep-th journal and the parametric result is likely to be influential. The main obstacle is the unverified positivity of the smeared kernel inequalities, which are load-bearing for the central claim; supplying a complete proof or reproducible numerical verification would make the paper publishable. I do not see a circularity problem: the bound takes κg as input and constrains the spectrum."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you read it. The headline bound, Eq. (38) and its parametric form Lambda_caus ~ sqrt(M_P F_pi / kappa_g), is not new in parametric content: [32] already got the same scale from time delay. What this paper adds is a dispersion-relation derivation of that scale, a clean negative result in Sec. 5 (no bound on kappa_g when gravitons are treated as external probes), and explicit consistency conditions for 5D holographic models, Eqs. (77)-(78) and (83). Those are real additions, and the 5D applications in Sec. 7 are worked out carefully.\n\nThe paper does several things well. It is honest about what is assumed: tree-level spectral decomposition, large-Nc style resonance sums, and a logarithmic IR regulator M_IR whose arbitrariness weakens the numerical coefficients but not the parametric scale. The smearing setup follows the established Caron-Huot-Mazac-Rastelli-Simmons-Duffin technology, and the authors correctly note that the previous time-delay result supports their parametric claim rather than competing with it. The negative result of Sec. 5 is a useful conceptual point: external gravitons cannot bound kappa_g, and dynamical gravity is essential.\n\nThe soft spot is real. The central inequality chain (30)-(31) needs A_s,u >= B_s,u for every exchanged state, all J >= 4 and all masses above threshold. The paper proves the J -> infinity limit analytically (Appendix A) and shows a figure at the lowest mass for moderate J, but for the intermediate-J region it relies on 'we have checked explicitly' without showing the check or supplying code. The u-channel integrand is a weighted average of a sign-changing expression against an oscillating Wigner d-function, so this is not a trivial pointwise inequality. A counterexample at some intermediate J and mass would invalidate Eq. (32) and with it the main bound. The tree-level/no-loops assumption is also a genuine limitation, though standard in this field and explicitly flagged by the authors.\n\nI don't think the paper is wrong; I think the main claim is not yet fully demonstrated. For a reader who takes the positivity conditions as a plausible conjecture, the paper is a solid contribution. It deserves a serious referee, and I would send it to review. The main request to the authors should be: either give a complete proof of (30)-(31) for all J and m, or make the numerical verification reproducible (code/scripts with a dense grid), or soften the presentation from 'bound' to 'bound assuming the numerically supported positivity conjecture.' I'd bring it to the reading group.","headline":"A serious, well-written bootstrap paper whose central bound is conditional on an unproven positivity condition in the intermediate-spin region; the parametric result is known from time-delay, but the derivation and model implications are worth referee time.","tokens_in":25029,"tokens_out":3554,"would_cite":true,"duration_ms":34276,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A U(1)-gravitational anomaly forces new spin-4 states below a universal cutoff in axion-type EFTs.","keywords":["chiral-gravitational anomaly","graviton scattering amplitudes","dispersion relations","smearing","causality bounds","axion effective field theory","large-Nc gauge theories","holographic models"],"falsifier":"A numerical scan that computes the smeared integrals in Eq. (29) for the functions Eq. (35) over the full grid of masses $m^2\\geq M_u^2$ and spins $J\\geq4$ and finds one point with $A_u(m,J)<B_u(m,J)$ would falsify the derivation, since the positivity inequality (31) is the step that converts the sum rule into the bound. Alternatively, a concrete UV model with a $U(1)$-gravitational anomaly that satisfies the smearing inequalities yet has no $J\\geq4$ state below $\\sqrt{M_P F_\\pi/\\kappa_g}$ would contradict the paper's central claim.","tokens_in":23918,"feed_emoji":"🌌","tokens_out":10169,"duration_ms":85533,"temperature":0.7,"pith_summary":"The paper tries to prove that every effective field theory containing a pseudo-Goldstone $\\eta$ with a $U(1)$-gravitational anomaly — for example an axion EFT — has a finite causal cutoff $\\Lambda_{\\rm caus}\\sim\\sqrt{M_P F_\\pi/\\kappa_g}$ below which new states of spin 4 or higher must appear. It first shows that the anomaly coefficient cannot be bounded while gravitons are treated as external, non-dynamical probes, making dynamical gravitons essential. The proof uses smeared dispersion relations in graviton-graviton and eta-graviton scattering to convert the graviton's $1/t$ pole into a positive-definite sum rule, yielding explicit bounds on the mass of the lightest such states. If correct, the result gives a universal scale below which axion EFTs break down, explains why large-$N_c$ gauge theories with glueballs evade the bound in a certain limit, and forces new consistency conditions on holographic 5D models.","feed_headline":"Axion EFTs must break down below sqrt(M_P f_a)","feed_subtitle":"Causality in graviton scattering forces spin-4 resonances below this universal scale in any theory with a U(1)-gravitational anomaly.","key_machinery":"The load-bearing object is the smeared dispersion relation for $2\\to2$ helicity amplitudes at fixed $t$. Analyticity and unitarity give sum rules in which the imaginary part is a sum over exchanged states weighted by Wigner $d$-functions, but the graviton $t$-channel pole introduces a $1/t$ singularity that forbids the usual $t\\to0$ limit and makes the $d$-functions sign-indefinite. Smearing functions $A(t)$, $B(t)$, $C(t)$ — chosen with specific zeros and boundary behaviour, Eqs. (35) and (46) — are integrated over $t\\in[-M_u^2,0]$ so that every state's contribution becomes positive and the $1/t$ pole is matched to the low-energy anomaly terms. The whole argument thereby reduces to verifying the positivity inequalities (30)–(31) and (44), which the paper checks numerically for sample spins and analytically at large $J$ through the Darboux form of the $d$-functions.","core_discovery":"On its own terms, the paper establishes that a $U(1)$-gravitational anomaly forces dynamical gravity into the problem and cannot be bootstrapped with gravitons as external sources. From smeared dispersion relations in $h^+h^+\\to h^+h^+$ and $\\eta h^+\\to\\eta h^+$ scattering it derives explicit bounds — Eq. (38), parametrically $\\Lambda_{\\rm caus}\\sim\\sqrt{M_P F_\\pi/\\kappa_g}$ — on the mass $M_u$ of the lightest $J\\geq4$ state coupled to two gravitons, and Eq. (48) for $J\\geq2$ states in $\\eta$–$h$ scattering. It then shows that in axion EFTs this is a genuine cutoff below the naive perturbative scale, that in large-$N_c$ gauge theories with $N_F\\ll N_c$ and a non-large 'tHooft coupling the presence of glueballs can push the quantum gravity scale below $\\Lambda_{\\rm caus}$ and evade the bound, and that in holographic 5D models $\\Lambda_{\\rm caus}$ becomes a new scale whose consistency imposes conditions such as Eq. (78) and Eq. (83).","pith_inferences":["(Editorial extension) The same smearing-dispersion logic should apply to mixed anomalies involving other Goldstones, so the generic scale $\\sqrt{M_P F/\\kappa}$ is likely a feature of any EFT with a Goldstone–graviton–graviton vertex, not only the $U(1)$ case.","(Editorial extension) If the bound holds, axion EFTs used in gravitational-wave or black-hole computations inherit this cutoff, which suggests that experimental searches should consider signatures of $J\\geq4$ partners near $\\sqrt{M_P f_a}$, even though their couplings are Planck-suppressed.","(Editorial extension) The flat-space 5D consistency condition Eq. (83) departs from the naive large-$N_c$ scaling in the paper; checking whether other string or 5D constructions satisfy it would either confirm the bound or reveal where the positivity assumptions fail."],"forward_implications":["Every EFT of $\\eta$ plus gravitons with $\\kappa_g\\neq0$ has a cutoff $\\Lambda_{\\rm caus}\\sim\\sqrt{M_P F_\\pi/\\kappa_g}$, so at least one $J\\geq4$ state coupled to two gravitons must exist at or below this scale.","For axion models ($F_\\pi/\\kappa_g \\sim f_a$) the axion EFT cannot be extrapolated beyond $\\sqrt{M_P f_a}$, a scale below the naive perturbative cutoff.","In large-$N_c$ gauge theories with $N_F\\ll N_c$ and a non-large 'tHooft coupling, glueballs shift the quantum gravity scale below $\\Lambda_{\\rm caus}$, so no bound on mesonic $J\\geq4$ states follows.","In holographic 5D duals, $\\Lambda_{\\rm caus}$ is a new cutoff unless the 5D parameters satisfy conditions such as Eq. (78) for warped models and Eq. (83) for flat models; the D3/D7 string embedding has string states below the bound.","If the lightest glueball is the dilaton, the bound constrains its coupling to two gravitons and hence $(a_{\\rm UV}-c_{\\rm UV})/N_c^2$ when higher-spin states are heavy."],"supporting_citations":[{"why":"Introduces the smearing technique that integrates sum rules over t to restore positivity in the presence of the graviton 1/t pole.","marker":"[21]"},{"why":"Provides the graviton-scattering dispersion relations and eikonal time-delay bound that the paper adapts to include the anomaly coefficient.","marker":"[22]"},{"why":"The chiral anomaly bootstrap whose spin/parity classification and sum-rule machinery are extended here from U(1)-gauge anomalies to U(1)-gravitational anomalies.","marker":"[18]"},{"why":"Source of the causality/time-delay constraints and the logarithmic IR divergence that appears in the smeared bound.","marker":"[31]"},{"why":"Previously derived the parametric scale sqrt(M_P F_pi/kappa_g) from time-delay arguments, which the paper recovers from dispersion relations.","marker":"[32]"},{"why":"Justifies treating Lambda_QG as a rigorous quantum-gravity cutoff, used to compare with Lambda_caus in large-Nc and holographic models.","marker":"[43]"},{"why":"Gives the dilaton-graviton coupling kappa_dilaton = sqrt(2)(a_UV - c_UV) used to translate the bound on a light glueball into a constraint on trace anomaly coefficients.","marker":"[48]"},{"why":"The D3/D7 string model used as an explicit holographic example where string states appear below the causality bound.","marker":"[54]"}],"fun_headline_variants":[],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The derivation depends on the claim, verified numerically only for sample values and analytically only in the $J\\to\\infty$ limit, that the smearing functions in Eq. (35) and Eq. (46) satisfy the positivity inequalities $A_s\\geq B_s$ and $A_u\\geq B_u$ for every allowed spin and mass; a counterexample at intermediate $J$ would void the bound.","fun_headline_variants_meta":{"raw":{"variants":[]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000318,"raw_usage":{"total_tokens":1829,"prompt_tokens":1009,"completion_tokens":820,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":625,"completion_tokens_details":{"reasoning_tokens":703}},"tokens_in":625,"tokens_out":820,"duration_ms":6919,"temperature":1.0,"reasoning_tokens":703,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:11:56.858934+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A numerical scan that computes the smeared integrals in Eq. (29) for the functions Eq. (35) over the full grid of masses $m^2\\geq M_u^2$ and spins $J\\geq4$ and finds one point with $A_u(m,J)<B_u(m,J)$ would falsify the derivation, since the positivity inequality (31) is the step that converts the sum rule into the bound. Alternatively, a concrete UV model with a $U(1)$-gravitational anomaly that satisfies the smearing inequalities yet has no $J\\geq4$ state below $\\sqrt{M_P F_\\pi/\\kappa_g}$ would contradict the paper's central claim.","supporting_citations":[{"cited_title":"Serra, J","cited_arxiv_id":null,"evidence_quote":"Previously derived the parametric scale sqrt(M_P F_pi/kappa_g) from time-delay arguments, which the paper recovers from dispersion relations."}],"review_version":1}