{"id":"511a43e3-7355-4626-81b0-d34a166a95c4","arxiv_id":"2504.17989","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Four-graviton one-loop closed superstring amplitudes in the celestial basis factor out the α' dependence, and the field theory limit commutes with the Mellin transform for all values of the cross-ratio.","lead":"This paper computes, for the first time, the one-loop scattering amplitude of four gravitons in closed superstring theory in the celestial basis, and shows that the string parameter α' factors out cleanly. It also proves that taking the low-energy (field theory) limit before or after converting to celestial variables gives the same result, confirming that a property already known for gluons also holds for gravitons.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The commutativity claim rests on an unproven Imτ→∞ dominance that (4.33) itself contradicts: the Mellin-transformed T-integral is dominated by T=1, not the cusp, and the singular locus of (rX−Y)^(β−4) is unanalyzed.","rationale":"The paper's main claim is a commutativity statement: M[lim_{α′→0} A_str] = lim_{α′→0} M[A_str]. The string-side computation (Section 4.4) reaches (4.37) through two formal steps: (i) exact Mellin integration over ω, giving (rX−Y)^{β−4}; (ii) replacement of the moduli-space integral by the Imτ→∞ asymptotics (4.29)–(4.33). Step (ii) is the load-bearing one. The approximation is only valid as Imτ→∞, but the resulting T-integral ∫_1^∞ dT T^{β−D/2−1} for D=4 and imaginary β is dominated by T=1, where the approximation is not valid. Since the T-integral is α′-independent after Mellin, any finite-T contribution is not suppressed as α′→0 and must either vanish identically—which is not shown—or spoil the limit. The singular locus rX−Y=0 adds a further uncontrolled boundary. The field-theory side (4.42) is derived from the worldline expression (3.12), independent of this moduli approximation, so the equality (4.37)=(4.42) hinges entirely on the unsupported dominance assumption. This is the same weak spot the reader identified, and the conditional verdict is appropriate unless the proposed test shows the finite-moduli contributions vanish.","tokens_in":18638,"tokens_out":19763,"duration_ms":197012,"concrete_test":"For a representative physical configuration (e.g., r=2, λ_i chosen so β=−i), evaluate numerically the exact integral in (4.26) restricted to Imτ∈[1,L] using the full χij of (3.8), and compare with the same restriction of the asymptotic expression (4.33). If the difference Δ(L) does not tend to zero as L grows, or if the contribution from Imτ∈[1,L] is not negligible relative to the large-T tail, then the Imτ→∞ dominance used in (4.35) is false and (4.37)=(4.42) is not established. A simpler analytic first check: verify that the T-integrand in (4.33) peaks at T=1, so the asymptotic region is not the dominant one.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that (4.37) equals (4.42) for all r>1—rests on replacing the moduli-space integral in (4.26) by its Imτ→∞ asymptotics (4.29)–(4.33). This step is not justified, and there is a concrete indication it fails: after the Mellin transform and the change of variables ρ_i=Imν_i/Imτ, T=Imτ, the T-integral in (4.33) is ∫_1^∞ dT T^{β−D/2−1}. For physical β=−iλ/2 and D=4, |T^{β−D/2−1}| = T^{−3}, so the integral is dominated by the lower endpoint T=1—precisely the region where the approximations (4.29)–(4.30) are invalid. Moreover, the integrand (rX−Y)^{β−4} has a branch-type singularity where rX−Y=0; the zero locus is not analyzed, and for imaginary β the real part of β−4 is −4, making such singularities naively non-integrable. The ω-integral identity (4.25) and the subsequent analytic continuation may then pick up boundary terms. If finite-T or singular contributions survive, the α′ absorption via η=(α′T)^{-1} in (4.35) does not capture them, and the equality with (4.42)—obtained from the worldline expression (3.12) without this moduli approximation—is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the four-graviton one-loop amplitude in type II closed superstring theory in the celestial basis. After a Mellin transform in the external energies, the authors obtain an expression in which the α' dependence factorizes as an overall power (α')^{β−5}, Eq. (4.27). They then take the α'→0 limit of the Mellin-transformed amplitude by approximating the genus-one moduli integrand by its Im τ→∞ asymptotics, arriving at Eq. (4.37). This is compared with a direct calculation of the celestial one-loop N=8 supergravity amplitude in the worldline formalism, Eq. (4.42), and the two are claimed to coincide after the distributional identity I(λ−iD) is used to enforce β=D/2. The paper concludes that, at one loop, the field theory limit commutes with the Mellin transform for all r>1, extending earlier open-string gluon results to closed-string gravitons.","tokens_in":18897,"tokens_out":13176,"duration_ms":126636,"significance":"If the main claim were established, this would be a valuable contribution to celestial string amplitudes. The α' factorization is explicit and clean, the organization of the loop expansion in terms of κ_{10}²/(α')⁴ is natural, and the use of the distributional identity from [27] is a good idea. The paper also provides a useful worldline derivation of the celestial one-loop N=8 supergravity amplitude. However, the central commutativity result rests on an unverified cusp-dominance assumption; the concern raised below is load-bearing. With that gap filled, the paper would meet the standard for publication; as it stands the main claim is not fully supported.","major_comments":[{"comment":"The central step of the paper is the replacement of the moduli integral in (4.26) by its Im τ→∞ asymptotics, but this replacement is never justified by a uniform bound on the fundamental domain. The concern is not merely technical: after the change of variables (4.31) and η=(α'T)^{-1}, the remaining T integral in (4.33) is ∫_1^∞ dT T^{β−D/2−1}, whose absolute value for D=4 and β=−iλ/2 is ∫_1^∞ dT T^{−3}; this integral is dominated by the lower endpoint T=1, precisely the region where (4.29)–(4.30) are not valid. Therefore the assertion that the α'→0 limit of the Mellin-transformed amplitude is governed by the cusp is not supported. The authors need to either prove that finite-T contributions vanish or are absorbed into the same distributional identity, or revise the derivation of (4.37).","section":"Section 4.4, Eqs. (4.29)–(4.33) and (4.35)"},{"comment":"The displayed equality between the moduli measure and ∫_1^∞ dT/T² ∫_0^1[dρ] + perm is stated without derivation and is not an identity on the fundamental domain. The fundamental domain is |τ|≥1, so the lower limit on T should be √3/2 after the Re τ integration, not 1; more importantly, the full integrand in (4.26) depends on Re τ and Re ν through the theta functions in (3.8), so the reduction to T and ρ_i is valid only after the cusp approximation has been made. The constants coming from dτ dτ̄ and ∏ d²ν_l are also omitted. This step needs to be made precise, since it is the basis for the α' absorption in (4.35).","section":"Section 4.4, Eq. (4.31)"},{"comment":"The analytic continuation of the ω-integral ignores the zero locus of rX−Y. Since χ in (3.8) is real and positive, rX−Y is real, and its zero locus is a real hypersurface; for β=−iλ/2 the naive power (rX−Y)^{β−4} has modulus |rX−Y|^{−4}, which is not locally integrable across that hypersurface. The paper does not analyze whether this locus contributes to the α'→0 limit or whether boundary terms from the analytic continuation in β survive in (4.33). This is load-bearing because the subsequent T-integral and the identification with (4.42) assume that only the cusp region matters.","section":"Section 4.4, Eq. (4.25)"}],"minor_comments":[{"comment":"There is a typo in the condition 's + t + u+ = 0'; it should read s+t+u=0.","section":"Section 2, after Eq. (2.10)"},{"comment":"References [24] and [29] appear in the bibliography but are not cited in the body of the text; please cite them in the relevant discussion or remove them.","section":"References"},{"comment":"The claim that the field theory limit commutes with the Mellin transform 'for all values of the conformally invariant cross-ratio' should be qualified: the derivation explicitly restricts to r>1 via the Heaviside function after Eq. (4.22) and to the distributional support β=D/2.","section":"Abstract and Section 5"},{"comment":"The lower limit of the T-integration is written as 1, but the fundamental domain (3.4) gives T≥√3/2 after integrating over Re τ; if this replacement is an approximation, it should be stated as such.","section":"Section 4.4, Eq. (4.31)"},{"comment":"The exponent notation such as r^{14−β}/3 is ambiguous without parentheses; the authors may wish to write r^{(14−β)/3}.","section":"Eqs. (4.37) and (4.42)"}],"recommendation":"major_revision","confidential_remarks":"The main result is conditional on a single technical step that is central and currently unjustified. If the authors can supply a rigorous treatment of the finite-T and singular-locus contributions, or if they can show that the distributional identity absorbs them, the paper would be a solid contribution. In its present form I cannot recommend acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take on arXiv:2504.17989. It's a reasonable extension of the celestial-string program to one-loop closed strings, and the main claim—that the field-theory limit commutes with the Mellin transform for all r—is plausible. But the proof has real analytic gaps that need referee attention.\n\nWhat's new: this is the first computation of the four-graviton one-loop closed superstring amplitude in the celestial basis. The α' factorization they find, with the dependence organized as an overall factor (α')^{β-5}, is consistent with earlier open-string work, and the paper makes a clean statement about the loop expansion. The matching between the string expression (4.37) and the N=8 supergravity celestial amplitude (4.42) is a non-trivial check, and having the supergravity result written down is itself useful.\n\nWhere I have concerns: the step from the full moduli integral to the Im τ→∞ asymptotic form is not justified. After the Mellin transform, the α' dependence has been factored out of the integrand, so the standard argument that the cusp dominates no longer applies. In fact, the T integral in (4.33) is dominated by T=1, not T→∞. The asymptotic expansion (4.29)-(4.30) is still accurate at T=1 since q=e^{-2π} is small, so this is not fatal, but the paper never proves that the replacement captures the full distributional limit. The measure equality (4.31) is asserted without derivation. The integrand (rX-Y)^{β-4} has a singular locus that is not analyzed; for imaginary β it is not classically integrable, and the paper relies entirely on the distributional identity from [27] without discussing how the ν-integral is defined. Finally, the field-theory side is computed in the worldline formalism derived from the string, so the cross-check is not fully independent.\n\nThese are fillable gaps, not a smoking gun. I don't see an obvious algebra error, and the final equality is consistent with the previous open-string result.\n\nWho this is for: people working in celestial holography and string amplitudes. It extends a known series of results and gives a concrete example to test against.\n\nRecommendation: send it to peer review. The referee should ask for a careful treatment of the moduli-space limit, including the finite-T contributions and the singular locus. With those fixed, this would be a solid contribution.","headline":"A plausible but not airtight one-loop closed-string celestial amplitude computation; the final commutativity claim is likely right, but the moduli-space asymptotics need real analytic support before publication.","tokens_in":19529,"tokens_out":16581,"would_cite":true,"duration_ms":162872,"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":"At one loop, the field-theory limit of the closed superstring four-graviton amplitude commutes with the Mellin transform and reproduces the celestial N=8 supergravity amplitude for every cross-ratio.","keywords":["celestial holography","closed superstring","one-loop amplitudes","four-graviton scattering","Mellin transform","conformal basis","N=8 supergravity","field theory limit"],"falsifier":"The equality of (4.37) and (4.42) would be falsified by any non-vanishing contribution to $J(r,\\beta)$ from the part of the fundamental domain away from $\\operatorname{Im}\\tau\\to\\infty$ in the limit $\\alpha'\\to0$: for instance, a numerical evaluation of the full torus integral at fixed $r>1$ and fixed $\\beta$ showing that the region near $\\tau=e^{i\\pi/3}$ or near the zero locus $rX-Y=0$ contributes at order $\\alpha'^{\\beta-5}$ or slower. A direct check would be to Mellin-transform the full string amplitude without the $\\operatorname{Im}\\tau\\to\\infty$ approximation and test numerically whether the difference from (4.42) vanishes as $\\alpha'\\to0$ for $r$ close to 1.","tokens_in":18372,"feed_emoji":"🌌","tokens_out":12322,"duration_ms":97484,"temperature":0.7,"pith_summary":"Celestial holography rewrites scattering amplitudes as correlation functions on the celestial sphere using a Mellin transform over the energies of the external particles. This paper asks whether the usual low-energy limit of string theory, $\\alpha' \\to 0$, survives that change of variables for four-graviton scattering at one loop in closed type II superstring theory. The authors show that it does: when the limit is taken by first isolating the region of the torus moduli space where $\\operatorname{Im}\\tau \\to \\infty$, the Mellin transform and the field-theory limit commute. The resulting celestial string amplitude is exactly the celestial one-loop $N=8$ supergravity amplitude, as distributions enforcing $\\beta = D/2$, for all values of the conformal cross-ratio $r=-s/t$. A by-product is the direct computation of one-loop graviton amplitudes in the conformal basis.","feed_headline":"One-loop string gravitons match supergravity in the celestial basis","feed_subtitle":"The right low-energy limit commutes with the Mellin transform for every cross-ratio, matching N=8 supergravity.","key_machinery":"The argument is carried by the reduced integral $J(r,\\beta)$ of equation (4.26), which contains all dependence on the torus moduli. Two logarithmic combinations of worldsheet $\\theta$ functions, $X$ and $Y$ defined in (4.21), package the Mandelstam dependence; in the $\\operatorname{Im}\\tau\\to\\infty$ region they become linear in the rescaled vertex positions $\\rho_i$ and in $T=\\operatorname{Im}\\tau$, reducing the integral to the parameter integral over the simplex. The change of variables $\\eta=\\alpha'\\omega(rX-Y)$ turns the Mellin transform into the Gamma function $\\Gamma(4-\\beta)$ times $(rX-Y)^{\\beta-4}$; then $\\eta=(\\alpha'T)^{-1}$ absorbs every remaining $\\alpha'$ into the upper limit of a Schwinger-parameter integral. The distributional identity $I(x)=4\\pi\\delta(x)$ for real $x$, extended to complex $x$, is the mechanism that finally imposes $\\beta=D/2$.","core_discovery":"The central claim of the paper is the equality of two expressions: equation (4.37), the $\\alpha'\\to0$ limit of the Mellin-transformed one-loop type II closed superstring amplitude, and equation (4.42), the Mellin transform of the one-loop $N=8$ supergravity amplitude. Both are written as the same rational function of $r$ and $1-r$ times the universal conformal prefactor, multiplied by an integral over the simplex $0\\le\\rho_1<\\rho_2<\\rho_3\\le1$ with integrand $[(1-r)\\rho_1\\rho_3 + r\\rho_1\\rho_2 - \\rho_2\\rho_3 - (\\rho_1-\\rho_2)]^{\\beta-4}$, up to permutations. The matching is enforced by the distributional kernel $I(\\lambda-iD)=\\int_0^\\infty d\\eta\\,\\eta^{\\frac{i}{2}\\lambda + \\frac{D}{2} - 1}$, which behaves as a delta function setting $\\beta=D/2$. Because this holds for every $r>1$, the one-loop graviton case does not need the forward-scattering, large-$r$ limit that was required at tree level.","pith_inferences":["Editorial inference: the same mechanism, dominance by $\\operatorname{Im}\\tau\\to\\infty$ plus a rescaling of the modular parameter to a Schwinger parameter, is likely to generalize to higher loops whenever the Mandelstam dependence enters through exponentiated log-ratios with linear large-$T$ scaling; a two-loop graviton computation would test this.","Editorial inference: because the equality is distributional and enforces $\\beta=D/2$, the conformal soft limits of the one-loop graviton amplitude should receive no string corrections in $\\alpha'$; extracting the soft limits from (4.37) would be a concrete check.","Editorial inference: the different behaviour at tree level (large $r$ needed) and one loop (all $r$) raises a puzzle for open-closed duality in the celestial basis, since worldsheet duality would seem to require compatible $r$-dependence in the two descriptions.","Editorial inference: the commutativity result is sensitive to the order of limits; for celestial observables whose Mellin kernel is not a delta function, string corrections might survive the low-energy limit, so the result may be special to this basis and this observable."],"forward_implications":["The $\\alpha'$ dependence of closed-superstring celestial amplitudes factorizes as an overall power: $(\\alpha')^{\\beta-1}$ at tree level and $(\\alpha')^{\\beta-5}$ at one loop, so the loop expansion is organized by the dimensionless ratio $\\kappa_{10}^2/(\\alpha')^4$.","The one-loop celestial string amplitude and the celestial $N=8$ supergravity amplitude coincide for every value of $r=-s/t$, not only in the forward-scattering region $r\\to\\infty$.","The features previously found for open-string gluons, namely factorization of $\\alpha'$ and commutativity of the Mellin transform with the field-theory limit, also hold for closed-string gravitons, marking them as universal properties of string amplitudes in the conformal basis.","The worldline-formalism Schwinger parameters of the field-theory amplitude emerge from the modular parameter $T$ after the rescaling $\\eta=(\\alpha'T)^{-1}$, which explains why the field-theory integration domain is recovered.","The one-loop field-theory graviton amplitude in the conformal basis exhibits double poles in the $\\beta$-plane, in agreement with the general expectation that logarithms of Mandelstam invariants produce such poles."],"supporting_citations":[{"why":"Identifies the tree-level obstruction to taking $\\alpha'\\to0$ in celestial amplitudes and the forward-scattering resolution; the problem this paper re-examines at one loop.","marker":"[4]"},{"why":"Shows for one-loop open-string gluons that the Mellin transform commutes with the field-theory limit and supplies the worldline-formalism comparison used here.","marker":"[8]"},{"why":"Establishes that the one-loop field-theory limit of type II strings is controlled by the $\\operatorname{Im}\\tau\\to\\infty$ region of moduli space, the key approximation of Section 4.4.","marker":"[15]"},{"why":"Provides the one-loop $N=8$ supergravity four-graviton amplitude that the celestial field-theory result must match.","marker":"[13]"},{"why":"Supplies the celestial loop integrals and the analytic structure in the $\\beta$-plane used to evaluate the Mellin transform.","marker":"[19]"},{"why":"Accounts for the double poles in the $\\beta$-plane produced by logarithms of Mandelstam invariants at one loop.","marker":"[20]"},{"why":"Gives the distributional interpretation of $I(\\lambda-iD)$ for non-real argument, which justifies treating the kernel as a delta function enforcing $\\beta=D/2$.","marker":"[27]"}],"fun_headline_variants":["Celestial basis: one-loop closed strings equal supergravity","Mellin and low-energy limits commute in one-loop gravity","Closed string gravitons match supergravity at one loop","One-loop closed strings: field theory limit commutes with Mellin","Celestial amplitudes: one-loop gravitons agree with N=8 supergravity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculation assumes that, in the $\\alpha'\\to0$ limit, the Mellin-transformed one-loop torus integral is dominated uniformly by the $\\operatorname{Im}\\tau\\to\\infty$ region of moduli space for every $r>1$, and that no other region, in particular the locus where $rX-Y=0$, contributes.","fun_headline_variants_meta":{"raw":{"variants":["Celestial basis: one-loop closed strings equal supergravity","Mellin and low-energy limits commute in one-loop gravity","Closed string gravitons match supergravity at one loop","One-loop closed strings: field theory limit commutes with Mellin","Celestial amplitudes: one-loop gravitons agree with N=8 supergravity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000197,"raw_usage":{"total_tokens":1382,"prompt_tokens":983,"completion_tokens":399,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":599,"completion_tokens_details":{"reasoning_tokens":310}},"tokens_in":599,"tokens_out":399,"duration_ms":4015,"temperature":1.0,"reasoning_tokens":310,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T10:28:36.171467+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The equality of (4.37) and (4.42) would be falsified by any non-vanishing contribution to $J(r,\\beta)$ from the part of the fundamental domain away from $\\operatorname{Im}\\tau\\to\\infty$ in the limit $\\alpha'\\to0$: for instance, a numerical evaluation of the full torus integral at fixed $r>1$ and fixed $\\beta$ showing that the region near $\\tau=e^{i\\pi/3}$ or near the zero locus $rX-Y=0$ contributes at order $\\alpha'^{\\beta-5}$ or slower. A direct check would be to Mellin-transform the full string amplitude without the $\\operatorname{Im}\\tau\\to\\infty$ approximation and test numerically whether the difference from (4.42) vanishes as $\\alpha'\\to0$ for $r$ close to 1.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes that the one-loop field-theory limit of type II strings is controlled by the $\\operatorname{Im}\\tau\\to\\infty$ region of moduli space, the key approximation of Section 4.4."}],"review_version":1}