{"id":"4502f555-3ee0-4108-8e4b-baebc16ed2b6","arxiv_id":"2504.16903","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The sub-subleading soft graviton theorem gains collinear correction terms, derived from distributional spinor derivative identities and shown to agree with the collinear part of the asymptotic charges found by Freidel, Pranzetti, and Raclariu.","lead":"This paper studies formulas for particle collisions in gravity when one particle becomes nearly zero-energy, and it finds that the standard formula needs a correction whenever two other particles travel in exactly the same direction. The correction matches a piece of the spacetime symmetry structure found in a separate line of work, bringing the two approaches into agreement.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (6.2)'s correction sum omits the {1,2} pair via the b>2 restriction, while Eq. (5.9) sums over all k≠a; for generic tree amplitudes with a 1/<12> pole this breaks relabeling invariance, so the universal corrected soft theorem as stated is incomplete.","rationale":"The reader's verdict was CONDITIONAL, with the weakest assumption identified as the Section 3 distributional identity and the interpretive choice to subtract delta terms. My stress-test focuses on a different, more concrete flaw: the final formula (6.2) has a summation range that is not permutation invariant. The paper's own general-MHV derivation (5.9) sums over all k≠a, while (6.2) restricts to b>2;b>a, excluding the pair (1,2). In the five-point MHV example this restriction is harmless because the amplitude has no 1/<12> pole; for generic tree-level amplitudes, such a pole is generically present, so the correction term should include δ(1,2). The paper never explains why this pair is singled out, and the claimed extension to all tree-level amplitudes in Section 6 is explicitly an expectation based on factorization, not a proof. This internal inconsistency reinforces the CONDITIONAL verdict and adds a concrete, checkable condition: the summation in (6.2) must be amended to cover all collinear pairs, or a justification for the exclusion of {1,2} must be supplied. I am not moving the verdict to REJECT because the underlying mechanism—distributional spinor derivatives generating collinear corrections—may well be correct once the sum is fixed; the issue is in the formulation of the universal claim, not in the core idea. Hence 'UNCHANGED' (still CONDITIONAL), with the specific revision requirement articulated in the concrete test.","tokens_in":21385,"tokens_out":32641,"duration_ms":280046,"concrete_test":"Check relabeling invariance of (6.2) on a concrete non-MHV amplitude, e.g. the six-point NMHV gravity amplitude with helicities (--++++). Compute the coefficient of δ(1,2) in (1/2)Σ_a[s a]/<s a>D^2_a M_6 using any explicit NMHV formula (e.g. a Hodges-type expansion); if nonzero, (6.2)'s b>2 restriction misses a required correction. As a lighter check, re-evaluate the five-point MHV example by isolating the separate a=1 and a=3 contributions to the δ(1,3) coefficient; if the two contributions each match half of Eq. (4.10), then the ordered sum (5.9) double-counts and the 1/2 in (6.2) must be reconciled with the unrestricted sum.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline formula (6.2) subtracts a second term with the double restriction b>2 and b>a, i.e. only unordered pairs {a,b} with b≥3 are included. This restriction appears in the five-point MHV example (4.10)-(4.12), where the amplitude has no 1/<12> pole and hence no δ(1,2) distributional term. But the general-MHV derivation in §5.2, Eq. (5.9), sums over all k≠a without any b>2 condition, and §6.1, Eq. (6.1), likewise shifts by a sum over all k≠a. For a generic non-MHV tree amplitude the 1/<12> pole is present, so acting with D^2_1 (or D^2_2) on that pole generates a non-vanishing δ(1,2) term of the form (1/2)[s1]/<s1>[s2]^2 f^{h_1 h_2}_{h_P}(t) δ(1,2) M_{n-1}. Dropping it is not justified and makes Eq. (6.2) not invariant under permutations of the hard-particle labels. Permuting 1↔3 changes the set of included pairs, while the left-hand side (the physical amplitude) transforms covariantly; for the equality to hold the two expressions would have to differ by an identity that singles out the pair {1,2}, which is not shown and is false for a generic pole configuration. This is therefore not just a cosmetic issue: the central theorem as stated cannot hold for all tree-level amplitudes unless the sum is symmetrized over all unordered pairs (or the factor 1/2 and the ordered sum in (5.9) are consistently adjusted). This gap directly affects the consistency check with [18] in §7 and must be resolved before (6.2) can be accepted as the universal corrected theorem.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper argues that the sub-subleading soft graviton theorem of Cachazo and Strominger must be corrected by collinear/distributional terms. Starting from the identity (3.3), the authors note that spinor derivatives acting on angle-bracket poles of amplitudes generate delta-function-supported terms, and they propose to subtract these from the soft factor. The general corrected formula is Eq. (1.6)/(6.2). The paper verifies the prescription in detail for the five-point MHV amplitude, gives a general MHV recombination argument in Section 5, argues heuristically in Section 6 that the result extends to all tree-level amplitudes, and identifies the correction with the collinear part of the asymptotic charge found in [18].","tokens_in":21689,"tokens_out":13449,"duration_ms":131431,"significance":"If the corrected theorem is valid, this is a significant result: it provides an amplitude-side derivation of the collinear charge component of [18], clarifies the status of the Cachazo-Strominger soft theorem at sub-subleading order, and gives a concrete example where distributional terms in the soft expansion cannot be dropped. The five-point MHV example is a genuine, fully worked check, including the exact ϵ=1 sum in Appendix B and comparison with [21]. The comparison with [18] is an external consistency check, not a circular use of the target result, and no parameters are fitted. At the same time, the universal claim in Eq. (6.2) currently has a load-bearing summation problem and the all-tree-level extension is only sketched, so the significance would be fully realized only after those issues are resolved.","major_comments":[{"comment":"The correction sum is restricted by 'b>2; b>a', so the only unordered pair excluded is {1,2}. This restriction is not derived anywhere: the general MHV formula (5.9) sums over all k≠a with no b>2 condition, and the general subtraction formula (6.1) also sums over all k≠a. The five-point MHV example used to motivate (4.12) has no 1/<12> pole, so it cannot justify dropping δ(1,2) terms. For a generic tree amplitude with a 1/<12> pole, D^2_1 (or D^2_2) acting on that pole produces a nonzero distributional term, and omitting it makes the right-hand side of (6.2) depend on which leg is labeled 1; interchanging labels 1 and 3 changes the set of subtracted pairs while the left-hand side transforms covariantly. The stress-test concern therefore lands: the universal corrected theorem as stated is not permutation invariant. The authors must either prove that all δ(1,2) terms vanish for arbitrary helicities, or symmetrize the correction sum over all unordered pairs and explain the relation between the ordered sum in (5.9) and the restricted sum in (6.2). This issue also affects the consistency check in Section 7, which relies on (6.2).","section":"§6.2, Eq. (6.2) (and Eq. (1.6))"},{"comment":"The recombination of distributional terms in the general MHV case is asserted rather than demonstrated. Eq. (5.8) contains single delta functions, derivatives of delta functions, and products of delta functions; the text states that the products vanish and the derivative terms combine with the single-delta terms, but the general-n computation is not shown. Because this recombination is the only derivation of the correction term for MHV amplitudes, it should be presented in full, or replaced by a cleaner argument, before it is imported into Eq. (6.2).","section":"§5.2, Eq. (5.9)"},{"comment":"The extension from MHV to all tree-level amplitudes is an expectation argument, not a proof. The text says 'one expects the result to carry over' and appeals to collinear factorization, but the soft derivative D^2_a acts before the collinear limit is taken, and the commutativity of these operations is precisely what needs to be established. If the theorem is claimed for all tree-level graviton amplitudes, a derivation for non-MHV helicity configurations must be supplied, or the claim should be restricted to MHV amplitudes.","section":"§6, 'General Tree-level Graviton Amplitudes'"}],"minor_comments":[{"comment":"The text first sets ∂_{\\tilde\\lambda_a}\\langle ab\\rangle^{-1}=0 in going from (2.9) to (2.10) and then reintroduces it as a distributional term in Section 3; this should be acknowledged explicitly as a choice of how the compact form is defined, to avoid the appearance of inconsistency.","section":"Footnote 5"},{"comment":"The two-line computation of D^2_a([ab]/\\langle ab\\rangle) omits the intermediate derivative-of-delta terms that the general discussion says must be kept; expanding one line or adding a sentence explaining the cancellation would make the example easier to follow.","section":"Eq. (4.11)"},{"comment":"The transition from (7.2) to (7.3) is very compressed, especially the replacement of the summand by f^{h_a h_b}_{h_P}(t)δ(s,b)δ(a,b) after four derivatives; a schematic derivation or a precise reference to the corresponding step in [18] should be given.","section":"Section 7, Eq. (7.3)"},{"comment":"The notation ∂δ(a,b) is used without definition; it should be specified as the derivative of the delta function with respect to the relevant holomorphic coordinate.","section":"Appendix C"},{"comment":"Only three helicity combinations of the graviton splitting functions are listed; the (a^-,b^-) combination relevant for the possible δ(1,2) terms in repeated-negative-helicity MHV configurations is not given, so the claimed vanishing of those terms in that sector cannot be checked from the text.","section":"Eq. (2.13)"},{"comment":"The symbol n denotes both the number of hard particles and the reference leg in the BCFW sum; this double use is confusing in Section 5, where n is the hard multiplicity while the reference leg is 4 or 5.","section":"Section 2, Eqs. (2.8)-(2.10)"}],"recommendation":"major_revision","confidential_remarks":"To the editor: the b>2 restriction in Eq. (6.2) is the main correctness issue. If it is a typo and the intended sum is over all unordered pairs, the paper is close to publishable after a revised Section 5.2 and a real Section 6 proof; if it is a substantive restriction, the universal claim as stated does not hold."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The mechanism is sound: the identity in Eq. (3.3) produces delta-function terms when the sub-subleading soft differential operator acts on angle-bracket poles of the amplitude, and once you decide to keep those terms instead of dropping them, you get collinear corrections to the Cachazo-Strominger theorem. The five-point MHV example is genuinely worked through, with enough detail in the appendices to reconstruct it, and the correction matches the collinear charge of [18]. That is a real, checkable result, and new as far as I can tell. The paper also tells you honestly what it has not proven: Section 8 leaves out the divergent two-soft-graviton piece, and Section 6 frames the all-tree-level extension as an expectation.\n\nThe soft spot is not minor. The universal formula (6.2) restricts the correction sum to b>2 and b>a, which excludes exactly the pair {1,2}. In the five-point MHV example that restriction is harmless because particles 1 and 2 are the negative-helicity lines and the MHV amplitude vanishes in that collinear limit. But the general MHV derivation, Eq. (5.9), sums over all k≠a, and the shift prescription in Eq. (6.1) also sums over all k≠a. Nothing in the paper justifies singling out particle 2. For a generic non-MHV amplitude there is no reason for the 1/<12> pole to be absent, the δ(1,2) term is non-vanishing, and dropping it breaks relabeling invariance: permute 2 and 3 and the excluded pair changes while the physical amplitude just permutes. So (6.2) cannot be the universal corrected theorem as stated. The fix, symmetrizing over all unordered pairs, is straightforward in principle, but the consistency check with [18] in Section 7 must be redone after it.\n\nTwo smaller gaps. The recombination into (5.9) for general MHV is asserted, not shown: the derivative-of-delta terms and the claimed vanishing of the D_a F_{n-1} piece need a real appendix. And the Section 3 prescription — keep the deltas, subtract them — is a convention whose only external anchor is consistency with [18]; the five-point example supports the choice but does not fully remove that circularity.\n\nBottom line: this is for people in soft theorems and celestial holography, and it deserves a serious referee. The referee should send it back for a corrected (6.2) over all pairs, a fuller derivation of (5.9), and a clean recheck of Section 7. The MHV-level content and the mechanism will likely survive and get cited either way.","headline":"The distributional-correction mechanism is sound and the five-point MHV check is solid, but the universal formula (6.2) drops the {1,2} pair via an unjustified b>2 restriction, so the all-tree-level claim fails as stated.","tokens_in":22345,"tokens_out":23462,"would_cite":true,"duration_ms":201326,"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":"The paper establishes that the sub-subleading soft graviton theorem must include collinear correction terms.","keywords":["soft theorems","graviton amplitudes","spinor helicity","collinear limits","MHV amplitudes","asymptotic symmetries","celestial holography","distributional identities"],"falsifier":"Take an explicit five-point MHV graviton amplitude, apply $\\frac{1}{2}\\sum_a [sa]\\langle sa\\rangle^{-1}D_a^2$ to the four-point amplitude, and collect the coefficient of each collinear delta function $\\delta(a,b)$ after imposing momentum conservation. If that coefficient is not $\\frac{1}{2}[sb]^3\\langle sb\\rangle^{-1} f^{h_a h_b}_{h_P}(t)$ times the appropriate three-point amplitude, the corrected soft theorem (1.6) fails; conversely, a direct canonical charge computation yielding a different coefficient of the collinear delta term after the four $\\bar z$-derivatives of (7.3) would falsify the claimed match.","tokens_in":21015,"feed_emoji":"⚛️","tokens_out":7890,"duration_ms":75155,"temperature":0.7,"pith_summary":"The paper argues that the standard sub-subleading soft graviton theorem, obtained from a BCFW decomposition of tree-level amplitudes, has been missing a set of distributional terms. When the soft differential operator acts on angle-bracket poles of the amplitude, spinor derivatives produce delta functions supported on collinear configurations of two hard particles; the theorem must subtract these terms to keep both sides free of distributions. With the subtraction, the amplitude-side soft factor reproduces the collinear part of the asymptotic charge found in canonical analyses of Einstein's equations. This matters because it removes an apparent mismatch between the soft theorem and the Ward-identity picture at sub-subleading order, and it sharpens what the soft theorem is an identity about.","feed_headline":"Soft graviton theorem gains collinear correction terms","feed_subtitle":"New delta-function terms reconcile the amplitude soft theorem with asymptotic charge analysis","key_machinery":"The load-bearing mechanism is a distributional identity for spinor derivatives,\n$$\\tilde\\lambda_i \\partial_{\\tilde\\lambda_j}\\frac{1}{\\langle jk\\rangle} = [ik]\\,\\delta(j,k),\\qquad \\delta(j,k)=\\pi\\$delta^{2}$(\\langle jk\\rangle),$$\nwhich the paper treats as a non-vanishing distributional relation even where earlier derivations set it to zero. Applied to the product of angle-bracket poles in an MHV amplitude, the identity turns derivatives of the soft operator into delta-function terms concentrated on collinear pairs; these recombine, through the collinear factorization of the amplitude, into the graviton splitting functions times a lower-point amplitude. The central move is the prescription that these distributional terms be subtracted from the definition of the soft factor, yielding an identity in which both sides are ordinary functions.","core_discovery":"On the paper's terms, the correct positive-helicity sub-subleading soft graviton theorem for tree-level amplitudes is\n$$$M^{{(2)}}$_{n+1} = \\frac{1}{2}\\sum_{a=1}^n \\frac{[sa]}{\\langle sa\\rangle} $D_a^{2}$ M_n - \\frac{1}{2}\\sum_{a=1}^n \\sum_{\\substack{b>2\\\\ b>a}} \\frac{[sb]^3}{\\langle sb\\rangle} $f^{{h_a h_b}}$_{h_P}(t)\\, \\delta(a,b)\\, M_{n-1}(\\ldots, $P^{{h_P}}$, \\ldots),$$\nwhere $D_a = \\tilde\\lambda_s \\partial_{\\tilde\\lambda_a}$ and the second sum subtracts the distributional collinear terms generated when $D_a^2$ acts on the amplitude's poles. The subtraction uses the graviton splitting functions $f^{h_a h_b}_{h_P}(t)$ with $t = \\omega_a/(\\omega_a+\\omega_b)$ and delta functions $\\delta(a,b)$ forcing particles $a$ and $b$ collinear. The paper derives this for all tree-level MHV amplitudes and argues by collinear factorization that it extends to general tree-level amplitudes, then shows that after translating to celestial-sphere coordinates and taking four $\\bar z$ derivatives, the correction matches the non-divergent collinear component of the canonical asymptotic charge.","pith_inferences":["Editorial inference: the subtraction rule effectively redefines the soft operator on amplitudes; checking that the shifted operator reproduces the celestial OPE algebra would give an independent test of the split between hard and collinear charge.","Editorial inference: because the derivation relies only on the distributional identity and collinear factorization, the same construction should produce subleading-order collinear corrections in Yang-Mills theory; the paper names this as a future direction, but it follows from the same mechanism.","Editorial inference: at loop level, where subleading soft factors are corrected, the same distributional terms are expected to appear shifted; self-dual gravity, where loop amplitudes are known, is a natural place to look for them."],"forward_implications":["The sub-subleading soft theorem for tree-level gravitons becomes an identity between distribution-free quantities only after the collinear subtraction; comparisons that omit it are comparing the soft factor to a shifted amplitude.","At leading and subleading order the correction vanishes, so the Weinberg and subleading soft theorems are unaffected; the new terms are specific to sub-subleading order in gravity.","The correction matches, after four celestial-sphere derivatives, the non-divergent collinear part of the canonical asymptotic charge, so the soft-theorem/Ward-identity correspondence survives at this order.","For general tree-level amplitudes the same subtraction is dictated by collinear factorization, so the result is not an artifact of the MHV form used in the proof.","By the same argument, acting with the soft factor on the collinear pole is equivalent to taking the soft limit after the collinear limit, resolving the order-of-limits ambiguity for these terms."],"supporting_citations":[{"why":"Supplies the BCFW derivation of the sub-subleading soft graviton theorem and the convention of dropping distributional spinor derivatives that this paper corrects.","marker":"[21]"},{"why":"Provides the canonical asymptotic-charge analysis whose collinear component is the target of the match in Section 7.","marker":"[18]"},{"why":"Source of the collinear charge term noted in Section 3 and in the Discussion as the expected counterpart of the distributional terms.","marker":"[19]"},{"why":"Gives the distributional identity for derivatives of inverse spinor components used in Section 3.","marker":"[26]"},{"why":"Provides the graviton splitting functions and collinear factorization used to recombine the delta-function terms.","marker":"[2]"},{"why":"Hodges formula for MHV graviton amplitudes, which the paper rewrites in pole-product form for the general MHV proof.","marker":"[25]"},{"why":"Sets the collinear-limit parametrization and notation used throughout the calculation.","marker":"[23]"}],"fun_headline_variants":["Sub-subleading soft theorem revised with collinear terms","Collinear corrections reconcile soft graviton theorem with charges","Soft graviton theorem gets collinear delta-function terms","Cachazo-Strominger soft theorem fixed by collinear singularities","Graviton soft theorem updated with collinear corrections"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument stands or falls on treating the distributional spinor identity as physically operative, so that derivatives acting on angle-bracket poles produce delta functions that must be subtracted; if the original convention of dropping these terms were the correct one, every collinear correction would vanish and the agreement with the asymptotic charge analysis would disappear.","fun_headline_variants_meta":{"raw":{"variants":["Sub-subleading soft theorem revised with collinear terms","Collinear corrections reconcile soft graviton theorem with charges","Soft graviton theorem gets collinear delta-function terms","Cachazo-Strominger soft theorem fixed by collinear singularities","Graviton soft theorem updated with collinear corrections"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000847,"raw_usage":{"total_tokens":3699,"prompt_tokens":974,"completion_tokens":2725,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":590,"completion_tokens_details":{"reasoning_tokens":2644}},"tokens_in":590,"tokens_out":2725,"duration_ms":18720,"temperature":1.0,"reasoning_tokens":2644,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T10:54:15.157842+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take an explicit five-point MHV graviton amplitude, apply $\\frac{1}{2}\\sum_a [sa]\\langle sa\\rangle^{-1}D_a^2$ to the four-point amplitude, and collect the coefficient of each collinear delta function $\\delta(a,b)$ after imposing momentum conservation. If that coefficient is not $\\frac{1}{2}[sb]^3\\langle sb\\rangle^{-1} f^{h_a h_b}_{h_P}(t)$ times the appropriate three-point amplitude, the corrected soft theorem (1.6) fails; conversely, a direct canonical charge computation yielding a different coefficient of the collinear delta term after the four $\\bar z$-derivatives of (7.3) would falsify the claimed match.","supporting_citations":[{"cited_title":"Di Francesco, P","cited_arxiv_id":null,"evidence_quote":"Gives the distributional identity for derivatives of inverse spinor components used in Section 3."}],"review_version":1}