{"id":"0b5a9bb8-6ecb-436f-9e56-e9cbbeead760","arxiv_id":"2505.23385","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Open-string AdS building blocks can be generated by Drinfeld associator recursions and closed-string ones by Deligne associator recursions, yielding all-order zeta-valued expansions.","lead":"This paper constructs explicit matrix recursions that generate the building blocks of open and closed string amplitudes on AdS from the Drinfeld and Deligne associators, the generating series of multiple zeta values. The result gives all-order low-energy expansions of these AdS integrals and a new proof that their coefficients are (single-valued) multiple zeta values.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"General-word associator construction is asserted, not proven: eqs. (4.19)-(4.24) are verified only for weight <= 3 words, so the all-order claim for every J_w/I_w is not yet secured.","rationale":"The reader's weakest_assumption identifies exactly the load-bearing gap: the general-word matrix blocks and boundary values (4.19)-(4.24) are asserted by pattern after three explicit examples and the weight-three appendix, without a proof for arbitrary words. My reading of Sections 4.1 and 4.2 confirms that the all-order associator generation of the AdS building blocks depends on those formulae: without them, the first component of C1 cannot be extracted from the Drinfeld recursion, and the Deligne recursion inherits the same unproven structure. I find no independent fatal flaw; the examples are explicit, the shift relations are cited from earlier work, and the single-valued lift is a natural step once the open-string construction is established. The concrete test I propose is a finite, mechanizable check that would either substantiate the pattern for longer words or reveal a counterexample. Because the gap is real but addressable and the surrounding evidence is credible, the appropriate verdict remains conditional acceptance, matching the reader's CONDITIONAL verdict.","tokens_in":27541,"tokens_out":4713,"duration_ms":43390,"concrete_test":"Write a small symbolic script that, for every word w of length <= 5, (i) builds F_w(x) from equation (4.17), (ii) computes dF_w/dx using only the defining integral (4.1) and relation (4.6), (iii) constructs e0 and e1 from the block rules (4.19)/(4.21), and (iv) checks the KZ equation componentwise; then (v) independently compute the small-x and small-(1-x) asymptotics of the integrals S_w[i](x), e.g. for w=0010 and w=0011, and compare with (4.23)/(4.24). If all words up to length 5 satisfy the KZ equation and the boundary-value formulas, the pattern is strongly supported; the first failure would locate the exact point where the claim breaks.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central claim is that for every word w in {0,1}*, the vector F_w(x) built from the integrals S_w[i](x) satisfies the KZ equation with matrix blocks (4.19)/(4.21), and that the regularized boundary values take the closed forms (4.22)-(4.24). Section 4.1 illustrates this for three words (e, 0, 001) and Appendix B lists all weight-three words, but no proof is given that the derivative relations (4.6) assemble into the stated global matrices for a general word, nor that the small-x asymptotics of the S-integrals produce the alternating 1/s^k entries of (4.24) for w = ...10^k. The even components of C1 in (4.22) are left unspecified. Since the all-order generation of J^{s,t}_w and, through the single-valued lift, I^{s,t}_w depends on these formulae, the central claim currently rests on pattern extrapolation. This is a specific, addressable gap rather than a contradiction; the low-weight examples and the internal consistency of the outlined recursion support a conditional accept.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims that the AdS string amplitude building blocks for open and closed strings are generated by the Drinfeld and Deligne associators, respectively. For a word w in {0,1}*, the author constructs a vector F_w(x) of regulated Selberg-type integrals with polylogarithm insertions, asserts that it satisfies the KZ equation (2.2) with matrix blocks given by pattern in eqs. (4.19) and (4.21), and states the regularized boundary values in eqs. (4.22)–(4.24). Applying the Drinfeld recursion then yields all-order low-energy expansions of the integrals J_w^{s,t} with multiple zeta values, and applying the single-valued map yields the Deligne recursion for the I-integrals with single-valued multiple zeta values. The construction is worked out in detail for words up to weight three (Examples 1–3 and Appendix B), and a higher-point analogue is illustrated with a two-fold integral (Example 4 and Appendix C).","tokens_in":27907,"tokens_out":15685,"duration_ms":139300,"significance":"If the general-word claim were fully established, the paper would provide a clean and unifying explanation of the MZV and svMZV expansions of AdS amplitude building blocks, extending the flat-space associator recursions to a curved background. The explicit matrix constructions, the worked weight-three catalog, and the single-valued lift are valuable and appear internally consistent in the low-weight cases. The paper ships no code or machine-checked proofs, but it does give concrete, checkable formulae and no fitted parameters, with the construction resting on standard facts about KZ solutions and the single-valued map. The main significance is conditional on closing the general-word proof gap.","major_comments":[{"comment":"The central claim is that for every word w in {0,1}*, the vector F_w(x) in eq. (4.17) satisfies the KZ equation (2.2) with the matrix blocks (4.19) and (4.21), and that the regularized boundary values take the closed forms (4.22)–(4.24). This is verified only for words of length at most three (Examples 1–3 and Appendix B); the general block matrices and boundary values are asserted by pattern, with the even components of C1 in (4.22) left as '···'. Since the all-order expansions of every J_w and, through the single-valued lift, every I_w depend on these formulae, the main theorem is not proved for words of arbitrary length. This is an addressable gap: an induction on the suffix structure of w using eq. (4.6) would likely close it, but the proof must be supplied or the claim explicitly restricted to the verified weights.","section":"Section 4.1, eqs. (4.19)–(4.24)"},{"comment":"The Deligne recursion for the closed-string building blocks inherits the same gap. The identification C1 = sv(C1) in eq. (4.33) requires the explicit form of all components of C1 for a general word, but eq. (4.22) specifies only the odd components and leaves the even components undefined. Without a closed form for the even components, the statement that the Deligne associator generates every I_w is not established. A proof or a precise statement of the proven domain is needed before the all-order claim for arbitrary I_w can be accepted.","section":"Section 4.2, eqs. (4.33)–(4.34)"}],"minor_comments":[{"comment":"The statement 'Z^sv = Q[(ζ^sv_w)] ⊂ Z' is incorrect as written: single-valued multiple zeta values do not form a subring of ordinary multiple zeta values. The single-valued map is not an inclusion; it is a ring homomorphism from Z to another ring. Please rephrase, e.g., 'Z^sv = sv(Z)'.","section":"Section 2, after eq. (2.1)"},{"comment":"The matrices displayed in the examples contain entries that appear inconsistent with eq. (4.6) unless a nontrivial scaling of the vector components is assumed. For instance, in eq. (4.10) the (1,2) entry of e0 is shown as s, whereas eq. (4.6) with w=0 gives a coefficient t for S_0[1] in the derivative of S_0[0]; ratios such as s/t and t/s also appear in later matrices. Please clarify the normalization convention used for the vector entries, or correct the matrices so that they directly match eq. (4.6).","section":"Section 4.1, eqs. (4.10), (4.14), (4.19), (4.21)"},{"comment":"The shorthand 'S_{w3,w4}[i3,i4](x) = s_{3 i1} s_{4 i2} S_{w3,w4}[i3,i4](x)' uses undefined indices i1 and i2; presumably the factors should be s_{3,i3} and s_{4,i4}. Please fix the notation.","section":"Section 4.1, eq. (4.29)"},{"comment":"The '···' notation for the even components of C1 is opaque. At least the examples in Appendix B give concrete expressions; consider defining the even components explicitly for a general word, or state that they are determined by the construction and give the general pattern.","section":"Section 4.1, eq. (4.22)"},{"comment":"The phrase 'another proof' in the abstract and introduction is stronger than what is currently demonstrated for arbitrary words. Until the general-word construction is proved (or its domain of validity is stated precisely), it would be more accurate to say that the paper gives a new recursive framework and proves the claim for low weights, with a conjectural extension to all orders.","section":"Abstract and Section 1"}],"recommendation":"major_revision","confidential_remarks":"This is a promising paper with a clear and elegant central idea, and the low-weight examples are concrete and checkable. The main issue is that the all-order claim is not yet supported by a proof for general words; the gap is specific and likely fillable by an induction on the suffix structure of the word. I recommend major revision rather than rejection. The paper fits the scope of JHEP, and the author's citation of his own related work is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper gives a new and plausible associator formalism for the AdS J- and I-integrals, but the all-order claim for arbitrary words rests on a pattern that is verified only through weight three. I think it deserves review, with the general-word proof requested.\n\nWhat's actually new: prior work proved the zeta-valued low-energy expansions and derived shift relations. This paper constructs explicit matrix representations such that the Drinfeld associator recursively generates the open-string building blocks and the Deligne associator generates the closed-string ones. That directly answers the question raised in ref. [32] and supplies all-order generating relations. The three worked examples and Appendix B are concrete and internally consistent, and the single-valued lift to the I-integrals is a natural and clean step.\n\nSoft spots: the general-case matrices (4.19)/(4.21) and boundary values (4.22)-(4.24) are stated by pattern, not proven. In particular, (4.22) leaves the even components of C1 unspecified, saying only that they depend on the word. The low-weight examples are convincing, but the claim 'for every word w' is currently an extrapolation. This is a specific, addressable gap, not a contradiction; I found no sign that the pattern fails. The higher-point section is honestly labeled as speculative, so I don't hold that against it. I also note the paper says it provides 'another proof' of zeta-valued expansions; that is fair, but the proof is conditional on the unproven general matrices.\n\nThe citation pattern looks fine: the flat-space recursions are properly credited, and the self-citation to [20] is relevant prior work. No fitted parameters, no circularity.\n\nWho this is for: anyone working on AdS amplitudes, single-valued periods, or associator methods. The paper is well written and the examples make the construction easy to check. It deserves a serious referee; I would send it to review and ask the author to prove the general block structure or at least state it as a conjecture with a precise proof sketch. That would turn a conditional result into a solid one.","headline":"A credible and useful associator formalism for AdS amplitude building blocks, but the all-order claim for general words rests on a pattern verified only through weight three.","tokens_in":28300,"tokens_out":1988,"would_cite":true,"duration_ms":19316,"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":"This paper shows that the AdS open- and closed-string amplitude building blocks are generated by the Drinfeld and Deligne associators, respectively, yielding all-order low-energy expansions with (single-valued) multiple zeta value…","keywords":["AdS string amplitudes","Drinfeld associator","Deligne associator","multiple zeta values","single-valued multiple zeta values","Knizhnik-Zamolodchikov equation","Selberg integrals","polylogarithm insertions"],"falsifier":"Compute the vector F_w(x) for a weight-four word such as w=0010 from its integral definition, take the regularized boundary values, and compare the first component of C1 with the first component of Φ(e0,e1)C0 using the proposed matrices; a mismatch at any order in s,t would falsify the claim.","tokens_in":27307,"feed_emoji":"","tokens_out":8482,"duration_ms":79184,"temperature":0.7,"pith_summary":"Open- and closed-string scattering on anti-de Sitter space is built from families of beta-function integrals with polylogarithm insertions, the J- and I-integrals. This paper claims that these families are not an arbitrary zoo: each building block is generated by a single recursive object. For open strings the Drinfeld associator, the generating series of multiple zeta values, relates two endpoint vectors of a carefully chosen integral vector, producing every J-integral in its low-energy expansion. For closed strings the same construction with the single-valued Deligne associator produces the I-integrals. If correct, this gives all-order closed-form expressions and a uniform explanation of why the coefficients are (single-valued) multiple zeta values.","feed_headline":"Associators generate the building blocks of AdS string amplitudes","feed_subtitle":"Open strings: Drinfeld associator. Closed strings: Deligne associator. Both yield all-order expansions.","key_machinery":"The load-bearing object is the Drinfeld associator Φ(e0,e1), the generating series of multiple zeta values in noncommutative letters e0,e1, together with its single-valued image the Deligne associator $Φ^{{sv}}$. The mechanism that connects them to AdS integrals is the vector F_w(x) of eq. (4.17), whose entries are the integrals S_{w'}[i](x) = ∫_0^x dy y^s(1-y)^t(x-y)^r L_{w'}(y)/(y-y_i) for every suffix w' of w. This vector solves the KZ equation with the block matrices (4.19) and (4.21), so the standard boundary-value relation C1 = Φ(e0,e1)C0 applies; mapping everything to its single-valued analogue turns it into C1 = $Φ^{{sv}}$(e0,e1)C0. The matrices encode how each prefix letter 0 or 1 feeds the next suffix, and it is this recursive bookkeeping that turns the associator expansion into the J- and I-integrals.","core_discovery":"The paper's central discovery is that the AdS amplitude building blocks are governed by the same two associators that organise flat-space string amplitudes. For every word w in {0,1}*, the vector F_w(x) whose entries are deformed Selberg integrals S_{w'}[i](x) over all suffixes w' of w solves the Knizhnik-Zamolodchikov equation with explicit block matrices e0 and e1. The regularized endpoint values C0 and C1 are therefore related by C1 = Φ(e0,e1) C0, with the odd components of C1 equal to s $J^{{s,t+1}}$_{w_i...w_n}; applying the shift relations converts these into the building blocks $J^{{s,t}}$_w. Since the same matrices serve the single-valued version, the Deligne recursion C1 = $Φ^{{sv}}$(e0,e1) C0 generates the closed-string building blocks $I^{{s,t}}$_w. The paper works out all weight-three words, a two-fold iterated example, and states the general block-matrix pattern.","pith_inferences":["A natural next step, hinted at by the paper's outlook, is to feed the J- and I-integrals into the physical amplitude combinations: if the block-matrix pattern persists, the full four-point AdS amplitudes, not just their building blocks, would carry an associator presentation, giving a direct handle on AdS analogues of KLT double-copy relations.","The fact that the same matrices e0,e1 appear in both the Drinfeld and the Deligne recursions suggests that single-valuedness acts only on the associator and boundary data, not on the differential system; this could simplify the search for the correct twisted cohomology of AdS integrals.","A concrete testable extension is to verify the stated pattern for all weight-four and weight-five words by computer algebra; agreement would strengthen the claim, while any deviation would signal that the function space of S-integrals needs additional elements."],"forward_implications":["Every J-integral J^{s,t}_w gets an explicit all-order low-energy expansion in s and t whose coefficients are multiple zeta values, obtained by evaluating the matrix element of the Drinfeld associator.","Every I-integral I^{s,t}_w gets an all-order expansion with single-valued multiple zeta values as coefficients, from the same matrix element with the Deligne associator.","The generating series J(s,t;e0,e1) and I(s,t;e0,e1) are expressed as matrix elements of the two associators, so building blocks can be produced to arbitrary order without redoing integrals.","The iterated J- and I-integrals defined in Section 4, which are analogues of higher-point amplitudes, are recursively generated from lower-depth integrals by the same associator mechanism.","The construction recovers and re-proves the previously established zeta-value and single-valued-zeta-value coefficient statements for the AdS building blocks."],"supporting_citations":[{"why":"Defines the J- and I-integrals, derives the shift relations, and proves I=sv(J), the identity that the paper lifts to associator form.","marker":"[32]"},{"why":"Supplies the flat-space open-string Drinfeld recursion that this paper generalizes to AdS building blocks.","marker":"[17]"},{"why":"Supplies the flat-space closed-string Deligne recursion that is lifted to AdS in Section 4.2.","marker":"[20]"},{"why":"Established that the low-energy coefficients of the open-string J-integrals are zeta values; the present paper gives another proof and an all-order generating expression.","marker":"[31]"},{"why":"Established that the closed-string I-integrals have single-valued zeta coefficients; the Deligne recursion provides an independent derivation.","marker":"[23]"},{"why":"Introduced the method of relating Selberg integrals to multiple zeta values via the KZ boundary-value setup used throughout.","marker":"[34]"},{"why":"Identified the Deligne associator as the generator of flat-space closed-string amplitudes, the structural precedent for the closed-string claim.","marker":"[12]"}],"fun_headline_variants":["Drinfeld and Deligne associators generate AdS string blocks","AdS string amplitude blocks from Drinfeld and Deligne associators","Associators: key to all-order AdS string blocks","How Drinfeld and Deligne associators build AdS amplitudes","Same associators, new geometry: AdS string blocks"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the vector of integrals built from all suffixes of a word satisfies the required differential equation with the proposed block matrices, and that its endpoint values take the simple closed forms in eqs. (4.22)-(4.24), including the alternating 1/s^k entries for words ending in k zeros; the paper demonstrates three examples and asserts the general pattern.","fun_headline_variants_meta":{"raw":{"variants":["Drinfeld and Deligne associators generate AdS string blocks","AdS string amplitude blocks from Drinfeld and Deligne associators","Associators: key to all-order AdS string blocks","How Drinfeld and Deligne associators build AdS amplitudes","Same associators, new geometry: AdS string blocks"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000905,"raw_usage":{"total_tokens":3830,"prompt_tokens":816,"completion_tokens":3014,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":432,"completion_tokens_details":{"reasoning_tokens":2925}},"tokens_in":432,"tokens_out":3014,"duration_ms":23665,"temperature":1.0,"reasoning_tokens":2925,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T12:45:51.589862+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the vector F_w(x) for a weight-four word such as w=0010 from its integral definition, take the regularized boundary values, and compare the first component of C1 with the first component of Φ(e0,e1)C0 using the proposed matrices; a mismatch at any order in s,t would falsify the claim.","supporting_citations":[{"cited_title":"Terasoma, Selberg Integrals and Multiple Zeta Values , Compositio Mathematica 133 (2002) 1","cited_arxiv_id":null,"evidence_quote":"Introduced the method of relating Selberg integrals to multiple zeta values via the KZ boundary-value setup used throughout."}],"review_version":1}