{"id":"62a3fd91-a2b8-48e6-b564-0d01851c2388","arxiv_id":"2411.19657","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For any integer d, the Kontsevich graph complex GC^2_d and the oriented graph complex OGC^2_{d+1} are connected by a zigzag of quasi-isomorphisms of dg Lie algebras.","lead":"This paper proves that the Kontsevich graph complex and its oriented version are equivalent as dg Lie algebras, lifting a known cohomology isomorphism to the chain level. It matters because graph complexes underlie deformation quantization, Grothendieck-Teichmüller theory, and the cohomology of moduli spaces.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The acyclicity of ker π1 (Prop 4.0.2) is the least secure step: the proof compresses a spectral sequence into an appeal to [M] §6.2.3, and the asserted form of the associated graded differential is not verified.","rationale":"The reader's weakest_assumption identifies both the acyclicity of ker π1 (Prop 4.0.2) and the Maurer-Cartan property of γ as external dependencies from [M]. The acyclicity of ker π1 is the more load-bearing concern for the central claim: it is the step that upgrades the previously known vector-space-level isomorphism to a chain-level quasi-isomorphism, and its proof in the paper is a compressed spectral sequence argument that ends with a citation to an unpublished preprint. The Maurer-Cartan property of γ is also necessary, but the paper at least states that it is a calculation from [M], and the twisted differential would be a natural object to verify independently. The main reason to keep the verdict CONDITIONAL rather than ACCEPT is that Prop 4.0.2 is not self-contained: the identification of the associated graded differential and the coker(δ1) description are asserted without proof, and the final acyclicity is deferred to an external source. These are not signs of fraud or sloppiness; they are unresolved dependencies that a careful reader cannot check from the text alone. The proposed test—explicitly computing the double associated graded differential and checking low-degree homology—would either confirm the argument or expose a concrete failure. Therefore the existing CONDITIONAL verdict is appropriate, and no change to the reader's assessment is needed.","tokens_in":11220,"tokens_out":7147,"duration_ms":61353,"concrete_test":"Write out the differential δ = [γ, ·] explicitly on the generators of zOGC_{d,d+1}, apply the two filtrations from Prop 4.0.2, and list all terms of the double associated graded differential. Verify that exactly δ1 and δ2 survive and that the coker(δ1) description is as stated. As a numerical check, compute H(ker π1) for d = 0 (or d = -1) in low loop order using the explicit graph differential; if any nonzero class appears, Prop 4.0.2 is false. If the check reproduces zero homology and the differential terms match, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central zigzag requires both arrows π1 and π2 to be quasi-isomorphisms of dg Lie algebras. Prop 4.0.1 for π2 is reasonably self-contained, but Prop 4.0.2 is not. The proof filters ker π1 by essential vertices, then by black vertices, and asserts that the double associated graded differential has exactly two terms δ1 and δ2, with δ1 the 'natural inclusion' of OGC^0_{d+1} into OGC_{d,d+1} and δ2 creating one inessential vertex from an edge between two essential vertices. The subsequent reduction to coker(δ1), identified with graphs having at least one white vertex and at least one black essential vertex, and the final acyclicity of that coker are justified only by 'the argument of §6.2.3 in [M]', an external preprint not reproduced here. No convergence or boundedness of the filtrations is discussed. If the double associated graded complex contains an additional differential term, or if the coker(δ1) description is incorrect, the claimed acyclicity (10) may fail, and π1 would not be a quasi-isomorphism. Since the known vector-space-level isomorphism [W2] does not supply this chain-level statement, the main theorem depends on this step.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs, for every integer d, a two-coloured graph complex zOGC_{d,d+1} together with explicit maps π1: zOGC_{d,d+1} → GC^0_d and π2: zOGC_{d,d+1} → OGC^0_{d+1}. The main theorem claims that both maps are quasi-isomorphisms of dg Lie algebras, so that GC^0_d and OGC^0_{d+1} are connected by a zigzag of quasi-isomorphisms; combining this with the known inclusions GC^2_d ⊂ GC^0_d and OGC^2_{d+1} ⊂ OGC^0_{d+1} yields the advertised quasi-isomorphism between GC^2_d and OGC^2_{d+1}. The construction of zOGC_{d,d+1} is presented in Section 3 as a deformation complex of a morphism of 2-coloured operads, and Section 4 proves the two quasi-isomorphism statements by spectral sequence and acyclicity arguments.","tokens_in":11492,"tokens_out":6065,"duration_ms":49449,"significance":"If the main theorem is correct, it upgrades the known cohomology-level isomorphism H(GC^0_d) ≅ H(OGC^0_{d+1}) to a chain-level statement in the category of dg Lie algebras, which is likely to be useful in deformation-theoretic applications such as the Grothendieck–Teichmüller Lie algebra and formality questions. The construction of the intertwining complex is explicit, and the general strategy—deformation theory of a 2-coloured operad morphism plus acyclicity of the two kernels—is natural. The paper is not circular: the central claim is a new chain-level object, and the arguments are mathematical derivations rather than restatements of the input. However, the proof currently delegates its most delicate computations to the authors' unpublished preprint [M], which makes independent verification difficult.","major_comments":[{"comment":"The proof that ker π1 is acyclic is the load-bearing step for the left arrow of the zigzag, and it is not self-contained. The double filtration (by essential vertices, then by black vertices) is asserted to have an associated graded differential with exactly two terms δ1 and δ2, and the identification of coker(δ1) with graphs having at least one white vertex and at least one black essential vertex is not verified; the final acyclicity of (OGC^{•◦}_{d,d+1}, δ2) is then deferred to 'the argument of §6.2.3 in [M]', an unpublished preprint. No convergence or boundedness statement for the two spectral sequences is supplied. Since the known cohomology-level isomorphism [W2] does not provide this chain-level statement, Proposition 4.0.2 is a genuine gap rather than a cosmetic omission.","section":"§4, Prop. 4.0.2"},{"comment":"The assertion that the degree-1 element γ is a Maurer–Cartan element, i.e. γ ˝ γ = 0, is what makes δ = [γ, ·] a differential and gives zOGC_{d,d+1} its dg Lie algebra structure. This fact is imported from §§5.1.1 and 5.2.2 of [M] without reproduction. It is not a peripheral computation: the entire proof of the main theorem is carried out in the twisted complex (zOGC_{d,d+1}, [·, ·], δ). The authors should either reproduce the computation or give a complete proof in this paper.","section":"§3.3, Eq. (8) and definition of γ"},{"comment":"The statement that 'It is elementary to check that π1 and π2 are both morphisms of dg Lie algebras' is not demonstrated. For π2 the claim is structurally clear from the semidirect product decomposition, but for π1 one must verify compatibility with the twisted differential δ = [γ, ·] and with the Lie bracket; in particular the contraction of inessential black vertices into edges between white vertices must be checked against the differential. A concise verification or a reference to a stated lemma is needed.","section":"§4, first paragraph"},{"comment":"The spectral sequence arguments use filtrations by the number of vertices or essential vertices, but the relevant graph spaces are products over the number of edges and vertices, so the filtrations are not bounded below and convergence is not automatic. The paper should state explicitly which filtration and convergence argument is being used before concluding acyclicity from the associated graded complex.","section":"§4, Props. 4.0.1 and 4.0.2"}],"minor_comments":[{"comment":"The opening pages contain numerous typos ('W e', 'o f', 'literat ure', 'th e', 'cohomolo gy'); the manuscript should be proofread before publication.","section":"Abstract and Introduction"},{"comment":"The text says the isomorphism is upgraded to 'quasi-isomorphic as Lie ∞ algebras', while the abstract and the main theorem state 'quasi-isomorphic as dg Lie algebras'; the relation between these two statements should be clarified.","section":"§2.2, last sentence"},{"comment":"Equation (2) and the surrounding orientation conventions are not readable because of missing labels in the displayed graphs; the figures should be redrawn.","section":"§2.1, Eq. (2)"},{"comment":"The notation Gra˝_{d,d+1}(m,n) is introduced with a product over p, but the generators are then said to have 'some number of unlabeled edges'; the relationship between p and the unlabeled edges should be clarified.","section":"§3.2"},{"comment":"Reference [M] is an unpublished preprint from 2023; since the paper relies on it for two pivotal computations, the authors should either mark clearly which results are used or make the paper self-contained.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"I recommend major revision. The mathematical idea is promising and the paper is generally well organized, but the two central computational inputs are taken from the authors' own unpublished preprint [M]. If the authors can make the proof self-contained—at least for Proposition 4.0.2 and the Maurer–Cartan property of γ—I would be prepared to support acceptance. There is no evidence of circularity; the stress-test concern about Prop. 4.0.2 lands and should be addressed head-on."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real content here is the chain-level upgrade: for each d there is a dg Lie algebra zOGC_{d,d+1} with quasi-isomorphisms to GC^0_d and OGC^0_{d+1}. That is strictly stronger than the cohomology-level isomorphism in Willwacher's earlier paper, and it is genuinely new. The 2-coloured operad construction is a real technique, not a repackaging. The paper also gives a fresh proof of the old isomorphism along the way.\n\nWhat the paper does well: the definition of zOGC is explicit, the maps π1 and π2 are simple and natural, and the proof that π2 is a quasi-isomorphism is mostly self-contained. The acyclicity argument for ker π2, using vertex-colour complexes and acyclic tensor factors, is clear and convincing. The attribution to prior work is honest: the paper states plainly which earlier results are being upgraded and which computations are imported.\n\nThe soft spot is Prop 4.0.2, the acyclicity of ker π1. The proof filters by essential vertices, then by black vertices, and asserts that the double associated graded differential has exactly two terms δ1 and δ2, with the stated descriptions. It then identifies coker δ1 and delegates the final acyclicity to §6.2.3 of the preprint [M]. No convergence or boundedness discussion is given, and the two-step spectral sequence is compressed into about a page. I do not think the argument is wrong, but this is a load-bearing gap: if the associated graded complex has an extra term, or the coker description is inaccurate, π1 is not a quasi-isomorphism and the main theorem fails. The Maurer-Cartan property of γ, which defines the differential, is likewise imported from [M] without reproduction.\n\nTwo smaller points. The claim that π1 and π2 are dg Lie algebra morphisms is stated as 'elementary to check' but not shown; this is a minor issue, since it probably is routine, but it should be spelled out or referenced. And the paper relies heavily on [M], an unpublished preprint by one of the authors. Self-citation is not itself a flaw, but here two essential technical inputs live only in that preprint, so a referee must be able to check them.\n\nWho this is for: people working on graph complexes, deformation theory of operads, and Grothendieck-Teichmüller theory. It deserves a serious referee. The central claim is plausible, the construction is meaningful, and the gaps look fillable rather than fatal. I would send it to peer review and ask the authors to either reproduce the imported computations from [M] or provide a fully verified version of [M] alongside.","headline":"Upgrades the known cohomology isomorphism to a chain-level zigzag; worth refereeing despite two imported technical lemmas from a preprint.","tokens_in":12044,"tokens_out":1796,"would_cite":true,"duration_ms":17082,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["17B70","18M85"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every integer d, the Kontsevich graph complex and the oriented graph complex are quasi-isomorphic as differential graded Lie algebras.","keywords":["graph complex","oriented graph complex","dg Lie algebra","quasi-isomorphism","deformation complex","2-coloured operad","Maurer-Cartan element","cohomology"],"falsifier":"Verify directly, in a small example such as d = 0 or d = 1, that γ ˝ γ = 0 in the deformation complex zOGC_{d,d+1}, and compute the cohomology of the subcomplex OGC_{d,d+1}^{•˝} with the differential δ2 that creates one inessential vertex from an edge between essential vertices. A nonzero cohomology class there, or a nonzero self-bracket for γ, would show the claimed quasi-isomorphism does not hold.","tokens_in":11017,"feed_emoji":"🕸️","tokens_out":9786,"duration_ms":77736,"temperature":0.7,"pith_summary":"The paper proves, for every integer d, that the Kontsevich graph complex $GC_d^{2}$ and the oriented graph complex OGC_{d+1}^2 are quasi-isomorphic as differential graded Lie algebras. This upgrades a previously known isomorphism of their cohomology groups to an isomorphism at the chain level, where the differential and Lie bracket are taken into account. The proof works by building an explicit intermediate dg Lie algebra of graphs, zOGC_{d,d+1}, with two colours of vertices and two kinds of edges, together with simple projections to both target complexes. Showing both projections are quasi-isomorphisms is the technical heart of the paper. A sympathetic reader should care because these complexes encode deformation-theoretic invariants in Poisson geometry, Lie bialgebra theory, and the rational homotopy of moduli spaces, so identifying them at chain level transfers structure between those settings.","feed_headline":"Graph complexes match as Lie algebras, not just in cohomology","feed_subtitle":"A new intermediate graph algebra links the Kontsevich and oriented complexes, upgrading cohomology to chain level","key_machinery":"The load-bearing object is the dg Lie algebra zOGC_{d,d+1}, defined as the connected deformation complex of a morphism from a 2-coloured operad Holie_{d,d+1} (controlling shifted homotopy Lie structures on pairs of graded spaces V[d] ⊕ W[d−1]) into a 2-coloured graph operad Grad,d+1. This complex decomposes as a semidirect product OGC_{d,d+1} ⋊ OGC_{d+1}^0; its differential is twisted by a Maurer–Cartan element γ, a degree-one graph element satisfying γ ˝ γ = 0, which makes the twisted differential square to zero. The two maps π1 and π2 extract the two sides of the diagram. The main technical work is the proof that the three relevant associated graded complexes—after filtering by vertex counts, essential black vertices, or black vertices—are acyclic, so that the two kernels have vanishing cohomology.","core_discovery":"The central claim is that there exists, for each d in Z, a dg Lie algebra zOGC_{d,d+1} fitting into a diagram $GC_d^{0}$ ← zOGC_{d,d+1} → OGC_{d+1}^0 in which both arrows are quasi-isomorphisms. The left arrow π1 sends a graph by erasing all inessential black vertices of valence two and replacing each one by an edge between the two white neighbours; the right arrow π2 simply projects the semidirect product OGC_{d,d+1} ⋊ OGC_{d+1}^0 onto its second factor. The proof of quasi-isomorphism proceeds by filtering the kernels of these projections and showing the associated graded complexes are acyclic. Because the inclusions $GC_d^{2}$ ⊂ $GC_d^{0}$ and OGC_{d+1}^2 ⊂ OGC_{d+1}^0 are known quasi-isomorphisms, the theorem delivers a zigzag of quasi-isomorphisms between the bivalent complexes $GC_d^{2}$ and OGC_{d+1}^2 as dg Lie algebras, and after truncation to loop orders at least two, between the trivalent complexes GC_d and OGC_{d+1}^3.","pith_inferences":["The authors do not claim it, but the explicit nature of the intertwining complex suggests the zigzag may be replaceable by a single direct chain map between GC_d^2 and OGC_{d+1}^2.","One could test the compatibility of the maps with known cohomology classes: for instance, the degree-zero class corresponding to grt1 should be matched explicitly by π1 and π2 in low loop orders.","A self-contained proof of the acyclicity arguments imported from the earlier preprint would make the theorem independent of that preprint; until then, the statement rests on those computations.","The construction may translate to other ground rings or to graph complexes with different vertex decorations, but the paper works only in characteristic zero."],"forward_implications":["The cohomology Lie algebras H^•(GC_d^2) and H^•(OGC_{d+1}^2) are isomorphic as graded Lie algebras, not merely as graded vector spaces.","The loop-order filtrations match: both maps preserve the loop-order grading, so the isomorphism holds graded by loop order.","The zigzag extends to the trivalent subcomplexes: for loop orders at least two, GC_d and OGC_{d+1}^3 are connected by quasi-isomorphisms of dg Lie algebras.","Deformation problems controlled by either complex—for example Lie bialgebra properad deformations or homotopy automorphisms of operads—are governed by equivalent dg Lie algebras, so invariants transfer across."],"supporting_citations":[{"why":"Supplies the cohomology-level isomorphism between GC_d^0 and OGC_{d+1}^0 that the paper upgrades to a chain-level quasi-isomorphism.","marker":"[W2]"},{"why":"Provides the two technical inputs on which Proposition 4.0.2 depends: the Maurer–Cartan property of γ and the acyclicity argument for the associated graded complex.","marker":"[M]"},{"why":"Gives the deformation-complex construction of GC_d^0 and the known quasi-isomorphism GC_d^2 ⊂ GC_d^0 used to extend the zigzag.","marker":"[W1]"},{"why":"Introduces the original graph complex in the context of deformation quantisation of Poisson structures, the object being upgraded.","marker":"[K1]"},{"why":"Supplies the operadic composition law for graphs used throughout the construction of the deformation complexes.","marker":"[K2]"},{"why":"Gives an earlier proof of the cohomology-level isomorphism between oriented and unoriented complexes, but only as graded vector spaces.","marker":"[Z]"}],"fun_headline_variants":["Graph complexes linked via new intermediate algebra","Oriented and Kontsevich complexes now quasi-isomorphic","Chain-level bridge between graph complexes found","New algebra unifies Kontsevich and oriented complexes","Graph complexes match beyond cohomology"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The main theorem rests on two computations borrowed from an earlier preprint: the check that the graph element γ satisfies the equation that makes its differential square to zero, and the proof that a certain filtered subcomplex of the kernel of π1 has zero cohomology; if either computation is wrong, the main equivalence could fail.","fun_headline_variants_meta":{"raw":{"variants":["Graph complexes linked via new intermediate algebra","Oriented and Kontsevich complexes now quasi-isomorphic","Chain-level bridge between graph complexes found","New algebra unifies Kontsevich and oriented complexes","Graph complexes match beyond cohomology"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000209,"raw_usage":{"total_tokens":1336,"prompt_tokens":801,"completion_tokens":535,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":417,"completion_tokens_details":{"reasoning_tokens":467}},"tokens_in":417,"tokens_out":535,"duration_ms":4627,"temperature":1.0,"reasoning_tokens":467,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T05:58:42.097545+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Verify directly, in a small example such as d = 0 or d = 1, that γ ˝ γ = 0 in the deformation complex zOGC_{d,d+1}, and compute the cohomology of the subcomplex OGC_{d,d+1}^{•˝} with the differential δ2 that creates one inessential vertex from an edge between essential vertices. A nonzero cohomology class there, or a nonzero self-bracket for γ, would show the claimed quasi-isomorphism does not hold.","supporting_citations":[],"review_version":1}