{"id":"21139f7f-a671-4191-8b2a-c7f953a3b250","arxiv_id":"2507.17568","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A new bimodule Hochschild cohomology and an operadic obstruction theory that prove intrinsic and almost formality theorems for DG bimodules over DG algebras.","lead":"The paper develops an obstruction theory for extending truncated A-infinity structures on bimodules over A-infinity algebras, with obstructions detected by a new bimodule Hochschild cohomology. This yields conditions under which DG algebras with a DG bimodule are forced to be formal, giving a bimodule analogue of Kadeishvili's intrinsic formality theorem.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Load-bearing gap: Theorem 4.1.5's spectral sequence is asserted to generalize [Mur20] from endomorphism operads to arbitrary graded operads with multiplication, but the proof is not reproduced; the central formality theorem depends on it.","rationale":"The reader's weakest_assumption identifies exactly the same load-bearing concern: the proof of Theorem 4.1.5, on which the central formality theorem rests, is an asserted generalization of [Mur20] with no reproduced argument. I considered other potential weak points—the omitted proof of Proposition 2.3.13 (minimal models of pairs), the use of the Complete Convergence Lemma to collapse the lim^1 term in Theorem 5.1.1, and the sign conventions in the operad complex—but each of these is either standard or inherits the same dependence on [Mur20]. The paper is otherwise internally coherent: the new bimodule Hochschild cohomology, the long exact sequence, and the formality theorems are logically connected, and the reliance on [Mur20] is explicit and honest. No internal inconsistency or obvious counterexample was found; the issue is verification, not suspected falsity. The proposed check (a non-endomorphism instance of Theorem 4.1.5) would settle whether the claimed generalization actually holds. If it passes, the main theorems are supported but still warrant a fully written proof in the paper or a transparent reference. Therefore the reader's CONDITIONAL verdict should stand unchanged.","tokens_in":58357,"tokens_out":18559,"duration_ms":194532,"concrete_test":"Verify Theorem 4.1.5 for a target that is not an endomorphism operad. Concretely, take O=ΛAss (the suspended associative operad, i.e., the free graded operad on a binary operation μ of degree 1 modulo μ{μ}=0) and compute the Bousfield-Kan fringed spectral sequence of the tower Map(Ak,ΛAss) directly from the cell structure of Ak and the definition of the Dwyer-Kan mapping spaces. Then compare the E1/E2 pages and the d2 formula with Theorem 4.1.5(9)-(13). If the E2 page is not H^{s+2,-t}(Λ^2Ass) or if d2 is not [m3,x], the 'same proofs work' assertion fails; if it matches, the generalization is supported but still needs a written proof for publication.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Theorem 1.1.3) is proved by reducing to Theorem 5.1.1, a general formality criterion for maps A∞→O whose proof uses the obstruction spectral sequence of Theorem 4.1.5 at every inductive step. The proof of Theorem 4.1.5 is not self-contained: it states that it is 'a compendium of [Mur20, Sections 4, 5 and 6]' and that despite [Mur20] sometimes working with the endomorphism operad O=E(V), 'the same proofs work for general graded operads since they only depend on the homotopy theory of the operad A∞' and one 'only needs to replace the Hochschild complex with the operad complex.' No detailed verification is given for the E1-page identification (Item 9), the E2 identification (Item 11), or the Gerstenhaber-bracket formula for d2 (Item 13) in the stated generality. Theorem 4.2.2 (fibre-wise obstructions) relies on the analogous assertion that the tower of homotopy fibres 'fits within this framework' because 'the required structure is preserved by taking homotopy fibres.' If any of these identifications secretly uses the concrete form E(V)(n)=hom(V⊗n,V) or the identification of cochains with multilinear maps, the obstruction theory and hence the intrinsic/almost formality theorems lack proof. This is the weakest point of the paper: it is an explicit, located assertion of 'same proofs work' with no reproduction, and no machine-checked formalization is provided.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops an obstruction theory for extending truncated minimal A-infinity bimodule structures over truncated minimal A-infinity algebras, using Bousfield-Kan type (truncated) spectral sequences associated to towers of mapping spaces Map(A_k, O) for a graded operad O with multiplication and an associative operadic ideal. It introduces the bimodule Hochschild cohomology HH^{n,r}(A|M), realized as the cohomology of the homotopy fibre of the map δ: C^{•,*}(A) → C^{•,*}_{A^e}(M,M), and shows that the second differential of the obstruction spectral sequence is the Gerstenhaber bracket with a bimodule universal Massey product. The main applications are simultaneous intrinsic formality for a graded algebra and a graded bimodule (Theorem 5.1.5, Theorem 1.1.3), almost formality theorems for DG algebras and DG bimodules (Theorems 5.2.11, 5.2.16, 5.2.23), and d-sparse variants (Section 6).","tokens_in":58687,"tokens_out":17999,"duration_ms":164874,"significance":"If the technical framework is sound, the paper is a significant contribution: it offers a new cohomological criterion for simultaneous intrinsic formality of a graded algebra and a graded bimodule, extends Muro's enhanced obstruction theory from endomorphism operads to arbitrary graded operads with multiplication and associative ideals, and provides explicit obstruction cocycles (Theorem 4.1.5(15)) and an explicit second differential (Theorem 4.1.5(13)). The new bimodule Hochschild cohomology and its long exact sequence with Hochschild cohomology and self-extensions are natural and likely to be useful. The paper is clearly organized and the statements of the main theorems are precise. The main risk is that the central technical engine, Theorem 4.1.5, is asserted to generalize [Mur20] with limited proof, and several later results (including the sparse variants) omit proofs as 'almost identical'; these are the points that need to be strengthened before the conclusions are fully established.","major_comments":[{"comment":"The proof of Theorem 4.1.5 states that [Mur20, Sections 4-6] extend verbatim from endomorphism operads to arbitrary graded operads with multiplication, because 'the same proofs work' and 'one only needs to replace the Hochschild complex with the operad complex.' This is load-bearing: Theorem 5.1.1 invokes Items 7 and 11 at every inductive step, and the intrinsic and almost formality theorems (Theorems 5.1.2, 5.1.4, 5.1.7, 5.2.5, 5.2.10, 5.2.15, and their sparse analogues) all depend on this generalization. Items 9, 11, 13, and 15 involve explicit identifications of E1, E2, the differential d2, and obstruction cocycles that in [Mur20] are proved using the concrete identification of cochains with multilinear maps on E(V). Please provide the detailed verification in the stated generality, or give precise theorem references in [Mur20] that cover arbitrary graded operads.","section":"Section 4.1 (Theorem 4.1.5)"},{"comment":"The proof of Theorem 4.2.2 asserts that the tower of homotopy fibres Str_{A_n,h}(I) 'also fits within this framework since the required structure is preserved by taking homotopy fibres,' but it does not show how the E1-page identification (Item 9) and the d2 formula (Item 13) follow from the fibre sequence and the short exact sequence (3.3.9). This fibre-wise spectral sequence is used in Theorems 5.1.6, 5.1.7, 5.2.19, and 5.2.21, so the omitted argument is not merely expository. Please spell out the induced spectral sequence of the top tower and the compatibility of differentials, or indicate which statements in [Mur20] imply these facts for homotopy fibres.","section":"Section 4.2 (Theorem 4.2.2)"},{"comment":"The hypothesis of Theorem 5.2.19 is stated as '{ {mg3 − mf3} } ∈ E^{2,-1}_{ΛO}(ΛI)', and the hypothesis of Theorem 6.2.10 as '{ {mg_{d+2} − mf_{d+2} } } ∈ E^{2,-1}_{ΛO}(ΛI)'. Since qf = qg forces the difference to be a cocycle in the ideal complex, the stated membership is automatic and imposes no condition on g. The base case of the induction in the proof of Theorem 5.2.19 requires the difference class to be zero, as used in Theorem 5.2.21; for the d-sparse case the appropriate bidegree is E^{d+1,-d}_{ΛO}(ΛI), not E^{2,-1}. These theorem statements should be corrected accordingly.","section":"Section 5.2 (Theorems 5.2.19 and 6.2.10)"},{"comment":"The proofs of the sparse analogues are omitted as 'almost identical' to the d=1 case, and Theorem 6.1.6(12)-(15) generalizes [JKM22, Proposition 5.2.2] to arbitrary d-sparse graded operads, with only Item 15 receiving a sketch. The d-sparse spectral sequence is not a formal consequence of the d=1 case: the page indices, the bidegrees of the differentials, and the ranges of k in Items 12-16 depend on d. Since Theorems 6.2.3-6.2.12 are presented as new results, the verification of Theorem 6.1.6 and the reduction of the sparse proofs should be included, or at least a precise statement of which arguments in [JKM22] carry over unchanged.","section":"Section 6 (Theorems 6.2.3, 6.2.4, 6.2.10, 6.2.12 and Theorem 6.1.6)"}],"minor_comments":[{"comment":"Theorem 1.1.2 is described as providing a 'necessary condition' for intrinsic formality of a graded bimodule, but it actually provides a sufficient condition, in analogy with Kadeishvili's Theorem 1.1.1.","section":"Introduction, page 2"},{"comment":"The differential of the bimodule Hochschild-Massey complex is printed as d: HH^{s,t}(A|M) → HH^{s+2,t-1}(A|M); according to Definition 6.2.1(2) and Definition 6.2.9 it should have bidegree (d+1,-d), i.e., d: HH^{s,t}(A|M) → HH^{s+d+1,t-d}(A|M).","section":"Definition 6.2.7(2)"},{"comment":"The vanishing condition '{ {mM3 − mN3} } = 0' is stated with target Ext^{2,-1} in the introduction and with target EM^{2,-1} in Theorem 5.2.23; the two formulations should be reconciled, since EM^{2,-1} is not literally the same notation as Ext^{2,-1}.","section":"Theorems 1.2.3 and 5.2.23"},{"comment":"There are several typographical errors: 'the the homotopy fibre' on page 8, 'Below, the by the universal Massey product' before Theorem 5.2.11, 'linear endomorphism operand' in the abstract, and the garbled sentence in Definition 2.3.10 ('Let A be an O-algebra A and an M an O-A-bimodule').","section":"Throughout"},{"comment":"The proofs of Theorem 2.3.4 and Proposition 2.3.13 are omitted as 'entirely analogous' or 'almost identical' to [Mur14] and Proposition 2.2.12; since these results justify the mapping-space characterization of quasi-isomorphism of pairs used in the main theorems, a brief indication of the modifications would improve self-containedness.","section":"Sections 2.3-2.4"}],"recommendation":"major_revision","confidential_remarks":"The paper relies heavily on [Mur20] and also on the unpublished or preprint works [JKM22] and [JM]. The refereeing process would be helped if the authors indicated which parts of [Mur20] are being used verbatim and which are genuinely new. The main technical gap is the asserted generalization of the spectral sequence machinery to arbitrary graded operads; this is fixable in principle but needs to be written out. I recommend major revision rather than rejection, as the central claims appear plausible and the gaps are of the kind that can be addressed within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about arXiv:2507.17568. The main results are genuinely new: the bimodule Hochschild cohomology HH^{n,r}(A|M) with its long exact sequence, the bimodule universal Massey product, and the simultaneous intrinsic-formality (5.1.5) and almost-formality (5.2.16, 5.2.23) theorems are not repackaged older results. The second thing is the real soft spot: the central proof rests on Theorem 4.1.5, whose generalization from the endomorphism operad case in [Mur20] is asserted with \"same proofs work\" rather than written out. The stress-test note lands; that is the load-bearing gap.\n\nWhat the paper does well: the strategy is sound and clearly laid out. The cohomology of a pair is a natural object, defined as a homotopy fibre of the action of the Hochschild complex on self-extensions, and the long exact sequence connecting it to HH(A) and Ext(M,M) is useful. The obstruction-theoretic perspective via mapping spaces of DG operads coheres, and the d-sparse variants are a sensible addition.\n\nSoft spots, in order of seriousness. First, Theorem 4.1.5 is a compendium of [Mur20], and the extension from E(V) to arbitrary graded operads with multiplication and an associative ideal hinges on items 9, 11, and 13—the E1-page, E2-page, and the Gerstenhaber formula for d2—none of which is proved here in the stated generality. If any of those identifications secretly uses the concrete form of the endomorphism operad, the formality theorems lack proof. I think the generalization is probably fine, since the proofs in [Mur20] are mostly about the A∞ operad's good behaviour, but the paper should either reproduce the argument or cite specific results in [Mur20] that cover it. Second, Theorem 4.2.2 justifies the fibre-wise tower by \"preserved by taking homotopy fibres\", which is again asserted rather than shown. Third, Section 6 omits proofs as \"almost identical\"; acceptable, but the sparse differentials cannot be checked without the earlier details. Fourth, several explicit cochain formulas are deferred to the companion [JM], in preparation; minor, but it slows verification.\n\nMy own verdict is conditional: the central machinery is likely sound, but the verification burden sits exactly at the asserted generalization. This paper deserves peer review, not desk rejection. Send it out; the referee should require a detailed proof or precise citations for Theorem 4.1.5's generalization, and ask for the deferred formulas to be stated.","headline":"The bimodule formality theorems are new and the architecture holds together, but the load-bearing spectral-sequence generalization from [Mur20] is asserted, not proved; worth a serious referee, who must push on that point.","tokens_in":59285,"tokens_out":3034,"would_cite":true,"duration_ms":33870,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["18M65","18N40","16E45","55S35","18G40","18G80"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes a cohomological criterion for simultaneous intrinsic formality of a graded algebra and a graded bimodule over it, using a new bimodule Hochschild cohomology and a fringed spectral sequence.","keywords":["A-infinity bimodules","intrinsic formality","bimodule Hochschild cohomology","universal Massey product","fringed spectral sequence","differential graded algebras","operads with multiplication"],"falsifier":"Find a graded algebra $A$ and a graded $A$-bimodule $M$ with $HH^{n+2,-n}(A|M)=0$ for all $n\\ge1$ but with two minimal $A_\\infty$-pairs over $(A,M)$ that are not gauge isomorphic; equivalently, compute the spectral sequence of Theorem 4.1.5 for that pair and exhibit a nonzero obstruction on a later page. Alternatively, test the asserted generality by computing the claimed $E_2$-page description and $d_2$ bracket formula for a non-endomorphism graded operad with an associative operadic ideal.","tokens_in":58100,"feed_emoji":"🧮","tokens_out":14536,"duration_ms":125414,"temperature":0.7,"pith_summary":"This paper develops an obstruction theory for extending truncated minimal $A_\\infty$-algebra and $A_\\infty$-bimodule structures, and uses it to prove formality theorems for differential graded algebras together with their bimodules. The central goal is a cohomological criterion: if a newly introduced bimodule Hochschild cohomology vanishes in the range $HH^{n+2,-n}(A|M)=0$ for $n\\ge1$, then every DG algebra with cohomology $A$ and every compatible DG bimodule with cohomology $M$ is quasi-isomorphic to the pair $(A,M)$ with zero differential. This simultaneously extends the classical intrinsic formality theorem for graded algebras to algebra-bimodule pairs and gives the analogous statement for bimodules over a fixed algebra. The obstructions live in a truncated fringed spectral sequence whose second page is governed by $HH^{\\bullet,*}(A|M)$ and whose second differential is the Gerstenhaber bracket with a bimodule analogue of the universal Massey product.","feed_headline":"Cohomology test decides when algebra-bimodule pairs are formal","feed_subtitle":"A new bimodule Hochschild cohomology and a Massey-product differential make simultaneous formality testable.","key_machinery":"The load-bearing object is the bimodule Hochschild cochain complex $C^{n,r}(A|M)=C^{n,r}(A)\\oplus C^{n-1,r}_{A^e}(M,M)$, defined as the mapping cocone of $\\delta$. It is the operadic cochain complex of the linear endomorphism operad $E(A,M)$, whose operations are multilinear maps built from the graded algebra $A$ and the graded bimodule $M$; a morphism $A_\\infty\\to E(A,M)$ is exactly a compatible pair of an $A_\\infty$-algebra structure on $A$ and an $A_\\infty$-bimodule structure on $M$. The argument is carried by a truncated fringed spectral sequence—one whose near-diagonal terms are pointed sets or abelian groups—associated to the tower of mapping spaces $\\mathrm{Map}(A_{k+1},E(A,M))\\to \\mathrm{Map}(A_k,E(A,M))$: its second page is mostly $HH^{\\bullet,*}(A|M)$, and its second differential $d_2$ is the Gerstenhaber bracket with the bimodule universal Massey product. This spectral sequence converts the extension problem for truncated minimal structures into cohomology vanishing, and passage to the homotopy fibre over the projection $E(A,M)\\twoheadrightarrow E(A)$ yields the fibre-wise theory for bimodules over a fixed $A_\\infty$-algebra.","core_discovery":"On the paper's own terms, the main discovery is that the obstruction theory for truncated minimal $A_\\infty$-structures has a bimodule analogue controlled by the homotopy fibre of the cochain map $\\delta\\colon C^{\\bullet,*}(A)\\to C^{\\bullet,*}_{A^e}(M,M)$, $c\\mapsto \\mathrm{id}_M\\cdot c - c\\cdot \\mathrm{id}_M$. The cohomology $HH^{\\bullet,*}(A|M)$ of this fibre fits into a long exact sequence connecting the Hochschild cohomology of $A$ with the self-extensions of $M$, and it appears as the main term of the second page of a fringed spectral sequence built from towers of mapping spaces of DG operads. The second differential is bracketing with the bimodule universal Massey product $\\{\\{m^{A\\ltimes M}_3\\}\\}\\in HH^{3,-1}(A|M)$. From this, the paper derives intrinsic formality: vanishing of $HH^{n+2,-n}(A|M)$ for $n\\ge1$ forces every pair $(B,N)$ with cohomology $(A,M)$ to be quasi-isomorphic to $(A,M)$, and a refined Hochschild–Massey vanishing gives an almost formality theorem when the universal Massey product data are matched. The same machinery is formulated for arbitrary graded operads with multiplication and an associative operadic ideal, and in $d$-sparse versions for cohomology concentrated in degrees divisible by a fixed integer $d$.","pith_inferences":["The long exact sequence around $HH(A|M)$ suggests that simultaneous formality is genuinely separate from componentwise formality: the connecting map $\\delta$ measures a possible asymmetry of $\\operatorname{Ext}_{A^e}(M,M)$ as an $HH(A)$-bimodule, so the pair can be jointly non-formal even when the algebra and the bimodule are each separately formal.","The operadic formulation predicts that the same $E_2$-page description and the same $d_2$ bracket formula should hold for any graded operad with multiplication and an associative operadic ideal; this is testable by computing the spectral sequence for a non-endomorphism operad of this kind.","The $d$-sparse statements imply that for examples concentrated in even degrees, the first nonzero differentials appear only on later pages, so formality can be forced by vanishing in a smaller range of bidegrees; this would be the operative form of the criterion in geometric settings with even-degree cohomology."],"forward_implications":["If $HH^{n+2,-n}(A|M)=0$ for all $n\\ge1$, then every DG algebra $B$ with $H^*(B)\\cong A$ and every DG $B$-bimodule $N$ with $H^*(N)\\cong M$ is quasi-isomorphic to the pair $(A,M)$ with zero differential; this is a simultaneous intrinsic formality theorem for algebra-bimodule pairs.","If $\\operatorname{Ext}^{n+1,-n}_{A^e}(M,M)=0$ for $n\\ge1$, then every DG $A$-bimodule with cohomology $M$ is formal, giving an intrinsic formality criterion for bimodules over a fixed graded algebra.","For DG algebras and bimodules that are not formal, the almost formality criterion uses the bimodule universal Massey product: if the bimodule Hochschild–Massey cohomology $HM^{n+2,-n}(H^*(A)|H^*(M),\\{\\{m^{A\\ltimes M}_3\\}\\})$ vanishes for $n>1$, then any pair with the same cohomology and the same bimodule universal Massey product is quasi-isomorphic.","The $d$-sparse variants say that when cohomology is concentrated in degrees divisible by $d$, the same classification holds under the analogous vanishing for $n>d$, with the universal Massey product of length $d+2$ playing the role of the cubic operation."],"supporting_citations":[{"why":"Supplies the fringed spectral sequence and the A-infinity obstruction theory that Theorem 4.1.5 compiles and generalizes from endomorphism operads to arbitrary graded operads with multiplication.","marker":"[Mur20]"},{"why":"Provides the Hochschild-cohomology intrinsic formality theorem for graded algebras that is extended to pairs and bimodules.","marker":"[Kad88]"},{"why":"Introduces the linear endomorphism DG operad E(V,W) and its operadic ideal, the language used to encode compatible A-infinity bimodule structures.","marker":"[BM09]"},{"why":"Provides the fringed spectral sequence and convergence results for towers of fibrations on which the obstruction theory and formality proofs rely.","marker":"[BK72]"},{"why":"Establishes the Hochschild–Massey cohomology and the almost formality theorem for DG algebras that the present paper generalizes to algebra-bimodule pairs.","marker":"[JKM22]"},{"why":"Records the classical Hochschild obstructions to extending truncated minimal A-infinity algebra structures that the new theory refines.","marker":"[Lef03]"}],"fun_headline_variants":["Bimodule formality via new cohomology theory","Obstruction theory for A-infinity bimodules decoded","New cohomology predicts algebra-bimodule formality","Spectral sequence decides bimodule formality","Massey product differential gates bimodule formality"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is Theorem 4.1.5: the spectral-sequence obstruction theory previously developed for endomorphism operads is asserted, without a fully written-out proof, to transfer to arbitrary graded operads with a multiplication and an associative operadic ideal. If that transfer fails, the obstruction theory and the formality theorems built on it collapse.","fun_headline_variants_meta":{"raw":{"variants":["Bimodule formality via new cohomology theory","Obstruction theory for A-infinity bimodules decoded","New cohomology predicts algebra-bimodule formality","Spectral sequence decides bimodule formality","Massey product differential gates bimodule formality"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000501,"raw_usage":{"total_tokens":2584,"prompt_tokens":1213,"completion_tokens":1371,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":829,"completion_tokens_details":{"reasoning_tokens":1289}},"tokens_in":829,"tokens_out":1371,"duration_ms":10820,"temperature":1.0,"reasoning_tokens":1289,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:45:17.818675+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find a graded algebra $A$ and a graded $A$-bimodule $M$ with $HH^{n+2,-n}(A|M)=0$ for all $n\\ge1$ but with two minimal $A_\\infty$-pairs over $(A,M)$ that are not gauge isomorphic; equivalently, compute the spectral sequence of Theorem 4.1.5 for that pair and exhibit a nonzero obstruction on a later page. Alternatively, test the asserted generality by computing the claimed $E_2$-page description and $d_2$ bracket formula for a non-endomorphism graded operad with an associative operadic ideal.","supporting_citations":[],"review_version":1}