{"id":"991da127-0679-4dea-b3e0-9196891300e7","arxiv_id":"2411.17359","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Constructs a trivial extension DG-algebra and a unitally positive A∞-algebra, yielding a direct proof that smooth 3-fold flops are classified by their contraction algebras.","lead":"For any module that repeats periodically, this paper builds two new algebraic objects: a DG-algebra T and an A∞-algebra N, then uses them to prove that smooth 3-fold flops are classified by their contraction algebras. The work gives a direct and general proof of the Donovan-Wemyss conjecture, a central result relating birational geometry to noncommutative algebra.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"The Reader's verdict ACCEPT with moderate confidence is appropriate. The weakest assumption in the paper is indeed the Setup 6.1/7.1 bridge via Lemma 7.1, because if the structural facts about contraction algebras were misapplied, the reconstruction of End_A(Q) from A_con would break. However, the paper's use of these facts is consistent with the literature: contraction algebras are symmetric, 3-CY duality applies to the NCCR, sphericity gives the Ext-vanishing pattern, and four-periodicity supplies the identification Ω^4S_j ≅ S_j. The sign-heavy constructions in §2–§3 were checked for internal consistency; Proposition 2.13, Theorem 3.6, and Theorem 3.8 are supported by explicit computations, and Theorem 6.6's dimension count is valid after the cohomological identifications are accepted. I therefore see no load-bearing concern that would justify changing the verdict. The proposed concrete test would provide independent confirmation of the most cited-dependent step, but it is a verification of standard results rather than a response to a detected flaw.","tokens_in":33273,"tokens_out":42878,"duration_ms":379545,"concrete_test":"Independently compute Tor^A_*(S_j,A_con) for a non-Atiyah flopping contraction, e.g. the cDV singularity with known NCCR and contraction algebra, by taking the minimal projective resolution Q_j of S_j over A, tensoring with A_con, and checking that H^0 and H^3 are each one-dimensional and isomorphic to S_j while H^1=H^2=0. This directly verifies Lemma 7.1 and the existence of the 4-periodic splicing used in Setup 6.1, settling the weakest cited step.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After close reading, I did not find an internal inconsistency or a concrete gap in the central argument. The proof of the Donovan-Wemyss conjecture rests most heavily on Lemma 7.1 and the surrounding Setup 6.1/7.1, which assert that for each simple S_j of the contraction algebra, the complex P_j = i^*Q_j has cohomology only in degrees 0 and 3, and that the resulting Ω^4S_j is isomorphic to S_j. This is exactly the step the Reader flagged. The cited structural facts — A_con symmetric ([A,3.3]), A 3-CY, A_con spherical ([DW3,4.7]), and four-periodicity ([D,1.1]) — are standard in the flops literature, and the chain of isomorphisms in Lemma 7.1 is internally coherent. The use of four-periodicity to identify the one-dimensional Ω^4S_j with S_j is legitimate: a four-periodic finite-dimensional algebra has Ω^4 ≅ id in the stable category, and with Ω^4S_j one-dimensional and non-projective, the stable isomorphism upgrades to an actual isomorphism. The later A∞-reconstruction Theorem 6.6 and the categorical upgrade in §8 are built on verified cohomology computations and dimension-counting arguments; I found no place where an identity or quasi-isomorphism is asserted without sufficient support. Thus the residual risk is not a detected error but the known reliance of the method on external results about contraction algebras.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces two general derived/homotopical objects: a trivial extension DG-algebra T associated to any periodic module over an algebra, and a 'unitally positive' A∞-algebra N extracted from the positive cohomology of a unital DG-category. The authors prove that T is quasi-isomorphic to the endomorphism DG-algebra End_Γ(P) of the periodic projective resolution, compute the cohomology of T, and show that N is well defined up to A∞-quasi-isomorphism. The central application is to birational geometry: under the flops setup, the unitally positive A∞-algebra associated to the contraction algebra recovers the DG-endomorphism algebra of the tilting NCCR resolution, which via known results of Booth, Hua–Keller, and Kalck–Yang yields the Donovan–Wemyss conjecture for smooth 3-fold flops, both in the single-curve case (Corollary 1.7) and the multi-curve case (Theorem 8.13). The proof is explicit, with detailed sign conventions, and the main reconstruction theorem is based on cohomology computations and finite-dimensionality rather than on circular reasoning.","tokens_in":10,"tokens_out":7558,"duration_ms":137426,"significance":"If correct, this is a substantial result: it gives a second, direct proof of the Donovan–Wemyss conjecture and, more importantly, introduces constructions of independent utility. The trivial extension DG-algebra T is built from only a single periodic module, and the unitally positive A∞-algebra N is defined for very general DG-categories, so the paper provides new tools beyond the specific birational application. The proof is notably self-contained in its main steps: Theorem 6.6 compares cohomology groups by explicit quasi-isomorphisms and dimension counts, and the final geometric conclusion rests on standard external theorems about derived contraction algebras and singularity categories. The authors are also careful to note the relationship with the independent proof by Jasso–Muro, emphasizing that the two approaches have essentially no technical overlap. The main strength of the paper is the clarity and explicit nature of the constructions; no fitted parameters or post-hoc assumptions appear in the argument.","major_comments":[],"minor_comments":[{"comment":"The displayed definitions of σ and τ appear identical, although the text immediately asserts |σ| = n and |τ| = −n and uses τ∘σ = Id_{≥n}. The arrow directions or signs in the display for τ should be corrected so that the displayed object matches its stated degree and properties.","section":"Notation 2.4"},{"comment":"The statement and proof contain a garbled arrow `←/mapsfromcharg` and the typo `Z-modues`; these should be fixed for readability.","section":"Theorem 3.6"},{"comment":"Setup 6.1 states that S is a finite-dimensional simple A-module, but Theorem 6.6 uses in equation (6.C) the fact that H^0(End_{A_con}(P)) = Hom_{A_con}(i^*S, i^*S) is one-dimensional. This requires that i^*S is simple as an A_con-module (or at least that its endomorphism ring is C). The flops application satisfies this, but the general statement of Setup 6.1 should include this hypothesis explicitly.","section":"Setup 6.1 / Theorem 6.6"},{"comment":"In equations (8.A) and (8.D) the cohomological degree is denoted inconsistently: the text uses Ext^i_{A_con}(S_i,S_j) where the displayed range is in k, and the same notational confusion appears in (8.B). Please make the degree variable uniform.","section":"Equations (8.A)–(8.D)"},{"comment":"The proof invokes that A is 3-CY without a specific citation; since this is a standard property of NCCRs, a reference (for example to Van den Bergh or Iyama–Reiten) would be helpful for the reader.","section":"Lemma 7.1"}],"recommendation":"minor_revision","confidential_remarks":"The paper is strong and the central argument appears sound. The main caveat is the reliance on external results about contraction algebras and singularity categories, but those are standard and clearly cited. The requested changes are local: correcting typographical/display errors and adding a missing hypothesis in Setup 6.1. I do not see a need for further mathematical revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know about this paper is that it works. It gives a second proof of the Donovan–Wemyss conjecture, but it is not a repackaged version of Jasso–Muro: the trivial extension DG-algebra T and the unitally positive A∞-algebra N are genuinely new, and they are built in enough generality to apply beyond flops. The proofs are careful, the sign conventions are managed explicitly rather than hand-waved, and the authors are honest about the relation to the earlier proof. That honesty matters, because the conjecture was already known; what is new is the framework and the directness, not the first resolution.\n\nThe paper earns its keep in three places. First, T provides an explicit DG model for End(P) whenever a module is periodic, and the cohomology computation in §3 is clean. Second, the unitally positive A∞-algebra N is a simple but useful idea: it cuts away negative cohomology and unwanted degree-zero summands, and Proposition 5.7 shows it is well defined up to quasi-isomorphism. Third, the reconstruction theorem 6.6 does the heavy lifting without assuming the conjecture, and the categorical upgrade in §8 extends the argument to the multi-curve case. I could not find a circular step or a fitted parameter.\n\nThe soft spots are proportionate. The main structural input is Setup 6.1/7.1, where the complex P_j has cohomology only in degrees 0 and 3. That relies on standard facts about contraction algebras—symmetry, sphericity, four-periodicity—so the argument is only as robust as those facts. They are well-established in the flops literature, so this is a dependency, not a flaw. The final step back to the base ring also leans on Booth and Hua–Keller; the paper does not re-prove those recovery theorems, but it cites them correctly. The only real complaint is that the sign-heavy A∞-machinery makes the paper hard to check thoroughly, though I found no actual error. The stress-test note agrees: no internal inconsistency surfaced.\n\nWho should read this? Anyone working on flops, contraction algebras, or derived reconstruction, and anyone interested in new constructions of DG-algebras from periodic modules. It deserves a serious referee. I would send it to review and ask for a careful check of §2 and the dimension-counting in 6.6, but I would not desk-reject it. I would probably cite the T and N constructions if I did related work, but I would not assign this paper the status of the first proof of the conjecture.","headline":"A well-executed second proof of Donovan-Wemyss via two genuinely new general constructions; the residual risk sits in standard flops facts and external recovery theorems, not in the paper's own logic.","tokens_in":34058,"tokens_out":1865,"would_cite":true,"duration_ms":21522,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14E30","14J30","16E45","16S38"],"pacs":[],"model":"deepseek-v4-flash","headline":"A new trivial-extension DG-algebra and a unitally positive A-infinity algebra prove that smooth 3-fold flops are classified by their contraction algebras.","keywords":["trivial extension DG-algebra","unitally positive A-infinity algebra","Donovan-Wemyss conjecture","3-fold flops","contraction algebras","periodic modules","noncommutative crepant resolutions","A-infinity minimal models"],"falsifier":"The conjecture would fall if two smooth one-curve 3-fold flops over complete local rings had isomorphic contraction algebras but non-isomorphic base rings; a more local check is to find a smooth 3-fold flop whose contraction algebra's simple module has nonzero Tor in a degree other than 0 or 3, which would break Lemma 7.1 and Setup 6.1.","tokens_in":33092,"feed_emoji":"🔁","tokens_out":11174,"duration_ms":102050,"temperature":0.7,"pith_summary":"This paper proves the Donovan–Wemyss conjecture: smooth 3-fold flopping contractions with one curve over a complete local base are classified, up to isomorphism, by their contraction algebras. The proof builds two new derived objects from arbitrary periodic modules: a trivial extension DG-algebra T, and a unitally positive A-infinity algebra N that keeps only the identity and the positive cohomology. In the flop setting, N reconstructs the endomorphism DG-algebra of the resolution's tilting bundle from the finite-dimensional contraction algebra alone. From that bridge, known results on derived contraction algebras and singularity categories force the base rings to be isomorphic whenever the contraction algebras are. The same machinery also gives the multi-curve version of the conjecture.","feed_headline":"Contraction algebras classify 3-fold flops","feed_subtitle":"A new DG-algebra and A-infinity invariant rebuild a flop's local ring from its contraction algebra.","key_machinery":"The load-bearing construction has two stages. First, given any module with a length-n periodic projective resolution, the paper forms the trivial extension DG-algebra T by taking the endomorphism DG-algebra of the periodic complex and adding a shifted bimodule copy of it, with an explicit multiplication and differential; this algebra is quasi-isomorphic to the more homological endomorphism DG-algebra and its cohomology is a window of Hom and Ext groups. Second, for any unital DG-algebra it passes to a strictly unital minimal model and restricts to the span of the identity plus all positive cohomology, producing the unitally positive A-infinity algebra N. The critical mechanism in the flop case is the A-infinity isomorphism between the endomorphism DG-algebra of the crepant resolution's resolution and N; once this bridge is in place, the classification follows by taking Koszul duals and applying the singularity-category recovery result.","core_discovery":"On its own terms, the paper's central discovery is an A-infinity reconstruction theorem: the strictly unital minimal model of the endomorphism DG-algebra of the noncommutative crepant resolution's resolution Q is A-infinity-isomorphic to the unitally positive A-infinity algebra built from the contraction algebra. In the flops setting the contraction algebra is symmetric, spherical, and four-periodic, so the complex obtained by tensoring Q with the contraction algebra is a length-four periodic projective resolution of the simple module. The positive cohomology of its endomorphism DG-algebra, together with the identity, determines the full A-infinity structure. Therefore two flops with isomorphic contraction algebras have A-infinity-quasi-isomorphic derived endomorphism algebras, and after Koszul duality and the known recovery theorem for singularity categories, their base rings are isomorphic.","pith_inferences":["Beyond the flop setting, the same recipe could be applied to any module with a periodic projective resolution; the unitally positive A-infinity algebra could serve as a general invariant that detects whether two periodic modules have quasi-isomorphic endomorphism DG-algebras.","The paper leaves open the computational question of whether the higher A-infinity products of N carry information invisible in the cohomology groups; if they do, they would distinguish flops with identical Ext-dimension vectors.","If Setup 6.1 holds in other dimensions or for non-smooth contractions, the same bridge would give analogous reconstruction theorems; the four-periodicity of contraction algebras is the geometric input that would need a counterpart there."],"forward_implications":["When the paper's proof is right, two smooth one-curve 3-fold flops over complete local rings have isomorphic base rings exactly when their contraction algebras are isomorphic, so the entire local geometry is encoded in a finite-dimensional algebra.","In the multi-curve case, the base ring is recovered from the contraction algebra up to iterated mutation, and isomorphic base rings are equivalent to derived equivalence of the contraction algebras' module categories.","The full Ext-algebra of the simple modules over the noncommutative crepant resolution is recoverable from the positive Ext-algebra of the contraction algebra together with the unit, so the derived endomorphism ring is an invariant of the contraction algebra.","Any periodic module over any algebra now yields an explicit DG-algebra, so periodic homological phenomena admit a universal differential graded model."],"supporting_citations":[{"why":"Introduces contraction algebras for 3-fold flops and proves that isomorphic bases give isomorphic contraction algebras.","marker":"[DW1]"},{"why":"Constructs the tilting bundle and the noncommutative crepant resolution used to set up the module theory.","marker":"[V]"},{"why":"Shows contraction algebras are spherical, which gives the Tor vanishing used in Setup 7.1.","marker":"[DW3]"},{"why":"Shows contraction algebras are four-periodic, which yields the length-four periodic resolution of the simple module.","marker":"[D]"},{"why":"Shows contraction algebras are symmetric, a fact used in the Tor calculation of Lemma 7.1.","marker":"[A]"},{"why":"Connects the derived contraction algebra to singularity categories, used in the final step of the classification.","marker":"[B]"},{"why":"Provides the recovery theorem that the singularity category of the base ring determines the ring.","marker":"[HK]"},{"why":"Relates relative singularity categories and derived deformation algebras, used when passing from A-infinity quasi-isomorphisms to geometry.","marker":"[KY]"},{"why":"Gives the criterion that quasi-isomorphic bounded complexes of projectives have quasi-isomorphic endomorphism DG-algebras.","marker":"[ST]"}],"fun_headline_variants":["Contraction algebra: complete flop invariant","3-fold flops classified by contraction algebras","A-infinity proof: contraction algebras classify flops","Flop classification via contraction algebras"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that for every smooth 3-fold flop, the simple module over the contraction algebra has a length-four periodic projective resolution obtained from the crepant resolution, with all cohomology vanishing outside degrees 0 and 3.","fun_headline_variants_meta":{"raw":{"variants":["Contraction algebra: complete flop invariant","3-fold flops classified by contraction algebras","A-infinity proof: contraction algebras classify flops","Flop classification via contraction algebras"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000799,"raw_usage":{"total_tokens":3464,"prompt_tokens":846,"completion_tokens":2618,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":462,"completion_tokens_details":{"reasoning_tokens":2575}},"tokens_in":462,"tokens_out":2618,"duration_ms":21185,"temperature":1.0,"reasoning_tokens":2575,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:15:03.617146+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The conjecture would fall if two smooth one-curve 3-fold flops over complete local rings had isomorphic contraction algebras but non-isomorphic base rings; a more local check is to find a smooth 3-fold flop whose contraction algebra's simple module has nonzero Tor in a degree other than 0 or 3, which would break Lemma 7.1 and Setup 6.1.","supporting_citations":[],"review_version":1}