{"id":"2456e747-366f-4397-9cac-6b44d02a8825","arxiv_id":"2506.22847","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The category of strictly unital A∞-categories with strict morphisms admits a cofibrantly generated model structure with quasi-equivalences as weak equivalences.","lead":"This paper proves that the category of strictly unital A∞-categories with strict morphisms carries a cofibrantly generated model structure. This gives a precise notion of cofibrant A∞-category and shows semi-free A∞-categories are cofibrant, resolving an open question.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Pushout formula (14) used for condition 4 is unproved and, as written, omits the ℳ(x,y) summand, so the inclusion inc is undefined; the proof of J-cell ⊆ W rests on this computation.","rationale":"The reader's weakest-assumption analysis identifies the same spot: the proof of condition 4 depends on an unproved and possibly incorrect pushout computation. My stress-test confirms this is the most load-bearing point. The displayed formula (14) is not merely incomplete but has a concrete m = 0 defect: it would make inc undefined. A correct pushout formula should include ℳ(x,y) and alternating paths through the contractible reduced endomorphism complex; with that fix, the quasi-isomorphism claim for inc becomes plausible. However, the paper does not prove that the pushout in A∞Catstrict is computed by such a formula, nor does it address how the A∞-relations of ℳ interact with the amalgamated tensor algebra. This is not a disagreement with the general strategy or with the known model structure on DG-categories; it is a gap in a central verification. The concern is fixable by a detailed universal-property computation, so the appropriate recommendation remains the conditional verdict given by the reader rather than outright rejection. I do not see a separate, more severe objection that would change the verdict: the smallness conditions are asserted but plausible, and the characterization of fibrations, while terse, follows the same pattern as Tabuada's DG-category model structure, provided the cited semi-free lifting lemma [15, Lemma 6.6] is valid as used.","tokens_in":9316,"tokens_out":19855,"duration_ms":218519,"concrete_test":"Let ℳ be the strictly unital A∞-category with objects x, z, y and hom-complexes freely generated by a ∈ ℳ(x,z), b ∈ ℳ(z,y), and c ∈ ℳ(x,y), with composition relation m2(b,a) = c. Compute the pushout 𝒫 of F′: 𝒜 → K_A∞ along the functor sending the unique object 3 to z, using the universal property of strict A∞-functors (i.e., characterize strict functors 𝒫 → 𝒩 in terms of pairs of strict functors ℳ → 𝒩 and K_A∞ → 𝒩 agreeing on 𝒜). Check whether 𝒫(x,y) contains ℳ(x,y) as a direct summand and whether the m = 0 term of formula (14) is ℳ(x,y) or ℳ(z,y) ⊗ ℳ(x,z). If it is the latter, formula (14) fails as written. Then determine whether the corrected formula also identifies c with the formal path b ⊗ a; if such an identification occurs, the pushout is not the naive free tensor product and condition 4 requires a different proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The main risk is the verification of condition 4 of the Recognition Theorem 2.1. To prove J-cell ⊆ W, the proof must show that for every pushout of a generating trivial cofibration F′: 𝒜 → K_A∞ (and similarly R(n)) along an arbitrary strict functor, the induced map inc: ℳ(x,y) → 𝒫(x,y) is a quasi-isomorphism. The argument depends entirely on the claimed pushout formula (14), which is asserted without proof and, as printed, cannot be correct: its m = 0 summand is displayed as ℳ(z,y) ⊗ ℳ(x,z), so 𝒫(x,y) does not contain ℳ(x,y) at all, and the map inc is not even defined. The intended formula must include ℳ(x,y) as the m = 0 term and then add alternating paths with at least one factor of the contractible complex K̄_A∞(z,z); with that correction inc would be an inclusion of a direct summand with contractible cokernel. But the actual pushout in A∞Catstrict is not obviously given by such a free tensor-product formula: it must also respect the A∞-operations and relations of ℳ, and strictness of the functors makes the amalgamation more delicate than a free tensor algebra. The proof cites [15] for this computation rather than verifying the universal property. Since condition 4 is load-bearing for the existence of the model structure, an incorrect or unjustified pushout computation would invalidate Theorem 3.1 as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to construct a cofibrantly generated model structure on A∞Catstrict, the category of strictly unital A∞-categories with strict A∞-functors over a commutative ring R, whose weak equivalences are quasi-equivalences. The proposed generating cofibrations are the DG-category generators Q and S(n) adapted to A∞-categories, and the generating trivial cofibrations are F′: 𝒜→K_A∞ and R(n). The proof invokes Hovey's Recognition Theorem and reduces the main work to verifying condition 4, namely that J′-cell maps are weak equivalences. The paper also derives consequences: all objects are fibrant, fibrations are degreewise-surjective isofibrations, and cofibrant A∞-categories have cofibrant hom-spaces, hence are cofibrant in the sense of the semi-free resolution framework.","tokens_in":9612,"tokens_out":5355,"duration_ms":63478,"significance":"If the main theorem is correct, it gives a genuine Quillen model structure on a category of A∞-categories with strict morphisms, a result that has been missing in the literature and would justify the terminology 'cofibrant A∞-category' in the sense of the author's earlier work. The paper is well organized and the overall strategy, using the Recognition Theorem and adapting the DG-category generators, is natural. The lifting characterizations in Lemma 3.2 are a useful and reasonably explicit part of the argument, and the connection to semi-free resolutions is a compelling motivation. However, the proof of the central condition 4 is not completed in the manuscript: the pushout formula on which it relies is asserted rather than proved and, as printed, is not even well formed. Because this gap is load-bearing, the main theorem cannot be accepted on the basis of the present text.","major_comments":[{"comment":"The displayed pushout formula (14) is load-bearing for condition 4 of the Recognition Theorem, but as printed it cannot be correct. For m=0 the formula gives 𝒫(0)(x,y)=ℳ(z,y)⊗ℳ(x,z), so the chain complex ℳ(x,y) does not occur as a summand and the map inc: ℳ(x,y)→𝒫(x,y) is not even defined. Even if one inserts the evident correction (ℳ(x,y) for the m=0 term and alternating contractible factors for m≥1), the assertion that this tensor-product formula computes the pushout in A∞Catstrict is not proved and is not immediate: the pushout must respect the A∞ operations and strictness of the maps, not merely describe the underlying graded modules. Since the quasi-isomorphism of inc is exactly what is needed to prove J′-cell⊂𝒲, Theorem 3.1 is not established as written.","section":"§3, Eq. (14)"},{"comment":"The same defect occurs in the pushout computation for the generating trivial cofibration R(n) after Eq. (16). The displayed formula again has an m=0 term of the form ℳ(T(5),y)⊗ℳ(x,T(4)), rather than ℳ(x,y), so the inclusion inc is not defined from the stated formula. The contractibility of 𝔻ₙ alone does not repair this: one first needs a correct description of the actual pushout in A∞Catstrict, and then a proof that the map on hom-complexes is a quasi-isomorphism. This part of condition 4 is therefore also unsupported.","section":"§3, pushout for R(n)"},{"comment":"The proof states 'It is easy to verify item 1. 2. and 3.' of Theorem 2.1, but conditions 2 and 3 require the domains of I and J′ to be small relative to I-cell and J′-cell respectively. No argument is supplied, and in A∞Catstrict this is not a formality: the relevant colimits are colimits of A∞-categories with strict morphisms, which are not computed simply as colimits of underlying chain complexes. A proof or a precise citation for smallness is needed.","section":"§3, proof of Theorem 3.1"},{"comment":"The computation displayed at the end of the construction of K_A∞ is self-contradictory. The text claims that m₁((r₁,f)−(f,r₂)) equals (f,g,f)−(f,g,f), and then writes '≠ 0'. As written, the right-hand side is 0. If a sign convention is intended that prevents cancellation, the signs must be displayed; if the terms do cancel, then the claimed need for the extra generator r₁₂ in the A∞ case is not demonstrated. This point is not merely cosmetic, since it is part of the justification for the chosen generating trivial cofibration F′.","section":"§3, definition of K_A∞"}],"minor_comments":[{"comment":"Corollary 3.3 refers to 'Lemma 3.1', but the relevant statement appears to be Lemma 3.2; the cross-reference should be corrected.","section":"§3, Corollary 3.3"},{"comment":"Definition 3.2 requires ℱ₁ to be a surjective quasi-isomorphism of complexes, but Lemma 3.2 and the later characterization of fibrations only use degreewise surjectivity of ℱ₁. The definition should be aligned with the statements in which it is used.","section":"§3, Definition 3.2"},{"comment":"In proving that H⁰(inc) is essentially surjective, the text says that N′(2) is quasi-isomorphic to N′(1)=z. This requires an explicit quasi-isomorphism in H⁰(𝒫) between the new object and the image object; the argument would be clearer if that morphism were identified.","section":"§3, essential surjectivity after Eq. (14)"},{"comment":"Reference [20] is listed with the arXiv identifier math/0310337, which is the same as the identifier given for [12]; this is likely a typo in the bibliography.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The main theorem is plausible and the paper's strategy is sensible, but the verification of condition 4 of the Recognition Theorem is currently a genuine gap, not just a missing detail. The author should either prove the pushout formula in A∞Catstrict from the universal property or supply a different argument that J′-cell maps are weak equivalences. The manuscript also relies heavily on the author's own prior work, especially [15] for completeness/cocompleteness and for properties of semi-free resolutions; the editor may wish to ensure that those cited results are indeed established and publicly available in the quoted form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear X,\n\nThe paper proves a model structure on A∞Catstrict with quasi-equivalences as weak equivalences, answering an open question from the author's earlier IMRN paper. The strategy is sensible: adapt Tabuada's DG-category model structure and Hinich's A∞-algebra case, with K_A∞ replacing the DG Kontsevich category. The statement of fibrations and cofibrant objects are what you'd expect. This is a real result, if the proof holds.\n\nWhat's new: the model structure itself, and the corollary that semi-free A∞-categories are cofibrant. That last point matters for Fukaya-category practice. The author knows the surrounding literature, and the paper is clearly written.\n\nThe soft spots are all in the proof of condition 4 of the Recognition Theorem, which is the load-bearing part. The pushout formula (14) for 𝒫(x,y) is asserted without derivation, and as printed it cannot be right: the m=0 summand is ℳ(z,y)⊗ℳ(x,z), which omits ℳ(x,y) entirely, so the map inc is not even defined. The intended formula should have ℳ(x,y) as the base term. More importantly, the formula treats the pushout as a free tensor product over the glued object. That may be true for strict A∞-categories (semi-free generation), but it needs proof: the A∞ operations of ℳ have to be respected, and that is not automatic. The paper cites [15] for this, but the universal property is not verified. Also, smallness of the domains for conditions 2 and 3 is just asserted as 'easy to verify,' and Theorem 3.4 is dismissed as 'the same as in DG-categories,' which needs an argument because the underlying objects are not DG-categories. There is also a typo in the m1 computation: m1 of the combination equals (f,g,f) − (f,g,f) and then says ≠ 0; that should be = 0, unless the expression is different.\n\nNone of these look irreparable. The central strategy is credible, and the author's previous work supplies the completeness/cocompleteness and semi-free machinery. The paper needs a serious referee and a revised version with the pushout computation done in full, the smallness arguments spelled out, and the typos fixed.\n\nFor you: if you work on homotopy theory of A∞-categories, this is worth a read and worth citing once the gap is closed. I'd send it to peer review.","headline":"First model structure on strict A∞-categories: right result, credible strategy, but the proof of the key pushout claim is sketched to the point of being a gap.","tokens_in":10112,"tokens_out":5219,"would_cite":true,"duration_ms":68070,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14F08","18G70","18N40"],"pacs":[],"model":"deepseek-v4-flash","headline":"The author proves that the category of strictly unital A∞-categories with strict morphisms carries a cofibrantly generated model structure, making semi-free A∞-categories cofibrant and every object fibrant.","keywords":["A∞-categories","model categories","Quillen model structure","quasi-equivalences","semi-free resolutions","strict A∞-functors","DG-categories","cofibrant objects"],"falsifier":"Run the pushout computation of (13) for a minimal test case—say $\\mathcal{M}$ the one-object A∞-category with hom-complex $(R,0)$—and verify the universal property of the pushout term by term; at the same time compute $H^n(K_{A_\\infty}(1,1)/R\\cdot 1_1)$ directly from the generators and the differential $m^1$. A nonzero cohomology group, or a mismatch with formula (14), would show that condition 4 fails and Theorem A is false.","tokens_in":9119,"feed_emoji":"∞","tokens_out":19312,"duration_ms":187791,"temperature":0.7,"pith_summary":"A∞-categories are structures in which composition is associative only up to a coherent hierarchy of homotopies, and they are studied over a commutative ring in homological algebra and symplectic geometry. The full category of A∞-categories with all A∞-functors was known to be badly behaved: it is not complete, since it lacks equalizers. This paper proves that the restriction to strict A∞-functors, which preserve the higher composition maps on the nose, fixes the obstruction: the category $\\mathrm{A_\\infty Cat}_{\\mathrm{strict}}$ carries a cofibrantly generated model structure. In this structure the weak equivalences are the quasi-equivalences, every object is fibrant, and the fibrations are surjective isofibrations; moreover every cofibrant object has cofibrant hom-spaces in $\\mathrm{Ch}(R)$. The payoff is that semi-free A∞-categories, previously used as resolutions, are now proven cofibrant, giving the informal term 'cofibrant A∞-category' a precise model-theoretic meaning.","feed_headline":"Strict A∞-categories gain a model structure","feed_subtitle":"Every category is fibrant; semi-free resolutions become cofibrant, giving a precise homotopy theory for A∞-categories.","key_machinery":"The proof is built on the Recognition Theorem for cofibrantly generated model structures. The generating cofibrations are the same as those of the standard DG-category model structure; the generating trivial cofibrations are an inclusion $F'\\colon \\mathcal{A}\\to K_{A_\\infty}$ together with the inclusions $R(n)\\colon \\mathcal{B}\\to \\mathcal{P}(n)$. Here $K_{A_\\infty}$ is a two-object semi-free A∞-category, a resolution of the category with two objects and two mutually inverse morphisms, generated by closed morphisms $j_{12}$, $j_{21}$ together with two degree-one generators whose differential $m^1$ makes the two morphisms inverse up to homotopy. The decisive step is condition 4 of the Recognition Theorem, which requires pushouts of generating trivial cofibrations to be quasi-equivalences. For the pushout along $F'$, the paper shows that the hom-spaces of the glued category are given by the tensor-product formula (14), in which the quotient complex $K_{A_\\infty}(1,1)/R\\cdot 1_1$ appears as a factor; because $K_{A_\\infty}(1,1)$ is homotopy equivalent to $(R,0)$ and the unit splits, this quotient is contractible, and so the inclusion into the pushout is a quasi-isomorphism.","core_discovery":"The central claim is Theorem A: the category $\\mathrm{A_\\infty Cat}_{\\mathrm{strict}}$ of strictly unital A∞-categories over a commutative ring $R$, with strict A∞-functors, admits a cofibrantly generated model structure whose weak equivalences are the quasi-equivalences. Fibrations are exactly the strict A∞-functors that are isofibrations and surjective on morphisms, and every object is fibrant. If $\\mathcal{A}$ is cofibrant, then each hom-object $\\mathcal{A}(x,y)$ is a cofibrant object in the category of unbounded chain complexes over $R$, hence in particular h-projective. The paper then deduces that the semi-free A∞-categories are cofibrant, and that the semi-free resolutions constructed in earlier work are cofibrant resolutions, so every A∞-category has a cofibrant replacement whose hom-spaces are cofibrant. This gives the previously informal term 'cofibrant A∞-category' a precise meaning.","pith_inferences":["Because every object is fibrant, the semi-free cofibrant replacement of an object is simultaneously a fibrant replacement; the author does not spell this out, but it means the homotopy category of $\\mathrm{A_\\infty Cat}_{\\mathrm{strict}}$ can be modelled entirely by semi-free categories.","The tensor-product formula (14) is concrete enough to serve as a recipe: for any category obtained by gluing along a semi-free resolution, hom-spaces reduce to tensor products with the contractible quotient $K_{A_\\infty}(1,1)/R\\cdot 1_1$, so the model structure can be checked on explicit examples by direct computation.","By analogy with the homological algebra of DG-categories, where cofibrant objects with cofibrant hom-spaces play a central role, this model structure is a natural foundation for a derived representation theory of A∞-categories; the paper stops once the model structure is established, but the explicit cofibrant objects make that extension a plausible next step.","The same Recognition Theorem argument might adapt to other settings with strict morphisms, such as strictly unital algebras over other algebraic operads, whenever the relevant pushouts are known to exist and the analogous quotient complexes are contractible."],"forward_implications":["Semi-free A∞-categories are cofibrant objects, and the semi-free resolutions previously used to compute homotopy categories are cofibrant resolutions, so every A∞-category has a cofibrant replacement by a semi-free category.","Since every object is fibrant, any cofibrant replacement is automatically a fibrant-cofibrant replacement; no separate fibrant replacement step is needed.","Cofibrant A∞-categories have cofibrant hom-objects in the category of unbounded chain complexes over $R$, and hence h-projective hom-spaces, so the term 'cofibrant A∞-category' now has a model-theoretic justification rather than being only an analogy with DG-categories.","Fibrations are characterized explicitly as strict isofibrations surjective on morphisms, so right lifting properties and the associated homotopy theory of $\\mathrm{A_\\infty Cat}_{\\mathrm{strict}}$ can be computed concretely.","Because the model structure is cofibrantly generated and the category is complete and cocomplete, standard Quillen model-categorical machinery such as homotopy limits, homotopy colimits, and derived functors applies to $\\mathrm{A_\\infty Cat}_{\\mathrm{strict}}$."],"supporting_citations":[{"why":"Supplies the Recognition Theorem, the criterion by which the model structure is verified.","marker":"[10]"},{"why":"Supplies the generating cofibrations for the DG-category model structure, which the paper reuses as the set $I$.","marker":"[19]"},{"why":"Used to prove that the class of surjective strict A∞-functors equals the right-injective class determined by the generating cofibrations.","marker":"[20]"},{"why":"Provides the semi-free A∞-categories and resolutions, the completeness and cocompleteness of the strict-morphism category, and the contractibility result for $K_{A_\\infty}(1,1)/R\\cdot 1_1$.","marker":"[15]"},{"why":"Provides the statement that cofibrant DG-categories have cofibrant hom-spaces, which the paper adapts to prove the same for cofibrant A∞-categories.","marker":"[22]"},{"why":"Establishes that the full category of A∞-categories lacks equalizers, the obstruction that motivates working with strict morphisms.","marker":"[2]"}],"fun_headline_variants":["Strict A∞-categories get model structure","A∞-categories: strict morphisms yield model structure","Cofibrant A∞-categories finally precise","Every A∞-category has a cofibrant replacement"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction rests on two assumptions: that pushouts of the generating trivial cofibration $F'$ are computed by the tensor-product formula (14), and that the quotient complex $K_{A_\\infty}(1,1)/R\\cdot 1_1$ is contractible, a fact cited to earlier work rather than proved here. If either of these fails, condition 4 of the Recognition Theorem, and with it the whole model structure, collapses.","fun_headline_variants_meta":{"raw":{"variants":["Strict A∞-categories get model structure","A∞-categories: strict morphisms yield model structure","Cofibrant A∞-categories finally precise","Every A∞-category has a cofibrant replacement"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001157,"raw_usage":{"total_tokens":4747,"prompt_tokens":850,"completion_tokens":3897,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":466,"completion_tokens_details":{"reasoning_tokens":3832}},"tokens_in":466,"tokens_out":3897,"duration_ms":32514,"temperature":1.0,"reasoning_tokens":3832,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T21:57:25.097293+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the pushout computation of (13) for a minimal test case—say $\\mathcal{M}$ the one-object A∞-category with hom-complex $(R,0)$—and verify the universal property of the pushout term by term; at the same time compute $H^n(K_{A_\\infty}(1,1)/R\\cdot 1_1)$ directly from the generators and the differential $m^1$. A nonzero cohomology group, or a mismatch with formula (14), would show that condition 4 fails and Theorem A is false.","supporting_citations":[{"cited_title":"Model categories","cited_arxiv_id":null,"evidence_quote":"Supplies the Recognition Theorem, the criterion by which the model structure is verified."},{"cited_title":"Ornaghi, The Homotopy Theory of A∞categories, Int","cited_arxiv_id":null,"evidence_quote":"Provides the semi-free A∞-categories and resolutions, the completeness and cocompleteness of the strict-morphism category, and the contractibility result for $K_{A_\\infty}(1,1)/R\\cdot 1_1$."},{"cited_title":"Localizations of the categories of A∞-categories and Internal Homs","cited_arxiv_id":null,"evidence_quote":"Establishes that the full category of A∞-categories lacks equalizers, the obstruction that motivates working with strict morphisms."}],"review_version":1}