{"id":"dbcaa7ea-e957-422e-9549-94e48713b362","arxiv_id":"1908.03021","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Dg manifolds form a homotopy site whose infinity category of stacks is equivalent to the Toen-Vezzosi category of stacks on dg algebras with finitely many generators in each degree.","lead":"This paper connects two different formalisms for derived algebraic geometry: explicit differential graded manifolds and Toen-Vezzosi derived stacks. It proves the two formalisms produce equivalent categories of stacks, so an explicit dg construction can be viewed as a derived stack.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3's functorial resolution rests on asserted subset-indexed decompositions (10)-(12) and a multiplicative compatibility claim in the surjectivity argument; these unverified steps carry the stack equivalence.","rationale":"Good-faith summary: the paper builds a category M of dg manifolds with a category-of-fibrant-objects structure (Thm 1), proves A ↪ M homotopically full and faithful (Thm 2), constructs a functorial cosimplicial resolution R: A → A^Δ inside A (Thm 3), proves A ↪ (A)^op homotopically full and faithful (Thm 4), constructs hypercovers from affine atlases (Thm 5), and derives St(A) ≃ St(M) (Thm 6), the Quillen adjunction with St((A)^op) and the equivalence St(A) ≃ St(rA^op) (Thm 7), and the derived-scheme statement (Thm 8). The abstract's central claim is this chain.\n\nThe least secure link is Theorem 3. I agree with the reader's locus but sharpen it: the risk is not 'length' but three concrete assertions inside the proof — preservation of decompositions (10)–(11), formula (12), and the product identity rα·rβ = α·β — none of which is actually derived, and which carry the two properties (projectivity of D^k_{n+1}; surjectivity of A_{n+1} → M_{n+1}) that keep the resolution inside A. Visible indexing inconsistencies (the union condition over subsets of {0,...,n} cannot produce {0,...,n+1}; the face count is off by one) make the step impossible to check as written. I found no counterexample: an attempt to violate the closure claims by adjoining a nilpotent degree-0 generator yields a map that fails the étale cohomology condition in graded-commutative algebra, so the surrounding claims are plausible. Peripheral structure (Reedy directions, function-complex invariance, cofinality arguments) checked out consistently. Hence no REJECT; CONDITIONAL stands.\n\nAdditional self-flagged limits weighed: the title promises shifted symplectic structures on derived Quot-stacks, which the paper does not deliver (and the announced basis [6] is said to be wrong); Theorem 8's proof skips the transfer of Theorem 5's hypercovers from M to (A)^op; the 'pseudo-model category' observation on rA is asserted without proof. These reduce the paper's scope relative to its title but do not by themselves falsify the proved chain.\n\nVerdict: UNCHANGED (CONDITIONAL). A specialist should check Theorem 3's combinatorial core — ideally via the explicit small-case computation proposed — before the stack equivalence is cited.","tokens_in":24302,"tokens_out":62617,"duration_ms":604795,"concrete_test":"Verify the inductive step of Theorem 3 on a minimal example by explicit symbolic computation. Take A = F[x, y] with |x| = 0, |y| = −1, dy = 0 (so A ∈ A, with A0 = F[x] and one projective generator in degree −1). Implement the construction of R(A) for n = 0, 1, 2 (in Sage or Macaulay2): compute the latching and matching objects L_1, M_1, L_2, M_2, the pullback (9) defining D^k_{n+1}, and check (1) the decompositions (10)–(11) exist at each step and are preserved; (2) formula (12) reproduces the actual pullback as a direct sum; (3) each D^k_{n+1} is a finitely generated projective A0-module; (4) A_2 → M_2 is surjective in negative degrees. In parallel, re-derive (12) from (9) analytically, tracking the face and degeneracy maps for Δ^1 and Δ^2 explicitly; the derivation must reproduce the |s| < n+1 restriction without invoking 'it is clear'.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is the equivalence of ∞-categories of stacks St(M) ≃ St(rA^op) (dg manifolds vs. the Toen–Vezzosi site of dg algebras with finitely generated cohomology), plus the statement that dg manifolds represent derived schemes (Theorems 6, 7, 8). The bridge is Theorem 3: a functor R: A → A^Δ producing special cosimplicial resolutions entirely within A; Theorem 4 (homotopical full faithfulness of A ↪ (A)^op) and hence the stack comparison rest on it.\n\nInside the proof of Theorem 3, the decisive claims are: (i) the subset-indexed direct-sum decompositions (10)–(11) of the modules G^k_m, introduced as an 'additional inductive assumption' but never proven to be preserved by the induction step; (ii) the isomorphism (12), expressing the pullback D^k_{n+1} as a direct sum of these pieces, stated without derivation from (9); (iii) the surjectivity of A_{n+1} → M_{n+1} in negative degrees, which needs a multiplicative compatibility of the decomposition: 'if s ∪ s' ≠ {0,...,n+1}, then at least one of α_s, β_{s'} is 0' and 'one immediately obtains rα·rβ = α·β' (p. 23). The displayed union condition is ambiguous: s, s' range over subsets of {0,...,n}, so their union cannot equal {0,...,n+1}; the definition of rα jumps from P(0,...,n) to P(0,...,n+1); and the count of n-faces of Δ^{n+1} containing s is n+2−|s|, not n+1−|s|. These slips sit at the exact point where the construction must stay inside A. If (12) fails, D^k_{n+1} need not be projective over A0; if the product claim fails, A_{n+1} → M_{n+1} need not be surjective. Either way the resolution leaves A and the stack equivalence is unsupported.\n\nThe paper's own caveats corroborate caution: footnote 2 only 'believes' the theorems survive with ordinary quasi-isomorphisms, and the Introduction announces that the main theorem of [6], the basis of the title's Quot-stack program, is wrong (private communication). These do not refute Theorem 3, but they raise the bar for an intricate, unformalized construction.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a theory of dg manifolds, defined as dg schemes whose degree-zero component is a smooth quasi-projective scheme and whose negative-degree components are locally almost free over it. The authors equip the category M of dg manifolds with fibrations and weak equivalences, prove that M is a category of fibrant objects (Theorem 1), and compare the simplicial localization of the affine subcategory A with that of M (Theorem 2) and with the Toen-Vezzosi model category of all non-positively graded dg algebras (Theorem 4). The central technical result is Theorem 3, which constructs a functorial special cosimplicial resolution R: A -> A^Delta staying inside the subcategory A. Using this resolution and an etale topology transferred from the affine dg algebra site, the paper proves a Quillen equivalence between stacks on A and on M (Theorem 6) and shows that stacks represented by dg manifolds are derived schemes (Theorem 8). The paper is the first part of a program toward explicit derived Quot-stacks and shifted symplectic structures; it also announces, based on a private communication, that the main theorem of the earlier derived Quot-scheme paper [6] is incorrect, which motivates deferring part (2) of the program.","tokens_in":24745,"tokens_out":3931,"duration_ms":45154,"significance":"If the results are correct, the paper provides a valuable bridge between the explicit, scheme-theoretic dg manifold language of Ciocan-Fontanine-Kapranov and the Toen-Vezzosi derived algebraic geometry framework. Theorems 1, 2, 5, 6, and 8 form an elegant conceptual structure: the use of categories of fibrant objects, cocycle categories, and affine atlases giving hypercovers is a sound and useful approach. The construction of Theorem 3, however, is load-bearing for the entire comparison, since Theorems 4, 6, and 8 rely on it. The proof of Theorem 3 is presented as a long combinatorial induction with several steps asserted as \"obvious\" or \"immediate\", and the text contains concrete set-theoretic and combinatorial errors at the critical point where the construction must remain in A. Until that proof is repaired or replaced, the central equivalences are conditional. The paper also contains no machine-checked verification of the intricate simplicial construction, which would have been especially helpful here.","major_comments":[{"comment":"The additional inductive assumption (10)-(11), giving decompositions G^k_m = \\oplus_{s in P(0,...,m)} G^k_{m,s}, is introduced but the proof never shows that the induction step preserves this decomposition. Equation (12), asserting an isomorphism D^k_{n+1} \\simeq \\oplus_{s} G^k_{|s|-1,{0,...,|s|-1}}, is stated without derivation from the pullback diagram (9). This is not a cosmetic gap: (12) is exactly what is needed to conclude that each D^k_{n+1} is a finitely generated projective A0-module, and hence that A_{n+1} remains in the subcategory A. A complete proof of (10)-(12) and their preservation under induction is required.","section":"Section 2, proof of Theorem 3, equations (10)-(12)"},{"comment":"The multiplicative compatibility condition is stated as: \"if s \\cup s' \\neq {0,...,n+1}, then at least one of \\alpha_s, \\beta_{s'} is 0\", but s and s' are subsets of {0,...,n}, so their union can never equal {0,...,n+1}. The subsequent definition of r\\alpha and r\\beta says the internal sums run over all n-faces of \\Delta^{n+1} containing s, and the count is given as n+1-|s|; the correct count is n+2-|s|. These errors occur precisely in the step proving r\\alpha \\cdot r\\beta = \\alpha \\cdot \\beta, which is needed to show that A_{n+1} \\to M_{n+1} is surjective in negative degrees. Without a corrected and complete argument, the theorem does not establish that the resolution stays inside A.","section":"Section 2, p. 23, surjectivity argument after (13)"},{"comment":"The proof asserts \"Therefore all maps in this diagram are weak equivalences\", based on the preceding compositions and the 2-out-of-6 property. However, the diagrams as drawn do not themselves imply that each individual arrow is a weak equivalence; in particular, the map Hom_{(A)^{op}}(R(X), X1) \\to N(D/X1) requires a separate justification beyond the composability of weak equivalences. The proof also depends essentially on the functoriality of R from Theorem 3 and on the injectivity claim of Lemma 2. Since Theorem 4 is the bridge to the Toen-Vezzosi stacks, this step needs to be written out explicitly and rigorously.","section":"Section 2, proof of Theorem 4"}],"minor_comments":[{"comment":"The notation \"A–pAqop\" is confusing; the standard A^op would be clearer, especially since the paper also uses the symbol A for the category of all dg algebras.","section":"Notation and conventions"},{"comment":"The symbol \"xE0_m\" is likely a typo for \\bar{E}^0_m or a similar decoration; the reader cannot tell what is intended.","section":"Section 2, p. 22"},{"comment":"The phrase \"over all the n-faces of \\Delta^{n+1} that contains (or s1 respectively)\" is incomplete and should be rewritten.","section":"Section 2, p. 23"},{"comment":"The proofs of parts (1) and (2) contain nearly identical repeated passages; a single lemma isolating the amphicity argument would improve readability.","section":"Section 1, proofs of Proposition 5"},{"comment":"The paper should state explicitly that the etale topology on M is transferred from the affine site by definition, so the independent content of Theorem 6 lies entirely in Theorem 5 (affine atlases give hypercovers). This would prevent a possible misreading that the topology itself is the main new contribution.","section":"Introduction and Remark 9"},{"comment":"The announcement that the main theorem of [6] is wrong is based on an unpublished communication; if this fact is needed for the motivation, it should at least be flagged as a personal communication and ideally verified in writing with the authors of [6] before publication.","section":"Introduction, paragraphs after Theorem 8"}],"recommendation":"major_revision","confidential_remarks":"The central issue is Theorem 3. The proof as written contains set-theoretically impossible conditions and an incorrect face count, and the preservation of the inductive decomposition is not demonstrated. These are load-bearing for the stack equivalence. I would not accept the paper before a complete, corrected proof of Theorem 3 is supplied, ideally with the simplicial combinatorics written out in full. The claim that the main theorem of [6] is wrong, based on a private communication, is also unusual; the editors may want to seek confirmation from the authors of [6] before this statement appears in print. The overall structure of the paper is promising, and the categorical framework (Theorems 1, 2, 5, 6, 8) is well designed, so a major revision with a repaired Theorem 3 is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"If Theorem 3 holds, this is an important bridge between explicit dg-manifold models and Toën–Vezzosi derived stacks. But the proof of Theorem 3 as written has combinatorial indexing problems that block verification, so the right call is a specialist referee, not acceptance on faith.\n\nWhat is new and worthwhile: the category M of dg manifolds with a category-of-fibrant-objects structure, the almost-affineness notion of weak equivalence, and the claimed equivalences St(M) ≃ St(A) ≃ St((rA)^op). Theorems 1, 2, 6, and 8 are sketched plausibly, and the atlas-to-hypercover argument (Lemma 3, Prop. 6, Thm. 5) is a genuinely nice way to get derived schemes out of dg manifolds. If the comparison works, it gives people who want explicit dg models a way into Toën–Vezzosi without redoing the foundations.\n\nThe soft spots sit in Theorem 3. The subset-indexed decompositions (10)–(11) are introduced as an 'additional inductive assumption' but the proof never shows the induction preserves them. The isomorphism (12) is asserted without derivation. In the surjectivity argument, α_s and β_{s'} are indexed by subsets of {0,...,n}, while rα and rβ are defined with s ∈ P(0,...,n+1); the union condition s ∪ s' ≠ {0,...,n+1} is then meaningless for subsets of {0,...,n}. The count of n-faces containing a subset of size |s| should be n+2−|s|, not n+1−|s|. These are not cosmetic typos; they occur exactly where the construction must stay inside A. They may be repairable, but as written the proof does not close.\n\nTwo smaller caveats. The introduction reports a private communication that the main theorem of [6] is wrong; that is honest, but it also means the title's promise of derived Quot-stacks is not delivered here and the motivating construction rests on unpublished follow-up work. Footnote 2 says the authors 'believe' Theorem 6 survives with ordinary quasi-isomorphisms; that is a work-around, not a proof.\n\nBottom line: this paper deserves a serious referee. The architecture is coherent and the goal is valuable. The referee should be asked to verify Theorem 3 before the stack equivalence is taken as established. I would not cite the equivalence in my own work yet. For a reading group, it is a maybe—good for a group that tolerates thirty pages of simplicial combinatorics.","headline":"Valuable bridge between dg manifolds and Toën–Vezzosi stacks, but Theorem 3's proof has unverified combinatorial steps that a specialist referee must check.","tokens_in":25405,"tokens_out":6113,"would_cite":false,"duration_ms":59183,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14A20","14J35","14J40","14F05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Dg manifolds can be organized into a homotopy site whose infinity category of stacks is equivalent to the standard affine derived geometry of finitely generated dg algebras.","keywords":["Derived Quot scheme","Derived Quot Stack","Simplicial localization","dg manifolds","homotopy site","category of fibrant objects","derived algebraic geometry","dg schemes"],"falsifier":"Concretely, compute the homotopy type of Map_{L(A)}(X,Y) from the explicit cocycle category and compare it with the homotopy type obtained from the ambient model category of all dg algebras, for a pair of affine dg manifolds X,Y with at least one non-zero negative-degree element; any mismatch would refute Theorem 4 and thus the stack equivalence.","tokens_in":24081,"feed_emoji":"🧮","tokens_out":8288,"duration_ms":82423,"temperature":0.7,"pith_summary":"This paper's first step is to make differential graded manifolds—schemes whose ambient underlying scheme is smooth and quasi-projective and whose negative-degree structure sheaf is locally almost free—into a homotopy site, and to prove that the resulting infinity category of stacks is equivalent to the infinity category of stacks on the site of differential non-positively graded algebras whose cohomologies have finitely many generators in each degree. Under this equivalence every dg manifold represents a derived scheme. The point is to supply a concrete, quasi-projective model for derived objects such as Quot-stacks of coherent sheaves, where explicit representatives of shifted symplectic forms can later be constructed. The paper also notes that a previously published dg Quot-scheme construction has a flawed main theorem, so the Quot-scheme application is deferred to a sequel.","feed_headline":"Dg manifolds give the same stacks as derived affine schemes","feed_subtitle":"A category of dg manifolds is proved a homotopy site matching stacks on finitely generated dg algebras.","key_machinery":"The central machinery is the subcategory A of affine dg manifolds, opposite to smooth almost-free dg algebras, together with the functor R: A -> A^$\\Delta$ that produces special cosimplicial resolutions entirely inside A (Theorem 3). The subtle point is functoriality: standard free-algebra resolutions leave A, so the proof is a long simplicial construction that keeps every resolution inside A. R lets the mapping spaces in the simplicial localization of A be computed from cosimplicial resolutions as in the ambient model category, proving Theorem 4. On the geometric side, the category-of-fibrant-objects structure on M supplies path objects via 'killing cocycles' and gives cocycle descriptions of mapping spaces; Proposition 6 and the almost-affineness condition then ensure that open affine atlases of a dg manifold form hypercovers, which is what makes the stack comparison work.","core_discovery":"The paper's central claim is that the infinity category of stacks on the site of dg manifolds is equivalent to the infinity category of stacks on the site of differential non-positively graded algebras whose cohomologies have finitely many generators in each degree, and that under this correspondence every dg manifold represents a derived scheme. To reach this conclusion, the paper introduces the category M of dg manifolds—dg schemes whose ambient classical scheme is smooth and quasi-projective and whose negative-degree pieces are locally almost free—and equips it with fibrations and weak equivalences so that both M and its affine subcategory A are categories of fibrant objects. It then proves that the inclusions A into M and A into the ambient category of all dg algebras induce homotopically full and faithful functors between simplicial localizations, and that these comparisons respect the etale topologies, giving equivalent categories of stacks. This establishes the first half of a two-step program: the second half, constructing explicit dg Quot-schemes with shifted symplectic forms, is postponed to a later paper.","pith_inferences":["Because the equivalence identifies dg manifolds with derived schemes, the promised dg Quot-schemes should carry the same shifted symplectic forms as their derived-stack counterparts; a natural next test is to compute the form on the dg manifold model directly.","The paper's stated failure of the earlier dg Quot construction means the equivalence alone does not produce derived Quot-stacks; the follow-up must show that the projective system of dg manifolds converges in the stack category to a derived scheme with the correct tangent complex.","The comparison suggests that quasi-projectivity can be used as a technical crutch inside derived geometry: any derived scheme that is quasi-projective over the base should admit a representing dg manifold, making concrete computations possible."],"forward_implications":["A dg manifold, under the established equivalence, represents a derived scheme in the ambient derived algebraic geometry (Theorem 8).","Stacks on the full category of dg manifolds are Quillen equivalent to stacks on affine dg manifolds (Theorem 6).","Mapping spaces between affine dg manifolds can be computed through the explicit functorial cosimplicial resolutions of Theorem 3, with no loss of homotopical information (Theorem 4).","The essential image of the affine dg manifolds inside all dg algebras consists exactly of dg algebras whose cohomology is finitely generated in each degree, so the equivalence is with that natural site.","The category of dg manifolds supports two compatible categories-of-fibrant-objects structures, one with weak equivalences and one with quasi-isomorphisms, giving flexibility in computing mapping spaces."],"supporting_citations":[{"why":"Supplies the S-topology, the simplicial model structure on prestacks, and the infinity category of stacks that the paper compares against.","marker":"[24]"},{"why":"Provides the geometric-stack and derived-scheme framework used in Theorems 6 and 8.","marker":"[25]"},{"why":"Introduces the notion of dg schemes and the dg Quot-scheme program; the paper builds its category of dg manifolds from this notion.","marker":"[6]"},{"why":"Gives the definition of derived scheme used in Theorem 8.","marker":"[22]"},{"why":"Supplies the cocycle description of mapping spaces in categories of fibrant objects, used in Theorems 2, 4, and 6.","marker":"[17]"},{"why":"Provides the theory of function complexes and special simplicial resolutions used in the construction of R in Theorem 3.","marker":"[10]"},{"why":"Gives the axioms of categories of fibrant objects and the factorization lemma used to build path objects in M.","marker":"[5]"}],"fun_headline_variants":["Dg manifolds match finite dg algebras in stack category","Two sites, equivalent stacks: dg manifolds and finite dg algebras","Dg manifolds represent derived schemes via stack equivalence","Finite dg algebras and dg manifolds: one stack category"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire comparison rests on Theorem 3's claim that a functorial cosimplicial resolution can be built entirely inside the subcategory of smooth almost-free dg algebras, and if that functoriality fails, the mapping-space equivalences and stack equivalence collapse.","fun_headline_variants_meta":{"raw":{"variants":["Dg manifolds match finite dg algebras in stack category","Two sites, equivalent stacks: dg manifolds and finite dg algebras","Dg manifolds represent derived schemes via stack equivalence","Finite dg algebras and dg manifolds: one stack category"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000691,"raw_usage":{"total_tokens":3067,"prompt_tokens":824,"completion_tokens":2243,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":440,"completion_tokens_details":{"reasoning_tokens":2169}},"tokens_in":440,"tokens_out":2243,"duration_ms":15706,"temperature":1.0,"reasoning_tokens":2169,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:26:48.414535+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Concretely, compute the homotopy type of Map_{L(A)}(X,Y) from the explicit cocycle category and compare it with the homotopy type obtained from the ambient model category of all dg algebras, for a pair of affine dg manifolds X,Y with at least one non-zero negative-degree element; any mismatch would refute Theorem 4 and thus the stack equivalence.","supporting_citations":[{"cited_title":"Advances in Mathematics 193, pp","cited_arxiv_id":null,"evidence_quote":"Supplies the S-topology, the simplicial model structure on prestacks, and the infinity category of stacks that the paper compares against."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the geometric-stack and derived-scheme framework used in Theorems 6 and 8."},{"cited_title":"Derived Quot schemes","cited_arxiv_id":null,"evidence_quote":"Introduces the notion of dg schemes and the dg Quot-scheme program; the paper builds its category of dg manifolds from this notion."},{"cited_title":"EMS Surv","cited_arxiv_id":null,"evidence_quote":"Gives the definition of derived scheme used in Theorem 8."},{"cited_title":"Cocycles in categories of fibrant objects","cited_arxiv_id":"1502.03925","evidence_quote":"Supplies the cocycle description of mapping spaces in categories of fibrant objects, used in Theorems 2, 4, and 6."},{"cited_title":"Function complexes in homotopical algebra","cited_arxiv_id":null,"evidence_quote":"Provides the theory of function complexes and special simplicial resolutions used in the construction of R in Theorem 3."},{"cited_title":"Abstract homotopy theory and generalized sheaf cohomol- ogy","cited_arxiv_id":null,"evidence_quote":"Gives the axioms of categories of fibrant objects and the factorization lemma used to build path objects in M."}],"review_version":1}