{"id":"bd8840cb-9b2f-4090-ac94-e43818400220","arxiv_id":"2505.24345","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The non-acyclicity class of a constructible etale sheaf is additive across distinguished triangles, under a cohomological smoothness assumption.","lead":"This paper proves that the non-acyclicity class, an invariant measuring how far a sheaf is from being locally acyclic, is additive over distinguished triangles of sheaves. The proof uses a new categorical framework of bivariant cohomological correspondences and a trace formula inspired by categorical traces.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.12 omits Assumption 2.22: additivity is proved only when g:Y→S is cohomologically smooth, so the theorem as stated is overbroad.","rationale":"The paper proposes a categorical trace-like formula for Yang–Zhao non-acyclicity classes and derives additivity from it. The nine-diagram machinery is presented with explicit proofs and appears internally coherent. The main substantive weakness is exactly the one identified by the reader: the proof requires Assumption 2.22, which is stated only in Section 2 and not included in the abstract or in Theorem 1.12. This is not a cosmetic omission: the trace formula in Proposition 2.28 and the localization step in Definition 2.30 use the isomorphisms (2.17.1) and (2.17.2), which follow from Assumption 2.22 but are not established for arbitrary g. In particular, Section 4.1 begins with 'Assume g satisfies assumption 2.22,' so the main theorem as stated in the introduction is broader than what is proved. There is no evidence of internal inconsistency in the categorical constructions when the assumption is present, and the reliance on Lu–Zheng and Yang–Zhao is explicit. Therefore the appropriate verdict remains conditional: the result is plausible and likely correct under cohomological smoothness of g, but the stated theorem and abstract need to be amended to include that hypothesis. No revised numerical or formal-verification evidence is available to strengthen the verdict beyond conditional.","tokens_in":14996,"tokens_out":15790,"duration_ms":216951,"concrete_test":"Let S=Spec k and Y=Spec k[ε]/(ε²) with g the structure morphism. Compute K_{Y/S}=g^!Λ and the bi-evaluation map (2.19.1) at F=Λ; if K_{Y/S} is not ⊗_Y-invertible or (2.19.1) is not an isomorphism, Assumption 2.22 is not automatic for arbitrary separated finite type g. Then test Theorem 1.12 for the trivial triangle Λ→Λ→0 on X=Y: either C_{X/Y/S}(Λ) is not defined without Assumption 2.22, or the claimed equality fails, showing the missing hypothesis is essential.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central theorem is stated without any condition on g:Y→S, but the categorical trace-like formula for the non-acyclicity class is built under Assumption 2.22: D(Y/S) is ⊗_Y-invertible and the bi-evaluation map (2.19.1) is an isomorphism, which holds when g is cohomologically smooth. Section 4 begins by imposing Assumption 2.22, and Propositions 4.3 and 4.4 rely on (2.17.1) and (2.17.2) being isomorphisms; these are only derived from Assumption 2.22 via Proposition 2.21. If g is not cohomologically smooth—for example a non-smooth finite morphism—there is no justification for identifying the categorical trace with the non-acyclicity class or for running the exact nine-diagram additivity argument. Thus the abstract's unqualified claim that additivity is proved is not supported by the body of the paper as written. The theorem should either explicitly carry Assumption 2.22 as a hypothesis or prove that the non-acyclicity class and its additivity survive without it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a bivariant ∞-categorical framework of cohomological correspondences to give a trace-like formula for the non-acyclicity classes introduced by Yang and Zhao, and then derives an additivity theorem for these classes under distinguished triangles. Section 2 constructs the bivariant category BivCohCorr_S and defines the non-localized and localized NA classes via categorical traces; Section 3 proves a nine-diagram lemma; Section 4 applies the lemma to prove additivity. The abstract and Theorem 1.12 state the additivity theorem without any hypothesis on the morphism g: Y → S, but the proof in Section 4 is carried out under Assumption 2.22, which is satisfied, for example, when g is cohomologically smooth.","tokens_in":15177,"tokens_out":5095,"duration_ms":58444,"significance":"The categorical trace formula is a genuinely useful reformulation: it packages the NA class as an instance of the same construction that gives relative characteristic classes, and the additivity theorem is the expected 'characteristic class' property. The paper contains a self-contained proof of Proposition 3.6, and the exact nine-diagram technique is a workable substitute for earlier homotopy-theoretic trace-additivity results. If the hypothesis issue is resolved, this will be a solid contribution to geometric ramification theory. The main value is conditional: the current statement overclaims what is proved, because the proof as written requires Assumption 2.22 on g.","major_comments":[{"comment":"The additivity theorem is proved only under Assumption 2.22, which requires D(Y/S) to be ⊗_Y-invertible and the bi-evaluation map (2.19.1) to be an isomorphism, but the abstract and Theorem 1.12 state the result without this hypothesis. Section 4 opens by imposing Assumption 2.22, and Propositions 4.3 and 4.4 depend on (2.17.1) and (2.17.2) being isomorphisms, which are derived from Assumption 2.22 via Proposition 2.21. As written, the central claim is overbroad; the theorem should either carry Assumption 2.22 explicitly as a hypothesis, or the author should prove that the trace formula and the additivity argument survive without it.","section":"§4.1, Assumption 2.22; Theorem 1.12; Abstract"},{"comment":"The equality '∆_{Y/S}(E_F ⊗_S Hom_S(E_F,S)) = i_* i^! ∆_{Y/S}(E_F ⊗_S Hom_S(E_F,S))' is not justified and is generally false, since the object on the left need not be supported on Z. The proof of Corollary 4.4 needs to explain explicitly how the class in H^0_Z is obtained: the vanishing of the U-restriction (Definition 2.30) should force the relevant trace to land in the Z-supported part. As written, this is a gap in the proof of part (2) of Theorem 1.12.","section":"§4.4, diagram (4.4.2)"}],"minor_comments":[{"comment":"The abbreviation 'c.f.' should be 'cf.' throughout, and there is a typo 'equiality' in Proposition 1.17.","section":"Throughout"},{"comment":"The assertion that ∆_{Y/S}(F|_U ⊗ Hom_S(F|_U,S)) = 0 uses the dualizability assumption on U and should be stated explicitly as a consequence of (2.23.1).","section":"Definition 2.30"},{"comment":"The paper relies on an unpublished note for Proposition 3.6; although a proof is included, adding the note's URL to the bibliography rather than only in the acknowledgments would improve traceability.","section":"Acknowledgments; Proposition 3.6"},{"comment":"The proof of Proposition 1.17 is only sketched; since the same nine-diagram technique is used later, a few more details there would help the reader.","section":"§1.20"}],"recommendation":"major_revision","confidential_remarks":"The mismatch between the abstract's unqualified additivity claim and the proof under Assumption 2.22 is the main obstacle to acceptance. The author should also be asked to clarify the localization step in Corollary 4.4. Both issues appear fixable within the manuscript's scope, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper is worth taking seriously, but it needs a revision to align the statements with the assumptions. The main result is genuinely new: additivity for Yang–Zhao non-acyclicity classes has not been proved elsewhere, and the bivariant category of cohomological correspondences is a natural tool for the task. The trace-like formula (Proposition 2.28) and the exact nine-diagram proof (Propositions 4.3–4.4) are substantive. The paper also gives a self-contained proof of the nine-diagram lemma it needs, even though the author learned the trick from a note by Yang; that is fine in my book.\n\nThe soft spot, which the stress-test note correctly identifies, is that Theorem 1.12 appears to state additivity without any condition on the map g:Y→S. But Section 4 begins by imposing Assumption 2.22, which requires D(Y/S) to be ⊗-invertible and the bi-evaluation map to be an isomorphism—conditions that hold when g is cohomologically smooth. The proof of the categorical description of the non-acyclicity class and the additivity argument both rely on this assumption. So the theorem as stated, and the abstract's unqualified claim, are overbroad. The fix is straightforward: either carry Assumption 2.22 explicitly into the theorem statements, or prove that the NA class and its additivity make sense without it. This matters, because non-smooth finite morphisms are exactly the kind of case the geometric ramification theory wants to handle.\n\nMinor points: the paper leans on known categorical trace additivity (May, Groth–Ponto–Schulman, Ramzi), so the novelty is in the application and formulation, not the underlying mechanism. That is fine for a paper like this. I did not see circularity: the result is genuinely derived from trace additivity, not assumed. No code or external verification, but that is not expected here.\n\nOverall: the paper deserves peer review, not desk rejection, but only after the authors fix the mismatch between the main theorem and the stated assumptions. I would tell the editor to send it out and ask the referee to check that the trace-like formula really recovers the NA class under the stated hypotheses, and whether additivity can be pushed beyond cohomologically smooth g. A serious referee can handle that.","headline":"Genuinely new additivity theorem for NA classes, but Theorem 1.12 omits the cohomological smoothness hypothesis that the proof actually uses.","tokens_in":15737,"tokens_out":1460,"would_cite":false,"duration_ms":20499,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14F20","14F05","18N70"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that non-acyclicity classes add over exact triangles of constructible sheaves.","keywords":["non-acyclicity class","categorical trace","cohomological correspondences","étale sheaves","additivity","constructible sheaves","Verdier duality","exact nine-diagrams"],"falsifier":"Take \\(S = \\mathrm{Spec}\\,k\\), \\(Y\\) a smooth curve over \\(S\\), \\(X\\) a relative curve over \\(Y\\), and an exact triangle \\(F'\\to F\\to F''\\) of locally constant étale sheaves on \\(X\\) that are ULA over both \\(Y\\) and \\(S\\). Compute the three localized classes \\(C^Z_{X/Y/S}\\) by the defining excision formula: the theorem predicts the middle class is the sum of the two outer classes. A single such computation where the equality fails would falsify Theorem 4.4.","tokens_in":14754,"feed_emoji":"➕","tokens_out":14310,"duration_ms":153497,"temperature":0.7,"pith_summary":"The paper establishes that the non-acyclicity class, a cohomological invariant measuring how a constructible sheaf fails to be acyclic along a closed subset relative to a fibration, is additive: for a distinguished triangle \\(F' \\to F \\to F''\\) of sheaves satisfying the relevant local acyclicity conditions, the class of \\(F\\) equals the class of \\(F'\\) plus the class of \\(F''\\). To prove this, the author rewrites the non-acyclicity class as a categorical trace-like composition using a bivariant version of cohomological correspondences, then shows that any such trace-like pairing is additive on exact triangles by passing through exact nine-diagrams. If correct, the result turns the non-acyclicity class into a genuine characteristic-class-valued invariant, parallel to the known additivity of cohomological characteristic classes, and provides a categorical explanation for why the class behaves like a ramification-theoretic conductor. The main theorem is stated under a cohomological smoothness assumption on the middle fibration.","feed_headline":"Non-acyclicity classes add over exact triangles","feed_subtitle":"A trace-like categorical formula turns this ramification invariant into a sum of two parts.","key_machinery":"The load-bearing objects are the bivariant category of cohomological correspondences—roughly, a category whose objects are pairs of a scheme and a sheaf and whose morphisms are correspondences together with a map of sheaves—and the exact nine-diagram lemma. \\(\\mathrm{BivCohCorr}_S\\) is a symmetric monoidal fibration whose object \\((X/Y;F)\\) packages a scheme \\(X\\) over \\(Y\\) over \\(S\\) together with a sheaf \\(F\\); it organizes the categories \\(\\mathrm{CohCorr}_Y\\) for all intermediate bases \\(Y\\) into one structure, so that the comparison maps between the dual over \\(Y\\) and the dual over \\(S\\) become natural isomorphisms under Assumption 2.22. The nine-diagram lemma (Proposition 3.6) takes an exact \\(3\\times 3\\) diagram whose rows and columns are exact triangles and produces an exact triangle on alternating sums of the entries; applied to the maps between coevaluation and evaluation associated to an endomorphism of an exact triangle, it yields the trace identity \\(\\mathrm{Tr}(g) = \\mathrm{Tr}(f) + \\mathrm{Tr}(h)\\) that is the heart of additivity.","core_discovery":"The paper's central claim is Theorem 4.4: under Assumption 2.22, for an exact triangle \\(F' \\to F \\to F''\\) of constructible complexes of finite Tor-amplitude on \\(X\\), if \\(F', F, F''\\) are dualizable (the categorical form of universal local acyclicity) in \\(\\mathrm{CohCorr}_S\\) and their restrictions to an open immersion \\(U \\subset X\\) are dualizable in \\(\\mathrm{CohCorr}_Y\\), then the localized non-acyclicity classes satisfy \\(C^Z_{X/Y/S}(F) = C^Z_{X/Y/S}(F') + C^Z_{X/Y/S}(F'')\\) in \\($H^{0}$_Z(X, K_{X/Y/S})\\). Equivalently, the non-localized class \\(C_{X/Y/S}\\) is additive on exact triangles of dualizable objects. The route is a categorical trace-like formula: the class is written as the composition \\((X;\\Lambda) \\to \\mathrm{Hom}_Y((X;F),(X;F)) \\to \\Delta_{Y/S}((X;F) \\otimes_S D(F/S)) \\to D(X/Y/S)\\) in the category of cohomological correspondences over \\(S\\), and the localized version is obtained by excision on the closed complement of \\(U\\). Additivity then follows from a general proposition about trace-like pairings built from such compositions.","pith_inferences":["A natural next step, not pursued in the paper, is to use the same trace-like formula to prove additivity for the geometric characteristic cycle class, since both classes are expected to coincide under the cycle-class map.","The bivariant category could serve as the natural home for a relative trace formula; the trace-like composition here is already close to a relative trace, so one might deduce product formulas and decomposition statements by the same adjunction arguments.","The proof only needs the two comparison maps (2.17.1) and (2.17.2) to be isomorphisms; testing whether a weaker assumption than cohomological smoothness of \\(g:Y\\to S\\) suffices would enlarge the class of fibrations to which the additivity theorem applies.","In arithmetic settings where the non-acyclicity class recovers classical ramification conductors, additivity would give a unified proof of the additivity of conductors under extensions of sheaves."],"forward_implications":["The additivity of localized non-acyclicity classes holds for any exact triangle of constructible sheaves satisfying the ULA hypotheses, so the invariant behaves like a cohomological characteristic class in the relative setting.","The non-localized class \\(C_{X/Y/S}\\) is additive as well, giving a relative analogue of the known additivity of categorical traces for cohomological characteristic classes.","Because the proof is categorical, the same exact-nine-diagram mechanism applies to any trace-like natural transformation in a stable symmetric monoidal \\(\\infty\\)-category, not only to the étale-sheaf construction.","Under Assumption 2.22, which is satisfied when the middle morphism \\(g:Y\\to S\\) is cohomologically smooth, the theorem gives a uniform additivity statement that does not require case-by-case geometric arguments."],"supporting_citations":[{"why":"introduces the non-acyclicity class whose additivity is the paper's theorem, and supplies the fibration construction that the trace formula rewrites.","marker":"[17]"},{"why":"provides the monoidal category of cohomological correspondences and the categorical-trace description of relative characteristic classes that the bivariant category extends.","marker":"[9]"},{"why":"proves additivity of categorical traces in a closely related setting, the statement the paper adapts to non-acyclicity classes.","marker":"[8]"},{"why":"supplies the earlier triangulated-category trace additivity theorem whose framework the nine-diagram proof refines.","marker":"[13]"}],"fun_headline_variants":["Trace-like formula makes ramification classes additive","Non-acyclicity invariants split along exact triangles","Additivity proven for etale non-acyclicity classes","Categorical trace yields sum rule for non-acyclicity","Exact triangles preserve non-acyclicity additivity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on Assumption 2.22, that the dualizing complex \\(D(Y/S)\\) is tensor-invertible in the category of cohomological correspondences over \\(Y\\) and the bi-evaluation map is an isomorphism, which holds when \\(g:Y\\to S\\) is cohomologically smooth; without this, the trace-like formula for the non-acyclicity class and the comparison maps that make it additive are not available.","fun_headline_variants_meta":{"raw":{"variants":["Trace-like formula makes ramification classes additive","Non-acyclicity invariants split along exact triangles","Additivity proven for etale non-acyclicity classes","Categorical trace yields sum rule for non-acyclicity","Exact triangles preserve non-acyclicity additivity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000781,"raw_usage":{"total_tokens":3414,"prompt_tokens":874,"completion_tokens":2540,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":490,"completion_tokens_details":{"reasoning_tokens":2470}},"tokens_in":490,"tokens_out":2540,"duration_ms":22732,"temperature":1.0,"reasoning_tokens":2470,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T12:25:16.165513+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take \\(S = \\mathrm{Spec}\\,k\\), \\(Y\\) a smooth curve over \\(S\\), \\(X\\) a relative curve over \\(Y\\), and an exact triangle \\(F'\\to F\\to F''\\) of locally constant étale sheaves on \\(X\\) that are ULA over both \\(Y\\) and \\(S\\). Compute the three localized classes \\(C^Z_{X/Y/S}\\) by the defining excision formula: the theorem predicts the middle class is the sum of the two outer classes. A single such computation where the equality fails would falsify Theorem 4.4.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"introduces the non-acyclicity class whose additivity is the paper's theorem, and supplies the fibration construction that the trace formula rewrites."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the monoidal category of cohomological correspondences and the categorical-trace description of relative characteristic classes that the bivariant category extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"proves additivity of categorical traces in a closely related setting, the statement the paper adapts to non-acyclicity classes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the earlier triangulated-category trace additivity theorem whose framework the nine-diagram proof refines."}],"review_version":1}