{"id":"219f08e2-6176-45bb-b595-206c5b19e3b5","arxiv_id":"2601.16437","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Over a field of characteristic p, all Kontsevich-Soibelman operations on periodic cyclic homology are generated by the p-fold equivariant cap product and commute with the Getzler-Gauss-Manin connection.","lead":"This paper classifies the Kontsevich-Soibelman operations on the periodic cyclic homology of a dg algebra: over a field of characteristic p they are generated by the p-fold equivariant cap product, and over the rationals they are trivial. It also shows these operations commute with the Getzler-Gauss-Manin connection, and uses this to give a new obstruction to realizing a middle cohomology class by a Lagrangian submanifold.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Prop. 6.20 only covers the [e0] generator; the [e1] generators in Thm 5.8/Cor 5.9 are never shown to be cap products or to descend to HH_per, so Theorem 1.1's generation statement has a missing step.","rationale":"The reader's weakest assumption concerned the equivariant quasi-equivalence between cacti and configuration spaces. That is a genuine gap, but I find a more immediate and more directly load-bearing gap in the assembly of Theorem 1.1: the generator theorem used to prove the F_p classification includes [e1]-type operations, while Proposition 6.20 identifies only the [e0] operation with the cap product. Unless [e1] operations are shown either to be expressible via cap products or to fail to descend to HH_per, the central 'generated by cap products' statement is not proven. This is a proof gap, not an indication that the theorem is false; it could be closed by a short computation or by a derivation property for B_p. There is also a secondary mismatch: Theorem 5.8 assumes H^*_{Σp}(CC_*(A)^{⊗p}) is free over R, a hypothesis absent from Theorem 1.1; this becomes relevant if R is an F_p-algebra but not a field. Both issues reinforce the conditional verdict, so I do not change the reader's verdict.","tokens_in":72704,"tokens_out":20288,"duration_ms":398144,"concrete_test":"Check whether the missing bridge is true. (a) Compute the image of Res_{C_p⊂S^1} on the class [e1]∈H^1(C_*(Cyl^∘(p,1))^{tCp}_{hΣp}); Lemma 5.10 does this for [e0], and the same computation should be carried out for [e1]. If [e1] is not in the image, its operations do not descend to HH_per, and Theorem 1.1 is consistent. (b) If it is in the image, verify the identity [Ξ_p([e0])([φ],−), B_p] = Ξ_p([e0])(B_p[φ],−) with B_p[φ]∈H^*_{C_p}(CC_*(A)^{⊗p}). A concrete case: take A = F_p with trivial operations and p = 3, compute both sides on HH^{C_3,per}_*(F_p); agreement would show the missing step is recoverable, disagreement would require modifying Theorem 1.1.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.1 (the F_p case) is stated to follow from Theorem 5.8 plus Proposition 6.20. But Theorem 5.8 says all C_p-KS operations are generated by two families: Ξ_p([e0])([φ],−) and Ξ_p([e1])([φ],−); Corollary 5.9 rewrites the second as [Ξ_p([e0])([φ],−), B_p]. Proposition 6.20 identifies only Ξ_p([e0])([φ],−) with the p-fold equivariant cap product. No result in §6.3 identifies the commutator with B_p as a cap product, nor proves that B_p preserves the direct summand HH_per_*(A) inside HH^{Cp,per}_*(A) corresponding to Res_{C_p⊂S^1}. Lemma 5.10 proves [e0] is in the image of restriction from S^1, but says nothing about [e1]. Thus, as written, the primary generator theorem contains a second generator that is not accounted for, so Theorem 1.1's 'generated by cap products' conclusion does not logically follow. This is a proof gap in the central classification, not a claim that the theorem is false.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies equivariant operations on periodic cyclic homology induced by the chain-level action of the two-colored Kontsevich-Soibelman operad. It introduces a new model for this operad based on two-colored cacti with spines, proves an equivariant quasi-equivalence to the configuration-space operad of disks on a disk/cylinder, and uses Cohen's computation of equivariant homology of configuration spaces to propose a classification of Kontsevich-Soibelman operations. The two main algebraic claims are: (1) over Q there are no nontrivial KS operations on HH^{per}_*, while over F_p the algebra of KS operations is generated by p-fold equivariant cap products; (2) every KS operation commutes with the Getzler-Gauss-Manin connection. The paper also gives applications to quantum Steenrod operations, to torsion in the integral cohomology of symplectic manifolds, and to the Lagrangian realization problem.","tokens_in":72967,"tokens_out":5656,"duration_ms":58059,"significance":"If the main results hold, this is a substantial contribution: it provides a unifying operadic framework for natural operations on periodic cyclic homology, establishes a previously unexpected automatic covariant constancy with respect to the Getzler-Gauss-Manin connection, and gives a structural explanation for the appearance of p-fold cap products. The explicit chain-level bookkeeping of C_p-actions and the construction of cyclic cacti with spines are valuable technical innovations, and the connection to quantum Steenrod operations and the Lagrangian realization problem gives the paper broad reach. However, the classification of KS operations is not fully established by the text as written: the central proof has gaps around the second generator [e1], the freeness hypothesis in Theorem 5.8, and the deferred parts of the equivariant quasi-equivalence. These gaps are repairable in principle, but they are load-bearing for Theorem 1.1.","major_comments":[{"comment":"The reduction of Theorem 1.1 to cap products is incomplete. Theorem 5.8 and Corollary 5.9 assert that all C_p-KS operations are generated by Ξ_p([e0])([φ],−) and by [Ξ_p([e0])([φ],−), B_p]. Proposition 6.20 identifies only Ξ_p([e0])([φ],−) with the p-fold equivariant cap product. No result in §6.3 identifies the commutator [Ξ_p([e0])([φ],−), B_p] with a cap product [ψ] ⋒^{C_p} −, nor proves that B_p preserves the summand HH^{per}_*(A) inside HH^{C_p,per}_*(A). Lemma 5.10 and Corollary 5.11 concern [e0] only. Therefore the statement 'generated by endomorphisms of the form [φ] ⋒^{C_p} −' in Theorem 1.1 does not logically follow from the cited results. This is a proof gap in the central classification, not a claim that the theorem is false; it can be repaired either by proving a closure property for the operation [−, B_p] or by directly expressing Ξ_p([e1])([φ],−) as a cap product and provi","section":"Theorem 1.1; §5.1–5.2 and §6.3"},{"comment":"Theorem 5.8 is stated only under the hypothesis that H^*_{Σ_p}(CC_*(A)^{⊗p}) is a free R-module. Theorem 1.1, which is claimed to follow from Theorem 5.8 and Proposition 6.20, contains no such freeness assumption and is stated for every dg algebra over a ring containing F_p. The proof of Theorem 5.8 uses the Künneth decomposition (5.29) and the freeness assumption in an essential way. The manuscript does not explain why the freeness hypothesis is automatic, or how the conclusion of Theorem 1.1 is obtained without it. This is a load-bearing mismatch between the stated theorem and the result actually proved in Section 5.","section":"Theorem 5.8 and Theorem 1.1"},{"comment":"The paper's classification depends on the equivariant quasi-equivalence between the cacti model and the configuration-space model, but that comparison is not fully proved in the text. Theorem 6.18 states an isomorphism of two-colored topological operads, yet the proof defers a 'technical issue involving rotation of basepoints' to Salvatore's paper, and Lemma 6.16 leaves the construction of the inverse homeomorphism to the reader. Since Theorem 5.8 uses the equivariant homology computation of the configuration spaces, the missing equivariant comparison is not a cosmetic matter. Please either supply the missing arguments or state precisely which statements of [Sal1] are being invoked and verify that they carry the required S^1 × S^1 and Σ_n equivariance.","section":"Theorem 4.29; §6.2, Theorem 6.18 and Lemma 6.16"}],"minor_comments":[{"comment":"The ring R((t,θ)) is used before its definition is fixed for p=2; in the introduction θ^2=t for p=2, while in §2.3 the complex is defined with θ^2=0. Please clarify the conventions and the precise form of the completed coefficient ring in Proposition 2.5.","section":"§2.3 and Proposition 2.5"},{"comment":"The cyclic/cocyclic structure maps are only described pictorially. Since the paper explicitly claims to keep track of the finite subgroup C_p ⊂ S^1, a combinatorial specification of the Λ and Λ^op structure maps would make the construction easier to verify.","section":"§4.2, Definition 4.22 and Lemma 4.24"},{"comment":"The statement of Theorem 5.18 lists generators Ξ_p([e0]) and Ξ_p([e1]) with [φ]∈H^*_{Σ_p}(CC_*^{⊗p}), but the twisted operations are defined using H^*_{Σ_p}(CC_*^{⊗p}⊗R(p)). The relation between these two inputs, though illustrated in Example 5.19, would benefit from an explicit statement of the map H^*_{Σ_p}(CC_*^{⊗p}⊗R(p)) → H^*_{Σ_p}(CC_*^{⊗p}).","section":"§5.2, Theorem 5.18"},{"comment":"The proof of Theorem 7.6 uses the unit [e_L] of an object of the Fukaya category. Footnote 5 acknowledges that the Fukaya category does not have strict units and refers to a homological unit. Since the argument requires [e_L] to be a genuine input for the OC^{S^1} map, a short explanation of which construction is used and why it satisfies the needed properties would be helpful.","section":"§7.4, Theorem 7.6 and Footnote 5"},{"comment":"The proof of Theorem 6.9 is summarized by reference to [Sal1]; because this theorem is the basis for the W-construction used in §6.2, a more detailed indication of the proof would improve readability.","section":"§6.1, Theorem 6.9"}],"recommendation":"major_revision","confidential_remarks":"The central algebraic classification (Theorem 1.1) currently has a genuine proof gap: the second generator Ξ_p([e1]) is never identified with a cap product, and the descent to HH_per is only established for [e0]. The freeness hypothesis in Theorem 5.8 also does not match the statement of Theorem 1.1. These issues are repairable within the manuscript's scope, so I do not recommend rejection, but the revision should address them explicitly. The paper is ambitious and contains valuable constructions; if the gaps are filled, it would be a strong contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: this is real work, not a throwaway. The two-colored cacti model for KS and the equivariant quasi-equivalence to configuration spaces are genuine technical novelties. Theorem 1.2—automatic covariant constancy with respect to Getzler-Gauss-Manin via Kaledin's first-order-neighborhood formulation—is clean, and I believe it. The use of Cohen/Rossi to produce explicit generators over F_p is informative, and the symplectic application with the Ottem-Rennemo example is interesting enough to justify the machinery.\n\nBut as written the paper does not prove Theorem 1.1. The stress-test note lands. Theorem 5.8 says every C_p-KS operation is generated by [e0] and [e1]; Corollary 5.9 rewrites [e1] as the commutator with B_p. Then Proposition 6.20 identifies only [e0] with the p-fold equivariant cap product. Nowhere is it shown that the commutator with B_p is a cap product, and nowhere is it shown that B_p preserves the HH_per summand inside HH^{Cp,per}. Lemma 5.10 covers [e0] only. So the claimed generation by cap products is not a logical consequence of the stated results. It may be true—perhaps [e1] operations vanish on HH_per, or B_p is zero there, or [e1] is in the image of the S^1 restriction after all—but the text does not say. This is a load-bearing gap in the central classification, not a cosmetic omission.\n\nThere are smaller gaps too. Theorem 6.18 explicitly defers a technical point to Salvatore; Lemma 6.16 leaves the inverse map to the reader; Lemma 5.7's proof is largely a picture. Those are probably fixable, but they sit in the chain. Section 7 leans on [Che2] and on the strict-unit issue in Footnote 5; that is an acceptable dependence, but Theorem 7.6 is compressed.\n\nThe paper deserves a serious referee, and I would send it. The referee should be asked to pin down the [e1]/B_p step: either prove the missing descent and the cap-product identification, or state explicitly what Theorem 1.1 is claiming after restriction. The cacti construction and Theorem 1.2 are strong enough that the paper merits careful review even if the central classification needs repair.","headline":"A substantial operadic reformulation with a plausible classification whose written proof has a real gap: the [e1] generators are never shown to descend to periodic cyclic homology or to be cap products.","tokens_in":73500,"tokens_out":6045,"would_cite":true,"duration_ms":61913,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["16E40","55P48","53D40","55S12"],"pacs":[],"model":"deepseek-v4-flash","headline":"Kontsevich–Soibelman operations on periodic cyclic homology are generated by p-fold equivariant cap products.","keywords":["periodic cyclic homology","Kontsevich–Soibelman operations","cacti operad","equivariant cap product","Getzler–Gauss-Manin connection","positive characteristic","Fukaya category","quantum Steenrod operations"],"falsifier":"Take a dg algebra over F_p with explicitly computed periodic cyclic homology and search for an endomorphism in the image of Ξ that cannot be written as a polynomial in operators [ϕ]^{C_p} ⋒^{C_p} −; or, at the topological level, check whether the inverse homeomorphism asserted in Lemma 6.16 can actually be made S^1×S^1-equivariant—if not, the generation theorem may lack a proof in its current form.","tokens_in":72528,"feed_emoji":"🌀","tokens_out":8257,"duration_ms":73490,"temperature":0.7,"pith_summary":"This paper sets out to describe every 'natural' endomorphism of the periodic cyclic homology of a differential graded algebra—those induced by the chain-level action of the two-colored Kontsevich–Soibelman operad. The headline claim is a dichotomy: over rings containing the rationals, no such operation is nonzero; over rings containing the field with p elements, the entire algebra of operations is generated by p-fold equivariant cap products with Hochschild cochains. A second, independent theorem says every such operation commutes with the Getzler–Gauss-Manin connection, so the operations are covariantly constant in families of algebras. The technical engine is a reformulation of the operad using cacti with spines and cyclic cacti, together with an equivariant quasi-equivalence to configurations of little disks on a disk and on a cylinder. If correct, this gives a closed-form classification of the operadic symmetries of periodic cyclic homology and places the p-fold cap product—already linked to quantum Steenrod operations—at the center of the story.","feed_headline":"One cap product spans all cyclic-homology symmetries","feed_subtitle":"In positive characteristic, p-fold cap products generate all cyclic-homology endomorphisms—and commute with the Gauss-Manin connection.","key_machinery":"The central object is the two-colored Kontsevich–Soibelman operad, replaced by a 'cacti with spines' model: cacti are unions of embedded circles with marked points and spines, and cyclic cacti add an output basepoint on the base lobe. The paper proves an equivariant quasi-equivalence from this two-colored cacti operad to the configuration-space operad of little disks on a disk and on a cylinder, tracking Σ_k and S^1×S^1 actions; the circle actions are modeled combinatorially using the cyclic category and its finite p-cyclic subcategory. This equivalence lets equivariant homology of KS(k,1) be computed by localization and classical configuration-space homology, producing generators [e0],[e1];","core_discovery":"The paper claims a complete classification of Kontsevich–Soibelman operations on periodic cyclic homology. Over Q ⊂ R, none exist; over F_p ⊂ R, all are generated by p-fold equivariant cap products [ϕ]^{C_p} ⋒^{C_p} −. The generator [e0] is exactly that cap product, and [e1] is its commutator with the residual circle operator, so everything reduces to cap products. Covariant constancy is proved independently: any operation lifts to the first-order neighborhood of the diagonal, giving compatibility with the Grothendieck-connection form of the Gauss-Manin connection.","pith_inferences":["Because the classification runs through equivariant localization, the same dichotomy—trivial over Q, cap-product-generated over F_p—should persist for A-infinity categories and for curved algebras once strict units are replaced by homotopy units; the paper only indicates this.","The commutator formula [e1] = [[e0], B_p] is a C_p-equivariant Cartan-type formula; it suggests a purely algebraic proof of covariant constancy of the p-fold cap product should exist without the configuration-space detour.","The identification of [e0] with the p-fold cap product is chain-level and explicit; one testable consequence is that arithmetic properties of quantum Steenrod operations, such as vanishing of θ-terms, should hold for any A whose periodic cyclic homology is generated by units, not just Fukaya categories.","If the deferred basepoint-rotation technicality cannot be resolved, the generator classification might still hold after replacing the cacti model by a weakly equivalent one; testing this would clarify whether the equivariant cacti comparison is necessary or merely convenient."],"forward_implications":["Over Q ⊂ R, no nonzero Kontsevich–Soibelman operation exists on periodic cyclic homology.","Over F_p ⊂ R, every Kontsevich–Soibelman operation is a composition and linear combination of p-fold equivariant cap products [ϕ]^{C_p} ⋒^{C_p} −.","All such operations are covariantly constant: each commutes with the Getzler–Gauss-Manin connection, in both the S^1 and C_p versions.","For a closed monotone symplectic manifold, p-torsion in integral cohomology forces failure of Abouzaid's generation criterion for the Fukaya category over F_p.","A Lagrangian-realizable middle cohomology class α over F_p must satisfy α ∪ im(β) = 0; concrete Fano examples show this constraint is not implied by classical submanifold realizability or by pairing with the symplectic form."],"fun_headline_variants":["In char p, cap products generate all cyclic-homology ops","Cyclic homology symmetries stem from p-fold caps","Getzler-Gauss-Manin connection tames KS operations","All Kontsevich-Soibelman ops reduce to cap products"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The classification rests on the equivariant quasi-equivalence between the two-colored cacti operad and the configuration-space operad (Theorem 4.29); one rotation-of-basepoints compatibility is deferred to another paper and an inverse homeomorphism in Lemma 6.16 is left to the reader, so the existence of this equivariant equivalence is the load-bearing premise.","fun_headline_variants_meta":{"raw":{"variants":["In char p, cap products generate all cyclic-homology ops","Cyclic homology symmetries stem from p-fold caps","Getzler-Gauss-Manin connection tames KS operations","All Kontsevich-Soibelman ops reduce to cap products"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000239,"raw_usage":{"total_tokens":1337,"prompt_tokens":715,"completion_tokens":622,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":459,"completion_tokens_details":{"reasoning_tokens":551}},"tokens_in":459,"tokens_out":622,"duration_ms":5664,"temperature":1.0,"reasoning_tokens":551,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T08:32:52.650589+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a dg algebra over F_p with explicitly computed periodic cyclic homology and search for an endomorphism in the image of Ξ that cannot be written as a polynomial in operators [ϕ]^{C_p} ⋒^{C_p} −; or, at the topological level, check whether the inverse homeomorphism asserted in Lemma 6.16 can actually be made S^1×S^1-equivariant—if not, the generation theorem may lack a proof in its current form.","supporting_citations":[],"review_version":1}