{"id":"ca83098c-680d-4d0f-9f9c-c1aca8317a9a","arxiv_id":"2501.13792","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Infinitesimal reduction-compatible star products are exactly bivector fields plus symmetric differential operators built from the characteristic distribution and the normal bundle, up to constraint equivalence.","lead":"This paper computes the second constraint Hochschild cohomology, which classifies infinitesimal star products compatible with coisotropic reduction. It gives a complete description of the first-order deformations of functions on a Poisson manifold that survive reduction, including symmetric terms invisible to the classical Hochschild-Kostant-Rosenberg isomorphism.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Equation (3.10), the global homotopy decomposition imported from [Dip+24], is false as stated for arbitrary cochains; surjectivity in Theorem 3.6 needs a cocycle version that the paper neither states nor proves.","rationale":"The reader's weakest assumption is that (3.10) is imported from a same-group preprint without hypotheses or proof, and that surjectivity depends on it. I agree that this is the main load-bearing dependency. My stress test sharpens the concern: (3.10) as printed is not merely unproved, it is false for arbitrary cochains, by a dimension count in the degree-(2,1) part of C^2_ca(M). The proof only needs the formula for d-closed φ, where a chain homotopy identity with the s(dφ) term would suffice, but the paper does not state that corrected identity. This is an internal gap, not just an unsupported citation, and it reinforces the CONDITIONAL verdict. I also noticed a smaller but real statement-level issue: the domain of U includes degree-one symmetric elements in Γ∞(TC^⊥), which are killed by d; the theorem should restrict to symmetric degree at least 2. This does not change the verdict because it is easily repaired once the homotopy issue is resolved. I keep the reader's CONDITIONAL since the central classification may well be true, but the proof as written rests on an unverified and, as stated, overbroad lemma.","tokens_in":12109,"tokens_out":26296,"duration_ms":247400,"concrete_test":"Verify the cocycle version of (3.10) from [Dip+24, arXiv:2410.15903]: identify the chain homotopy s explicitly, check that for d-closed φ∈C^2_ca(M)_N it yields φ = hkr(pr φ) + d(sφ), and re-run the surjectivity argument of Theorem 3.6 with that exact s. As a numerical check in V=span{x,y}, compute whether (x∨y)⊗x lies in im d; it does not, confirming the all-cochain formulation is false. If [Dip+24] does not provide such an s on arbitrary constraint manifolds with simple distributions, the theorem lacks support.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The surjectivity step of Theorem 3.6 starts from (3.10): every φ1⊗...⊗φn ∈ C^n_ca(M) is written as hkr(pr1(φ1)∧...∧pr1(φn)) + dH(φ). As a statement about all cochains this is dimensionally impossible. In a two-dimensional vector space V with basis x,y, take φ=(x∨y)⊗x ∈ S^2V⊗V ⊂ C^2_ca(M). Then pr1(x∨y)=0, so the hkr term vanishes, but φ is not in the image of d:S^3V → S^2V⊗V ⊕ V⊗S^2V: S^2V⊗V has dimension 6, whereas the degree-3 part of im d has dimension 4, and a basis computation shows (x∨y)⊗x is not hit. Thus (3.10) can only be intended for d-closed φ, where a chain homotopy would give φ = hkr(pr φ) + dH + s(dφ) with s(dφ)=0. The paper applies (3.10) only to a cocycle φ∈C^2_ca(M)_N, so this correction may suffice, but the needed homotopy identity is not stated with hypotheses or proven. Separately, the theorem's domain SΓ∞(D)∨Γ∞(TC^⊥) as written contains degree-one elements ψ with dψ=0, giving U(0,ψ)=0; injectivity requires restricting to S^{≥2} or quotienting by degree one.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces constraint manifolds M=(M,C,D) and studies infinitesimal deformations of the associated constraint algebra C∞(M) that are compatible with coisotropic reduction. The main result, Theorem 3.6, claims an isomorphism U: X2(M)_N ⊕ SΓ∞(D)∨Γ∞(TC^⊥) → H2_diff(M)_N given by U(X,ψ)=hkr(X)+[Op∇(dψ)], with Corollary 3.10 identifying constraint equivalence classes of infinitesimal constraint star products with that vector space. The paper also proves Proposition 3.8 for the relative cohomology H2_diff(M)_0 and Proposition 3.9 relating constraint equivalence to exactness in the constraint Hochschild complex.","tokens_in":12440,"tokens_out":14989,"duration_ms":132295,"significance":"If Theorem 3.6 is correct, the paper gives a concrete and useful description of infinitesimal deformations compatible with coisotropic reduction, including symmetric bidifferential operators of arbitrarily high order that are invisible in ordinary Hochschild cohomology and that become trivial after reduction. The conceptual identification of the new symmetric contribution SΓ∞(D)∨Γ∞(TC^⊥), the illustrative Example 3.2, and the clean deformation-theoretic translation in Proposition 3.9 are valuable parts of the paper. However, the proof of the central surjectivity statement rests on an externally imported global homotopy decomposition that is stated in a form that is false, and the theorem statement itself has an injectivity problem because its second summand contains degree-one elements on which U vanishes. These issues are load-bearing and require substantive repair before the main claim can be accepted.","major_comments":[{"comment":"The domain of U in (3.21) includes the degree-one summand Γ∞(TC^⊥), but for ψ∈Γ∞(TC^⊥) we have dψ=0 by Lemma 3.3(ii), so U(0,ψ)=0. Thus U is not injective as stated and Theorem 3.6 is false without a restriction to symmetric degree at least 2. The same correction must be made in Proposition 3.8 and Corollary 3.10, or the notation SΓ∞(D)∨Γ∞(TC^⊥) must be explicitly defined to mean the direct sum over k≥2 of S^{k-1}Γ∞(D)∨Γ∞(TC^⊥).","section":"§3.2, Theorem 3.6 and Corollary 3.10"},{"comment":"Equation (3.10) is stated for every cochain φ1⊗...⊗φn in C^n_ca(M), but in that generality it is false. In a two-dimensional vector space V with basis x,y, take φ=(x∨y)⊗x ∈ S^2V⊗V ⊂ C^2_ca(M). Then pr1(x∨y)=0, so the HKR term vanishes, while φ is not in the image of d:S^3V→S^2V⊗V⊕V⊗S^2V: a direct basis computation shows that the image has dimension 4 in a 6-dimensional target and that the component (x∨y)⊗x has no preimage. The surjectivity proof of Theorem 3.6 applies (3.10) only to a cocycle, so a cocycle-level homotopy identity with hypotheses could suffice, but that correct statement is neither proved nor explicitly cited. Since (3.10) is the starting point for the surjectivity argument, the authors must either prove the corrected identity or cite the precise theorem from [Dip+24] with the necessary hypotheses.","section":"§3.2, Eq. (3.10)"},{"comment":"The surjectivity argument contains several compressed steps that are load-bearing and are not justified as written. After obtaining ψ_{T/N} ∈ (SX(M))_{T/N} with dψ_{T/N} ∈ C^2_ca(M)_N, the proof asserts that one may assume ψ_{T/N}=X1∨...∨Xk is a factorizing tensor with k≥2; for a sum of monomials, dψ∈N does not automatically imply that d of each monomial lies in N, so this reduction needs a separate argument. The later claim that the ℓ=1 sum collapses to a single summand with exactly one Xi∈Γ∞(TC^⊥) and all other Xi∈Γ∞(D) is also asserted rather than proved. These steps are essential for concluding ψ_{T/N} ∈ S^{n-1}Γ∞(D)∨Γ∞(TC^⊥), so the proof of surjectivity is incomplete as written.","section":"§3.2, proof of Theorem 3.6"}],"minor_comments":[{"comment":"The proof states that hkr(X)=[0] implies hkr(X)=0 because hkr(X) is totally antisymmetric while δD is not; this is false, since exact cochains can be antisymmetric, for example δD=0 for D a derivation. The conclusion follows more simply from the classical HKR quasi-isomorphism after observing that exactness in the constraint subcomplex implies exactness in the full Hochschild complex, so the proof should be replaced by that argument.","section":"§3.1, Proposition 3.1(iii)"},{"comment":"The phrase 'ker(d|SX)=X(M)' in the proof of Theorem 3.6 should be formulated more precisely: d vanishes on S^1X(M) and is injective on S^kX(M) for k≥2. As written, it could be confused with a statement about a differential having a large kernel on all symmetric degrees.","section":"§3.2, Lemma 3.3"},{"comment":"The notation X(M)_{T/N} is used in the proof without being defined; the reader can infer it from (S^1X(M))_{T/N}=Γ∞(TC^⊥), but this should be stated explicitly.","section":"§3.2, Eq. (3.16) and surrounding notation"},{"comment":"The operator ∂²/(∂x_{n0}∂x_{nT}) is said to lie in DiffOp^1; this is correct for univariate differential operators but may confuse readers who expect DiffOp^n to denote n-linear operators of order one in each argument. A brief clarifying phrase would help.","section":"Example 3.2"},{"comment":"The citation to [Dip+24] should include the specific theorem or proposition number in which the global homotopy decomposition is proved, since the statement as reproduced in this paper is not correct for arbitrary cochains.","section":"§3.2, Eq. (3.10)"}],"recommendation":"major_revision","confidential_remarks":"The main theorem depends crucially on the global homotopy decomposition (3.10) imported from [Dip+24], a preprint by the same group. The version stated in the manuscript is false, so the editor may wish to verify that the corrected cocycle-level statement appears in [Dip+24] with full hypotheses, or that the authors provide a self-contained proof in the revision. Apart from this, the paper's framework is coherent and the intended result is plausible, but the proof and statement of Theorem 3.6 need substantial revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new thing here is a theorem: for a constraint manifold, the second constraint Hochschild cohomology is X2(M)_N plus a symmetric piece built from D and TC^⊥. That symmetric contribution—bidifferential operators of arbitrarily high order that vanish in ordinary Hochschild cohomology—is genuinely new and the natural completion of the H0/H1 result in [DEW22]. The HKR-injectivity part is clean, and the interpretation of the symmetric classes as genuinely different infinitesimal star products that become equivalent after forgetting reduction is well drawn.\n\nWhere it gets soft: (3.10) is stated for all cochains, but that cannot be right—the stress-test counterexample (x∨y)⊗x in two dimensions is a legitimate cochain that the leading term misses. The paper only applies (3.10) to a d-closed cochain, so a homotopy version with the cocycle condition explicitly stated may be enough, but that version is neither stated nor proved. Since surjectivity of U leans on this decomposition, the proof as written has a real gap. Second, the domain of U in Theorem 3.6 includes degree-one symmetric elements, but dψ=0 for those, so U(0,ψ)=0; the injectivity argument itself restricts to k≥2. The statement needs to be repaired to S^≥2 or the degree-one part quotiented out. Minor, but it makes the theorem literally false as printed. Third, the step 'we can assume ψT/N is factorizing' and the collapse argument are compressed; probably fillable, but it should be spelled out.\n\nI don't see a hidden contradiction in the main idea, and the dependence on [Dip+24] for the global homotopy is not a problem by itself—it's the same group's preprint, but the hypotheses need to be precise. I would send this to review, but only after the author fixes (3.10) and the degree-one issue. A serious referee should check whether the cocycle version of (3.10) is actually proved in [Dip+24]; if yes, the paper becomes a solid extension. Worth putting in front of a deformation quantization audience.","headline":"First computation of H2_diff(M)_N with a new symmetric piece, but the proof needs a repaired (3.10) and a domain fix before it's fully trustworthy.","tokens_in":12992,"tokens_out":3033,"would_cite":false,"duration_ms":25144,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53D55","16E40","53D17"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper classifies the infinitesimal star products on a Poisson manifold that are compatible with coisotropic reduction, showing they form a vector space of constraint bivector fields plus symmetric high-order differential operators.","keywords":["deformation quantization","infinitesimal star products","coisotropic reduction","constraint Hochschild cohomology","Poisson manifolds","constraint symbol calculus","Hochschild-Kostant-Rosenberg theorem"],"falsifier":"A direct computation of $H^2_{\\mathrm{diff}}(\\mathcal{M})_N$ for a constraint manifold that lacks a global tubular-neighbourhood splitting of $TC=D\\oplus D^\\perp$ (or for which decomposition (3.10) is false) would settle the matter: any result other than $X^2(\\mathcal{M})_N\\oplus S\\Gamma^\\infty(D)\\vee\\Gamma^\\infty(TC^\\perp)$ falsifies Theorem 3.6 in full generality.","tokens_in":11880,"feed_emoji":"✨","tokens_out":10998,"duration_ms":86838,"temperature":0.7,"pith_summary":"This paper classifies the infinitesimal star products on a Poisson manifold that are compatible with coisotropic reduction: given a coisotropic submanifold $C$ with a simple characteristic distribution $D$, a star product is compatible when it descends to a star product on the reduced manifold $M_{\\mathrm{red}}=C/D$. The main theorem identifies the equivalence classes of such first-order deformations with a vector space built from two pieces: constraint bivector fields $X^2(\\mathcal{M})_N$, and symmetric elements $S\\Gamma^\\infty(D)\\vee\\Gamma^\\infty(TC^\\perp)$ formed from the characteristic distribution and the normal directions to the constraint submanifold. The second piece is new: it yields symmetric bidifferential operators of arbitrarily high order that ordinary Hochschild cohomology cannot detect, yet they are genuinely inequivalent once reduction compatibility is required. The paper thereby makes the deformation-theoretic part of 'quantization commutes with reduction' computable in a general coisotropic setting.","feed_headline":"Infinitesimal star products surviving reduction are classified","feed_subtitle":"Reduction-compatible quantizations split into constraint bivectors plus invisible symmetric high-order operators.","key_machinery":"The carrying object is the constraint symbol calculus: a torsion-free constraint covariant derivative $\\nabla$ gives an isomorphism $\\mathrm{Op}_\\nabla$ from a constraint version of the tensor algebra $T^\\bullet SX(\\mathcal{M})$ onto constraint multi-differential operators, turning the Hochschild differential into the differential $d$ built from the reduced shuffle coproduct. The classical Hochschild-Kostant-Rosenberg map $\\mathrm{hkr}$ embeds constraint multivector fields into the cohomology, and the additional cohomology is shown to come exclusively from differentials $d\\psi$ of elements $\\psi\\in S\\Gamma^\\infty(D)\\vee\\Gamma^\\infty(TC^\\perp)$. The proof's engine is the global homotopy decomposition $\\varphi=\\mathrm{hkr}(\\mathrm{pr}_1(\\varphi_1)\\wedge\\cdots\\wedge\\mathrm{pr}_1(\\varphi_n))+dH(\\varphi)$, which lets every constraint cocycle be split into an antisymmetric HKR part and a differential of a symmetric part; Lemma 3.5 then controls where these symmetric differentials can land.","core_discovery":"The central claim is Theorem 3.6: for every constraint manifold $\\mathcal{M}=(M,C,D)$ and every torsion-free constraint covariant derivative $\\nabla$, the map $U(X,\\psi)=\\mathrm{hkr}(X)+[\\mathrm{Op}_\\nabla(d\\psi)]$ is an isomorphism from $X^2(\\mathcal{M})_N\\oplus S\\Gamma^\\infty(D)\\vee\\Gamma^\\infty(TC^\\perp)$ onto the second constraint Hochschild cohomology $H^2_{\\mathrm{diff}}(\\mathcal{M})_N$. This means every class controlling an infinitesimal constraint deformation is uniquely a sum of an antisymmetric Hochschild-Kostant-Rosenberg part coming from a constraint bivector and a symmetric part produced by differentiating a symmetric product of one transverse-to-$C$ vector field with any number of distribution-tangent vector fields. The symmetric classes are exact in the ordinary Hochschild complex but not in the constraint subcomplex, so they correspond to infinitesimal star products that are equivalent as ordinary deformations yet inequivalent when the equivalence must preserve reduction. Corollary 3.10 turns this into the classification of constraint equivalence classes of infinitesimal constraint star products, and Proposition 3.9 explains why the first nonvanishing difference of two constraint star products is governed by this cohomology.","pith_inferences":["A natural extension is to test whether the same two-summand pattern computes higher constraint Hochschild cohomologies $H^k_{\\mathrm{diff}}(\\mathcal{M})_N$; the mechanism here suggests that symmetric elements in additional tensor slots would appear.","If this infinitesimal classification lifts to formal star products, it would turn 'quantization commutes with coisotropic reduction' into a classification statement rather than an existence question.","Because the $\\psi$-classes vanish under ordinary equivalence, reduction schemes that track only antisymmetric first-order data would miss them; a concrete test is to compute $H^2_{\\mathrm{diff}}(\\mathcal{M})_N$ on a torus with a linear distribution and compare the result with the ordinary Hochschild cohomology."],"forward_implications":["Constraint equivalence classes of infinitesimal constraint star products are exactly $X^2(\\mathcal{M})_N\\oplus S\\Gamma^\\infty(D)\\vee\\Gamma^\\infty(TC^\\perp)$, by Corollary 3.10.","There exist infinitesimal constraint star products that are inequivalent under reduction-compatible equivalences even though they are equivalent as ordinary star products; these are precisely the nonzero $\\psi$-classes.","All these infinitesimal constraint star products become equivalent after reduction: the canonical map to $H^2_{\\mathrm{diff}}(\\mathcal{M}_{\\mathrm{red}})$ sends $(X,\\psi)$ to the class of $X$.","The sub-cohomology $H^2_{\\mathrm{diff}}(\\mathcal{M})_0$, governing deformations inside the vanishing ideal, is isomorphic to $X^2(\\mathcal{M})_0\\oplus S\\Gamma^\\infty(D)\\vee\\Gamma^\\infty(TC^\\perp)$ by Proposition 3.8."],"supporting_citations":[{"why":"Supplies the Hochschild-Kostant-Rosenberg quasi-isomorphism $\\mathrm{hkr}$ that identifies the antisymmetric part of every deformation class.","marker":"[HKR62]"},{"why":"Introduces constraint Hochschild cohomology, proves the degree-zero and degree-one cases, and defines the deformation-theoretic framework used in Proposition 3.9.","marker":"[DEW22]"},{"why":"Provides the global homotopy decomposition (equation (3.10)) that is the key input for the surjectivity step of Theorem 3.6.","marker":"[Dip+24]"},{"why":"Develops the constraint symbol calculus for multi-differential operators that the paper extends and applies.","marker":"[Dip23]"},{"why":"Gives the adapted dual basis and constraint vector-bundle constructions used to produce constraint covariant derivatives.","marker":"[DK23]"},{"why":"Supplies the classical full symbol calculus isomorphism $\\mathrm{Op}_\\nabla$ for smooth manifolds, which the constraint calculus builds on.","marker":"[Pal65]"}],"fun_headline_variants":["Reduction-compatible star products: split into bivector + invisible part","Every reduction-compatible star product has a hidden symmetric piece","Star products that respect reduction: all classes now known","Reduction reveals hidden distinctions in star products","Infinitesimal star products under coisotropic reduction fully classified"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The surjectivity argument assumes the global homotopy decomposition (equation (3.10)), imported without proof from another preprint, holds for every cocycle on the present class of constraint manifolds; if that decomposition fails, the classification has no support.","fun_headline_variants_meta":{"raw":{"variants":["Reduction-compatible star products: split into bivector + invisible part","Every reduction-compatible star product has a hidden symmetric piece","Star products that respect reduction: all classes now known","Reduction reveals hidden distinctions in star products","Infinitesimal star products under coisotropic reduction fully classified"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00085,"raw_usage":{"total_tokens":3636,"prompt_tokens":827,"completion_tokens":2809,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":443,"completion_tokens_details":{"reasoning_tokens":2726}},"tokens_in":443,"tokens_out":2809,"duration_ms":18714,"temperature":1.0,"reasoning_tokens":2726,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T15:35:32.182053+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct computation of $H^2_{\\mathrm{diff}}(\\mathcal{M})_N$ for a constraint manifold that lacks a global tubular-neighbourhood splitting of $TC=D\\oplus D^\\perp$ (or for which decomposition (3.10) is false) would settle the matter: any result other than $X^2(\\mathcal{M})_N\\oplus S\\Gamma^\\infty(D)\\vee\\Gamma^\\infty(TC^\\perp)$ falsifies Theorem 3.6 in full generality.","supporting_citations":[],"review_version":1}