{"id":"7377f935-7c92-417b-ad4b-b27b4f5ea28c","arxiv_id":"1908.02325","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"Every locally free Poisson module on a klt Poisson projective variety is rationally Morita equivalent to a flat partial sheaf, a meromorphic flat connection, a co-Higgs sheaf, or a special meromorphic co-Higgs sheaf.","lead":"This paper introduces a new 'birational Morita equivalence' for holomorphic Poisson modules and uses it to classify such modules on mildly singular projective varieties into flat, connection-like, or co-Higgs types. The result connects Poisson geometry with foliation theory and bundle theory, and it gives a geometric description of rank-two cases.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.8's Morita span uses an unnormalized pushforward: on a degree-d quasi-étale cover f_*f^*σ = dσ, not σ; the proof needs to rescale both σ̃ and the lifted connection, so the gap is real but repairable.","rationale":"The reader's weakest assumption identifies the same load-bearing step: the construction of the Morita span in Theorem 4.8 depends on f being a Poisson morphism, and the equality f_*σ̃ = σ is not justified for a finite cover of degree d > 1. Our pass confirms that the natural pullback has pushforward dσ, so the text as written contains a genuine gap. We also note an additional consequence the reader did not spell out: repairing f_*σ̃ = σ by scaling the bivector forces a corresponding rescaling of the lifted Poisson connection; otherwise flatness with respect to the scaled bivector fails. This makes the gap more than a typo, but it is likely repairable because all four cases in Theorem 4.8 are invariant under nonzero rescaling of the Poisson structure and the connection. The same issue propagates to Corollaries 5.11 and 5.12, which invoke Theorem 4.8. We considered whether the failure to prove transitivity of rational Morita equivalence is more serious, but transitivity is not needed for the theorem's one-way reduction of a Poisson module to one of the four normal forms. Overall, the central claim is plausible and the flaw is a missing normalization rather than a demonstrated counterexample, so the reader's CONDITIONAL verdict is appropriate and we recommend no change.","tokens_in":17972,"tokens_out":17045,"duration_ms":208093,"concrete_test":"Isolate the step in Theorem 4.8 where f_*σ̃ = σ is claimed. Take a quasi-étale cover f: Z → X of degree 2 with σ a generically nondegenerate Poisson bivector, and compute f_*f^*σ on the smooth locus: it equals 2σ, not σ. Then set τ = (1/2)f^*σ and define the lifted connection matrix θ̃ = (1/2)θ. Verify that δ_τ(θ̃) + θ̃∧θ̃ = 0 follows from the original flatness δ_σ(θ) + θ∧θ = 0. If this verification succeeds, the proof is repairable by an explicit normalization; if it fails, the reduction to (Y,σ2) is invalid.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Theorem 4.8, after [22, Proposition 8.14] produces a quasi-étale cover f: Z = W×Y → X, the text asserts: “Since f is a quasi-étale cover there is a Poisson bivector σ̃ ∈ H^0(Z, ∧²T_Z) such that f_*σ̃ = σ.” For a finite surjective morphism of degree d > 1, the natural pullback satisfies f_*f^*σ = dσ on the étale locus, so the unnormalized claim is false. This matters because the span (X ← Z → Y) is a rational Morita equivalence only if f is Poisson; without f_*σ̃ = σ, the reduction of (E,∇) to (E0,∇0) on Y does not follow, and the four-case split in Theorem 4.8 loses its base. The gap is repairable: take τ = (1/d)f^*σ and define the pulled-back connection by scaling the local matrices by 1/d as well; flatness is preserved because δ_τ(θ̃) + θ̃∧θ̃ = c²(δ_σ(θ) + θ∧θ) with c = 1/d. But the paper does not perform this normalization, and it is not purely cosmetic: the module structure on f^*E must also be rescaled, not merely the bivector. As written, the proof is incomplete at its central reduction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a birational notion of rational Morita equivalence for Poisson modules on complex normal projective Poisson varieties and proves a structural dichotomy theorem (Theorem 1.1, proved as Theorem 4.8): every locally free Poisson module on a klt Poisson projective variety is rationally Morita equivalent to at least one of four normal forms: (a) a flat holomorphic sheaf on a transcendental Poisson variety; (b) a meromorphic flat connection on a generically symplectic variety; (c) a co-Higgs sheaf on a variety with trivial Poisson structure; or (d) a meromorphic co-Higgs sheaf on a transcendental Poisson variety with a Poisson-flatness condition. The proof passes through Druel's structural result for foliations with canonical singularities, a quasi-étale product cover, and Polishchuk's pushforward of Poisson modules. The paper then specializes to rank-two sl2-Poisson modules, obtaining explicit triples of rational vector fields, normal forms for the connection, and a geometric description of the induced symplectic foliation on the projectivization. It also claims a leaf bijection and an isomorphism of Casimir function spaces for rational Morita equivalence.","tokens_in":18282,"tokens_out":12776,"duration_ms":149958,"significance":"If the main theorem is made correct, the paper gives a useful unifying framework: holomorphic flat connections, meromorphic flat connections, and co-Higgs sheaves appear as different manifestations of Poisson modules, and the connection to Druel's and Polishchuk's results is natural. The rank-two normal forms and the projective-foliation interpretation are concrete and potentially useful. The paper is not circular and does not fit parameters; it relies transparently on external theorems. However, the central proof has a normalization gap in the construction of the quasi-étale span, and Proposition 4.2 contains an overclaim about leaf bijections. These issues are technically significant, but the main theorem is plausibly repairable, so the paper deserves a major revision rather than rejection.","major_comments":[{"comment":"After invoking [22, Proposition 8.14], the proof asserts: \"Since f: Z→X is a quasi-étale cover there is a Poisson bivector σ̃∈H^0(Z,∧²T_Z) such that f_*σ̃=σ.\" For a finite morphism of degree d, the natural pullback satisfies f_*(f^*σ)=d·σ on the étale locus, so the unnormalized assertion is false when d>1. This is load-bearing: under Definition 4.1, the span (X←Z→Y) is a rational Morita equivalence only if f is a Poisson morphism, and without f_*σ̃=σ the subsequent pushforward (π2)_*(f^*E,∇̃) need not define a Poisson module on Y. The gap is repairable by taking σ̃=(1/d)f^*σ and lifting the connection as (1/d)f^*∇; flatness is then preserved, since both the Poisson differential and the quadratic curvature term rescale consistently. The paper should carry out this normalization explicitly. As written, the proof is incomplete at the central reduction, and the issue propagates to all cases of Theorem 4.8 and to Corollary 5.11.","section":"Section 4, proof of Theorem 4.8"},{"comment":"The claimed bijection between leaves of the symplectic foliations is asserted without proof and is not a formal consequence of the equality f^{-1}F_{σ1}=h^{-1}F_{σ2}. For example, take S=P^1 with the zero Poisson structure and let f=h be the double cover z↦z². Then f and h are dominant Poisson morphisms and f^{-1}F=h^{-1}F is the point foliation, but the induced map on the set of leaves is two-to-one over all but one point of the target, so there is no bijection of leaves. The statement should either be proved under additional hypotheses (for instance étale or submersion hypotheses) or removed from Proposition 4.2.","section":"Section 4, Proposition 4.2(1)"}],"minor_comments":[{"comment":"The definition says \"h_*ρ = σ1\" for the second arrow; this should presumably be \"h_*ρ = σ2\".","section":"Section 4, Definition 4.1"},{"comment":"In the sentence \"E1 is rationally Morita equivalent to ...\", the symbol E1 should be (E,∇); as written it is an undefined notation.","section":"Section 4, proof of Theorem 4.8"},{"comment":"In the co-Higgs case the text says \"This shows the part (c),\" but the dimension-two pullback foliation corresponds to part (a), not part (c).","section":"Section 5, proof of Corollary 5.12"},{"comment":"The notation \"TY|B\" should be clarified as the relative tangent sheaf T_{Y/B} (or T_{Y/B}⊗ something); as written it is ambiguous.","section":"Theorem 1.1(d) and Corollary 5.12(d)"},{"comment":"The phrase \"a rational vector field v tangent to (Y,0)\" is unclear; presumably v is a rational vector field on Y with the zero Poisson structure, but the notation (Y,0) is not introduced.","section":"Corollary 5.11(c)"},{"comment":"The displayed type \"∇:E→E⊗TX(D)\" should be balanced with the usage elsewhere, which suggests ∇:E→TX⊗E(D); if the tensor factors are intentionally ordered differently, the convention should be stated.","section":"Definition 5.1"}],"recommendation":"major_revision","confidential_remarks":"The normalization issue in Theorem 4.8 is the main obstacle; it is concrete and repairable. The false leaf bijection in Proposition 4.2 should be corrected or withdrawn, but it is not the central engine of the classification. I would be willing to look at a revised version with these points addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe short version: there is a real idea here, and the main proof has a gap that is central but likely fixable. The paper introduces rational Morita equivalence for Poisson modules on klt projective varieties, then uses Druel's structure theorem to split a Poisson module into four birational normal forms: flat along a transcendental foliation, meromorphic flat connection on a generically symplectic variety, co-Higgs on a trivial Poisson variety, or a mixed meromorphic co-Higgs case. That classification is genuinely new. The rank-two sl2-Poisson module analysis in Section 5 is the best part: the triple of vector fields, the projectivized Poisson bivector, and the corollaries tying these modules to transversely projective structures and Riccati foliations are clean and worth keeping.\n\nWhere it gets soft: the proof of Theorem 4.8. After applying [22, Prop 8.14] to get a quasi-étale cover f: Z = W×Y -> X, the text asserts that the pulled-back bivector σ̃ satisfies f_*σ̃ = σ. For a degree-d cover, f_*f^*σ = dσ on the étale locus, so the claim is off by a factor of d. The fix is simple: take σ̃ = (1/d)f^*σ and rescale the lifted connection by 1/d as well; flatness is preserved because both the bivector and the local matrices scale quadratically. But the paper doesn't do this, and without the normalization the span is not a Poisson morphism, so the reduction to (Y, σ2) does not follow as written. The gap is repairable, not fatal, but it sits at the load-bearing step.\n\nTwo smaller issues. Proposition 4.2 asserts a bijection between leaves for any rational Morita equivalence, with no proof; that is plausible for the quasi-étale span used in the theorem, but not for arbitrary dominant Poisson morphisms. And the relation is called an equivalence without a transitivity argument; the paper only needs one direction, so the terminology overreaches. Neither undercuts the main idea.\n\nNo circular reasoning; the external citations (Druel, Polishchuk, Wang) are legitimate, and the author's own prior work is cited appropriately. The examples, especially the Hilbert scheme one, are thoughtful.\n\nWho this is for: Poisson algebraic geometers, and people working on co-Higgs bundles or holomorphic foliations. It deserves a serious referee: the theorem is plausible, the rank-two material is valuable, and a careful referee can either fix the normalization or push the author to do it. I would send it out, with a note that the proof of Theorem 4.8 needs a substantial rewrite.","headline":"A genuinely new framework for Poisson modules, with a central but repairable gap in the proof of the main theorem.","tokens_in":18817,"tokens_out":6740,"would_cite":true,"duration_ms":74086,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["17B63","53D17","70G45","37F75"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that every locally free Poisson module on a klt Poisson projective variety is rationally Morita equivalent to one of four explicit elementary objects: a flat holomorphic sheaf, a meromorphic flat connection, a co-Higgs…","keywords":["Poisson modules","rational Morita equivalence","klt Poisson varieties","co-Higgs sheaves","meromorphic flat connections","holomorphic foliations","symplectic foliations","sl2-Poisson modules"],"falsifier":"Compute the pushforward of the pulled-back Poisson bivector along a degree-two quasi-étale cover of the sort used in Theorem 4.8: if the result is twice the original bivector rather than the original, the cover is not a Poisson morphism and the claimed Morita reduction fails at that step.","tokens_in":17752,"feed_emoji":"📐","tokens_out":14453,"duration_ms":129642,"temperature":0.7,"pith_summary":"Poisson modules are vector bundles with a flat connection that differentiates along the Hamiltonian vector fields of a Poisson structure. This paper introduces rational Morita equivalence, a birational weakening of the usual Morita equivalence: two Poisson varieties are equivalent when a common normal variety maps to both by dominant Poisson morphisms. On projective varieties with mild singularities (klt varieties whose Poisson structure is either generically symplectic or has canonical foliation singularities), the main theorem states that every locally free Poisson module is rationally Morita equivalent to one of four explicit objects: a flat holomorphic sheaf, a sheaf with a meromorphic flat connection, a co-Higgs sheaf, or a meromorphic co-Higgs sheaf with an integrability condition. This matters because it reduces arbitrary Poisson modules to objects whose birational geometry is already accessible. As an application, rank-two $\\mathfrak{sl}_2$-Poisson modules are classified up to these normal forms, and the induced symplectic foliation on their projectivization is described.","feed_headline":"Poisson modules reduce to four elementary types","feed_subtitle":"On mildly singular projective varieties, any Poisson module is rationally Morita equivalent to one of four explicit objects.","key_machinery":"The paper's central object is rational Morita equivalence: a roof $(S,\\varrho)$ of a normal variety with a possibly meromorphic Poisson bivector and two dominant Poisson morphisms to $(X,\\sigma_1)$ and $(Y,\\sigma_2)$. Its load-bearing identity is the pushforward condition $f_*\\varrho=\\sigma_1$ and $h_*\\varrho=\\sigma_2$ that makes the roof a Poisson correspondence. The proof is carried by the symplectic foliation $\\mathcal{F}_\\sigma$ of the Poisson bivector: a klt Poisson variety has either generically symplectic structure or a foliation with canonical singularities, which allows a quasi-étale cover $Z=W\\times Y\\to X$ over which the pulled-back foliation becomes a product. The decisive dichotomy is that a flat Poisson connection $\\nabla:\\mathcal{E}\\to T_X\\otimes\\mathcal{E}$ either lands in $T_{\\mathcal{F}_\\sigma}\\otimes\\mathcal{E}$, giving a flat partial connection, or else projects to a section of $\\mathcal{N}_{\\mathcal{F}_\\sigma}\\otimes\\operatorname{End}(\\mathcal{E})$ with $\\varphi\\wedge\\varphi=0$, i.e. a co-Higgs field. For rank-two $\\mathfrak{sl}_2$-modules the concrete machinery is the trace-free connection matrix $\\nabla=\\delta+\\begin{pmatrix} v_1 & v_2 \\\\ v_0 & -v_1 \\end{pmatrix}$ and the induced Poisson bivector $\\Sigma=\\sigma+(v_0+2v_1z+v_2z^2)\\wedge\\frac{\\partial}{\\partial z}$ on $\\mathbb{P}(\\mathcal{E})$, which encodes the symplectic foliation of the projectivized bundle.","core_discovery":"Theorem 1.1 (restated as Theorem 4.8) is the central claim: given a locally free Poisson module $(\\mathcal{E}, \\nabla)$ on a klt Poisson projective variety $(X,\\sigma)$, there is a rationally Morita equivalent model on which $(\\mathcal{E},\\nabla)$ takes one of four forms: (a) a flat holomorphic sheaf on a transcendental Poisson variety; (b) a meromorphic flat connection on a generically symplectic variety; (c) a co-Higgs sheaf (a sheaf with a field $\\varphi\\in H^0(X,T_X\\otimes\\operatorname{End}(\\mathcal{E}))$ satisfying $\\varphi\\wedge\\varphi=0$) on a variety with trivial Poisson structure; or (d) a meromorphic co-Higgs sheaf $(\\mathcal{E}_0,\\psi)$ on a transcendental Poisson variety $(Y,\\sigma_0)$, equipped with a rational map $\\zeta:Y\\dashrightarrow B$ with $\\dim B=\\dim \\mathcal{F}_{\\sigma_0}$, whose co-Higgs field $\\psi$ is tangent to the pullback tangent sheaf and is annihilated by a meromorphic Poisson connection $D_0$ with $D_0(\\psi)=0$. The proof reduces the symplectic foliation $\\mathcal{F}_\\sigma$ to a product over a quasi-étale cover, then applies a dichotomy: a flat Poisson connection either factors through the tangent sheaf of the foliation, producing a flat partial connection, or it induces a section $\\varphi$ of $\\mathcal{N}_{\\mathcal{F}_\\sigma}\\otimes \\operatorname{End}(\\mathcal{E})$ satisfying $\\varphi\\wedge\\varphi=0$. The four cases of the theorem come from combining this dichotomy with whether the pushed-forward Poisson structure is zero, generically symplectic, or transcendental.","pith_inferences":["The dichotomy in the proof—flat connection either tangent to the symplectic foliation or inducing a co-Higgs field—suggests that the category of Poisson modules may admit a semiorthogonal decomposition into a flat part and a co-Higgs part, mirroring the four cases of Theorem 1.1.","One testable extension is to apply the rank-two normal form to a concrete Poisson threefold: write an explicit $\\mathfrak{sl}_2$-Poisson connection, extract the triple $(v_0,v_1,v_2)$, and verify which of the cases in Corollary 5.11 is realized; this would give a computational check of the structure theorem.","The methods seem adaptable to reflexive coherent Poisson modules, not just locally free ones, because the key co-Higgs field construction uses a codimension-at-least-two singular set and flatness rather than local freeness; proving the theorem in that setting would broaden its scope to torsion-free sheaves."],"forward_implications":["If Theorem 1.1 is correct, invariants of flat sheaves, meromorphic connections, and co-Higgs sheaves become invariants of arbitrary Poisson modules on klt Poisson projective varieties, up to rational Morita equivalence.","For rank-two meromorphic $\\mathfrak{sl}_2$-Poisson modules, the normal forms show that the geometry is controlled by a triple of rational vector fields satisfying the equations $\\delta(v_0)=v_0\\wedge v_1$, $\\delta(v_1)=2v_0\\wedge v_2$, and $\\delta(v_2)=v_1\\wedge v_2$, linking the theory to transversely projective foliations.","The induced symplectic foliation on $\\mathbb{P}(\\mathcal{E})$ for a rank-two $\\mathfrak{sl}_2$-Poisson module is either a two-dimensional pullback of a curve foliation, a codimension-one Riccati foliation, or the pullback of a foliation generated by a rational Poisson vector field and the symplectic foliation; in dimension $2k+1$ the induced map to the base is Poisson.","The proof yields functors from the category of Poisson modules on $(X,\\sigma)$ either to co-Higgs bundles on the base $Y$ (when the pushed-forward Poisson structure is trivial) or to meromorphic connections along the degeneracy divisor (when it is generically symplectic), so categorical questions about Poisson modules can be transferred to those better-studied categories."],"supporting_citations":[{"why":"Supplies the quasi-étale cover that reduces the symplectic foliation to a product, the first step of Theorem 4.8.","marker":"[22]"},{"why":"Provides the pushforward construction for Poisson modules and the trace-free connection form used throughout Section 5.","marker":"[40]"},{"why":"Gives the flatness argument producing a co-Higgs field with vanishing wedge-square and the Poisson connection that annihilates it.","marker":"[48]"},{"why":"Establishes the correspondence between Poisson flat connections and logarithmic or meromorphic flat connections on generically symplectic varieties.","marker":"[41]"},{"why":"Defines the transversely projective structures used to interpret the rank-two $\\mathfrak{sl}_2$-Poisson normal forms.","marker":"[46]"}],"fun_headline_variants":["Poisson modules reduce to four models via rational Morita equivalence","Rational Morita equivalence tames Poisson modules to four types","Poisson modules on klt varieties: rational Morita classification","Four explicit Poisson module forms from a rational Morita equivalence"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The reduction rests on the assumption that the quasi-étale cover that simplifies the symplectic foliation is a Poisson morphism, so that the pulled-back Poisson structure pushes forward to the original one.","fun_headline_variants_meta":{"raw":{"variants":["Poisson modules reduce to four models via rational Morita equivalence","Rational Morita equivalence tames Poisson modules to four types","Poisson modules on klt varieties: rational Morita classification","Four explicit Poisson module forms from a rational Morita equivalence"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000906,"raw_usage":{"total_tokens":3939,"prompt_tokens":1032,"completion_tokens":2907,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":648,"completion_tokens_details":{"reasoning_tokens":2838}},"tokens_in":648,"tokens_out":2907,"duration_ms":21517,"temperature":1.0,"reasoning_tokens":2838,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:49:24.841226+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the pushforward of the pulled-back Poisson bivector along a degree-two quasi-étale cover of the sort used in Theorem 4.8: if the result is twice the original bivector rather than the original, the cover is not a Poisson morphism and the claimed Morita reduction fails at that step.","supporting_citations":[{"cited_title":"Polishchuk, Algebraic geometry of Poisson brackets","cited_arxiv_id":null,"evidence_quote":"Provides the pushforward construction for Poisson modules and the trace-free connection form used throughout Section 5."},{"cited_title":"W ang, Generalized holomorphic structures , J","cited_arxiv_id":null,"evidence_quote":"Gives the flatness argument producing a co-Higgs field with vanishing wedge-square and the Poisson connection that annihilates it."},{"cited_title":"Pym, Poisson Structures and Lie Algebroids in Complex Geometry , Phd","cited_arxiv_id":null,"evidence_quote":"Establishes the correspondence between Poisson flat connections and logarithmic or meromorphic flat connections on generically symplectic varieties."},{"cited_title":"Sc´ ardua,Transversely affine and transversely projective holomorphi c foliations , Ann","cited_arxiv_id":null,"evidence_quote":"Defines the transversely projective structures used to interpret the rank-two $\\mathfrak{sl}_2$-Poisson normal forms."}],"review_version":1}