{"id":"6f2c9ceb-3234-4054-b579-35d86fe54ef9","arxiv_id":"1908.05258","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Bogomolov-Guan manifolds have unobstructed holomorphically symplectic deformations and admit a Beauville-Bogomolov-Fujiki form satisfying the Fujiki formula.","lead":"This paper proves that certain non-Kähler, simply-connected holomorphically symplectic manifolds, called Bogomolov-Guan manifolds, behave like hyperkähler manifolds: their deformations are unobstructed and their second cohomology carries a Beauville-Bogomolov-Fujiki form. A reader interested in complex geometry will find a new class of manifolds where period-map and Fujiki-form techniques from hyperkähler geometry work without the Kähler condition.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.3's central Fujiki formula rests on Proposition 4.4, whose proof as written is invalid: the proposition is false without a homogeneity hypothesis, and the n-th root/analytic-continuation step does not establish a global quadratic form.","rationale":"I agree with the reader that Proposition 4.4 is the weakest point of the paper. The concern is stronger than a mere proof gap: the proposition as stated is false, since it lacks a homogeneity assumption on Q. The intended application, however, uses only the homogeneous polynomial Q(η)=∫η^{2n}, so the main theorem may still be repairable. The homogeneous version is plausible: if the Hodge rotations are taken with respect to a fixed metric, their infinitesimal generators over an open set of 2-planes span ∧²H², hence generate so(H²), whose invariant homogeneous polynomials are generated by the quadratic form. But the paper does not supply this argument, nor does it address the dependence of the rotation subgroups on a varying Hodge metric. The geometric parts, especially Theorem 3.9 and Proposition 5.9, are coherent and are not the source of my concern. Because the gap is localized in §4.2 and likely repairable, I would not reject the paper; the appropriate action is to require a complete proof or a precise reference for the homogeneous version of Proposition 4.4, together with an explicit homogeneity hypothesis. This leaves the verdict CONDITIONAL.","tokens_in":22949,"tokens_out":22006,"duration_ms":237101,"concrete_test":"Run the following check in a computer algebra system. (1) Verify that Q=r^4+r^2 on V=R^3 is invariant under G=SO(3), that every 2-plane admits the required S^1 subgroup, and that no quadratic q satisfies Q=λq^2; this refutes Proposition 4.4 as stated. (2) For the repaired homogeneous version needed in Theorem 4.3, compute in the same system the Lie algebra generated by the infinitesimal generators J_P over an open set S of 2-planes in V=R^4 with n=2: if the generators span all of so(V), the invariant ring in degree 4 is spanned by q^2 and the theorem is recoverable; if the span is smaller and a non-q^2 invariant of degree 4 exists, Theorem 4.3 fails. This distinguishes a fixable over-broad statement from a genuine obstruction.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Proposition 4.4 is the only bridge from openness of the period map (Corollary 3.12) to the BBF form in Theorem 4.3. As stated, the proposition is false: take V=R^3, G=SO(3), S=Gr(2,V), and Q=r^4+r^2. It is G-invariant, and every plane admits the required S^1 rotation subgroup, but Q is not proportional to q^2 for any quadratic q, because the conclusion forces homogeneity of degree 4. The proof in §4.2 also does not establish the homogeneous case. 'Any rotation-invariant polynomial on R^2 is a power of quadratic form' is only true for homogeneous polynomials; the definition q = n√(±Q) requires a global choice of sign and branch, and the claim that q is quadratic on each P∈S does not show the same function extends to a polynomial on V. The step 'd²/dxdy q = 0, hence by analytic continuation everywhere' is asserted: q is only defined on the union of the planes in S (when n is even, only up to sign), no extension across V\\US or regularity at zeros is proved, and the argument ignores that the rotation subgroups in the application are not given relative to a fixed metric on H^2. Since Corollary 5.10 inherits Theorem 4.3, this gap is load-bearing; a complete proof or reference is required.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies compact non-Kähler holomorphically symplectic manifolds, focusing on Bogomolov–Guan (BG) manifolds. It proves unobstructedness of holomorphically symplectic deformations under hypotheses that hold for BG manifolds, establishes a local Torelli statement, and then uses the resulting openness of the period map to derive a Fujiki-type formula and a Beauville–Bogomolov–Fujiki (BBF) form on H^2 for a class of non-Kähler holomorphically symplectic manifolds (Theorem 4.3). Applied to BG manifolds, this yields a smooth deformation space and a BBF form (Corollary 5.10). The deformation-theoretic part largely follows and adapts the Kaledin–Verbitsky framework, while the BBF-form part relies on a new algebraic proposition about polynomial invariants of Lie groups.","tokens_in":23238,"tokens_out":12900,"duration_ms":131771,"significance":"If the proof of Proposition 4.4 can be completed, the paper gives a meaningful extension of hyperkähler deformation theory and of the Beauville–Bogomolov–Fujiki formalism to non-Kähler holomorphically symplectic manifolds. The deformation argument is mostly self-contained and follows the well-established [KV] approach; the BBF form is derived from the period map and the algebraic lemma rather than obtained by fitting constants, which is a genuine virtue. The main risk to the central claim is the algebraic invariant-theory lemma, whose proof as written is incomplete.","major_comments":[{"comment":"Proposition 4.4 is false as stated. Take V=R^3, G=SO(3) with the standard representation, S=Gr(2,V), and Q(x)=||x||^4+||x||^2. For each P in S the subgroup of rotations about the normal axis acts by rotations on P and trivially on V/P, so the hypotheses hold, but Q is not proportional to q^n for any quadratic form q. The missing hypothesis is that Q is homogeneous of degree 2n, and the proposition also never defines n. Since Theorem 4.3 applies the proposition to Q(η)=∫_M η^{2n}, which is homogeneous of degree 2n, the statement is repairable by adding this hypothesis, but as written it is false.","section":"§4.2, Proposition 4.4"},{"comment":"The proof of the homogeneous case is also incomplete. The assertion that any rotation-invariant polynomial on R^2 is a power of a quadratic form is only true for homogeneous polynomials. The function q:=n-th root of ±Q is not shown to be well defined: for even n the sign requires a global choice on the whole set US, the behavior at points where Q=0 is not analyzed, and no argument is given that the plane-by-plane quadratic functions q|_P glue to a single quadratic polynomial on V. The step 'this function is real algebraic, hence by analytic continuation d^2/dxdy q=0 everywhere' is not justified, because q is only defined on US and, for even n, only up to sign, not on V, and its regularity at Q=0 is not established. The possible variation of the constant λ(P) from plane to plane is precisely what must be controlled, and the proof does not address it.","section":"§4.2, proof of Proposition 4.4"},{"comment":"The first sentence of the proof asserts that after shrinking U all complex structures I in U satisfy the assumptions of Theorem 4.3. This is not automatic for non-Kähler deformations: it is not shown that Hodge decomposition on H^2, the equality h^{2,0}=h^{0,2}=1, or the vanishing of ∂-exact holomorphic 3-forms are open conditions in the holomorphically symplectic deformation space. In particular, Corollary 3.12 says the period map to Gr(2,H^2) has open image, but it does not by itself imply h^{2,0}(I)=1 for every nearby I; without that, the associated Hodge rotation ρ_I does not act trivially on V/P, and the hypotheses of Proposition 4.4 are not satisfied.","section":"§4.1, proof of Theorem 4.3"}],"minor_comments":[{"comment":"The reference key [G] is used for two different papers, one by P. Gauduchon and one by É. Ghys; these should have distinct keys.","section":"References"},{"comment":"There are typos such as 'Clealry' in Claim 2.18, 'characterstic' in Section 5.2, and 'impliying' in the proof of Corollary 3.12.","section":"§2.4 and §5.2"},{"comment":"The commutative diagram in (2.2) is difficult to parse; a clearer layout would help the reader follow the construction of the Bogomolov–Guan manifold.","section":"Diagram (2.2)"},{"comment":"The differential of the period map to the Grassmannian is described only verbally; writing out the tangent map would make the openness argument easier to verify.","section":"Corollary 3.12"}],"recommendation":"major_revision","confidential_remarks":"The main results are interesting and likely correct, but the paper's weakest point is Proposition 4.4. If the authors supply a correct proof with the homogeneity hypothesis and a rigorous extension argument, the paper should be suitable for publication. The reliance on [KV] is acceptable, since that theorem has independent standing."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nYou should know two things. First, the paper does something real: it proves unobstructedness of holomorphically symplectic deformations and a local Torelli theorem for compact non-Kähler holomorphically symplectic manifolds under much weaker hypotheses than [KV], and it checks those hypotheses for Bogomolov-Guan manifolds. That part reads as a genuine extension and is likely correct. Second, the advertised BBF form for BG manifolds rests on Proposition 4.4, and as written that proposition is false and its proof is not salvageable in the present form.\n\nThe good news first. Theorem 3.9 is a clean generalization of the Kaledin-Verbitsky unobstructedness result, with the holomorphically symplectic Tian-Todorov lemma at the center. Lemma 3.10 is terse but the argument is standard. Proposition 5.9 gives what is needed for BG manifolds: Hodge decomposition on H^2 and vanishing of holomorphic 3-forms. The topological and complex-geometric input from Guan and Bogomolov is handled honestly. If the BBF gap is fixed, the paper will be a solid contribution.\n\nNow the soft spot. Proposition 4.4 as stated is false: take V=R^3, G=SO(3), S an open set of 2-planes, and Q=r^4+r^2. Every plane admits the required S^1 rotation subgroup, and Q is G-invariant, but Q is not proportional to q^2 for any quadratic q. The missing hypothesis is homogeneity of Q. In the application Q(η)=∫_M η^{2n} is homogeneous of degree 2n, so the proposition could be repaired. But the proof in 4.2 is also incomplete: the n-th root is not shown to exist globally, the claim that it is a second-degree polynomial on each plane does not imply a single polynomial on all of V, and the analytic continuation step is asserted. This is not a minor typo; Theorem 4.3 and Corollary 5.10 inherit it.\n\nAlso minor: the notation in Lemma 3.10 is ambiguous, and the diagram in 2.4.2 is hard to parse. Those are not serious.\n\nThe citation pattern is fine; [KV] is a published independent result, so relying on it is not a problem.\n\nWho should read this? Anyone working on non-Kähler symplectic geometry or deformation theory of non-Kähler compact complex manifolds. It deserves a serious referee: the deformation theorem is valuable on its own, and the BBF part is a plausible theorem with a fixable gap. I would ask the authors to correct Proposition 4.4 and expand the proof, then send it out for review.\n\nBest,\n[You]","headline":"A real deformation-theoretic advance for non-Kähler holomorphically symplectic manifolds, with the advertised BBF form resting on a false algebraic proposition as stated; the gap is fixable but load-bearing.","tokens_in":23797,"tokens_out":3691,"would_cite":false,"duration_ms":36646,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["32G05","14J42","53C26","14C05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that Bogomolov-Guan manifolds, despite being non-Kähler, behave like hyperkähler manifolds: their holomorphically symplectic deformations are unobstructed, their period map is locally a diffeomorphism, and their second…","keywords":["holomorphically symplectic manifolds","non-Kähler geometry","Bogomolov-Guan manifolds","Beauville-Bogomolov-Fujiki form","Fujiki formula","local Torelli theorem","deformation theory","Kodaira-Thurston surface"],"falsifier":"Compute ∫_Q $η^{{2n}}$ for a spanning set of classes η in $H^{2}$ of a Bogomolov-Guan manifold: if no symmetric bilinear form q satisfies ∫_Q $η^{{2n}}$=λq(η,η)^n for all η, or equivalently the polarized multi-linear Fujiki identity fails, then Theorem 4.3 is false. A simpler algebraic check would be to exhibit a polynomial on $R^{4}$ invariant under rotations on an open set of 2-planes that is not proportional to a power of a quadratic form.","tokens_in":22753,"feed_emoji":"📐","tokens_out":9049,"duration_ms":83438,"temperature":0.7,"pith_summary":"The paper sets out to show that the main structural tools of hyperkähler geometry survive in a non-Kähler setting. Its target is the class of Bogomolov-Guan manifolds: compact, simply connected, holomorphically symplectic manifolds built from a Kodaira-Thurston surface via the Hilbert scheme and a cyclic covering. The paper proves that their holomorphically symplectic deformations are unobstructed, that the period map is locally a diffeomorphism, and that all sufficiently small complex deformations remain holomorphically symplectic. From this deformation theory it derives the Fujiki formula: a symmetric bilinear form q on $H^{2}$(M) such that ∫_M $η^{{2n}}$=λq(η,η)^n, the non-Kähler analogue of the Beauville-Bogomolov-Fujiki form. This extends the deformation-theoretic backbone of hyperkähler geometry to a family of manifolds previously thought to lie outside its reach.","feed_headline":"A quadratic form controls volumes on non-Kähler symplectic manifolds","feed_subtitle":"Deformations of Bogomolov-Guan manifolds are unobstructed, and their period map is a local diffeomorphism.","key_machinery":"The argument runs on two mechanisms. First, a holomorphically symplectic version of the Tian-Todorov lemma: using the symplectic form Ω to identify vector-valued (0,1)-forms with (1,1)-forms, the Schouten bracket becomes [a,b]=δ(a∧b)-(δa)∧b-(-1)^{|a|}a∧δb with δ=[Λ_Ω,∂], so each Maurer-Cartan obstruction is expressed as a ∂-exact term. Under the cohomological hypotheses of Theorem 3.9 those obstructions vanish, giving unobstructedness of holomorphically symplectic deformations. Second, an algebraic proposition (Proposition 4.4): a polynomial Q invariant under a Lie group that rotates an open family of 2-planes in $H^{2}$(M,R) and fixes their complements must be proportional to q(η,η)^n for a quadratic form q. The paper obtains this open family of 2-planes from Corollary 3.12, which shows that the period map's image is open in the Grassmannian Gr(2,$H^{2}$(M,R)).","core_discovery":"The central claim is Theorem 4.3: for a compact holomorphically symplectic manifold M of complex dimension 2n with Hodge decomposition on $H^{2}$(M), no ∂-exact holomorphic 3-forms, and $H^{{0,2}}$(M)=$H^{{2,0}}$(M)=C, there exists a symmetric bilinear form q on $H^{2}$(M) and a fixed constant λ such that ∫_M $η^{{2n}}$=λq(η,η)^n for every η∈$H^{2}$(M). Applied to a Bogomolov-Guan manifold Q, this gives a Beauville-Bogomolov-Fujiki form with λ=1, a smooth Kuranishi deformation space, and a period map that is a local diffeomorphism; all sufficiently small complex deformations of Q are again holomorphically symplectic. The paper thus claims that these non-Kähler manifolds reproduce two signature features of hyperkähler geometry: unobstructed deformations governed by a period map, and a quadratic form controlling top cup products of two-dimensional cohomology classes.","pith_inferences":["The same cohomological hypotheses single out a general class: any compact holomorphically symplectic manifold satisfying them would inherit a Beauville-Bogomolov-Fujiki form, so the result is not tied to the details of the Bogomolov-Guan construction.","If the conjectured non-degeneracy of q holds, the local period map would give the moduli space a natural analytic structure with a period domain of dimension b_2-2, close to the hyperkähler Teichmüller picture.","A direct algebraic test of Proposition 4.4 could settle the paper's most delicate step independently of geometry: finding any G-invariant polynomial that is an nth power of a quadratic form on every plane in an open set but not globally would show the theorem needs repair.","Guan's original nilmanifold-based examples could be checked against the same Fujiki formula; the paper treats Bogomolov's Hilbert-scheme construction, so the scope across all known non-Kähler holomorphically symplectic examples is not fully pinned down."],"forward_implications":["Bogomolov-Guan manifolds have unobstructed holomorphically symplectic deformations, with the period map locally a diffeomorphism.","Every sufficiently small complex deformation of a Bogomolov-Guan manifold remains holomorphically symplectic.","The Fujiki formula ∫_Q η^{2n}=q(η,η)^n holds, so one quadratic form controls all top self-intersections of degree-2 classes.","Polarizing the Fujiki formula expresses every integral of a product of 2n classes from H^2 in terms of q, exactly as in the hyperkähler case.","The image of the period map is open in the Grassmannian of 2-planes in H^2(Q,R), making the local period domain explicit."],"supporting_citations":[{"why":"Supplies the holomorphically symplectic deformation-theoretic setup, including the Hamiltonian vector-field sheaf and the formal period-map isomorphism, that Theorem 3.9 generalizes to compact non-Kähler manifolds.","marker":"[KV]"},{"why":"Gives Bogomolov's construction of the smooth n^2 covering Q of the characteristic-foliation leaf space, which is the Bogomolov-Guan manifold studied in the paper.","marker":"[Bo3]"},{"why":"Provides Guan's non-Kähler holomorphically symplectic examples and the Hodge-decomposition fact for H^2 used in Proposition 5.9 for Bogomolov-Guan manifolds.","marker":"[Gu3]"},{"why":"Shows that the Hilbert scheme of a holomorphically symplectic surface is holomorphically symplectic, the starting point of the Bogomolov-Guan construction.","marker":"[Bea]"},{"why":"Establishes the Fujiki formula in the hyperkähler case, the model statement that Theorem 4.3 extends to non-Kähler holomorphically symplectic manifolds.","marker":"[F]"},{"why":"Supplies the original Tian-Todorov lemma, which the paper adapts to the holomorphically symplectic setting to prove unobstructedness.","marker":"[Ti]"},{"why":"Provides the Tian-Todorov unobstructedness argument for Calabi-Yau manifolds, serving as the template for the deformation-theory proof in Section 3.","marker":"[To2]"},{"why":"Supplies the Hodge-decomposition theorem for compact complex surfaces, extended to Hilbert schemes in Proposition 5.2 and used for Bogomolov-Guan manifolds.","marker":"[BHPV]"}],"fun_headline_variants":["Non-Kähler symplectic manifolds get a BBF form","Deformations of non-Kähler symplectic manifolds are unobstructed","Local Torelli and Fujiki formula for non-Kähler symplectic manifolds","Bogomolov-Guan manifolds share hyperkähler deformation theory"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the algebraic lemma that a polynomial invariant under rotations on an open family of 2-planes must be the nth power of a quadratic form; the proof assumes the nth root is a well-defined polynomial on the whole space, and that step is only sketched.","fun_headline_variants_meta":{"raw":{"variants":["Non-Kähler symplectic manifolds get a BBF form","Deformations of non-Kähler symplectic manifolds are unobstructed","Local Torelli and Fujiki formula for non-Kähler symplectic manifolds","Bogomolov-Guan manifolds share hyperkähler deformation theory"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000983,"raw_usage":{"total_tokens":4186,"prompt_tokens":974,"completion_tokens":3212,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":590,"completion_tokens_details":{"reasoning_tokens":3125}},"tokens_in":590,"tokens_out":3212,"duration_ms":22711,"temperature":1.0,"reasoning_tokens":3125,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:20:13.772028+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute ∫_Q $η^{{2n}}$ for a spanning set of classes η in $H^{2}$ of a Bogomolov-Guan manifold: if no symmetric bilinear form q satisfies ∫_Q $η^{{2n}}$=λq(η,η)^n for all η, or equivalently the polarized multi-linear Fujiki identity fails, then Theorem 4.3 is false. A simpler algebraic check would be to exhibit a polynomial on $R^{4}$ invariant under rotations on an open set of 2-planes that is not proportional to a power of a quadratic form.","supporting_citations":[],"review_version":1}