{"id":"5417da9f-9c94-4dc3-8885-fa2792b7dba9","arxiv_id":"2505.08113","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"On completely solvable almost abelian solvmanifolds, a symplectic form satisfies the hard-Lefschetz condition if and only if the defining action is semisimple.","lead":"The authors prove that for a family of geometric spaces built from solvable Lie groups, when the defining action is not a simple diagonal one, every compatible symplectic structure fails the hard-Lefschetz property, a cohomological rigidity shared by Kähler manifolds. This completes a dichotomy for the k=1 completely solvable case and comes with explicit examples of such spaces.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.21's normal-form proof asserts d(ex_i)=d(x_i), which fails already for m=2 with b1≠0; the splitting used in Theorem 4.25 is therefore unsupported.","rationale":"I read the paper in good faith. The main theorem is plausible, the circuit analysis is inventive, and the lattice constructions in Section 5 are interesting. The reader's weakest assumption was the external dependency on Proposition 3.1 from the preprint [3]; I agree that dependency is real, but I found a more concrete internal problem: the proof of Theorem 4.21 contains a false differential computation. For a simple Jordan block pair with m=2 and a generic symplectic form, the new basis vectors ex_i defined in the proof do not satisfy d(ex_i)=d(x_i), and the linear map φ is not a Lie algebra homomorphism. Since Theorem 4.25 invokes Theorem 4.21 to split an arbitrary symplectic form into a maximal circuit on a subalgebra h1 and a closed nondegenerate form on the complement, the absence of a bracket-preserving change of basis leaves the hard-Lefschetz failure at degree 2 unproved for the 0∉spec(A0) nonsemisimple case. This is the most load-bearing weakness because it is internal and demonstrable, not merely a citation gap. The theorem may still be true—an automorphism-group argument might repair the normal form—but the text as written does not establish it. I therefore recommend CONDITIONAL acceptance, requiring a corrected proof of Theorem 4.21 or a revised argument for Theorem 4.25. If the normal form cannot be salvaged, the main theorem would need substantial reworking, but I do not see evidence that the statement itself is false.","tokens_in":30457,"tokens_out":26094,"duration_ms":237671,"concrete_test":"Set m=2, λ=1, and ω = f1∧f2 + g1 + g2 (so b1=b2=1) on g_A with A = diag(J_2(1), J_2(−1)). Compute d(ex3), d(ex4) from the definition in Theorem 4.21 and compare with d(x3), d(x4); the equality fails. Then check whether some automorphism of g_A (i.e., an element of the centralizer of A0 acting on the dual basis) can conjugate ω to f1∧f2 + c(g2) for c≠0. If no automorphism can eliminate the g1 coefficient, the normal form statement itself is false; if one can, the theorem may be recoverable but needs a corrected proof that exhibits the automorphism explicitly.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central proof in the case 0∉spec(A0) with nonsemisimple A0 rests on Theorem 4.21, which claims that any symplectic form can be written, after a change of basis, as a sum of maximal unmixed circuits plus an exact term. The proof of the p=1 case defines new dual vectors by ex_{2m+1-i} = (sum over j) b_{j+i-1} x_{m+j} and states 'It is straightforward to check that d(ex_i)=d(x_i) for all 1≤i≤2m', concluding that the linear map φ is a Lie algebra homomorphism. This assertion is false in general. Take m=2, λ≠0, and ω = f1∧f2 + b1 g1 + b2 g2 with b2≠0; b1 is a free parameter because ω is closed and nondegenerate for any b1. The construction gives ex3 = b2 x3 and ex4 = b1 x3 + b2 x4. Using the bracket relations (10), one computes d x3 = λ f1∧x3 − f1∧x4, d x4 = λ f1∧x4, while d(ex3) = b2(λ f1∧x3 − f1∧x4) and d(ex4) = b1 λ f1∧x3 + (b2 λ − b1) f1∧x4. These equal d(x3), d(x4) only when b1=0 and b2=1. For a generic symplectic form b1 is nonzero, so φ does not commute with the exterior derivative and is not a Lie algebra homomorphism. Consequently, the bracket-preserving splitting g_A = h1 ⋉ h2 used in Theorem 4.25 is not justified by the cited computation. The propagation argument via Lemma 3.5 requires an honest subalgebra h1, and the paper does not provide a corrected automorphism argument. This is an internal gap in the proof of the main theorem, independent of the external dependency on Proposition 3.1.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies symplectic forms on almost abelian completely solvable Lie algebras g_A = R ⋉_A R^m and the associated solvmanifolds. The main theorem (Theorem 4.27) asserts that, for unimodular completely solvable almost abelian Lie algebras, every symplectic form satisfies the hard-Lefschetz condition if and only if A is semisimple. The semisimple direction is taken from Kasuya, and the nonsemisimple direction is proved by a detailed analysis of H^1, H^2, H^{2n-2}, and H^{2n-1}, with the failure located at degree 1 when 0 lies in the spectrum of A_0 and at degree 2 otherwise. The paper also constructs lattices for several of the Lie groups in question, yielding explicit symplectic solvmanifolds without the hard-Lefschetz property for any symplectic form.","tokens_in":30897,"tokens_out":20192,"duration_ms":194465,"significance":"If the main theorem is correct, it provides a complete answer for the class of almost abelian completely solvable Lie groups with a one-dimensional semisimple factor, sharpening the known results of Benson-Gordon and Kasuya and exhibiting a clean dichotomy: either all symplectic forms are hard-Lefschetz or none are. The cohomological analysis via 'circuits' is concrete and potentially reusable, and the lattice constructions give nontrivial examples. The paper is also careful to state the precise degree at which the Lefschetz map fails. However, as detailed below, the current proof contains a false key normal-form statement and a missing case, and it depends on an unpublished external proposition, so the significance is conditional on a substantial repair.","major_comments":[{"comment":"The proof's central assertion that d(ex_i)=d(x_i) is false in general, and the theorem as stated is false under the reading required by Definition 4.9 and by the later use in Theorem 4.25. Take m=2, lambda != 0, and basis with brackets [f1,x1]=lambda x1+x2, [f1,x2]=lambda x2, [f1,x3]=-lambda x3+x4, [f1,x4]=-lambda x4. The form omega=delta+b1 x1^ x3+b2(x1^ x4-x2^ x3) is symplectic for every b1 provided b2 != 0. The proof defines ex3=b2 x3 and ex4=b1 x3+b2 x4. With the convention of Section 3, d x3=lambda f1^ x3-f1^ x4 and d x4=lambda f1^ x4, whereas d(ex3)=b2 lambda f1^ x3-b2 f1^ x4 and d(ex4)=b1 lambda f1^ x3+(b2 lambda-b1) f1^ x4. These coincide with d x3 and d x4 only when b2=1 and b1=0, so the map phi is not a Lie algebra homomorphism. Moreover, by Theorem 4.14 the class [x1^ x3]=[g1] is nonzero in H^2(g_A), so for b1 != 0 the class of omega has a nonzero multiple of [g1], while the right-hand side of (17) has none; an exact term cannot remove this component. Since Theorem 4.25 uses (17)-(18) to obtain the splitting to which Lemma 3.5 is applied, this is a load-bearing gap in the proof of the main theorem. The authors should either correct Theorem 4.21 or replace the argument in Theorem 4.25 by a direct computation that does not rely on this normal form.","section":"Section 4.4, Theorem 4.21"},{"comment":"Theorem 4.24 allows M to be nonzero with v=0, but its proof invokes Proposition 4.19(i), which is stated only under the hypothesis v != 0. When v=0 and Z (hence M) is nonzero, Proposition 4.19(i) does not provide the orthogonal decomposition g_A = g_M ⋉ W used in the proof. In that case closed, non-exact 2-forms of the type f1^nu with nu in W^* can occur (see Remark 4.7), so the claimed splitting and the propagation step require a separate argument. Please either supply the missing v=0 case or restrict the statement of Theorem 4.24 and adjust the main theorem accordingly.","section":"Section 4.5, Theorem 4.24"},{"comment":"The proofs of the main results rely on Proposition 3.1, which is cited from the unpublished preprint [3], and the companion reference [4] is listed as 'In preparation'. This proposition is used to reduce every symplectic form to the form omega = f1^ f2 + omega_0 with A_0 in sp(omega_0) and to obtain the normal form in equation (5). Since neither proposition is proved in the manuscript, the main theorem is not self-contained. Please include a proof of Proposition 3.1 or replace the citation by a published, verifiable source; at minimum, state precisely which statements from [3] and [4] are needed and why they are available to the reader.","section":"Section 3, Proposition 3.1"}],"minor_comments":[{"comment":"The notation x_i is used both for Lie algebra basis vectors and for the dual 1-forms, which makes the computation d(ex_i)=d(x_i) very hard to follow. Please use separate notation for the dual basis, or state explicitly that all computations are performed in the dual space V^* g_A.","section":"Section 4.1 and proof of Theorem 4.21"},{"comment":"Theorem 2.13 is stated for nilmanifolds, but in Theorem 4.24 it is applied to the nilpotent Lie algebra g_M. Please state the Lie-algebra version used here and give a reference or a short justification, since the underlying manifold gamma\\G_M may not exist for arbitrary M.","section":"Section 2.2, Theorem 2.13"},{"comment":"Reference [4] is listed as 'In preparation'. Please update the publication status of both [3] and [4] before final submission, or indicate that the relevant statements have been incorporated into the present paper.","section":"References [3] and [4]"},{"comment":"In Case (iii) the sentence stating that the symplectic solvmanifold fails the hard-Lefschetz property 'as a consequence of either Theorem 4.24 or Theorem 4.25' should specify which theorem applies in the mixed case and, if both t != 0 and some m_i >= 2 occur, what the degree of failure is.","section":"Section 5, Case (iii)"}],"recommendation":"major_revision","confidential_remarks":"The central dichotomy is plausible and valuable, and the cohomological computations are detailed. However, the manuscript currently contains a false key theorem (Theorem 4.21) whose proof is used in a load-bearing way, a missing case in Theorem 4.24, and a dependence on unpublished work. I therefore recommend a major revision rather than rejection, but only if the authors can replace the flawed normal-form argument with a correct proof of the needed splitting (or with a direct computation for Theorems 4.24-4.25). If these points cannot be repaired, the paper should not be published. It may also be worth asking the authors to make the manuscript self-contained with respect to Proposition 3.1."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me cut to the chase. The main theorem is the kind of result people will want to cite: for G = R ⋉_φ R^m completely solvable and unimodular, symplectic solvmanifolds are either hard-Lefschetz for every symplectic form (semisimple φ, by Kasuya) or fail for every symplectic form. The paper also pins down where failure happens: degree 1 when 0 ∈ spec(A0) and the nilpotent part is nonzero, degree 2 otherwise. That dichotomy, and the circuit technique for H²(g_A), are genuinely new, and the cohomological computations in Section 4.2 are mostly careful. The lattice examples in Section 5 are a real bonus.\n\nThe proof has a soft spot that is not just a missing citation. Theorem 4.21 claims that after a change of basis any symplectic form is δ plus one maximal circuit plus an exact term, and the proof asserts d(ex_i) = d(x_i) to conclude the change is a Lie algebra automorphism. This is false. Take m = 2 and ω = δ + b1 g1 + b2 g2 with b2 ≠ 0. The construction gives ex3 = b2 x3 and ex4 = b1 x3 + b2 x4. With the Jordan brackets [f1, x3] = -λ x3 + x4 and [f1, x4] = -λ x4, one computes d(ex4) = b1 λ f1∧x3 + b2 λ f1∧x4, which equals d(x4) = λ f1∧x4 only when b1 = 0 and b2 = 1. Generic symplectic forms have b1 ≠ 0, so φ is not a Lie algebra homomorphism. The splitting in Theorem 4.25 relies on this to apply Lemma 3.5 with an honest subalgebra h1; that step is unsupported. For p > 1, the new basis mixes different double blocks, so the alleged ω1 on one block is not even supported there. This is an internal gap, independent of the paper's reliance on the unpublished preprint [3] for Proposition 3.1. That reliance is a second concern: a load-bearing structural fact is cited rather than proved. The use of Theorem 2.13, a nilmanifold theorem, at the Lie algebra level in Theorem 4.24 is also not justified in the text.\n\nNone of this makes me think the main theorem is false. The degree-1 case looks solid, and the degree-2 failure likely survives via a direct computation with arbitrary coefficients b_l instead of the flawed basis change. But the paper as written is not fully verified. A serious referee should take it, because the result is significant and the fix may be short. I would not cite the main theorem yet, though the H² computations could be cited with care. The reading group would enjoy testing the normal-form claim on the m = 2 example.","headline":"A significant converse theorem for almost abelian symplectic solvmanifolds, but the proof of the key normal-form theorem contains a concrete false computation; the result is plausible and deserves refereeing, not acceptance as-is.","tokens_in":31452,"tokens_out":13443,"would_cite":false,"duration_ms":117137,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53D05","22E25","22E40"],"pacs":[],"model":"deepseek-v4-flash","headline":"For completely solvable solvmanifolds G=R ⋉_φ R^m, every symplectic form satisfies the hard-Lefschetz condition exactly when φ is semisimple; otherwise every symplectic form fails it, at degree 1 or degree 2.","keywords":["hard-Lefschetz condition","solvmanifold","almost abelian Lie group","symplectic form","semisimple action","Jordan canonical form","cohomology","lattice"],"falsifier":"Take a unimodular completely solvable almost abelian Lie algebra g_A whose A_0 has a Jordan block J_m(λ) with m≥2, write ω=$f^{1}$∧$f^{2}$+g_m(x,x), and compute the class [$ω^{{m-1}}$∧(x_1∧x_{m+1})] in $H^{{2m}}$(g_A); the paper proves it is zero while [x_1∧x_{m+1}] is not, so finding a case where this wedge is non-exact would refute Theorem 4.25. Alternatively, exhibit one non-semisimple unimodular completely solvable A and one symplectic form whose Lefschetz maps at all degrees are isomorphisms; the theorem predicts none exists.","tokens_in":30264,"feed_emoji":"📐","tokens_out":6421,"duration_ms":54881,"temperature":0.7,"pith_summary":"This paper establishes a dichotomy for symplectic solvmanifolds coming from almost abelian Lie groups G=R ⋉_φ R^m that are completely solvable and unimodular: if the defining action φ is semisimple, every symplectic form satisfies the hard-Lefschetz condition, and if φ is not semisimple, no symplectic form does. The authors prove the converse direction that was previously missing, completing the k=1 case of a question raised by Kasuya's theorem. They also show where the failure occurs: at degree 1 when the nilpotent part of the structure matrix acts on the zero eigenspace, and at degree 2, never at degree 1, when the nonzero spectrum has nontrivial Jordan blocks. The proof reduces the cohomological question to linear algebra: any symplectic form can be written, up to exact terms, as a sum of \"circuits\" built from double generalized eigenspaces, and a top-degree wedge product in one circuit is visibly exact.","feed_headline":"Non-semisimple actions kill hard Lefschetz on solvmanifolds","feed_subtitle":"For G=R⋉R^m, checking whether one matrix is diagonalizable decides the hard-Lefschetz property for every symplectic form.","key_machinery":"The load-bearing object is the \"circuit\": for two double elementary canonical subspaces X_+⊕X_- and Y_+⊕Y_- of the structure matrix A_0, the circuits are alternating sums g_l(x,y)=Σ_{i=1}^l (-1)^{i+1} x_i∧y_{m+l+1-i}. These are closed, non-exact 2-forms whose cohomology classes are linearly independent and span the relevant part of $H^{2}$. Proposition 3.1 supplies the starting point: every symplectic form on g_A can be normalized to ω=$f^{1}$∧$f^{2}$+ω_0 with A_0∈sp(ω_0). A change of basis then turns any symplectic form into $f^{1}$∧$f^{2}$ plus top-length unmixed circuits plus an exact term. The crucial identity is δ∧Γ_{m,2m}=-d($f^{2}$∧Γ_{m,2m-1}), which exhibits the exact kernel element responsible for the degree-2 failure, while the zero-eigenvalue case uses the nilpotent subalgebra g_M and a propagation lemma that transfers failures from a subalgebra to the whole Lie algebra.","core_discovery":"The central claim is Theorem 4.27: on a unimodular completely solvable almost abelian Lie algebra g_A with structure matrix A, a symplectic form satisfies the hard-Lefschetz condition if and only if A is semisimple. Since the de Rham cohomology of the solvmanifold is computed by left-invariant forms, the same statement holds for every symplectic form on every such solvmanifold Γ\\G. The paper's contribution is the \"only if\" direction, proved by describing which Jordan blocks force a Lefschetz map to have kernel. If A has a nonzero nilpotent part on the zero generalized eigenspace, the subalgebra g_M is non-abelian nilpotent with an ω-orthogonal complement, so the nilmanifold theorem of Benson and Gordon propagates a degree-1 failure. If the nonzero part of A has a Jordan block of size at least 2, the symplectic form contains an unmixed circuit of length m≥2, and the wedge $ω^{{m-1}}$∧(x_1∧x_{m+1}) is exact even though [x_1∧x_{m+1}] is not, giving a degree-2 failure.","pith_inferences":["The same circuit machinery may answer the authors' question for k>1: if the proof generalizes, non-semisimplicity of φ: R^k→Aut(R^m) would force failure of hard Lefschetz for every symplectic form on such solvmanifolds.","The dichotomy suggests that within almost abelian solvmanifolds the hard-Lefschetz condition is a purely representation-theoretic property of the action, not a property of the chosen symplectic structure.","A testable extension is to check whether the exactness identity δ∧Γ_{m,2m}=-d(f^2∧Γ_{m,2m-1}) persists when A has complex eigenvalues; if it does, the non-completely-solvable case would also fail hard Lefschetz at degree 2."],"forward_implications":["For every completely solvable almost abelian solvmanifold of even dimension, the hard-Lefschetz property is independent of the symplectic form: either all symplectic forms have it or none do.","A symplectic form on such a solvmanifold is hard-Lefschetz exactly when the defining action is semisimple, so a single matrix computation decides the entire cohomological property.","The failure of hard Lefschetz is concentrated at degree 1 or degree 2: in the zero-eigenvalue nilpotent case at degree 1, and in the nonzero-Jordan-block case at degree 2 with degree 1 always an isomorphism.","Many concrete examples now exist: the lattice constructions in Section 5 produce symplectic solvmanifolds with no hard-Lefschetz symplectic form, using blocks J_{2t}(0) or J_m(t_k)⊕J_m(-t_k).","Combined with Kasuya's theorem, the result gives a complete characterization for k=1, so any counterexample to the broader question would have to lie outside the completely solvable almost abelian family or have k>1."],"supporting_citations":[{"why":"Supplies Proposition 3.1, the structural fact that every symplectic form on an almost abelian Lie algebra can be reduced to ω=f^1∧f^2+ω_0 with A_0∈sp(ω_0); the whole proof rests on this normalization.","marker":"[3]"},{"why":"Kasuya's theorem gives the semisimple-implies-hard-Lefschetz direction that the paper completes by proving the converse.","marker":"[17]"},{"why":"Benson-Gordon's nilmanifold theorem provides the degree-1 failure used in the zero-eigenvalue case through the nilpotent subalgebra g_M.","marker":"[6]"},{"why":"Hattori's isomorphism transfers cohomological conclusions from the Lie algebra g_A to the solvmanifold Γ\\G.","marker":"[16]"},{"why":"Bock's lattice criterion is the tool used in Section 5 to exhibit solvmanifolds with the predicted failure.","marker":"[8]"},{"why":"Supplies the Jordan canonical form conditions for matrices in sp(m,R), which determine which spectra and Jordan blocks can occur in the completely solvable case.","marker":"[4]"}],"fun_headline_variants":["Semisimple action iff hard Lefschetz on solvmanifolds","Non-semisimple action: hard Lefschetz fails for all symplectic forms","Check a matrix: hard Lefschetz on solvmanifolds iff semisimple","Hard Lefschetz fails at degree 1 or 2 when action isn't semisimple"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on the cited Proposition 3.1, which says that every symplectic form on an almost abelian Lie algebra can be normalized to ω=$f^{1}$∧$f^{2}$+ω_0 with A_0∈sp(ω_0); if that structural classification were false or incomplete, the circuit analysis and both main theorems would need reworking.","fun_headline_variants_meta":{"raw":{"variants":["Semisimple action iff hard Lefschetz on solvmanifolds","Non-semisimple action: hard Lefschetz fails for all symplectic forms","Check a matrix: hard Lefschetz on solvmanifolds iff semisimple","Hard Lefschetz fails at degree 1 or 2 when action isn't semisimple"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000934,"raw_usage":{"total_tokens":4093,"prompt_tokens":1135,"completion_tokens":2958,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":751,"completion_tokens_details":{"reasoning_tokens":2866}},"tokens_in":751,"tokens_out":2958,"duration_ms":20854,"temperature":1.0,"reasoning_tokens":2866,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T22:03:26.983257+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a unimodular completely solvable almost abelian Lie algebra g_A whose A_0 has a Jordan block J_m(λ) with m≥2, write ω=$f^{1}$∧$f^{2}$+g_m(x,x), and compute the class [$ω^{{m-1}}$∧(x_1∧x_{m+1})] in $H^{{2m}}$(g_A); the paper proves it is zero while [x_1∧x_{m+1}] is not, so finding a case where this wedge is non-exact would refute Theorem 4.25. Alternatively, exhibit one non-semisimple unimodular completely solvable A and one symplectic form whose Lefschetz maps at all degrees are isomorphisms; the theorem predicts none exists.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Kasuya's theorem gives the semisimple-implies-hard-Lefschetz direction that the paper completes by proving the converse."},{"cited_title":"Benson and C","cited_arxiv_id":null,"evidence_quote":"Benson-Gordon's nilmanifold theorem provides the degree-1 failure used in the zero-eigenvalue case through the nilpotent subalgebra g_M."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Hattori's isomorphism transfers cohomological conclusions from the Lie algebra g_A to the solvmanifold Γ\\G."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Bock's lattice criterion is the tool used in Section 5 to exhibit solvmanifolds with the predicted failure."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Jordan canonical form conditions for matrices in sp(m,R), which determine which spectra and Jordan blocks can occur in the completely solvable case."}],"review_version":1}