{"id":"5e24cb62-362d-4a2d-bcbd-8b492636a28b","arxiv_id":"2608.08665","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Compact quotients of complex homogeneous spaces with compact isotropy are Kähler whenever they admit both a balanced metric and a pluriclosed metric.","lead":"This paper proves that a large family of compact complex manifolds, the quotients of complex homogeneous spaces with compact isotropy, must have a Kähler metric whenever they admit two weaker metric types: balanced and pluriclosed. It settles the Fino-Vezzoni conjecture in this setting and also shows the pluriclosed flow converges to a flat Kähler metric when the first Chern class vanishes.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem H's reduction to an ODE on the G-invariant Aeppli slice requires the unstated naturality of the pluriclosed flow under biholomorphisms; without it the actual PDE flow need not stay G-invariant.","rationale":"The paper's main structural argument is sound: Theorem F derives the existence of a unique balanced (hence Kähler) metric in each G-invariant Aeppli slice from the concavity of log det and the boundedness of the slice, and the symmetrisation operator is carefully constructed. The most load-bearing point in Theorem H is the reduction of the genuine pluriclosed flow to a finite-dimensional ODE. The paper proves the tangent vector field lies in A at every point of S, but this only defines an ODE on S. To identify the PDE solution with the ODE solution, one must know that the PDE solution starting from a G-invariant metric remains G-invariant; that follows from naturality of the flow under biholomorphisms and uniqueness, neither of which is stated. This is not a deep mathematical flaw, as the naturality is a standard property, but it is a genuine gap in the written proof. The reader's weakest assumption identified exactly this issue, so I agree with the reader's assessment. The verdict remains CONDITIONAL: the paper should add a short justification of the naturality (or at least a reference for it) and fix the typo in the definition of A. No other part of the argument appears to require modification.","tokens_in":10380,"tokens_out":29512,"duration_ms":311058,"concrete_test":"Verify the naturality identity φ^*(ρ_B^ω)^{1,1} = (ρ_B^{φ^*ω})^{1,1} for a biholomorphism φ, using the local expression of the Bismut-Ricci form in terms of the Chern connection and its torsion. If the identity holds, insert a one-sentence proof in Section 3; if it fails, Theorem H must be restated for the invariant flow only.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 3, the proof of Theorem H reduces the pluriclosed flow to an ODE on the G-invariant Aeppli slice S by asserting that 'the pluriclosed flow restricts to an ODE on S.' This requires that a solution starting at a G-invariant metric stays G-invariant for all time. The tangent-space computation (ρ_B^ω)^{1,1} ∈ A for ω ∈ S only shows that the ODE vector field is tangent to S; it does not show that the actual PDE solution, which is the object whose existence is claimed in Theorem H, remains in S. The missing step is naturality of the flow under the G-action: for any biholomorphism φ of X, φ^*ω(t) should solve the flow whenever ω(t) does. Then, since G acts by biholomorphisms, the pullback g^*ω(t) is also a solution with the same initial data as ω(t), and uniqueness forces g^*ω(t) = ω(t). This naturality is standard—it follows from a local computation using the Chern connection—but it is neither stated nor proved in the paper. Without it, the ODE describes only the 'invariant' flow, and Theorem H's assertion about the actual pluriclosed flow from an invariant initial metric is not justified.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves Theorem C: under Assumptions B, if the compact quotient Y=Γ\\X of a complex homogeneous space X with compact isotropy admits both a balanced metric and a pluriclosed metric, then Y is Kähler. The engine is Theorem F, which shows that in the finite-dimensional space H of G-invariant real (1,1)-forms, every G-invariant Aeppli slice contains a unique balanced metric, and this metric is Kähler when the slice is generated by a pluriclosed metric. Theorem H asserts that, under the additional hypothesis c1(Y)=0, the pluriclosed flow from any G-invariant pluriclosed metric is immortal, stays in the G-invariant Aeppli slice, and converges to a flat Kähler metric. The proof strategy is to identify balanced metrics with critical points of the volume functional (equivalently of det on H), prove strict concavity of log det to get uniqueness, and then use L=log det as a Lyapunov function for the induced ODE.","tokens_in":10603,"tokens_out":11278,"duration_ms":119523,"significance":"The result is a substantial advance on the Fino–Vezzoni conjecture: it covers all compact discrete quotients of complex homogeneous spaces with compact isotropy, a much broader class than the solvmanifold and nilmanifold cases treated earlier. Theorem H also extends the recent result of Fino and Vezzoni by removing the vanishing first Bott–Chern class assumption and by covering the full homogeneous setting. The proof is remarkably elementary: after symmetrisation, the main theorems reduce to finite-dimensional convex analysis (strict concavity of log det on a bounded affine slice) plus a Lyapunov argument, and the symmetrisation operator is constructed in detail rather than merely invoked. The proof is self-contained in its key steps and contains no fitted parameters or circular dependencies on the results it claims to prove.","major_comments":[{"comment":"The assertion 'Therefore, the pluriclosed flow restricts to an ODE on S' is not justified as written. The preceding computation shows only that the vector field X_ω=-(ρ_B^ω)^{1,1} is tangent to the affine slice S. To conclude that the actual parabolic flow starting at a G-invariant metric evolves inside S, one must prove that the pluriclosed flow is natural with respect to biholomorphisms: if φ is a biholomorphism of X and ω(t) is a solution, then φ^*ω(t) is also a solution, so by uniqueness of the flow g^*ω(t)=ω(t) for every g∈G. This naturality is standard and follows from the local Chern-connection formula for the Bismut–Ricci form, but it is neither stated nor proved in the paper. Without this step, the ODE describes only the invariant flow, and Theorem H's claims about the genuine PDE flow from a G-invariant initial metric are not established. Please add this lemma explicitly, or give a precise citation, before using the ODE reduction.","section":"Section 3, proof of Theorem H"},{"comment":"The statement that the flow starting at the unique Kähler metric ω* is constant depends on the same missing naturality. The argument 'the pluriclosed flow preserves the Kähler condition and ω* is the unique Kähler metric in S, so the flow starting at ω* is constant' is valid only if one already knows that the flow from ω* remains in S. That is exactly the point that requires biholomorphic naturality; as written, the reasoning is circular. Once the naturality lemma is added, this paragraph becomes correct.","section":"Lemma 3.1, first paragraph"},{"comment":"The identity d/dt log det(ω(t)) = |T_{ω(t)}|^2_{ω(t)} is imported from [Str16, Lemma 6.1] without stating the lemma. The paper notes that the Laplacian term vanishes by homogeneity, but for the reader to verify the sign and the vanishing, the relevant formula from [Str16] should be quoted or at least stated precisely. This is not a block to the argument, but it would improve the exposition.","section":"Lemma 3.1, Lyapunov computation"}],"minor_comments":[{"comment":"The displayed definition of A uses ∂α+∂̄α for a G-invariant (1,0)-form α; this expression is not a real (1,1)-form. It should be ∂̄α+∂\\bar α, equivalently (d(α+\\bar α))^{1,1}, as in the correct definition in Section 2.","section":"Section 1, displayed definition of A"},{"comment":"The line '0=f'(0)=f'(1)' should read 'f'(0)=f'(1)=0' for clarity.","section":"Proof of Lemma 2.4"},{"comment":"The normalization ∫_Y μ=1 is introduced after μ is defined; it would be clearer to state the normalization before writing the integral formula.","section":"Proposition A.1(iii)"},{"comment":"The term 'G-invariant Aeppli slice' is used without a short explanation of its cohomological meaning; a sentence noting that it is the invariant analogue of an Aeppli cohomology class would orient the reader.","section":"General exposition"}],"recommendation":"major_revision","confidential_remarks":"The only substantive issue is the missing naturality lemma in Theorem H: the passage from the PDE flow to the finite-dimensional ODE on the invariant slice needs an explicit statement that the pluriclosed flow commutes with pullback by biholomorphisms. This is a standard and local fact, so I expect a straightforward revision to address it. The rest of the argument is coherent and the results are significant; I see no circularity, fitted parameters, or unsupported central claims outside this gap."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline is simple: this paper proves the Fino–Vezzoni conjecture for compact quotients of complex homogeneous spaces with compact isotropy, and that is a real step change. It unifies a long list of special cases (nilmanifolds, solvmanifolds, semisimple groups with regular J, etc.) under one argument. The symmetrization operator in Appendix A is the key new tool, and the way it turns a balanced metric on Y into a G-invariant balanced metric on X is clean.\n\nThe core of the proof is sound. Proposition 2.1 (balanced = critical point of volume on the Aeppli slice) is correct and follows from a nondegenerate pairing. Lemma 2.4 (strict concavity of log det) is the right engine: bounded slice plus strict concavity gives a unique critical point. The Lyapunov function argument in Lemma 3.1 is also coherent, assuming the reduction to an ODE.\n\nThe stress-test note is right. In Section 3, the paper says 'the pluriclosed flow restricts to an ODE on S' after showing only that the vector field at points of S lies in A. That shows the ODE stays in the slice, but not that the PDE solution starting at a G-invariant metric stays G-invariant. The missing step is naturality under biholomorphisms: pull back a solution by g ∈ G, get another solution with same initial data, and uniqueness forces equality. This is standard but it must be stated. Without it the identification of the actual flow with the ODE is not justified. Easy fix.\n\nTwo smaller points: the displayed definition of A in the introduction has a typo (missing bars on the second ∂). And the paper cites [Str16, Lemma 6.1] for the |T|^2 identity; that is fine, but the flatness of ω* is used essentially and it would help to spell out why the Laplacian term vanishes (they do include one sentence, but the logic could be expanded).\n\nThe citation pattern is careful and contextual. Self-citations appear where they are needed, not as support for the main claim. No fitted parameters, no invented entities.\n\nThis paper deserves a serious referee. With the naturality sentence added, the proof is complete as far as I can see. Send it out.","headline":"Theorem C is a major and convincing advance on the Fino–Vezzoni conjecture; Theorem H has a small unstated naturality gap that should be patched, not a fatal flaw.","tokens_in":11136,"tokens_out":4035,"would_cite":true,"duration_ms":40204,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C55","53C30","53C44"],"pacs":[],"model":"deepseek-v4-flash","headline":"On compact quotients of complex homogeneous spaces with compact isotropy, the Fino–Vezzoni conjecture holds: balanced plus pluriclosed forces Kähler, and with trivial first Chern class the pluriclosed flow converges to a flat Kähler metric.","keywords":["Fino–Vezzoni conjecture","balanced metric","pluriclosed metric","SKT metric","homogeneous complex manifold","pluriclosed flow","Kähler metric","Aeppli cohomology"],"falsifier":"A concrete counterexample would be a compact quotient satisfying Assumptions B that admits both a balanced and a pluriclosed metric but no Kähler metric; such a space would refute Theorem C. To test the flow part, one could take any $G$-invariant pluriclosed metric on a space with $c_1(Y)=0$ and check numerically or analytically whether the pluriclosed flow leaves the $G$-invariant Aeppli slice at positive time, which would show Theorem H's ODE reduction is invalid.","tokens_in":10157,"feed_emoji":"📐","tokens_out":10895,"duration_ms":99442,"temperature":0.7,"pith_summary":"The paper proves the Fino–Vezzoni conjecture for compact quotients of complex homogeneous spaces with compact isotropy: if the quotient admits both a balanced Hermitian metric and a pluriclosed Hermitian metric, then it admits a Kähler metric. This matters because the conjecture asks whether two weaker non-Kähler conditions always combine to force Kählerness, and the paper answers it for every manifold of this homogeneity type. The argument also covers quotients of real Lie groups carrying left-invariant complex structures and compact homogeneous spaces with a transitive compact Lie group. Under vanishing real first Chern class, the paper shows that the pluriclosed flow starting from any invariant pluriclosed metric exists for all time and converges smoothly to a flat Kähler metric.","feed_headline":"Balanced and pluriclosed metrics force Kähler on homogeneous quotients","feed_subtitle":"Shows the Fino–Vezzoni conjecture holds for compact quotients of homogeneous spaces with compact isotropy.","key_machinery":"The load-bearing object is the $G$-invariant Aeppli slice $S=(\\omega_0+\\mathcal{A})\\cap P$, where $\\mathcal{A}$ consists of the $(1,1)$-parts $\\partial\\alpha+\\bar\\partial\\bar\\alpha$ of $G$-invariant $(1,0)$-forms $\\alpha$, and $P$ is the cone of $G$-invariant Hermitian metrics. On this finite-dimensional slice, the volume functional, realized as a determinant on the space of Hermitian operators commuting with the isotropy representation, has a unique critical point because the existence of a $G$-invariant balanced metric forces the slice to be bounded and the logarithm of the determinant to be strictly concave. Proposition 2.1 identifies balanced metrics with exactly these critical points. A symmetrization operator, built by averaging over the compact quotient $\\Gamma\\backslash G$ against a bi-invariant volume form, pushes arbitrary balanced or pluriclosed metrics on $Y$ to $G$-invariant ones while preserving the relevant properties. For the flow, the same log-determinant function serves as a proper strict Lyapunov function for the ODE induced on the slice, and the Chern torsion identity supplies its monotonicity.","core_discovery":"The central claim is Theorem C: under Assumptions B, a compact quotient $Y=\\Gamma\\backslash X$ that admits both a balanced and a pluriclosed metric must also admit a Kähler metric. The proof establishes a stronger slice statement (Theorem F): after symmetrizing a balanced metric to a $G$-invariant one, every $G$-invariant Aeppli slice contains a unique balanced metric, and when the slice is generated by a pluriclosed metric that balanced metric is Kähler. When $c_1(Y)=0$, the pluriclosed flow stays inside the finite-dimensional Aeppli slice, log-volume is a proper strict Lyapunov function for the induced ODE, and the unique Kähler metric in the slice is the global attractor. Because that limit metric is homogeneous and Ricci-flat, it is flat, giving Theorem H.","pith_inferences":["The slice argument suggests that the balanced representative in each $G$-invariant Aeppli class is unique; checking whether this uniqueness extends to full non-invariant Aeppli classes would connect the result to volume invariants on arbitrary compact complex manifolds.","Because the proof only needs a finite-dimensional slice and strict concavity of log-volume, the same strategy may apply to non-homogeneous compact manifolds admitting a finite-dimensional group of biholomorphisms whose slice geometry is bounded.","A natural test is to compute the pluriclosed flow numerically for a low-dimensional solvmanifold satisfying the assumptions; the Lyapunov argument suggests the convergence rate should be exponential, a statement the paper does not make."],"forward_implications":["For $Y=\\Gamma\\backslash G$ with a left-invariant complex structure and $\\Gamma$ a cocompact lattice, balanced plus pluriclosed implies a flat Kähler metric.","For compact homogeneous spaces with a transitive compact Lie group, balanced plus pluriclosed implies Kähler, and the space is a product of a complex torus and a generalized flag manifold.","When $c_1(Y)=0$, every $G$-invariant pluriclosed metric flows immortally to a flat Kähler limit, generalizing earlier results that required vanishing first Bott–Chern class.","The conjecture for this homogeneous class is reduced to finite-dimensional data: one $G$-invariant balanced metric bounds every invariant Aeppli slice, giving uniqueness of balanced representatives in each slice."],"supporting_citations":[{"why":"Introduces the Fino–Vezzoni conjecture that the paper resolves for homogeneous quotients and supplies the problem statement.","marker":"[FV15]"},{"why":"Establishes that a Hermitian metric simultaneously balanced and pluriclosed is Kähler, used to identify the unique slice metric as Kähler.","marker":"[AI01]"},{"why":"Provides the flatness theorem for Kählerian homogeneous spaces of unimodular Lie groups used in the lattice case.","marker":"[Han57]"},{"why":"Shows a positive $(n-1,n-1)$-form has a unique Hermitian root, used to keep balancedness after symmetrization.","marker":"[Mic82]"},{"why":"Supplies the evolution identity for log-determinant under the pluriclosed flow in terms of Chern torsion, the core of the Lyapunov estimate.","marker":"[Str16]"},{"why":"Gives the fact that homogeneous Ricci-flat Kähler metrics are flat, used for the limit metric in Theorem H.","marker":"[Bes87]"},{"why":"Provides the result that a cocompact lattice makes a Lie group unimodular, needed for the bi-invariant volume form in the symmetrization integral.","marker":"[Mil76]"},{"why":"Gives the previous flow theorem under vanishing first Bott–Chern class that Theorem H generalizes and drops that assumption.","marker":"[FV26]"}],"fun_headline_variants":["Balanced + pluriclosed ⇒ Kähler on homogeneous quotients","Fino–Vezzoni proven for compact homogeneous quotients","Pluriclosed flow converges to flat Kähler when c1=0","When balanced and pluriclosed meet, homogeneous quotients are Kähler"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The full flow theorem rests on the assumption that the pluriclosed flow preserves $G$-invariance of the initial metric, so that the infinite-dimensional PDE really reduces to the finite-dimensional ODE; the paper invokes the standard naturality of the flow under biholomorphisms but does not prove it.","fun_headline_variants_meta":{"raw":{"variants":["Balanced + pluriclosed ⇒ Kähler on homogeneous quotients","Fino–Vezzoni proven for compact homogeneous quotients","Pluriclosed flow converges to flat Kähler when c1=0","When balanced and pluriclosed meet, homogeneous quotients are Kähler"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001292,"raw_usage":{"total_tokens":5190,"prompt_tokens":772,"completion_tokens":4418,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":388,"completion_tokens_details":{"reasoning_tokens":4337}},"tokens_in":388,"tokens_out":4418,"duration_ms":35714,"temperature":1.0,"reasoning_tokens":4337,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:30:51.335256+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete counterexample would be a compact quotient satisfying Assumptions B that admits both a balanced and a pluriclosed metric but no Kähler metric; such a space would refute Theorem C. To test the flow part, one could take any $G$-invariant pluriclosed metric on a space with $c_1(Y)=0$ and check numerically or analytically whether the pluriclosed flow leaves the $G$-invariant Aeppli slice at positive time, which would show Theorem H's ODE reduction is invalid.","supporting_citations":[],"review_version":1}