{"id":"1e1bc789-817a-4c10-8fe0-b57f1f488aa6","arxiv_id":"2501.01017","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"For admissible, chi-semi-convex solutions of complex Hessian equations with gradient terms on compact Hermitian manifolds, the paper proves uniform second-order estimates.","lead":"This paper proves a new bound on the second derivatives of solutions to a family of complex Hessian equations on curved spaces called Hermitian manifolds. The bound is the key missing step for solving these equations by a standard continuity method when the solution is only semi-convex.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central estimate depends on Lemma 1.2, whose proof imports Zhang's real-case inequalities without checking the Hermitian/torsion setting and whose application in (4.15) shows sign/normalization mismatches; without a repaired derivation Theorem 1.1 is not established.","rationale":"The reader's verdict CONDITIONAL is appropriate: the paper's main result is plausible and follows a standard continuity-method framework, but the new technical core, Lemma 1.2, is imported from real-case work without full verification in the Hermitian setting. My independent reading confirms this as the weakest point and identifies an additional sign/normalization discrepancy between Lemma 1.2 and its use in (4.15). This does not prove the theorem false, but it means the proof as written is not yet verifiable. The recommendation remains CONDITIONAL, so the verdict is unchanged.","tokens_in":14727,"tokens_out":18624,"duration_ms":151450,"concrete_test":"Check the implication (1.5) => (4.15): multiply Lemma 1.2 by sigma_k/lambda_1, set sigma_k = psi, and write out the resulting lower bound for -lambda_1^{-1} sum_{p != q} sigma^{pp,qq} D_1 chi_{pp} D_1 chi_{qq}. If the signs of the K|D_1 psi|^2/(lambda_1 sigma_k) and sum_{p>1} sigma^{pp}|D_1 chi_{pp}|^2/lambda_1^2 terms are opposite to those displayed in (4.15), the proof as written does not follow. This single check isolates the load-bearing step.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim depends on Lemma 1.2, used in (4.15) to control bad third-order terms. The proof of Lemma 1.2 is not self-contained: it begins with 'By (3.27) in [43]' and, in the nontrivial case (3.4), says 'refer to the deduction of (3.73) in [43]' without reproducing whether those real-Hessian inequalities survive for Hermitian W with non-symmetric covariant derivatives and torsion. If (1.5) fails, the estimate (4.15) has no basis and Theorem 1.1 does not follow. A second, independent difficulty is visible in the application: (4.15) is presented as a consequence of Lemma 1.2, but after multiplying (1.5) by sigma_k/lambda_1 the terms appear with the opposite signs: Lemma 1.2 gives +K|D_j sigma_k|^2/sigma_k^2 and +(1-epsilon_0) sum_{i>1} sigma^{ii}|D_1 chi_{ii}|^2/(lambda_1 sigma_k), whereas (4.15) displays -K|D_1 psi|^2/(lambda_1 sigma_k) and -(1-epsilon_0) sum_{p>1} sigma^{pp}|D_1 chi_{pp}|^2/lambda_1^2. Unless an unstated rearrangement converts one form into the other, the proof of (4.15) is not justified.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies uniform second-order a priori estimates for admissible, χ-semi-convex solutions of the complex Hessian equation (1.2) on compact Hermitian manifolds, where the right-hand side depends on the gradient and the unknown function. The main result, Theorem 1.1, asserts a bound |DDu| ≤ C under χ′ ≥ εω and χ-semi-convexity, generalizing earlier estimates that required χ ∈ Γ_{k+1}. The proof follows the standard maximum-principle route with an auxiliary function Q = log λ1 + φ(|Du|²) + ϕ(u), a perturbation argument for the largest eigenvalue, and a new 'modified concavity inequality' (Lemma 1.2) that is meant to control the bad third-order terms. The paper is organized into a preliminary section, a section proving Lemma 1.2, and a section proving Theorem 1.1.","tokens_in":15122,"tokens_out":8834,"duration_ms":76118,"significance":"If the proof is correct, the result is a meaningful improvement: it replaces the convexity assumption χ ∈ Γ_{k+1} by the weaker χ-semi-convexity, matching recent developments for real k-Hessian equations, and it does so in the presence of gradient terms and torsion on Hermitian manifolds. The paper is clearly written and follows a natural strategy, with a sensible choice of auxiliary functions and a perturbation argument for the eigenvalue degeneracy. The main weakness is that the central Lemma 1.2 is not self-contained: it imports the key inequalities (3.27) and the deduction of (3.73) from Zhang's real-case paper [43] without reproducing the Hermitian adaptation, and the application of Lemma 1.2 in (4.15) involves a normalization and sign rearrangement that is not explained. These points are load-bearing because Theorem 1.1 rests directly on (1.5). No machine-checked proofs or code are supplied; the contribution is a classical analytic proof whose verification currently depends on unavailable details.","major_comments":[{"comment":"The proof of Lemma 1.2 is not self-contained. The first line invokes (3.27) of Zhang [43], and the nontrivial case (3.4) is dismissed with 'refer to the deduction of (3.73) in [43]' without reproducing the argument. Zhang's inequalities are proved for real k-Hessian equations with symmetric Hessians and real test vectors, whereas the present setting involves a Hermitian tensor W with non-symmetric covariant derivatives and torsion terms coming from (2.4). Since (1.5) is the engine behind (4.15) and hence behind Theorem 1.1, the paper must either prove the Hermitian version of these inequalities or state and prove a precise transfer lemma. As written, the central estimate rests on an unverified import.","section":"3, Lemma 1.2 and Eq. (1.5)"},{"comment":"The displayed calculation preceding (3.5) uses parameters a and M and constants C1, C2, C3, C3' without defining them or explaining their dependencies. The text says 'by assuming λ1 > Mk and M large enough' and later 'by choosing M ≥ 2C3'(c0+1)/(c0ε0)', but the order of choices is not fixed and the parameter a is never assigned a value. Because the final comparison in (3.5) depends on these choices, the proof of (1.5) under the assumption (3.4) cannot be checked as written.","section":"3, proof of Lemma 1.2, display before (3.5)"},{"comment":"The application of Lemma 1.2 in (4.15) does not match the statement of the lemma. Lemma 1.2 contains the term -∑ σ^{pp,qq}ω_{ppj}ω_{qqj}/σ_k, while (4.15) has -λ1^{-1}∑σ^{pp,qq}D1χ_{pp}D1χ_{qq} without the /σ_k factor. The lemma's positive K|Djσk|²/σk² term becomes -K|D1ψ|²/(λ1σk) in (4.15), and the lemma's positive (1-ε0)∑_{i>1}... term appears with a negative sign. These discrepancies can be reconciled by multiplying (1.5) by σ_k, moving terms to the other side, and using D1σk = D1ψ, but this rearrangement is not stated. Moreover, such a step needs the boundedness and positivity of σk = ψ, which is available but not explicitly invoked at that point. As written, (4.15) does not follow by direct substitution.","section":"4, Eq. (4.15)"},{"comment":"Lemma 2.3, in particular inequalities (2.9) and (2.10), is imported from Dong [10] without proof. These formulas encode the commutator and torsion terms that are specific to the Hermitian setting and are used directly in the proof of Theorem 1.1. The paper should either prove them in an appendix or state the precise hypotheses under which they apply to equation (1.2); relying on a computation lemma with no verification of the current hypotheses leaves a gap in the main proof.","section":"2, Lemma 2.3"}],"minor_comments":[{"comment":"There is a typo: 'a natural problem is weather we can weaken' should read 'whether we can weaken'.","section":"1, Introduction, p.2"},{"comment":"The notation σ^{pq}_k is introduced in (2.6) for ∂σ_k/∂χ_{pq}, but the connection to F^{ij} in Lemma 2.1 is not explicit; this creates confusion in formulas such as (2.9) and (4.4) where σ^{pq}_k and F are used interchangeably.","section":"2, Notation"},{"comment":"The chain σ^{ii}_k ≥ σ^{11}_k ≥ kσ_k/(nλ1) used in (4.16) is asserted without proof. It can be justified from Proposition 2.1(8) together with the eigenvalue ordering, but the paper should spell this out.","section":"4, Eq. (4.16)"},{"comment":"Reference [43] is an arXiv preprint (arXiv:2408.10781v1, 2024). Since the proof of Lemma 1.2 depends critically on inequalities (3.27) and (3.73) of that paper, the authors should clarify the version used and, ideally, cite the published version if one exists.","section":"References, [43]"}],"recommendation":"major_revision","confidential_remarks":"The main technical uncertainty is whether Zhang's real-case concavity inequalities survive the Hermitian/torsion setting. If the authors can supply a complete derivation of (1.5), the paper is likely publishable, but as it stands the central lemma is not independently checkable. The sign and normalization mismatch in (4.15) should be resolved explicitly, not only because it blocks verification but because a reader cannot tell whether the intended inequality is actually the rearranged version of (1.5). The reliance on an unpublished arXiv preprint for the key inequalities should also be addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take on arXiv:2501.01017. The genuinely new thing is Lemma 1.2, a complex-Hermitian concavity inequality under chi-semi-convexity, and its use in Theorem 1.1 to weaken Dong's Gamma_{k+1} assumption to chi-semi-convex for the gradient-dependent Hessian equation (1.2). That is a real, if incremental, step in a working subfield; the Corollary with sigma_{k+1}>-A is a natural consequence. The paper is clearly written, and the Q = log lambda_1 + phi(|Du|^2) + phi(u) maximum-principle apparatus is standard and mostly competently handled. The choice epsilon_0 = 3 beta is sensible.\n\nI checked the step (4.15) that a stress-test flagged for sign mismatch, and the signs are actually consistent. You use Lemma 1.2 to bound -1/lambda_1 * sum from below, so the negative K|D_1 psi|^2 and negative (1-epsilon_0) sum terms are expected; epsilon_0 = 3 beta then cancels the leftover sigma^{11} term. No problem there.\n\nThe real soft spot is Lemma 1.2 itself. The proof is not self-contained: it starts with \"By (3.27) in [43]\" and, in the nontrivial case, says \"refer to the deduction of (3.73) in [43]\" without reproducing the argument. The constants a, M, C_1, C_2, C_3 appear in the display before (3.5) without being pinned down; a is a free parameter and the final inequality holds for all a, but the reader cannot track the C's. More substantively, Zhang's inequalities are for real vectors xi_i^2, while Lemma 1.2 needs them for complex omega_{ppj}; because the expression contains products omega_{ppj} omega_{qqj} without conjugates, it is not automatic that the real proof survives. Torsion is a red herring at this stage -- Lemma 1.2 is pure Hermitian matrix algebra -- but the complex conjugacy issue is real and needs to be checked line by line. This is the load-bearing new inequality; if it fails, Theorem 1.1 collapses.\n\nBottom line: the paper deserves a serious referee and is likely fixable by writing out Lemma 1.2 properly. If the lemma is real, the theorem is probably true. I would not cite it as-is, but I would send it out.","headline":"A plausible and useful weakening of the convexity assumption for complex Hessian second-order estimates, held up by a non-self-contained concavity lemma that needs referee verification.","tokens_in":806,"tokens_out":1779,"would_cite":false,"duration_ms":155300,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35J15","53C55","58J05","35B45"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that admissible, χ-semi-convex solutions of complex Hessian equations on compact Hermitian manifolds satisfy a uniform second-order bound.","keywords":["complex Hessian equations","second order estimates","semi-convexity","Hermitian manifolds","concavity inequality","a priori estimates","elementary symmetric function"],"falsifier":"The most direct check is algebraic: for $n=2$ and $k=2$, compute both sides of (1.5) explicitly for a Hermitian matrix $W$ with eigenvalues in $\\Gamma_k$, $\\lambda_n>-A$, and $\\lambda_1$ large, for each index $j$ and for arbitrarily small $\\epsilon_0>0$; a single violation would refute Lemma 1.2, and since that lemma is the step importing the real estimates, Theorem 1.1 would be left without proof.","tokens_in":14487,"feed_emoji":"📐","tokens_out":11668,"duration_ms":104500,"temperature":0.7,"pith_summary":"This paper proves a uniform second-order estimate, a bound of the form $|D\\bar D u|\\le C$, for smooth admissible solutions of complex Hessian equations on compact Hermitian manifolds. The estimate is established under the assumption that the form $\\chi$ is semi-convex, meaning all its eigenvalues are bounded below by a constant $-A$, rather than under the stronger assumption $\\chi\\in\\Gamma_{k+1}(M)$ used in earlier work. Because the equation carries gradient terms on both sides and the background metric may have torsion, the main difficulty is controlling third-order terms; the paper's response is a modified concavity inequality for the $k$-th elementary symmetric function. A uniform $C^2$ bound of this kind is the key a priori estimate needed to run the continuity method for the equation.","feed_headline":"Semi-convex Hessian solutions get uniform second-order bounds","feed_subtitle":"A new concavity inequality weakens the standing convexity assumption for complex Hessian equations","key_machinery":"The load-bearing object is the modified concavity inequality (Lemma 1.2) for the $k$-th elementary symmetric function $\\sigma_k$ of the eigenvalues of a Hermitian tensor. The inequality says that when the eigenvalues lie in the Gårding cone $\\Gamma_k$ with $\\lambda_1\\ge\\cdots\\ge\\lambda_n>-A$ and $\\lambda_1$ is large, the negative second-derivative terms from differentiating the equation twice are bounded below by a positive multiple of $\\sigma_k^{11}|\\omega_{11j}|^2/(\\lambda_1\\sigma_k)$ plus lower-order terms, with the loss factor $(1-\\epsilon_0)$ for arbitrarily small $\\epsilon_0$. The proof transfers the real-case estimates (3.27) and (3.73) of [43] to complex Hermitian tensors, and the final section feeds this inequality into the auxiliary function $Q=\\log\\lambda_1+\\varphi(|\\nabla u|^2)+\\phi(u)$, using the perturbation argument of [6] to make the largest eigenvalue smooth and the Hermitian commutation formulas of [37] together with the symmetric-function identities of [2] to control all third-order terms.","core_discovery":"The central claim is Theorem 1.1: on a compact Hermitian manifold $(M,\\omega)$ of complex dimension $n$, if $\\chi'(z,u)\\ge\\epsilon\\omega$ and $u$ is an admissible, $\\chi$-semi-convex $C^\\infty$ solution of the Hessian equation $\\chi^k\\wedge\\omega^{n-k}=\\psi(z,Du,u)\\omega^n$, then all second covariant derivatives of $u$ are bounded by a constant depending only on $(M,\\omega)$, $n$, $k$, $\\epsilon$, $\\chi'$, $\\psi$, $a$, $\\sup_M|u|$, and $\\sup_M|Du|$. The proof establishes a complex version of the real-variable concavity inequality, adjusting the coefficients so that the good third-order terms are retained even though complex conjugacy removes some terms that the real proof could use. A direct consequence is that the second-order estimate survives under the weaker and more natural semi-convexity assumption instead of the full cone condition $\\chi\\in\\Gamma_{k+1}(M)$.","pith_inferences":["If the transferred real-variable estimates genuinely survive in the complex Hermitian setting, the same concavity inequality should be reusable for parabolic or degenerate variants of (1.2) where the $\\Gamma_{k+1}$ assumption was previously the bottleneck.","The coefficient optimization from $2$ to $1-\\epsilon_0$ suggests the method may tolerate position-dependent lower bounds on the eigenvalues, which would be useful for geometric applications where curvature terms enter the estimates.","One could stress-test the inequality in the model case $n=k=2$ with explicit formulas for $\\sigma_k$; if the constants are not sharp, the proof may yield stronger estimates, and if a violation appears, it would pinpoint the transfer from the real case as the fragile step."],"forward_implications":["If Theorem 1.1 is correct, the uniform second-order estimate requires only semi-convexity of $\\chi$, not the stronger condition $\\chi\\in\\Gamma_{k+1}(M)$ used in earlier second-order estimates.","Corollary 1 follows directly: the same estimate holds whenever $\\sigma_{k+1}(\\chi)>-A$, since this lower bound implies $\\chi$-semi-convexity.","The bound is the main a priori estimate in the continuity method, so a proof of existence of smooth admissible solutions of (1.2) reduces to establishing zero-order and gradient estimates.","The estimate covers equations with the gradient term $\\psi(z,Du,u)$ on the right-hand side and with torsion terms from the Hermitian metric, extending the earlier $\\Gamma_{k+1}$ result to the semi-convex regime.","The constant in the estimate does not depend on higher derivatives of $u$, only on the fixed data and on $\\sup_M|u|$ and $\\sup_M|Du|$, so the bound remains stable along a continuity path."],"supporting_citations":[{"why":"Supplies the real-variable concavity estimates (3.27) and (3.73) that Lemma 1.2 transfers to the complex Hermitian setting; this transfer is the load-bearing unproved step.","marker":"[43]"},{"why":"Introduces the semi-convexity framework and the concavity inequality for real Hessian equations that the present inequality modifies.","marker":"[26]"},{"why":"Provides the computational lemmas (2.9)-(2.10) used directly in Section 4 and the earlier second-order estimate under $\\chi\\in\\Gamma_{k+1}(M)$ that Theorem 1.1 weakens.","marker":"[10]"},{"why":"Gives the perturbation argument that makes the largest eigenvalue smooth at the maximum point of the auxiliary function.","marker":"[6]"},{"why":"Supplies the Hermitian commutation formulas (2.4)-(2.5) relating mixed covariant derivatives and torsion.","marker":"[37]"},{"why":"Provides the symmetric-function derivative formulas (2.2)-(2.3) used to differentiate the equation twice and to expand the second-order terms.","marker":"[2]"}],"fun_headline_variants":["Semi-convexity suffices for uniform Hessian bounds","Relaxed convexity still yields second-order estimates","Weaker assumption, same second-order control for Hessian PDEs","Semi-convex approach to Hessian estimates on Hermitian manifolds","New concavity inequality relaxes Hessian convexity requirement"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the real-variable concavity estimates (3.27) and (3.73) of [43] continue to hold for complex Hermitian tensors with torsion, a transfer the proof invokes rather than reproduces; if that transfer fails, the key inequality (1.5) and with it Theorem 1.1 collapse.","fun_headline_variants_meta":{"raw":{"variants":["Semi-convexity suffices for uniform Hessian bounds","Relaxed convexity still yields second-order estimates","Weaker assumption, same second-order control for Hessian PDEs","Semi-convex approach to Hessian estimates on Hermitian manifolds","New concavity inequality relaxes Hessian convexity requirement"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000881,"raw_usage":{"total_tokens":3745,"prompt_tokens":821,"completion_tokens":2924,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":437,"completion_tokens_details":{"reasoning_tokens":2837}},"tokens_in":437,"tokens_out":2924,"duration_ms":20555,"temperature":1.0,"reasoning_tokens":2837,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:36:44.836707+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The most direct check is algebraic: for $n=2$ and $k=2$, compute both sides of (1.5) explicitly for a Hermitian matrix $W$ with eigenvalues in $\\Gamma_k$, $\\lambda_n>-A$, and $\\lambda_1$ large, for each index $j$ and for arbitrarily small $\\epsilon_0>0$; a single violation would refute Lemma 1.2, and since that lemma is the step importing the real estimates, Theorem 1.1 would be left without proof.","supporting_citations":[{"cited_title":"$C^2$ estimates for $k$-Hessian equations and a rigidity theorem","cited_arxiv_id":"2408.10781","evidence_quote":"Supplies the real-variable concavity estimates (3.27) and (3.73) that Lemma 1.2 transfers to the complex Hermitian setting; this transfer is the load-bearing unproved step."},{"cited_title":"Lu, Curvature estimates for semi-convex solutions of Hessian equations in hyperbolic space, Calc","cited_arxiv_id":null,"evidence_quote":"Introduces the semi-convexity framework and the concavity inequality for real Hessian equations that the present inequality modifies."},{"cited_title":"Dong, Second order estimates for complex Hessian equations with gradient terms on both sides, J","cited_arxiv_id":null,"evidence_quote":"Provides the computational lemmas (2.9)-(2.10) used directly in Section 4 and the earlier second-order estimate under $\\chi\\in\\Gamma_{k+1}(M)$ that Theorem 1.1 weakens."},{"cited_title":"Chu, A simple proof of curvature estimates for convex solution of k-Hessian equation, Proc","cited_arxiv_id":null,"evidence_quote":"Gives the perturbation argument that makes the largest eigenvalue smooth at the maximum point of the auxiliary function."},{"cited_title":"Tosatti, B","cited_arxiv_id":null,"evidence_quote":"Supplies the Hermitian commutation formulas (2.4)-(2.5) relating mixed covariant derivatives and torsion."},{"cited_title":"Ball, Differentiability properties of symmetric and isotropic functions, Duke Math","cited_arxiv_id":null,"evidence_quote":"Provides the symmetric-function derivative formulas (2.2)-(2.3) used to differentiate the equation twice and to expand the second-order terms."}],"review_version":1}