{"id":"26c7a361-2bda-4d09-81b1-543283fbca93","arxiv_id":"2505.12817","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For the complex Monge-Ampere eigenvalue problem on a strictly convex domain in C^2, the unique eigenfunction u satisfies that -log(-u) is strictly convex.","lead":"This paper proves that on a smooth strictly convex domain in C^2, the eigenfunction of the complex Monge-Ampere Dirichlet problem is log-concave, meaning v = -log(-u/4) has a positive definite Hessian everywhere. The authors concede in Remark 1.2 that the same fact was already proven in all dimensions by Chen, Li and Ma, so the theorem itself is not new and the contribution is an independent proof by a different method.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Equation (2.6)/(2.7) drops a factor of 16: the constant rank theorem is proved for F = expression - λ rather than for the actual transformed complex Monge-Ampère equation, whose determinant is one sixteenth of the displayed expression.","rationale":"The paper's central claim is independently corroborated by Chen-Li-Ma [12], so I do not see evidence that the theorem itself is false. The reader's conditional verdict is therefore appropriate. However, the proof as written has a sharper internal inconsistency than the general dependence on [42] identified by the reader: the real-coordinate equation (2.6) and the operator F in (2.7) are missing the factor 16 that the authors themselves use in Lemma 4.1. Since Theorems 3.1, 3.2, and Corollary 3.3 are stated exclusively for this F, the constant rank theorem is proved for a PDE that is not the transformed complex Monge-Ampère equation unless λ is rescaled. This is a concrete, checkable defect rather than an unsupported assertion. I also note the companion typo in (2.7), where v should tend to +∞ on ∂Ω, not to 0; the proof of Theorem 1.1 later uses the correct +∞ normalization. Because the missing factor only changes λ by a positive constant, the positivity-based constant rank argument may survive a global rescaling, but this must be verified by tracing λ through the algebraic identities of Section 3. For this reason I would keep the verdict at CONDITIONAL rather than accept the manuscript as written, but I do not see a reason to move to REJECT: the claim is credible, the ball case is internally correct, and the identified issue is repairable if the rescaling is consistent.","tokens_in":30147,"tokens_out":25234,"duration_ms":256606,"concrete_test":"Independently compute det(H) for H_{ij} = 1/4[(v_{x_ix_j}+v_{y_iy_j}) - (v_{x_i}v_{x_j}+v_{y_i}v_{y_j})] + i/4[(v_{x_iy_j}-v_{x_jy_i}) - (v_{x_i}v_{y_j}-v_{y_i}v_{x_j})] and verify that it equals one sixteenth of the bracket in (2.6). Then replace every λ in the F of Section 3 by 16λ and re-check the key identities used in Claim 1 and Claims 2-3, especially line (3.20), the coefficient simplifications for M1 and M2, and the final positivity estimates P2(M1) and P3(M2). If all inequalities remain valid, the factor-16 issue is a typographical rescaling and the proof can be repaired; if any positivity estimate fails, the constant rank theorem does not apply to (1.2) and Theorem 1.1 is unproved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing problem is that the operator F used throughout Section 3 is not the transformed complex Monge-Ampère operator. Starting from (2.2), det(v_{z_i\\bar z_j} - v_{z_i}v_{\\bar z_j}) = λ, and using (2.3)-(2.5), each entry of the relevant 2x2 matrix is one quarter of a real bilinear expression; hence its determinant is one sixteenth of the bracket displayed in (2.6). Equation (2.6) should therefore read 16λ = (...) and the F in (2.7) should be (...) - 16λ. The paper instead writes λ = (...), and Theorems 3.1, 3.2 and Corollary 3.3 are stated and proved for F = (...) - λ. Lemma 4.1 confirms the missing factor: equation (4.1) contains 16λu^2 and the ODE (4.3) contains 8λ. Unless λ in Section 3 is secretly 16 times the eigenvalue of (1.2), Corollary 3.3 is a theorem about a different PDE, so the application to v_t in Section 4 is not justified as written. A global rescaling λ -> 16λ may repair the argument because the proof mostly uses λ > 0, but the identities (3.20), Claim 1, and the M1/M2 positivity computations must be rechecked with this replacement; the correction is not automatic.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims that for a bounded, smooth, strictly convex domain in C^2, the unique solution u of the complex Monge-Ampère Dirichlet eigenvalue problem det(u_{i\\bar j}) = λ(Ω)(-u)^2 is such that v = -log(-u/4) is strictly convex in Ω (Theorem 1.1). The proof combines a boundary-strip convexity estimate, a constant rank theorem for the real Hessian equation satisfied by v, and a deformation argument from the ball. The constant rank theorem is the main technical content: it treats the cases where the real Hessian has minimal rank 2 or 3, using auxiliary functions adapted to the complex structure, following the framework of Bian and Guan. The ball case is handled in Lemma 4.1, and the paper explicitly notes in Remark 1.2 that a more general result was subsequently obtained by Chen, Li and Ma, whose theorem in complex dimension 2 coincides with Theorem 1.1.","tokens_in":30320,"tokens_out":8239,"duration_ms":84078,"significance":"If the proof is correct, the paper provides an independent proof of log-concavity of the eigenfunction in C^2 and, more importantly, a constant rank theorem for a fully nonlinear equation that is not concave in the real Hessian. The ball case, Lemma 4.1, is explicit and checkable, and the overall strategy is classical. However, the novelty of the main theorem is limited by the acknowledged result of Chen, Li and Ma [12], which already covers Theorem 1.1 as a special case. The potential value of the paper therefore lies in the constant rank technique itself, which must be correct and self-contained. The reader's verification of the ball case appears sound, but the central constant rank theorem has a scaling inconsistency with the transformed equation, as detailed below.","major_comments":[{"comment":"The constant rank theorem is proved for the wrong equation by a factor of 16. From (2.3)–(2.5), A = 1/4(U+W) + (√-1/4)(V-V^T), so det(A) = 1/16[(v11+v33)(v22+v44) - (v12+v34)^2 - (v14-v32)^2]. Since the transformed equation (2.2) is det(A) = λ, equation (2.6) should have 16λ on the left, and F in (2.7) should be the bracket minus 16λ, not the bracket minus λ. Lemma 4.1 confirms the missing factor: equation (4.1) contains 16λu^2 and the ODE (4.3) contains 8λ. Thus Theorems 3.1, 3.2, and Corollary 3.3 are established for a different PDE. The error may be repairable by replacing λ with 16λ throughout Section 3, because the argument mostly uses λ > 0, but the identities (3.20), Claim 1, and Claims 2–3 must be rechecked with this replacement; the correction is not merely notational.","section":"§2, (2.6)–(2.7), and Lemma 4.1"},{"comment":"The deformation argument assumes uniform C^3 bounds for v_t on Ω_t, stated only as \"According to the a priori estimates for the eigenvalue problem (1.2) established by Chu, Liu and McCleerey [14]\". The proof needs a precise statement of the estimate, including its dependence on the geometry and its uniformity in t, and verification that the normalized transformed family v_t = -log(-u_t/4) with inf v_t = 2 log 2 satisfies it. Without this, the passage from strict convexity for t < t0 to convexity but not strict convexity at the first loss time t0 is not fully justified.","section":"§4, proof of Theorem 1.1"},{"comment":"The proof of Corollary 3.3 depends on imported machinery from the authors' preprint [42]: the rank identity rank(∇^2v) = rank(K) + 2, the adapted coordinate system, and Lemmas 2.2–2.3. These are load-bearing for the constant rank theorem but are not proved in the present paper and their precise hypotheses are only summarized. If [42] is not yet published, the manuscript should include proofs of these statements or at least state them with full hypotheses; as it stands, the constant rank theorem is not self-contained.","section":"§2, (2.8)–(2.9), Lemmas 2.2–2.3"}],"minor_comments":[{"comment":"The word \"probelm\" in the first paragraph should be \"problem\".","section":"§1.1"},{"comment":"In the expansion of φ_{ij}, the term (v11j+v33j)(v22i+v33i) appears to be a typo; it should presumably be (v22i+v44i) by symmetry with the preceding terms.","section":"§3.1, display after (3.9)"},{"comment":"The notation σ_2^1(B) and σ_1^3(B) is used without definition; the paper defines σ_k(W|i) and σ_k(W|ij), but not the superscript notation used in Lemma 2.3. Please clarify.","section":"§2, Lemma 2.3"},{"comment":"Because Remark 1.2 acknowledges that Theorem 1.1 is already contained in the more general result of Chen, Li and Ma [12], the introduction should state explicitly that the paper's contribution is the constant rank proof rather than the log-concavity theorem itself.","section":"§1.2, Remark 1.2"}],"recommendation":"major_revision","confidential_remarks":"The factor-of-16 inconsistency between Section 2 and Lemma 4.1 is the main technical obstruction: it means the constant rank theorem is proved for a different equation, and while rescaling λ likely repairs the argument, the authors must redo the relevant computations. The paper's novelty is also limited by [12], which the authors themselves acknowledge; if the constant rank proof is correct and self-contained, that may still be a publishable contribution, but the introduction should be repositioned. Finally, the heavy reliance on the authors' preprint [42] for rank identities and auxiliary-function estimates should be addressed editorially, either by requiring self-contained proofs or by ensuring a stable citable version of [42] is available."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the theorem is not new. Remark 1.2 says Chen, Li and Ma [12] proved strict real log-concavity of the first complex 2-Hessian eigenfunction in arbitrary dimension, and in C^2 that is exactly Theorem 1.1. What this paper offers is an independent constant-rank proof of that special case, building on the authors' preprint [42]. I think the proof has a real but fixable error.\n\nI checked the factor 16. From u=-4e^{-v}, the equation det(u_{i\\bar j})=lambda(-u)^2 becomes det(v_{z_i\\bar z_j}-v_{z_i}v_{\\bar z_j})=lambda. Writing that 2x2 determinant in real coordinates gives 16lambda on the left, not lambda. The paper's (2.6) omits the 16, and F in (2.7) subtracts lambda instead of 16lambda. So Section 3 proves a constant rank theorem for the wrong PDE. Lemma 4.1 uses the correct factor: (4.1) has 16lambda u^2. The two parts are inconsistent as written. This is not obviously fatal: the proof mostly uses lambda>0 and algebraic identities, and a global rescaling lambda -> 16lambda should repair the statement. But (3.20), Claim 1, and Claims 2-3 must be rechecked after that replacement, and Corollary 3.3 as written does not apply to v_t.\n\nThe ball case is correct: the ODE (4.3), the derivative computation at the first zero of v'', and the contradiction all line up. The deformation-plus-boundary-strip skeleton is the standard Caffarelli-Friedman/Korevaar-Lewis argument and would work if the constant rank theorem survives. The main soft spots are the imports: rank identity (2.9), the adapted coordinates, and Lemmas 2.2-2.3 come from [42], and Theorems 3.1-3.2 are long computations I could not fully verify. The uniform C^3 bounds from [14] are asserted without a precise statement, and the withdrawal of the earlier arXiv version is unexplained. None of these is obviously fatal, but together they mean the proof needs careful refereeing.\n\nWho is this for? Someone working on convexity properties of complex Hessian equations and on constant rank techniques. The result itself will be cited through [12]; this paper's value is as an independent, technique-heavy proof of a special case. I would send it to a serious referee, with instructions to verify the factor 16 and the dependence on [42].","headline":"The theorem is already in Chen-Li-Ma by the authors' own admission, and the proof as written drops a factor of 16 in the real-coordinate equation, but the argument looks repairable and the paper deserves a serious referee.","tokens_in":31083,"tokens_out":11461,"would_cite":false,"duration_ms":102005,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B50","32W20"],"pacs":[],"model":"deepseek-v4-flash","headline":"The first complex Monge-Ampère eigenfunction is log-concave on every smooth strictly convex domain in C².","keywords":["log-concavity","complex Monge-Ampère operator","eigenvalue problem","constant rank theorem","strict convexity","plurisubharmonic functions","deformation method","real Hessian"],"falsifier":"Numerically compute the first eigenfunction $u$ of $\\det(u_{i\\bar j})=\\lambda(\\Omega)(-u)^2$ on a smooth strictly convex domain that is not a ball, say a convex ellipsoid in $\\mathbb{C}^2$ with unequal axes; then evaluate the smallest eigenvalue of the real Hessian of $v=-\\log(-u/4)$ on a fine interior grid. A single interior point where that eigenvalue is $\\le 0$, while the boundary strip is strictly convex, would contradict Theorem 1.1 and its constant-rank corollary.","tokens_in":29723,"feed_emoji":"📐","tokens_out":13293,"duration_ms":120505,"temperature":0.7,"pith_summary":"This paper proves that the first eigenfunction of the complex Monge-Ampère Dirichlet problem is log-concave on any bounded, smooth, strictly convex domain in two complex dimensions. Concretely, if $u$ solves $\\det(u_{i\\bar j})=\\lambda(\\Omega)(-u)^2$ with $u=0$ on $\\partial\\Omega$, then the rescaled function $v=-\\log(-u/4)$ has a strictly positive definite real Hessian throughout the interior. The proof converts the complex equation into a real fully nonlinear equation for $v$, proves a constant rank theorem for the real Hessian of $v$, and then deforms any such domain into a ball, showing that strict convexity cannot be lost along the deformation. The property matters because log-concavity is the shape condition underlying Brunn–Minkowski-type inequalities for eigenvalues, and it gives the complex Monge-Ampère operator the same shape property long known for Laplacian eigenfunctions on convex domains.","feed_headline":"First complex Monge-Ampère eigenfunction is log-concave in C²","feed_subtitle":"A logarithmic rescaling forces strict convexity on every smooth strictly convex domain in two complex dimensions.","key_machinery":"The load-bearing mechanism is the auxiliary matrix $K = I - B A^{-1} B A^{-1}$, formed from the complex Hessians $A=(v_{z_i\\bar z_j})$ and $B=(v_{z_i z_j})$, together with the rank identity $\\operatorname{rank}(\\nabla^2 v)=\\operatorname{rank}(K)+2$ that reduces the $4\\times4$ real Hessian to a $2\\times2$ complex object. In coordinates adapted via the Autonne–Takagi factorization (a complex linear change of coordinates that diagonalizes the real Hessian), $\\nabla^2 v$ is diagonal, and its minimal rank can only be $2$ or $3$. For each case the paper constructs an auxiliary function $\\phi$ — $\\operatorname{tr}(K)+\\det(K)/\\operatorname{tr}(K)$ for minimal rank $2$, and $\\det(K)$ for minimal rank $3$ — proves the differential inequality $\\sum_{i,j}F_{ij}\\phi_{ij}\\le C(\\phi+|\\nabla\\phi|)$, and invokes the strong maximum principle to conclude that $\\nabla^2 v$ has constant rank. The boundary strict-convexity estimates then contradict constant degeneracy, so the deformation from the ball preserves strict convexity.","core_discovery":"The central claim is Theorem 1.1: on a bounded, smooth, strictly convex domain $\\Omega \\subset \\mathbb{C}^2$, the unique plurisubharmonic (complex-convex) solution $u \\in C^\\infty(\\Omega)\\cap C^{1,1}(\\bar\\Omega)$ of the eigenvalue problem $\\det(u_{i\\bar j})=\\lambda(\\Omega)(-u)^2$ in $\\Omega$, with $u<0$ and $u=0$ on $\\partial\\Omega$, has the property that $v=-\\log(-u/4)$ is strictly convex in $\\Omega$. Equivalently, the positive function $-u$ is strictly log-concave. The proof reduces the question to a constant rank theorem: for solutions of the transformed real equation $F(\\nabla^2 v,\\nabla v)=0$ in $\\mathbb{R}^4$, the real Hessian $\\nabla^2 v$ cannot change rank inside the domain. Strict convexity is known outright for the ball, boundary estimates force full rank near $\\partial\\Omega$, and the deformation from a ball to any smooth strictly convex domain then rules out any interior loss of strict convexity.","pith_inferences":["If the rank-identity strategy scales, the same strict log-concavity should hold for the complex Monge-Ampère eigenfunction in all dimensions $n\\ge 2$; the two-dimensional case here is the natural test case for that mechanism.","Strict log-concavity of the eigenfunction is the analytic input from which a Brunn–Minkowski inequality for the first complex Monge-Ampère eigenvalue in $\\mathbb{C}^2$ is expected to follow by standard factorization arguments; the paper does not draw that consequence.","The mechanism reveals that real-Hessian convexity is compatible with unitary-invariance of the complex equation: it is the logarithmic change of variables, not a real-linear symmetry, that exposes the convexity, suggesting similar results for other complex fully nonlinear Dirichlet problems on convex domains."],"forward_implications":["On every bounded, smooth, strictly convex domain in $\\mathbb{C}^2$, the level sets $\\{x : -u(x) \\ge c\\}$ of the first complex Monge-Ampère eigenfunction are strictly convex bodies.","The real Hessian of $v=-\\log(-u/4)$ has constant rank throughout the domain, so it cannot pass from full rank near the boundary to a degenerate rank in the interior.","Strict convexity of $v$ persists along every smooth deformation of the domain that keeps it smooth and strictly convex, giving a whole family of log-concave eigenfunctions."],"supporting_citations":[{"why":"Supplies the rank identity rank(∇²v)=rank(K)+2, the adapted coordinate system, and Lemmas 2.2–2.3 used in Theorems 3.1–3.2.","marker":"[42]"},{"why":"Provides the constant-rank framework and the auxiliary-function construction that the paper adapts, along with Lemmas 2.4–2.5 for the differential estimates.","marker":"[5]"},{"why":"Establishes existence, uniqueness, and the uniform C³ and C^{1,1} a priori bounds for solutions along the domain deformation used in the proof of Theorem 1.1.","marker":"[14]"},{"why":"Origin of the constant-rank-plus-deformation method and of the boundary strict-convexity estimate recorded as Lemma 4.2.","marker":"[9]"},{"why":"Extends the constant-rank-plus-deformation approach to higher dimensions; the deformation step's contradiction follows its scheme.","marker":"[23]"},{"why":"Provides the boundary convexity estimate (Lemma 4.2) that forces full rank of ∇²v near ∂Ω.","marker":"[22]"}],"fun_headline_variants":["Monge-Ampère eigenfunctions in C² are strictly log-concave","Log-concavity proven for complex Monge-Ampère eigenfunctions","Eigenfunction log-concavity for complex Monge-Ampère on C² domains","Complex Monge-Ampère: eigenfunction is log-concave in C²","Strict log-concavity of Monge-Ampère eigenfunctions in two complex dimensions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof stands or falls on the constant rank theorem for the transformed real equation: if the real Hessian of $v$ could change rank at an interior point, the deformation argument would have no route to rule out the first loss of strict convexity.","fun_headline_variants_meta":{"raw":{"variants":["Monge-Ampère eigenfunctions in C² are strictly log-concave","Log-concavity proven for complex Monge-Ampère eigenfunctions","Eigenfunction log-concavity for complex Monge-Ampère on C² domains","Complex Monge-Ampère: eigenfunction is log-concave in C²","Strict log-concavity of Monge-Ampère eigenfunctions in two complex dimensions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000987,"raw_usage":{"total_tokens":4155,"prompt_tokens":881,"completion_tokens":3274,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":497,"completion_tokens_details":{"reasoning_tokens":3162}},"tokens_in":497,"tokens_out":3274,"duration_ms":24435,"temperature":1.0,"reasoning_tokens":3162,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:29:26.241142+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically compute the first eigenfunction $u$ of $\\det(u_{i\\bar j})=\\lambda(\\Omega)(-u)^2$ on a smooth strictly convex domain that is not a ball, say a convex ellipsoid in $\\mathbb{C}^2$ with unequal axes; then evaluate the smallest eigenvalue of the real Hessian of $v=-\\log(-u/4)$ on a fine interior grid. A single interior point where that eigenvalue is $\\le 0$, while the boundary strip is strictly convex, would contradict Theorem 1.1 and its constant-rank corollary.","supporting_citations":[{"cited_title":"A microscopic convexity principle for nonlinear partial differential equations.Invent","cited_arxiv_id":null,"evidence_quote":"Provides the constant-rank framework and the auxiliary-function construction that the paper adapts, along with Lemmas 2.4–2.5 for the differential estimates."},{"cited_title":"The eigenvalue problem for the complex Hessian operator onm-pseudoconvex manifolds.J","cited_arxiv_id":null,"evidence_quote":"Establishes existence, uniqueness, and the uniform C³ and C^{1,1} a priori bounds for solutions along the domain deformation used in the proof of Theorem 1.1."},{"cited_title":"Caffarelli and Avner Friedman","cited_arxiv_id":null,"evidence_quote":"Origin of the constant-rank-plus-deformation method and of the boundary strict-convexity estimate recorded as Lemma 4.2."},{"cited_title":"Korevaar and John L","cited_arxiv_id":null,"evidence_quote":"Extends the constant-rank-plus-deformation approach to higher dimensions; the deformation step's contradiction follows its scheme."},{"cited_title":"Korevaar","cited_arxiv_id":null,"evidence_quote":"Provides the boundary convexity estimate (Lemma 4.2) that forces full rank of ∇²v near ∂Ω."}],"review_version":1}