{"id":"1e9d3924-ff76-47ed-9cb9-14840d8ee7eb","arxiv_id":"2506.01434","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A new monotone formula for the k-Hessian equation yields ball characterizations for exterior overdetermined problems and recovers sharp geometric inequalities.","lead":"This paper proves that if a solution to the exterior k-Hessian equation has constant boundary gradient, the domain must be a ball, for 1 ≤ k < n/2. It also develops a two-parameter family of monotone quantities that yield sharp geometric inequalities for k-convex star-shaped domains.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3 hinges on inequality (6), whose proof rests on Lemma 6: the C^2 asymptotic expansion and full regularity of every level set, cited from the unpublished preprint [34].","rationale":"The reader identified Lemma 6 and the level-set regularity as the weakest assumption, and I concur. Tracing the logic of Theorem 3, the only PDE input that produces the sharp inequality forcing the Alexandrov-Fenchel equality is inequality (6), which is a direct consequence of the monotone formula. The monotone formula in turn requires the asymptotic expansion of the solution at infinity and the smoothness of all level sets to compute the limit (31) and to justify integration by parts over the level sets. All of these are imported from unpublished preprints [21] and [34]; the manuscript supplies no proof. This is not an internal inconsistency, but it is a genuine correctness risk: if Xiao's asymptotic lemma is only true in C^{1,1} or if a level set is not regular for some s, the core geometric inequality and the overdetermined result are unproven. I did not find a flaw in the algebraic derivations of (40)-(41) or in the use of the Alexandrov-Fenchel inequality; the stated chain is coherent conditional on Lemma 6 and the C^0-C^2 estimates. The reader's conditional verdict is therefore appropriate, and no adjustment is needed beyond insisting that the cited preprints be made available and the key lemma explicitly checked.","tokens_in":14189,"tokens_out":29870,"duration_ms":272569,"concrete_test":"Independently verify Lemma 6 for a non-spherical smooth convex domain: apply the Kelvin transform v(y)=|y|^{n/k-2} u(y/|y|^2) so the asymptotic at infinity becomes the boundary behavior at y=0, then prove from the C^{1,1} estimates of [21] and comparison with the explicit radial solution that v is C^2 at 0 with the matching first and second derivatives, and that each level set {v=s} is smooth near 0. If the second derivative error fails to be o(|x|^{-n/k}) or some level set is singular, recompute Lemma 8's limit (31); any change would invalidate inequality (6) and hence Theorem 3.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central result, Theorem 3, is derived by combining the integral identities (32) and (34) with the geometric inequality (6). Inequality (6) follows from the monotone formula in Theorem 1, whose proof uses the divergence theorem on each level set Σ_t and the limit computation (31) in Lemma 8. Both steps depend on Lemma 6, taken verbatim from Xiao's unpublished preprint [34]: the asymptotic expansion u = -ρ|x|^{2-n/k} + o(|x|^{2-n/k}) in C^2 topology and the assertion that every level set {u=s}, s∈[-1,0), is regular. Specifically, (25)-(27) are inserted into (14)-(15) to obtain the leading-order expressions in Lemma 8, and the regularity of all level sets is needed for F(t) to be well-defined and for the divergence theorem in Proposition 7. The manuscript does not prove these facts; it only cites [21] and [34]. If the second-order error in (27) is not uniform, or if some level set develops a singularity, then the limit (31) is not justified, inequality (6) is unsupported, and the chain (40)-(41) leading to the Alexandrov-Fenchel equality in Theorem 3 has no foundation. This is the single most load-bearing assumption in the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the exterior Dirichlet problem (1) for the homogeneous k-Hessian equation with 1 ≤ k < n/2 on the complement of a smooth bounded domain Ω. It introduces a two-term level-set quantity F(t) (equation (4)) and proves, under k-convexity and star-shapedness of Ω, a monotonicity formula for F(t) together with a lower bound whose equality case is claimed to characterize spheres. From this it derives the geometric inequalities (6) and (7) with equality cases, and then proves the main overdetermined result Theorem 3: if a solution of (1) additionally satisfies |∇u| = c on ∂Ω and Ω is smooth, bounded, and convex, then Ω is a ball. The proof of Theorem 3 combines two integral identities (Lemmas 9 and 10) with the geometric inequality (6) and a special Alexandrov-Fenchel inequality, following the strategy of Brandolini-Nitsch-Salani for the anisotropic capacity.","tokens_in":14416,"tokens_out":13001,"duration_ms":136358,"significance":"If the result is correct, Theorem 3 is a genuine extension of Reichel's overdetermined capacity theorem and of Brandolini-Nitsch-Salani's anisotropic-capacity result to the k-Hessian setting, and it answers the non-Weinberger case k ≥ 2. The monotone quantity (3) is a meaningful new construction: it genuinely combines two curvature integrals and its limit at infinity is computed rather than imposed. The paper also provides detailed PDE computations in the approximation argument. The main weakness is that the load-bearing asymptotic and regularity facts are imported from two unpublished arXiv preprints, and the equality case of the monotone formula is asserted rather than proved. Such a paper is publishable after the gaps identified below are fixed and the main lemma is made verifiable.","major_comments":[{"comment":"The proof of Theorem 1, and therefore the geometric inequality (6) used in Theorem 3, depends on the C^2 asymptotic expansion (25)-(27) and on the regularity of every level set {u = s}, s ∈ [-1,0), taken from the unpublished preprint [34]. Lemma 8 passes the o(1) terms in (25)-(27) through the integrals defining H_k, H_{k-1}, and the surface measures, and it uses the outer-minimizing argument (28)-(30) without a uniformity justification. Since the limit (31) is the only bridge between the PDE solution and the claimed inequality, the central argument is unsupported unless Lemma 6 is proved in the paper or replaced by a verifiable published statement. I ask the authors to provide a self-contained proof of, at least, the asymptotic expansion and the uniform remainder estimates, or to cite a published version of [34] that contains these statements.","section":"Section 3, Lemma 6 and Lemma 8"},{"comment":"The equality characterization in Proposition 7 is not proved. The proof ends with the assertion that if F(t) is constant then Σ_t is umbilical, but no argument is given for this implication, nor for why umbilicality of all level sets together with the star-shapedness and k-convexity assumptions forces Ω to be a ball. This equality statement is used in the 'if and only if' assertions of Theorems 1 and 2. Either supply the missing argument in detail or, if the main goal is Theorem 3, restrict the equality claims to the cases actually needed.","section":"Section 3, Proposition 7"},{"comment":"The final step of the proof of Theorem 3 invokes 'special Alexandrov-Fenchel inequalities' without stating them explicitly, only citing Section 7.4 of [28]. The exact inequality, its equality condition, and the verification that the convexity hypotheses of [28] hold for Ω are load-bearing, since this is the step that yields the equality forcing Ω to be a ball. Please write out the inequality and the equality case as a lemma, and make the reduction for k = 1 to [11] explicit.","section":"Section 4, proof of Theorem 3"},{"comment":"Lemma 9 and its use of (33) require the boundary convergence lim_{ε→0} |∇u_ε| = |∇u| = c on ∂Ω. This convergence is asserted with a citation to [21] and [34] but is not proved in the manuscript. Since (40), which is essential for Theorem 3, follows from the boundary terms in (32) and (34), this convergence should be justified in the paper or by a published reference.","section":"Section 4, Lemma 9"}],"minor_comments":[{"comment":"In the proof of Lemma 8, the sentence 'as t large enough, H > 0 along Σ_t' should read 'as t close to 0 from below' or 'as t → 0^-'.","section":"Section 3, Lemma 8"},{"comment":"In the estimate following (A8), the quantity written as H_1 should be H_{k-1} (or the appropriate curvature term from the preceding displayed formula); the notation should be corrected for consistency.","section":"Appendix A, Lemma 11"},{"comment":"In equation (37) the middle integral appears to have a missing factor u_l: the first term should involve S^{ij}_k u_i u_l x_l ν_j, not S^{ij}_k u_i x_l ν_j. Please check the displayed formula.","section":"Section 4, Lemma 10, equation (37)"},{"comment":"References [21] and [34] are arXiv preprints; if they have appeared in journals in the meantime, the published versions should be cited.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper's central theorem is contingent on two unpublished preprints, especially [34], for the asymptotic expansion and level-set regularity that feed into the limit (31). The authors should be asked to include a complete proof of the needed asymptotic/regularity statement or to replace it by a published reference, and to fill in the equality-case argument in Proposition 7. These issues are fixable within the scope of the paper, so I do not recommend rejection, but they block verification of the main claim as it stands."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version. The paper's real contribution is a two-parameter monotone formula F(t) for the exterior homogeneous k-Hessian problem (Theorem 1), and a ball characterization for the overdetermined problem for 2 ≤ k < n/2 under convexity (Theorem 3). The overdetermined result is new: Weinberger's approach fails for k ≥ 2, and the authors get around it by combining two integral identities with the geometric inequality (6). That is a solid piece of work.\n\nWhat it does well: the main computation in Proposition 7 is detailed and the structure is convincing. The authors are also honest about the provenance—Remark 2(1) says plainly that Theorem 3 uses Ma-Zhang's inequality, and the proofs of the integral identities are attributed to [21] and [34]. The geometric inequalities (6) and (7) are restatements of known results, but deriving them from one monotone formula is a real unifying simplification, not just a repackaging.\n\nThe soft spots are real but proportionate. The load-bearing assumption is Lemma 6, copied from Xiao's unpublished preprint [34]: a C^2 asymptotic expansion of u and regularity of every level set. Lemma 8 uses the expansion to compute the limits (31), and without uniform second-order control the limits do not follow. The paper does not prove these estimates, nor does it restate the precise hypotheses from [34] that make them true. A referee cannot fully check Theorem 1 without accessing [21] and [34]. The equality case in Proposition 7 is also asserted in one line (\"we can see that Σ_t is umbilical\"), which is too terse; that only affects the equality clauses in Theorem 2, not Theorem 3, but it should be fixed.\n\nI don't see a circularity problem. The monotone formula is genuinely constructed, and the overdetermined theorem uses external Alexandrov-Fenchel inequalities. The stress-test note about Lemma 6 is correct as a concern, but it identifies a documentation gap rather than a demonstrated error.\n\nWho is this for: anyone working on k-Hessian equations, exterior overdetermined problems, or geometric inequalities via level-set methods. It deserves a serious referee. The referee should verify Lemma 6 against [34], ask the authors to make the dependence explicit, and request a proof of the equality case in Proposition 7. Send it to peer review.","headline":"New monotone formula plus a genuine ball characterization for exterior k-Hessian problems; the proof leans on unverified estimates from two unpublished preprints, but that's a fixable gap.","tokens_in":14967,"tokens_out":3520,"would_cite":true,"duration_ms":37366,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35J60","35N25","52A40"],"pacs":[],"model":"deepseek-v4-flash","headline":"For the homogeneous $k$-Hessian equation in an exterior domain, the overdetermined condition $|\\nabla u|=c$ on the boundary is satisfied if and only if the domain is a ball.","keywords":["k-Hessian equation","overdetermined problem","exterior domain","ball characterization","monotone formula","geometric inequality","Aleksandrov-Fenchel inequality","k-admissible solution"],"falsifier":"Find a smooth bounded convex non-ball $\\Omega$ for which a numerical or analytic solution of (1) satisfies $|\\nabla u|=c$ on $\\partial\\Omega$; Theorem 3 asserts that no such pair exists. A sharper check is to test the asymptotic expansion of Lemma 6 on a non-star-shaped $(k-1)$-convex domain, since the proof's limit computation would break if that expansion fails.","tokens_in":13975,"feed_emoji":"🔵","tokens_out":13462,"duration_ms":119939,"temperature":0.7,"pith_summary":"The paper sets out to prove a sharp rigidity statement: for $1\\le k < n/2$, a smooth bounded convex domain $\\Omega$ in $\\mathbb{R}^n$ admits a $k$-admissible solution (a solution with the right Hessian ellipticity) of the exterior homogeneous $k$-Hessian equation with constant boundary gradient if and only if $\\Omega$ is a ball. The route is a new two-term monotone formula $F(t)$, built from level-set integrals of the $(k-1)$-st and $k$-th mean curvatures weighted by powers of $|\\nabla u|$, whose monotonicity yields sharp weighted and Minkowski-type geometric inequalities with equality only for balls. The paper then combines two integral identities with these inequalities to force the overdetermined problem onto the equality case of an Aleksandrov-Fenchel inequality. This matters because the classical P-function approach to overdetermined exterior problems appears not to work for $k\\ge2$, and the result extends the $k=1$ anisotropic-capacity characterization to the full range $1\\le k<n/2$ under convexity.","feed_headline":"Only balls solve exterior k-Hessian with fixed boundary gradient","feed_subtitle":"Monotone formula yields sharp geometric inequalities and pins down which convex domains allow constant boundary flux.","key_machinery":"The carrying object is the level-set functional $F(t)$ with two curvature integrands, weighted by powers of the gradient, and with coefficients $C_1(t)$, $C_2(t)$ prescribed by an ODE system. The ODE choice makes the derivative of $F(t)$ a sum of non-positive terms up to an approximation error that vanishes with $\\varepsilon$, so $F(t)$ is non-increasing on $[-1,0)$. The second ingredient is the asymptotic expansion $u(x)=-\\rho|x|^{2-n/k}+o(|x|^{2-n/k})$, which fixes the $t\\to0$ limit of $F(t)$ and the growth of the level sets, converting monotonicity into sharp lower bounds. For Theorem 3, the machinery is a pair of integral identities obtained by divergence theorem and Rellich-Pohozaev-type computation on $S^{ij}_k$ and $S^{ij}_{k-1}$; together with the Minkowskian integral formula they isolate the constant $c$ as a ratio of quermassintegrals, and the weighted inequality of Theorem 2 runs against the Aleksandrov-Fenchel inequality until equality forces all principal curvatures to be equal.","core_discovery":"The central claim is that the exterior Dirichlet problem for $S_k(\\nabla^2u)=0$, $u=-1$ on $\\partial\\Omega$, $u\\to0$ at infinity, solved by a $k$-admissible function $u$, is rigid under the extra boundary condition $|\\nabla u|=c$: for smooth bounded convex $\\Omega$ and $1\\le k<n/2$, the only such domain is the ball. The proof establishes a general monotone formula for $F(t)=C_1(t)\\int_{\\Sigma_t} H_k|\\nabla u|^a + C_2(t)\\int_{\\Sigma_t} H_{k-1}|\\nabla u|^{a+1}$ along level sets $\\Sigma_t=\\{u=t\\}$, with $C_1$ and $C_2$ chosen to solve a first-order ODE system so that $F$ is non-increasing in $t$. Using the asymptotic $u(x)=-\\rho|x|^{2-n/k}+o(|x|^{2-n/k})$, the paper computes the limiting value of $F(t)$ and obtains, for $a\\ge k(n-k-1)/(n-k)$, the sharp weighted inequality $\\frac{n-k}{n-2k}\\int_\\Sigma |\\nabla u|^{a+1}H_{k-1}\\le \\int_\\Sigma |\\nabla u|^a H_k$ and a generalized Minkowski inequality, each with equality exactly on balls. For the overdetermined theorem, two further integral identities give an explicit formula for the boundary gradient $c$ in terms of $\\int H_{k-1}/\\int H_{k-2}$; combining that formula with the weighted inequality reverses a special Aleksandrov-Fenchel inequality, so equality must hold in that inequality, which forces $\\Omega$ to be a ball.","pith_inferences":["Editorial inference: the same ODE-cancellation construction should extend to other homogeneous curvature equations whose solutions share an inverse-power asymptotic, giving monotone formulas for anisotropic or $p$-capacitary exterior problems.","Editorial inference: the equality case analysis suggests a quantitative stability statement—domains that nearly satisfy the overdetermined condition should be near-balls, with a closeness exponent controlled by the gap in the Aleksandrov-Fenchel inequality.","Editorial inference: the paper's reliance on level-set regularity at infinity for all $s\\in[-1,0)$ means the sharp inequalities in Theorems 1-2 are tied to the exterior solution theory for $(k-1)$-convex star-shaped domains; extending that theory to more general domains would automatically extend the inequalities."],"forward_implications":["For every $k$-convex star-shaped domain and every admissible exponent $a$, inequality (6) holds with equality only for balls; this recovers and sharpens the previously known weighted inequality for the exterior $k$-Hessian potential.","The generalized Minkowski inequality (7) holds for $k$-admissible exterior solutions, with the same sphere equality case, giving a level-set proof of a sharp quermassintegral bound.","The overdetermined problem (1)+(10) has no non-ball convex solution: any constant-flux exterior $k$-Hessian configuration in the allowed parameter range must be spherical.","The boundary gradient $c$ is not free data; it is determined by the domain through $c=\\frac{n-2k}{k}\\cdot\\frac{k-1}{n-k+1}\\cdot\\frac{\\int H_{k-1}}{\\int H_{k-2}}$, an explicit rigidity formula.","If the special Aleksandrov-Fenchel inequalities used in the final step can be proved under $k$-convexity and star-shapedness, the convexity hypothesis in Theorem 3 can be relaxed, as the paper notes."],"supporting_citations":[{"why":"Supplies the existence and approximation scheme for the exterior k-Hessian Dirichlet problem and the weighted geometric inequality that the monotone formula generalizes.","marker":"[21]"},{"why":"Supplies existence, the inverse-power asymptotic form of the solution, and level-set regularity over the full range of level values, used to compute the limiting value of the monotone quantity and the boundary identities.","marker":"[34]"},{"why":"Provides the k = 1 overdetermined anisotropic-capacity result and the integral-identity and geometric-inequality method that the paper adapts to all admissible k.","marker":"[11]"},{"why":"Supplies the Minkowskian integral formula and the special Aleksandrov-Fenchel inequalities whose equality case forces the domain to be a ball.","marker":"[28]"},{"why":"Represents the classical P-function maximum-principle route that the paper identifies as failing for k at least 2, motivating the alternative approach.","marker":"[31]"}],"fun_headline_variants":["Exterior k-Hessian rigidity: only balls allow constant boundary flux","Ball uniqueness for exterior k-Hessian with prescribed gradient","New monotone formula nails shape of k-Hessian domains","Constant boundary gradient forces ball in k-Hessian exterior problem"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on the exterior solution $u$ having regular level sets throughout and obeying the asymptotic $u(x)=-\\rho|x|^{2-n/k}+o(|x|^{2-n/k})$; if either fails for some admissible domain, the limiting value of $F(t)$ and the boundary integral identities no longer follow.","fun_headline_variants_meta":{"raw":{"variants":["Exterior k-Hessian rigidity: only balls allow constant boundary flux","Ball uniqueness for exterior k-Hessian with prescribed gradient","New monotone formula nails shape of k-Hessian domains","Constant boundary gradient forces ball in k-Hessian exterior problem"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000231,"raw_usage":{"total_tokens":1540,"prompt_tokens":1057,"completion_tokens":483,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":673,"completion_tokens_details":{"reasoning_tokens":411}},"tokens_in":673,"tokens_out":483,"duration_ms":4814,"temperature":1.0,"reasoning_tokens":411,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T11:43:53.242030+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find a smooth bounded convex non-ball $\\Omega$ for which a numerical or analytic solution of (1) satisfies $|\\nabla u|=c$ on $\\partial\\Omega$; Theorem 3 asserts that no such pair exists. A sharper check is to test the asymptotic expansion of Lemma 6 on a non-star-shaped $(k-1)$-convex domain, since the proof's limit computation would break if that expansion fails.","supporting_citations":[{"cited_title":"Brandolini, C","cited_arxiv_id":null,"evidence_quote":"Provides the k = 1 overdetermined anisotropic-capacity result and the integral-identity and geometric-inequality method that the paper adapts to all admissible k."},{"cited_title":"Schneider","cited_arxiv_id":null,"evidence_quote":"Supplies the Minkowskian integral formula and the special Aleksandrov-Fenchel inequalities whose equality case forces the domain to be a ball."},{"cited_title":"Weinberger","cited_arxiv_id":null,"evidence_quote":"Represents the classical P-function maximum-principle route that the paper identifies as failing for k at least 2, motivating the alternative approach."}],"review_version":1}