{"id":"c7cf001f-831f-4206-982b-cc6b8744cc2a","arxiv_id":"2507.08993","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The exterior Dirichlet problem for Hessian equations admits a unique smooth k-convex solution when the obstacle is star-shaped and strictly (k-1)-convex, not necessarily convex, for 2 <= k <= n.","lead":"This paper proves that a nonlinear PDE called a Hessian equation has a smooth solution outside a non-convex hole in space, as long as the hole is star-shaped and its boundary satisfies a curvature condition weaker than convexity. The result extends a classical existence theorem to a new class of domains and sharpens the known smoothness of the solution.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Bounded-domain approximants u_R in (3.1) are asserted but never constructed; the monotone limit in §3.4 needs this existence, and the paper supplies no theorem, citation, or continuity argument establishing it.","rationale":"The reader’s weakest-assumption analysis focuses on the strict (k−1)-convex subsolution inequality (2.9), which is plausible and well-motivated. My reading puts the most load-bearing missing piece earlier: the bounded-domain approximating solutions u_R of (3.1) are asserted but never proved to exist, and the a priori estimates in Sections 3.1–3.3 are conditional on that existence. The omitted viscosity-subsolution property of \\bar u in Remark 3.2 is one of the ingredients needed to apply the standard existence theory, so it is directly tied to the same gap. I do not see a counterexample to the main theorem, and the missing steps are likely fillable by standard methods; the correct verdict remains CONDITIONAL. A decisive check is to write down the applicable existence theorem and verify its hypotheses for (3.1), including the non-C^2 interface of the subsolution. If that verification succeeds, the paper is acceptable after adding the missing citations and proof; if it fails, Theorem 1.4 lacks a proof for arbitrary large R.","tokens_in":17009,"tokens_out":31688,"duration_ms":401869,"concrete_test":"State the standard existence theorem for σ_k(D^2u)=1 on a bounded C^2 domain with prescribed C^2 boundary data and check it against (3.1) for arbitrary R: (i) verify that φ and \\bar u_R are admissible boundary values in the sense of that theorem; (ii) verify that the function \\bar u defined in (2.21) satisfies the theorem’s subsolution hypothesis in the C^2 regions and has the correct viscosity subsolution property across ∂E1, supplying the missing step in Remark 3.2; (iii) if the cited theorem requires a C^2 subsolution, either mollify \\bar u across ∂E1 or extend the argument directly. If these checks pass, the gap is purely expository and the main theorem is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central existence proof in Section 3 is conditional on a family u_R of smooth strict k-convex solutions of (3.1): σ_k(D^2u_R)=1 in Ω_R=E_R\\bar D, u_R=φ on Γ, u_R=\\bar u_R on ∂E_R. Section 3 opens with “We will show that for any R… there exists…”, but what follows are only a priori C^0/C^1/C^2 estimates (Lemmas 3.1–3.9). No existence theorem is stated or cited, and no continuity or Perron argument is run. The monotonicity and convergence in §3.4 (“By the standard maximum principle … u_R→u”) presuppose these approximants. This is not a minor expository gap: if (3.1) has no admissible k-convex solution for some R—for instance, if the outer boundary data \\bar u_R is not admissible in the nonconvex ring—the limiting argument collapses. The natural repair is to invoke the known k-convex Dirichlet regularity theory (e.g., Caffarelli–Nirenberg–Spruck [5] and Bao–Li–Li [1]) with \\bar u as a subsolution and \\bar u_R as a supersolution. Remark 3.2 explicitly defers the proof that \\bar u is a viscosity subsolution (“with a small modification”), so even this prerequisite is not established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the exterior Dirichlet problem for Hessian equations, σ_k(D^2u)=1 in R^n\\overline{D} with u=φ on ∂D, where D is a bounded, smooth, star-shaped, strictly (k-1)-convex domain in R^n, n≥3, and the solution is required to have the quadratic growth prescribed by a matrix A∈A_k. The main result (Theorem 1.4) claims existence and uniqueness of a smooth strictly k-convex solution with the asymptotic behavior (1.2), extending earlier results of Bao–Li–Li [1] and Caffarelli–Li [4] from strictly convex domains to non-convex (k-1)-convex domains. Theorem 1.5 claims decay estimates for all derivatives of the remainder. The proof constructs a piecewise-smooth subsolution by gluing a radial power-type subsolution in E_1\\overline{D} with a generalized symmetric subsolution in R^n\\E_1, then considers approximating Dirichlet problems on bounded annular domains E_R\\overline{D}. It derives uniform C^0, C^1, C^2 estimates for these approximating solutions and passes to a monotone limit to obtain a smooth solution of the exterior problem. The final section derives the asymptotic expansion near infinity and the derivative estimates.","tokens_in":17259,"tokens_out":19456,"duration_ms":211248,"significance":"If the proof is completed, the result is a genuine extension of the known exterior Hessian equation theory: it replaces the strict convexity condition on D, which is used in all prior works cited in the paper, by the weaker strict (k-1)-convexity plus star-shapedness, and it yields C^∞ regularity rather than merely viscosity regularity. The construction of the subsolution in the non-convex ring, especially the use of the boundary principal curvature in the strict (k-1)-convexity condition, is the main technical novelty and is plausible. The paper also makes concrete use of the linear asymptotic expansion for the solution, following the strategy of Caffarelli–Li [4]. The central claims are significant for the field if the gaps identified below are filled.","major_comments":[{"comment":"The existence of a smooth strictly k-convex solution u_R of (3.1) is asserted at the beginning of Section 3, but no proof or theorem citation is provided. Lemmas 3.1–3.9 are all conditional on the existence of such u_R. No Perron method, continuity argument, or reference to the standard k-convex Dirichlet existence theory (e.g., Caffarelli–Nirenberg–Spruck [5] or Bao–Li–Li [1]) is given. Since the monotone limit in Section 3.4 requires the family {u_R}, this gap is load-bearing. The gap appears repairable by invoking the known solvability theorem with \\bar{u} as a viscosity subsolution and \\bar{u}_R as a supersolution, but as written the existence of the approximating solutions is not established.","section":"Section 3, Eq. (3.1)"},{"comment":"The uniqueness statement in Theorem 1.4 is stated but never proven. Section 3 proves existence of the limiting solution and Section 4 proves the asymptotic behavior; there is no comparison argument showing that two smooth strictly k-convex solutions with the same prescribed quadratic growth must coincide. This is a missing proof for a stated part of the main theorem and must be added.","section":"Theorem 1.4 and Sections 3–4"},{"comment":"Lemma 3.1 uses \\bar{u} as a subsolution in a maximum principle argument, but \\bar{u} is only piecewise smooth and has a jump in normal derivative across ∂E_1. The proof treats the interface with a limiting argument that is not fully justified. Remark 3.2 states that a small modification of the proof of Lemma 3.1 would show \\bar{u} is a viscosity subsolution, but that modification is not supplied. Since \\bar{u} being a viscosity subsolution is a prerequisite both for Lemma 3.1 and for applying standard existence theory to (3.1), this missing proof affects the core existence argument.","section":"Remark 3.2 and Lemma 3.1"},{"comment":"The estimate (4.3) gives |D^mE(x)| ≤ C|x|^{-(2β-2+m)} with β strictly less than n/2 for k<n, because β < k/(2h_k) ≤ n/2 and the case h_k=k/n gives β<n/2. The sentence 'Repeating the argument above by replacing β with n/2 gives Theorem 1.5' is not justified: the construction in Section 2.2 does not allow β=n/2. A more detailed argument is needed to upgrade (4.3) to the stronger decay (1.3), presumably via the linear asymptotic expansion obtained from Lemma 3.6 of [4].","section":"Section 4, Theorem 1.5 derivation"}],"minor_comments":[{"comment":"The symbol φ is used both for the boundary data and for the function b^N in the Claim inside the proof of Proposition 2.1; this overloading makes the proof harder to follow and should be fixed by using a different letter for the power function.","section":"Section 2.1.2"},{"comment":"The formula for μ(α,β) should be parenthesized clearly; as printed, 'µ(α, β) = ∫_1^∞ [(1 + αt^{-β})^{1/k} − 1]dt − 1' is ambiguous and is followed by '< ∞' which is unnecessary and confusing.","section":"Proposition 2.2"},{"comment":"The monotonicity assertion 'when R_1<R_2 ... u_{R_2}<u_{R_1}' needs a brief justification: one must use that \\bar{u}_{R_2}<\\bar{u}_{R_1} on ∂E_{R_1}, together with the comparison principle on Ω_{R_1}.","section":"Section 3.4"},{"comment":"The normalization μ(α,β)=0 is introduced without comment; it should be explained that this shift is absorbed into the constant c in (1.2).","section":"Section 4"},{"comment":"There are several typographical issues, for example 'Futhermore' in Lemma 3.4 and the stray 'r <' in the displayed equation of Lemma 3.4.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper relies substantially on techniques and results from [1], [4], and [11]; [1] has an overlapping author with the present paper and [11] is by the second author. This is not by itself a problem because the main theorem is an extension to a new geometric setting rather than a reduction to those papers. The main issue is completeness: the existence of the approximating bounded-domain solutions, the uniqueness assertion, the viscosity-subsolution property of \\bar{u}, and the final upgrade of the decay estimates all need to be spelled out. These are likely fixable, so the paper is appropriate for a major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take on arXiv:2507.08993. The genuine new content is the relaxation of the domain assumption from strict convexity to strict (k-1)-convexity plus star-shapedness for the exterior Hessian Dirichlet problem, and the upgrade to smooth classical solutions. The construction of the subsolution by gluing \\bar\\phi and \\bar\\phi_1 is intricate and, as far as I can see, correct; the estimates in Section 3 are standard but competently adapted. This is a real extension of Bao-Li-Li, not a repackaged version. Self-citation is not an issue; the paper leans on [1], [4], [11] exactly where it should.\n\nThe soft spots are real but repairable. The existence of the bounded-domain solutions u_R to (3.1) is asserted at the start of Section 3 and never established or referenced. The paper jumps straight to a priori estimates for a solution that has not been shown to exist. The fix is well known: use \\bar u as a viscosity subsolution and \\bar u_R as a supersolution, then Perron plus Caffarelli-Nirenberg-Spruck regularity. The same missing step shows up in Remark 3.2, which defers the viscosity-subsolution property of \\bar u with 'a small modification.' That should be written out or at least cited precisely. Separately, uniqueness in Theorem 1.4 is stated but never proved; the proof section only covers existence. The asymptotic part, Theorem 1.5, is dispatched in one sentence ('repeating the argument with β replaced by n/2') which is terse but plausible.\n\nNone of these are fatal. The main theorem is very likely correct and the geometric relaxation is meaningful. The paper deserves a serious referee, and a referee should ask for the missing existence argument, the uniqueness proof, and a cleaner treatment of the subsolution property in Remark 3.2. I'd read a revised version again. It is worth citing for the k-convex exterior problem. I'd maybe bring it to a PDE reading group, though the details are not exciting for a general audience.","headline":"Genuinely new geometric relaxation to strict (k-1)-convexity, with smooth solutions; the core is correct but the paper omits existence of its bounded-domain approximants and the stated uniqueness.","tokens_in":17845,"tokens_out":2255,"would_cite":true,"duration_ms":26526,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35J60","35J96","35B40","35D40","35B65"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that the exterior Dirichlet problem for Hessian equations has a smooth solution outside a star-shaped strictly (k-1)-convex domain.","keywords":["Hessian equations","exterior Dirichlet problem","non-convex ring","star-shaped domain","strictly (k-1)-convex","viscosity subsolution","asymptotic decay","elliptic PDE"],"falsifier":"Take $n=3$, $k=2$, and a smooth star-shaped domain whose boundary has positive mean curvature except at one flat umbilic point where both principal curvatures vanish (for instance a smoothed flat-capped surface). At that point, inequality (2.9) has $c_1=0$, so the radial subsolution $b^N-1+\\varphi$ should fail to be 2-convex along the normal approach; checking whether any other term restores $k$-convexity would settle whether strictness of $(k-1)$-convexity is genuinely needed.","tokens_in":16760,"feed_emoji":"📐","tokens_out":11374,"duration_ms":117937,"temperature":0.7,"pith_summary":"The paper establishes existence of smooth solutions to the exterior Dirichlet problem for the Hessian equation $\\sigma_k(\\lambda(D^2u))=1$ on $\\mathbb{R}^n\\setminus D$, when the bounded domain $D$ is star-shaped and strictly $(k-1)$-convex rather than convex. Previous results for Hessian equations of this type required strict convexity of the boundary and, except in the Monge–Ampère case, produced only viscosity solutions. The new theorem supplies a strictly $k$-convex solution in $C^\\infty(\\mathbb{R}^n\\setminus D)$ with prescribed quadratic growth at infinity, for every $2\\le k\\le n$. A sympathetic reader should care because this moves the solvability boundary from convex geometry to a milder curvature condition that still leaves enough control for elliptic estimates.","feed_headline":"Non-convex rings admit smooth Hessian solutions","feed_subtitle":"Existence and uniqueness now hold on star-shaped rings whose boundary is only strictly (k-1)-convex, not convex.","key_machinery":"The central object is the radial-graph rescaling $b=r/\\rho(\\theta)$, where $\\partial D$ is written as $r=\\rho(\\theta)$. Its Hessian has an explicit block form in spherical coordinates in terms of the principal curvature matrix of the boundary, and the decisive inequality (2.9) bounds $\\sigma_m(D^2(b^N-1+\\varphi))$ from below by $\\frac{M^{m-1}}{r^{m-1}}(c_1 B/\\rho^2-c_0 M/r)$, where $c_1=\\min\\sigma_{m-1}(\\kappa)>0$ encodes strict $(k-1)$-convexity. This gives the subsolution inside $E_1\\setminus\\bar D$; outside $E_1$ the authors use generalized symmetric subsolutions $\\omega_{\\alpha,\\beta}(s)=\\int_1^s(1+\\alpha t^{-\\beta})^{1/k}dt$, and the two pieces are glued across $\\partial E_1$ by a normal-derivative jump that preserves the viscosity subsolution property.","core_discovery":"Theorem 1.4 states that for $n\\ge 3$, $2\\le k\\le n$, and a bounded, smooth, star-shaped, strictly $(k-1)$-convex domain $D\\subset\\mathbb{R}^n$, given any $A\\in A_k$, any $b\\in\\mathbb{R}^n$, and any $\\varphi\\in C^\\infty(\\partial D)$, there exists $c_*$ such that for every $c>c_*$ there is a unique strictly $k$-convex solution $u\\in C^\\infty(\\mathbb{R}^n\\setminus D)$ of $\\sigma_k(\\lambda(D^2u))=1$ in $\\mathbb{R}^n\\setminus\\bar D$, $u=\\varphi$ on $\\partial D$, with $\\limsup_{|x|\\to\\infty}|x|^{n-2}|u-(\\tfrac12 x^T A x+b\\cdot x+c)|<\\infty$. The solution is obtained as the monotone, locally smooth limit of solutions on bounded ellipsoidal rings $E_R\\setminus\\bar D$, and Theorem 1.5 gives the corresponding decay for all derivatives of the correction term $E=u-(\\tfrac12x^T A x+c)$: $\\limsup_{|x|\\to\\infty}|x|^{n-2+m}|D^m E(x)|<\\infty$ for every $m\\ge 1$.","pith_inferences":["If the same machinery transfers, Hessian quotient equations $\\sigma_k/\\sigma_l=1$ outside strictly $(k-1)$-convex star-shaped domains should be solvable by the same radial-graph construction, since the argument uses only the cone property and Maclaurin inequalities.","The strictness of $(k-1)$-convexity enters only through the positive constant $c_1$; a natural extension would be to allow isolated points where $\\sigma_{k-1}(\\kappa)=0$ and compensate with an extra term in the subsolution.","The threshold $c_*$ is geometric in nature: making it explicit in terms of $|\\varphi|_{C^2}$, the curvature minimum, and the quadratic data would turn the existence criterion into a checkable quantitative one."],"forward_implications":["For $2\\le k<n$, existence no longer demands convexity of $D$: any star-shaped domain whose boundary has positive $(k-1)$-curvature is admissible, so rings with some negative principal curvatures are allowed.","The approximating solutions on $E_R\\setminus\\bar D$ converge smoothly on compact sets, so the limiting exterior solution is classical, not merely a viscosity solution.","The prescribed quadratic asymptotic is sharp: the correction term and all of its derivatives decay at the rates in (1.2)–(1.3), so the solution class with that asymptotic is controlled at infinity.","For each $c>c_*$ the solution is unique; varying $c$ produces a one-parameter family of exterior solutions with the same boundary data and the same quadratic part.","When $k=n$, strict $(k-1)$-convexity reduces to strict convexity, and the theorem recovers the classical exterior Monge–Ampère existence result with boundary smoothness of the solution."],"supporting_citations":[{"why":"Introduces the class $A_k$ and the generalized symmetric subsolutions $\\omega_{\\alpha,\\beta}$ used outside the ellipsoid, including the key estimates (2.11)–(2.16).","marker":"[1]"},{"why":"Supplies the gluing construction across $\\partial E_1$ and the linear elliptic lemma used to obtain the asymptotic decay of the correction term.","marker":"[4]"},{"why":"Provides the boundary $C^2$ estimates and barrier arguments for Hessian Dirichlet problems that control the approximating solutions on both boundaries.","marker":"[5]"},{"why":"Is the source of the spherical-coordinate Hessian computation and the curvature lower bound (2.9) that makes $b^N-1+\\varphi$ a subsolution in $E_1\\setminus\\bar D$.","marker":"[11]"},{"why":"Gives the second fundamental form of a radial graph $\\rho(\\theta)$, which is the starting point for the entire subsolution computation.","marker":"[2]"},{"why":"Is the Schauder theory used to pass from $C^2$ estimates to smooth convergence and to the derivative decay of the correction term.","marker":"[7]"}],"fun_headline_variants":["Existence and uniqueness for Hessian equations on non-convex rings","Star-shaped non-convex domains: smooth Hessian solutions","Hessian exterior problem solved on non-convex rings","Smooth solutions for Hessian equations on star-shaped non-convex rings"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole construction depends on the boundary $\\partial D$ having strictly positive $(k-1)$-curvature, so that $c_1=\\min\\sigma_{k-1}(\\kappa)>0$; if some boundary point has zero $(k-1)$-curvature, the key lower bound (2.9) loses its positive term and the subsolution, and hence the gluing argument, can collapse.","fun_headline_variants_meta":{"raw":{"variants":["Existence and uniqueness for Hessian equations on non-convex rings","Star-shaped non-convex domains: smooth Hessian solutions","Hessian exterior problem solved on non-convex rings","Smooth solutions for Hessian equations on star-shaped non-convex rings"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000894,"raw_usage":{"total_tokens":3821,"prompt_tokens":879,"completion_tokens":2942,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":495,"completion_tokens_details":{"reasoning_tokens":2870}},"tokens_in":495,"tokens_out":2942,"duration_ms":25083,"temperature":1.0,"reasoning_tokens":2870,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:08:54.429754+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $n=3$, $k=2$, and a smooth star-shaped domain whose boundary has positive mean curvature except at one flat umbilic point where both principal curvatures vanish (for instance a smoothed flat-capped surface). At that point, inequality (2.9) has $c_1=0$, so the radial subsolution $b^N-1+\\varphi$ should fail to be 2-convex along the normal approach; checking whether any other term restores $k$-convexity would settle whether strictness of $(k-1)$-convexity is genuinely needed.","supporting_citations":[{"cited_title":"G.; Li, H","cited_arxiv_id":null,"evidence_quote":"Introduces the class $A_k$ and the generalized symmetric subsolutions $\\omega_{\\alpha,\\beta}$ used outside the ellipsoid, including the key estimates (2.11)–(2.16)."},{"cited_title":"Y .An extension to a theorem of J¨orgens, Calabi, and Pogorelov.Comm","cited_arxiv_id":null,"evidence_quote":"Supplies the gluing construction across $\\partial E_1$ and the linear elliptic lemma used to obtain the asymptotic decay of the correction term."},{"cited_title":"The Dirichlet problem for nonlinear second-order elliptic equa- tions","cited_arxiv_id":null,"evidence_quote":"Provides the boundary $C^2$ estimates and barrier arguments for Hessian Dirichlet problems that control the approximating solutions on both boundaries."},{"cited_title":"Generalized Minkowski inequality via degenerate Hessian equations on exterior domains","cited_arxiv_id":"2207.05673","evidence_quote":"Is the source of the spherical-coordinate Hessian computation and the curvature lower bound (2.9) that makes $b^N-1+\\varphi$ a subsolution in $E_1\\setminus\\bar D$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the second fundamental form of a radial graph $\\rho(\\theta)$, which is the starting point for the entire subsolution computation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Is the Schauder theory used to pass from $C^2$ estimates to smooth convergence and to the derivative decay of the correction term."}],"review_version":1}