{"id":"8d3c0d2d-5312-446d-aac7-21fb2ad3a7fc","arxiv_id":"2507.16402","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For quasilinear equations -div(a(|∇u|)∇u)=f, the stress field a(|∇u|)^k∇u is shown to belong to W^{1,2}(Ω) up to the boundary for the optimal range of k, under weak W^{2,X} boundary regularity, with a convex-domain version requiring no boundary regularity.","lead":"This paper proves global second-order regularity for solutions of a broad class of quasilinear elliptic equations, including degenerate p-Laplacian type operators, under both Dirichlet and Neumann boundary conditions. The results extend known sharp estimates from smooth domains to domains whose boundary curvature is integrable only in certain weak Lorentz spaces, and add integrability estimates for the inverse gradient when the source term has a sign.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The approximation step in Theorem 1.1 needs uniform control of strong curvature norms that Lemma 4.2 does not provide; without it, the passage from smooth to W^{2,X} domains is incomplete.","rationale":"The reader's conditional verdict is reasonable: the displayed inequalities (5.63) and (5.68) appear reversed and (4.42) contains a sign typo, so a careful correction pass is needed. My stress-test identifies an additional, more structural gap in the approximation argument. The proof of Theorem 1.1 proceeds through three approximation layers: smooth data on a smooth domain, smooth domains approximating the W^{2,X} domain, and finally general f. The second layer needs estimates that are uniform in m. The constants in the preceding estimates depend on ||tr B||_{L^{n-1,1}} and on Ψ_Ω, while the stated approximation controls only K_{Ω_m} and the weak Lorentz norms. This is not a known consequence of (4.52), because capacity-quotient bounds do not control stronger pointwise or strong-Lorentz norms of the second fundamental form. If the missing preservation is supplied, the argument can likely be completed; the issue is therefore a proof gap rather than evidence against the theorem. The reader's weakest assumption about the smallness of K_Ω is related but distinct: it concerns the absorption in Theorem 4.1, whereas my concern concerns whether the same smallness, plus the strong curvature norms, is uniform across the approximating sequence. For this reason I record partial agreement and leave the verdict unchanged at CONDITIONAL.","tokens_in":31867,"tokens_out":20338,"duration_ms":223561,"concrete_test":"Verify whether the approximation lemma actually supplies uniform control of the strong curvature norms: for Ω with ∂Ω ∈ W^{2,L^{n-1,1}} (n ≥ 3; L log L for n = 2), take the smooth domains Ω_m from Lemma 4.2 and check whether sup_m ||B_m||_{L^{n-1,1}(∂Ω_m)} is finite. If this is not a consequence of (4.51)-(4.52) and the cited result in [21], then Step 2 of Theorem 1.1 lacks a necessary uniform estimate; either supply such a preservation lemma or replace (5.55)/(5.75) with bounds depending only on K_Ω and Ψ_Ω.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing gap is in Step 2 of the proof of Theorem 1.1 (and in the analogous Step 2 of Theorem 1.7). The uniform W^{1,2} bound (5.80) for the approximating solutions u_m is quoted from (5.75), but the constant in (5.75) contains the strong boundary norms ||tr B||_{L^{n-1,1}(Ω)} (n ≥ 3) and the L log L norm (n = 2) via the global Lipschitz estimate (5.55). The authors assert that uniformity in m is guaranteed by properties (4.51) and (4.52), yet those properties only control the Lipschitz constants, diameters, and the capacity quotient K_{Ω_m}(r) ≤ C K_Ω(r). They do not imply uniform bounds on the stronger norms ||tr B_m||_{L^{n-1,1}(∂Ω_m)} or ||tr B_m||_{L log L(∂Ω_m)}. The only approximation statement actually cited ([21]) preserves the weaker L^{n-1,∞} and L^{1,∞} log L norms. If these stronger norms are not preserved, the constant in (5.80) may depend on m, and the subsequential limit argument does not produce a finite bound on the original domain. This is a missing link in the main approximation argument, not a mere artifact of the boundary regularity condition.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves global second-order regularity for the stress field a(|∇u|)^k ∇u for a class of quasilinear elliptic equations with homogeneous Dirichlet or Neumann boundary conditions. The main result, Theorem 1.1, asserts that a(|∇u|)^k ∇u ∈ W^{1,2}(Ω) for a range of exponents k determined by the growth indices ia, sa of a, provided the boundary belongs to a Lorentz-Zygmund class W^{2,X} with X = L^{n-1,1} (n≥3) or X = L log L (n=2). The proof is built on a global integral inequality (Theorem 4.1) for smooth approximating problems, with boundary curvature terms absorbed through the capacity quotient K_Ω(r). A sign condition on the source term yields integrability of the inverse of the gradient (Theorem 1.7), and convex domains are treated without boundary regularity assumptions (Theorems 1.6 and 1.8).","tokens_in":32102,"tokens_out":25484,"duration_ms":255408,"significance":"If the proof is completed, the results are significant: they extend sharp second-order boundary regularity for the p-Laplacian to general quasilinear operators with (p-1)-type growth, recover the Cianchi-Maz'ya stress-regularity theorem as the case k=1, and provide new information on the inverse of the gradient. The algebraic core in Theorem 4.1 is carefully presented, and the lower bound (4.39) for the quadratic form is sound. The convex-domain results are attractive because they require no boundary regularity. The main caveat is that the approximation step from smooth to W^{2,X} domains is not fully justified, as explained below.","major_comments":[{"comment":"The uniform-in-m bound (5.80) is not established. The constant in (5.75) depends, through the global Lipschitz estimate (5.55), on the strong boundary norms ∥tr B∥_{L^{n-1,1}(∂Ω)} for n≥3 and the L log L norm for n=2. Lemma 4.2 only provides (4.51)-(4.52), namely uniform Lipschitz characteristics, diameters, and the capacity quotient K_{Ω_m}(r) ≤ C K_Ω(r). The approximation property explicitly invoked in Step 2 preserves only the weaker L^{n-1,∞} and L^{1,∞} log L norms. These controlled quantities do not imply uniform bounds on the strong L^{n-1,1} or L log L norms of tr B_m, so the constant in (5.80) may depend on m and the subsequential limit argument does not yield a finite bound on Ω. The authors need to supply an approximation lemma preserving the strong curvature norms, or replace (5.55) by a global Lipschitz estimate whose constant is controlled by the quantities supplied by Lemma 4.2.","section":"§5, Step 2 of Theorem 1.1 (and §6, Step 2 of Theorem 1.7)"},{"comment":"The displayed inequalities (5.63) and (5.68) are incorrect as written. In Case 1, where 0 ≤ ia < sa, condition (3.23) implies aε(t)^γ ≤ a(1)^γ (1+∥∇u∥_{L∞}^2)^{saγ/2} for γ ≥ 0; it does not give the stated lower bound with exponent ia, and the direction of the inequality is reversed. The same problem occurs in Case 2 at (5.68) for γ ≤ 0. The desired estimates (5.65) and (5.70) are nevertheless true and follow from the corrected upper bound on aε^γ, so this is a local but necessary correction in the proof of the central estimate.","section":"§5, Eqs. (5.63) and (5.68)"}],"minor_comments":[{"comment":"The denominator in the two displayed formulas of (4.42) should be (ε + |∇u|²)^{α/2}, not (ε − |∇u|²)^{α/2}.","section":"§4, Eq. (4.42)"},{"comment":"In the statement of Lemma 6.1, u is a scalar function, not a vector field; the wording should be corrected.","section":"§6, Lemma 6.1"},{"comment":"The sentence introducing the approximation properties says 'see [21]' but the actual approximation lemma used is Lemma 4.2, which is attributed to [1]; the attribution should be made precise to avoid confusion.","section":"§5, Step 2 of Theorem 1.1"},{"comment":"The interval for β in (1.11) in the mixed case ia < 0 < sa is displayed with parentheses that may be read as open endpoints; please clarify which endpoints are included, especially since the subsequent estimates use closed intervals in places.","section":"§1, Theorem 1.7"}],"recommendation":"major_revision","confidential_remarks":"The principal obstacle is the approximation step: the constant in (5.80) is not shown to be independent of m. If the authors can prove an approximation lemma that preserves the strong L^{n-1,1} (or L log L) curvature norms, or can replace the global Lipschitz estimate with one whose constant is controlled by the quantities in Lemma 4.2, the argument appears sound. The reversed inequalities in (5.63) and (5.68) are local and easily fixed. I would not recommend rejection, but the revision must close the approximation gap."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper proves genuinely new results: for the class -div(a(|∇u|)∇u)=f, with a satisfying (1.4), it shows a(|∇u|)^k∇u ∈ W^{1,2}(Ω) under boundary regularity only W^{2,X} (X=L^{n-1,1} or L log L), and also gives inverse-gradient integrability under a sign condition. The p-Laplace case falls out with the sharp range of k, and the convex-domain version with no boundary regularity is a nice addition. The central global inequality (Theorem 4.1) is proved in detail, and the algebraic lower bound (4.39) checks out.\n\nThe soft spots are real but different in severity. First, several displayed inequalities look reversed: (5.63) and (5.68) in the proof of Theorem 1.1. These appear to be typos—the subsequent argument needs the opposite inequality—but they make the text hard to verify. There is also a sign typo in (4.42). These are fixable.\n\nSecond, and more seriously, the approximation step in Step 2 of Theorem 1.1 (and the analogous step in Theorem 1.7) has a missing uniform estimate. The constant in the key bound (5.75) depends on the strong boundary norm ||tr B||_{L^{n-1,1}(∂Ω)} (or L log L if n=2), through the global Lipschitz estimate (5.55). For the approximating domains Ω_m, the authors need a uniform bound on ||tr B_m||_{L^{n-1,1}(∂Ω_m)}. Lemma 4.2 only gives control of the Lipschitz character, the diameter, and the capacity quotient K_{Ω_m}(r) ≤ C K_Ω(r). It does not give uniform control of the strong norms, and the cited approximation result from [21] preserves only the weaker L^{n-1,∞} and L^{1,∞} log L norms. The note after (5.80) says uniformity follows from (4.51)–(4.52), but those properties do not bound the strong curvature norms. Without that bound, the constant in (5.80) may depend on m, and the subsequential limit argument does not produce a finite bound on Ω. This is a gap in the main theorem, not a cosmetic issue. The stress-test note is correct on this point, and I checked the relevant parts of the manuscript.\n\nWho is this for? Regularity theorists working on quasilinear elliptic equations and boundary estimates. It deserves a serious referee—the main idea is good and the result is likely true—but the referee should insist on either a corrected approximation argument or a version of the Lipschitz estimate that depends only on weak-type norms. I would send it to peer review, but expect the authors to have to do substantial work.","headline":"A genuinely new extension of second-order boundary regularity for quasilinear elliptic equations, but the approximation argument has a missing uniform curvature-norm estimate that needs fixing.","tokens_in":32715,"tokens_out":9133,"would_cite":false,"duration_ms":79879,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B65","35J25","35J62"],"pacs":[],"model":"deepseek-v4-flash","headline":"Quasilinear stress fields gain a square-integrable derivative up to the boundary, provided the boundary curvature lies in a sharp borderline space, with convex domains needing no smoothness at all.","keywords":["quasilinear elliptic equations","second-order regularity","boundary estimates","Orlicz-Sobolev spaces","p-Laplace equation","second fundamental form","Lorentz-Zygmund spaces","convex domains"],"falsifier":"Compute $K_\\Omega(r)$ for a domain whose boundary second fundamental form lies in $L^{n-1}$ but not in $L^{n-1,1}$ (for $n=2$, in $L^1$ but not $L\\log L$); if $\\limsup_{r\\to0^+}K_\\Omega(r)$ is not below the threshold, solve the $p$-Laplace equation with a smooth nonzero right-hand side and test whether $|\\nabla u|^{p-2+k}\\nabla u\\in W^{1,2}(\\Omega)$ for the exponents $k$ in (1.9). A failure at the claimed exponent, or a check that the boundary term in (4.46) does not stay absorbable, would refute the theorem's boundary hypotheses; conversely, confirming the estimate on such borderline domains would support the sharpness of the Lorentz-Zygmund condition.","tokens_in":2411,"feed_emoji":"📐","tokens_out":2739,"duration_ms":103332,"temperature":0.7,"pith_summary":"This paper establishes global second-order regularity up to the boundary for solutions of $-\\operatorname{div}(a(|\\nabla u|)\\nabla u)=f$ with either zero Dirichlet or zero Neumann data, for any coefficient $a\\in C^1(0,\\infty)$ whose logarithmic derivative stays above $-1$ and is bounded above. The regularity is expressed on the flux object $a(|\\nabla u|)^k\\nabla u$: it has a square-integrable weak derivative for a range of exponents $k$ that depends on the growth indices of $a$. The boundary condition needed is that the weak second fundamental form of $\\partial\\Omega$ lies in a borderline Lorentz-Zygmund space, $L^{n-1,1}$ for $n\\ge 3$ and $L\\log L$ for $n=2$, together with a smallness condition derived from it. In the model case $a(t)=t^{p-2}$, the result recovers the sharp boundary regularity known for the $p$-Laplacian. If the domain is merely convex, the same conclusion holds with no boundary regularity assumption at all.","feed_headline":"Quasilinear stress fields gain a derivative at the boundary","feed_subtitle":"Boundary curvature in a sharp borderline space is enough, and convex domains need no smoothness at all.","key_machinery":"The central device is a global weighted Hessian estimate for the regularized equation, proved for smooth domains and smooth solutions and then transferred to weak solutions by approximation. The estimate controls $\\int_\\Omega \\frac{a_\\varepsilon(|\\nabla u|)}{(\\varepsilon+|\\nabla u|^2)^{\\alpha/2}}|D^2u|^2\\,dx$ up to boundary terms in which the second fundamental form $B$ of $\\partial\\Omega$ appears through identity (4.44); the trace estimate (3.30) bounds these boundary terms by the capacity quotient $K_\\Omega(r)$, so the smallness condition $\\limsup_{r\\to0^+}K_\\Omega(r)<c$ is exactly what makes the boundary integrals absorbable. A pointwise linear-algebra lemma (Lemma 3.7) supplies the ellipticity of the quadratic form in the Hessian, and the approximation family $a_\\varepsilon$ from Lemma 3.1 preserves the growth indices $i_a,s_a$ while regularizing the equation. The resulting uniform bounds on $a_\\varepsilon(|\\nabla u_\\varepsilon|)^k\\nabla u_\\varepsilon$ in $W^{1,2}$ pass to the limit through compactness and almost-everywhere convergence of gradients.","core_discovery":"On a bounded Lipschitz domain whose boundary has weak second derivatives in the Lorentz-Zygmund space $X$ with $X=L^{n-1,1}$ for $n\\ge 3$ and $X=L\\log L$ for $n=2$, every weak solution of the Dirichlet or Neumann problem satisfies $a(|\\nabla u|)^k\\nabla u\\in W^{1,2}(\\Omega)$ for every $k$ in the intervals (1.9), whenever $f\\in W^{1,1}(\\Omega)\\cap L(\\Omega)$. This is a global, up-to-the-boundary result for the stress field. In the special case $a(t)=t^{p-2}$ it gives the optimal $p$-Laplace boundary regularity; for $k=1$ it reproduces the known characterization $a(|\\nabla u|)\\nabla u\\in W^{1,2}$ under the stated source condition. For bounded convex domains, the sign of the boundary curvature replaces the integrability condition and the same flux regularity follows without any regularity assumption on $\\partial\\Omega$ (Theorem 1.6). When the source term has a definite local sign, the inverse weight $1/a(|\\nabla u|)$ is locally integrable to the powers listed in (1.11), yielding $u\\in W^{2,q}(\\Omega)$ for $1\\le q<(s_a+1)/s_a$ in the degenerate regime.","pith_inferences":["Because the argument needs only the smallness of $K_\\Omega(r)$, the geometric condition could be checked numerically from boundary curvature data alone; a domain whose curvature concentrates enough to violate the smallness condition would be a candidate counterexample.","The convex-domain result suggests that the real mechanism is curvature sign rather than curvature magnitude; one might expect analogues for other second-order elliptic problems where a favourable curvature sign is available, though the paper does not state this.","The borderline $L^{n-1,1}$/$L\\log L$ scale is sharp for the method, and constructing examples with curvature in $L^{n-1}$ but not $L^{n-1,1}$ would test whether the integrability assumption is necessary.","The inverse-gradient bounds are likely to interact with fine properties of the zero set of $\\nabla u$; they could be used to quantify the size of the singular set in degenerate problems."],"forward_implications":["For the $p$-Laplace operator, $a(t)=t^{p-2}$, the theorem gives exactly the optimal range of exponents $k$ for which $|\\nabla u|^{p-2+k}\\nabla u\\in W^{1,2}(\\Omega)$, matching the known sharp boundary result and its counterexamples.","Taking $k=1$ recovers, under the paper's domain and source assumptions, the known result that the stress field $a(|\\nabla u|)\\nabla u$ lies in $W^{1,2}(\\Omega)$ when $f\\in L^2(\\Omega)$.","On any bounded convex domain, the conclusion holds with no boundary regularity assumption, so convexity alone guarantees the second-order flux estimate for both Dirichlet and Neumann problems.","If the source term has a fixed sign in a neighbourhood, the inverse weight $1/(a(|\\nabla u|))^\\beta$ is integrable for the exponents $\\beta$ in (1.11); consequently, when $a(t)$ vanishes at $t=0$ and $s_a\\ge 1$, the solution lies in $W^{2,q}(\\Omega)$ for every $q<(s_a+1)/s_a$.","The boundary-regularity threshold is scale-critical: it requires $L^{n-1,1}$ or $L\\log L$, the borderline Lorentz refinement of the natural $L^{n-1}$ integrability of the second fundamental form."],"supporting_citations":[{"why":"Establishes the baseline that $a(|\\nabla u|)\\nabla u\\in W^{1,2}(\\Omega)$ holds when $f\\in L^2(\\Omega)$; this is the result the paper sharpens.","marker":"[21]"},{"why":"Supplies the sharp p-Laplace boundary regularity that Theorem 1.1 recovers as the special case $a(t)=t^{p-2}$.","marker":"[44]"},{"why":"Provides global Lipschitz estimates for the gradient and the regularized approximation $a_\\varepsilon$ used throughout the proof.","marker":"[19]"},{"why":"Gives the optimal second-order regularity framework for the p-Laplace system and the technical lemmas the proof builds on.","marker":"[22]"},{"why":"Is the vectorial p-Laplacian counterpart whose approach the paper extends to the general class of operators.","marker":"[56]"},{"why":"Supplies global second-order estimates and the capacity/trace machinery used to control boundary terms in anisotropic problems.","marker":"[3]"},{"why":"Provides the pointwise differential inequality, Lemma 3.7, that keeps the Hessian quadratic form elliptic.","marker":"[7]"},{"why":"Supplies the smooth approximation of Lipschitz domains with controlled weak curvatures and capacity estimates used in the regularization step.","marker":"[1]"}],"fun_headline_variants":["Quasilinear stress gains boundary regularity from curvature alone","Convex domains bypass smoothness in quasilinear boundary estimates","Sharp second-order boundary estimates for quasilinear equations","Boundary curvature alone yields optimal p-Laplace regularity","Second-order flux regularity up to the boundary via weak curvature"],"cache_read_input_tokens":34688,"weakest_assumption_plain":"The load-bearing premise is that the boundary curvature is small at small scales, in the precise sense that the capacity quotient $K_\\Omega(r)$ tends below a fixed constant as $r\\to0^+$; this is inherited from $\\partial\\Omega\\in W^{2,X}$, and if it fails the boundary integrals in the proof cannot be absorbed.","fun_headline_variants_meta":{"raw":{"variants":["Quasilinear stress gains boundary regularity from curvature alone","Convex domains bypass smoothness in quasilinear boundary estimates","Sharp second-order boundary estimates for quasilinear equations","Boundary curvature alone yields optimal p-Laplace regularity","Second-order flux regularity up to the boundary via weak curvature"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001144,"raw_usage":{"total_tokens":4723,"prompt_tokens":899,"completion_tokens":3824,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":515,"completion_tokens_details":{"reasoning_tokens":3747}},"tokens_in":515,"tokens_out":3824,"duration_ms":28046,"temperature":1.0,"reasoning_tokens":3747,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:11:14.842898+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $K_\\Omega(r)$ for a domain whose boundary second fundamental form lies in $L^{n-1}$ but not in $L^{n-1,1}$ (for $n=2$, in $L^1$ but not $L\\log L$); if $\\limsup_{r\\to0^+}K_\\Omega(r)$ is not below the threshold, solve the $p$-Laplace equation with a smooth nonzero right-hand side and test whether $|\\nabla u|^{p-2+k}\\nabla u\\in W^{1,2}(\\Omega)$ for the exponents $k$ in (1.9). A failure at the claimed exponent, or a check that the boundary term in (4.46) does not stay absorbable, would refute the theorem's boundary hypotheses; conversely, confirming the estimate on such borderline domains would support the sharpness of the Lorentz-Zygmund condition.","supporting_citations":[{"cited_title":"Cianchi, V .G","cited_arxiv_id":null,"evidence_quote":"Establishes the baseline that $a(|\\nabla u|)\\nabla u\\in W^{1,2}(\\Omega)$ holds when $f\\in L^2(\\Omega)$; this is the result the paper sharpens."},{"cited_title":"Montoro, L","cited_arxiv_id":null,"evidence_quote":"Supplies the sharp p-Laplace boundary regularity that Theorem 1.1 recovers as the special case $a(t)=t^{p-2}$."},{"cited_title":"Cianchi, V","cited_arxiv_id":null,"evidence_quote":"Provides global Lipschitz estimates for the gradient and the regularized approximation $a_\\varepsilon$ used throughout the proof."},{"cited_title":"Cianchi, V","cited_arxiv_id":null,"evidence_quote":"Gives the optimal second-order regularity framework for the p-Laplace system and the technical lemmas the proof builds on."},{"cited_title":"Global second order optimal regularity for the vectorial $p$-Laplacian","cited_arxiv_id":"2502.17067","evidence_quote":"Is the vectorial p-Laplacian counterpart whose approach the paper extends to the general class of operators."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies global second-order estimates and the capacity/trace machinery used to control boundary terms in anisotropic problems."},{"cited_title":"Balci, A","cited_arxiv_id":null,"evidence_quote":"Provides the pointwise differential inequality, Lemma 3.7, that keeps the Hessian quadratic form elliptic."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the smooth approximation of Lipschitz domains with controlled weak curvatures and capacity estimates used in the regularization step."}],"review_version":1}