{"id":"a469d95c-c543-46df-8b05-a862a5bb4ac8","arxiv_id":"2509.00292","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Boundary value problems for the magnetic Laplacian admit uniform nontangential maximal estimates in Lipschitz domains when the magnetic field is of finite type, extending Jerison-Kenig theory.","lead":"A mathematician proves that solutions to the magnetic Schrödinger equation on rough domains satisfy sharp boundary estimates that stay uniform as the semiclassical scale h shrinks. The result extends a classic theorem of Jerison and Kenig for the plain Laplacian to operators with magnetic fields of finite type.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.2, the key m-weighted Poincaré inequality, is imported from an unreviewed preprint [8], and every main estimate depends on it; the d=2 descending step is also only sketched.","rationale":"The reader's CONDITIONAL verdict is essentially correct. The d≥3 proof is a coherent chain, and I did not find an internal contradiction in the estimates once (3.12) is granted. However, the chain is only as secure as Theorem 3.2, which is quoted from an unpublished preprint [8] by the same author. This is not an ad hominem point: it is a reproducibility and verification gap. The manuscript contains no derivation of (3.12), and the proof of the main theorems repeatedly uses it in ways that cannot be bypassed by the other arguments in the paper. I therefore agree that acceptance should be conditional on independent verification of [8, Theorem 3.8] and on the constant being independent of the localization parameters used in §10. I also flag a secondary gap: the d=2 cases, especially Dirichlet descent, are asserted rather than proved, and the H^1 regularity of the lifted boundary data on the product domain is nontrivial at the edges. The finite-type condition (1.4) is correctly identified by the reader as a limitation, but because it is an explicit hypothesis of the theorems rather than an unstated or circular assumption, I do not treat it as the main load-bearing concern. Overall, if [8] holds as stated, the results appear plausible; if it does not, the main theorems lack support. Hence the reader's CONDITIONAL verdict should stand unchanged.","tokens_in":32207,"tokens_out":17201,"duration_ms":207842,"concrete_test":"Inspect arXiv:2505.03690, Theorem 3.8, and verify that its hypotheses match (3.2)-(3.6) on B(0,4R0) and that its constant is independent of R0 and of any extension of A outside Ω; then re-derive (3.12) for the model field B(x)=x^κ on a ball and check uniformity in the scaling β. If [8] assumes a global B_q weight or a vector potential defined on all of R^d, the finite-type localization in §10 is not justified. Separately, complete the d=2 Dirichlet lift by proving F∈H^1(∂O) and the side/cap trace identities used to bound T^hF.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central results rest on inequality (3.12), which is not proved in this paper: Theorem 3.2 says only 'Proof. See [8, Theorem 3.8]'. This is a load-bearing, non-reproduced input. The m-weighted Green/Neumann estimates (5.23)-(5.26), (6.14)-(6.16), the Rellich estimates (7.10), (7.20)-(7.21), and the final nontangential maximal estimates (8.7), (8.13), (9.3), (9.5) all invoke (3.12) or estimates derived from it. If [8, Thm 3.8] requires stronger hypotheses than (3.2)-(3.6), or if its constant depends on R0 or on an extension of A outside Ω, then the localization argument in §10 breaks down. The finite-type condition (1.4) itself is a stated assumption and Lemma 10.1 does supply (3.2)-(3.6) after the β-rescaling, so the unspecified step is precisely the validity of (3.12) under those conditions. Additionally, the d=2 case is dismissed by 'the method of descending' with details omitted; in particular, the lifted Dirichlet data F(x1,x2,x3) is asserted to lie in H^1(∂O) and to satisfy the tangential-boundary identities used in the proof of Theorem 1.2, but this is not demonstrated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves uniform-in-h nontangential maximal estimates for weak H^1 solutions of the Neumann and Dirichlet boundary value problems for the magnetic Laplacian (hD+A)^2u=0 in bounded Lipschitz domains, under the assumption that the magnetic field B=∇×A is of finite type on the closure of the domain. The proof introduces the auxiliary function m(x,B), reduces the problem by rescaling to the operator (D+βA)^2 with β=h^{-1}, and then combines m-weighted Poincaré and Caccioppoli inequalities with Rellich-type identities, Green/Neumann function estimates, and an approximation argument for Lipschitz domains. The case d≥3 is treated in detail; the case d=2 is handled by the method of descent. The main results are Theorems 1.1 and 1.2, together with the derived Dirichlet-to-Neumann norm equivalence in Remark 1.5.","tokens_in":32633,"tokens_out":11132,"duration_ms":132322,"significance":"If the result is correct, it gives a substantial extension of the classical Jerison–Kenig nontangential estimates for harmonic functions to the semiclassical magnetic Laplacian, with constants uniform in h and in the geometry of the domain. The paper is new even for smooth domains and provides a coherent framework of m-weighted Rellich estimates, Green/Neumann bounds, and localization under finite-type conditions. The manuscript is technically rich: the Rellich identities, the smooth-domain estimates, and the approximation argument are carefully structured and are credible for d≥3. The main weakness is that the central m-weighted Poincaré inequality is imported from an unreviewed preprint by the same author, and the d=2 case is only sketched. These are load-bearing for the stated theorems.","major_comments":[{"comment":"The m-weighted Poincaré inequality (3.12) is the first load-bearing input of the paper, but it is not proved here: Theorem 3.2 says only 'Proof. See [8, Theorem 3.8]', where [8] is an unreviewed arXiv preprint by the same author. This inequality is used repeatedly in subsequent arguments: Theorem 3.5 (3.16), the Green/Neumann function estimates (5.23)–(5.26) and (6.14)–(6.16), the Rellich estimates (7.10) and (7.20)–(7.21), the final maximal estimates (8.7), (8.13), (9.3), (9.5), and the localization in Theorem 1.1 (10.13). The short remark after Theorem 3.2 clarifies one point about the curl of B outside Ω, but it does not provide the proof or verify that the hypotheses of [8, Thm 3.8] match (3.2)–(3.6). Because all main claims depend on (3.12), the manuscript is not self-contained at a central point. Please include a complete proof of (3.12) in the paper, or point to a peer-reviewed re","section":"§3, Theorem 3.2, Eq. (3.12)"},{"comment":"The two-dimensional case is dismissed in a few lines by 'the method of descending'. For the Dirichlet problem, the lifted data F on O=Ω×(0,r0) is asserted to lie in H^1(∂O) and to satisfy the tangential-boundary norm identities; for both problems, the boundary norms on the flat lids Ω×{0,r0} are claimed to be controlled by interior norms and by an energy estimate. These are non-obvious steps: the trace of F on the lids depends on the unknown interior values of u, and the required absorption argument needs the trace of (hD+A)u on ∂Ω to be controlled by the nontangential maximal function. The details are omitted, including the verification that the lifted Dirichlet data is in the correct Besov/trace space at the edges. Since the d=2 case is part of the main theorems, these steps must be supplied or the scope of Theorems 1.1–1.2 explicitly restricted to d≥3.","section":"§10, Proofs of Theorems 1.1 and 1.2, d=2 case"},{"comment":"The localization argument in the proof of Theorem 1.1 relies on Lemma 10.1 to obtain the local scaling conditions (3.2)–(3.6) for βB. The proof of Lemma 10.1 is compressed: the transition from the Taylor-polynomial estimate (10.4) to the full doubling condition (10.2) and the lower bound (10.3) for all 0<r<r0 is only sketched ('The general case ... follows by a covering argument'). The covering argument requires the lower bound (10.6) on balls of all relevant radii, including r near r0; this is plausible but not written out. Since the rescaling (10.9) and hence the applicability of Theorem 9.3 depend on this lemma, please expand the proof so that the constants in (10.2)–(10.3) are seen to be uniform in x and r without hidden dependence on r0.","section":"§10, Lemma 10.1 and rescaling step"}],"minor_comments":[{"comment":"Typo/grammar: 'No addition condition beyond A∈C^1(Ω;R^d) is not needed' should read 'No additional condition beyond A∈C^1(Ω;R^d) is needed.'","section":"Remark 1.6"},{"comment":"The statements say the constants depend on 'C0 in (3.3)', but the assumptions are stated with 'C0 in (3.2)'. Since (3.2) implies (3.3), this is not a mathematical issue, but the notation should be aligned.","section":"Theorem 9.3 and Theorem 9.4"},{"comment":"The symbol M(w) is introduced with w(x)=u(x)m(x,B), but in the proof the smoothing function is called em and the function w is redefined as u em. This is a bit confusing; please rename one of the objects.","section":"§8, Lemma 8.1"},{"comment":"After (10.13), the phrase 'using the energy estimate' is terse. The estimate ∫Ω |(D+βA)u|^2 ≤ C∫∂Ω |βg|^2 follows from the weak formulation together with (2.7) and the m-weighted Poincaré inequality, but a one-line derivation would improve readability.","section":"§10, proof of Theorem 1.1"}],"recommendation":"major_revision","confidential_remarks":"The main technical risk is the dependence of the central inequality (3.12) on an unreviewed preprint by the same author. I would suggest asking the author to include a proof of [8, Thm 3.8] in an appendix, or to provide a published reference. The d=2 descending argument also needs fuller justification. These are fixable within the manuscript's scope, so I do not recommend rejection, but the current version is not self-contained enough for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Zhongwei Shen has a genuinely new result here: uniform-in-h nontangential maximal estimates for boundary value problems for the magnetic Laplacian on bounded Lipschitz domains, under a finite-type condition on the magnetic field. That extends the Jerison–Kenig theory in a nontrivial way, and it is new even for smooth domains. The machine is the author's m-function, and the application to boundary value problems is the contribution.\n\nThe paper is well built for d ≥ 3. The Rellich identities, the Green/Neumann function estimates, the duality arguments and the localization under the finite-type condition all hang together. Constants are tracked so they depend only on the domain, the finite-type parameters and ‖B‖_{C^{κ+1}}. The approximation step from smooth to Lipschitz domains is standard but done carefully. I believe the main estimates are very likely correct.\n\nNow the soft spots, in proportion. The biggest one is Theorem 3.2, the m-weighted Poincaré inequality (3.12). It is stated with 'Proof. See [8, Theorem 3.8]' where [8] is an unreviewed preprint by the same author. That inequality is load-bearing: the Green and Neumann function estimates, the Rellich estimates and the final maximal estimates all use it. A referee cannot independently assess the paper as it stands unless [8] is verified. The author does tell us that the proof in [8] only uses B = curl A in Ω, so there is no issue extending A, but that is not enough. The second soft spot is the d = 2 case, which is handled by 'the method of descending' with details omitted. In particular the lifted Dirichlet data on the three-dimensional cylinder is asserted to lie in H^1 and to satisfy the tangential-boundary identities, but that is not shown. That is probably fixable, but as written it is a gap. The finite-type assumption itself is a real restriction—fields with a zero of infinite order at a boundary point are not covered—but that is stated clearly and is not a flaw.\n\nI would send this to referees. The right structure is a conditional acceptance: ask the author to either include a proof of Theorem 3.2 or report its precise hypotheses and whether [8] is under review, and to spell out the d = 2 descending argument. The work is citable, and I'd bring it to a reading group.","headline":"Solid new extension of Jerison–Kenig to the magnetic Laplacian with uniform h estimates, but it leans on an unreviewed preprint inequality and a sketchy d=2 case.","tokens_in":33033,"tokens_out":2566,"would_cite":true,"duration_ms":28524,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35P25"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves uniform nontangential maximal estimates for the semiclassical magnetic Laplacian under a finite-type magnetic field condition, extending the classical Lipschitz-domain result for the Laplacian.","keywords":["magnetic Laplacian","semiclassical analysis","nontangential maximal function","Lipschitz domain","finite-type magnetic field","Neumann problem","Dirichlet problem","Rellich identities"],"falsifier":"Check the quoted proof of (3.12) for the model field B(x)=x_1^κ in a ball: if the inequality fails for a sequence ψ oscillating at scale m(x,B)^{-1}, the central chain breaks. Alternatively, test numerically whether the constants in (1.6) remain bounded as h→0 for a field with a zero of order κ+1 at a boundary point; a blow-up would show the finite-type threshold is sharp.","tokens_in":32127,"feed_emoji":"🧲","tokens_out":8289,"duration_ms":94113,"temperature":0.7,"pith_summary":"The paper's goal is a boundary regularity theorem for the magnetic Schrödinger operator (hD+A)^2 that is uniform in the semiclassical parameter h. For a bounded Lipschitz domain whose magnetic field B=∇×A is of finite type on the closed domain, the author proves that weak H^1 solutions of the Neumann and Dirichlet problems satisfy nontangential maximal estimates for the combined quantity |(hD+A)u|+h m(x,h^{-1}B)|u|, with constants independent of h. Such estimates were known for the plain Laplacian in Lipschitz domains, and the paper transfers them to the magnetic setting. The result matters because uniform boundary estimates are the kind of tool needed to study spectral and resolvent behavior of magnetic Schrödinger operators as h→0. The estimates are claimed to be new even for smooth domains.","feed_headline":"Magnetic Laplacian gains uniform boundary estimates under finite-type fields","feed_subtitle":"L² control of solutions on Lipschitz domains holds with constants independent of h—new even in the smooth case.","key_machinery":"The central object is m(x,B), the largest radius r such that |B|≤r^{-2} on B(x,r). It serves as the local magnetic scale; m(x,h^{-1}B) replaces h^{1/2} or h as the natural small parameter in the estimates. The chain of proof uses an m-weighted Poincaré inequality (3.12), boundary Rellich identities adapted to the magnetic operator, interior pointwise bounds for v_h, and Green/Neumann function estimates with decay (1+|x-y|m(x,B))^{-ℓ}. Together they convert boundary-data control into nontangential control of v_h, uniformly in h.","core_discovery":"Under the assumption that B=∇×A is of finite type on the closure of a bounded Lipschitz domain—meaning that some finite collection of derivatives of B cannot all vanish at the same point—the paper proves that weak H^1 solutions of the semiclassical magnetic Neumann and Dirichlet problems satisfy uniform non-tangential maximal estimates. The combined quantity v_h=|(hD+A)u|+h m(x,h^{-1}B)|u| is controlled in L^2(∂Ω) by the boundary data alone, with constants independent of h∈(0,h0). The proof reduces the semiclassical parameter to a rescaling of the potential, localizes near the boundary on the scale m(x,h^{-1}B)^{-1}, and uses Rellich identities plus m-weighted bounds on the Green and Neumann","pith_inferences":["An implicit corollary is that the m scale, not h^{1/2}, is the natural semiclassical length near the boundary: the estimates are unchanged when the field is rescaled, so h enters only through β=h^{-1}.","A likely extension is L^p versions of (1.6)-(1.9) for p in a range around 2; the paper proves only L^2, but the machinery of maximal functions and Rellich identities is standard for such extrapolation.","The finite-type threshold appears necessary: for fields with infinite-order zeros the m scale can be exponentially small, so uniform h-independent control should fail; testing this would sharpen the boundary of the theorem.","Through the Dirichlet-to-Neumann equivalence, these estimates may feed into spectral asymptotics or resolvent estimates for magnetic Schrödinger operators in domains, where uniform boundary control as h→0 is a recurring need."],"forward_implications":["For h below a threshold h0, the L^2 norm of the nontangential maximal function of v_h is controlled by the boundary data, with no h-dependent loss.","The Dirichlet-to-Neumann map Λ_h satisfies ||Λ_h f|| ≈ ||T^h f|| + h||m(x,h^{-1}B)f||, a magnetic analogue of the classical boundary equivalence.","The estimates hold on Lipschitz domains, not just smooth ones, and are new in the smooth case as well.","The two-dimensional case is obtained from three dimensions by a cylindrical extension, the method of descending.","The finite-type hypothesis is quantitative: constants depend only on Ω, the pair (κ,c0), and ||B||_{C^{κ+1}(Ω)}."],"supporting_citations":[{"why":"Supplies the classical Laplacian nontangential estimates in Lipschitz domains that the paper extends to the magnetic operator.","marker":"[4, 5]"},{"why":"Provides the m-weighted Poincaré inequality (3.12) on which the whole chain of estimates rests.","marker":"[8]"},{"why":"Introduces the auxiliary function m(x,B) and supplies its comparison and Hölder-type inequalities used throughout.","marker":"[9, 10]"},{"why":"Gives harmonic-function nontangential maximal estimates used to control the maximal function of the solution from its boundary values.","marker":"[6]"},{"why":"Supplies the approximation of a Lipschitz domain by smooth domains with uniform Lipschitz character used in Section 9.","marker":"[13]"},{"why":"Underpins the construction and pointwise bounds of the Green function for the magnetic operator in divergence form.","marker":"[1, 3]"},{"why":"Provides a smooth substitute of m(x,B) with |∇m|≤Cm^2, used in the maximal-function estimate of Lemma 8.1.","marker":"[11]"}],"fun_headline_variants":["Uniform boundary control for magnetic Laplacian on Lipschitz domains","h-independent estimates for magnetic Laplacian: new even for smooth domains","Semiclassical magnetic Laplacian: boundary estimates independent of h","Finite-type fields make magnetic Laplacian boundary bounds uniform in h"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The result collapses if the finite-type condition (1.4) fails at some point of the closed domain, or if the m-weighted Poincaré inequality (3.12), quoted from the author's unpublished preprint [8], does not hold as stated.","fun_headline_variants_meta":{"raw":{"variants":["Uniform boundary control for magnetic Laplacian on Lipschitz domains","h-independent estimates for magnetic Laplacian: new even for smooth domains","Semiclassical magnetic Laplacian: boundary estimates independent of h","Finite-type fields make magnetic Laplacian boundary bounds uniform in h"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000238,"raw_usage":{"total_tokens":1334,"prompt_tokens":715,"completion_tokens":619,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":459,"completion_tokens_details":{"reasoning_tokens":553}},"tokens_in":459,"tokens_out":619,"duration_ms":6465,"temperature":1.0,"reasoning_tokens":553,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T13:44:58.846134+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check the quoted proof of (3.12) for the model field B(x)=x_1^κ in a ball: if the inequality fails for a sequence ψ oscillating at scale m(x,B)^{-1}, the central chain breaks. Alternatively, test numerically whether the constants in (1.6) remain bounded as h→0 for a field with a zero of order κ+1 at a boundary point; a blow-up would show the finite-type threshold is sharp.","supporting_citations":[{"cited_title":"Kenig, Harmonic analysis techniques for second order elliptic boundary value prob- lems, CBMS Regional Conference Series in Mathematics, vol","cited_arxiv_id":null,"evidence_quote":"Gives harmonic-function nontangential maximal estimates used to control the maximal function of the solution from its boundary values."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the approximation of a Lipschitz domain by smooth domains with uniform Lipschitz character used in Section 9."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides a smooth substitute of m(x,B) with |∇m|≤Cm^2, used in the maximal-function estimate of Lemma 8.1."}],"review_version":1}