{"id":"0c9f448a-276f-40b2-b9ba-5e8a70b5bafd","arxiv_id":"1908.02268","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In 1-sided chord-arc domains, Carleson measure estimates for bounded solutions of real elliptic equations are equivalent to the elliptic measure being an A∞ weight, yielding new perturbation and transpose-invariance results.","lead":"This paper proves that for elliptic equations in rough \"1-sided chord-arc\" domains, the elliptic measure belongs to the Muckenhoupt class A∞ if and only if all bounded solutions satisfy a Carleson measure estimate. It also shows that this property is preserved under perturbation of the coefficients and passes between an operator and its transpose under Carleson conditions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.1 rests on unpublished [HMT1] boundary estimates (Lemmas 2.16–2.24); acceptance should remain conditional on their verification.","rationale":"The reader's weakest assumption correctly identifies the unpublished [HMT1] estimates as the load-bearing foundation of the paper. My independent reading confirms that Lemma 2.24 is cited, not proved, and that its parts are used at multiple essential points in the proof of Theorem 1.1. The dependence is structural: without Lemma 2.24(a), the lower bound in Lemma 3.10 collapses; without Lemma 2.24(b)–(c), the bounded-solution-to-Carleson direction cannot replace δ by the Green function; without Lemma 2.24(c)–(d), the dyadic A∞ condition cannot be transferred to the continuous elliptic-measure A∞ definition. I found no internal inconsistency in the arguments conditional on those lemmas, and the perturbation theorems are coherent reductions to Theorem 1.1. The appropriate verdict is therefore the same conditional acceptance: the central claims are plausible and well supported once the HMT1 estimates are verified, but full verification is currently blocked by an unavailable reference. The paper gives real evidence through detailed proofs and reductions, but the missing foundation prevents unconditional acceptance.","tokens_in":36320,"tokens_out":9197,"duration_ms":98462,"concrete_test":"Obtain the [HMT1] manuscript and have an independent expert verify Lemmas 2.16, 2.17, and 2.24 in the stated generality: real, non-symmetric, bounded measurable elliptic operators in 1-sided chord-arc domains. As a focused analytical check, re-derive Lemma 2.24(b) (the comparison r^{n-1}G_L(X,X_Δ) ≈ ω_L^X(Δ)) without invoking [HMT1], using only published capacity-density and boundary-Harnack results. If [HMT1] cannot be produced or this re-derivation requires extra assumptions such as exterior corkscrew or Lipschitz coefficients, then Theorem 1.1 is not established in the generality claimed and the paper should add a self-contained appendix or downgrade the claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central equivalence of Theorem 1.1 is built on Lemma 2.24, whose four parts are stated with the note \"The proofs ... may be found in [HMT1]\" (Section 2.4), and [HMT1] is listed as \"work in progress, 2014\". Parts (a)–(d) supply the boundary Hölder decay, Green-function/elliptic-measure comparison, doubling, and boundary comparison for real non-symmetric operators. These estimates are used essentially throughout: (3.18) and (3.21) in Lemma 3.10 need Lemma 2.24(a) and Lemma 2.16; the dyadic-to-continuous A∞ transfer in (3.15) needs Lemma 2.24(c)–(d); the proof of (b)=>(a) needs Lemma 2.24(b)–(c) and Lemma 2.17 at (3.55). If any of these quantitative estimates fails for non-symmetric L in a 1-sided CAD, the equivalence (a)⇔(b) has no proof. Lemma 2.16 is cited to Bourgain/HKM/Zhao, but the non-symmetric, Ahlfors-regular-boundary version is also attributed to [HMT1]. This is a foundational dependency on an unavailable manuscript, not a circularity, and the paper does not reproduce the lemmas. Theorems 1.3 and 1.6 reduce to Theorem 1.1 and Proposition 4.18, so they inherit the same dependency.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies real (not necessarily symmetric) uniformly elliptic divergence-form operators in 1-sided chord-arc domains. Theorem 1.1 asserts the equivalence between (a) Carleson measure estimates for the gradients of all bounded weak solutions and (b) membership of the associated elliptic measure in the Muckenhoupt class A∞ with respect to surface measure. The forward direction is proved by a new dyadic good-cover argument; the reverse direction is proved by a reduction to discrete Carleson measures using sawtooths and published lemmas from [HMT2]. Theorems 1.3 and 1.6 are derived as consequences of a more general perturbation result, Theorem 4.13, which gives the preservation of A∞ under Carleson perturbations of the coefficients and under transposition when the divergence of the antisymmetric part satisfies a Carleson condition. Corollary 1.9 then removes an assumption in [HMT2].","tokens_in":36509,"tokens_out":11619,"duration_ms":116259,"significance":"If Theorem 1.1 is fully justified, this is a substantial extension of the Kenig-Kirchheim-Pipher-Toro result to 1-sided chord-arc domains and to non-symmetric operators, with the equivalence of Carleson estimates and A∞ as the central tool. The proof strategy is genuinely novel: the dyadic good-cover construction in Lemma 3.10 and the sawtooth-based reduction in Section 3.2 are interesting in their own right, and Theorem 4.13 gives a unified and conceptually simpler route to the perturbation results. There are no fitted parameters and no circularity: the new implications reduce to previously established estimates, and the Carleson-from-A∞ argument is an actual proof. The main caveat is that the foundational boundary estimates are cited to the unpublished manuscript [HMT1], so the significance of the theorem is currently conditional on the validity and availability of that manuscript.","major_comments":[{"comment":"The central equivalence depends on Lemmas 2.16, 2.17, and 2.24, all of which are either entirely cited to, or in the non-symmetric 1-sided CAD case attributed to, the unpublished manuscript [HMT1] ('work in progress, 2014'). These lemmas are used essentially at (3.18), (3.21), (3.15), and (3.55), so the proofs of both directions of Theorem 1.1 cannot be completed without them. The paper does not reproduce their proofs, and I found no published reference that covers the non-symmetric 1-sided CAD case. This is a load-bearing external dependency, not a circularity, but it blocks unconditional acceptance. The authors should either include proofs of these lemmas in the paper, cite a published or otherwise available version of [HMT1], or explicitly state Theorem 1.1 as conditional on [HMT1].","section":"Section 2.4 and Theorem 1.1"},{"comment":"In the proof that Q_i^k ∩ Q_j^{k+1} ≠ ∅ forces Q_i^k ⊂ Q_j^{k+1}, the displayed inequality a^{-k} μ(Q_i^k) < μ( ~F ∩ Q_i^k) ≤ a^{-k-1} μ(Q_i^k) has an upper bound that does not follow from (3.7) or (3.8). The valid argument is that (3.7) for Q_i^k gives μ( ~F ∩ Q_i^k) > a^{-k} μ(Q_i^k) > a^{-(k+1)} μ(Q_i^k), so Q_i^k itself satisfies the density property used to define F_{k+1}, contradicting maximality of Q_j^{k+1} when Q_j^{k+1} ⊊ Q_i^k. The displayed inequality should be corrected. Since Lemma 3.5 supplies the good cover used in Lemma 3.10, the proof of (a) ⇒ (b) in Theorem 1.1 depends on this repair.","section":"Section 3.1, Lemma 3.5"}],"minor_comments":[{"comment":"The heading reads 'Proof of Proof of Theorem 1.1'; the duplicated 'Proof of' should be removed.","section":"Section 3.1, text before (3.12)"},{"comment":"The chain G0(X_I)/ℓ(I) ≈ G0(X_I)/δ(X_I) ≈ ω0(Δ_Q)/σ(Q) ≈ 1 compresses several nontrivial steps: Lemma 2.24(b) is stated with X_Δ (the corkscrew point of Δ) rather than with X_I, so one also needs Harnack's inequality and the symmetry identity (2.22) to justify replacing X_Δ by X_I. This should be expanded for readability and verifiability.","section":"Section 3.2, Eq. (3.55)"},{"comment":"In (3.36) the lower bound is labelled u(X_{~Q_i^ℓ}), but in the context of Case 2 the lower bound is for u(X_{~P_i^ℓ}) = u(X_{~Q_i^ℓ}); the label is confusing and should be corrected.","section":"Section 3.2, Lemma 3.24, Case 2"},{"comment":"The chain T_Q ⊂ T_Q^* ⊂ T_Q^{**} ⊂ T_Q^{**} ⊂ κ0 B_Q ∩ Ω repeats T_Q^{**}; the intended second inclusion is likely T_Q^{**} ⊂ T_Q^{***} or the intermediate region should be named consistently.","section":"Section 2.3, Eq. (2.11)"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the decisive issue is the unpublished dependency [HMT1]. If the authors can supply a public version of [HMT1] or append the needed boundary estimates, and the Lemma 3.5 inequality is corrected, I would view the paper as acceptable. The manuscript is otherwise careful and the main results are consistent with the known symmetric and Lipschitz cases."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a serious paper with a real new theorem, but before you rely on it you need to know that its foundation is an unpublished manuscript [HMT1]. The central result, Theorem 1.1, gives an iff in 1-sided chord-arc domains: every bounded solution of Lu = -div(A∇u) satisfies the Carleson measure estimate (1.2) iff the elliptic measure ω_L ∈ A∞. This extends Kenig-Kirchheim-Pipher-Toro from Lipschitz/graph domains to 1-sided CADs, and the converse direction gets a genuinely new proof. Theorems 1.3 and 1.6 are applications: non-symmetric perturbation stability and transpose equivalence under Carleson conditions. The work is not circular: the new Carleson-from-A∞ argument assumes ω_L ∈ A∞ and derives the Carleson estimate, and the perturbation reduction is a clean integration-by-parts argument with the antisymmetric part. The dyadic good-cover lemma and the sawtooth technology are well executed.\n\nThe soft spot is exactly the one in the reader's report: Lemma 2.24, parts (a)–(d), state the boundary Hölder decay, Green-function/elliptic-measure comparison, doubling, and boundary comparison for real non-symmetric operators, and the proofs are in [HMT1], listed as work in progress 2014. These lemmas are used essentially throughout Theorem 1.1 — see (3.18), (3.21), (3.15), and (3.55). If any of those estimates fails for non-symmetric L in a 1-sided CAD, the equivalence has no proof. This is not circularity, but it does mean the paper cannot be fully verified from its own text. The authors are explicit about the dependence, which is honest. For a referee, this is the main issue: conditional accept, with the condition being that [HMT1] becomes available or the key lemmas are reproduced.\n\nThe proofs that are present are detailed and coherent; the reduction to discrete Carleson measures (Lemma 3.39) and the use of HMT2's machinery are clean. No free parameters, no invented entities. The statements are consistent with the symmetric and Lipschitz cases. The citation pattern is honest; self-citations are to independently established results, and the [HMT1] reliance is flagged.\n\nWho is this for? People working on elliptic measure, A∞ weights, and boundary value problems on rough domains. If you need a perturbation theorem for non-symmetric operators or transpose equivalence, this is the paper. It deserves a serious referee. My recommendation: send it to review, but ask the authors to provide the [HMT1] estimates, either by posting the manuscript or appending the proofs. If they do, this is an accept.","headline":"Serious and genuinely new results in non-symmetric elliptic measure theory, but the load-bearing estimates live in an unpublished companion; accept conditionally on their release.","tokens_in":37131,"tokens_out":2303,"would_cite":true,"duration_ms":23780,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["31B05","35J08","35J25","42B99","42B25","42B37"],"pacs":[],"model":"deepseek-v4-flash","headline":"In 1-sided chord-arc domains, Carleson estimates are equivalent to A∞ elliptic measure, even for non-symmetric coefficients.","keywords":["elliptic measure","Poisson kernel","Carleson measures","A-infinity weights","1-sided chord-arc domains","non-symmetric elliptic operators","perturbation of elliptic operators","uniform rectifiability"],"falsifier":"Exhibit a 1-sided chord-arc domain $\\Omega$ and a real uniformly elliptic matrix $A$ (not necessarily symmetric) such that every bounded weak solution satisfies (1.2) but $\\omega_L\\notin A_\\infty(\\partial\\Omega)$, or vice versa; because the two properties are claimed equivalent for all such operators, a single counterexample would decide against the theorem.","tokens_in":36039,"feed_emoji":"","tokens_out":7251,"duration_ms":69664,"temperature":0.7,"pith_summary":"This paper proves that, in a 1-sided chord-arc domain, a real elliptic operator $Lu=-{\\rm div}(A\\nabla u)$ with possibly non-symmetric coefficients has elliptic measure in the Muckenhoupt class $A_\\infty(\\partial\\Omega)$ exactly when every bounded weak solution satisfies a Carleson measure estimate for $|\\nabla u|^2\\delta(X)$. The result extends a previous square-function criterion from bounded Lipschitz domains and graphs to the rougher 1-sided chord-arc setting, and removes the symmetry assumption on the coefficient matrix. On this equivalence the paper builds two applications: a perturbation theorem showing that the $A_\\infty$ property is stable under coefficient disagreements that are Carleson in a quadratic sense, and a theorem identifying when an operator and its transpose (or its symmetric part) have $A_\\infty$ elliptic measure simultaneously. A corollary removes an auxiliary hypothesis in an earlier result: under a slightly stronger Carleson condition, $A_\\infty$ ellipticity for the operator already forces the domain to be a genuine chord-arc domain.","feed_headline":"Carleson estimates are equivalent to A∞ elliptic measure in rough domains","feed_subtitle":"For non-symmetric elliptic operators, this gives perturbation stability and a link between an operator and its transpose.","key_machinery":"The load-bearing mechanism is the passage between a continuum Carleson estimate and dyadic sawtooth combinatorics. The paper uses dyadic grids on the Ahlfors-regular boundary, Carleson boxes $T_Q$, sawtooth domains $\\Omega_{\\mathcal{F},Q}$, and a \"good $\\varepsilon_0$-cover\"—a nested chain of level sets defined through the dyadic maximal operator, in which each next level loses a fixed fraction of the previous elliptic measure. Lemma 3.10 converts a small elliptic-measure set into a bounded solution whose conical square function is large on that set, which is exactly what forces the surface measure to be small. In the converse direction, the Green function $G_{L_0}$ replaces $\\delta(X)$ as the weight, and an integration-by-parts identity for antisymmetric matrices, involving the column-wise divergence ${\\rm div}_C D$, is what lets the perturbation terms be absorbed as Carleson-measure contributions.","core_discovery":"On the paper's own terms, the central discovery is Theorem 1.1: for any real uniformly elliptic divergence-form operator $L$ in a 1-sided chord-arc domain, condition (a)—every bounded weak solution satisfies the Carleson measure estimate (1.2)—is equivalent to condition (b)—$\\omega_L\\in A_\\infty(\\partial\\Omega)$. The forward direction is proved by constructing, for any small elliptic-measure set $F$, a good $\\varepsilon_0$-cover whose nested levels are turned into a harmonic function $u=\\omega_L^X(S)$ with large conical square function on $F$, forcing $F$ to have small surface measure. The reverse direction replaces the distance-to-boundary weight $\\delta$ by the Green function $G_{L_0}$ on sawtooth subdomains and uses integration by parts with adapted cutoffs. The same reversal, run with $L_0$ in place of $L$, yields the general perturbation statement Theorem 4.13, from which Theorems 1.3 and 1.6 follow.","pith_inferences":["Because the proof of Theorem 1.1 is quantitative, the $A_\\infty$ constants and the Carleson constant should depend only on the ellipticity, the 1-sided chord-arc domain constants, and the Carleson norm; comparing these constants across the perturbation theorem may give effective stability bounds.","The Green-function substitution suggests a template for nonlinear divergence-form operators, such as the $p$-Laplacian, wherever a Green function and boundary Harnack estimates are available.","Theorem 1.6 indicates that the antisymmetric part of the matrix is visible to elliptic measure only through its column-wise divergence; one could test whether matrices with the same antisymmetric divergence produce the same $A_\\infty$ behaviour."],"forward_implications":["For any real (not necessarily symmetric) elliptic operator in a 1-sided chord-arc domain, the Carleson measure estimate (1.2) and $\\omega_L\\in A_\\infty(\\partial\\Omega)$ are interchangeable criteria.","If $\\omega_{L_0}\\in A_\\infty$ and the disagreement $\\varrho(A_1,A_0)^2/\\delta$ is a Carleson measure, then $\\omega_{L_1}\\in A_\\infty$; the perturbation need not be small.","When the antisymmetric part is locally Lipschitz and its column divergence satisfies (1.8), the three properties $\\omega_L\\in A_\\infty$, $\\omega_{L^\\top}\\in A_\\infty$, and $\\omega_{L_{\\rm sym}}\\in A_\\infty$ are equivalent.","Under the stronger gradient-Carleson hypothesis (1.11), $\\omega_L\\in A_\\infty$ alone implies the domain is a chord-arc domain; the transpose condition previously assumed is redundant.","Small antisymmetric perturbations with locally Lipschitz entries and Carleson-controlled divergence preserve the $A_\\infty$ property of elliptic measure."],"supporting_citations":[{"why":"Supplies the quantitative elliptic-measure, Green-function, and comparison estimates (Lemmas 2.16, 2.17, 2.24) that the proof of Theorem 1.1 invokes as black boxes.","marker":"[HMT1]"},{"why":"Provides the A∞-implies-NTA theorem used in Corollary 1.9 and the sawtooth Poincaré and stopping-time lemmas used in the converse direction.","marker":"[HMT2]"},{"why":"The previous symmetric-operator perturbation result whose scope and proof are extended here to non-symmetric coefficients.","marker":"[CHM]"},{"why":"The square-function/A∞ criterion for bounded Lipschitz domains and Lipschitz graphs that the paper generalizes to 1-sided chord-arc domains.","marker":"[KKPT]"},{"why":"Supplies the construction of Carleson boxes, sawtooth regions, and dyadic grids used throughout the argument.","marker":"[HM3]"},{"why":"Establishes the dyadic cube decomposition on spaces of homogeneous type, used for the boundary dyadic grid on an Ahlfors regular set.","marker":"[Chr]"}],"fun_headline_variants":["Non-symmetric operators: Carleson iff A∞ measure","Rough domains: Carleson control ties to A∞ elliptic measure","Equivalence of Carleson estimates and A∞ in 1-sided chord-arc","For non-symmetric elliptic operators, Carleson ↔ A∞ measure","Carleson estimates characterize A∞ measure for non-symmetric operators"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on a suite of boundary estimates for elliptic measure and Green functions in 1-sided chord-arc domains—doubling, boundary Harnack, and comparison—quoted from an unpublished manuscript [HMT1]; the main equivalence would collapse if those estimates are not available with the stated quantitative control.","fun_headline_variants_meta":{"raw":{"variants":["Non-symmetric operators: Carleson iff A∞ measure","Rough domains: Carleson control ties to A∞ elliptic measure","Equivalence of Carleson estimates and A∞ in 1-sided chord-arc","For non-symmetric elliptic operators, Carleson ↔ A∞ measure","Carleson estimates characterize A∞ measure for non-symmetric operators"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000935,"raw_usage":{"total_tokens":4060,"prompt_tokens":1064,"completion_tokens":2996,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":680,"completion_tokens_details":{"reasoning_tokens":2900}},"tokens_in":680,"tokens_out":2996,"duration_ms":18983,"temperature":1.0,"reasoning_tokens":2900,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:49:19.607365+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a 1-sided chord-arc domain $\\Omega$ and a real uniformly elliptic matrix $A$ (not necessarily symmetric) such that every bounded weak solution satisfies (1.2) but $\\omega_L\\notin A_\\infty(\\partial\\Omega)$, or vice versa; because the two properties are claimed equivalent for all such operators, a single counterexample would decide against the theorem.","supporting_citations":[],"review_version":1}