{"id":"d140c3fd-0be1-4aa2-b33c-5b15ca6cec41","arxiv_id":"2601.16167","paper_version":3,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":6,"one_line_summary":"For uniformly non-flat Ahlfors-David regular boundaries of dimension between n−1−δ0 and n−1 in R^n (n≥3), harmonic measure is concentrated on a set of dimension strictly less than the boundary dimension.","lead":"This paper claims that for uniformly non-flat, Ahlfors-David regular boundaries of dimension between n−1−δ0 and n−1 in R^n, the harmonic measure is concentrated on a set of strictly smaller dimension than the boundary. The genuinely new case is the codimension-one endpoint s = n−1, with an explicit — although astronomically small — allowed gap δ0.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1 claims dim ω < κs with κ < 1, but Lemma 13 only proves dim ω < s; the quantitative bound is never derived.","rationale":"The reader identified several concerns, choosing the unproved corkscrew condition as the weakest assumption. I agree that the corkscrew condition is asserted without proof and is used essentially in Theorem 9. However, the most load-bearing concern about the central claim is the gap between Lemma 13 and Theorem 1: the theorem promises dim ω < κs with κ ∈ (0,1), but the only dimension-drop result proved is dim ω < s. Even if the corkscrew condition and all geometric estimates are correct, the paper does not deliver the quantitative part of its own headline theorem. This is an internal inconsistency in the argument as written, independent of external geometric assumptions. It is concrete and testable: one can trace the constants through Lemma 13. The reader's rationale also notes this issue ('Lemma 13 gives only dim ω < s'), so there is partial agreement. I recommend keeping the CONDITIONAL verdict: the paper contains a substantial plausible route, but the proof as printed does not establish the stated quantitative conclusion. The corkscrew issue is also real and should be addressed, but the κs gap is the more immediate blocker to accepting Theorem 1.","tokens_in":26485,"tokens_out":15472,"duration_ms":153312,"concrete_test":"Re-examine Lemma 13 (and the earlier Tolsa/Bourgain dimension-drop lemmas it cites) to determine whether, under the hypothesis (117) with the specific M, η produced by Theorem 9, one can conclude dim ω ≤ α(M, η) · s for an explicit α < 1. If no such α is obtained, or if the only valid conclusion is dim ω < s, then Theorem 1's κ < 1 is unproved. A direct check: substitute the M and η from Theorem 9 into the proof of Lemma 13 and compute the resulting upper bound on dim ω; if it is s (or an unspecified constant), the theorem's assertion fails as stated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim of Theorem 1 is that there exists κ ∈ (0,1) with dim ω < κs. The proof, however, ends with Lemma 13, whose statement and proof conclude only dim ω < s. No argument in Section 6 (or anywhere else) derives a strictly smaller dimension bound. The symbol κ is used earlier in Lemma 7 as a geometric scaling factor (Eq. 16), and the theorem's κ is never connected to the proof. Moreover, Lemma 13 is stated conditionally on a density-increment property (Eq. 117) with unspecified constants M, η; even if that property is established by Theorem 9, the quantitative value of κ would depend on M, η, and the AD-regular constant, but the paper does not compute or even assert such dependence. Thus the headline result—the explicit dimension drop below s—is not derived. This is not a matter of an obscure estimate; it is a missing logical step between the final lemma and the theorem statement. If Lemma 13 can only give dim ω < s, then the theorem should be weakened accordingly, and the phrase 'dim ω < κs' is unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to prove a quantitative dimension drop for harmonic measure on uniformly non-flat Ahlfors-David regular boundaries of dimension s with n-1-delta0 <= s <= n-1, for n>=3. The method is potential-theoretic: using the Dirichlet Green function, the author isolates corner regions near a corkscrew point, uses uniform non-flatness and AD-regularity to force a non-uniform gradient of the Poisson kernel, and derives a density-increment contradiction resembling Azzam's argument. The proof then invokes a change-of-pole lemma and a density-increment dimension-drop lemma to conclude the theorem. The central advertised novelty is an explicit quantitative bound on the dimension-gap parameter delta0 and a dimension-drop factor kappa<1, obtained without Riesz-transform or compactness machinery.","tokens_in":26633,"tokens_out":18833,"duration_ms":175379,"significance":"If fully correct, the theorem would extend Azzam's dimension-drop result to boundaries at or below the codimension-one threshold, with explicit dependence of the admissible dimension gap on the non-flatness parameter, and would provide an elementary alternative to existing Riesz-transform techniques. The local double-integral argument in Lemma 10 is a genuine non-circular idea, and the explicit choices of w and delta0 in Section 5 show a serious quantitative intention. However, the manuscript as written does not deliver the stated theorem: the quantitative dimension-drop factor is not derived, the displayed delta0 is not the one produced by the proof, and the corkscrew condition is asserted without support. These are load-bearing gaps, not presentation issues.","major_comments":[{"comment":"Theorem 1 asserts the existence of kappa in (0,1) such that dim omega < kappa s. The proof ends with Lemma 13, whose statement and proof conclude only dim omega < s. The symbol kappa is introduced in Lemma 7 as a geometric scaling factor (Eq. (16)) and is never connected to the theorem's kappa. The final reduction to [Tol24, Lemma 2.8] is not shown to yield a dimension bound strictly below s, and no dependence of a quantitative bound on the constants M and eta in Eq. (117) is computed. The headline dimension-drop factor is therefore not derived; the theorem must either be proved with a quantitative estimate or weakened to dim omega < s.","section":"§6, Lemma 13"},{"comment":"The claim that boundaries of codimension >=1 automatically satisfy the interior corkscrew condition is made without proof or citation. This is load-bearing: the corner construction of Section 5 (Eq. (13)) and Lemma 8 require a pole p1 at distance r1 from the boundary with B(p1,r1) empty, and Theorem 9's proof uses the corkscrew condition to obtain a uniform lower bound on |A(U)-\\hat A(U)| in terms of l(Q). For a connected domain whose AD-regular boundary has dimension s <= n-1, nearly touching boundary components can prevent uniform corkscrew balls; Eq. (1) alone does not obviously exclude this. The authors must either prove the corkscrew condition from the hypotheses or add it as an explicit assumption.","section":"§3, before Definition 1"},{"comment":"The delta0 displayed in Theorem 1 and the delta0 derived in the proof are different. The proof concludes at Eqs. (89) and (95) that it is enough to take delta0 approximately beta_1^{3n log C / beta_1^n}, whereas Theorem 1 displays delta0 of the form beta^{4n log C1} beta_1^n (or beta/(4n log C1) beta_1^n as rendered). For small beta these are vastly different, with the displayed value far larger (less restrictive) than the value the proof supports. The quantitative claim of the theorem must be reconciled with the proof's sufficient bound.","section":"§5 (2), Eqs. (89)–(95)"},{"comment":"In the s=n-1 case the lower bound in Eq. (42) scales like beta_1^{w(1+s)} = beta_1^{wn}. Immediately afterward, the summed lower bound and theta0 in Eq. (47) use beta_1^{2w}. Since n>=3 and beta_1<1, beta_1^{wn} is much smaller than beta_1^{2w}; the larger exponent is not justified by the projection argument. The subsequent estimates (e.g., Eqs. (50) and (62)) rely on this exponent, so the contradiction in the core density-increment argument is not established as written.","section":"§5 (1), Eq. (42) ff."}],"minor_comments":[{"comment":"The statement 'there exists a1<=kappa<=1/2' should read '01' is undefined.","section":"§6, Lemma 13"},{"comment":"The sentence 'This completes the proof of Theorem 1' follows only a one-sentence reduction to [Tol24, Lemma 2.8]; given that the density-increment hypothesis in Lemma 13 is the exact input to the theorem, a more detailed proof of this reduction is needed.","section":"§6, end"},{"comment":"The equation numbering '(3.4)' appears in Section 2; renumber consistently.","section":"§2, Eq. (3.4)"}],"recommendation":"major_revision","confidential_remarks":"The advertised quantitative theorem is not supported by the current proof: the dimension-drop factor kappa is never derived, the displayed delta0 does not match the delta0 obtained in the proof, and the corkscrew assertion is unproved. These are central, not cosmetic. I recommend major revision with the request that the authors either supply the missing quantitative arguments or revise the theorem to match what the proof actually establishes. The local double-integral idea is worth preserving."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing to know: this paper has a genuinely new endpoint case and a plausible elementary strategy, but as written it does not prove its own Theorem 1. The final lemma gives only dim ω < s, while the theorem claims dim ω < κs for some κ < 1, and no argument in Section 6 bridges that gap. The κ in the theorem is never connected to the proof; the same symbol appears earlier in Lemma 7 as a geometric scaling factor with a different value. There is also an unflagged kernel inconsistency: §2 defines the Green function using the R^{n+1} fundamental solution, while all estimates use the R^n kernel.\n\nWhat is actually new: the endpoint s = n−1 under uniform non-flatness, and the range just below it, with an explicit δ0, is absent from the cited literature. Azzam handles codimension < 1, and Tolsa treats boundaries contained in a hyperplane. The paper avoids Riesz-transform machinery and compactness arguments, and the double-integral contradiction in Lemma 10 has the right order of magnitude: for an (n−1)-dimensional AD cube, the inner Riesz potential scales like L. The honesty of the limitation notes (R^2 omitted, (n−1, n) omitted) also weighs in the author's favor, and the citation pattern appears appropriate.\n\nThe soft spots, in proportion: the missing κs derivation is load-bearing, not a cosmetic issue. Lemma 13's condition (117) depends on unspecified constants M, η, and even if the density-increment property is granted, no quantitative κ is extracted. The corkscrew condition is asserted for codimension ≥ 1 without proof or citation; the corner construction in Section 5 and the lower bound on |A(U) − q1| in Theorem 9 rely on it, and thin gaps in the complementary set could break it. The killed-domain capacity-density condition in Lemma 12 is unverified. The displayed δ0 still contains an unspecified constant. These are fixable-sounding issues, and I did not find a demonstrated false step in the central estimate chain, but the text is OCR-degraded and internally mislabeled, so fine-grained verification was not possible here.\n\nWho this is for: specialists in harmonic measure and geometric measure theory. A serious referee could check whether the κs bound can be extracted from Lemma 13 with additional work, or whether the theorem must be weakened to dim ω < s. My recommendation: send it to peer review rather than desk reject it — the endpoint question is important and the strategy deserves referee time — but expect major revision before it is publishable.","headline":"Plausible endpoint result for dimension drop at codimension-one, but the headline κs bound is never derived and the kernel is inconsistent; worth a referee but not citable yet.","tokens_in":27302,"tokens_out":2260,"would_cite":false,"duration_ms":24085,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["31B15","28A78","35J25"],"pacs":[],"model":"deepseek-v4-flash","headline":"Theorem 1 claims harmonic measure drops dimension on uniformly non-flat AD-regular boundaries with codimension at least one, for an explicit range of boundary dimensions.","keywords":["harmonic measure","dimension drop","Ahlfors-David regular","uniform non-flatness","potential theory","Green function","corkscrew condition","dyadic cubes"],"falsifier":"One concrete way to test the claim: construct a connected domain Ω⊂R^3 with a 2-dimensional AD-regular boundary satisfying the uniform non-flatness condition (1) but for which some boundary point x has no interior ball B(y,c r)⊂Ω∩B(x,r) with a universal c>0 (e.g., a boundary with arbitrarily thin gaps between components). If such a domain fails the conclusion dim ω<κs, the theorem is false. A direct numerical check would compute the Poisson-kernel density ratio Θ on nested dyadic cubes for such a pinched gap, and see whether the density jump of Theorem 9 actually occurs.","tokens_in":26185,"feed_emoji":"📉","tokens_out":5936,"duration_ms":60730,"temperature":0.7,"pith_summary":"This paper proves that harmonic measure in R^n (n≥3) loses dimension whenever the domain's boundary is Ahlfors-David regular of dimension s in the interval n−1−δ0 ≤ s ≤ n−1 and is uniformly non-flat at every location and scale. The theorem gives a concrete, quantitative δ0 that depends only on the non-flatness parameter β and the regularity constant C1, and it avoids Riesz-transform and compactness machinery, relying instead on elementary potential theory: a corner construction, multipole expansions of the fundamental solution, and a double-integral averaging argument. If correct, this is the first quantitative dimension-drop result for boundaries of codimension at least one in the uniformly non-flat setting.","feed_headline":"Non-flat AD-regular boundaries force harmonic-measure dimension drop","feed_subtitle":"Explicit δ0 extends the dimension-drop result to boundaries of codimension ≥ 1, without Riesz transforms.","key_machinery":"The argument rests on a corner construction: for a corkscrew point p1 and a nearest boundary point q1, the boundary is decomposed into dyadic annular regions A_m(q1) with radii κ^m t1, where κ is chosen from the Ahlfors-David regularity constant. The uniform non-flatness condition forces a point of the boundary in each annulus at controlled distance from the plane through q1 perpendicular to p1q1, giving a lower bound on the gradient of the fundamental-solution potential at q1. Taylor/multipole expansions of the fundamental solution up to second order split potential differences into a gradient term and a second-order term; the annulus geometry plus Ahlfors regularity bounds the second-order","core_discovery":"The central claim is Theorem 1: fix n≥3, C1>1, and 0<β<1, set β1=1/(4⌈1/β⌉+2) and δ0=β1^{3n log C1 / β1^n}. For every connected domain Ω⊂R^n whose boundary is (s,C1)-Ahlfors-David regular with n−1−δ0 ≤ s ≤ n−1 and satisfies the uniform non-flatness condition b^β_{ωΩ}(x,r) ≥ β, there exists κ∈(0,1) such that dim ω_Ω < κs. That is, there is a Borel set K with dim K < κs and ω_Ω(K^c)=0: the harmonic measure, although supported on the full boundary, lives entirely on a strictly lower-dimensional subset. The proof works by showing that on every dyadic boundary cube there is a subcube—at a length scale controlled by the parameters—where the average Poisson kernel differs from the average over the","pith_inferences":["A natural extension is the plane case n=2, where the logarithmic fundamental solution would require modified estimates; the author notes this but does not carry it out, and the same corner construction plausibly yields an analogous dimension drop there.","The explicit dependence of δ0 on β and C1 suggests an effective version of the theory: one can track how κ changes as β→0, which may be useful for numerical experiments on fractal boundaries.","The unproved 'automatic' corkscrew condition is a hidden geometric premise; if there exist AD-regular, uniformly non-flat boundaries of dimension n−1 without a uniform interior corkscrew constant, the theorem's scope would be narrower than stated and a two-sided capacity condition would be needed.","The change-of-pole argument could be extracted as a standalone lemma transferring density oscillations from boundary balls to arbitrary poles in AD-regular domains, which may have applications beyond the dimension-drop problem."],"forward_implications":["If Theorem 1 is correct, the dimension drop for harmonic measure holds for uniformly non-flat AD-regular boundaries with dimension in [n−1−δ0, n−1], a regime of codimension at least one previously treated only under extra geometric structure.","The explicit formula for δ0 gives a concrete range of s for which the drop is guaranteed, rather than an existential constant from compactness or Riesz-transform methods.","The theorem implies that on such domains harmonic measure is singular with respect to the s-dimensional Hausdorff measure on the boundary: there is a set K with ω(K)=1 and H^s(K)=0.","The density-oscillation property established in Theorem 9 is exactly the input to the standard dimension-drop criterion, so the proof yields a quantitative route to dimension estimates for harmonic measure.","Because the proof avoids Riesz transforms, the same corner-and-averaging mechanism may be adaptable to other elliptic operators and to the logarithmic (n=2) case."],"fun_headline_variants":["Explicit δ0 forces harmonic-measure dimension drop on AD boundaries","No Riesz transforms: harmonic-measure dimension drop on non-flat AD boundaries","Uniformly non-flat AD boundaries force harmonic-measure dimension drop","Explicit δ0 extends dimension drop to codimension ≥1 AD boundaries","Dimension drop on non-flat AD boundaries with explicit quantitative bounds"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof assumes without a supplied argument that every AD-regular boundary of dimension s ≥ n−1 automatically satisfies the interior corkscrew condition with a uniform constant, so the corkscrew point sits at distance comparable to the boundary cube's side length; if that uniformity fails, the corner lower bounds collapse.","fun_headline_variants_meta":{"raw":{"variants":["Explicit δ0 forces harmonic-measure dimension drop on AD boundaries","No Riesz transforms: harmonic-measure dimension drop on non-flat AD boundaries","Uniformly non-flat AD boundaries force harmonic-measure dimension drop","Explicit δ0 extends dimension drop to codimension ≥1 AD boundaries","Dimension drop on non-flat AD boundaries with explicit quantitative bounds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00196,"raw_usage":{"total_tokens":7485,"prompt_tokens":718,"completion_tokens":6767,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":462,"completion_tokens_details":{"reasoning_tokens":6673}},"tokens_in":462,"tokens_out":6767,"duration_ms":43445,"temperature":1.0,"reasoning_tokens":6673,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T08:42:32.395540+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete way to test the claim: construct a connected domain Ω⊂R^3 with a 2-dimensional AD-regular boundary satisfying the uniform non-flatness condition (1) but for which some boundary point x has no interior ball B(y,c r)⊂Ω∩B(x,r) with a universal c>0 (e.g., a boundary with arbitrarily thin gaps between components). If such a domain fails the conclusion dim ω<κs, the theorem is false. A direct numerical check would compute the Poisson-kernel density ratio Θ on nested dyadic cubes for such a pinched gap, and see whether the density jump of Theorem 9 actually occurs.","supporting_citations":[],"review_version":1}