{"id":"4267979c-2e48-4d10-be30-e33376b001f5","arxiv_id":"2505.16473","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For weighted inhomogeneous Dirichlet approximation, the Hausdorff f-measure of non-improvable matrices is zero when a certain series converges and full when it diverges.","lead":"This number theory paper proves a zero-full law for the Hausdorff measure of sets of weighted, inhomogeneous Dirichlet non-improvable matrices, under a decay condition on the approximating function. The result generalizes a theorem of Kim and Kim, but the proof relies on an unproved quasi-independence lemma that needs scrutiny.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 5.4, the quasi-independence estimate at the core of the divergence proof, is stated without proof and is used as a black box in the paragraph entitled 'Proof of Lemma 5.4', which actually proves Lemma 5.1; Theorem 1.6 is therefore conditional on that estimate.","rationale":"The paper's stated goal is a weighted inhomogeneous zero-full law for Hausdorff measure, extending Kim-Kim. The convergence half is a standard covering argument and appears sound. The divergence half is more delicate: it constructs a full-measure limsup set W(Φ), then uses the mass transference principle to transfer full measure to a shrunk limsup set inside D^b_{α,β}(ψ)^c. The linchpin is Lemma 5.4's pairwise intersection estimate, which is the only input that makes Lamperti's lemma applicable. The reader's weakest assumption identifies precisely this: Lemma 5.4 is asserted without proof, and the subsequent proof labeled 'Proof of Lemma 5.4' actually proves Lemma 5.1 while using Lemma 5.4 as a black box. I checked the surrounding text and found no other source or argument supplying the estimate. Because the estimate is load-bearing and unverified, the verdict should remain CONDITIONAL rather than ACCEPT; it should not be REJECT, since the statement is plausible and the missing proof may be routine, albeit essential. The concrete test I propose isolates the most dangerous case—parallel integer multiples—where a naive product bound can fail, and checks whether the gcd condition repairs it. If that check passes, the remaining step is to write out the general proof of Lemma 5.4; if it fails, the divergence argument is invalid. Thus my concern agrees with the reader's, and I recommend no change to the conditional verdict.","tokens_in":14611,"tokens_out":14565,"duration_ms":120642,"concrete_test":"Verify Lemma 5.4 by direct computation in the two extremal cases: (i) m=2, n=1, u1=(1,0), u2=(N,0) with the gcd condition; (ii) m=2, n=1, u1=(1,0), u2=(N,1) with det=1. For each, compute L(R'(u1,δ1)∩R'(u2,δ2)) as a function of δ1, δ2, N and check whether it is bounded by a constant independent of N times δ1δ2 for all δ1, δ2 > 0. If case (i) is not O(δ1δ2) because the gcd condition fails to exclude overlapping bands, Lemma 5.4 is false and the divergence argument collapses; if both cases pass, write the general proof and check the constant for the parallel case u2 = k u1, which Lemma 5.4 permits.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The divergence part of Theorem 1.6 rests on Lemma 5.4, which asserts the quasi-independence estimate L(R'(u1,Φ(u1))∩R'(u2,Φ(u2))) ≪ ∏_j ϕ_j(u1)ϕ_j(u2) for all u1 ≠ ±u2. This inequality is exactly what allows Lemma 5.3 (Lamperti) to convert the divergent measure sum from Lemma 5.10 into positive measure, then Lemma 5.1 into full Lebesgue measure, and finally Theorem 2.1 plus Lemma 5.11 into full Hausdorff measure. However, no proof or citation is supplied for Lemma 5.4. The paragraph in Section 5.1 labeled 'Proof of Lemma 5.4' proves Lemma 5.1 while invoking Lemma 5.4 as a black box. The concern is not merely a missing routine detail: the bound must hold uniformly for all integer directions, including the case u2 = k u1 with k ≥ 2, which Lemma 5.4 permits since only u1 ≠ ±u2 is excluded. For parallel slabs the intersection can behave like the smaller width rather than the product of widths; the gcd condition in R' may rescue the estimate by excluding resonant sub-hyperplanes, but the paper gives no argument. Until Lemma 5.4 is proved or traced to a verifiable source, the divergence half of Theorem 1.6, and with it the zero-full law, is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper establishes a zero-full law for the Hausdorff f-measure of the set D^b_{α,β}(ψ)^c of inhomogeneous Dirichlet non-improvable affine forms with weights, under a decay assumption on ψ and a dimension-function range (mn−a) ⪯ f ⪯ (mn−a+1). The proof combines the Diophantine transference principle of Cassels with a reduction to limsup sets of hyperplane neighbourhoods. The convergence half is a direct covering argument using the series criterion. The divergence half constructs a limsup set W_{n,m}(Φ) of full Lebesgue measure, applies Lamperti's lemma via a quasi-independence estimate, and then uses the mass transference principle from balls to rectangles to obtain full Hausdorff f-measure. The main technical tool, Lemma 5.4, is stated without proof and is used as a black box in the proof of Lemma 5.1.","tokens_in":14905,"tokens_out":5734,"duration_ms":44881,"significance":"If the proof is completed, the result would substantially generalize the unweighted theorem of Kim and Kim (Adv. Math., 2022) to weighted vectors and to Hausdorff measures, answering their open question. The convergence part appears sound and is a standard covering argument. The paper contains genuinely new intermediate constructions, especially the definition of the auxiliary functions Φ and the reduction of the divergence part to the full-measure statement for W_{n,m}(Φ). However, the divergence half is currently conditional on the unproved quasi-independence estimate in Lemma 5.4, which is exactly the step that permits the application of Lamperti's lemma. The result is therefore not yet established as stated.","major_comments":[{"comment":"Lemma 5.4 is a load-bearing quasi-independence estimate: it is used in the proof of Lemma 5.1 to convert the divergent measure sum from Lemma 5.10 into positive measure via Lemma 5.3, and then into full Lebesgue measure via Theorem 5.2. However, the paragraph labeled 'Proof of Lemma 5.4' actually proves Lemma 5.1 while invoking Lemma 5.4 as a black box. No proof or precise reference is given for Lemma 5.4 itself. Please supply a complete proof of Lemma 5.4, or a precise citation to a source where both the lower bound and the intersection bound are proved.","section":"Section 5.1, Lemma 5.4 and the paragraph headed 'Proof of Lemma 5.4'"},{"comment":"Lemma 5.4 asserts two estimates: the lower bound (φ(|u1|)/|u1|)^n ∏_j φ_j(u1) ≤ L(R'(u1,Φ(u1))) and the intersection bound L(R'(u1,Φ(u1)) ∩ R'(u2,Φ(u2))) ≪ ∏_j φ_j(u1)φ_j(u2) for all u1 ≠ ±u2. The lower bound is needed in Lemma 5.10 to replace the measure by the product ∏ φ_j(u) up to constants; no counting argument for the admissible v with gcd(u,v_j)=1 is supplied. The intersection bound is asserted uniformly even for collinear pairs such as u2 = k u1 with k ≥ 2, which the lemma explicitly permits; for parallel slabs the intersection can behave like the smaller width rather than the product of widths, and the gcd condition in R' is the only apparent mechanism that could rescue the product estimate. The paper gives no argument that this gcd condition suffices. This point must be addressed before the divergence half of Theorem 1.6 can be considered proved.","section":"Section 5.1, Lemma 5.4 (lower bound and collinear case)"}],"minor_comments":[{"comment":"Lemma 5.11 relies crucially on Proposition 2.2, which is cited to the author's preprint [7]. Since [7] is not a peer-reviewed publication, please clarify its status or include a proof of Proposition 2.2 in an appendix so that the dependence of Theorem 1.6 on this estimate is fully transparent.","section":"Section 2, Proposition 2.2 and Section 5.2, Lemma 5.11"},{"comment":"The statement 'there are ≍ 2^ℓ integers u ∈ [2^ℓ, 2^{ℓ+1}]' is ambiguous: it should be clarified whether the claim is that every integer u in that dyadic range has the stated property, or that a positive proportion does. The proof seems to show the latter for a set Λ of positive density, and the wording should be adjusted accordingly.","section":"Section 5.1, Lemma 5.8"},{"comment":"There are minor typographical issues, including 'hyperrectanlges' (in Section 2, before Proposition 2.2) and inconsistent use of 'limsup' vs. 'lim sup'. These do not affect the mathematics.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper's main result is significant and the convergence half is sound, but the divergence half rests on Lemma 5.4, which is stated without proof and used circularly in the labeled proof. I recommend major revision: the author should either prove Lemma 5.4 or provide a verifiable reference. I would also ask the editor to verify the accessibility of the author's preprint [7], since Proposition 2.2 is load-bearing in Lemma 5.11."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper does real work—Theorem 1.6 is the first weighted inhomogeneous zero-full law for Dirichlet non-improvable affine forms, and the series criterion with γ_u is a genuine generalization that reduces to the Kim–Kim unweighted statement (Remark 2 checks out). The convergence half is a straightforward covering argument and I think it is correct. The construction in Section 5.1, using hyperrectangle content and the mass transference principle from balls to rectangles, is a serious adaptation of the author's earlier weighted framework.\n\nThe problem is Lemma 5.4. It is the quasi-independence estimate that makes the divergence half go: Lamperti's lemma (Lemma 5.3) needs the pairwise intersection bound to pass from the divergent measure sum to positive measure. But the lemma is stated without proof or citation. Worse, the paragraph in Section 5.1 titled 'Proof of Lemma 5.4' does not prove it; it proves Lemma 5.1 and uses Lemma 5.4 as a black box. That is circular in the present text. The concern is not cosmetic: the estimate must hold for all u1,u2 with u1 ≠ ±u2, including roughly parallel vectors. For parallel hyperplane slabs the intersection measure can behave like the smaller width instead of the product of widths. The gcd condition in R' may rescue the estimate by excluding resonant sub-hyperplanes, but the paper gives no argument either way. So the divergence half of Theorem 1.6 is unsupported as written.\n\nThe rest of the machinery looks plausible. The use of the Euler quotient and the density argument in Lemma 5.10 is standard. If the author can supply a proof of Lemma 5.4, the paper would be in very good shape. For a reader in metric Diophantine approximation, the result is directly relevant, but I would not cite the theorem until the lemma is fixed. I'd send the paper to a serious referee with a request to focus on that lemma before acceptance. Not a desk reject—this is important enough and the convergence half is solid.","headline":"A genuine weighted extension of Kim–Kim with a solid convergence half, but the divergence half rests on an unproved quasi-independence lemma, so the main theorem is currently conditional.","tokens_in":15445,"tokens_out":3571,"would_cite":false,"duration_ms":27365,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11J20","11K60","37A17"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes that weighted inhomogeneous Dirichlet non-improvable affine forms obey a zero-full law for Hausdorff measure, with a single series deciding which case occurs.","keywords":["Dirichlet's theorem","inhomogeneous Diophantine approximation","zero-full law","Hausdorff measure","weighted approximation","mass transference principle","Diophantine transference","affine forms"],"falsifier":"Take $u_1,u_2\\in\\mathbb{Z}^m$ with $|u_1|,|u_2|\\to\\infty$, $u_1\\ne\\pm u_2$, and $u_1$ nearly parallel to $u_2$, and compute the ratio $L^{mn}(R'(u_1,\\Phi(u_1))\\cap R'(u_2,\\Phi(u_2))) / \\prod_{j=1}^n \\phi_j(u_1)\\phi_j(u_2)$. If the ratio is unbounded, Lemma 5.4 is false and the divergence argument cannot be completed; if it stays bounded for every such family, the missing quasi-independence estimate is confirmed.","tokens_in":14380,"feed_emoji":"📐","tokens_out":16842,"duration_ms":119226,"temperature":0.7,"pith_summary":"This paper proves a zero-full law for the Hausdorff $f$-measure of the set of inhomogeneous Dirichlet non-improvable affine forms with weights: for shifted affine maps $q\\mapsto Aq+b$ for which Dirichlet's simultaneous approximation cannot be improved by a faster-decaying error function, the fractal-size measure is either zero or the measure of the whole unit cube. For any shift $b\\in\\mathbb{R}^m\\setminus\\mathbb{Z}^m$, any decreasing continuous approximating function $\\psi$ with $\\psi(t)/\\psi(2t)\\ge\\lambda>1$, and any dimension function $f$ in a specified range, the dichotomy is decided by the convergence or divergence of a single explicit series over integer vectors. This answers the weighted Hausdorff-measure question posed in [11, §5.3] and extends the earlier unweighted zero-one law to general weights and arbitrary dimension functions. The proof converts non-improvability into asymptotic approximation via a Diophantine transference principle and then applies a mass transference principle from balls to rectangles.","feed_headline":"Weighted Dirichlet non-improvable forms obey zero-full law","feed_subtitle":"A single convergence series decides, for every weighted inhomogeneous case, whether Hausdorff measure is zero or full.","key_machinery":"The proof is carried by three linked mechanisms. First, the Diophantine transference principle (Theorem 3.1) shows that for $b\\notin\\mathbb{Z}^m$ the non-improvable set is almost a limsup set of hyperplane neighborhoods: $A$ is non-improvable exactly when there are infinitely many $u\\in\\mathbb{Z}^m$ with $\\|A_{*,j}\\cdot u\\|_{\\mathbb{Z}}<c\\,t(u)^{-\\beta_j}$ for all $j$. Second, an auxiliary weight vector $\\Phi=(\\phi_1,\\ldots,\\phi_n)$ built from $\\gamma_u(\\beta,f)$ defines a larger limsup set $W_{n,m}(\\Phi)$ whose Lebesgue measure is shown to be full; the key steps are a quasi-independence estimate for the sets $R'(u,\\Phi(u))$ and a divergence argument using an almost-independence lemma that turns divergent measure sums into positive measure. Third, the mass transference principle from balls to rectangles upgrades full Lebesgue measure to full Hausdorff $f$-measure, using the estimate for the Hausdorff $f$-content of a hyperrectangle (Proposition 2.2) and the choice of $\\varpi_u$ so that a contained hyperrectangle has $f$-content comparable to $\\varpi_u^{mn}$.","core_discovery":"The central claim is Theorem 1.6. Let $\\psi$ be decreasing and continuous with $\\lim_{t\\to\\infty}\\psi(t)=0$ and $\\psi(t)/\\psi(2t)\\ge\\lambda>1$ for $t\\ge1$; let $f\\prec mn$ be a dimension function with $(mn-a)\\preceq f\\preceq(mn-a+1)$ for some $1\\le a\\le n-1$. Then for every $b\\in\\mathbb{R}^m\\setminus\\mathbb{Z}^m$, $H^f(D^b_{\\alpha,\\beta}(\\psi)^c)=0$ if $\\sum_{u\\in\\mathbb{Z}^m\\setminus\\{0\\}}\\gamma_u(\\beta,f)|u|^n<\\infty$, and $H^f(D^b_{\\alpha,\\beta}(\\psi)^c)=H^f([0,1]^{mn})$ if the same series diverges. Here $H^f$ is the Hausdorff $f$-measure, the fractal-size measure built from the dimension function $f$, and $\\gamma_u(\\beta,f)$ is an explicit minimum over $1\\le j\\le n$ of products involving $f$, $t(u)^{-\\beta_j}/|u|$, and the ratios $t(u)^{\\beta_j-\\beta_\\ell}$; $t(u)$ is the smallest $t$ with $|u_i|<\\psi(t)^{-\\alpha_i}$ for every $i$. The result is a genuine zero-full dichotomy: no intermediate Hausdorff $f$-measure values are possible, and the same series governs every inhomogeneous shift.","pith_inferences":["Editorial extension: the same transference-plus-mass-transference architecture is likely to yield zero-full laws for neighbouring weighted sets whenever a full-measure limsup cover can be constructed.","Editorial extension: the theorem leaves open whether the decay hypothesis $\\psi(t)/\\psi(2t)\\ge\\lambda>1$ is necessary; testing slowly decaying functions such as $\\psi(t)=(\\log t)^{-c}$ would reveal whether the almost-independence step is essential.","Editorial extension: for $b\\in\\mathbb{Z}^m$ the transference principle degenerates, so the present method does not apply and a different mechanism would be needed in the homogeneous case.","Editorial extension: because $\\gamma_u(\\beta,f)$ is explicit, the theorem can be turned into concrete Hausdorff-dimension thresholds for power-law $\\psi$ and $f(r)=r^s$, producing practical weighted dimension formulas."],"forward_implications":["For every $b\\notin\\mathbb{Z}^m$, the Hausdorff $f$-measure of the non-improvable set is either $0$ or $H^f([0,1]^{mn})$, so no intermediate measure values occur.","The weighted Hausdorff-measure question stated in [11, §5.3] is answered in the affirmative under the decay condition $\\psi(t)/\\psi(2t)\\ge\\lambda>1$.","In the equal-weight case $\\alpha_i=1/m$, $\\beta_j=1/n$, and $f(r)=r^s$, the new series is equivalent to the series in [11, Theorem 1.4], recovering the earlier unweighted Hausdorff law.","Because the deciding series does not depend on the shift $b$, the same criterion applies uniformly to every inhomogeneous shift outside the integer lattice.","The result covers general dimension functions in the stated range, not just power functions, so it is a statement about Hausdorff measures as well as dimensions."],"supporting_citations":[{"why":"Supplies the Diophantine transference principle used in Corollary 3.2 to convert non-improvability into asymptotic approximation conditions.","marker":"[3]"},{"why":"Provides the density result for integers with $\\varphi(u)/u$ large, used to select the set $\\Lambda$ in Lemma 5.10.","marker":"[5]"},{"why":"Supplies Proposition 2.2, the hyperrectangle Hausdorff $f$-content estimate that underpins the divergence half.","marker":"[7]"},{"why":"Gives the unweighted Hausdorff result and the question answered here, and provides counting and reduction lemmas used in Sections 4 and 5.","marker":"[11]"},{"why":"Introduces the mass transference principle from balls to arbitrary shapes, quoted in Theorem 2.1 and used in the divergence proof.","marker":"[14]"},{"why":"Provides the almost-independence lemma that turns a divergent sum of measures into positive measure for the limsup set.","marker":"[15]"},{"why":"Gives the zero-one law for limsup sets stated as Theorem 5.2, which upgrades positive Lebesgue measure to full measure.","marker":"[17]"},{"why":"Supplies the Hausdorff-measure version of the mass transference principle from balls to arbitrary shapes, also cited in Theorem 2.1.","marker":"[19]"}],"fun_headline_variants":["Zero-full law for weighted Dirichlet non-improvable forms","Answer to Kim-Kim: weighted non-improvable forms obey zero-full law","Weighted inhomogeneous Dirichlet forms: Hausdorff measure is 0 or full","Zero-full dichotomy for inhomogeneous Dirichlet non-improvable forms","Hausdorff measure of weighted non-improvable forms: only 0 or full"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the quasi-independence estimate in Lemma 5.4, which asserts that the intersection measure of $R'(u_1,\\Phi(u_1))$ and $R'(u_2,\\Phi(u_2))$ is at most a constant times the product of the two measures whenever $u_1\\ne\\pm u_2$; the paper states this bound without proof, and the passage labelled as its proof uses the lemma itself while proving a different statement. If the bound fails for large nearly parallel vectors, the divergence part of the theorem no longer follows.","fun_headline_variants_meta":{"raw":{"variants":["Zero-full law for weighted Dirichlet non-improvable forms","Answer to Kim-Kim: weighted non-improvable forms obey zero-full law","Weighted inhomogeneous Dirichlet forms: Hausdorff measure is 0 or full","Zero-full dichotomy for inhomogeneous Dirichlet non-improvable forms","Hausdorff measure of weighted non-improvable forms: only 0 or full"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000582,"raw_usage":{"total_tokens":2728,"prompt_tokens":925,"completion_tokens":1803,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":541,"completion_tokens_details":{"reasoning_tokens":1701}},"tokens_in":541,"tokens_out":1803,"duration_ms":10655,"temperature":1.0,"reasoning_tokens":1701,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T15:01:17.254871+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $u_1,u_2\\in\\mathbb{Z}^m$ with $|u_1|,|u_2|\\to\\infty$, $u_1\\ne\\pm u_2$, and $u_1$ nearly parallel to $u_2$, and compute the ratio $L^{mn}(R'(u_1,\\Phi(u_1))\\cap R'(u_2,\\Phi(u_2))) / \\prod_{j=1}^n \\phi_j(u_1)\\phi_j(u_2)$. If the ratio is unbounded, Lemma 5.4 is false and the divergence argument cannot be completed; if it stays bounded for every such family, the missing quasi-independence estimate is confirmed.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Diophantine transference principle used in Corollary 3.2 to convert non-improvability into asymptotic approximation conditions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the density result for integers with $\\varphi(u)/u$ large, used to select the set $\\Lambda$ in Lemma 5.10."},{"cited_title":"Hausdorff measure and Fourier dimensions of limsup sets arising in weighted and multiplicative Diophantine approximation","cited_arxiv_id":"2504.09411","evidence_quote":"Supplies Proposition 2.2, the hyperrectangle Hausdorff $f$-content estimate that underpins the divergence half."},{"cited_title":"Kim and W","cited_arxiv_id":null,"evidence_quote":"Gives the unweighted Hausdorff result and the question answered here, and provides counting and reduction lemmas used in Sections 4 and 5."},{"cited_title":"Koivusalo and M","cited_arxiv_id":null,"evidence_quote":"Introduces the mass transference principle from balls to arbitrary shapes, quoted in Theorem 2.1 and used in the divergence proof."},{"cited_title":"Lamperti","cited_arxiv_id":null,"evidence_quote":"Provides the almost-independence lemma that turns a divergent sum of measures into positive measure for the limsup set."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the zero-one law for limsup sets stated as Theorem 5.2, which upgrades positive Lebesgue measure to full measure."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Hausdorff-measure version of the mass transference principle from balls to arbitrary shapes, also cited in Theorem 2.1."}],"review_version":1}