{"id":"80127064-b362-4c7c-bd4e-ef0dece5bad0","arxiv_id":"2505.05425","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For every p0, there are rectangle-based Busemann-Feller bases on T^ω whose L^p differentiation range is exactly [p0,∞] or (p0,∞], and these ranges exhaust the six possible forms for complete metric measure spaces.","lead":"This paper constructs differentiation bases of axis-parallel rectangles on the infinite-dimensional torus that differentiate L^p functions exactly when p lies above any prescribed threshold. It also proves that, in any complete metric measure space, the set of exponents for which a basis differentiates L^p can only take one of six possible forms.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Borel–Cantelli step in Theorem 1.1 asserts that the upper average f(x) is at least 1 from individual rectangles with large averages, but omits the required verification that the nested rectangles supplied by (A3) have diameters tending to 0; the needed bound is implicit in Lemma 4.3 but should…","rationale":"The reader's weakest assumption and my own reading converge on the same step: after the Borel–Cantelli argument, the proof jumps from individual rectangles Q with Avg f(Q) >= 1 to the upper average f(x) >= 1 without explicitly constructing a contracting sequence. This is not a stylistic nit: the upper average is defined through limits over sequences (S_n) contracting to x, and condition (A3) gives only nesting, not diameter decay. The needed estimate is almost certainly true from Lemma 4.3, since every rectangle used at step j lies in a dyadic rectangle Q_k with k >= m_j + d_j and m_j -> infinity, so diam(Q) <= C 2^{-m_j}. However, because this is not written, the proof as it stands relies on an unverified geometric fact. The concrete check above would settle it in a few lines. If the bound holds, Theorem 1.1 and the examples in Theorem 1.3 are sound; if it failed, the non-differentiation argument would collapse. Given that the gap is real but readily fixable, I recommend conditional acceptance rather than unconditional acceptance.","tokens_in":20464,"tokens_out":34333,"duration_ms":350226,"concrete_test":"Check whether the diameter bound implicit in Lemma 4.3 is true: for each j and each Q in S_{j,n}, write Q as a subset of g + Q_k with k >= m_j + d_j as in Lemma 4.3, and compute diam_{rho_{T^omega}}(Q) <= sum_{d=1}^{k+2} 2^{-d} * 2^{-k-1} + sum_{d=k+3}^{infty} 2^{-d} * (1/2) <= 2^{-k} + 2^{-k-2} <= C 2^{-m_j}. Then verify that the nested sequence chosen for x in limsup F*_j satisfies diam(Q_j) -> 0, so (Q_j) -> x in the sense of the definition and the averaging argument for f(x) >= 1 is valid. If the estimate fails for some sequence of k's, the proof of failure of differentiation is incomplete.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The weakest point is in the proof of Theorem 1.1, Section 5, immediately after the display defining f in (5.1). To prove that p is not in diff(B), the text invokes the second Borel–Cantelli lemma and then says: for x in infinitely many selected E_{j,n}, the upper average f(x) is at least 1 because for each such j there is a rectangle Q in S_{j,n} containing x with Avg f(Q) >= 1. By the definition in Section 1.3, this conclusion requires a sequence (Q_j) of sets from B(x) that actually contracts to x, i.e. whose diameters with respect to rho_{T^omega} tend to 0. Condition (A3) only provides nestedness (Q_{j'} subset Q_j for j'>j); it does not by itself imply diameter decay. The missing estimate is available: Lemma 4.3 produces all rectangles inside dyadic rectangles Q_k with k >= m_j + d_j, and such rectangles have diameter O(2^{-k}) <= C 2^{-m_j} under rho_{T^omega}, while m_j tends to infinity. However, this verification does not appear in the paper. Without it, neither the non-differentiation conclusion nor even the assertion that B is a differentiation basis is fully justified. This is an expositional gap, not a demonstrated counterexample.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies differentiation of integrals on the infinite-dimensional torus T^ω. Theorem 1.1 constructs, for every p0 ∈ [1,∞), Busemann–Feller bases B_≥ and B_> consisting of axis-parallel rectangles such that B_≥ differentiates L^p(T^ω) if and only if p ≥ p0 and B_> differentiates L^p(T^ω) if and only if p > p0. Theorem 1.3 classifies the possible sets diff(B) for arbitrary differentiation bases in complete metric measure spaces as one of six forms, and conversely realizes each form. The proof combines a monotonicity property (Proposition 2.1), a weak-type maximal operator criterion (Lemma 3.2 and Proposition 3.3), a construction of nested (ε,d)-configurations (Proposition 4.4), and a Borel–Cantelli argument. Corollary 1.2 transfers the examples to the unit interval with bases of finite unions of intervals. Several examples clarify measurability issues and the special role of p = ∞.","tokens_in":20775,"tokens_out":15531,"duration_ms":135345,"significance":"If the results are correct, this is a substantial contribution to differentiation theory. The construction overcomes the finite-dimensional obstruction to Busemann–Feller rectangle bases and recovers and extends Hayes' classical examples in a new setting. The classification theorem for complete spaces is complete and the converse constructions are explicit. The paper is well organized, makes good use of the authors' earlier sharp weak-type estimates, and presents the negative direction through a transparent quantitative divergence and the second Borel–Cantelli lemma. The only substantive gap is a missing verification in the contraction step of the proof of Theorem 1.1, which is easily repairable from the construction in Lemma 4.3.","major_comments":[{"comment":"To conclude f(x) ≥ 1 from the existence, for infinitely many j, of a rectangle Q_j in S_{j,n_j} containing x with Avg f(Q_j) ≥ 1, the proof must exhibit a sequence of sets from B(x) contracting to x. Condition (A3) supplies nestedness but not diameter decay, and the definition of contraction in Section 1.3 requires radii tending to 0. The needed estimate is implicit in Lemma 4.3: every rectangle in the configurations constructed there is contained in a dyadic rectangle of side length O(2^{-k}) with k ≥ m_j + d_j, so its diameter with respect to ρ_{T^ω} is O(2^{-m_j}), and m_j tends to infinity by the choice of m_j. Please add this verification explicitly; it is load-bearing for the non-differentiation conclusion and also for the assertion that B is a differentiation basis.","section":"Section 5, proof of Theorem 1.1, paragraph after (5.1)"}],"minor_comments":[{"comment":"The sentence \"let Q0 ∈ R0 be such that |Q0| ≤ 2^{-(d-1)^2-1} and so Q0 has at least d nonfree coordinates\" is imprecise: the condition alone does not logically force the existence of d nonfree coordinates without additional explanation from the cited construction.","section":"Section 4, Definition 4.1"},{"comment":"The mutual independence of the events F*_j is asserted as \"routine to check\"; since this independence is the key probabilistic ingredient in the Borel–Cantelli step, a short explanation of why the dyadic product structure yields independence would improve the exposition.","section":"Section 4, Proposition 4.4, condition (A4)"},{"comment":"The transfer argument to the unit interval is summarized as \"routine\"; a brief justification that the weak-type bounds and the Borel–Cantelli construction survive the transfer would help the reader.","section":"Section 5, proof of Corollary 1.2"},{"comment":"The measure µ_I on I is defined by setting the measure of a set to infinity when its intersection with I is uncountable; it would be useful to note explicitly that this gives a complete measure, since the completeness of X is used in the verification.","section":"Section 2, Example 2.3"},{"comment":"There are minor stylistic issues such as informal uses of \"iff\" and \"resp.\", and the choice of m_j in Theorem 1.1 is justified tersely; these do not affect the mathematics.","section":"Throughout"}],"recommendation":"minor_revision","confidential_remarks":"The only substantive issue is the missing contraction verification in the proof of Theorem 1.1. Once the authors add the short estimate from Lemma 4.3, the paper will be ready. I concur with the reader's positive assessment; the reliance on prior published results is appropriate, and the classification theorem is a noteworthy contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe paper is a solid piece of harmonic analysis. Its main achievements are the construction of Busemann–Feller bases of rectangles on the infinite-dimensional torus with an arbitrary prescribed differentiation threshold p0, and the classification of all possible ranges diff(B) for complete metric measure spaces into six shapes, with converses. The classification theorem is genuinely new and clean; the rectangle examples circumvent the finite-dimensional obstruction from strong maximal estimates, which is a nice point.\n\nThe proof strategy is mostly sound. Positive directions import a sharp weak-type estimate from prior work of the same group; that theorem is published and independently proved, so the importation is legitimate. The negative directions rest on a Borel–Cantelli argument using the independent events produced in Proposition 4.4, and the construction is explicit. The paper is also careful about measurability pathologies (Examples 2.5–2.7), a real virtue in this area.\n\nThe soft spot is the one flagged in the stress test. In the proof of Theorem 1.1, after defining f, the text asserts f(x) ≥ 1 for x in infinitely many selected configuration-unions. That inference requires a sequence of rectangles from the basis that actually contracts to x in the metric of T^ω. Condition (A3) gives nestedness but not diameter decay. The missing verification is available: Lemma 4.3 produces all rectangles inside dyadic rectangles of scale k, with k ≥ m_j + d_j, so their diameters are at most C 2^{-m_j}, and m_j → ∞. But that estimate is never written down. This is an expositional gap rather than a flaw; a referee should ask the authors to spell it out.\n\nThe reliance on earlier self-cited results for diff(R0), diff(R), and the weak-type bound is fine: those are published and independently argued. No circularity.\n\nThis paper deserves a serious referee. It is for anyone working on differentiation of integrals, maximal operators on infinite-dimensional tori, or Busemann–Feller bases. I would send it out, with a request to fill the contraction verification in Section 5.","headline":"Solid classification and sharp counterexamples in infinite-dimensional differentiation; one expositional gap around diameter decay, easily fixed.","tokens_in":21273,"tokens_out":3548,"would_cite":true,"duration_ms":32266,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["43A75","42B25"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every p0 ≥ 1, the paper constructs axis-parallel rectangle bases on the infinite-dimensional torus that differentiate L^p exactly when p ≥ p0 (or p > p0), and classifies all possible differentiation ranges in complete metric measure…","keywords":["differentiation basis","Lebesgue differentiation theorem","infinite-dimensional torus","Busemann–Feller basis","maximal operator","weak-type inequality","Rubio de Francia basis","complete metric measure space"],"falsifier":"Check whether, for the basis B_≥ constructed with d_j = j and ε_j = $j^{{-1/p0}}$/2, the nested rectangles Q_{j,n} containing a given x from condition (A3) have T^ω-diameters tending to 0 for almost every x. If for a positive-measure set of x these diameters stay bounded below, then the Borel–Cantelli step only proves f(x) ≥ 1 rather than producing a contracting sequence of averaging sets, and the claimed failure of differentiation for p < p0 would not follow from the written argument.","tokens_in":20272,"feed_emoji":"📐","tokens_out":7259,"duration_ms":64022,"temperature":0.7,"pith_summary":"This paper asks when the Lebesgue differentiation theorem—the statement that a function can be recovered as the limit of its averages over small neighborhoods—holds for a given family of averaging sets, and for which Lebesgue spaces L^p. The authors construct, in the infinite-dimensional torus, families of axis-parallel rectangles whose averages recover L^p functions precisely when p lies above a prescribed threshold p0, with either a strict or non-strict inequality. Because such rectangle bases cannot exhibit thresholds in any finite dimension, the infinite-dimensional setting is essential. The construction also yields a complete classification: in any complete metric measure space, the set of exponents for which a basis differentiates integrals must be one of six intervals (or the empty set), and each possibility is realized.","feed_headline":"Rectangular bases split L^p differentiation exactly at any p0","feed_subtitle":"A new construction realizes every possible L^p differentiation range in complete spaces, from none to all exponents above a threshold.","key_machinery":"The engine is the Rubio de Francia bases R0 and R on T^ω, together with what the paper calls (ε,d)-configurations: a rectangle Q0 together with d translates Qi that overlap Q0 in a fixed fraction ε of its measure. The paper quotes a sharp weak-type bound for the maximal operator of such a configuration, then builds, via a covering lemma and an inductive construction, a family S that tiles each rectangle by configurations, nests configurations across levels, and makes selected unions mutually independent. The independence condition lets the second Borel–Cantelli lemma force a specially built function f = sup $ε_j^{{-1}}$ 1_{F_j} to have a positive-measure set of points whose averages over shrinking rectangles stay at least 1, while f belongs to L^p for p below the threshold; the weak-type bound and a truncation argument give differentiability for p above it.","core_discovery":"The central claim is Theorem 1.1: for every p0 ∈ [1,∞), there exist Busemann–Feller bases B_≥ and B_> in the infinite-dimensional torus T^ω, whose elements are axis-parallel rectangles, such that B_≥ differentiates L^p(T^ω) if and only if p ≥ p0, and B_> differentiates L^p(T^ω) if and only if p > p0. Theorem 1.3 goes further: for any differentiation basis in a complete metric measure space, the set diff(B) of exponents p for which B differentiates L^p must be one of ∅, {∞}, [p0,∞], (p0,∞], [p0,∞), or (p0,∞), and each of these six forms is realized by a suitable complete space and basis. The paper also transfers the construction to the unit interval, producing Busemann–Feller bases of finite unions of intervals with the same L^p thresholds.","pith_inferences":["The unstated diameter check on the nested rectangles in the Borel–Cantelli step is worth making explicit: if the rectangles containing a point do not shrink to that point in the T^ω metric, the lower bound f(x) ≥ 1 would not translate into a failure of differentiation at x, so the argument as written depends on this property holding almost everywhere.","The classification suggests that in complete spaces, differentiation of L^p is monotone in p, with larger p easier; any failure of monotonicity would have to come from non-completeness, and the paper's examples indicate completeness is nearly necessary.","The same configuration machinery might produce bases with prescribed Orlicz-space differentiation ranges, since the weak-type estimates used here are quantitative and depend on the parameter ε.","The construction separates differentiation from maximal-operator boundedness: diff(B) can be a half-open interval while max(B) is closed, so the two notions are genuinely distinct."],"forward_implications":["Every possible differentiation range from the empty set to (p0,∞) is realized by a complete metric measure space, so the classification is sharp.","In any finite-dimensional torus, no Busemann–Feller rectangle basis can have a threshold p0 > 1, so such thresholds are an infinite-dimensional phenomenon.","The construction transfers to the unit interval, yielding Busemann–Feller bases of finite unions of open intervals with the same L^p differentiation thresholds.","For countable bases, the weak-type range of the associated maximal operator is contained in the differentiation range, and the gap between them can be controlled.","If the underlying space is a countable union of open sets with finite measure, only four of the six ranges (∅, {∞}, [p0,∞], (p0,∞]) can occur."],"supporting_citations":[{"why":"establishes diff(R0) = [1,∞] on the infinite-dimensional torus, the base case p0 = 1 and the starting point of the construction","marker":"[FR20]"},{"why":"proves diff(R) = ∅, used to realize the empty range (C1)","marker":"[Ko21]"},{"why":"supplies the sharp weak-type estimates for (ε,d)-configurations that drive the threshold calculations","marker":"[KRR23]"},{"why":"provides the maximal-operator analysis of intermediate Rubio de Francia bases on T^ω used in the inductive construction","marker":"[KMPRR23]"},{"why":"classical examples in the one-dimensional torus that the new rectangle bases reproduce and extend","marker":"[Ha52]"},{"why":"the strong maximal theorem in finite dimensions, which is the obstruction to rectangle bases with threshold p0 > 1","marker":"[CF75]"},{"why":"source of the Busemann–Feller formalism and the fact that density bases differentiate L∞","marker":"[dG75]"}],"fun_headline_variants":["Every L^p differentiation range realized for complete spaces","Six possible exponent sets for integral differentiation","Rectangular bases achieve any p0 threshold exactly","Classification of L^p differentiation ranges completed","Any p0 threshold possible for rectangular differentiation bases"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that a specially built function has a positive-measure set of points where averaging rectangles fail to converge relies on the nested rectangles that contain each point actually shrinking to that point in the metric of the infinite-dimensional torus; the paper does not explicitly verify this diameter condition.","fun_headline_variants_meta":{"raw":{"variants":["Every L^p differentiation range realized for complete spaces","Six possible exponent sets for integral differentiation","Rectangular bases achieve any p0 threshold exactly","Classification of L^p differentiation ranges completed","Any p0 threshold possible for rectangular differentiation bases"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000331,"raw_usage":{"total_tokens":1890,"prompt_tokens":1040,"completion_tokens":850,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":656,"completion_tokens_details":{"reasoning_tokens":781}},"tokens_in":656,"tokens_out":850,"duration_ms":7697,"temperature":1.0,"reasoning_tokens":781,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:06:49.572216+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check whether, for the basis B_≥ constructed with d_j = j and ε_j = $j^{{-1/p0}}$/2, the nested rectangles Q_{j,n} containing a given x from condition (A3) have T^ω-diameters tending to 0 for almost every x. If for a positive-measure set of x these diameters stay bounded below, then the Borel–Cantelli step only proves f(x) ≥ 1 rather than producing a contracting sequence of averaging sets, and the claimed failure of differentiation for p < p0 would not follow from the written argument.","supporting_citations":[],"review_version":1}