{"id":"21b14151-e393-4e3f-8579-ea1a9d6e3891","arxiv_id":"1908.04592","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For every compact set E in the real line with dim_A E > 0, there exist measures supported on E with any prescribed upper Assouad dimension D > dim_A E, and for any d < dim_L E, a measure with lower Assouad dimension d.","lead":"This paper proves that on any compact subset of the real line with positive upper Assouad dimension, one can construct a probability measure supported on the set whose upper Assouad dimension is any prescribed value larger than the set's, and similarly for the lower Assouad dimension. For fractal geometers, this settles a natural existence question about how extreme the local scaling of a measure can be.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.1 ends with 'standard arguments now show' but the omitted estimates (comparability of adjacent masses, bounded overlap of balls, propagation of alternating weights) are exactly what is needed to prove dim_L = d and dim_A = D; the written proof is incomplete.","rationale":"The reader's verdict of CONDITIONAL with moderate confidence is appropriate. The main body of the paper (Theorem 3.2) is detailed and, as far as I can verify, correct: the KRS construction, Lemma 3.1, Lemma 3.5, and the two cases of the proof all check out. The D=∞ case is handled separately and is also coherent. The genuine soft spot is the final step of Theorem 4.1, where the author writes 'standard arguments now show' without supplying the estimates needed to convert the local weight assignments into global Assouad dimension bounds. This is not a contradiction or an appeal to a false lemma, but it is an omission of a nontrivial argument. The missing estimates are the kind that can hide hidden assumptions about the uniformity of the measure across all points and all scales. However, there is no reason to believe the claim is false: the construction places all weights in [s^D, s^d], which strongly suggests the standard dyadic-cube arguments go through. Thus the paper should be accepted conditional on a referee or the authors filling in these details. I agree with the reader that this is the weakest assumption.","tokens_in":12808,"tokens_out":48048,"duration_ms":438445,"concrete_test":"Independently complete the proof of Theorem 4.1 by deriving, for arbitrary x in E and 0<r<R, the bounds C1 (R/r)^d <= µ(B(x,R))/µ(B(x,r)) <= C2 (R/r)^D. Specifically, verify two estimates: (i) a ball B(x,R) intersects at most two level-J intervals and at most N0(s) level-K intervals, and each partially-covered interval contributes a mass comparable to the full interval; (ii) along any nested path, the product of weights is at least (s^D)^{#levels} and at most (s^d)^{#levels}. If either estimate fails, construct a concrete counterexample (e.g., a ball centered near a transition between a d-block and a D-block) and check whether the ratio violates the required bound. If the derivation succeeds, the reader's concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is Theorem 4.1: simultaneous prescription of lower Assouad dimension d and upper Assouad dimension D. The proof reduces to a KRS construction with weights alternating between s^d and s^D along an interior path, then concludes 'Since there are arbitrarily long paths with weights s^d and s^D respectively, standard arguments now show that dim_L µ = d and dim_A µ = D.' This is the load-bearing step. The paper does not verify the technical estimates required for the equalities: (1) for every x in E and every 0<r<R, the ratio µ(B(x,R))/µ(B(x,r)) must be bracketed by constants times (R/r)^D and (R/r)^d, uniformly in x,R,r; (2) the mass of a ball that partially intersects a KRS interval must be comparable to the mass of the full interval, which requires a detailed analysis of how weights z_w, y_w, x_w, p are distributed among boundary and interior children; (3) the number of level-k intervals intersecting a ball of radius r must be bounded by a constant independent of k. Without these, the constructed measure cannot be certified to have the prescribed dimensions. The comparability of adjacent intervals is asserted in a sentence but not proven, and the propagation of the alternating path weights into global bounds is not derived. All earlier parts of the paper, including the upper-dimension construction (Theorem 3.2), Lemma 3.5 on the proportionality constant, and the KRS estimates in Lemma 3.1, appear internally consistent.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the problem of prescribing Assouad dimensions of measures with prescribed compact support E ⊆ R. It first proves (Theorem 3.2) that if dim_A E > 0, then for every D > dim_A E, including D = ∞, there is a probability measure µ with support E and dim_A µ = D. The proof uses the generalized nested cubes construction of Käenmäki–Rajala–Suomala (KRS), assigning weights along specially chosen paths and using a proportionality constant ζ for the density of splitting levels. The paper then proves (Theorem 4.1) that for 0 < d < dim_L E ≤ dim_A E < D there is a single measure µ on E with dim_L µ = d and dim_A µ = D, by alternating weights s^d and s^D along an interior path. It also gives an example (Proposition 3.8) of an infinite compact set on which every measure has upper Assouad dimension either 0 or ∞, showing that the hypotheses cannot be dropped.","tokens_in":13121,"tokens_out":10530,"duration_ms":106348,"significance":"If the results are fully established, they provide a clean and essentially sharp answer to a natural question about the attainable Assouad dimensions of measures on compact subsets of the line. The KRS-cube framework is well chosen, and the paper contains several genuinely useful ingredients: Lemma 3.1 gives quantitative control of the number of children in a KRS construction, Lemma 3.5 proves positivity of the proportionality constant under dim_A E > 0, and the upper-dimension construction in Theorem 3.2 is carefully structured. Proposition 3.8 is a valuable counterexample showing that the main theorem is not vacuous and that the positivity assumption on dim_A E is needed. The main advertised simultaneous statement, however, is not fully proved as written, because Theorem 4.1 ends with an omitted 'standard arguments' step that carries the load of the conclusion.","major_comments":[{"comment":"The proof of Theorem 4.1 ends with 'standard arguments now show that dim_L µ = d and dim_A µ = D', but the estimates needed for these equalities are not supplied. To establish dim_L µ = d one needs a uniform lower bound µ(B(x,R))/µ(B(x,r)) ≥ c (R/r)^d for every x ∈ E and all sufficiently small 0 < r < R, and to establish dim_A µ = D one needs the complementary uniform upper bound with exponent D. The written argument only asserts that adjacent intervals have comparable µ-measures and that there are arbitrarily long paths carrying weights s^d and s^D; it does not derive the corresponding bounds for balls that only partially intersect KRS intervals, does not control the number of level-k intervals intersecting a ball, and does not show how the alternating weights propagate to arbitrary x, r, R. Because the simultaneous prescription of both dimensions is the main advertised contribution of the theorem, this omission is load-bearing, and the proof is incomplete as written.","section":"§4, Theorem 4.1"}],"minor_comments":[{"comment":"In the case without arbitrarily long boundary paths, the displayed exponent ζ(1+2^{-J})(j-k) should be ζ(1+2^{-J})(k-j), since k > j and the path from Iv to Iw has length k-j; with the sign as written, the displayed lower bound for µ(Iw) is not meaningful.","section":"§3, proof of Theorem 3.2"},{"comment":"The sentence 'A similar statement holds for the measure s of any two siblings' should refer to the measure µ (or to the measures of the two siblings), not to 's'.","section":"§3, Lemma 3.3"},{"comment":"The bounds s^D ≤ y_w,z_w ≤ s^d are asserted with the comment 'it is easy to see', but the verification is not shown; since these bounds are used for the comparability of adjacent interval masses, a short derivation would improve the readability and verifiability of the proof.","section":"§4, Theorem 4.1"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version: the paper proves what it claims. For any compact E in [0,1] with positive upper Assouad dimension, and any D greater than dim_A E, there is a probability measure supported on E whose upper Assouad dimension is exactly D. Similarly for lower Assouad dimensions, and both can be prescribed simultaneously when the lower dimension of E is positive. This generalizes earlier results that only worked for discrete sets like {q^n} or produced dimensions arbitrarily close to the set's. The upper-dimension proof (Theorem 3.2) is carefully written and looks correct. The proportionality constant zeta extracted from the KRS construction is a genuinely nice device, and Proposition 3.8 gives a sharp counterexample showing the result cannot extend to all infinite compact sets.\n\nThe soft spot is Theorem 4.1. After building the measure with alternating blocks of weights s^d and s^D along an interior path, the proof stops with 'standard arguments now show' and a brief comparability claim for adjacent intervals. The stress-test is right that the global equalities dim_L mu = d and dim_A mu = D require uniform bounds over all centers and scales, not just along the chosen paths. But I think the gap is manageable: the construction fixes the minimum weight at s^D and the maximum at s^d, so any path of length L changes mass by at most (s^d)^L from above and at least (s^D)^L from below. Combined with the comparability of adjacent intervals, that yields the necessary bounds. It should still be written out; a referee should ask for it. There is also a sign typo in the exponent in the proof of Theorem 3.2 (where (j-k) appears, it should be (k-j)), which is confusing but harmless.\n\nBottom line: this is a solid, honest paper with a real result. I would send it to a careful referee, and I would cite it if I were working in the area.","headline":"Solid subfield result on prescribed Assouad dimensions of measures; main theorem holds up, but Theorem 4.1's last step is too compressed.","tokens_in":13669,"tokens_out":4979,"would_cite":true,"duration_ms":48798,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["28A78","28A80"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a compact set with positive upper Assouad dimension supports measures whose upper Assouad dimension can be any prescribed value larger than that of the set, and likewise for the lower Assouad dimension.","keywords":["Assouad dimension","regularity dimension","measures on fractals","nested cubes","doubling measures","compact sets","upper Assouad dimension","lower Assouad dimension"],"falsifier":"Test the boundary case by computing the upper Assouad dimension of E = {0} union {$\\alpha$^(M^n)} from Proposition 3.8. That proposition shows every measure on E has upper Assouad dimension 0 or infinity, so a positive value of dim_A E would contradict Theorem 3.2; the expected value, dim_A E = 0, would confirm that the theorem's positive-dimension hypothesis is genuinely needed rather than merely technical.","tokens_in":12599,"feed_emoji":"📐","tokens_out":9629,"duration_ms":100098,"temperature":0.7,"pith_summary":"The paper asks whether the Assouad dimension of a measure is forced by the set that carries it. Its answer is no: whenever a compact set E subset [0,1] has positive upper Assouad dimension, there is a probability measure supported on E whose upper Assouad dimension is exactly any prescribed value larger than that of E, including infinity. The same method yields, for any lower Assouad dimension below that of E, a measure on E with that lower dimension, and when both targets are positive a single measure can realize both prescribed dimensions simultaneously. The proof decomposes E into nested, dyadic-like intervals and chooses the proportion of mass assigned to each child interval so that the fastest ball-mass ratios have exactly the target exponent. A closing example shows that merely being compact and infinite is not enough: some infinite compact sets support only measures of upper Assouad dimension 0 or infinity.","feed_headline":"Sets carry measures of any larger Assouad dimension","feed_subtitle":"A nested-cube weighting scheme dials a measure's dimension to any prescribed value above its support's own.","key_machinery":"The carrying mechanism is the nested-cube construction (the KRS construction): a dyadic-like decomposition of E into nested intervals labelled by finite words, with lengths comparable to s^k and each interval splitting into at most N children. The measure is defined by assigning each child a weight, so that the masses of every sibling group sum to 1. The load-bearing object is the proportionality constant zeta of the construction, the limiting fraction of levels along arbitrarily long paths at which intervals split; for sets of positive upper Assouad dimension, zeta is positive. Setting the splitting weight to a = s^(D/zeta) makes the fastest mass decay along such paths equal s^(-D k), which translates to the desired exponent D in ball-mass ratios, while the boundedness of all other weights keeps every ratio at or below that exponent.","core_discovery":"The central claim is Theorem 3.2: if E subset [0,1] is compact and its upper Assouad dimension is positive, then for every D larger than that dimension there is a probability measure mu with support E and upper Assouad dimension exactly D. Theorem 4.1 is the simultaneous version: for 0 < d < dim_L E and D > dim_A E, a single measure can have lower Assouad dimension d and upper Assouad dimension D. The proof builds the measure by assigning weights to the children of each interval in a nested-cube decomposition of E. Along selected paths the weights are chosen so that mass decays with the target exponent; everywhere else the weights are bounded between a minimum and maximum that keeps every ball-mass ratio below that exponent. Thus the Assouad dimensions of a measure are not intrinsic to its support but are parameters the construction can dial in, subject only to the inequalities d < dim_L E and D > dim_A E.","pith_inferences":["Implicit in the paper's closing remark is a broader conjecture: the same weighting scheme may prescribe Assouad dimensions on any compact doubling metric space admitting a nested-cube construction with positive proportionality constant, even when the set's upper Assouad dimension is zero.","The proportionality constant zeta is introduced as a property of a particular construction; a natural question the paper leaves open is whether zeta is independent of the choice of nested-cube decomposition for a given set, and whether it connects to known homogeneity characteristics of E.","A direct test of the omitted estimates in Theorem 4.1 would be to track the mass ratio of adjacent intervals along a path whose prescribed weights alternate between s^d and s^D; the simultaneous statement is fully justified only if this ratio remains bounded independently of the path length."],"forward_implications":["For any compact E subset [0,1] with positive upper Assouad dimension, the possible upper Assouad dimensions of probability measures supported on E include every real number above dim_A E and also infinity.","If the lower Assouad dimension of E is positive, the two dimensions can be tuned independently: one measure can have any lower Assouad dimension d < dim_L E and any upper Assouad dimension D > dim_A E at the same time.","Measures of lower Assouad dimension 0 are easy to produce on any infinite support: add a point mass to any measure supported on E; the lower dimension drops to 0, and if the added point is not isolated, the upper dimension becomes infinity.","The positive-dimension hypothesis is essential: the paper constructs an infinite compact set E on which every measure has upper Assouad dimension either 0 or infinity, so no intermediate dimension is attainable."],"supporting_citations":[{"why":"Supplies the generalized nested-cube construction on which every measure in the paper is built.","marker":"[9]"},{"why":"Defines the upper regularity dimension framework and gives the q^n motivating example with dimension log p / log q that Theorem 3.2 generalizes.","marker":"[3]"},{"why":"Proves the earlier approximation result for doubling spaces that this paper sharpens from arbitrarily close to exact prescribed dimensions.","marker":"[11]"},{"why":"Provides the lower Assouad dimension analogue for doubling measures that motivates Theorem 4.1.","marker":"[1]"},{"why":"Gives the lower regularity and predetermined-regularity framework used for the lower-dimension half of the construction.","marker":"[7]"}],"fun_headline_variants":["Measures on a set can have any larger Assouad dimension","Assouad dimension of measures is tunable, given the set","For any D above dim_A E, a measure exists with dim_A = D","Your set's dimension doesn't limit your measure's dimension","Construct measures with prescribed Assouad dimensions on any set"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof of the simultaneous upper-and-lower result assumes that the chosen weights keep the masses of neighbouring construction intervals comparable all the way down every path, and the paper states the resulting dimension equalities as following by standard arguments without deriving those bounds.","fun_headline_variants_meta":{"raw":{"variants":["Measures on a set can have any larger Assouad dimension","Assouad dimension of measures is tunable, given the set","For any D above dim_A E, a measure exists with dim_A = D","Your set's dimension doesn't limit your measure's dimension","Construct measures with prescribed Assouad dimensions on any set"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001061,"raw_usage":{"total_tokens":4376,"prompt_tokens":800,"completion_tokens":3576,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":416,"completion_tokens_details":{"reasoning_tokens":3486}},"tokens_in":416,"tokens_out":3576,"duration_ms":26689,"temperature":1.0,"reasoning_tokens":3486,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:38:09.722137+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Test the boundary case by computing the upper Assouad dimension of E = {0} union {$\\alpha$^(M^n)} from Proposition 3.8. That proposition shows every measure on E has upper Assouad dimension 0 or infinity, so a positive value of dim_A E would contradict Theorem 3.2; the expected value, dim_A E = 0, would confirm that the theorem's positive-dimension hypothesis is genuinely needed rather than merely technical.","supporting_citations":[{"cited_title":"Jonsson, Besov spaces on closed subsets of R ^ n , Trans","cited_arxiv_id":null,"evidence_quote":"Supplies the generalized nested-cube construction on which every measure in the paper is built."},{"cited_title":"Bylund and J","cited_arxiv_id":null,"evidence_quote":"Defines the upper regularity dimension framework and gives the q^n motivating example with dimension log p / log q that Theorem 3.2 generalizes."},{"cited_title":"a enm \\","cited_arxiv_id":null,"evidence_quote":"Proves the earlier approximation result for doubling spaces that this paper sharpens from arbitrarily close to exact prescribed dimensions."},{"cited_title":"Assouad, U.E..R","cited_arxiv_id":null,"evidence_quote":"Provides the lower Assouad dimension analogue for doubling measures that motivates Theorem 4.1."},{"cited_title":"Quantum signature of exciton condensation","cited_arxiv_id":"1807.09191","evidence_quote":"Gives the lower regularity and predetermined-regularity framework used for the lower-dimension half of the construction."}],"review_version":1}