Pith. sign in

REVIEW 2 major objections 4 minor 26 references

The sequence property for fractal dimensions

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For many fractal dimensions, a compact set contains a convergent sequence of the same dimension.

desk verdict The sequence property is proved for genuinely new classes (Assouad, intermediate, profiles) with clean compactness/capacity arguments; the only real gap is Corollary 7.7, whose simultaneous profile claim needs continuity or a weaker statement. read the letter →

arxiv 2608.06051 v1 pith:CXO5SBVT submitted 2026-08-06 math.MG

Kenneth Falconer , Yuyang Liu This is my paper · ORCID
classification math.MG MSC 28A8028A78
keywords sequencepropertyfractaldimensionupperboxAssouadspectrumintermediatedimensionsprofilesfinitestabilitypoints
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

For a whole family of fractal dimensions—upper box dimension, Assouad dimension, the Assouad spectrum, upper intermediate dimensions, and box-dimension profiles—the paper proves that every compact set contains a convergent sequence of points whose dimension equals the dimension of the whole set. The proof rests on a localisation lemma: for any dimension that is monotone and finitely stable, some point $x$ of a compact set $E$ has the property that every ball centred at $x$ meets $E$ in a set of full dimension. The same machinery decides when one sequence can witness several dimensions at once: if $E$ is dimension-homogeneous, meaning every ball centred in $E$ contains a bi-Lipschitz copy of $E$, then a single convergent sequence has the same intermediate dimensions, Assouad spectrum, and dimension profiles as $E$ itself. The paper also constructs sets where no single sequence can witness two different dimensions, so the homogeneity condition is not merely technical.

What carries the argument

The load-bearing object is the dimension point: a point $x$ in a compact set $E$ such that $\dim(E \cap B(x,r)) = \dim E$ for every $r > 0$. Lemma 2.5 proves such a point exists for every monotone, finitely stable dimension, so the whole construction reduces to placing finite sets of near-full dimension inside the nested balls $B(x, 2^{-n})$ and enumerating their union. A second mechanism, Lemma 7.4, interleaves countably many sequences converging to the same point into one sequence that contains a tail of each, which is what allows one sequence to witness many dimensions at once. The dimension-homogeneous hypothesis—every ball centred in $E$ contains a bi-Lipschitz image of $E$—makes every point of $E$ a common dimension point for all the dimensions considered. Two auxiliary tools carry the individual cases: a capacity estimate showing a finite subset can capture a fixed fraction of a compact set's capacity at a given scale, and the Hausdorff-metric continuity of intermediate pre-measures, which lets dense finite subsets approximate the pre-measure of the whole set.

What would settle it

Run the construction of Corollary 7.6 on a self-affine carpet whose intermediate dimension function is known at several parameter values, producing the sequence $(a_i)$, and numerically estimate $\dim^\theta\{a_i\}$ at those values; the theorem predicts exact equality at every $\theta$, so any strict shortfall refutes the simultaneous claim. For the single-dimension theorem, take a compact set with known upper box dimension, choose the maximal $r_n$-separated sets used in Theorem 3.2, and check that the enumerated union has exactly that upper box dimension.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that monotone, finitely stable dimensions with the finite-set property satisfy a sequence property: for every non-empty compact $E$ with $\dim E > 0$ there is a point $x \in E$ and a sequence $(a_k) \subset E$ with $a_k \to x$ and $\dim(a_k) = \dim E$. Lemma 2.5 guarantees that every compact set has a dimension point; the proofs then place finite point sets $P_n \subset B(x, 2^{-n})$ whose $n$-scale dimension is within $1/n$ of the full dimension, enumerate their union as a sequence, and use finite stability to discard finitely many extraneous points. For upper box dimension the finite sets are maximal $r_n$-separated sets; for Assouad dimension and the Assouad spectrum they come from the failure of the defining covering inequality; for intermediate dimensions a Hausdorff-metric continuity lemma for pre-measures produces them; and for dimension profiles a capacity lemma does the same. The final section interleaves the individual witnessing sequences with a tail-combining lemma, so a dimension-homogeneous set has one convergent sequence realising $\dim^\theta E$ for all $\theta \in (0,1]$ and, in $\mathbb{R}^n$, realising every profile $\dim_B^s E$ for $0 < s \le n$.

Load-bearing premise

The simultaneous witnessing results assume the set is dimension homogeneous—every ball centred in $E$ contains a bi-Lipschitz copy of $E$—so every point of $E$ is a common dimension point for all the dimensions under consideration; without such a common point the paper's examples show that simultaneous witnessing can fail.

Editorial extensions

If this is right

  • Upper box dimension, Assouad dimension, each Assouad spectrum level, each upper intermediate dimension $\theta \in (0,1]$, and each box-dimension profile $\dim_B^s$ satisfy the sequence property on compact sets.
  • Hausdorff and packing dimensions cannot have the sequence property because they are countably stable, and lower box dimension fails because it is not finitely stable; the paper locates the dividing line at finite stability.
  • For dimension-homogeneous compact sets, including self-similar, self-affine and graph-directed iterated-function-system attractors, one convergent sequence simultaneously realises every intermediate dimension $\dim^\theta E$ and every Assouad spectrum value of $E$.
  • In $\mathbb{R}^n$, for such sets a single sequence also realises every profile $\dim_B^s E$ for $0 < s \le n$, so the projections of the sequence have almost-surely the same dimensions as the corresponding projections of $E$.
  • For intermediate dimensions, even without a common dimension point a countable subset of $E$ can share $\dim^\theta E$ for all $\theta \in (0,1]$ by taking a countable dense set of parameter values and using continuity.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The authors leave implicit that the same construction scheme should apply to any future monotone, finitely stable dimension with the finite-set property; the proofs for box, Assouad, intermediate and profile dimensions read as templates rather than isolated facts.
  • A natural test of the limits of the simultaneous result is to ask whether the bi-Lipschitz local structure in the dimension-homogeneity hypothesis can be relaxed to mere existence of a common dimension point; the counterexamples show some condition is needed, but not that full homogeneity is minimal.
  • Because a single sequence encodes the whole intermediate-dimension function of a homogeneous fractal, one could try to turn the theorem into a numerical recipe: sample the constructed sequence, estimate $\dim^\theta$ on the finite point set at several $\theta$, and use continuity to interpolate across the spectrum.
  • The sequence property gives a way to think of countable sets as dimension carriers for box-like dimensions, in sharp contrast to Hausdorff dimension, where every countable set has dimension zero; this may clarify how finite stability rather than countability controls the dimension of point sets.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 4 minor

Summary. The paper introduces and studies the 'sequence property' for several fractal dimensions: a dimension has this property if every compact set of positive dimension contains a convergent sequence with the same dimension. The authors prove the sequence property for upper box dimension (Theorem 3.2), Assouad dimension (Theorem 4.2), the Assouad spectrum (Theorem 4.4), upper intermediate dimensions in locally compact spaces (Theorem 5.2), and box dimension profiles in R^n (Theorem 6.4). The main technical device is a localisation lemma (Lemma 2.5) showing that finitely stable dimensions always have a 'dimension point' x in every compact set, around which the set has full dimension at all scales. The paper also treats simultaneous representation by a single sequence: Corollaries 7.5 and 7.6 show that when several dimensions share a common dimension point, one sequence can witness all of them, provided the relevant functions of the parameter (θ or ϑ) are continuous. Corollary 7.7 makes a similar claim for dimension profiles, and Examples 7.1-7.3 show that without a common dimension point simultaneous representation can fail.

Significance. If the results are correct, they give a unified and largely constructive explanation of when a countable point set can have the same fractal dimension as a compact set, extending earlier work of Ivanov on upper box dimension. The localisation lemma and the merge lemma for sequences are clean and reusable tools, and the negative examples in Section 7 are valuable because they show that the 'common dimension point' condition is genuinely necessary. The use of intermediate pre-measure continuity (Lemma 5.1) and of capacity-based finite subsets (Lemma 6.3) gives the proofs a solid analytic grounding. The main caveat is that the simultaneous result for dimension profiles, Corollary 7.7, is not proved as written because it relies on an unstated continuity property in the profile parameter s.

major comments (2)
  1. [Corollary 7.7] The proof of Corollary 7.7 is not complete. It is dismissed as 'similar' to Corollary 7.6, but Corollary 7.6 crucially uses the known continuity of θ ↦ dim^θ F and ϑ ↦ dim^ϑ_A F to pass from equality on a countable dense set of parameters to equality for all parameters. For dimension profiles, Section 6 establishes monotonicity and finite stability of s ↦ dim^s_B F, but it neither states nor proves continuity of s ↦ dim^s_B F for a fixed compact set F, nor does it cite such a result. Consequently, applying Lemma 7.4 and Corollary 7.5 over a countable dense set of s-values yields equality only on that dense set. Monotonicity alone does not suffice: two monotone functions can agree on a dense set and differ at a jump. The statement of Corollary 7.7 therefore requires an additional continuity proof or a direct construction that avoids the dense-set argument; otherwise the claim is unproved as written.
  2. [Section 7, after Corollary 7.5] The use of the 'dimension homogeneous' hypothesis for dimension profiles is under-justified. The definition says every ball B(x,r) with x in E contains a bi-Lipschitz image of E, and the text then asserts that this makes every point a dimension point 'for a wide range of dimension definitions'. For Corollary 7.7 one needs specifically that dim^s_B(E∩B(x,r)) = dim^s_B E. This requires that dim^s_B is invariant under bi-Lipschitz embeddings and that a bi-Lipschitz copy of E contained in B(x,r) has the same profile as E. Section 6 defines dim^s_B via capacities and proves finite stability and the sequence property, but it does not state or prove bi-Lipschitz invariance of dim^s_B, nor is a reference given. A citation or a short proof is needed for this load-bearing step in Corollary 7.7.
minor comments (4)
  1. [Section 6, Theorem 6.4 proof] The first sentence reads 'Let E⊂R^n be compact with dim^s_B = t > 0'; it should read 'with dim^s_B E = t > 0'.
  2. [Example 7.3 proof] There is a garbled duplicated phrase in the proof: 'As in Example, 7.2, if (a_k)⊂E is a convergent sequence with a_k → x∈E then either a_k ∈ E_1 for all sufficiently large k so Let (a_k)_k ⊂E with a_k →x∈E. If x∈E_1...' The sentence needs to be rewritten.
  3. [Section 6.1] The heading 'Propostition 6.2' contains a typo and should read 'Proposition 6.2'.
  4. [Corollary 7.6(i)] The proof invokes Theorem 5.2 for θ_j in a countable dense subset of (0,1], but Theorem 5.2 is stated only for θ∈(0,1). The endpoint θ=1 is covered by Theorem 3.2, and the continuity cited from [12] is on (0,1]; this endpoint issue should be stated explicitly so the reader knows how θ=1 is included.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the sequence-property theorems are derived directly from definitions and standard external results, with at most a non-circular proof gap in Corollary 7.7.

full rationale

The paper's central claims are not circular. The individual sequence-property results (Theorems 3.2, 4.2, 4.4, 5.2, 6.4) are proved constructively from the definitions of the relevant dimensions, using only monotonicity/finite stability, the general dimension-point Lemma 2.5, and standard external results (e.g., capacity subadditivity, weak density of finitely supported measures, Hausdorff-metric continuity of intermediate pre-measures). No fitted parameter is renamed as a prediction, and no cited theorem assumes the sequence property. The simultaneous-representation results rest on the elementary merging Lemma 7.4 and Corollary 7.5, then pass from a countable dense set of parameter values to all values using externally cited continuity results: continuity of theta-mapsto-dim^theta F for intermediate dimensions (Falconer-Fraser-Kempton) and continuity of the Assouad spectrum. Those continuity facts are independent of the sequence property and are not equivalent to the conclusions. The only substantive issue is a proof gap rather than circularity: Corollary 7.7 says 'The proof is similar to that of Corollary 7.6', but the needed continuity of s-mapsto-dim^s_B F in s is neither stated nor derived, so equality on a countable dense set of s-values would not automatically extend to all 0<s<=n. This is an omitted proof or citation, not a self-referential reduction, and it does not undermine the independent construction in Theorem 6.4. Overall, the derivation chain is self-contained against the stated assumptions and external benchmarks; no circular step was found.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No free parameters and no invented entities are introduced. The axioms are either standard mathematical background or external results cited from the literature. The central new content is the derivation of sequence properties from these axioms.

assumptions (6)
  • domain assumption The dimension notions considered (upper box, Assouad, Assouad spectrum, upper intermediate, dimension profiles) are monotone, finitely stable, and vanish on finite sets.
    Used throughout Sections 2-6 to guarantee dimension points (Lemma 2.5) and the sequence constructions.
  • domain assumption Intermediate dimensions can be defined on locally compact metric spaces and satisfy the pre-measure characterization (5.3)-(5.4).
    Needed in Section 5.1; the authors rely on Banaji [2] for the metric-space extension.
  • standard math The Hausdorff metric space C(X) of non-empty compact subsets of a compact X is compact.
    Used in Lemma 5.1 to extract convergent subsequences of covers.
  • domain assumption Rutar's characterization of attainable Assouad spectra guarantees existence of the set E_2 in Example 7.1.
    Cited external result [25], not proved in the paper.
  • domain assumption The intermediate dimension of the sequence set {2 + n^{-1/3}} union {2} is 3theta/(3theta+1), and intermediate dimensions are continuous in theta on R^n.
    Used in Example 7.2 and Corollary 7.6; cited from [11,12].
  • domain assumption There exist compact sets with prescribed box dimensions and projection dimensions realizing the extremes in (6.1).
    Used in Example 7.3; cited from [9,13,19].

how reviews work

0 comments
Cite this review

Pith. "Pith review of The sequence property for fractal dimensions." pith.science (2026). https://pith.science/paper/CXO5SBVT

@misc{pith2026260806051,
  author       = {Pith},
  title        = {Pith review of: The sequence property for fractal dimensions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CXO5SBVT}},
  note         = {Machine review of arXiv:2608.06051}
}
abstract

For various definitions of fractal dimension that are finitely stable, including upper box dimension, upper intermediate dimensions and Assouad spectra, we show that, given a compact subset $E$ of a metric space, typically $R^n$, there is a convergent sequence of points contained in $E$ of the same dimension as $E$ itself. Moreover, under certain conditions it is possible for a sequence to witness the dimension of $E$ for many definitions of dimension simultaneously, for example in a self-affine set $E$ there is a single convergent sequence that has the same Assouad spectrum or intermediate dimension values as $E$ itself.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 25 canonical work pages

  1. [18]

    A.V. Ivanov. On the intermediate values of the box dimensions,Sib. Math. J. 64(2023), 593–597

    work page 2023

  2. [1]

    P. Assouad. ´Etude d’une dimension m´ etrique li´ ee ` a la possibilit´ e de plongements dansR n, C. R. Acad. Sci. Paris S´ er. A-B,288(1979), 731-734

    work page 1979

  3. [2]

    A. Banaji. Generalised intermediate dimensions,Monatsh Math.202(2023) 465– 506

    work page 2023

  4. [3]

    Banaji and I

    A. Banaji and I. Kolossv´ ary. Intermediate dimensions of Bedford-McMullen car- pets with applications to Lipschitz equivalence,Adv. Math.449(2024), no. 109735

    work page 2024

  5. [4]

    Burrell, K.J

    S. Burrell, K.J. Falconer and J. Fraser. Projection theorems for intermediate dimensions,J. Fractal Geom.8(2021), 95–116

    work page 2021

  6. [5]

    C. Chen, M. Wu and W. Wu. Accessible values for the Assouad and lower di- mensions of subsets,Real Anal. Exchange45(2020), 85–99

    work page 2020

  7. [6]

    Du Plessis.An Introduction to Potential Theory, Oliver and Boyd, Edinburgh, 1970

    N. Du Plessis.An Introduction to Potential Theory, Oliver and Boyd, Edinburgh, 1970

    work page 1970

  8. [7]

    Edgar.Measure, Topology and Fractal Geometry, Springer-Verlag, New York, 1990

    G.A. Edgar.Measure, Topology and Fractal Geometry, Springer-Verlag, New York, 1990

    work page 1990

Show all 26 references
  1. [8]

    Falconer.Fractal Geometry: Mathematical Foundations and Applications, John Wiley & Sons, Hoboken, NJ, 3rd

    K.J. Falconer.Fractal Geometry: Mathematical Foundations and Applications, John Wiley & Sons, Hoboken, NJ, 3rd. ed., 2014

    2014

  2. [9]

    Falconer

    K.J. Falconer. A capacity approach to box and packing dimensions of projections and other images, inAnalysis, Probability and Mathematical Physics on Fractals, pp 1–19, P.A. Ruiz, J.P. Chen, L.G. Rogers, R.S. Strichartz & A. Teplyaev (eds), World Scientific, Singapore, 2020

    2020

  3. [10]

    Falconer

    K.J. Falconer. A capacity approach to box and packing dimensions of projections of sets and exceptional directions,J. Fractal Geom.8(2021), 1–26

    2021

  4. [11]

    Falconer

    K.J. Falconer. Intermediate Dimensions: A Survey, inThermodynamic Formal- ism, pp 469–493, M. Pollicott and S. Vaienti (eds)Lecture Notes in Mathematics 2290, Springer, Cham, 2021

    2021

  5. [12]

    Falconer, J.M

    K.J. Falconer, J.M. Fraser and T. Kempton. Intermediate dimensions,Math. Z. 296(2020), 813–830. 17

    2020

  6. [13]

    Falconer and J.D

    K.J. Falconer and J.D. Howroyd. Projection theorems for box and packing di- mensions,Math. Proc. Cambridge Philos. Soc.119(1996), 287–295

    1996

  7. [14]

    Falconer and Shuqin Zhang

    K.J. Falconer and Shuqin Zhang. Frostman and Fourier characterisations of frac- tal dimensions,Math. Z.(2026), 312.79

    2026

  8. [15]

    Fraser.Assouad Dimension and Fractal Geometry, Cambridge University Press, Cambridge, 2021

    J.M. Fraser.Assouad Dimension and Fractal Geometry, Cambridge University Press, Cambridge, 2021

    2021

  9. [16]

    J.M. Fraser. Fractal geometry of Bedford-McMullen carpets, inThermodynamic Formalism, 495–516,Lecture Notes in Math.2290, Springer, Cham, 2021

    2021

  10. [17]

    Fraser and H

    J.M. Fraser and H. Yu. New dimension spectra: finer information on scaling and homogeneity,Adv. Math.329(2018), 273–328

    2018

  11. [19]

    J¨ arvanp¨ a¨ a

    M. J¨ arvanp¨ a¨ a. On the upper Minkowski dimension, the packing dimension, and othogonal projections,Ann. Acad. Sci. Fenn. A Dissertat.99(1994)

    1994

  12. [20]

    Marstrand

    J.M. Marstrand. Some fundamental geometrical properties of plane sets of frac- tional dimensions,Proc. London Math. Soc.(3)4(1954), 257–302

    1954

  13. [21]

    P. Mattila. Hausdorff dimension, orthogonal projections and intersections with planes,Ann. Acad. Sci. Fenn. A Math.1(1975), 227–244

    1975

  14. [22]

    P. Mattila. Hausdorff dimension, projections, and the Fourier transform,Publ. Mat.48(2004), 3–48

    2004

  15. [23]

    Mitchell and L

    A. Mitchell and L. Olsen. Coincidence and noncoincidence of dimensions in com- pact subsets of [0,1], arXiv 1812.09542

    arXiv

  16. [24]

    Parthasarathy.Probability Measures on Metric Spaces, Academic Press, New York and London, 1967

    K.R. Parthasarathy.Probability Measures on Metric Spaces, Academic Press, New York and London, 1967

    1967

  17. [25]

    A. Rutar. Attainable forms of Assouad spectra,Indiana Univ. Math. J.73 (2024), 1331–1356

    2024

  18. [26]

    D.W. Spear. Sets with different dimensions in [0,1],Real Anal. Exchange24 (1998/99), 373–389. Kenneth Falconer Mathematical Institute, University of St Andrews, St Andrews, Fife, KY16 9SS, UK kjf@st-andrews.ac.uk Yuyang Liu Mathematical Institute, University of St Andrews, St ...

    1998

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.