Pith. sign in

REVIEW 1 major objections 8 minor 29 references

Microscopic weighting: a canonical measure for metric spaces

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · glm-5.2

2026-07-07 15:39 UTC pith:UW7JBYLV

load-bearing objection New canonical invariant of finite metric spaces; clean proofs; one gap in Theorem 3.1 that doesn't affect the main results. the 1 major comments →

arxiv 2607.05349 v1 pith:UW7JBYLV submitted 2026-07-06 math.MG math.CO

The microscopic weighting on a metric space

Emily Roff , Simon Willerton This is my paper
classification math.MG math.CO MSC 51F9905C50
keywords magnitude of metric spacesnegative typedistance energySchoenberg embeddingweightingconcentrationmicroscopic weightingdistance matrix
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper introduces the microscopic weighting, a canonical signed measure of total mass one that can be attached to almost any finite metric space by taking the small-scale (t→0) limit of the weightings used to define the magnitude function. The authors prove that every finite metric space of strictly negative type admits such a weighting—this covers all finite subsets of Euclidean space, hyperbolic space, and finite trees. They then show that when it exists, the microscopic weighting admits two complementary geometric interpretations: it specifies the circumcentre of the Schoenberg polytope (a convex polytope canonically associated to the space), and it is the unique energy-maximizing signed measure for the distance-energy integral I. In particular, the maximal energy M(X) equals the derivative at zero of the magnitude function, μ'_X(0), tying together two previously unrelated invariants. The paper also begins extending the theory to compact subsets of Euclidean space, where the microscopic weighting must be understood as a Schwartz distribution rather than a measure, and verifies key analogues for the three-dimensional unit ball.

Core claim

The central discovery is that the small-scale limit of magnitude weightings, when it exists, is a gauging for the distance matrix D with concentration equal to μ'_X(0). For spaces of strictly negative type this limit always exists and equals con(D)·D^{-1}·1. This single object simultaneously encodes the circumcentre of the Schoenberg polytope, the energy-maximizing measure for the distance-energy integral, and the derivative of the magnitude function at zero—three facets of one structure. The bridge between them is the identity M(X) = μ'_X(0), proved for all finite spaces of strictly negative type.

What carries the argument

The argument runs through three linked objects: the similarity matrix Z(t) = exp_⊙(-tD), whose weightings w(t) = Z(t)^{-1}·1 are studied as t→0; the distance matrix D and its gaugings (vectors v with 1^T v = 1 and Dv = c·1, where c is the concentration); and the energy integral I(ν) = ∫∫ d(y,y') dν(y) dν(y'). The key algebraic fact is Theorem 2.7: magnitude and concentration are reciprocal, mag(A) = 1/con(A). The existence proof for strictly negative type spaces uses the invertibility of the bordered matrix N = [[0, 1^T],[1, D]] (Proposition 4.7) and the resulting characterization: strictly negative type ⟺ det(D)≠0 and con(D)≠∞.

Load-bearing premise

The extension to compact subsets of Euclidean space defines the microscopic weighting as a weak-* limit of distributions in a Bessel potential space. This definition and the verification for the three-dimensional ball require careful handling of function-space duality and normal derivatives. The general compact theory is incomplete: gaugings and concentration are not yet defined for compact spaces, and the equality μ'_Y(0) = M(Y) is verified only numerically for the first 21-

What would settle it

Find a finite metric space of strictly negative type whose distance matrix has finite concentration but for which the limit lim_{t→0} w(t) does not exist, or for which the maximizing measure for I does not equal con(D)·D^{-1}·1. Alternatively, find an odd-dimensional ball B^{2p+1} for p ≥ 21 where μ'_{B^{2p+1}}(0) ≠ M(B^{2p+1}).

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • The identity M(X) = μ'_X(0) connects the magnitude function (a categorically motivated invariant) with the distance-energy integral (a classical potential-theoretic quantity), potentially allowing tools from either domain to transfer.
  • The microscopic weighting provides a canonical, scale-independent measure on any finite metric space of strictly negative type, which could serve as a principled weight vector for boundary detection, diversity quantification, or other applications where an ad hoc choice was previously needed.
  • For compact Euclidean sets where no energy-maximizing measure exists (like B^3), the microscopic weighting as a distribution may fill the role of an energy-maximizing distribution, extending the variational interpretation beyond the finite setting.
  • The Schoenberg-polytope interpretation gives a concrete geometric meaning to the microscopic weighting: it is the barycentric coordinate vector of the circumcentre, making the abstract limiting procedure visually and computationally accessible.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If Conjecture 6.2 (μ'_Y(0) = M(Y) for all compact spaces of strictly negative type) holds, it would imply that the derivative of magnitude at zero is a universal potential-theoretic invariant, calculable by either magnitude-theoretic or energy-optimization methods.
  • The fact that the microscopic weighting on B^3 is supported entirely on the boundary S^2 suggests that for compact Euclidean regions, the microscopic weighting may generically concentrate on the boundary, providing a distributional analogue of boundary detection observed in finite approximations.
  • The connection to resistance curvature on graphs (Theorem 5.12) hints that microscopic weightings may recover or relate to other discrete curvature notions on graphs equipped with different metrics, potentially unifying several ad hoc curvature definitions under one limiting procedure.
  • The conjecture that finite concentration of D is the only obstruction to existence of a microscopic weighting (Conjecture 3.3) would, if true, give a purely algebraic criterion for when the magnitude function is differentiable at zero.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

1 major / 8 minor

Summary. The paper introduces the microscopic weighting on a finite metric space—the small-scale limit of the weightings used to define the magnitude function—and proves that it exists for every finite space of strictly negative type (including all finite subsets of Euclidean and hyperbolic space, and all finite trees). The microscopic weighting, when it exists, is shown to be the unique gauging for the distance matrix, with concentration equal to the derivative of the magnitude function at zero. Three geometric interpretations are provided: via the Schoenberg polytope (circumcentre in barycentric coordinates), as an energy-maximizing measure for the distance-energy integral, and—in the case of graphs with resistance distance—as recording resistance curvature. The paper also lays groundwork for extending the theory to compact subsets of Euclidean space, where the microscopic weighting must be understood as a distribution rather than a measure; this is verified for the three-dimensional ball. The central results for finite spaces of strictly negative type rest on a power-series argument (Theorem 3.8) and standard linear algebra combined with the Hjorth–Lisoň–Markvorsen–Thomassen characterization of strictly negative type (Proposition 4.7).

Significance. The paper makes a substantial contribution to the theory of magnitude of metric spaces. The microscopic weighting is a natural and canonical invariant, and the paper provides multiple complementary viewpoints on it. The identification M(X) = μ'_X(0) (Theorem 5.10) connecting maximal distance-energy with the derivative of magnitude is a particularly clean result. The extension to compact Euclidean subsets via Meckes' distributional framework, while explicitly conjectural in part, is a sensible first step and the verification for B³ (Theorems 6.9–6.10) is a nice concrete result. The connection to resistance curvature (Theorem 5.12) is a pleasing bonus. The proofs are clean and well-structured throughout. The main results are falsifiable and parameter-free in the sense that they identify a specific vector with specific geometric content.

major comments (1)
  1. The proof of Theorem 3.1, specifically the step deriving equation (3.2) and the subsequent limit-taking, factors lim_{t→0} [Z(t)w'(t)] as lim Z(t) · lim w'(t). This requires w'(t) to converge as t→0, which is not established in the proof. As the stress-test note observes, Z(t) approaches the singular matrix 11^T, so convergence of w'(t) is not automatic. The authors should either add a hypothesis (e.g., D invertible, where analyticity from Theorem 3.8 applies) or note that the argument is valid under the conditions of the main results. Since Theorems 4.11 and 5.10 operate entirely in the invertible-D regime where w(t) is analytic at t=0, the central claims are not affected, but the statement of Theorem 3.1 as written is more general than what the proof establishes.
minor comments (8)
  1. The abstract states the microscopic weighting can be associated to 'almost any' finite metric space. It would help the reader to state more precisely what 'almost any' means here (e.g., generic in the sense of Roff–Yoshinaga [25], or equivalently con(D) ≠ ∞ and det(D) ≠ 0).
  2. In Example 2.9, the formula for con(D(a,b)) has a special case at (a,b) = (3/2, 2) where con = 1, but the 'otherwise' formula gives 2(9/4 - 3)/(6+6-12) = 0/0. A brief note that this is a 0/0 indeterminate form resolved by direct computation would help the reader.
  3. Figure 1 (right panel) is somewhat hard to parse: the meaning of the shaded region, the red dotted line, and the blue dashed curve are explained in the text but not labeled in the figure itself. Adding labels or a legend would improve readability.
  4. In the proof of Proposition 5.9, the Lagrange multiplier argument shows that a maximizer must be a gauging, but the converse (that a gauging is a maximizer) uses the negative type inequality. The logic is correct but the two directions are interleaved; separating them more clearly would aid readability.
  5. Section 6.1: the formula μ'_{B_{2p+1}}(0) = ∏_{i=1}^p 2i/(2i-1) is stated as verified for p = 0,...,20 using SageMath code from [29]. It would be appropriate to state whether this is a proven identity or a numerical observation, and if the latter, what precision was used.
  6. The term 'gauging' (Definition 2.3) is introduced as new terminology. While the concept appears in the literature under other names (e.g., 'd-invariant measure' in Nickolas–Wolf), a brief remark on the relationship to existing terminology in Section 5.2 would help readers familiar with that literature.
  7. In Theorem 6.9, the conditions on h and k_b (that h ≡ 1 on a neighbourhood of B³, and k_b = d(b,−) on a neighbourhood of S²) are somewhat involved. Stating more explicitly why these regularity conditions are needed (and whether they are artifacts of the proof or essential) would improve readability.
  8. Reference [25] (Roff–Yoshinaga) is cited as providing the power series technique adapted in Lemma 3.7. Since one of the authors of the present paper is also an author of [25], the relationship between the results here and those in [25] could be stated more explicitly, particularly regarding what is genuinely new in the adaptation from magnitude to weightings.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for a careful and constructive report. The referee's single major comment identifies a genuine gap in the proof of Theorem 3.1, which we will fix. The central results of the paper are unaffected.

read point-by-point responses

  1. Referee: The proof of Theorem 3.1, specifically the step deriving equation (3.2) and the subsequent limit-taking, factors lim_{t→0} [Z(t)w'(t)] as lim Z(t) · lim w'(t). This requires w'(t) to converge as t→0, which is not established in the proof. As the stress-test note observes, Z(t) approaches the singular matrix 11^T, so convergence of w'(t) is not automatic. The authors should either add a hypothesis (e.g., D invertible, where analyticity from Theorem 3.8 applies) or note that the argument is valid under the conditions of the main results. Since Theorems 4.11 and 5.10 operate entirely in the invertible-D regime where w(t) is analytic at t=0, the central claims are not affected, but the statement of Theorem 3.1 as written is more general than what the proof establishes.

    Authors: The referee is entirely correct. The proof of Theorem 3.1 takes the limit of equation (3.2) by factoring lim_{t→0} [Z(t)w'(t)] as (lim Z(t))·(lim w'(t)), which requires the convergence of w'(t) as t→0. The existence of lim_{t→0} w(t) alone does not guarantee this, and since Z(t) → 11^T is singular, the convergence of w'(t) is not automatic. We acknowledge this gap. We will revise the paper as follows. We will add the hypothesis that the derivative w'(t) converges as t→0 to the statement of Theorem 3.1, and note explicitly that this hypothesis is satisfied whenever the distance matrix D is invertible and has finite concentration, by the analyticity argument of Theorem 3.8. Since Theorems 4.11, 5.10, and 5.12 all operate in the invertible-D regime (where w(t) is analytic at t=0 by the power-series expansion in Lemma 3.7 and Theorem 3.8), the central results of the paper are unaffected. We will also add a remark after Theorem 3.1 clarifying that the hypothesis is automatically satisfied in the settings of interest, and that Conjecture 3.3, if true, would remove the need for the additional hypothesis. revision: yes

Circularity Check

0 steps flagged

No significant circularity found; the derivation chain is self-contained with one minor, non-load-bearing self-citation.

full rationale

The paper's central results (Theorems 3.8, 4.11, 5.10) are derived from standard linear algebra and externally established results (Schoenberg's theorem [26], Hjorth et al.'s Proposition 4.7 [13], Nickolas-Wolf's energy framework [22-24], Meckes' weighting spaces [21]). The one self-citation to Roff-Yoshinaga [25] provides the power series technique adapted in Lemma 3.7, but the adaptation is substantive: [25] analyzed the magnitude function's limit, while this paper analyzes the weighting vector's limit, yielding the strictly stronger Theorem 3.8. The key identity con(D) = μ'_X(0) (Theorem 3.1(3)) is derived by differentiating Z(t)w(t) = 1 and taking limits, not by definition. The energy-maximization result (Theorem 5.10) chains through Proposition 5.9, whose proof uses a standard Lagrange multiplier argument plus the negative-type inequality—no fitted parameters or definitional equivalences. The compact-space theory (Section 6) is explicitly labeled as conjectural groundwork with one verified example (B³). No step reduces to its inputs by construction, and no prediction is a renamed fit. The self-citation [25] is not load-bearing for the central claims since Theorem 3.8's proof is given in full and could stand without it. Score 1 reflects the minor self-citation that provides context but does not force any result.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 3 invented entities

No free parameters are introduced. The axioms are standard mathematical results or domain results from cited literature. The invented entities (microscopic weighting, concentration, gauging) are all given independent characterizations and are not postulated ad hoc.

axioms (5)
  • standard math Standard linear algebra: orthogonal splitting of R^n into im(A) ⊕ ker(A) for symmetric A
    Used in the proof of Theorem 2.7 (magnitude = 1/concentration).
  • domain assumption Schoenberg's theorem: a finite metric space embeds isometrically into Euclidean space via d^{1/2} iff it is of negative type
    Invoked in Section 5.1 to define the Schoenberg polytope and establish Theorem 5.4.
  • domain assumption Proposition 4.7 (Hjorth et al. [13]): a finite space of negative type is of strictly negative type iff the bordered matrix N = [[0, 1^T],[1, D]] is invertible
    Load-bearing for Proposition 4.10 and Theorem 4.11, which characterize strictly negative type via invertibility and finite concentration of D.
  • domain assumption Meckes' framework: weightings for compact subsets of Euclidean space exist as elements of Bessel potential spaces H^{-(m+1)/2}
    Used in Section 6.2 to define microscopic weightings for compact subsets of R^m via weak-* limits.
  • standard math Dominated convergence theorem
    Used in the proof of Theorem 6.10 to exchange limits and integrals when showing Alexander's maximizing family converges to the microscopic weighting on B^3.
invented entities (3)
  • Microscopic weighting independent evidence
    purpose: A canonical, scale-independent signed measure on a finite metric space, defined as the small-scale limit of weightings
    Its existence is proved for spaces of strictly negative type (Theorem 4.11), and it is characterized via three independent routes: as a gauging for D, as the circumcentre of the Schoenberg polytope, and as an energy-maximizing measure. It makes the falsifiable prediction M(X) = μ'_X(0), verified numerically for odd-dimensional balls.
  • Concentration of a symmetric matrix (con(A)) independent evidence
    purpose: A matrix invariant dual to magnitude, defined as the scalar c in Av = c·1 for a gauging v with 1^T v = 1
    The duality mag(A) = 1/con(A) is proved in Theorem 2.7 from standard linear algebra. It is not postulated but derived.
  • Gauging for a symmetric matrix independent evidence
    purpose: A vector v satisfying 1^T v = 1 and Av = c·1; the 'twin' of a weighting
    The concept appears in the distance geometry literature under other names (d-invariant measures, aspherical distance matrices). The paper gives it a stable name and proves it coincides with energy-maximizing measures for negative type spaces (Proposition 5.9).

pith-pipeline@v1.1.0-glm · 36088 in / 3181 out tokens · 448431 ms · 2026-07-07T15:39:54.443879+00:00 · methodology

0 comments
read the original abstract

We introduce the microscopic weighting, a canonical signed measure of mass one that can be associated to almost any finite metric space. The microscopic weighting is obtained as the small-scale limit of the weightings used to define the magnitude function. We give general criteria for its existence, proving in particular that every finite space of strictly negative type admits a microscopic weighting; this includes every finite subset of Euclidean or hyperbolic space and every finite tree. Heuristically speaking, the microscopic weighting distributes its mass as widely as possible across a space, assigning greater weight to sparse or outlying regions and emphasizing points on the boundary. Indeed, we show that on a finite space of negative type the microscopic weighting can be characterized (when it exists) as an optimizing measure for an energy integral determined by the distance function. Alternatively, it can be characterized in terms of the geometry of the Schoenberg embedding. Each of these interpretations also clarifies the information carried by the derivative of the magnitude function at zero. Though our main focus in this paper is on finite metric spaces, we lay the groundwork to extend the theory to compact subsets of Euclidean space. In that setting, we observe that the microscopic weighting must be understood as a distribution rather than as a measure.

Figures

Figures reproduced from arXiv: 2607.05349 by Emily Roff, Simon Willerton.

Figure 1
Figure 1. Figure 1: The five-point metric space 𝑃(𝑎, 𝑏) described in Exam￾ple 2.9. Unmarked edges represent distances equal to 1, while 𝑎 and 𝑏 can take any values in the interval (0, 2]. On the right is part of the parameter space. See Examples 2.9 and 3.6 for discussion. That the one-point property can fail was demonstrated first by an example due to Willerton [16, Example 2.2.8]. More recently, Roff and Yoshinaga have show… view at source ↗
Figure 2
Figure 2. Figure 2: The small-scale behaviour of the magnitude function for four spaces in the family 𝑃(𝑎, 𝑏). See Example 3.6 for discussion. (This space appears in [25, Example 3.4] as a ‘smallest’ space for which the one￾point property fails.) Meanwhile, the distance matrix of 𝑃D = 𝑃(2, 3/2) has zero concentration, but the one-point property does hold for 𝑃D. Thus, con(𝐷) = 0 does not imply failure of the one-point propert… view at source ↗
Figure 3
Figure 3. Figure 3: The Schoenberg polytopes associated to two members of the family of spaces in Example 2.9. The vertices of Δ(𝑃E) lie on a 2-sphere, while those of Δ(𝑃C) do not. See Example 5.6. As 𝐺 is real and symmetric, it can be decomposed as 𝐺 = 𝑄Λ𝑄T where the columns of 𝑄 are orthonormal eigenvectors for 𝐺 and Λ is the diagonal matrix of eigenvalues. The key to Schoenberg’s theorem is that 𝐷 is conditionally negative… view at source ↗
Figure 4
Figure 4. Figure 4: The microscopic weightings on grid approximations to two regions in the plane. The area of the disc at each point is proportional to the weight at that point. Red represents a positive weight, blue a negative weight. Notice that the largest weights appear to be concentrated on the boundary. 5.3. Microscopic weightings and discrete curvature. Our third interpretation relates the microscopic weighting to a n… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

29 extracted references · 29 canonical work pages · 2 internal anchors

  1. [1]

    Michael Adamer, Edward De Brouwer, Leslie O’Bray, and Bastian Rieck,The magnitude vector of images, Journal of Applied and Computational Topology8 (2024), 447–473

    work page 2024

  2. [2]

    2, 317–320

    Ralph Alexander,Two notes on metric geometry, Proceedings of the American Mathematical Society64 (1977), no. 2, 317–320

    work page 1977

  3. [3]

    Stolarsky,Extremal problems of distance geometry related to energy integrals, Transactions of the American Mathematical Society193(1974), 1–31

    Ralph Alexander and Kenneth B. Stolarsky,Extremal problems of distance geometry related to energy integrals, Transactions of the American Mathematical Society193(1974), 1–31

    work page 1974

  4. [4]

    2, 449–494

    Juan Antonio Barceló and Anthony Carbery,On the magnitudes of compact sets in Euclidean spaces, American Journal of Mathematics140 (2018), no. 2, 449–494

    work page 2018

  5. [5]

    Göran Björck,Distributions of positive mass, which maximize a certain generalized energy integral, Arkiv för Matematik3(1956), 255–269

    work page 1956

  6. [6]

    Ingwer Borg and Patrick J. F. Groenen,Modern multidimensional scaling: Theory and applications, second ed., Springer Series in Statistics, Springer, New York, 2005

    work page 2005

  7. [7]

    Eric Bunch, Jeffery Kline, Daniel Dickinson, Suhaas Bhat, and Glenn Fung,Weighting vectors for machine learning: numerical harmonic analysis applied to boundary detection, Preprint arXiv:2106.00827, 2021

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  8. [8]

    thesis, Oxford University, 2022

    Karel Devriendt,Graph geometry from effective resistances, Ph.D. thesis, Oxford University, 2022

    work page 2022

  9. [9]

    KarelDevriendtandRenaudLambiotte, Discretecurvatureongraphsfromtheeffectiveresistance ,Journal of Physics: Complexity3 (2022), article no. 025008

    work page 2022

  10. [10]

    G Friedlander and M

    F. G Friedlander and M. Joshi,Introduction to the theory of distributions, second ed., Cambridge University Press, 1998

    work page 1998

  11. [11]

    3–4, 887–896

    Aicke Hinrichs, Peter Nickolas, and Reinhard Wolf,A note on the metric geometry of the unit ball, Mathematische Zeitschrift268 (2010), no. 3–4, 887–896

    work page 2010

  12. [12]

    P. G. Hjorth, S. L. Kokkendorff, and S. Markvorsen,Hyperbolic spaces are of strictly negative type, Proceedings of the American Mathematical Society130(2002), no. 1, 175–181

    work page 2002

  13. [13]

    Poul Hjorth, Petr Lison˘ek, Steen Markvorsen, and Carsten Thomassen,Finite metric spaces of strictly negative type, Linear Algebra and its Applications270(1998), 255–273

    work page 1998

  14. [14]

    Hirokazu Katsumasa, Emily Roff, and Masahiko Yoshinaga,Is magnitude ‘generically continuous’ for finite metric spaces?, Preprint arXiv:2501.08745, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  15. [15]

    DJ Klein and M Randić,Resistance distance, Journal of Mathematical Chemistry12(1993), 81–95

    work page 1993

  16. [16]

    Tom Leinster,The magnitude of metric spaces, Documenta Mathematica18 (2013), 857–905

    work page 2013

  17. [17]

    ,Entropy and diversity: The axiomatic approach, Cambridge University Press, 2021

    work page 2021

  18. [18]

    Meckes,Maximizing diversity in biology and beyond, Entropy18 (2016), no

    Tom Leinster and Mark W. Meckes,Maximizing diversity in biology and beyond, Entropy18 (2016), no. 3, 88 (en)

    work page 2016

  19. [19]

    4, 1271–1309

    Tom Leinster and Emily Roff,The maximum entropy of a metric space, The Quarterly Journal of Mathematics 72(2021), no. 4, 1271–1309

    work page 2021

  20. [20]

    3, 733–757

    Mark Meckes,Positive definite metric spaces, Positivity17 (2013), no. 3, 733–757

    work page 2013

  21. [21]

    Meckes,Magnitude, diversity, capacities, and dimensions of metric spaces, Potential Analysis42 (2014), no

    Mark W. Meckes,Magnitude, diversity, capacities, and dimensions of metric spaces, Potential Analysis42 (2014), no. 2, 549–572

    work page 2014

  22. [22]

    I, Bulletin of the Australian Mathematical Society80 (2009), no

    Peter Nickolas and Reinhard Wolf,Distance geometry in quasihypermetric spaces. I, Bulletin of the Australian Mathematical Society80 (2009), no. 1, 1–25

    work page 2009

  23. [23]

    II, Mathematische Nachrichten284 (2011), 332–341

    , Distance geometry in quasihypermetric spaces. II, Mathematische Nachrichten284 (2011), 332–341

    work page 2011

  24. [24]

    III, Mathematische Nachrichten284 (2011), no

    , Distance geometry in quasihypermetric spaces. III, Mathematische Nachrichten284 (2011), no. 5–6, 747–760

    work page 2011

  25. [25]

    6, 1841–1855

    Emily Roff and Masahiko Yoshinaga,The small-scale limit of magnitude and the one-point property, Bulletin of the London Mathematical Society57(2025), no. 6, 1841–1855

    work page 2025

  26. [26]

    Sur la définition axiomatique d’une classe d’espace distanciés vectoriellement applicable sur l’espace de Hilbert

    I. J. Schoenberg,Remarks to Maurice Fréchet’s article “Sur la définition axiomatique d’une classe d’espace distanciés vectoriellement applicable sur l’espace de Hilbert”, Annals of Mathematics36 (1935), no. 3, 724–732

    work page 1935

  27. [27]

    Tarazaga, T

    P. Tarazaga, T. L. Hayden, and Jim Wells,Circum-euclidean distance matrices and faces, Linear Algebra and its Applications232 (1996), 77–96

    work page 1996

  28. [28]

    Simon Willerton,Spread: A measure of the size of metric spaces, International Journal of Computational Geometry and Applications25(2015), 207–225

    work page 2015

  29. [29]

    ,ThemagnitudeofoddballsviaHankeldeterminantsofreverseBesselpolynomials ,DiscreteAnalysis (2020), no. 5. ER: School of Mathematics and the Maxwell Institute for Mathematical Sciences, University of Edinburgh, Scotland SW: School of Mathematical and Physical Sciences, University of Sheffield, England

    work page 2020