Pith. sign in

REVIEW 1 major objections 3 minor 24 references

Strong invariants and Tverberg numbers in convexity spaces

T0 review · 1 major / 3 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read Five boundedness parameters of a convexity space—VC dimension, strong Helly, strong Carathéodory, comatching, and strong Radon—are one invariant; separable spaces get linear-in-t Tverberg bounds.

desk verdict Clean, genuinely unifying equivalence theorem plus a real Tverberg advance under S3-separability; both main proofs hold up, and the paper deserves peer review. read the letter →

arxiv 2607.29403 v1 pith:NYH5QBJB submitted 2026-07-31 math.CO cs.CGmath.MG

Minho Cho , Andreas F. Holmsen , Attila Jung , Hong Liu This is my paper
classification math.COcs.CGmath.MG MSC 52A3552A01
keywords convexityspaceVCdimensionstrongHellynumberCarathéodoryRadonTverbergbipartiteincidencegraphS3-separable
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

This paper argues that five a priori different measurements of a convexity space—VC-dimension, strong Helly number, strong Carathéodory number, comatching number, and strong Radon number—are boundedness-equivalent: if any one is at most d, all are, with the strong Radon number shifted by one. The proof exposes a common mechanism: a bipartite incidence graph between points and convex generators, in which intersections and convex hulls become the same common-neighborhood operation. On the Tverberg side, the paper proves that in any convexity space satisfying a weak separation axiom (S3), the t-part Tverberg number grows at most linearly in t, with a polynomial factor in the Radon number: r_t = O(dh log h)t for Helly number h and halfspace VC-dimension d, hence O(r^2 log r)t when the Radon number is r. A sympathetic reader cares because this is the first bound of the conjectured weak-Eckhoff order O(rt) for axis-parallel box convexity in every dimension, and because the equivalence theorem lets a bound proved in any one of five languages transfer immediately to all the others.

What carries the argument

The bipartite incidence model. Given a convexity space (X,C) and a generating family G, form the bipartite graph with vertex classes X and G, joining x to C when x lies in C. Then an intersection of generators is the common neighborhood N(F), and a convex hull is N(N(Y)); the identity N(N(N(A))) = N(A) makes hulls and intersections two faces of one operation. All five invariants become the non-existence of a single induced configuration, a comatching, which proves the equivalence without a chain of unrelated implications. For the Tverberg bound, the mechanism is a Helly-type centerpoint combined with a packed probabilistic ε-net lemma: under S3 separation, any net for the halfspaces through

What would settle it

Construct an S3-separable convexity space with bounded Helly number and bounded halfspace VC-dimension whose t-part Tverberg number is not O(t). Concretely, compute r_t for axis-parallel box convexity in R^4: the theorem forces r_t = O(t), so observing r_t/t unbounded would refute it.

Watch

Extended reading notes

Core claim

The central claim is Theorem 3.4: in any convexity space with a generator family, the following are equivalent: VC-dimension of convex sets at most d; every set in convex position has size at most d; comatching number at most d; exact Helly certificates of size at most d for convex sets and for generators; exact Carathéodory certificates of size at most d; strong Radon number at most d+1; and layered Tverberg decompositions on both the point side and the convex-set side. A second main result, Theorem 5.3, states that if the space is S3-separable with Helly number h and its halfspaces have VC-dimension d, then r_t = O(dh log h)t; in terms of the Radon number r this is O(r^2 log r)t, attaining

Load-bearing premise

The proof of the S3 Tverberg bound relies on the separation axiom that any point outside a convex set can be separated from it by complementary halfspaces; if that separation fails, a set meeting all halfspaces through the centerpoint need not have the centerpoint in its convex hull, and the argument collapses.

Editorial extensions

If this is right

  • A bound proved in any one of the five languages—VC-dimension, strong Helly, strong Carathéodory, comatching, or strong Radon—automatically transfers to the other four, together with layered and colorful Tverberg-type consequences.
  • For Hamming-ball convexity, known exact Helly certificates of size 2q+1 force the same bound for exact hull certificates and for the comatching and strong Radon numbers.
  • For axis-parallel box convexity in R^k, the theorem gives r_t = O(rt) uniformly in the dimension, the first dimension-uniform estimate of weak-Eckhoff order for boxes; the previous direct theory reached only dimension three.
  • For the geodesic convexity of the graph whose vertices are the 2-subsets of [n] and whose edges join intersecting pairs, the theorem yields r_t = O(n log n)t, recovering the optimal linear scale up to one logarithmic factor.
  • The O(t^4)-point realization of the counterexample separates the local Radon obstruction from the global Tverberg obstruction, showing that the ordinary Radon number alone cannot control Tverberg numbers.

Reading between the lines

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

  • Beyond the paper: the equivalence theorem suggests a transfer principle for computational geometry and learning-theoretic settings—any bound or algorithm expressed through one of the five parameters can be re-expressed in the others, potentially simplifying implementations.
  • Beyond the paper: because the S3 centerpoint–net argument converts a halfspace-meeting set into a convex-hull certificate, the same proof strategy may extend to other closure systems admitting complementary halfspaces, such as geodesic convexities in graphs, whenever the relevant halfspace family has bounded VC-dimension.
  • Beyond the paper: one may test the sharpness of Theorem 5.3 by computing r_t for axis-parallel boxes in R^4; the theorem predicts linear growth in t, so a superlinear rate would pinpoint the limit of the centerpoint–net method, while a linear rate with a smaller constant would suggest that the log h factor can be removed.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

1 major / 3 minor

Summary. The paper develops a unified theory of strong finite-certificate invariants for convexity spaces. Its first main result, Theorem 3.4, establishes the equivalence of bounded VC-dimension, strong Helly number, strong Carathéodory number, comatching number, and strong Radon number (with the expected additive-one shift), together with layered Tverberg-type decompositions. The proof is based on a bipartite incidence model between points and a generator family, which also yields duality results and a polynomial-size O(t^4) realization of Bukh's counterexample to the Calder–Eckhoff conjecture. The second main result, Theorem 5.3, proves that for an S3-separable convexity space with Helly number h and halfspace VC-dimension d, the t-th Tverberg number satisfies r_t = O(dh log h) t, and hence r_t = O(r^2 log r) t in terms of the Radon number r. The paper also derives colorful corollaries and, in an appendix, selection lemmas, weak epsilon-nets, and quantitative (p,q) theorems from the same centerpoint-net mechanism.

Significance. If the main results stand, the paper makes two substantial contributions. First, Theorem 3.4 is a clean and useful unification: five a priori distinct strong invariants are shown to be one parameter, and the equivalence is proved self-contained from definitions, with the finite-configuration caveat explicitly noted in Remark 3.5. Second, Theorem 5.3 gives the first Tverberg bound for separable convexity spaces that is simultaneously linear in t and polynomial in r, with constants tracked through the centerpoint lemma and the packed-net argument. The incidence/duality framework is attractive and the O(t^4) Bukh construction is a genuine simplification. The main concern is a gap in the proof of one of the colorful corollaries, Theorem 3.13; this is localized and does not appear to affect the core equivalence or the S3-separability Tverberg theorem, but it must be fixed before publication.

major comments (1)
  1. [§3.3, Theorem 3.13] The proof of the containment statement is incomplete. After selecting an inclusion-maximal rainbow hull M and reducing via strong Carathéodory to M = conv{p1,...,pd}, with p_{d+1} in S_i removed, maximality shows that for every q in S_i, conv{p1,...,pd,q} = M, hence S_i ⊆ M. But this argument is applied only to the color i of the removed point. For j ≠ i, replacing p_j by q ∈ S_j gives a rainbow hull that need not contain M, so maximality yields no information about S_j. The proof therefore establishes only the 'Moreover' clause (some color is contained in the hull of a d-tuple from the other colors), not the stated conclusion that a single rainbow (d+1)-tuple contains ∩_i conv S_i. Since Theorem 3.14 uses only the 'Moreover' clause, the colorful Tverberg theorem may still be valid, but Theorem 3.13 itself is unproved as stated. Please supply a correct argument for the full statement or
minor comments (3)
  1. [§1.1 and Abstract] The Hamming-ball bound is written as '2q+1' in several places; it should be 2^{q+1} (or 2^q+1 if that is intended). This is a notation issue but should be fixed to avoid ambiguity.
  2. [§3.3, Theorems 3.15 and 3.16] The proof is omitted as 'exactly the same' as Theorems 3.13 and 3.14. Given the gap in Theorem 3.13, these results inherit the issue; the authors should either provide the proof or explicitly state that the 'Moreover' part is the only ingredient used.
  3. [§3.3, Theorem 3.13] The first sentence of the proof says 'This finishes the proof' after proving only the 'Moreover' clause. This is misleading and should be rewritten once the theorem is repaired.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; the main theorems are proved from definitions and standard external inputs.

full rationale

The central derivation chain is self-contained. Theorem 3.4 is proved directly from definitions: (1)⇔(2) via the characterization of convex position as shattering using conv(A)∩S=A; (2)⇔(3) by taking C_i=conv({p_j:j≠i}); (3)⇔(4),(5) via inclusion-minimal subfamilies and Observation 3.3; (2)⇔(6) by singleton partitions; and (7) plus the layered items follow by iterating exact certificates. No item is defined in terms of another, and no parameter is fitted to a predicted quantity. Theorem 5.3 is an independent epsilon-net/centerpoint argument: Lemma 5.1 uses only the Helly number and S3-separation, Lemma 5.2 is the standard probabilistic epsilon-net theorem, and the packing step is a direct expectation argument. The bound r_t=O(dh log h)t is obtained without using the Tverberg bound as an input. The corollary O(r^2 log r)t follows from h≤r−1 and VC(B)≤r−1, both proved from definitions. The only self-citations ([14], [16]) appear in examples and in the appendix's application of a fractional Helly theorem; they serve as external inputs for corollaries, not as load-bearing content of the main theorems. No equation reduces by construction to an input, and no prediction is a renamed fit. Therefore no significant circularity is present.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

No data-fitted free parameters appear; all constants in the asymptotic bounds are absolute. The assumptions are either standard results or explicit domain hypotheses of the theorems. The dual convexity space and generator-relative dual Radon number are constructions, not unverifiable postulates.

assumptions (7)
  • domain assumption Convexity space closure axioms: ∅, X ∈ C; closure under arbitrary intersections; closure under unions of chains.
    The framework of the entire paper; all theorems are stated for such spaces.
  • domain assumption S3-separability: every point outside a convex set can be separated from it by complementary halfspaces.
    Used in Lemma 5.1(2) and Theorem 5.3 to convert halfspace-meeting sets into convex-hull certificates.
  • domain assumption Helly number h and halfspace VC-dimension d are finite.
    Required for the centerpoint lemma and the epsilon-net theorem in Theorem 5.3.
  • standard math Standard probabilistic epsilon-net theorem and Sauer-Shelah lemma.
    Invoked in Lemma 5.2 and Lemma A.2 for the packing of disjoint nets and for the rainbow net bound.
  • standard math Levi's inequality h <= r-1 and the halfspace shattering bound VC(B) <= r-1.
    Invoked in Corollary 5.4 to convert the h,d form into the Radon-number form; h <= r-1 is cited without proof.
  • standard math Unique minimal generator for finite convexity spaces.
    Claim 2.1, needed for the dual convexity space and duality invariance; a standard fact about finite closure systems.
  • domain assumption Fractional Helly theorem for S3-separable spaces (Holmsen [16]).
    Used only in Corollary A.7 to apply the quantitative (p,q) theorem to Radon-number-bounded spaces.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Strong invariants and Tverberg numbers in convexity spaces." pith.science (2026). https://pith.science/paper/NYH5QBJB

@misc{pith2026260729403,
  author       = {Pith},
  title        = {Pith review of: Strong invariants and Tverberg numbers in convexity spaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NYH5QBJB}},
  note         = {Machine review of arXiv:2607.29403}
}
abstract

Helly, Carath\'eodory, and Radon numbers encode three kinds of finite certificates in a convexity space: for the emptiness of an intersection, for membership in a convex hull, and for the existence of intersecting hulls. We study exact versions of these certificates, in which a subfamily must preserve the whole intersection or a subset must preserve the whole hull. Our first main result shows that, for finite configurations in an arbitrary convexity space, five a priori different boundedness conditions are equivalent: VC-dimension, strong Helly number, strong Carath\'eodory number, comatching number, and strong Radon number (with the expected additive-one shift). We also obtain equivalent layered Tverberg-type decompositions and colorful consequences. The common mechanism is exposed by the bipartite incidence graph between points and a generating family. For finite spaces, the unique minimal generator yields a natural dual convexity space; we characterize double dualization and prove that the strong parameters are duality invariant. The same model gives a polynomial-size, $O(t^4)$, realization of Bukh's counterexample to the Calder-Eckhoff partition conjecture. Finally, we obtain the first Tverberg bound for separable convexity spaces that is simultaneously linear in the number of parts and polynomial in the Radon number. If an $S_3$-separable convexity space has Helly number $h$ and its halfspaces have VC-dimension $d$, then $r_t=O(dh\log h)\,t$; in particular, Radon number $r$ gives $r_t=O(r^2\log r)\,t$. The bound attains the weak-Eckhoff scale $O(rt)$ whenever the Helly number is bounded. For axis-parallel box convexity in $\mathbb{R}^k$, gives the optimal order $r_t=O(rt)$ uniformly in every dimension. This appears to be the first dimension-uniform estimate of weak-Eckhoff order for box convexity, whereas the previous direct theory was confined to dimension three.

Figures

Figures reproduced from arXiv: 2607.29403 by the authors.

Figure 1
Figure 1. An illustration of the bipartite graph model of Bukh’s construction. [PITH_FULL_IMAGE:figures/full_fig_p015_1.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

24 extracted references · 7 linked inside Pith

  1. [16]

    Helly type problems in convexity spaces.arXiv preprint arXiv:2408.05871, 2024

    Andreas F Holmsen. Helly type problems in convexity spaces.arXiv preprint arXiv:2408.05871, 2024. 19

    arXiv 2024

  2. [1]

    The Helly number of hamming balls and related problems.arXiv preprint arXiv:2405.10275, 2024

    Noga Alon, Zhihan Jin, and Benny Sudakov. The Helly number of hamming balls and related problems.arXiv preprint arXiv:2405.10275, 2024

    arXiv 2024

  3. [2]

    Extended VC-dimension, and radon and tverberg type theorems for unions of convex sets.arXiv preprint arXiv:2506.17777, 2025

    Noga Alon and Shakhar Smorodinsky. Extended VC-dimension, and radon and tverberg type theorems for unions of convex sets.arXiv preprint arXiv:2506.17777, 2025

    arXiv 2025

  4. [3]

    Vapnik-chervonenkis density in some theories without the independence prop- erty, i.Transactions of the American Mathematical Society, 368(8):5889–5949, 2016

    Matthias Aschenbrenner, Alf Dolich, Deirdre Haskell, Dugald Macpherson, and Sergei Starchenko. Vapnik-chervonenkis density in some theories without the independence prop- erty, i.Transactions of the American Mathematical Society, 368(8):5889–5949, 2016

    2016

  5. [4]

    A generalization of Carathéodory’s theorem.Discrete Mathematics, 40(2- 3):141–152, 1982

    Imre Bárány. A generalization of Carathéodory’s theorem.Discrete Mathematics, 40(2- 3):141–152, 1982

    1982

  6. [5]

    Helly-type problems.Bulletin of the American Mathematical Society, 59(4):471–502, 2022

    Imre Bárány and Gil Kalai. Helly-type problems.Bulletin of the American Mathematical Society, 59(4):471–502, 2022

    2022

  7. [6]

    Radon partitions in convexity spaces.arXiv preprint arXiv:1009.2384, 2010

    Boris Bukh. Radon partitions in convexity spaces.arXiv preprint arXiv:1009.2384, 2010

    arXiv 2010

  8. [7]

    Some elementary properties of interval convexities.Journal of the London Mathematical Society, 2(3):422–428, 1971

    JR Calder. Some elementary properties of interval convexities.Journal of the London Mathematical Society, 2(3):422–428, 1971

    1971

Show all 24 references
  1. [8]

    Separation axiomS3 for geodesic convexity in graphs, 2024

    Victor Chepoi. Separation axiomS3 for geodesic convexity in graphs, 2024

    2024

  2. [9]

    Combinatorial properties of nonarchimedean convex sets.Pacific Journal of Mathematics, 323(1):1–30, 2023

    Artem Chernikov and Alex Mennen. Combinatorial properties of nonarchimedean convex sets.Pacific Journal of Mathematics, 323(1):1–30, 2023

    2023

  3. [10]

    De Loera, Reuben N

    Jesús A. De Loera, Reuben N. La Haye, David Rolnick, and Pablo Soberón. Quantitative Tverbergtheoremsoverlatticesandotherdiscretesets.Discrete & Computational Geometry, 58(2):435–448, 2017

    2017

  4. [11]

    A Helly type theorem for hypersurfaces.Journal of Com- binatorial Theory, Series A, 45(1):27–30, 1987

    Mikhail Deza and Peter Frankl. A Helly type theorem for hypersurfaces.Journal of Com- binatorial Theory, Series A, 45(1):27–30, 1987

    1987

  5. [12]

    Radon’s theorem revisited

    Jürgen Eckhoff. Radon’s theorem revisited. InContributions to Geometry: Proceedings of the Geometry-Symposium held in Siegen June 28, 1978 to July 1, 1978, pages 164–185. Springer, 1979

    1978

  6. [13]

    A partition theorem of tverberg-type for boxes inR3.Discrete Mathematics, 241(1-3):267–288, 2001

    Jürgen Eckhoff. A partition theorem of tverberg-type for boxes inR3.Discrete Mathematics, 241(1-3):267–288, 2001

    2001

  7. [14]

    Helly-type theorems for monotone properties of boxes.arXiv preprint arXiv:2503.22571, 2025

    Nóra Frankl and Attila Jung. Helly-type theorems for monotone properties of boxes.arXiv preprint arXiv:2503.22571, 2025

    2025 arXiv

  8. [15]

    Online optimisation in convexity spaces

    Thomas Gärtner and Olana Missura. Online optimisation in convexity spaces. InProceed- ings of the NIPS Workshop on Discrete and Combinatorial Problems in Machine Learning (DISCML), 2014

    2014

  9. [17]

    Partition numbers for trees and ordered sets.Pacific Journal of Mathe- matics, 96(1):115–140, 1981

    Robert Jamison. Partition numbers for trees and ordered sets.Pacific Journal of Mathe- matics, 96(1):115–140, 1981

    1981

  10. [18]

    A colorful extension of VC-dimension and geomet- ric applications.arXiv preprint arXiv:2607.10496, 2026

    Chaya Keller and Shakhar Smorodinsky. A colorful extension of VC-dimension and geomet- ric applications.arXiv preprint arXiv:2607.10496, 2026

    2026 arXiv

  11. [19]

    On Helly’s theorem and the axioms of convexity.J

    Friedrich W Levi. On Helly’s theorem and the axioms of convexity.J. Indian Math. Soc, 15(Pt A):65–76, 1951

    1951

  12. [20]

    Radon numbers grow linearly.Discrete & Computational Geometry, 68(1):165–171, 2022

    Dömötör Pálvölgyi. Radon numbers grow linearly.Discrete & Computational Geometry, 68(1):165–171, 2022

    2022

  13. [21]

    Colorful Helly via induced matchings

    Cosmin Pohoata, Kevin Yang, and Shengtong Zhang. Colorful Helly via induced matchings. arXiv preprint arXiv:2501.17149, 2025

    2025 arXiv

  14. [22]

    Helly-type theorems for separatedd-intervals.arXiv preprint arXiv:2501.03207, 2025

    Wei Rao. Helly-type theorems for separatedd-intervals.arXiv preprint arXiv:2501.03207, 2025

    2025 arXiv

  15. [23]

    A generalization of Radon’s theorem.Journal of the London Mathematical Society, 1(1):123–128, 1966

    Helge Tverberg. A generalization of Radon’s theorem.Journal of the London Mathematical Society, 1(1):123–128, 1966

    1966

  16. [24]

    Elsevier, 1993

    Marcel LJ van De Vel.Theory of convex structures, volume 50. Elsevier, 1993. A Appendix: Selection, weak nets, and piercing consequences Thecenterpoint–netprinciple, Lemma5.1, alsoyieldsselectionandpiercingstatements. Through- outthissubsection,(X,C)isanS 3-separableconvexitys...

    1993

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.