Pith. sign in

REVIEW 4 minor 18 references

Inequalities on Six Points in a $\mathrm{CAT}(0)$ Space

T0 review · 0 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper proves that a five-parameter family of quadratic six-point inequalities holds in every CAT(0) space, and that none of them follows from any five-point condition.

desk verdict Genuinely new six-point CAT(0) inequalities with a sound non-implication proof; only a minor expository gap. read the letter →

arxiv 2411.13877 v2 pith:D3YEIPZX submitted 2024-11-21 math.MG

Tetsu Toyoda This is my paper · ORCID
classification math.MG MSC 30L1553C2351F99
keywords CAT(0)spacesquadraticmetricinequalitiessix-pointisometricembeddingsbarycenterinequalitydistanceconvexityoctahedroncomparisonofnon-positivecurvature
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

The paper proves a new family of quadratic metric inequalities on six points in every $\mathrm{CAT}(0)$ space — a complete geodesic metric space of non-positive curvature — with parameters $a,b,c,s,t\in[0,1]$ satisfying $a\le s$. The main point is not just that the inequalities hold, but that each one is genuinely a six-point condition: for every parameter choice with $0

What carries the argument

The load-bearing object is the inequality (1.2) itself, a weighted sum of squared distances with parameters $a,b,c,s,t$. Its proof is carried by two $\mathrm{CAT}(0)$ tools: the barycenter inequality (2.4), which bounds the squared distance between the barycenters of two probability measures by the weighted sum of squared pairwise distances between their support points, and the $\mathrm{CAT}(0)$ distance-convexity inequality (2.1), together with Proposition 2.3 comparing points at fractions $a$ and $s$ along a geodesic. A barycenter here is the point minimizing the weighted sum of squared distances to the support points of the measure. The proof chooses barycenters $z=\operatorname{bar}((1-c)\delta_{z_0}+c\delta_{z_1})$ and $w=\operatorname{bar}((1-t)\delta_{y_1}+(1-s)t\delta_{x_0}+st\delta_{x_1})$, forcing every intermediate inequality to become an equality on a Euclidean octahedral configuration where two barycenter equations hold. That equality is what makes the sharpness construction work: increasing only the squared distance between $z_0$ and $z_1$ by a small positive amount leaves the right-hand side unchanged and makes the left-hand side strictly larger, so the inequality fails while all five-point subsets still behave well.

What would settle it

For a concrete parameter tuple such as $(a,b,c,s,t)=(1/4,1/2,1/3,3/4,1/5)$, produce explicit coordinates satisfying the intersection condition (2.5) and the two barycenter identities; this would verify the asserted existence. Alternatively, for any such configuration, compute the left-hand minus right-hand difference of (1.2) under $d_\varepsilon$: at $\varepsilon=0$ the difference is zero, while for $\varepsilon>0$ it equals $ab(1-c)c(2\varepsilon\|z_0-z_1\|+\varepsilon^2)>0$, so the claimed violation is directly checkable and would be contradicted only if the configuration does not exist.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1.3: for any $\mathrm{CAT}(0)$ space $X$ and any points $x_0,x_1,y_0,y_1,z_0,z_1$, the squared-distance inequality (1.2) holds whenever $a\le s$. Conversely, for any $a,b,c,s,t\in(0,1)$ with $a<s$, there exists a six-point metric space $L$, taken from a family of octahedral configurations described in Section 7.2 of the updated version of [1], such that every five-point subset of $L$ embeds isometrically into a $\mathrm{CAT}(0)$ space but $L$ violates (1.2) for those parameters. Consequently the family (1.2) cannot be derived from any five-point properties, including the $\mathrm{CAT}(0)$ 4-point ($\boxtimes$) condition. The same proof yields that these spaces satisfy every inequality of the form (1.3) and still violate (1.2), so the new inequalities are not consequences of that general family either. The paper also notes that any metric space satisfying the octahedron graph comparison satisfies (1.2).

Load-bearing premise

The sharpness half relies on the assertion, justified only by a figure, that for every choice of the five parameters with $0<a<s<1$ one can place six points in ordinary three-dimensional space so that a certain line segment crosses a quadrilateral exactly once and two weighted-average relations hold; if this placement fails for some parameters, the non-implication statement collapses.

Editorial extensions

If this is right

  • Any metric space that embeds isometrically into a $\mathrm{CAT}(0)$ space must satisfy (1.2) for all $a\le s$, so the family is a new necessary condition for $\mathrm{CAT}(0)$ embeddability.
  • The sharpness construction shows that the validity of all five-point $\mathrm{CAT}(0)$ quadratic inequalities is insufficient for six-point $\mathrm{CAT}(0)$ embeddability, with (1.2) as an explicit witness.
  • Because the six-point octahedral spaces used in the proof satisfy all inequalities of the form (1.3) yet violate (1.2), the catalogue of $\mathrm{CAT}(0)$ quadratic inequalities is strictly larger than the family (1.3).
  • Any metric space satisfying the octahedron graph comparison also satisfies (1.2); whether the full family (1.2) suffices for six-point $\mathrm{CAT}(0)$ embeddability is left open.

Reading between the lines

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

  • The equality case of the derivation suggests that (1.2) is sharp on octahedral configurations; a natural extension is to search for further explicit $\mathrm{CAT}(0)$ quadratic inequalities by forcing equality on other Euclidean polytopes.
  • The existence of the $\mathbb{R}^3$ configurations used for sharpness is asserted from a figure rather than proved by coordinates; making those coordinates explicit would turn the sharpness construction into a ready-to-use test set for numerical embedding algorithms.
  • One could test numerically whether the family (1.2), together with the $\boxtimes$-inequalities, is sufficient for six-point $\mathrm{CAT}(0)$ embeddability; the paper states no guess, so a counterexample or a proof would settle the open question.
  • The same barycenter-combination technique may produce explicit $n$-point $\mathrm{CAT}(0)$ inequalities for $n>6$ by choosing more probability measures whose barycenters coincide on a Euclidean configuration; this is an extension the paper does not pursue.
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

0 major / 4 minor

Summary. The paper establishes a family of quadratic metric inequalities on six points in any CAT(0) space. For parameters a, b, c, s, t in [0,1] with a ≤ s, and any six points x0, x1, y0, y1, z0, z1, the inequality (1.2) is proved by combining the CAT(0) comparison inequality (2.1), the barycenter inequality (2.4), and Proposition 2.3. The proof is an explicit algebraic chain (3.1)–(3.10), with the edge cases a = 0 and b = 0 treated separately. The author then shows sharpness of the inequality: for parameters in (0,1) with a < s, there is a six-point metric space (L, dL) such that every five-point subset embeds isometrically into a CAT(0) space, but (1.2) fails for (L, dL). This confirms that the new inequalities are genuine six-point CAT(0) conditions that do not follow from any five-point conditions. The paper also derives a corollary showing that even the full family of Andoni–Naor–Neiman inequalities (1.3) is insufficient, and it relates the new inequality to the octahedron graph comparison.

Significance. If the main result is correct, as the algebraic derivation indicates, this is a meaningful contribution: it supplies the first explicit examples of CAT(0) quadratic metric inequalities on six points that do not follow from the ⊠-inequalities nor from the five-point embeddability criterion. The proof is transparent and verifiable: the chain (3.1)–(3.10) is fully written out, equality in the Euclidean case is verified, and the sharpness construction is explicit modulo one geometric assertion that is true and easily supplied. The connection to Lebedeva's six-point spaces and to the O3-comparison is natural and places the result in a broader context. The paper represents a solid, narrow advance in the metric characterization problem for CAT(0) spaces. I verified the algebraic steps (3.1)–(3.10) and found them consistent; the edge case discussion covers the necessary parameter ranges. The main expositional weakness is the unproved assertion of the existence of the Euclidean configuration, which should be addressed before publication.

minor comments (4)
  1. [Section 3, proof of Theorem 1.3, moreover part] The existence of six points x0, x1, y0, y1, z0, z1 in R3 satisfying (2.5) and the two barycenter identities is asserted as 'clearly' with reference to Figure 3. Since this configuration is load-bearing for the sharpness construction, please replace the assertion by an explicit coordinate construction or a short verification. For example, take u=(1,0,0), v=(0,1,0), n=(0,0,1), set x0=-a u, x1=(1-a)u, y0=-b v, y1=(1-b)v, w=(t(s-a),(1-t)(1-b),0), z0=w+c n, z1=w-(1-c)n; then both barycenter identities hold by direct substitution, and (2.5) holds because the open segment (z0,z1) meets the plane of the quadrilateral exactly at w, which lies in the interior of the quadrilateral.
  2. [Theorem 2.4 and its use] Theorem 2.4 refers to the '(2+2)-point comparison' and '(4+2)-point comparison' without definition in the present text, pointing instead to [1, Section 6.2]. Although this is acceptable for a specialist journal, a short explanatory sentence would improve self-containedness, especially because Theorem 2.4 is central to the non-implication claim.
  3. [Equation (2.5)] The condition |(z0,z1) ∩ (conv({x0,y0,x1,y1}) \ I)| = 1 is terse. A short remark clarifying that it forces z0 and z1 to lie on opposite sides of the plane of the quadrilateral, with the segment piercing the interior of the quadrilateral, would make the geometric setup easier to follow.
  4. [Proof of Proposition 1.8, equation (3.11)] The quantification 'for any p, q, r in {x,y,z} with q ≠ r' is slightly confusing because p, q, r denote letters rather than points; please clarify by writing the inequalities with explicit indices, e.g., ∥p~_i - q~_j∥ ≤ dX(p_i, q_j) for p,q ∈ {x,y,z} with p ≠ q and i,j ∈ {0,1}.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: (1.2) is derived from standard CAT(0) comparison and barycenter inequalities; the non-implication uses independent prior theorems.

full rationale

The proof of Theorem 1.3 combines Definition 2.1, the barycenter inequality (2.4), inequality (2.2), and Proposition 2.3, itself proved from Definition 2.1, to derive (3.10) and hence the target inequality (1.2); the target inequality is never assumed in the derivation. The coefficients of (1.2) are produced by the algebra of weighted barycenters and the equality case in Lebedeva's configuration, and designing an inequality to be sharp does not make the proof circular. The moreover part depends on Lebedeva's external construction (Theorem 2.4) and on the author's Theorem 1.2, a previously published characterization of 5-point embeddability, which is independent support rather than a restatement of the present claim. Theorem 2.5 from the author's prior work [18] is used only in the auxiliary Corollary 1.5, and its conclusion is not the target inequality. The only expository gap is the assertion that six points in R3 satisfying (2.5) and the two barycenter identities exist; this is a missing coordinate check, not a circular step, and explicit coordinates can be supplied. No step reduces (1.2) to itself, to a fitted parameter, or to a self-citation chain.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The central inequality is derived from the CAT(0) comparison inequality and Sturm's barycenter inequality, which are standard. The non-implication result imports Lebedeva's six-point spaces and the author's earlier five-point characterization. No free parameters are fitted to data; the parameters a,b,c,s,t are arbitrary variables.

assumptions (5)
  • standard math CAT(0) comparison inequality (Definition 2.1, inequality (2.1))
    The central derivation depends on the defining comparison inequality for CAT(0) spaces, used throughout Section 3.
  • standard math Existence and inequality for barycenters of finitely supported probability measures in CAT(0) spaces (Sturm, used in (2.3) and (2.4))
    The proof relies on the barycenter inequality to bound distances between weighted averages of points; this is a prior theorem, not proved here.
  • domain assumption Lebedeva's Theorem 2.4: for points satisfying (2.5), small epsilon perturbations (L,d_epsilon) satisfy the (2+2)- and (4+2)-point comparisons
    This theorem is cited from the arXiv version of Alexander-Kapovitch-Petrunin [1]; it is load-bearing for the moreover part of Theorem 1.3.
  • domain assumption The author's prior theorems: Theorem 1.2 (five-point CAT(0) characterization) and Theorem 2.5 ([18], Lebedeva spaces satisfy Andoni-Naor-Neiman inequalities)
    These prior results are used to show that every five-point subset of the constructed space embeds and that Corollary 1.5 holds.
  • domain assumption Geometric existence: for each a,b,c,s,t with a<s, there are six points in R3 satisfying (2.5) and the two barycenter identities
    Asserted as 'clearly' true in the proof of Theorem 1.3 with reference to Figure 3; no explicit coordinates or proof are given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Inequalities on Six Points in a $\mathrm{CAT}(0)$ Space." pith.science (2026). https://pith.science/paper/D3YEIPZX

@misc{pith2026241113877,
  author       = {Pith},
  title        = {Pith review of: Inequalities on Six Points in a $\mathrmCAT(0)$ Space},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D3YEIPZX}},
  note         = {Machine review of arXiv:2411.13877}
}
abstract

We establish a family of inequalities that hold true on any $6$ points in any $\mathrm{CAT}(0)$ space. We prove that the validity of these inequalities does not follow from any properties of $5$-point subsets of $\mathrm{CAT}(0)$ spaces. In particular, the validity of these inequalities does not follow from the $\mathrm{CAT}(0)$ $4$-point condition.

Figures

Figures reproduced from arXiv: 2411.13877 by the authors.

Figure 1
Figure 1. The octahedron graph O3. Proposition 1.8. If a metric space (X, dX) satisfies the O3-comparison, then X satis￾fies the inequality (1.2) for any a, b, c, s, t ∈ [0, 1] with a ≤ s. Proposition 1.8 follows from Theorem 1.3 and the definition of the O3-comparison immediately. We will prove it at the end of Section 3. 1.3. Towards a characterization of 6-point subsets of CAT(0) spaces. It is nat￾ural to ask whether the v… view at source ↗
Figure 2
Figure 2. Lebedeva’s six points in R 3 . Theorem 2.4 (Lebedeva). Suppose x0, x1, y0, y1, z0, z1 ∈ R 3 are distinct six points that satisfy (2.5). Set L = {x0, x1, y0, y1, z0, z1}. Then there exists C ∈ (0, ∞) such that for any ε ∈ (0, C], (L, dε) defined by (2.6) is a metric space that satisfies the (2 + 2)-point comparison and the (4 + 2)-point comparison, but does not admit a distance-preserving embedding into any CAT(0) sp… view at source ↗
Figure 3
Figure 3. The points x0, x1, y0, y1, z0, z1 in R 3 . Let L = {x0, x1, y0, y1, z0, z1}, and define dε : L × L → [0, ∞) by (2.6) for each ε ∈ [0, ∞). We have observed that equality holds in (1.2) if we choose x0, x1, y0, y1, z0, z1 as above, and set dX = d0. Since the coefficient of dX(z0, z1) 2 in (1.2) is positive, it follows that for any ε ∈ (0, ∞), (1.2) does not hold true if we choose x0, x1, y0, y1, z0, z1 as above, and s… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 18 canonical work pages

  1. [1]

    S. B. Alexander, V. Kapovitch and A. Petrunin, Alexandrov meets Kirszbraun, in Proceedings of the G¨ okova Geometry-Topology Conference 2010, 88–109, Int. Press, Somerville, MA, 2011; updated version: arXiv:1012.5636v2, 2017

    work page Pith review arXiv 2010

  2. [2]

    S. B. Alexander, V. Kapovitch and A. Petrunin, Alexandrov geometry—foundations, Graduate Studies in Mathematics, 236, Amer. Math. Soc., Providence, RI, [2024] ©2024

    work page 2024

  3. [3]

    Andoni, A

    A. Andoni, A. Naor and O. Neiman, Snowflake universality of Wasserstein spaces, Ann. Sci. ´Ec. Norm. Sup´ er. (4)51 (2018), no. 3, 657–700

    work page 2018

  4. [4]

    Baˇ c´ ak, Old and new challenges in Hadamard spaces, Jpn

    M. Baˇ c´ ak, Old and new challenges in Hadamard spaces, Jpn. J. Math.18 (2023), no. 2, 115–168

    work page 2023

  5. [5]

    M. R. Bridson and A. Haefliger, Metric spaces of non-positive curvature , Grundlehren der math- ematischen Wissenschaften, 319, Springer, Berlin, 1999

    work page 1999

  6. [6]

    D. Y. Burago, Y. D. Burago and S. V. Ivanov, A course in metric geometry , Graduate Studies in Mathematics, 33, Amer. Math. Soc., Providence, RI, 2001

    work page 2001

  7. [7]

    M. Gromov, Metric structures for Riemannian and non-Riemannian spaces , translated from the French by Sean Michael Bates, Progress in Mathematics, 152, Birkh¨ auser Boston, Boston, MA, 1999

    work page 1999

  8. [8]

    Gromov, J

    M. Gromov, J. Math. Sci. (N.Y.) 119 (2004), no. 2, 178–200; translated from Zap. Nauchn. Sem. S.-Peterburg. Otdel. Mat. Inst. Steklov. (POMI) 280 (2001), 100–140, 299–300

    work page 2004

Show all 18 references
  1. [9]

    Kleiner, The local structure of length spaces with curvature bounded above, Math

    B. Kleiner, The local structure of length spaces with curvature bounded above, Math. Z. 231 (1999), no. 3, 409–456

    1999

  2. [10]

    N. D. Lebedeva and A. Petrunin, 5-point CAT(0) spaces after Tetsu Toyoda, Anal. Geom. Metr. Spaces 9 (2021), no. 1, 160–166

    2021

  3. [11]

    N. D. Lebedeva and A. Petrunin, Trees meet octahedron comparison, J. Topol. Anal., online ready, https://doi.org/10.1142/S1793525323500231, 2023

    2023 doi

  4. [12]

    N. D. Lebedeva and A. Petrunin, Graph comparison meets Alexandrov, Sib. Math. J. 64 (2023), no. 3, 624–628

    2023

  5. [13]

    N. D. Lebedeva, A. Petrunin and V. Zolotov, Bipolar comparison, Geom. Funct. Anal. 29 (2019), no. 1, 258–282

    2019

  6. [14]

    Y. G. Reshetnyak, Non-expansive maps in a space of curvature no greater than K, Sibirsk. Mat. ˇZ. 9 (1968), 918–927

    1968

  7. [15]

    Sturm, Probability measures on metric spaces of nonpositive curvature, in Heat kernels and analysis on manifolds, graphs, and metric spaces (Paris, 2002) , 357–390, Contemp

    K.-T. Sturm, Probability measures on metric spaces of nonpositive curvature, in Heat kernels and analysis on manifolds, graphs, and metric spaces (Paris, 2002) , 357–390, Contemp. Math., 338, Amer. Math. Soc., Providence, RI, 2003

    2002

  8. [16]

    Toyoda, An intrinsic characterization of five points in a CAT(0) space, Anal

    T. Toyoda, An intrinsic characterization of five points in a CAT(0) space, Anal. Geom. Metr. Spaces 8 (2020), no. 1, 114–165. INEQUALITIES ON SIX POINTS IN A CAT(0) SPACE 11

    2020

  9. [17]

    Toyoda, A non-geodesic analogue of Reshetnyak’s majorization theorem, Anal

    T. Toyoda, A non-geodesic analogue of Reshetnyak’s majorization theorem, Anal. Geom. Metr. Spaces 11 (2023), no. 1, Paper No. 20220151, 22 pp

    2023

  10. [18]

    Toyoda, The Andoni-Naor-Neiman inequalities and isometric embeddability into a CAT(0) space, arXiv:2404.13871, 2024

    T. Toyoda, The Andoni-Naor-Neiman inequalities and isometric embeddability into a CAT(0) space, arXiv:2404.13871, 2024. Kogakuin University, 2665-1, Nakano, Hachioji, Tokyo, 192-0015 Japan Email address : toyoda@cc.kogakuin.ac.jp

    2024 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.