Pith. sign in

REVIEW 1 major objections 4 minor 15 references

The Quasi-hyperbolicity Constant of a Metric Space

T0 review · 1 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read The paper introduces the quasi-hyperbolicity constant and shows that it is a rough isometry invariant in [1,2], equal to 1 on unbounded Gromov hyperbolic spaces, and computable exactly for L_p spaces and snowflaked lines.

desk verdict A clean new invariant with exact Lp and snowflake-line computations; the proofs are direct except for one intricate optimization that looks right. read the letter →

arxiv 1908.04440 v2 pith:GKZZTYDE submitted 2019-08-12 math.MG math.GT

classification math.MGmath.GT MSC 51K0546B2051F9951M10
keywords quasi-hyperbolicityconstantfour-pointinequalityGromovhyperbolicroughisometryinvariantJamesroundnessL_pspacessnowflakemetric
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 introduces the quasi-hyperbolicity constant $C(X,d)$, the smallest multiplier $\mu$ in a four-point inequality that may carry an additive error $\delta$, and proposes it as a rough-isometry invariant measuring how far a metric space deviates from Gromov hyperbolicity. For unbounded spaces $C$ lies in $[1,2]$, and it equals $1$ for every unbounded Gromov hyperbolic space. The paper's exact calculations are $C(L_p)=\max\{2^{1/p},2^{1-1/p}\}$ for any separable $L_p$ space of dimension at least two, $C(B)\ge\sqrt{2}$ for every Banach space of dimension at least two, and $C(\mathbb{R},d_E^\alpha)=m^\alpha$ for the $\alpha$-snowflake of the Euclidean line, where $m$ solves $(m-1)^\alpha+(m+1)^\alpha=2$. These results give a computable large-scale invariant that distinguishes Euclidean, Banach, and snowflaked geometries.

What carries the argument

The load-bearing object is the $(\mu,\delta)$-four-point inequality $xy+zw\le\mu\max\{xz+yw,xw+yz\}+2\delta$ together with the ratio $\Delta(x,y,z,w)=(xy+zw)/\max\{xz+yw,xw+yz\}$. For Banach spaces the proof uses the James constant $J(B)$ as a lower bound and the roundness invariant $r(B)$ as an upper bound via $C(B)\le 2^{1/r(B)}$. For the snowflaked line, the argument reduces $C(\mathbb{R},d_E^\alpha)$ to a two-variable maximization of two rational functions $F$ and $G$ over a compact domain, and Lemmas 6.10 and 6.13 show the maximum is attained on the diagonal $t=s$, converting the problem into the scalar equation $(m-1)^\alpha+(m+1)^\alpha=2$.

What would settle it

For any fixed $\alpha\in(0,1)$, numerically maximize $\Delta(x,y,z,w)=(|x-y|^\alpha+|z-w|^\alpha)/\max\{|x-z|^\alpha+|y-w|^\alpha,|x-w|^\alpha+|y-z|^\alpha\}$ over four points of $\mathbb{R}$; if the maximum exceeds $m^\alpha$ with $(m-1)^\alpha+(m+1)^\alpha=2$, the theorem is false. Equivalently, evaluate the left side of equation (6.14) on a dense grid of $(a,b)$ with $-1<a<b<1$ and $a+b>0$; any zero would break the sign analysis behind Lemma 6.13.

Watch

Extended reading notes

Core claim

The central claim is that the infimum over $\mu$ in the $(\mu,\delta)$-four-point inequality defines a meaningful numerical invariant of a metric space, and that this invariant can be computed exactly for broad classes of spaces. In the paper's terms: for any separable measure space with $\dim L_p\ge 2$, $C(L_p(\Omega,\Sigma,\mu))=\max\{2^{1/p},2^{1-1/p}\}$; for any Banach space of dimension at least two, $C(B)\ge\sqrt{2}$; for any CAT(0) space, $C_0\le\sqrt{2}$; and for the $\alpha$-snowflake of the real line, $C(\mathbb{R},d_E^\alpha)=m^\alpha$ with $(m-1)^\alpha+(m+1)^\alpha=2$. The constant is a rough isometry invariant, takes values in $[1,2]$ on unbounded spaces, and equals $1$ for unbounded Gromov hyperbolic spaces; a proper CAT(0) space with $C=1$ is necessarily Gromov hyperbolic.

Load-bearing premise

The exact snowflake-line value rests on Lemma 6.13, which asserts—by sign analysis of equation (6.14) and monotonicity of $t\mapsto (1-t^{1-\alpha})/(1-t)^{1-\alpha}$—that the maximum on the curve $F=G$ occurs at $t=s$; if that optimization step hides an error, the formula $C(\mathbb{R},d_E^\alpha)=m^\alpha$ is unproved.

Editorial extensions

If this is right

  • Unbounded Gromov hyperbolic spaces have $C=1$, so within a rough isometry class the value $1$ detects large-scale hyperbolicity; for proper CAT(0) spaces the converse also holds.
  • Every Banach space of dimension at least two has $C(B)\ge\sqrt{2}$, and if the dimension is at least three then equality $C(B)=\sqrt{2}$ forces $B$ to be a Hilbert space.
  • For separable $L_p$ spaces, $C(L_p)=\max\{2^{1/p},2^{1-1/p}\}$, so the constant is minimized at $p=2$ with value $\sqrt{2}$ and approaches $2$ as $p\to 1$ or $p\to\infty$.
  • For any metric space and any $0<\alpha\le 1$, $C_0(X,d^\alpha)\le 2^\alpha$, and this upper bound is sharp for $(\mathbb{R}^n,d_\infty^\alpha)$ with $n\ge 2$.
  • Snowflaking the real line strictly raises the quasi-hyperbolicity constant above $1$ for every $0<\alpha<1$; in particular $C(\mathbb{R},d_E^{1/2})=\sqrt{5}/2$.

Reading between the lines

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

  • The combination of the James-constant lower bound and roundness upper bound suggests a numerical scale on which Hilbert spaces are the roundest Banach spaces; whether $C(B)=\sqrt{2}$ characterizes Hilbert spaces in all dimensions is a natural open test.
  • The diagonal-maximization step for the snowflake line is a transferable strategy: any one-dimensional metric with a scaling action and a two-variable ratio may reduce to a scalar equation, so the same method could give exact constants for snowflakes of other normed lines.
  • The conjecture $C(\mathbb{R}^n,d_2^\alpha)=2^{\alpha/2}$ for $n\ge 2$ is directly testable by the same $\Delta$-ratio numerics; the line case's deviation from $2^{\alpha/2}$ shows the constant genuinely depends on dimension, not only on the snowflake exponent.
  • Because Proposition 2.9 identifies $C=C_0$ for snowflaked normed spaces, the restricted constant $C_0$ is the right computational target for such spaces, and differences between $C_0$ and $C$ measure small-scale versus large-scale geometry.
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

1 major / 4 minor

Summary. The paper introduces the quasi-hyperbolicity constant C(X,d) of a metric space, defined as the infimal µ for which a (µ,δ)-four-point inequality holds for some δ ≥ 0, together with a restricted version C0(X,d) with δ = 0. The authors prove basic properties: bounded spaces have C = 0, unbounded spaces have C ∈ [1,2], C is a rough isometry invariant, C = C0 for four-point scalable spaces, and unbounded Gromov hyperbolic spaces have C = 1. They show C0 ≤ √2 for Ptolemaic 2-round spaces and hence for subspaces of CAT(0)-spaces, and C ≥ √2 for every Banach space of dimension at least two, with a Hilbert-space rigidity statement in dimension at least three. For Lp-spaces of dimension at least two they obtain the exact value C(Lp) = max{2^{1/p}, 2^{1−1/p}}. For snowflaked metrics they prove the general bound C0(X,d^α) ≤ 2^α and compute the exact value for the α-snowflake of the Euclidean line: C(R, d_E^α) = m^α, where m ≥ 1 solves (m−1)^α + (m+1)^α = 2. The paper also contains examples clarifying the difference between C and C0, the behavior under quasi-isometry, and a lower bound C0 ≥ √2 for Riemannian manifolds of dimension greater than one.

Significance. If the results hold, the paper introduces a useful numerical rough-isometry invariant that quantifies deviation from Gromov hyperbolicity and takes values in [1,2] for unbounded spaces. The Lp calculation is a highlight: it combines external sharp results on the James constant and roundness with explicit four-point configurations, and it is free of fitted parameters. The exact snowflake-line value in Theorem 6.6 is a nontrivial optimization result and is likely to be of independent interest. The paper also carefully documents the failure of quasi-isometry invariance for non-intrinsic spaces and gives clean examples separating C from C0. Overall, the central claims are well supported by explicit and checkable arguments.

major comments (1)
  1. [Lemma 6.10 and Theorem 6.6] The proof of Lemma 6.10 states that the boundary of D1 is D0 ∪ {(t,s) ∈ D : t+s = 1} and that F(t,s) = 1 if and only if t+s = 1. This is incorrect: on D, F takes the value 1 on the segment t+s = 0 (for example F(0,0) = 1), while along t+s = 1 the values of F are generally larger than 1. Since the proof uses the claim F = 1 on this boundary component to discard it in favor of D0, the written proof is not valid as it stands. The intended argument is readily repaired by replacing t+s = 1 with t+s = 0; with that correction, the reduction of the maximum to D0 goes through. The correction should be made explicitly because this lemma is load-bearing for the exact value in Theorem 6.6.
minor comments (4)
  1. [Proposition 2.2(i)] In the proof of Proposition 2.2(i), the text reads "xy + zw ≤ (xz + yw) + (xw + zw)" and later repeats "xw + zw"; both occurrences should read "xw + yz".
  2. [Lemma 6.13] The proof of Lemma 6.13 asserts without justification that Fs(a,b) − Gs(a,b) ≠ 0 before applying the implicit function theorem. This is a necessary condition for the argument; please add a short verification or a reference to a calculation establishing that the difference does not vanish on D0.
  3. [Lemma 6.13] After treating the case a < b in detail, the proof states that the expression in (6.14) is positive if a > b without showing the corresponding variable transformation. A brief indication of the symmetric argument would improve readability and completeness.
  4. [Question 3.7] There is a typo: "Quesition 3.7" should be "Question 3.7".

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the paper's central claims are supported by independent external results and explicit computations.

full rationale

The quasi-hyperbolicity constant C(X,d) is defined directly from the (µ,δ)-four-point inequality, and the paper's main results are obtained either by elementary inequalities or by importing independent, externally established Banach-space theorems. The Lp-space formula, Corollary 5.12, combines the lower bound C(B) ≥ J(B) with the James-constant results of Gao–Lau and Komuro–Saito–Tanaka, plus the roundness upper bound C(B) ≤ 2^{1/r(B)} using Enflo's and Lennard–Tonge–Weston's roundness values; none of these inputs already contains the target formula, and the lower bound for ℓp^2 is exhibited by an explicit four-point configuration in Proposition 5.2. The snowflake-line calculation, Theorem 6.6, is a self-contained optimization over the compact domain D, with the reduction to the diagonal F=G curve proved in Lemmas 6.10 and 6.13 rather than assumed. No fitted parameter is renamed as a prediction, no result is justified solely by a self-citation (the cited Nica–Špakula, Bridson, Enflo, Gao–Lau, Komuro–Saito–Tanaka, and Lennard–Tonge–Weston results are prior external work), and no uniqueness theorem from the authors' own prior work is invoked. The occasional textual slips in Lemma 6.10's boundary description do not make the derivation circular; they are local presentation issues rather than reduction of a conclusion to its premise.

Assumptions & free parameters 0 free parameters · 9 assumptions · 0 invented entities

No numbers are fitted; m in Theorem 6.6 is the unique solution of a stated equation. The paper's original contributions are the definition of C and C0 and the comparisons; the ledger lists the external theorems it relies on. The only implicit extra premise is the invariance of hyperbolicity under rough isometry for non-intrinsic spaces, flagged in Proposition 3.1.

assumptions (9)
  • standard math A metric space is δ-hyperbolic iff it satisfies the (1,δ)-four-point inequality.
    Background characterization cited to Väisälä (2.12); used to define C and prove Proposition 2.6.
  • standard math CAT(0) spaces are Ptolemaic and 2-round.
    Used in Theorem 4.2 and Corollary 4.3; follows from Euclidean subembeddings in BH99.
  • standard math J(B) ≥ √2 for every non-trivial Banach space.
    Gao-Lau Theorem 2.5, used in Theorem 5.8.
  • standard math If dim B ≥ 3 and J(B)=√2 then B is Hilbert.
    Komuro-Saito-Tanaka, used in Theorem 5.8.
  • standard math r(Lp)=p for 1≤p≤2 and r(Lp)=1/(1-1/p) for 2≤p≤∞.
    Enflo and Lennard-Tonge-Weston results, used in Corollary 5.12 via Theorem 5.11.
  • standard math Every separable Lp space is isometric to one of the spaces in list (5.13).
    Classification from JL01 §4, used in Corollary 5.12 to get lower bounds from ℓ²_p subspaces.
  • standard math A proper CAT(0) space that is not Gromov hyperbolic contains an isometric Euclidean plane.
    Bridson Flat Plane Theorem, used in Proposition 3.8.
  • standard math (R^n,d₂^α) embeds isometrically into Hilbert space for 0<α≤1.
    Schoenberg 1937, used in Proposition 6.4.
  • domain assumption A space roughly isometric to a Gromov hyperbolic space is itself Gromov hyperbolic even without intrinsicness.
    Invoked implicitly in Proposition 3.1; not proven or cited there, though it follows from Lemma 2.13 with µ=1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Quasi-hyperbolicity Constant of a Metric Space." pith.science (2026). https://pith.science/paper/GKZZTYDE

@misc{pith2026190804440,
  author       = {Pith},
  title        = {Pith review of: The Quasi-hyperbolicity Constant of a Metric Space},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GKZZTYDE}},
  note         = {Machine review of arXiv:1908.04440}
}
abstract

We introduce the quasi-hyperbolicity constant of a metric space, a rough isometry invariant that measures how a metric space deviates from being Gromov hyperbolic. This number, for unbounded spaces, lies in the closed interval $[1,2]$. The quasi-hyperbolicity constant of an unbounded Gromov hyperbolic space is equal to one. For a CAT$(0)$-space, it is bounded from above by $\sqrt{2}$. The quasi-hyperbolicity constant of a Banach space that is at least two dimensional is bounded from below by $\sqrt{2}$, and for a non-trivial $L_p$-space it is exactly $\max\{2^{1/p},2^{1-1/p}\}$. If $0 < \alpha < 1$ then the quasi-hyperbolicity constant of the $\alpha$-snowflake of any metric space is bounded from above by $2^\alpha$. We give an exact calculation in the case of the $\alpha$-snowflake of the Euclidean real line.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 14 canonical work pages

  1. [1]

    Bridson and Andr\'e Haefliger, Metric spaces of non-positive curvature, Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences], vol

    Martin R. Bridson and Andr\'e Haefliger, Metric spaces of non-positive curvature, Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences], vol. 319, Springer-Verlag, Berlin, 1999. 1744486

  2. [2]

    Blumenthal, Theory and applications of distance geometry, Second edition, Chelsea Publishing Co., New York, 1970

    Leonard M. Blumenthal, Theory and applications of distance geometry, Second edition, Chelsea Publishing Co., New York, 1970. 0268781

  3. [3]

    I. D. Berg and I. G. Nikolaev, Quasilinearization and curvature of A leksandrov spaces , Geom. Dedicata 133 (2008), 195--218. 2390077

  4. [4]

    Bridson, On the existence of flat planes in spaces of nonpositive curvature, Proc

    Martin R. Bridson, On the existence of flat planes in spaces of nonpositive curvature, Proc. Amer. Math. Soc. 123 (1995), no. 1, 223--235. 1273477

  5. [5]

    Per Enflo, On the nonexistence of uniform homeomorphisms between L_ p -spaces , Ark. Mat. 8 (1969), 103--105. 0271719

  6. [6]

    Ji Gao and Ka-Sing Lau, On the geometry of spheres in normed linear spaces, J. Austral. Math. Soc. Ser. A 48 (1990), no. 1, 101--112. 1026841

  7. [7]

    Gromov, Hyperbolic groups, Essays in group theory, Math

    M. Gromov, Hyperbolic groups, Essays in group theory, Math. Sci. Res. Inst. Publ., vol. 8, Springer, New York, 1987, pp. 75--263. 919829

  8. [8]

    Johnson and Joram Lindenstrauss, Basic concepts in the geometry of B anach spaces , Handbook of the geometry of B anach spaces, V ol

    William B. Johnson and Joram Lindenstrauss, Basic concepts in the geometry of B anach spaces , Handbook of the geometry of B anach spaces, V ol. I , North-Holland, Amsterdam, 2001, pp. 1--84. 1863689

Show all 15 references
  1. [9]

    Naoto Komuro, Kichi-Suke Saito, and Ryotaro Tanaka, On the class of B anach spaces with J ames constant 2 , Math. Nachr. 289 (2016), no. 8-9, 1005--1020. 3512046

  2. [10]

    C. J. Lennard, A. M. Tonge, and A. Weston, Generalized roundness and negative type, Michigan Math. J. 44 (1997), no. 1, 37--45. 1439667

  3. [11]

    Bogdan Nica and J\' a n S pakula, Strong hyperbolicity, Groups Geom. Dyn. 10 (2016), no. 3, 951--964. 3551185

  4. [12]

    Takashi Sato, An alternative proof of B erg and N ikolaev's characterization of CAT(0) -spaces via quadrilateral inequality , Arch. Math. (Basel) 93 (2009), no. 5, 487--490. 2563595

  5. [13]

    I. J. Schoenberg, On certain metric spaces arising from E uclidean spaces by a change of metric and their imbedding in H ilbert space , Ann. of Math. (2) 38 (1937), no. 4, 787--793. 1503370

  6. [14]

    38, Longman Scientific & Technical, Harlow; copublished in the United States with John Wiley & Sons, Inc., New York, 1989

    Nicole Tomczak-Jaegermann, Banach- M azur distances and finite-dimensional operator ideals , Pitman Monographs and Surveys in Pure and Applied Mathematics, vol. 38, Longman Scientific & Technical, Harlow; copublished in the United States with John Wiley & Sons, Inc., New York,...

  7. [15]

    a is \"a l \

    Jussi V \"a is \"a l \"a , Gromov hyperbolic spaces, Expo. Math. 23 (2005), no. 3, 187--231. 2164775

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.