Pith. sign in

REVIEW 3 minor 13 references

Optimal differentiability of isotropic positive definite functions on even-dimensional spheres

T0 review · 0 major / 3 minor · reviewed 2026-08-27 · deepseek-v4-flash

Pith's one-line read For every even sphere and every k, there exists a positive definite function whose interior smoothness is exactly the guaranteed bound and no higher.

desk verdict The long-missing even-dimensional optimality examples are here, with a proof that holds up. read the letter →

arxiv 2608.26092 v1 pith:74RSP5FH submitted 2026-08-26 math.CA

classification math.CA MSC 42A8233C4542C10
keywords positivedefinitefunctionsspheresSchoenbergcoefficientsGegenbauerpolynomialsdifferentiabilitydimensionwalksevendimensionsPoissonkernel
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 settles the last open case of a long-standing regularity problem. For an isotropic positive definite function on the sphere S^d whose even continuation has 2k derivatives at zero, a known theorem guarantees 2k+floor((d-1)/2) continuous derivatives in the interior; optimality of that order was known for every odd dimension but not for even dimensions. This paper constructs, for every even d=2m and every k>=0, a strictly positive definite function that reaches exactly the guaranteed order: it is $C^{{2k+m-1}}$ on (0,pi), its derivative of order 2k+m fails at $\theta$=pi/2, its even continuation is 2k but not 2k+2 times differentiable at zero, and it is not positive definite in dimension d+1. The construction starts from a sparse Legendre series on $S^{2}$ whose equator is a genuine point of nondifferentiability and lifts it to all even dimensions by dimension-walk operators.

What carries the argument

Schoenberg's coefficient characterization turns the cone Psi_d into nonnegative sequences against normalized Gegenbauer polynomials, and the moment criterion Theorem 2.2 ties 2q-fold differentiability at zero to summability of $n^{{2q}}$ times the coefficients. The dimension walks, turning bands from S^d to $S^{{d+2}}$ and spherical montee from S^D to $S^{{D-2}}$, transfer the missing derivative upward and the pole regularity upward, with pointwise lemmas preserving the failure of a derivative at $\theta$=pi/2 under each operation. The analytic core is Lemma 4.2, a Poisson-regularization statement for $S^{2}$: if the boundary function is differentiable at a point with a uniform remainder, the tangential derivatives of its Poisson means converge to the directional derivative.

What would settle it

Compute the one-sided difference quotients ($\varphi$(pi/2+h)-$\varphi$(pi/2))/h for the explicit seed function phi in (4.2), using partial sums through very large j; if both sides converge to the same finite value, Proposition 4.3 is false and Theorem 1.2 collapses. More broadly, any single function in Psi_{2m} whose even continuation has 2k derivatives at zero but which lies in $C^{{2k+m}}$ on (0,pi) would refute the claimed optimality.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1.2: for every even dimension d=2m>=2 and every k>=0 there exists a function psi_{d,k} in Psi_d^+ with all five sharpness properties, hence 2k+floor((d-1)/2) is the largest differentiability order forced by the hypothesis at the origin. The engine is a two-dimensional seed $\varphi$($\theta$)=sum_{j>=0} a_j P_{4j+1}(cos $\theta$) with a_j=(j+1)^{-3/2}/zeta(3/2). The coefficients are summable, so Schoenberg's theorem puts phi in Psi_2, but the derivative series at the equator diverges because Legendre derivative values grow like $j^{{1/2}}$. The paper proves the derivative genuinely fails to exist by showing that if it existed, the tangential derivative of the Poisson means would converge to it, while monotone convergence forces those means to tend to -infinity. Turning bands transfer the missing derivative to every even dimension, and the spherical montee raises the pole regularity by two while raising the order of the missing equatorial derivative by one.

Load-bearing premise

The entire construction depends on Lemma 4.2, which assumes that differentiability of the boundary function at the equator has a remainder that is uniform over all approach directions; if that uniform control can fail even when the pointwise derivative exists, the contradiction proving that the two-dimensional seed is nondifferentiable would not go through.

Editorial extensions

If this is right

  • For every even d=2m and every k>=0, there is a strictly positive definite isotropic kernel whose interior regularity is exactly the guaranteed order: C^{2k+m-1} but not C^{2k+m}.
  • The sharp index is now known in all dimensions: r_*(d,k)=2k+floor((d-1)/2) for every d>=1 and k>=0.
  • In the even-dimensional construction, the first non-guaranteed derivative can always be made to fail at the fixed interior point theta=pi/2.
  • The examples are not positive definite in the next dimension, so the bound is genuinely dimension-specific and not inherited by extension to S^{d+1}.
  • For the base case k=0, d=2M, the theorem provides functions in Psi_{2M}^+ that are C^{M-1} on (0,pi) but do not have an M-th derivative at the equator.

Reading between the lines

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

  • The sparse-coefficient seed suggests a general recipe: a single residue class of Gegenbauer degrees, with coefficients at the critical exponent between summability and derivative divergence, can encode a point of nondifferentiability while keeping the kernel positive definite; the same recipe may yield sharp regularity examples on other two-point homogeneous spaces.
  • A testable extension would be to rotate or conformally adjust the seed so that the missing derivative occurs at any prescribed interior point rather than only at the equator; the paper does not state such a localization, but its transfer mechanisms suggest the obstruction is movable.
  • In covariance-model terms, the result implies there exist Gaussian random fields on every even-dimensional sphere whose sample-path differentiability at one interior direction is exactly one derivative lower than their pole regularity predicts, which may matter for procedures assuming uniform interior smoothness.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Request a human review

A listed scientist reviews the paper for a fee and the review publishes here regardless of verdict. See the reviewers or get listed.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

0 major / 3 minor

Summary. The paper proves optimality of the differentiability bound for isotropic positive definite functions on even-dimensional spheres. It constructs, for every d = 2m ≥ 2 and k ≥ 0, a strictly positive definite isotropic function ψ_{d,k} on S^d whose even continuation is 2k times differentiable at zero, which lies in C^{2k+m-1}(0,π), whose (2k+m)-th derivative at π/2 does not exist, whose even continuation is not 2k+2 times differentiable at zero, and which is not positive definite in dimension d+1. The construction starts with a sparse Legendre series on S^2 with coefficients at the threshold of summability, proves nondifferentiability at the equator via a Poisson-integral lemma (Lemma 4.2), and transfers the example to all even dimensions using turning bands and spherical montée. The paper thereby establishes that the bound in Theorem 1.1 is optimal in every dimension, completing earlier odd-dimensional results.

Significance. If correct, this settles an open problem left open in Ziegel (2014) and Trübner–Ziegel (2017). The proof is detailed and appears complete: the key analytic lemma is proved with explicit uniform estimates, the seed construction is explicit with coefficients normalized by ζ(3/2), and the dimension-walk chains rest on established theorems. The construction is parameter-free in the sense that no constants are fitted to force the conclusion; every assertion is backed by explicit inequalities. The result is likely to be of interest to approximation theory and to the theory of covariance models on spheres.

minor comments (3)
  1. [Section 6, proof of Corollary 6.1] The phrase 'The first row includes the circled = 1by taking m = 0' appears to be a typo; it should read 'the case d = 1'.
  2. [Throughout] Several inline formulas are run together with surrounding text (e.g., 'from(4.8)', 'π/2)does not exist', 'fork = 0'); these spacing errors should be corrected in production.
  3. [Section 4.2, Lemma 4.2] The differentiability hypothesis in Lemma 4.2 is the uniform-directional remainder condition defined just above the lemma; since this is stronger than ordinary manifold differentiability, it would help to note explicitly in Proposition 4.3 that the verification of this uniform condition is the reason for the uniform remainders in Step 2 of that proof.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity identified; the construction is self-contained and uses prior theorems only as external tools.

full rationale

The paper's central claim, Theorem 1.2, is a new construction of isotropic positive definite functions on even-dimensional spheres with sharp differentiability properties. The derivation chain is self-contained: the two-dimensional seed is built from an explicit Schoenberg sequence (4.1)–(4.2), whose failure of differentiability at the equator is proved directly via the spherical Poisson integral in Lemma 4.2 and Proposition 4.3. The dimension-walk steps in Lemmas 3.2, 3.4, and 3.5 are proven within the paper using pointwise differentiation and the explicit montée formula (3.2)–(3.3), and they transfer the seed to all even dimensions. Prior results are invoked as standard tools: Schoenberg's theorem characterizes \Psi_d, the moment criterion of Trübner and Ziegel is quoted as Theorem 2.2, and Ziegel's interior regularity theorem is used as an upper bound; none of these inputs contains the constructed counterexample or the conclusion of Theorem 1.2. The proof of Lemma 4.2 supplies its own estimates (4.10)–(4.11) and does not rely on the target result. No fitted parameters are renamed as predictions, and no load-bearing argument reduces to a self-citation. The manuscript does not cite the author's own prior work in a way that carries the argument. Therefore the paper exhibits no circularity.

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

The central claim is proven by an explicit construction. The free-parameter list is empty: the coefficients a_j = (j+1)^(-3/2)/ζ(3/2) are a deliberate threshold choice, not a fit to data or a parameter of the result. The paper relies on established theorems from the literature, all cited, and introduces no new entities. The only delicate new lemma, the Poisson localization lemma, is proven in the paper.

assumptions (6)
  • standard math Schoenberg's theorem characterizes Ψ_d as nonnegative Gegenbauer expansions.
    Used to verify that the constructed series belongs to Ψ_2 and is transferable via dimension walks.
  • standard math Trübner-Ziegel moment criterion: the even continuation has a finite 2q-th derivative at 0 iff Σ n^{2q} b_n < ∞.
    The paper uses this to prove that montée raises pole regularity by two (Lemma 3.4).
  • standard math Ziegel's interior regularity theorem gives floor((d-1)/2) continuous derivatives for any ψ ∈ Ψ_d.
    Used in Proposition 3.1 and again in the lower-bound part of Theorem 1.2.
  • standard math Trübner-Ziegel's montée formula for the Schoenberg coefficients of I_S f when the dimension is lowered by two.
    Used in Proposition 3.3 and in the induction of Theorem 1.2.
  • standard math Chen-Menegatto-Sun criterion: strict positive definiteness follows from having positive Schoenberg coefficients at infinitely many even and odd degrees.
    Used to show the constructed functions are strictly positive definite.
  • standard math Standard Gegenbauer polynomial estimates and Stirling's formula from DLMF [4].
    Used to prove Lemma 4.1 and the convergence/divergence of the seed series.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Optimal differentiability of isotropic positive definite functions on even-dimensional spheres." pith.science (2026). https://pith.science/paper/74RSP5FH

@misc{pith2026260826092,
  author       = {Pith},
  title        = {Pith review of: Optimal differentiability of isotropic positive definite functions on even-dimensional spheres},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/74RSP5FH}},
  note         = {Machine review of arXiv:2608.26092}
}
abstract

We prove optimality of the differentiability bound for isotropic positive definite functions on every even-dimensional sphere. If the even continuation of such a function on the $d$-dimensional sphere is $2k$ times differentiable at zero, then the function has $2k+\lfloor(d-1)/2\rfloor$ continuous interior derivatives; previously, optimality was known only in odd dimensions. We construct a function on the two-dimensional sphere whose first derivative does not exist at the equator and transfer it to all even dimensions by turning bands and spherical mont\'ee. The resulting examples are strictly positive definite, have $2k$ but not $2k+2$ derivatives at zero, and are not positive definite in the next dimension.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

  1. [1]

    R. K. Beatson and W. zu Castell,Dimension hopping and families of strictly positive definite zonal basis functions on spheres, J. Approx. Theory221(2017), 22–37

  2. [2]

    D. Chen, V. A. Menegatto, and X. Sun,A necessary and sufficient condition for strictly positive definite functions on spheres, Proc. Amer. Math. Soc.131(2003), 2733–2740

  3. [3]

    Dai and Y

    F. Dai and Y. Xu,Approximation Theory and Harmonic Analysis on Spheres and Balls, Springer Monographs in Mathematics, Springer, 2013

  4. [4]

    NIST Digital Library of Mathematical Functions, Release 1.2.7,https://dlmf.nist.gov/

  5. [5]

    Emery, A

    X. Emery, A. P. Peron, and E. Porcu,Dimension walks on hyperspheres, Comput. Appl. Math. 41(2022), Paper No. 199

  6. [6]

    Gneiting,On the derivatives of radial positive definite functions, J

    T. Gneiting,On the derivatives of radial positive definite functions, J. Math. Anal. Appl.236 (1999), 86–93

  7. [7]

    Gneiting,Strictly and non-strictly positive definite functions on spheres, Bernoulli19 (2013), 1327–1349

    T. Gneiting,Strictly and non-strictly positive definite functions on spheres, Bernoulli19 (2013), 1327–1349

  8. [8]

    Jäger,A note on the derivatives of isotropic positive definite functions on the Hilbert sphere, SIGMA15(2019), 081, 7 pages

    J. Jäger,A note on the derivatives of isotropic positive definite functions on the Hilbert sphere, SIGMA15(2019), 081, 7 pages

Show all 13 references
  1. [9]

    Matheron,Les Variables Régionalisées et Leur Estimation, Masson et Cie, Paris, 1965

    G. Matheron,Les Variables Régionalisées et Leur Estimation, Masson et Cie, Paris, 1965

  2. [10]

    G.Matheron,Quelques aspects de la montée, NoteGéostatistique120, CentredeGéostatistique, Fontainebleau, France, 1972

  3. [11]

    I. J. Schoenberg,Positive definite functions on spheres, Duke Math. J.9(1942), 96–108

  4. [12]

    Trübner and J

    M. Trübner and J. F. Ziegel,Derivatives of isotropic positive definite functions on spheres, Proc. Amer. Math. Soc.145(2017), 3017–3031

  5. [13]

    J. F. Ziegel,Convolution roots and differentiability of isotropic positive definite functions on spheres, Proc. Amer. Math. Soc.142(2014), 2063–2077. Department of Mathematics, School of Science, North China University of Technol- ogy, Beijing 100144, China Email address:yge@n...

Pith tools

Reviewed August 27, 2026 · model on record in the stance chip above.