Schoenberg Parameters Overview

• Schoenberg parameters are explicit coordinate systems that encode positivity-preserving structures, linking total positivity, kernel expansions, and operator approximations.
• They arise in various contexts such as Laguerre–Pólya factorization for Pólya frequency functions, Gegenbauer expansions on spheres, and Toeplitz total positivity.
• Their applications extend to deriving sharp inequalities in polynomial geometry, noncommutative analysis, and robust statistical methods.

Schoenberg parameters are parameter systems arising in several lines of work initiated by I. J. Schoenberg. In contemporary usage, the phrase does not denote a single invariant; rather, it refers to explicit coefficient sets, node systems, or generator data that appear in Schoenberg-type characterizations of total positivity, positive definite kernels, polynomial critical-point inequalities, approximation operators, and noncommutative positivity. In the most classical analytic setting, they are the parameters in the Hadamard factorization of a Laguerre–Pólya entire function associated with a totally positive function by Laplace transform (Gröchenig, 2020).

1. Range of meanings

In the cited literature, “Schoenberg parameters” denotes several technically distinct objects.

Context	Parameters	Role
Totally positive functions	$(C,\gamma,\delta,(\delta_j))$ with $m$ in the factorization of $\Psi$	Parametrize Laguerre–Pólya functions and totally positive functions
Positive definite kernels on spheres	$a_k$, $a_{k,l}$, $b_{n,d}$, $\varphi_{n,d}$	Coefficients or function-valued coefficients in Gegenbauer expansions
Toeplitz total positivity	$(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$	Edrei–Schoenberg factorization of totally nonnegative Toeplitz symbols
Polynomial geometry	$C_p$, or the $b_k$ in $m$0	Sharp constants or derivative-root parameters
Approximation operators	$m$1	Sampling nodes in Schoenberg and CCC-Schoenberg operators

This distribution of meanings is explicit in the literature on totally positive functions (Gröchenig, 2020), products of spheres (Guella et al., 2015), sphere–group kernels (Berg et al., 2015), Toeplitz parametrizations (Rietsch, 29 Sep 2025), Schoenberg type inequalities (Tang, 14 Aug 2025), and CCC-Schoenberg operators (Bosner, 2021). A plausible implication is that the term now functions as a family resemblance notion: it marks explicit coordinate systems attached to positivity-preserving structures rather than a single universal definition.

2. Totally positive functions and Laguerre–Pólya factorization

For a function $m$2, total positivity means that for every $m$3 and all strictly increasing sequences $m$4 and $m$5, one has

$m$6

If $m$7 is also integrable, it is a Pólya frequency function (Gröchenig, 2020).

Schoenberg’s fundamental theorem identifies the Laplace transforms of Pólya frequency functions with reciprocals of Laguerre–Pólya entire functions. If $m$8 is totally positive and integrable, then

$m$9

in a vertical strip, where $\Psi$0 is an entire function in the Laguerre–Pólya class with $\Psi$1; conversely, every such reciprocal Laplace transform comes from an integrable totally positive function (Gröchenig, 2020).

The associated factorization is

$\Psi$2

with

$\Psi$3

The paper describes the real parameters $\Psi$4 as a parametrization by the set $\Psi$5, and states that these parameters uniquely encode all totally positive functions via the Laplace transform relationship (Gröchenig, 2020). In the Fourier domain the same data appear in

$\Psi$6

A one-sided version holds for functions supported on $\Psi$7: $\Psi$8 with $\Psi$9 and $a_k$0 (Gröchenig, 2020). The same paper emphasizes the variation diminishing property of Pólya frequency functions and an operator-theoretic characterization via $a_k$1, which is zero-decreasing.

The number-theoretic significance is that the completed Riemann $a_k$2-function belongs to the Laguerre–Pólya class exactly when the Riemann hypothesis holds. Schoenberg’s theorem then yields the equivalence: the Riemann hypothesis holds if and only if there exists a Pólya frequency function $a_k$3 such that

$a_k$4

for $a_k$5, where $a_k$6 is the imaginary part of the first zero of $a_k$7 on the critical line; equivalently,

$a_k$8

is a Pólya frequency function if and only if the Riemann hypothesis holds (Gröchenig, 2020).

3. Positive definite kernels on spheres and groups

In Schoenberg’s theorem for a single sphere, a continuous isotropic positive definite kernel on $a_k$9 has isotropic part

$a_{k,l}$0

with $a_{k,l}$1 and

$a_{k,l}$2

The coefficients $a_{k,l}$3 are called Schoenberg’s parameters in the product-of-spheres literature (Guella et al., 2015).

On $a_{k,l}$4, the corresponding characterization is

$a_{k,l}$5

where $a_{k,l}$6 and

$a_{k,l}$7

Thus the single-index coefficients become doubly indexed coefficients $a_{k,l}$8, which the paper presents as the generalized Schoenberg parameters for products of spheres (Guella et al., 2015). For $a_{k,l}$9 and $b_{n,d}$0, Gegenbauer polynomials are replaced by monomials: $b_{n,d}$1

For isotropic positive definite kernels on a single sphere, the normalized expansion

$b_{n,d}$2

defines the $b_{n,d}$3-Schoenberg coefficients $b_{n,d}$4; when $b_{n,d}$5, the sequence $b_{n,d}$6 is a probability mass function (Arafat et al., 2018). The same paper gives the integral formula

$b_{n,d}$7

and, for $b_{n,d}$8, identifies the $b_{n,d}$9-Schoenberg coefficients with Fourier coefficients. It also provides explicit formulas expressing $\varphi_{n,d}$0-Schoenberg coefficients as linear combinations of $\varphi_{n,d}$1-Schoenberg coefficients and uses them to study the exponential and Askey families, together with lower bounds for curvature at the origin of locally supported kernels (Arafat et al., 2018).

A further extension replaces scalar coefficients by positive definite functions on a locally compact group $\varphi_{n,d}$2. For $\varphi_{n,d}$3 with

$\varphi_{n,d}$4

positive definiteness on $\varphi_{n,d}$5 is equivalent to an expansion

$\varphi_{n,d}$6

where each $\varphi_{n,d}$7 is a continuous positive definite function on $\varphi_{n,d}$8 and $\varphi_{n,d}$9 (Berg et al., 2015). The sequence $(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$0 is called the $(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$1-Schoenberg sequence of functions, or the $(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$2-Schoenberg functions. For $(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$3 the expansion becomes

$(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$4

4. Toeplitz total positivity and asymptotic representation theory

For an infinite normalized upper-triangular Toeplitz matrix with generating series

$(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$5

Edrei’s theorem gives the factorization

$(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$6

where $(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$7, the sequences $(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$8 and $(\gamma,\boldsymbol{\alpha},\boldsymbol{\beta})$9 are nonnegative, monotonically decreasing, and satisfy $C_p$0 (Rietsch, 8 Sep 2025). In this setting, the pairs of sequences $C_p$1 are the Schoenberg parameters, and nonzero parameters correspond to the roots and poles of the naturally associated generating function (Rietsch, 29 Sep 2025).

These parameters are central in the classification of characters of the infinite symmetric group. Thoma’s theorem identifies the extremal characters of $C_p$2 using the same parameter pairs, now constrained by

$C_p$3

and the cycle values are

$C_p$4

(Rietsch, 29 Sep 2025). Vershik–Kerov then interpret the same data asymptotically through normalized Frobenius coordinates of Young diagrams: $C_p$5 (Rietsch, 29 Sep 2025).

Recent work describes two further asymptotic realizations through quantum cohomology of the flag variety. If $C_p$6 are diagonal parameters extracted from minors of finite totally positive Toeplitz truncations, then

$C_p$7

Similarly, for normalized quantum parameters $C_p$8,

$C_p$9

(Rietsch, 29 Sep 2025).

The tropical analogue replaces positive reals by a valued semifield. In that setting the tropical Schoenberg parameters are weakly increasing sequences $b_k$0 and $b_k$1, and the infinite tropical Toeplitz data are recovered by

$b_k$2

The paper proves an asymptotic relation between finite tropical parameters and tropicalizations of the Schoenberg parameters, and identifies the finite tropical parametrization map with Lusztig’s weight map (Rietsch, 8 Sep 2025). This suggests that the classical Schoenberg parameters admit parallel analytic, asymptotic, and tropical realizations.

5. Polynomial geometry and Schoenberg type inequalities

In the geometry of polynomials, Schoenberg’s original inequality concerns a polynomial of degree $b_k$3 with zeros $b_k$4 and critical points $b_k$5 under the centroid condition $b_k$6. The quadratic form is

$b_k$7

and the centroid-free form is

$b_k$8

(Tang, 14 Apr 2025).

In this literature, “Schoenberg parameters” can denote the sharp constants in inequalities of the form

$b_k$9

under the centroid condition (Tang, 14 Aug 2025). An interpolation-based treatment yields the sharp constants

$m$00

thereby extending the theory to all $m$01 (Tang, 14 Aug 2025). The same line of work reports a new proof of the order-$m$02 inequality and states that the result completely resolves an open problem posed by Kushel and Tyaglov.

Complementary results establish an order-six Schoenberg type inequality and a novel order-one inequality. Under $m$03,

$m$04

and an explicit sextic inequality is given with equality if and only if all zeros are collinear (Tang, 14 Apr 2025). The same paper also derives a majorization-based inequality for elementary symmetric functions,

$m$05

and connects the resulting estimates to Sendov’s conjecture (Tang, 14 Apr 2025).

A different noncommutative usage occurs in the C$m$06-algebraic Schoenberg conjecture. If

$m$07

with $m$08, then the elements $m$09 are called the Schoenberg parameters for the derivative (Krishna, 2022). The conjectured inequalities are

$m$10

and

$m$11

The paper proves the conjecture for degree $m$12, where equality holds (Krishna, 2022).

6. Operators, noncommutative correspondences, and applied extensions

In approximation theory, the classical Schoenberg operator

$m$13

is built from Greville nodes $m$14 and normalized B-splines $m$15 on equidistant knots (Nagler et al., 2013). The same paper lists parameters associated with the operator: the mesh width $m$16, the stability constant $m$17, and the quantity $m$18 computed from B-spline differences. These govern estimates such as the lower bound

$m$19

and the equivalence

$m$20

for the second order modulus of smoothness (Nagler et al., 2013).

For CCC-splines, the CCC-Schoenberg operator is

$m$21

and the nodes

$m$22

are explicitly called the Schoenberg parameters (Bosner, 2021). They are usually taken as CCC-Greville points, although any strictly increasing sequence with the stated support property may be used if the operator reproduces $m$23. Their placement controls approximation, shape preservation, and quasi-collocation error bounds such as

$m$24

(Bosner, 2021).

Several noncommutative generalizations retain the same parametrizing role. For generalized Schur products, the noncommutative Schoenberg theorem states that a positivity-preserving noncommutative function on Schur spectral contractions must have a power-series expansion

$m$25

so the coefficients $m$26 are the decisive positivity parameters (Pascoe, 2019). For semigroups on cones of $m$27-positive, $m$28-superpositive, or $m$29-entanglement breaking maps, the generator condition

$m$30

is described as a Schoenberg parameterization of the generator $m$31 (Bhat et al., 2023). On dual groups, Schoenberg correspondence identifies conditionally positive functionals $m$32 as generators of convolution semigroups of states through

$m$33

with convolution exponential defined by

$m$34

(Schürmann et al., 2012).

In robust statistics, Schoenberg transformations

$m$35

act on squared Euclidean distances, and the parameters of the family—either the measure $m$36 or special-case scalars such as $m$37, $m$38, or $m$39—are explicitly called Schoenberg parameters (Bavaud, 2011). Examples include the power transform $m$40 with $m$41, the exponential transform $m$42, and the logarithmic transform $m$43. These parameters determine the transformed inertia

$m$44

the weights

$m$45

and the transition between distributed and concentrated regimes of robust location estimates (Bavaud, 2011).

Across these settings, Schoenberg parameters serve as explicit coordinates on positivity phenomena: they encode total positivity, positive definiteness, spectral generators, approximation nodes, or sharp contraction constants. This suggests a unifying editorial description: Schoenberg parameters are the data that make Schoenberg-type classification theorems concrete.