pith. sign in

arxiv: 2409.18136 · v1 · submitted 2024-09-08 · 🧮 math.CA · math.FA

Inequalities for exponential polynomials with applications to moment sequences

Ognyan Kounchev , Hermann Render , Tsvetomir Tsachev This is my paper

Pith reviewed 2026-05-23 20:36 UTC · model grok-4.3

classification 🧮 math.CA math.FA MSC 26D1534A4044A60
keywords exponential polynomialsmoment sequenceslinear differential operatorspolynomial inequalitiespositivity preservationmoment problem
0
0 comments X

The pith

If the (n+1)th derivative of the solution to a linear differential equation stays non-negative on an interval, then a weighted sum of its lower derivatives bounds every non-negative polynomial of degree at most n from below.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that when Φ_Λ_n solves the nth-order linear differential equation with constant coefficients given by the product of (d/dx - λ_j), satisfies the initial conditions that its derivatives up to order n-1 vanish at zero while the nth equals one, and has non-negative (n+1)th derivative on [0,B], the linear combination sum a_k k! Φ_Λ_n^{(n-k)}(x) lies above any polynomial R(x) = sum a_k x^k that is non-negative on the same interval. This supplies a systematic source of inequalities that involve exponential polynomials and immediately yields a construction for moment sequences. The construction takes a positive measure μ supported on an interval of length less than B and produces numbers s_k that must themselves be the moments of some other positive measure ν on that interval. A reader would care because the result ties the sign of a single higher derivative to a global positivity statement that preserves moment properties.

Core claim

The paper claims that if Φ_Λ_n is the unique real-valued solution of LΦ = 0 with L the product over j of (d/dx - λ_j), satisfying Φ^{(j)}(0) = 0 for j = 0 to n-1 and Φ^{(n)}(0) = 1, and if Φ_Λ_n^{(n+1)}(x) ≥ 0 on [0,B], then for every polynomial R(x) = sum_{k=0}^n a_k x^k that remains non-negative on [0,B] the inequality sum_{k=0}^n a_k k! Φ_Λ_n^{(n-k)}(x) ≥ R(x) holds throughout the interval. From the same inequality the authors obtain further bounds on exponential polynomials and prove that the numbers s_k formed by integrating k! Φ_Λ_n^{(n-k)}(x-a) against any non-negative measure μ whose support lies in an interval of length less than B constitute the moment sequence of some non-negative

What carries the argument

Φ_Λ_n, the unique solution to the nth-order constant-coefficient differential equation LΦ=0 with the prescribed initial conditions at zero; its non-negative (n+1)th derivative supplies the sign condition that turns the linear combination into a lower bound for non-negative polynomials.

If this is right

  • The inequality directly produces new pointwise bounds on exponential polynomials.
  • Any positive measure μ supported on an interval of length less than B generates, via the integrals s_k, a genuine moment sequence of another positive measure on the same interval.
  • The construction supplies a positivity-preserving map from measures to moment sequences of fixed length n.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same sign condition on the highest derivative might be used to bound other classes of functions, such as analytic functions whose Taylor remainders satisfy analogous positivity.
  • Specific choices of the λ_j could recover classical inequalities that involve pure exponentials or polynomials multiplied by exponentials.
  • The moment-sequence result suggests a possible route to constructing positive definite kernels on intervals by integrating the same Φ weights.
  • Testing the boundary case where Φ^{(n+1)} touches zero at isolated points would reveal how sharp the interval length B can be made.

Load-bearing premise

The (n+1)th derivative of Φ_Λ_n must remain non-negative throughout the interval [0,B].

What would settle it

A concrete counter-example consisting of explicit λ_j, an interval [0,B], and a non-negative polynomial R of degree at most n for which the computed linear combination dips below R at some interior point would falsify the inequality.

read the original abstract

Let $\Phi_{\Lambda_{n}}$ be the unique solution of the differential operator $L=\prod_{j=0}^{n}\left( \frac{d}{dx}-\lambda_{j}\right) $ such that $\Phi_{\Lambda_{n}}^{\left( j\right) }\left( 0\right) =0$ for $j=0,...,n-1,$ and $\Phi_{\Lambda_{n}}^{\left( n\right) }\left( 0\right) =1.$ Assume that $\Phi_{\Lambda_{n}}$ is real-valued and $\Phi_{\Lambda_{n} }^{\left( n+1\right) }\left( x\right) \geq0$ for all $x\in\left[ 0,B\right] .$ Then, if a polynomial $R\left( x\right) = {\displaystyle\sum_{k=0}^{n}} a_{k}x^{k}$ is non-negative on the interval $\left[ 0,B\right] ,$ the inequality \[ {\displaystyle\sum_{k=0}^{n}} a_{k}k!\Phi_{\Lambda_{n}}^{\left( n-k\right) }\left( x\right) \geq R\left( x\right) \] holds for $x\in\left[ 0,B\right] $. From this we derive several interesting inequalities for exponential polynomials. An important consequence is that for a non-negative measure $\mu$ over the interval $\left[ a,b\right] $ with $b-a<B$ the sequence defined by \[ s_{k}:=\int_{a}^{b}k!\Phi_{\Lambda_{n}}^{\left( n-k\right) }\left( x-a\right) d\mu\left( x\right) \] for $k=0,...,n$ is a moment sequence, i.e. there exists a non-negative measure $\nu$ with support in $\left[ a,b\right] $ such that $s_{k}=\int_{a} ^{b}\left( t-a\right) ^{k}d\nu\left( t\right) $ for $k=0,....,n.$

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The manuscript proves a conditional inequality: if Φ_Λ_n solves the linear ODE LΦ=0 with the given initial conditions at 0, is real-valued, and satisfies Φ_Λ_n^{(n+1)}(x) ≥ 0 on [0,B], then for any polynomial R(x)=∑_{k=0}^n a_k x^k that is nonnegative on [0,B], the weighted sum ∑ a_k k! Φ_Λ_n^{(n-k)}(x) ≥ R(x) holds on [0,B]. It derives corollaries for exponential polynomials and shows that the sequence s_k obtained by integrating k! Φ_Λ_n^{(n-k)}(x-a) against a nonnegative measure μ on [a,b] with b-a < B is a moment sequence (i.e., representable by some nonnegative ν supported on [a,b]).

Significance. If the central conditional result holds, the paper supplies a positivity-preserving comparison between nonnegative polynomials and exponential polynomials generated by constant-coefficient ODEs, together with an explicit construction of moment sequences on compact intervals. The moment-sequence consequence follows from the inequality by integration against μ and an appeal to Haviland’s theorem; this is a clean, falsifiable application that could be useful in the classical moment problem.

minor comments (3)
  1. [abstract, §3] The statement of the main theorem (abstract and §2) should explicitly record that the interval length condition b-a < B is required for the moment-sequence conclusion, as the inequality itself is stated only up to B.
  2. [§1] Notation for the differential operator L and the multi-index Λ_n is introduced without a dedicated preliminary subsection; a short §1.1 collecting the standing assumptions on the roots λ_j would improve readability.
  3. [§4] The proof of the moment-sequence claim invokes Haviland’s theorem but does not cite the precise reference or state the version used; adding the citation would make the argument self-contained.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive assessment of the manuscript, accurate summary of the main result, and recommendation for minor revision. No major comments were listed in the report.

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained

full rationale

The central inequality is stated conditionally on the explicit assumption that Φ_Λ_n is real-valued and Φ_Λ_n^{(n+1)}(x) ≥ 0 on [0,B], with Φ defined as the unique solution of the given linear ODE with the listed initial conditions at x=0. The moment-sequence consequence follows by integrating the inequality against a nonnegative measure μ (yielding nonnegativity of the linear functional on nonnegative polynomials of degree ≤ n) and invoking the external Haviland theorem; neither step reduces the claimed result to a fitted parameter, a self-citation, or a redefinition of the inputs. No load-bearing self-citations or ansatzes appear in the abstract or the described derivation chain.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The result rests on the standard existence-uniqueness theory for linear ODEs with constant coefficients and on the explicit positivity assumption on the (n+1)th derivative; no free parameters or new entities are introduced.

axioms (2)
  • standard math Φ_Λ_n is the unique solution of LΦ=0 with Φ^{(j)}(0)=0 for j=0..n-1 and Φ^{(n)}(0)=1
    This is the definition given in the abstract and is standard for the Cauchy function of a linear differential operator.
  • domain assumption Φ_Λ_n is real-valued and Φ_Λ_n^{(n+1)}(x) ≥ 0 on [0,B]
    This positivity hypothesis is stated explicitly as the condition under which the inequality holds.

pith-pipeline@v0.9.0 · 5927 in / 1497 out tokens · 33426 ms · 2026-05-23T20:36:45.243444+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [1]

    from Russian ed

    Akhiezer, N.I.: The problem of moments and some related questions in analysis, Oliver & Boyd, Edinburgh, 1965.(Transl. from Russian ed. Moscow 1961)

    work page 1965

  2. [2]

    M.; Kounchev, O.; Render, H.: Bernstein operators for expo- nential polynomials

    Aldaz, J. M.; Kounchev, O.; Render, H.: Bernstein operators for expo- nential polynomials. Constr. Approx. 29 (2009), no. 3, 345–367

    work page 2009

  3. [3]

    A.; Gay, R.: Complex analysis and special topics in har- monic analysis , Springer-Verlag, New York, 1995

    Berenstein C. A.; Gay, R.: Complex analysis and special topics in har- monic analysis , Springer-Verlag, New York, 1995

    work page 1995

  4. [4]

    Carnicer, J.M., Mainar, E., Pe˜ na, J.M.: Critical Length for Design Purposes and Extended Chebyshev Spaces, Constr. Approx. 20 (2004), 55–71. 12

    work page 2004

  5. [5]

    Carnicer, J.M., Mainar, E., Pe˜ na, J.M.: On the critical length of cy- cloidal spaces. Constr. Approx. 39 (2014), 573–583

    work page 2014

  6. [6]

    Calcolo 54(4) (2017), 1521–1531

    Carnicer, J.M., Mainar, E., Pe˜ na, J.M.: Critical lengths of cycloidal spaces are zeros of Bessel functions. Calcolo 54(4) (2017), 1521–1531

    work page 2017

  7. [7]

    Dyn, N.; Levin, D.; Luzzatto, A.: Exponentials reproducing subdivision scheme, Found. Comput. Math. 3 (2003) 187–206

    work page 2003

  8. [8]

    Dyn, N.; Kounchev, O.; Levin, D.; Render, H.: Regularity of gen- eralized Daubechies wavelets reproducing exponential pol ynomials with real-valued parameters, Appl. Comput. Harm. Anal. 37 (2014) 288–306

    work page 2014

  9. [9]

    1, Stanford Univ

    Karlin, S.: Total positivity, Vol. 1, Stanford Univ. Press, Standford 1968

    work page 1968

  10. [10]

    Applications to Numerical and Wavelet Analysis , Academic Press, London–San Diego, 2001

    Kounchev, O.I.: Multivariate Polysplines. Applications to Numerical and Wavelet Analysis , Academic Press, London–San Diego, 2001

    work page 2001

  11. [11]

    Ark Mat 48, 97–120 (2010)

    Kounchev, O.; Render, H.: A moment problem for pseudo-positive def- inite functionals. Ark Mat 48, 97–120 (2010)

    work page 2010

  12. [12]

    Kounchev, O.; Render, H.: Polyharmonic functions of infinite order on annular regions, Tˆ ohoku Math. J. 65 (2013), 199–229

    work page 2013

  13. [13]

    Kounchev, O.; Render, H.: A symmetry property for polyharmonic functions vanishing on equidistant hyperplanes, Math. Nachr. 290 (2017), 1087–1096

    work page 2017

  14. [14]

    Kounchev, O.; Render, H.: Interpolation of data functions on parallel hyperplanes, J. Approx. Theory 246 (2019), 43–61

    work page 2019

  15. [15]

    Kounchev, O.; Render, H.: Error estimates for interpolation with piece- wise exponential splines of order two and four , J. Comput. App. Math., Vol. 391, 1 August 2021, 113464

    work page 2021

  16. [16]

    Kounchev, O.; Render, H.; Tsachev, Ts.: On a class of L− splines of order 4: fast algorithms for interpolation and smoothing , BIT Numeri- cal Mathematics, volume 60, pages 879–899 (2020)

    work page 2020

  17. [17]

    McCartin, B.J.: Theory of exponential splines, J. Approx. Theory 66 (1991), 1–23. 13

    work page 1991

  18. [18]

    Micchelli, Ch.: Cardinal L− splines, In: Studies in Spline Functions and Approximation Theory, Eds. S. Karlin et al., Academic Pr ess, NY, 1976, pp. 203-250

    work page 1976

  19. [19]

    W.: Functional Data Analysis , Springer Verlag, Second Edition 2005

    Ramsay, J.O.; Silverman, B. W.: Functional Data Analysis , Springer Verlag, Second Edition 2005

    work page 2005

  20. [20]

    Schumaker, L.L.: Spline Functions: Basic Theory , Interscience, New York, 1981

    work page 1981

  21. [21]

    Unser, M.; Blu, T.: Cardinal Exponential Splines: Part I – Theory and Filtering Algorithms, IEEE Transactions on Signal Processing, 53 (2005), 1425–1438. 14

    work page 2005