Bernstein Inequality on Parabolic Domains

Yuan Xu · 2026

Read Bernstein Inequality on Parabolic Domains by Yuan Xu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-04268.txt.

Introduction

Xu proves several sharp Bernstein inequalities over parabolic domains. The domains cover bounded and unbounded paraboloids plus parabolic surfaces. His main tool is a second-order differential operator obeyed by a chosen set of orthogonal polynomials in the weighted space. Such inequalities bound how fast a polynomial can change, and sharp means the constants cannot be beaten. That bound is the heart of the paper.

Catalog record

CREATORYuan Xu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE61834
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.04268

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Bernstein Inequality on Parabolic Domains Thanks: The author was partially supported by Simons Foundation Grant #849676 Yuan Xu Abstract . Several families of sharp Bernstein inequalities are established on the weighted L 2 L^{2} space over parabolic domains, which include bounded or unbounded rotational paraboloids and parabolic surfaces. The main tool is a second-order differential operator satisfied by a specific basis of orthogonal polynomials in weighted L 2 L^{2} space. 2010 Mathematics Subject Classification 33C45, 42C05, 42C10 1. Introduction Bernstein inequalities on several classical domains have been revisited in recent studies, due to a new revelation of the existence of stronger inequalities than those in the literature. The phenomenon was first observed in [ 17 ] for the triangle, as a byproduct of a study on the rotational cone in {\mathbb{R}}^{d+1} , which motivated a follow-up study [ 7 ] that established sharp Bernstein inequalities on the simplex in {\mathbb{R}}^{d} that are stronger than long accepted inequalities (cf. [ 5 ] ) for this classical domain. Even more, stronger inequalities turn out to exist [ 2 ] even for the unit ball in {\mathbb{R}}^{d} , one of the most studied domains and a prototype for other rotational domains. There are two types of results in these studies, both based on properties of orthogonal polynomials defined via a class of weight functions on the given domain. The first type consists of Bernstein inequalities in weighted L^{p} norm that hold for all doubling weights, and their proofs are based on highly localized kernels for a particular weight function. The existence of latter kernels is established via the addition formula , which is the closed-form formula for the reproducing kernel of the orthogonal projection operator, mimicking the addition formula for spherical harmonics, that exists only on a few regular domains. The second type consists of sharp Bernstein inequalities in the L^{2} norm, which is sharp in the sense that an inequality becomes an equality for certain extremal polynomials. Most of the latter results can be established via the spectral operator , which is a second-order linear derivation operator that has orthogonal polynomials as eigenfunctions, with the eigenvalues depending only on the total degree of orthogonal polynomials. For the simplex and the unit ball, new sharp Bernstein inequalities arise from new decompositions of the classical spectral operators on these domains. Both the addition formula and the spectral operator provide powerful tools for analysis, but they exist only on a few regular domains. For d=2 , the existence of the spectral operator was characterized in [ 9 ] . For d>3 , no classification is known. While the spectral operators for the unit ball and the simplex are classical, the one for the rotational cone was discovered only recently [ 14 ] . The purpose of the present paper is to establish sharp Bernstein inequalities on rotational parabolic domains, which consist of solid paraboloids in {\mathbb{R}}^{d+1} , defined by {\mathbb{U}}^{d+1}=\left\{(x,t):\|x\|\leq\sqrt{t},\quad 0\leq t\leq b,\quad x\in{\mathbb{R}}^{d}\right\}, and parabolic surfaces {\mathbb{U}}_{0}^{d+1} of these paraboloids, where b could be infinity. Orthogonal polynomials for {\mathbb{U}}^{2} , bounded by a parabola and a line when b=1 , are semi-classical and studied already in [ 8 ] . The orthogonal structure for d\geq 2 has been explored more recently [ 16 ] . While an explicit basis of orthogonal polynomials can be given explicitly for two classes of weight functions, called the Jacobi and the Laguerre polynomials, they do not satisfy a full addition formula, nor do they possess a spectral operator. The lack of tools explains why analysis in this domain is far less developed; see [ 16 ] for what is known. Although there is no spectral operator, the specific orthogonal basis, the Jacobi and the Laguerre polynomials on parabolic domains, satisfy a second-order linear differential operator, but with the eigenvalues depending on one more index, rather than only on the degree of the polynomials. It turns out that this operator can be used to establish sharp Bernstein inequalities in the L^{2} norm, following more or less the same paradigm based on the spectral operator. This leads to the main results of this work. Our results are restricted only to inequalities in the L^{2} norm, and they are the first ones established on the parabolic domains and new even in the parabolic domain in the plane. At the moment, we are not aware of an effective tool for establishing inequalities in the L^{p} norm, for p\neq 2 , on parabolic domains. These L^{2} inequalities, especially the additional weight functions that accompany the derivatives, provide a possible guidance for the formulation of inequalities in the L^{p} norm. The paper is organized as follows. In the next section, we discuss how to deduce a sharp Bernstein inequality in the L^{2} norm from the spectral operator, and illustrate the approach by reviewing the results on the unit ball, which also serve as a preliminary for the latter sections. The sharp Bernstein inequalities on the paraboloids are established in Section 3, and those on the parabolic surfaces are established in Section 4. 2. Preliminary: sharp L^{2} Bernstein inequalities Let \Omega be a domain in {\mathbb{R}}^{d} and let W be a non-negative weight function defined on \Omega so that {\langle}f,g{\rangle}_{\Omega,W}=\int_{\Omega}f(x)g(x)W(x)\mathrm{d}x is a well-defined inner product on L^{2}(\Omega,W) . Let \Pi_{n}(\Omega) be the space of polynomials, restricted on \Omega , of degree at most n in d variables. For each n\in{\mathbb{N}}_{0} , let {\mathcal{V}}_{n}(\Omega,W) denote the subspace of \Pi_{n}(\Omega) that consists of orthogonal polynomials of degree n with respect to the inner product {\langle}\cdot,\cdot{\rangle}_{\Omega,W} . Let \operatorname{proj}_{n}:L^{2}(\Omega,W)\mapsto{\mathcal{V}}_{n}(\Omega,W) be the orthogonal projection operator. If \{P_{j}^{n}:1\leq j\leq d(n)\} is an orthogonal basis of {\mathcal{V}}_{n}(\Omega,W) , where d(n):=\dim{\mathcal{V}}_{n}(\Omega,W) , then the Fourier orthogonal expansion of f\in L^{2}(\Omega,W) is given by f=\sum_{n=0}^{\infty}\operatorname{proj}_{n}f,\quad\quad\hbox{where}\quad\operatorname{proj}_{n}f=\sum_{j=1}^{d(n)}\frac{{\langle}f,P_{j}^{n}{\rangle}_{\Omega,W}}{{\langle}P_{j}^{n},P_{j}^{n}{\rangle}_{\Omega,W}}P_{j}^{n}. A second-order linear derivation operator {\mathfrak{D}} is called a spectral operator of L^{2}(\Omega,W) if there exist nonnegative numbers {\lambda}(n) such that (2.1) {\mathfrak{D}}P=-{\lambda}(n)P,\qquad\forall P\in{\mathcal{V}}_{n}(\Omega,W),\qquad n=0,1,2,\ldots. That is, {\mathfrak{D}} has eigenvalue {\lambda}(n) with eigensapce {\mathcal{V}}_{n}(\Omega,W) for all n\in{\mathbb{N}}_{0} . The spectral operator exists only for rather special domains \Omega and/or the weight W . When it exists, it is self-adjoint in L^{2}(\Omega,W) by orthogonality and ( 2.1 ). Furthermore, the following lemma holds. Lemma 2.1 . Assume the spectral operator {\mathfrak{D}} exists. For f\in L^{2}(\Omega,W) , -\int_{\Omega}{\mathfrak{D}}f(x)\cdot f(x)W(x)\mathrm{d}x=\sum_{m=0}^{\infty}\lambda(m)\int_{\Omega}\left|\operatorname{proj}_{m}f(x)\right|^{2}W(x)\mathrm{d}x. In particular, if f\in\Pi_{n}(\Omega) , then \left|\int_{\Omega}{\mathfrak{D}}f(x)\cdot f(x)W(x)\mathrm{d}x\right|\leq\max_{0\leq m\leq n}\lambda(m)\int_{\Omega}|f(x)|^{2}W(x)\mathrm{d}x. The identity in the lemma follows from orthogonality since \operatorname{proj}_{m}f\in{\mathcal{V}}_{m}(\Omega,W) so that {\mathfrak{D}}\operatorname{proj}_{m}f=-\lambda(m)\operatorname{proj}_{m}f , whereas the inequality follows from the Parseval identity of the Fourier orthogonal series in L^{2}(\Omega,W) . The inequality, simple as it looks, is one of the main ingredients for deriving the sharp Bernstein inequality in L^{2} -norm. The other ingredient is the explicit self-adjoint form of the operator {\mathfrak{D}} . We illustrate how the latter works by working with an example: the unit ball {\mathbb{B}}^{d} of {\mathbb{R}}^{d} equipped with the classical weight function W_{\mathbb{B}}^{\mu}(x)=(1-\|x\|^{2})^{\mu-\frac{1}{2}},\qquad\mu>-\tfrac{1}{2}. Several orthogonal bases for {\mathcal{V}}_{n}({\mathbb{B}}^{d},W_{\mathbb{B}}^{\mu}) can be given explicitly. One of them is given in terms of the product of the Gegenbauer polynomials by parametrizing {\mathbb{B}}^{d} in the Cartesian coordinates. For x=(x_{1},\ldots,x_{d})\in\mathbb{R}^{d} , define by {\mathbf{x}}_{j} a truncation of x , namely \mathbf{x}_{0}=0 and \mathbf{x}_{j}=(x_{1},\ldots,x_{j}) , 1\leq j\leq d . Associated with {\mathbf{k}}=(k_{1},\ldots,k_{d}) , define {\mathbf{k}}^{j}:=(k_{j},\ldots,k_{d}) , 1\leq j\leq d , and {\mathbf{k}}^{d+1}:=0

Read the complete text