Geometric computation of Christoffel functions on planar convex domains

From MaRDI portal
Publication:2037071



Abstract: For arbitrary planar convex domain, we compute the behavior of Christoffel function up to a constant factor using comparison with other simple reference domains. The lower bound is obtained by constructing an appropriate ellipse contained in the domain, while for the upper bound an appropriate parallelepiped containing the domain is constructed. As an application we obtain a new proof of existence of optimal polynomial meshes in planar convex domains.


The Christoffel function is \[\lambda_n(x, D):=\left( \sum_{j=1}^N\,p_j(x)^2\right)^{-1},\] on a compact disk \(D\subset\mathbb{R}^d\) with non-empty interior. On \(D\) the Lebesgue measure is used and \(\{p_j\}_{j=1}^N\) is an orthonormal basis of the space \(\Pi_{n,d}\) of real algebraic polynomials of total degree at most \(n\) in \(d\) variables. The inner product is \(\langle f,g\rangle =\int_D\,fgdx,\ N=\hbox{dim}\, \Pi_{n,d}=\binom{n+d}{d})\). In this paper the author establishes a characterization of the Christoffel function for an arbitrary, planar convex domain (i.e. \(d=2\)), using a refinement of the methods used in his paper [J. Math. Anal. Appl. 455, No. 2, 1984--2000 (2017; Zbl 1379.41010)] and in [\textit{A. Prymak} and \textit{O. Usoltseva}, ``Pointwise behavior of Christoffel function on planar convex domains, in: \textit{M. Abell} (ed.) et al., Topics in classical and modern analysis. In memory of Yingkang Hu. Collected papers presented at the international conference in approximation theory, Savannah, GA, USA, May 8--11, 2017. 293--302 (2019), \url{https://www.researchgate.net/publication/320163447_Pointwise_behavior_of_Christoffel_function_on_planar_convex_domains}]. After a thorough Introduction, the main results are given in Theorem 2.1. For any planar convex body \(D\) and any interior point \(x\in D\) \[U(x,D)\leq cL(x,D),\] where \(c\) is an absolute constant. Theorem 3.1. Suppose \(D\) is a convex compact susbset satisfying \(B\subset D\subset 2B\). For any \(n\geq 1\) and arbitrary \(x\in D\) define \(\tau_n(x)=x\) if \(x\in (1-2^{-4}n^{-2})D\), and \(\tau_n(x=tx)\) where \(t>0\) is the largest integer satisfying \(tx \in (1-2^{-4}n^{-2})D\). Then \[\lambda_n(x,D)\sim n^{-2} L(\tau_n(x,D)\sim n^{-2}U(\tau_nx,D).\] The notation used is: \[L(x,D)=\sup\, \{(1-\|\mathcal{L}^{-1}x\|)^{1/2}|\hbox{det}\ \mathcal{L}|:\,\mathcal{L}\in\mathbb{A}, x\in\mathcal{L}B\subset D\}, \] \[U(x,D)=\inf\,\{((\mathcal{U}^{-1}x)_1)^{1/2}((\mathcal{U}^{-1}x)_2)^{1/2}|\hbox{det}\ \mathcal{U}|:\,\mathcal{U}\in\mathbb{A},x\in\mathcal{U}(\frac{1}{2}S),D\subset\mathcal{U} S\},\] with \(\mathbb{A}\) the collection of all non-degenerate affine transforms of \(\mathbb{R}^2\), \(B\) the unit ball and \(S\) the unit square in \(\mathbb{R}^2\).











This page was built for publication: Geometric computation of Christoffel functions on planar convex domains

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2037071)