Weighted Fejér constants and Fekete sets (Q485548)

From MaRDI portal





scientific article
Language Label Description Also known as
default for all languages
No label defined
    English
    Weighted Fejér constants and Fekete sets
    scientific article

      Statements

      Weighted Fejér constants and Fekete sets (English)
      0 references
      0 references
      9 January 2015
      0 references
      The weighted Lagrange interpolation polynomials on a system of nodes \(X=\{x_{k,n}:k=1,2,\dots ,n; n\in\mathbb{N}\}\) is \(L_{k,w,X}(x) =L_{k,w}(x)=w(x) \cdot\frac{\ell_{k}^{2}(x)}{w(x_{k})}\), where \(\ell_{k}(x)\), \(k=1,2,\dots ,n\), are the fundamental polynomials of Lagrange interpolation, and \(w(x)=e^{-Q(x)}\) is an approximating weight. If \(\| L_{k,w}\| _{\infty}=1\), then \((L_{k,w})'(x_{k})=0\) and the following constants \(C_{k},_{w}=\frac{\omega'' (x_{k})}{\omega'(x_{k})}+\frac{w'(x_{k}) }{w( x_{k}) }\) (Fejér constants) are null. A weighted Fekete set associated with a weight \(w\) on a set \(\Sigma \subset\mathbb{C}\) is the set \(\mathcal{F}_{n,w}=\{x_{1},\dots ,x_{n} \}\subset\Sigma\) such that the supremum \[ d_{n,w}=\sup_{z_{1},\dots,z_{n}\in\Sigma}d_{n,w}(z_{1},z_{2,}\dots, z_{n})=\sup_{z_{1},\dots ,z_{n}\in\Sigma}(\Pi| z_{i}-z_{j}| w( z_{i})w(z_{j}))^{\frac{2}{n(n-1)}} \] is attained at the set \(\mathcal{F}_{n,w}\). For the special weights \(w_{n}\) (defined by using \(w\) and the \(n\)th orthonormal polynomial \(p_{n,w}\) with respect to \(w\)), the Fekete set is formed by the zeros of \(p_{n,w}\). Let \(Y_{n,w}^{-}(x)=w(x)\sum_{k=1}^{n}\frac{\ell_{k}^{2} (x)}{w(x_{k})}= w(x)H_{n}\left( \frac{1}{w},(\frac{1}{w})';x\right)\) be the Hermite interpolation polynomial of \(\frac{1}{w}\) on the nodes \(\{x_{k}\}_{k=1}^{n}\). The main results of the paper may be summarized as it follows: If \(w\) is an admissible, approximating weight (cf. Def. 1) on \((a,b)\) and \(\left(\frac{1}{w}\right)^{(2n)}\geq0\) on \((a,b),\) then the following assertions are equivalent: (A) \(C_{k,w}=0,\) \(k=1,2,\dots ,n;\) (B) \(\mathcal{F}_{n,w^{\frac{1}{2(n-1)}}}=\{x_{1},x_{2},\dots ,x_{n}\}\); (C) \(Y_{n,w}(x)\) is \(w\)-stable and most economical (cf. Def. 6), i.e, it has minimal norm and minimum degree; (E) \(\{x_{1},x_{2},\dots ,x_{n}\}\) is \(1(w)\)-normal. Similar results hold for the special weight \(w_{n}\) satisfying the condition \(\left(\frac{1}{w_{n}}\right)^{(2n)}\geq0\) on \((a,b)\). Finally, the convergence of a weighted Grünwald interpolation process is proved.
      0 references
      interpolation
      0 references
      Hermite-Fejér
      0 references
      stable and most economical
      0 references
      Fekete sets
      0 references
      Grünwald operator
      0 references

      Identifiers