Peano kernel error analysis for quadratic nodal spline interpolation

From MaRDI portal





For \(n \geq 2\), let \[ a=x_0<x_1 < \dots < x_{2n}=b, \qquad \xi_i=x_{2i}, \qquad i=0, \dots, n. \] In a series of papers [see e.g. \textit{J. M. de Villiers} and \textit{C. H. Rohwer}, J. Comput. Appl. Math. 18, 107-119 (1987; Zbl 0637.41009)], an arbitrary order \(k\) nodal spline interpolation operator \(V_n\) has been introduced. For each \(f \in C[a,b]\), \(V_n(f)\) is a spline of degree \( \leq k\) and nodes in \(x_i\), which interpolates \(f\) at \(\xi_i\) and has some additional properties of locality and optimal order polynomial reproduction. In this paper the authors consider specifically the quadratic case, i.e.\ \(k=2\). They use the Peano kernel representation in order to obtain Jackson-type estimates for \(\|f-V_n(f)\|_\infty\). The constants obtained this way grow linearly with the parameter \[ R=\max \bigg\{{ \xi_{i+1}-\xi_i \over \xi_{j+1}-\xi_j}: 0 \leq i, j \leq n-1, \;|i-j|=1 \bigg\}, \] and in some important cases are significantly smaller than those obtained by \textit{J. M. de Villiers} [J. Approximation Theory 74, No.~3, 266-279 (1993; Zbl 0790.41008)] by different techniques.











This page was built for publication: Peano kernel error analysis for quadratic nodal spline interpolation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1305504)