Local \(C^2\)-smooth spline quasi-interpolation methods (Q6656527)

Quasi-interpolation and splines are at the core of modern numerical analysis. The former is an absolutely first-class approach to approximating and smoothing multivariate data of any form and efficient algorithms are available for carrying out these important tasks. The latter provide piecewise polynomials spaces of approximating functions that are relatively easy to handle and have been successfully studied and used in applications which range from engineering to computer-aided geometric design to medical applications.\N\NQuasi-interpolation is available in any dimension, but the authors restrict in this particular article their work to the univariate case.\N\NThis excellent paper addresses the very interesting task of finding twice-differentiable splines that can be used in quasi-interpolation reproducing certain polynomials. It is well-known that the best concepts of quasi-interpolation are developed with the constraint of reproducing polynomials (``polynomial precision''). This avails the analyst of proving concergence results based on local Taylor expansions. In order to get unique solutions, some extra degrees of freedom need to be removed in the authors' construction and this is carried out by new interpolation conditions, locality constraints and/or additional polynomial precision.\N\NThe authors illustrate the remarkable efficiency of the new quasi-interpolation designs by numerical examples. Also, the coefficients of the quasi-interpolants (or, rather, their quasi-Lagrange functions) are explicitly computed and provided to the interested reader.