Quadrature formula for sampled functions
Résumé
This paper deals with efficient quadrature formulas
involving functions that are observed only at fixed sampling points. The approach that we develop is derived from efficient continuous
quadrature formulas, such as Gauss-Legendre or Clenshaw-Curtis quadrature. We select nodes at sampling positions that are as close as possible to those of the associated classical quadrature and we update quadrature weights accordingly. We supply the theoretical quadrature error formula for this new approach. We show on examples the
potential gain of this approach.