Computing Discrepancies Related to Spaces of Smooth Periodic Functions
A notion of discrepancy is introduced, which represents the integration error on spaces of r-smooth periodic functions. It generalizes the diaphony and constitutes a periodic counterpart to the classical L2-discrepancy as well as r-smooth versions of it introduced recently by Paskov [Pas93]. Based on previous work [FH96], we develop an efficient algorithm for computing periodic discrepancies for quadrature formulas possessing certain tensor product structures, in particular, for Smolyak quadrature rules (also called sparse grid methods). Furthermore, fast algorithms of computing periodic discrepancies for lattice rules can easily be derived from well—known properties of lattices. On this basis we carry out numerical comparisons of discrepancies between Smolyak and lattice rules.
KeywordsQuadrature Formula Quadrature Rule Reproduce Kernel Hilbert Space Sparse Grid Integration Error
Unable to display preview. Download preview PDF.
- [GR65]I.S. Gradshteyn and I.M. Ryshik. Tables of integrals, series and products. Academic Press, NY, London, 1965.Google Scholar
- [Hei95]S. Heinrich. Efficient algorithms for computing the L2 discrep-ancy. TR 267–95 Universität Kaiserslautern, 1995. (to appear in Math. Comp.).Google Scholar
- [NR96]E. Novak and K. Ritter. High dimensional integration of smooth functions over cubes. to appear in Num. Math., 1996.Google Scholar
- [War72]T.T. Warnock. Computational investigations of low discrepancy point sets. In S.K. Zaremba, editor, Applications of Number Theory to Numerical Analysis. Academic Press, New York, 1972.Google Scholar