Pointwise convergence of multiple Fourier series: Sufficient conditions and an application to numerical integration (Q1206889): Difference between revisions

The authors are interested in obtaining sufficient conditions for the pointwise convergence of square partial sums of multiple Fourier series, in any number of dimensions in the presence of discontinuities. The result is motivated by a problem in numerical integration, which is called ``lattice rules'' for multiple integration and is related to the theory of equidistribution. A lattice rule for numerical integration of a periodic function is an equal-weight quadrature rule \({1 \over n} \sum^{n-1}_{j = 0} f(x^{(j)})\), where \(\{x^{(0)}, \dots,x^{(n-1)}\} = L \cap [0,1]^ S\) and \(L\) is a lattice in \(\mathbb{R}^ S\) containing \(\mathbb{Z}^ S\) as a subset. Based on previous papers [\textit{I. H. Sloan}, J. Comput. Appl. Math. 12/13, 131-143 (1985; Zbl 0597.65014), \textit{I. H. Sloan} and \textit{P. J. Kachoyan}, SIAM J. Numer. Anal. 24, 116-128 (1987; Zbl 0625.65020)] the authors indicate under which circumstances an integral over a unit cube can be compared with an equal-weight quadrature rule of above type by representing the error term essentially as series expansion in terms of the Fourier transform for the dual lattice (with the origin excluded).