Quasi-Monte Carlo methods for the numerical integration of multivariate Walsh series (Q1921099)

The authors consider quasi-Monte Carlo methods for the numerical integration of multivariate Walsh series by approximating the integrals via net-sequences. Best possible bounds are shown for the integration error \(R_N (f)\) when the integration points form a \((t, m, s)\)-net in \([0, 1]^s\) in base \(b\). More precisely it is proved that \[ R_N (f)\;\ll\;b^{t(\alpha- (1/2))} {{(\log N)^{s-1}} \over {N^{\alpha- (1/2)}}}, \] provided that \(f\) belongs to an \(E_s^\alpha\)-class with respect to the Walsh series in base \(b\). This improves an earlier paper of the authors and \textit{C. Traunfellner} [Math. Comput. 63, 277-291 (1994; Zbl 0806.65013)].