Discrepancy of weighted matrix nets
For any invertible real \(s\times s\) matrix \(A\), consider the lattice \(\Lambda= \{Am:m\in{\mathcal H}^S\}\) and write \(X= \Lambda\cap U^s\) with \(U^S= [0,1]^S\). A matrix net generated by \(A\) is a pair \((X,\rho)\), where \(\rho: X\to\mathbb{R}\) is a weight function on the points of \(X\). The authors consider translation invariant classes \(\Phi\) of Riemann integrable, 1-periodic functions \(f:\mathbb{R}^S\to \mathbb{R}\) and the error term for numerical integration \[ R(f;X,\rho)= \sum_{x\in X} \rho(x)f(x)- \int_{U^S} f(t)dt. \] The authors show that if the inverse of \(A\) has integer entries, then one can always choose equal weights to minimize \(R(\Phi; X,\rho)= \sup_{f\in\Phi}| R(f; X,\rho)|\) (for every class \(\Phi\)). Furthermore, in various ``natural cases (for \(X\)) the optimal equal weights have to be chosen as \(\rho(x)={1\over N}\) with \(N=\text{card }N\).
- Weighted discrepancy of Faure-Niederreiter nets for a certain sequence of weights
- Disjoint weighing matrices
- scientific article; zbMATH DE number 3966090
- Netted matrices
- Discrepancy of hyperplane nets and cyclic nets
- scientific article; zbMATH DE number 515358
- scientific article; zbMATH DE number 4137791
- Weighted discrepancy and high-dimensional numerical integration
- scientific article; zbMATH DE number 4151340
- scientific article; zbMATH DE number 1166315
This page was built for publication: Discrepancy of weighted matrix nets
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4211780)
