scientific article; zbMATH DE number 440739

From MaRDI portal
Publication:3140904





The author proves the following refinement of the Erdős-Koksma inequality in uniform distribution: Let \(\Delta^{(\infty)}_ M\) be the usual discrepancy of a point set \(x_ 1,\dots,x_ M\) in \([0,1]\). Then for every integer \(N\) we have: \[ \displaylines{\biggl| \Delta^{(\infty)}_ M - \sup_ Y\bigl|\sum^ M_{m=1} \sum^ N_{\textstyle {{n=-N}\atop n\neq 0}} (2\pi in)^{- 1}\cdot\widehat{j}\bigl({n\over N+1}\bigr)\cdot e^{2\pi in(y-x_ m)}\bigr|\biggr|\leq\hfill\cr \hfill {} \sup_ Y(2N + 2)^{- 1}\cdot \sum^ M_{m=1} \sum^ \infty_{n=-\infty} (N+1)\cdot{\sin(\pi(N+1)(Y-X_ m + n))\over \pi(N+1)(Y-x_{m+n})}} \] where \(\widehat{j}(t): = \pi t(1-| t|)\cot \pi t + | t|\). So this result, in contrast to the classical Erdős-Turán inequality also provides lower bounds for the discrepancy. Similar estimates are given for general \(L^ p\) discrepancies. For the entire collection see [Zbl 0778.00034].











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3140904)