We consider the problem of approximating the uniform distribution in the d-dimensional unit cube by a discrete distribution consisting of finitely many points. As measure of quality of our approximation we use the so-called star discrepancy. We discuss an algorithm that constructs small discrete distributions exhibiting a low star discrepancy. The algorithm is based on recent results on randomized roundings respecting hard constraints.
Benjamin Doerr, 03/27/2007 19:56
Benjamin Doerr, 03/26/2007 00:53
Benjamin Doerr, 03/21/2007 19:29
Benjamin Doerr, 03/11/2007 22:19 -- Created document.