A Sharp Discrepancy Bound for Jittered Sampling
arXiv:2103.15712 · doi:10.1090/mcom/3727
Abstract
For , a jittered sampling point set having points in is constructed by partitioning the unit cube into axis-aligned cubes of equal size and then placing one point independently and uniformly at random in each cube. We show that there are constants and such that for all and all the expected non-normalized star discrepancy of a jittered sampling point set satisfies \[c \,dm^{\frac{d-1}{2}} \sqrt{1 + \log(\tfrac md)} \le {\mathbb E} D^*(P) \le C\, dm^{\frac{d-1}{2}} \sqrt{1 + \log(\tfrac md)}.\] This discrepancy is thus smaller by a factor of than the one of a uniformly distributed random point set of points. This result improves both the upper and the lower bound for the discrepancy of jittered sampling given by Pausinger and Steinerberger (Journal of Complexity (2016)). It also removes the asymptotic requirement that is sufficiently large compared to .
References in corpus (6)
- Quasi-Monte Carlo Feature Maps for Shift-Invariant Kernels
- A Lower Bound for the Discrepancy of a Random Point Set
- Discrepancy Bounds for a Class of Negatively Dependent Random Points Including Latin Hypercube Samples
- On Negatively Dependent Sampling Schemes, Variance Reduction, and Probabilistic Upper Discrepancy Bounds
- The Supremum Norm of the Discrepancy Function: Recent Results and Connections
- Probabilistic Lower Bounds for the Discrepancy of Latin Hypercube Samples