Efficient spectral tests for multiple recursive generators

Lih Yuan Deng, Bryan R. Winter, Jyh Jen Horng Shiau, Henry Horng-Shing Lu, Nirman Kumar, Ching Chi Yang*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Abstract.: Large-order maximum-period Multiple Recursive Generators (MRGs) have become popular in the area of computer simulation. They have the nice properties of high-dimensional equi-distribution, generating efficiency, long period, and portability. The spectral test is a commonly used criterion for ranking pseudo-random number generators. Procedures for computing the spectral test values of MRGs are available in the literature but may not be efficient when the order of the MRG is large. In this article, we propose a novel method for the spectral test computation of MRGs that is simple, intuitive, and particularly efficient for MRGs with few non zero terms. With the proposed method, we are able to provide a list of ready-to-use “better” generators with respect to the spectral test performance among the DX generators of order k for various values of k.

Original languageEnglish
JournalCommunications in Statistics - Theory and Methods
DOIs
StateAccepted/In press - 2024

Keywords

  • Basis reduction
  • DX-k generators
  • LLL algorithm
  • Lattice structure
  • Normal vector

Fingerprint

Dive into the research topics of 'Efficient spectral tests for multiple recursive generators'. Together they form a unique fingerprint.

Cite this