Parallel and Distributed Quasirandom Numbers

Name: Prof. Michael Mascagni Address: Department of Computer Science and School of Computational Science Florida State University Tallahassee, FL 32306-4530 USA Offices: 498 Dirac Science Library/172 Love Building Phone: +1.850.644.3290 FAX: +1.850.644.0098 e-mail: mascagni@fsu.edu

Title: Parallel and Distributed Quasirandom Numbers

Abstract:

Software support for quasirandom number generation on regular, serial, computers is based on downloading special software from a collection of rather disjointed web sites. For parallel and distributed systems, the situation is much worse, as there does not exist any software at all for quasirandom number generation. The SPRNG library has become a de facto standard for pseudorandom number generation on parallel, and soon distributed, computing platforms. We describe a new project to provide quasirandom numbers for parallel and distributed systems based on
(1) blocking existing quasirandom number generations methods,
(2) parameterization of existing quasirandom number generations methods, and
(3) using full-period pseudorandom number generation methods to provide low discrepancy point sets.