NHSE ReviewTM 1996 Volume Second Issue

Random Number Generators for Parallel Computers

| <- HREF="node1.html" Prev | Index | Next -> |
NHSE ReviewTM: Comments · Archive · Search


Summary 

Random number generators use iterative deterministic algorithms for producing a sequence of pseudo-random numbers that approximate a truly random sequence. Ideally the sequence should be uniformly distributed, uncorrelated, reproduceable, portable, easily changed by adjusting an initial seed value, easily split into many independent subsequences, have a large period of repetition, pass all empirical tests for randomness, and be generated rapidly using limited computer memory.

Parallel random number generators should in addition have no correlations between the sequences on different processors, produce the same sequence for different numbers of processors, and not require any data communication between processors.

Developing random number generators that satisfy all of these requirements is a very difficult problem, particularly for parallel computers.

The main algorithms used for sequential random number generators are the following:

The main techniques used for parallelizing random number generators involve distributing the sequences of random numbers produced by a sequential generator among the processors in the following different ways:

Random number generators, particularly for parallel computers, should not be trusted. It is strongly recommended that all simulations be done with two or more different generators, and the results compared to check whether the random number generator is introducing a bias.

On a sequential computer, good generators to use are:

All of the parallel random number generators covered in this review have some limitations or possible problems. Recommended generators to use on a parallel computer are:

Some software implementing these recommended generators is available from the National HPCC Software Exchange (NHSE).

More work needs to be done on developing better random number generators for parallel computers, and subjecting these generators to more thorough empirical testing.

Copyright © 1996


| <- HREF="node1.html" Prev | Index | Next -> |
NHSE ReviewTM: Comments · Archive · Search


Paul Coddington, paulc@npac.syr.edu