Dynamic Creation of Pseudorandom Number Generators
We propose a new scheme Dynamic Creation (DC) of pseudorandom number generators (PRNG) for large scale Monte Carlo simulations in parallel or distributed systems. DC receives user’s specification such as word size, period, size of working area, together with a process ID (or a set of IDs). Then DC creates a PRNG (or a set of PRNGs, respectively) satisfying the specification, so that ID number is encoded in the characteristic polynomial of PRNG. Thus, different IDs assure highly independent PRNGs. Each PRNG is a small Mersenne Twister, which we proposed previously.
Unable to display preview. Download preview PDF.
- 2.Knuth, D.E.: Seminumerical Algorithms. 3rd ed., Vol 2, The Art of Computer Programming, Addison Wesley, Reading, MA., (1997)Google Scholar
- 6.Tezuka, S.: Lattice structure of pseudorandom sequence from shift register generators, Proceedings of the 1990 Winter Simulation Conference, IEEE (1990) 266 – 269Google Scholar