A Family of Hybrid Random Number Generators with Adjustable Quality and Speed
This addresses the need for efficient, high-quality random number generation in simulations, though it is incremental as it combines existing approaches.
The paper tackles the tradeoff between speed and quality in random number generation for large-scale stochastic simulations by introducing a family of hybrid generators, achieving high-quality output at a cost comparable to fast conventional generators as demonstrated by standard tests.
Conventional random number generators provide the speed but not necessarily the high quality output streams needed for large-scale stochastic simulations. Cryptographically-based generators, on the other hand, provide superior quality output but are often deemed too slow to be practical for use in large simulations. We combine these two approaches to construct a family of hybrid generators that permit users to choose the desired tradeoff between quality and speed for a given application. We demonstrate the effectiveness, performance, and practicality of this approach using a standard battery of tests, which show that high quality streams of random numbers can be obtained at a cost comparable to that of fast conventional generators.