One of the first of these physical random-number generators, called Random.
One of the tricks in designing physical random-number generators such as Random.
Reeds had encountered a similar problem with the Marsaglia-Zaman random-number generators in a different type of computation.
In an additional twist on the curious behavior of random-number generators, Shu Tezuka of the IBM Tokyo Research Laboratory in Japan and Pierre L'Ecuyer of the University of Montreal in Quebec have now proved that the Marsaglia-Zaman random-number generators are "essentially equivalent" to linear congruential methods.
The uncertainty about how subtle, hidden patterns among digits spewed out by various random-number generators may influence simulation results presents researchers using so-called Monte Carlo methods with a serious dilemma, especially when the answer is not known.
That number, multiplied by the for "congruential" random-number generators.
Marsaglia and Zaman's new class of random-number generators hinges on the use of what are called Fibonacci sequences.
When applied to seeds, or starting numbers, about 1,800 bits long, such a procedure forms the basis of "lagged-Fibonacci" random-number generators.