Random Number Generation (2006)

by Norman Matloff.

The basic building block for generating random numbers from various distributions is a generator of uniform random numbers on the interval (0,1), i.e. from the distribution U(0,1). How is this accomplished? Due to the finite precision of computers, we cannot generate a true continuous random variate. However, we will generate a discrete random variate which behaves very close to U(0,1). There is a very rich literature on the generation of random integers, commonly called pseudorandom numbers because they are actually deterministic. Pseudorandom numbers can be divided by their upper bound to generate U(0,1) variates. [via]

http://heather.cs.ucdavis.edu/~matloff/156/PL...