Efficient Algorithms for Joint Density Approximations of Discrete Independent Random Variables
Janche Sang, Vernon Rago, John Spragius · Purdue e-Pubs (Purdue University System) · 1992
We present an efficient algorithm to compute the joint density function of a set of discrete independent random variables.The computation is accurate upto the coverage specified.The teclmique is based on a simple mathematical structure and associated property knolVIl as the diamond property.The algorithm generates the product density, or slates of the joint random variable, in order of decreasing probability, stopping when a prescribed degree of accuracy has been obtained.Since the algorithm can be seen to exhibit a high degree of inherent parallelism, we propose easily implementable versions of the original algorithm for parallel machines.