Beta Variate Generation via Exponential Majorizing Functions

Abstract

Two acceptance/rejection algorithms for generating random variates from the beta distribution are developed. The algorithms use piece-wise linear and exponential majorizing functions coupled with a piece-wise linear minorizing function. The algorithms are exact to within the accuracy of the computer and are valid for all parameter values greater than one Marginal execution times are relatively insensitive to parameter values and are faster than any previously published algorithms.