bythenumbers Generating Gaussian Random Variates

In two recent columns, we have looked at approximations to error functions [1] and to inverse error functions. The former is closely related to the cumulative distri bution function (cdf) of a Gaussian random variable. The latter is closely related to the Gaussian inverse cumulative distribution function. In this installment, we discuss various approaches to generating pseudorandom variates that follow the Gaussian distribution. A variate is simply the numerical outcome of a random variable from a single experiment (or draw). The need to generate Gaussian random variates arises often in science and engineering applications, particularly in the analysis of various systems via Monte Carlo simulation. © 2009 IEEE.