Random Variate Generation from Generalized Inverse Gaussian Distribution

Abstract

We consider the problem of random variate generation from generalized inverse Gaussian (GIG) distribution. The CDF of this distribution does not admit an explicit form, so the standard approach to simulation based its inverse is not the right tool for this problem. Instead, we follow the rejection simulation method, based on the probability density function, which is given explicitly for this distribution. We study the efficiency of this method, and identify an optimal numerical procedure within this framework.