Random Variate Generation for Bayesian Nonparametric Reliability Analysis
Patrick J. Munson · Defense Technical Information Center (DTIC) · 2005
Simulation modeling requires accurate input analysis to ensure validity of the study. Hence, the mantra "garbage in = garbage out." Much of the research and simulation code that has been written to date has been focused on traditional parametric methods. Here we investigate Bayesian nonparametric methods for input modeling and reliability analysis. Bayesian nonparametric methods have been shown in many cases to produce better predictive models. Also, for use in a Bayesian setting, we have written C++ classes for random variate generation. These contain functions for standard and truncated distributions as well as functions for statistical data handling. Although we have written the code for Bayesian algorithms, the functions can be used anywhere a good source of random variates is needed. Included is a detailed description of class implementation and usage along with complete source code.