Computational Support for the Study of Lifetime Distribution Characteristics

摘要

The report develops and makes available programs that treat the support functions of a set of survivor distributions: Weibull, Gamma, and Lognormal. The issues of model characterization functions, maximum likelihood estimation, bias reduction, and censored samples are treated generally. The general material is made explicit for the distributions named. It features open code, allowing the user to pursue plans of his own. The paper also contains some items of more general interest. First, a technique is developed that offers substantial reduction in the dependence of the initialization values for the success of the Newton-Raphson iteration technique. Second, high-precision, numerical analysis techniques are developed for the parallel computation of several derivatives of the Incomplete Gamma function.