Kevin D. Reilly
James J. Buckley
Xidong Zheng
Francisco Hernandez

Monte Carlo, Statistics and Fuzzy Uncertain Probabilities

Proc. Huntsville Simulation Conference, 2002



Abstract

The senior author's earliest exposure to 'Monte Carlo and statistics' was tutorial: how to develop and theorize about elementary statistics from a consummate computational approach. It included developing through computation, sampling distributions of the mean, (sample) variance, Student's t, Chi-Squared, and Pearson's correlation coefficient. In follow-up studies, computations were extended to newer frameworks, e.g., the "numerical recipes," with additional experiments with Fisher's F in general linear model studies. Since some work of this kind is designed to act as a "co- processor" to other computations, e.g., stochastic simulations, our most recent work involves simulation languages, the choice, in this paper being Wolverine, Inc.'s relatively new SLX (Simulation Language with eXtensibility), a system which includes a subset of GPSS couched in a general milieu of object-directed computing. A brief overview of relevant features is presented before we conclude with a departure relating to the potentially extensive opportunity presented by the newly developing fuzzy uncertain probability theory. Here, we plan to parallel conventional Monte Carlo approaches to the maximum extent possible. Software engineering and forms of "intelligent processing" (AI) features arose over the course of studies, e.g., involving logic programming, expert guidance of choices, e.g., statistical test choices and their configurations.

Key Words: Monte Carlo techniques, standard statistics, Fuzzy probability and statistics, simulation programming languages, combined (integrated) simulation styles