Abstract
Scientists must model increasingly complex nonlinear systems in their attempt to solve real world problems. Especially when dealing with nonlinear systems, it is sometimes impossible to find a solution analytically. In such cases simulation may be the only way to realistically model the underlying mechanism of a problem and to find acceptable solutions. Simulation is not restricted to a certain field or subject but is part of the interdisciplinary science called operations research. Although simulation is already widespreadly used, in the field of statistics it is still uncommon, to used modeling techniques in order to find a solution. The simulation system SIMPLEX II is a universal simulator, which supports the user in model description, design and execution of experiments and data presentation and animation. The following paper will give an introduction into the simulation system and illustrate an example of how simulation can be used to improve the solution of statistical problems. Parameter estimation usually includes an average value for the parameter and the variance to determine the accuracy of the estimation. However, when dealing with complex nonlinear systems, only slight changes in the parameters can already cause significant differences in the response of the model. Sensitivity analysis can determine regions of instability in the parameter range space and therefore point to parameters, which have to be estimated with increased care. Using a stochastic simulation model we will demonstrate how to use simulation for a sensitivity analysis.
Schinagl, Josef Georg (1995). The simulation system SIMPLEX II. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1995 -THESIS -S356.