ABSTRACT
Modelers wish to specify their models in a symbolic high-level language, while analytic Markov process solution techniques require a low-level, numerical representation. The translation between these description levels is a major problem. We describe a general, but surprisingly powerful methodology for specifying system-level models and generating the associated transition rate matrix. The basic approach will be shown to have significant advantages in that it provides the basis for modular, extensible modeling tools. With this methodology, modeling tools tailored to particular application domains can be quickly and easily constructed. An implementation in Prolog of a system based on this methodology is described and some example applications are given.
