Abstract
In this paper we discuss how to automatically generate system dynamics models
using a kind of genetic algorithm known as a genetic program. This allows both the
structure and the parameters of the system dynamics models under study to be
evolved. This paper builds on previous work that introduced the use of genetic
programs to automatically generate system dynamics models. The paper’s
contribution is that it discusses how to automatically generate anticipatory system
dynamics in weakly constrained, data-sparse domains. The paper also describes
how this technique might be applied to an example domain, namely that of
transnational organized crime. This paper reports the status of work in progress. At
the time of submission, the designs described in this paper were partially, but not
fully, implemented