There is an increasing need in the scientific community for the comprehensive simulation of complex, dynamic, physical systems. Often such simulations are built through model coupling, that is, the merging of existing models so that their concurrent simulations affect each other. Model coupling is, however, a nontrivial task that is not adequately supported by existing frameworks which often require direct manipulation of model source code. We have developed an interface-based approach to model coupling that avoids this hurdle, allowing for the fast-prototyping of coupled models.
Our approach to model coupling allows the scientist to work with a novel model representation, called the Potential Coupling Interface (PCI). The PCI is an abstraction that exposes only those aspects of a model relevant to coupling, and it is the basis for specifying couplings in the Coupling Description Language (CDL). This research aims to influence the way in which model coupling is practiced in the scientific community.
This research was supported in part by NSF grants ACI-0081487 and SBE-0318372