Towards a software framework for Model Interoperability

Abstract

Modern mathematical models for simulation of various systems are becoming increasingly complex and intricate. Since no one model can simulate every aspect of a system, the need for these models to be broken into their component parts is imperative for accuracy and maintainability. However, no model can operate alone without data from an outside source, and so further efforts in the Model Interoperability area of research are necessary. Model Interoperability is dedicated to finding methods with which to couple two or more models. This coupling would allow the models to utilize each other's data to produce more comprehensive and more accurate results. There are several methods for model interoperability, including monolithic, component, scheduled, and communication. Also, there are solutions to the problem of model coupling that use one or more of those methods, but most of them require significant code modification. The goal of the software framework proposed in this paper is to minimize the need for code modification and creation, provide a user interface through which to couple the models, and establish a library of models and other activities that the users of the system can utilize for their own model coupling needs.