Multi-agent simulation of complex heterogeneous models in scientific computing

Electronic protoyping is becoming a part of every scientific inquiry and product design, and is the focus of research in the new scientific field of Computational Science and Engineering. The new grand challenge here is the rapid prototyping of manufactured artifacts and the rapid solution to problems with numerous interrelated elements. This, in turn, requires the fast, accurate simulation of physical processes and design optimization using knowledge and computational models from multiple disciplines in science and engineering. In this paper, we formulate a mathematical and software framework for complex rapid prototyping. Its design utilizes the current computer network infrastructures and high performance computation technologies. Its functionality includes adaptability and intelligence with respect to end-users and hardware platforms. We present the architecture of this framework, named SciAgents, using a multi-agent software model encapsulating a collaborating mathematical method. We also briefly discuss some issues related to legacy software reuse that we faced in the implementation. The design of SciAgents allows wholesale reuse of scientific software and provides a natural approach to parallel and distributed problem solving.