This project will explore issues in designing and implementing a coordination language for multidisciplinary applications. Multidisciplinary applications are becoming increasingly popular among computational scientists, due to their ability to accurately model complex systems involving a variety of multiple disciplines. For example, realistic aircraft modeling is achieved only by coordinating models from different disciplines, such as aerodynamics, propulsion, and structures. While data parallel languages, such as High Performance Fortran, can be successfully applied to many numerical applications, they are ill-equipped to specify and manage the complex coordination required by multidisciplinary applications. Since they lack "Programming-in-the-large" features, most current multidisciplinary applications are executed in a disjoint fashion with file I/O providing the necessary "plumbing" between the various segments of the application. This project will develop an alternative, language-based approach that concentrates on the coordination of the various discipline codes, resulting in a clean and efficient execution model that supports multiple levels of parallelism.
