Alan Sussman
Computer Science Department and UMIACS
University of Maryland
email: als@cs.umd.edu
Simulation of physical systems has become the third leg of investigation in many scientific disciplines, along with theory and experimentation. We will address the problem of providing software support for simulation of complex physical systems requiring multiple physical models, potentially at multiple scales and resolutions. Employing multiple physical models presents several difficult challenges, both in modeling the physics correctly, and in efficiently using multiple simulation codes to model a complete physical system. In particular, the individual models must be coupled, to allow them to exchange information either at boundaries where the models align in physical space, or in areas where the models overlap in space. We describe our efforts to address three main parts of the problem, being developed within the Maryland InterComm project: (1) comprehensive support for specifying at runtime what data is to be transferred, (2) flexibly and efficiently deciding when the data should be moved, and (3) effectively deploying multiple coupled simulation codes in a high performance distributed computing environment (the Grid). A main goal of the project is to minimize the changes to each individual simulation code, because those codes are generally complex and difficult to modify.