MESSENGERS is a distributed programming system based on the principles of self-migrating threads, which carry their own identity and behavior in the form of programs. This enables them to navigate freely in the underlying computational network, communicate with one another, and invoke native- mode C functions. Thus MESSENGERS is a general coordination paradigm for distributed applications development. Its main advantages are dynamic composability of applications, enabling an interactive and incremental development of arbitrarily extensible applications, and the ability to exploit clusters of heterogeneous workstations. This project investigates the capabilities of this new technology primarily in the context of individual-based simulation, that is, simulation where the application is modeled as a collection of autonomous individuals (e.g. particles in physics), each having its own state and behavior. It focuses on issues of performance and scalability by developing new techniques for load balancing and dynamic resource utilization that exploit the self-migration nature of the subcomputations.
