9803815 This research concerns models for concurrent computation. The central problems are (1) building truly concurrent models for concurrent languages, and (2) building models for unbounded nondeterminism. Traditional models for concurrency resolve the parallel composition of two processes into all possible interleavings of the sequential actions of each. True concurrency directly implements concurrent computation as a primitive operator. This research will combine domain-theoretic approaches with automata-theoretic techniques and trace theory to build new models for concurrent computation based on true concurrency. The goal is a better understanding of the difference between nondeterminism and parallelism. Potential applications are models of object-based concurrency and modeling the parallelization of serial code. Applications to modeling security issues, especially denial of service attacks are also being explored. A process is unboundedly nondeterministic if it has infinitely many choices of possible actions to perform. Models for this construct could be used to extend existing models for bounded nondeterminism to ones supporting unbounded nondeterminism. This would afford the potential to reason about processes exhibiting unbounded finitary behavior without also exhibiting infinitary behavior; for example, a process which is capable of doing an action any finite number of times, but incapable of doing it forever. Applications involve specification and refinement, as well as fairness.***
