Semantics Models for Concurrency

9803815本研究涉及并发计算模型。中心问题是(1)为并发语言构建真正的并发模型，以及(2)为无界不确定性构建模型。传统的并发模型将两个进程的并行组合分解为每个进程的顺序动作的所有可能的交错。真并发直接将并发计算作为基本运算符实现。本研究将领域理论方法与自动机理论技术和跟踪理论相结合，在真实并发的基础上建立新的并发计算模型。目标是更好地理解不确定性和并行性之间的区别。潜在的应用是基于对象的并发模型和串行代码的并行化建模。对安全问题（尤其是拒绝服务攻击）建模的应用也在探索中。如果一个进程有无限多个可能的操作选择要执行，那么它就是无限不确定性的。此构造的模型可用于将现有的有界不确定性模型扩展到支持无界不确定性的模型。这将使我们有可能在不表现出无限行为的情况下，对表现出无界有限行为的过程进行推理；例如，一个进程能够执行某一动作任意有限次，但不能永远执行。应用程序涉及规范和细化，以及公平性

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.***