Nonprocedural programming, such as functional, has been an attractive approach to programming parallel computers. Programs written in such languages exhibit implicit parallelism which can be exploited for their efficient parallel execution. This project will investigate the applicability of equational style to parallel programming. Equational programming is a style of nonprocedural programming in which a system of oriented equations of the form l -> r, where l and r are terms, constitute a program. Equational languages have logical semantics which is intuitive and easy to understand. A new class of equations called path sequential systems has been identified. This class of equations is naturally suitable for parallel execution of equational programs on both shared memory as well as message passing architectures. Efficient strictness analysis techniques have been developed which work well for flat parallelism in equational programs. In continuation of this research, a parallelizing compiler and run- time system for a first order equational language that is constructor based will be implemented. The implementation will be done on a shared memory and on a message passing parallel computer. A major thrust of this investigation will be on load balancing and minimizing inter-processor communications. Specifically, dynamic load balancing strategies that are based on regular communication patterns of path sequential systems will be investigated. Finally, the performance of the equational programming system will be examined by executing numerical and nonnumerical applications such as sparse matrix computations and toy theorem provers.
