Parallel language constructs for paradigm integration and deterministic computations

We describe parallel extensions of sequential programming languages for writing pro-grams that integrate dierent programming paradigms and that execute in heterogeneous environments comprising both distributed and shared memory. The extensions can be used to write programs with dynamic process and communication structures. Programs can use shared-memory, message passing, and data parallel programming paradigms, and can be written in a way that permits the compiler and run-time system to verify that they are deterministic. The extensions also provide the programmer with control over how data and processes are mapped to processors, and hence how computational resources are allocated to dierent parts of a program. A subset of these ideas has been incorporated in an extension to Fortran called Fortran M. However, the underlying sequential notation is not central to the ideas. 1