Type-based primitives for distributed data structures: a functional approach

We formulate type-based primitive functions for distributed processing of complex data structures. These primitives are derived directly from the type definitions of the data structures, and they naturally lead to distributed implementations. Our main results are: 1) a distribution model to support efficient aggregate operations over complex data structures. The model is simple yet general, and easy to implement. 2) A parallel functional programming environment assembled from readily available hardware and software systems. Functional programs execute in parallel and exhibit good speedup in this environment if they use the proposed type-based primitives.