Transformational derivation of (parallel) programs using skeletons

We describe a framework for the derivation of programs for arbitrary (in particular, parallel) architectures, motivated by a generalization of the derivation process for sequential algorithms. The central concept in this approach is that of a skeleton: on the one hand, a higher-order function for targeting transformational derivations at, on the other hand representing an elementary computation on the architecture aimed at. Skeletons thus form a basis for intermediate languages, that can be implemented once and for all, as a process separate from individual program developments. The available knowledge on the derivation of (higher-order) functional programs can be used for deriving parallel ones. In: H. Wijshoff, ed.: Computing Science in the Netherlands 1993, pp. 97-108. Copies on email request or (probably easiest) by clicking ftp://ftp.win.tue.nl/pub/math.prog.construction/skel.ps.Z.