This paper presents an approach to translating the data associated with a problem-solving procedure into efficient implementations. The approach involves reduction of problem-domain data strutures to implementations by way of intermediate modelling structures. Formalists are introduced for describing modelling structures -- abstract representations of data characteristics and relationships, independent of any specific implemetation -- and implementation structures -- machine-processable represetations. Based upon these formalisms, algorithms are presented for recognizing known modelling structures, for synthesizing implementations for modelling structures not recognized, and for combining several implementations according to structure membership and variable binding relationships. Design considerations influencing these formalisms and algorithms are discussed.

A storage structures formalism is described which can be used as a precise symbolic representation of low-level storage organizations and also as a description of storage requirements to a storage allocation mechanism. The formalism is based on three forms of memory management (sequential, linked, and associative) and associated referencing mechanisms (indexing, pointing, and hashing). Several examples of the formalism are presented. The meaning, or interpretation, of a storage structure as used to implement a modelling structure is discussed. Examples of alternative implementation structures (storage structures and their interpretations) for a particular modelling structure are presented.

This research concerns the problem of specifying the information relationships and their transformations included, explicitly and implicitly, in any problem-solving procedure. Our view of data representation is that problem representations (in a problem domain) are mapped to a machine representation (in an implementation domain) through various modelling representations (in modelling domains). Modelling domain representations make easier the discovery of acceptable implementation representations. We propose to focus on the study of mappings of data representations and their transformations from the modelling to implementation domains. Specifically, our goal is to answer four questions:

1) what are appropriate formalisms for describing modelling domains and implenentation domains,

2) what knowledge exists of each of these domains and how can this knowledge be represented in our formalisms,

3) how do we map specific representations from one domain to the other, and

4) how can we test the completeness of our formalisms and mapping process.

In answering the fourth question, we propose to construct an interactive system for generating alternative implementation domain representations from a modelling domain representation and selecting that which comes closest to the user's desired program performance criteria in an actual programming context.

Structured process description employs a small set of constructs which facilitate functional programming with concurrency. These constructs can be mapped directly onto a process interconnection architecture. Using these structured connectives, maximum pipelining throughput can be achieved. The constructs are presented and illustrated with example programs.