TOWARDS MODELS OF REALISTIC COMPUTING MACHINES IN COMPUTER SCIENCE

Abstract

The paper presents an approach to system modelling in design of both hardware and software systems. It is based on the definition of models of machines that can be directly implemented. The paper shows how to render less abstract and more realistic the abstract machines defined by theoreticians, so that they can capture implementation and technological-oriented aspects, such as testability, and allow an easy transition to final implementations. A realistic abstract machine for lambda-calculus is then presented and the design of system for lambda-expressions evaluation is illustrated. The architecture chosen for the system is based on a collection of finite state automata, evolving concurrently and communicating via a broadcast system. Some conclusive remarks about the use of realistic models arc finally drawn.