TRANSLATION OF SUPERCOMBINATOR PROGRAM INTO G-CODE

Abstract

The growing trend towards parallel processing has also led to increased interest in the other fields associated to it. Programming the available hardware in parallel processing architectures to achieve maximum efficiency is one of the major challenges, to which Functional Languages are being proposed as the suitable solutions. These declarative languages have simple semantics, strong mathematical base and properties to exploit the inherent parallelism in. the problem, without producing any undue side—effects. The functional languages are first translated into an intermediate form called lambda calculus and then implemented using graph reduction techniques. The Beta reduction is the basic rule for evaluation, which may follow either Applicative or Normal orderreduction. However, the lambda calculus being inefficient for implementation purposes, Supercombinators are used as an alternative form of intermediate representation. Considering the focus of current research to implement Functional languages on sequential machines, the suitability of supercombinators towards this, end makes it all the more attractive. A supercombinator program can be compiled into a fixed sequence of codes called G—code executable in an abstract sequential machine known as the G—machine. In this dissertation, the translator for supercombinator program into G-Code has been developed