Comparing Implementations of a Calculator for Exact Real Number Computation

Abstract

As one of the first theoretical programming languages for exact real number computation, Real PCF showed to be impractical due to the parallel construct needed for even basic operations. Later, Marcial et al. [12], [11] proposed a variant of Real PCF avoiding the parallelism by introducing a non-deterministic constructor into the language. This language is called LRT. In this paper we present an implementation of a calculator for exact real number computation based on LRT and compare its efficacy with an application of the standard use of real numbers in an imperative programming language. Finally, our implementation is compared with a standard implementation of exact real number computation based on the sign digit representation.