NUMERICAL-ANALYSIS USING NONPROCEDURAL PARADIGMS

This article presents a survey on the innovative features of a handful of languages that offer new features that can be valuable in numerica l analysis, and a survey of the pros and cons of the languages with re gards to work in numerical analysis. Language features such as polymor phism, first-class functions, and object-oriented programming offer im proved writability, readability, reliability, and maintenance of compu ter software. The article discusses language features and uses, and in cludes a comparison of current implementations. It is intended both as an introduction to nonprocedural language features for persons workin g in numerical mathematics and as an exploration of some of the langua ge requirements of numerical mathematics for persons working in langua ge development. The article discusses C++, Fortran 77, Fortran 90, Has kell, Lisp/CLOS, Modula-3, Sather, and SML with respect to a variety o f numerical analysis tasks: interpolation, optimization, array access and update, iteration, recursion, random number generation, and Gaussi an elimination on sparse matrices.