Scientific software frameworks and grid computing: Improving programming productivity

Abstract

Scientific research applications, or codes, are notoriously difficult to develop, use, and maintain. This is often because scientific software is written from scratch in traditional programming languages such as C and Fortran, by scientists rather than expert programmers. By contrast, modern commercial applications software is generally written using toolkits and software frameworks that allow new applications to be rapidly assembled from existing component libraries. In recent years, scientific software frameworks have started to appear, both for grid-enabling existing applications and for developing applications from scratch. This paper compares and contrasts existing scientific frameworks and extrapolates existing trends. Show more