An analysis of scientific computing environments : a consumer's view

Abstract

In the last three decades, with rapid advancements in the hardware and software industry, a large number of commercial and free open source languages and software have evolved. Many of these are Very High Level Languages (VHLLs) which can be easily used for scientific computing purposes such as algorithm testing and engineering computations. However, this vast pool of resources has not been utilized to its full potential. In this analysis, we will be looking at various simple and complex problems and how they can be approached in various languages. All the results will be uploaded on a website in the form of a wiki intended to be accessible to everyone. By analyzing standard problems encountered frequently in scientific computing, this wiki provides the users a performance based report which they can use to choose the best option for their particular applications. Simultaneously, a lexicon of standard codes will help them in learning those options which they want to use so that fear is not a barrier. The analysis also addresses some incompatibility issues within languages and their impact. This work is a preliminary investigation as part of Professor Alan Edelman's participation in the Numerical Mathematics Consortium. We expect the scientific computing community to benefit from this research as a whole, as this analysis will give them better alternatives for their computational needs.