A Proposed Taxonomy for Software Development Risks for High-Performance Computing (HPC) Scientific/Engineering Applications

Because the development of large-scale scientific/engineering application codes is an often difficult, complicated, and sometimes uncertain process, success depends on identifying and managing risk. One of the drivers of the evolution of software engineering, as a discipline, has been the desire to identify reliable, quantifiable ways to manage software development risks. The taxonomy that follows represents an attempt to organize the sources of software development risk for scientific/engineering applications around three principal aspects of the software development activity: the software development cycle, the development environment, and the programmatic environment. These taxonomic classes are divided into elements and each element is further characterized by its attributes.