Software project management and quality engineering practices for complex, coupled multiphysics, massively parallel computational simulations: Lessons learned from ASCI

Many institutions are now developing large-scale, complex, coupled multiphysics computational simulations for massively parallel platforms for the simulation of the performance of nuclear weapons and certification of the stockpile, and for research in climate and weather prediction, magnetic and inertial fusion energy, environmental systems, astrophysics, aerodynamic design, combustion, biological and biochemical systems, and other areas. The successful development of these simulations is aided by attention to sound software project management and software engineering. We have developed lessons learned from a set of code projects that the Department of Energy National Nuclear Security Agency has sponsored to develop nuclear weapons simulations over the last 50 years. We find that some, but not all, of the software project management and development practices (rather than processes) commonly employed for non-technical software add value to the development of scientific software and we identify those that we judge add value. Another key finding, consistent with general software industry experience, is that the optimal project schedule and resource level are solely determined by the requirements once the requirements are fixed.