Software Development Strategies for High-Energy Physics Simulations Based on Quantum Field Theory

摘要

Let us review the main elements of strategies, which we identified in the development of high energy physics simulations. Based on exemplary Monte Carlo simulation projects, several conjectures for the general aspects of work on scientific software are presented, including openness to change, robustness against external changes, modular development processes, and the constant evolution of tests. The analyzed projects were developed over more than 30 years, much longer than the typical lifetime of computer industry software. Some general observations are drawn together with quantitative evaluation on how the quality of software, measured mainly by the precision of the simulation, grows together with the complexity of software and tests. We find these observations interesting, even if attempts to generalize may remain to a degree speculative, as every researcher is tuning that to the specific needs of their own application. In the future, this can be validated, adjusted, or extended by the experiences gained in other projects.