Using high-performance parallel-in-time integration to model acoustics in expanding volumes

Initial-value problems describing physical systems have long suffered from their inablility to be parallelized. The Parareal algorithm and its variations provide a means for parallel-in-time integration of these initial-value problems, leading to significantly decreased computation time. Further parallelization and performance improvements can be achieved using distributed networks of GPUs. Due to its potential complexities, the case of expanding volumes with a temporally constant speed of sound is considered to highlight these methods.