Parallel Time Integration: An Approaching Paradigm Shift for Scientific Computing
Robert D. Falgout, USDOE National Nuclear Security Administration (NNSA), Jacob B. Schroder · 2023
This note argues that parallel-in-time methods will be necessary for doing highfidelity time-dependent simulations in the future.A "proof" is given to support the argument and to provide a framework for debate.The effect of a parallel-in-time paradigm on scientific computing practice is also discussed. Skepticism about parallel-in-timeSkepticism about the viability and necessity of parallel-in-time methods continues.Here are some of the questions and comments the authors have heard on this topic:• What about causality?Parallel-in-time methods converge to the same space-time solution as time stepping, so causality is not broken.