A parareal waveform relaxation algorithm for semi-linear parabolic partial differential equations (Q442706)

The authors combine the parareal algorithm and a kind of waveform relaxation (WR) algorithm to develop a new parallel algorithm, which can be carried out in parallel in two different directions for semi-linear parabolic partial differential equations. Sharp bounds on errors are presented, which indicate the approximately superlinear convergence for the new algorithm on a bounded time domain. The convergence of the parareal WR algorithm on an unbounded time domain is also analyzed. Considering the parallel efficiency, the new approach is superior to the classical parareal algorithm. It is observed numerically that much less running time is needed on a massively parallel computer than the time used by the corresponding WR algorithm to achieve the desired accuracy.