Poster (Other)

FZJ-2023-05038

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png pySDC: Prototyping Spectral Deferred Corrections

Baumann, T. (Corresponding author)FZJ* ; Speck, R.FZJ* ; Lunet, T. ; Ruprecht, D. ; Wimmer, L.

2023

This record in other databases:

Please use a persistent id in citations: doi:10.34734/FZJ-2023-05038

Abstract: Spectral deferred corrections (SDC) is a very malleable time-stepping scheme that is especially well-suited for stiff problems. A wide range of parallel-in-time (PinT) extensions exist, from small-scale parallelism across the method to the large-scale PinT algorithm PFASST. SDC iteratively solves fully-implicit Runge-Kutta (RK) schemes with low-order methods. pySDC is a Python library that allows interested users to test how well SDC works for their specific problem or algorithm with very little effort. It includes implementations of PinT algorithms in both simulated parallel versions as well as MPI-based parallel versions. In addition, implementations of many popular serial RK methods are part of pySDC to enable a quick and fair comparison in an internally consistent framework. The code is very actively developed by collaborators in Juelich, Hamburg and Wuppertal. pySDC is publicly hosted on GitHub and extensively tested with continuous integration to ensure reproducibility and maintainability of the code and the results obtained with it. Contribution guidelines and tutorials exist to invite and guide new users. The modular nature allows developers to work on specific features without awareness of all details in the code, which encourages also smaller projects, such as master's theses.

---

Contributing Institute(s):

1. Jülich Supercomputing Center (JSC)

Research Program(s):

1. 5112 - Cross-Domain Algorithms, Tools, Methods Labs (ATMLs) and Research Groups (POF4-511) (POF4-511)
2. TIME-X - TIME parallelisation: for eXascale computing and beyond (955701) (955701)

Appears in the scientific report 2023

---

Database coverage:

---