NHR PerfLab Seminar: Breaking through the barrier of time integration for climate and weather simulations (March 26, online)

Symbolic picture for the article. The link opens the image in a large view.

TitleBreaking through the barrier of time integration for climate and weather simulations (online)

Speaker: Univ.-Prof. Dr. Martin Schreiber, Applied Mathematics | High Perf. Scientific Comp. for PDEs | Université Grenoble Alpes | Lab. Jean Kuntzmann | INRIA | France

Date and time: Tuesday, March 26, 2024, 3:30-4:30 p.m. CET 



Weather and climate simulations face new challenges due to changes in computer architectures caused by physical limitations. From a pure computing perspective, algorithms are required to cope with stagnating or decreasing per-core speed and increasing on-chip parallelism. These and other showstopping trends will continue and already led to research on partly disruptive mathematical and algorithmic reformulations of dynamic cores, e.g., using (additional) parallelism along the time dimension.

This presentation provides an overview and introduction of a selection of the following promising newly developed and evaluated time integration methods for equations related to prototypical dynamical cores, all aimed at improving the ratio of wall clock time vs. error: Exponential Integration, Rational Approximation of Exponential Integration (REXI), Parallel Full Approximation Scheme in Space and Time (PFASST), Semi-Lagrangian methods combined with Parareal and MultiGrid Reduction In Time (MGRIT). We get improved time-vs.-error rates, but sometimes with additional challenges on the way which need to be further overcome. Overall, our results motivate further investigation and combination of these methods for operational weather/climate systems.

I gratefully acknowledge collaborators related to this presentation:
Jed Brown, Finn Capelle, François Hamon, Richard Loft, Michael Minion, Nathanaël Schaeffer, Andreas Schmitt, Joao G.C. Steinsträsser, Pedro S. Peixoto, Raphael Schilling.

Short Bio:Prof. Dr. Martin Schreiber

Martin Schreiber is a full professor at the Université Grenoble Alpes (UGA) on the adventurous path between high-performance computing and applied mathematics for solving PDEs. His primary research is on developing (parallel-in-)time integration methods incorporating the HPC and applied mathematics perspective.

He started his doctorate at the Technical University of Munich (TUM), where he developed new algorithms for highly efficient hybrid parallelization of dynamical h-adaptive meshes for explicit time steppers combined with a novel approach for dynamic schedulers and resource utilization. In 2015, he was appointed as a proleptic lecturer at the University of Exeter in the department of mathematics, where he and his international collaborators developed a variety of novel time integration methods in an HPC/math co-design fashion, most of them targeting to improve the performance of weather models. After briefly returning to TUM, he was appointed a full professor at the Université Grenoble Alpes in “High-performance scientific computing for PDEs”.

For a list of past and upcoming NHR PerfLab seminar events, see: https://hpc.fau.de/research/nhr-perflab-seminar-series/