Postdoctoral Position: Multiphysics simulations for geodynamics on heterogeneous exascale systems

• the efficient implementation of new models, methods, and algorithms into existing high-performance frameworks,
• the fast prototyping of new ideas in individual code,
• an interest in the entire simulation pipeline: starting from simple algorithms to state-of-the-art applications in geophysics executed on supercomputers,
• as well as the presentation of the results to the scientific community in journals, at conferences, and as lecture contributions for students.

• A Ph.D. in the field of Applied Mathematics, Computer Science, Computational Science and Engineering, or similar.
• Knowledge of numerics as part of a degree program. In particular, knowledge about finite-element analysis is an absolute must. Familiarity with iterative solvers, preconditioners, multigrid methods, and mixed-precision approaches is optional but extremely useful.
• We require good programming skills and experience in C++ and its common programming idioms and design patterns. Knowledge of MATLAB, Python, Julia, or a similar scripting language is necessary for prototyping.
• Interest and affinity for high-performance computing are necessary for the position. You should have experience with the roofline model and familiarity with a profiler. Experience with GPUs is a bonus.
• Of course, you need fluency in written and spoken English to communicate your ideas in this interdisciplinary project.

Note that we expect from candidates either a strong background in high-performance computing or in numerical analysis. If you think that you have the requirements above, you should apply for this position., * a motivational letter,

• a curriculum vitae, and
• relevant publications, for instance, your Ph.D. thesis.

• a young, dynamic, and international team of scientists from various disciplines, as well as cooperation with international partners,
• an exciting and varied project with many different aspects and possibilities for expansion,
• modern hardware and infrastructure at the workplace, from compute- and GPU servers to supercomputers,
• a salary according to the public service rates (level E13), and
• everything at a renowned and modern university in a beautiful city close to the Alps.

The BMBF-funded position is part of the CoMPS project, which is a multidisciplinary project combining the fields of mathematics, computer science, geophysics, and high-performance computing. It aims to improve the performance of the matrix-free finite-element-based framework HyTeG, in particular by techniques for data reduction through surrogate operators. Furthermore, we aim to extend the framework with mixed-precision approaches, which will be implemented and analyzed. Secondly, we aim at a volume coupling of the high-performance packages waLBerla, HyTeG, and MesaPD to solve complex multiphysics problems. The coupling is done across package boundaries. This also requires more sophisticated approaches in load-balancing.