High Performance Computing (HPC) Engineer

Parallel input-output of Uhaina code

for global storm surge simulations

Contact: Matthieu Haefele (matthieu.haefele@cnrs.fr)

Location: University de et des Pays de L'adour (Pau site, France)

Salary: 2000 - 2300 € net / month according to experience

Duration : 2 years

Starting date: September - November 2023

Context

The latest advances in the study of the consequences of climate change indicate that storm surge events will be more frequent and more powerful. The impact of these phenomena on coasts will increase and it is necessary to predict as much as possible the places where this impact will be important in order to be able to anticipate it and to take the adequate measures. This is one of the missions of the Bureau de Recherches Géologiques et Minières (BRGM) and the Uhaina code is developed and used for this purpose.

The Lagoon project contributes to the improvement of this code in order to make possible the simulation of this type of pheomena at global scale. The aim is to design advanced numerical methods capable of solving the problem and to implement them efficiently in the parallel code. In particular, inputs and outputs will become critical for simulations performance on the whole globe.

Work plan

The first step will be to measure the performance of several Inputs-Output (IO) technologies on the data closest to the full globe simulation in order to determine the relevant technology(ies). We will also evaluate the ability of these technologies to perform in-situ and/or in-transit IO with online post-processing. In a second step, the technology(ies) will be integrated into the code. More precisely, the integration work will be done in the AeroSol library, on which Uhaina is based. Finally, this new methodology will be documented and users will be supported in adapting their current post-processing.

Required skills

• C++ programming language
• Parallel computing (including MPI)
• Team work

Would be a plus :

• Python language
• Parallel IO technologies