ESCAPE2 Summer school program
Towards exascale computing for numerical weather prediction
The school will take place online
Accurate numerical weather prediction relies on massively parallel computing resources and models that can efficiently harness that computational power. As high-performance computing architectures approach the exascale (1018 floating-point operations per second), forecast models are undergoing significant scientific and technological change in order to run at finer and finer resolution within tight operational constraints.
The school aims to introduce students to computational tools currently used in operations and research on weather forecast models, including numerical algorithms, uncertainty quantification models, and high-performance computing tools. Theoretical lectures will be complemented by hands-on training where the students will develop small group projects using the techniques presented in the course.
Goals and structure
The goal of the school is to introduce doctoral students and other young researchers in applied mathematics and computational physics to the scientific computing problems arising in the development of numerical weather prediction models that can exploit the computational power of forthcoming exascale architectures. International experts in weather prediction, scientific computing, numerical mathematics, high performance computing and software engineering will give a comprehensive introduction to the main challenges in this area and will lead working groups, in which students will have an opportunity to work directly with examples of advanced implementations of numerical methods.
Luca Bonaventura, Politecnico di Milano, Italy
Tommaso Benacchio, Politecnico di Milano, Italy