Numerical discretisation libraries are a cornerstone of the HPC landscape with C++ based projects such as MFEM and others being routinely used for extreme scale simulations at the frontier of computational sciences. So why should we reinvent the wheel in Rust?
In this talk we want to motivate the development of Rust native numerical discretisation libraries at the example of our own ND project, an early stage framework of support libraries for grid based discretisation methods in Rust. We will discuss the use of Rust native features, such as its advanced traits system, in the development of ND, interfacing other libraries, deployment on HPC systems, and our view on the future of the project and the Rust based HPC ecosystem.
