Abstract of MGBOUT Project

The aim of this proposal is to implement in the BOUT++ code multigrid techniques to improve its performances and to prepare for new physical studies. BOUT++ is a flexible, modular framework to solve plasma physics relevant differential equations. It includes problem specific differential operators, boundary conditions and geometry. The CCFE version is used to study turbulence and coherent structure motions in the edge of magnetic fusion machines. Preliminary work with the HLST will be focused on training CCFE staff on multigrid techniques and HLST staff on BOUT++. In a second phase, geometrical multigrid techniques should be implemented in the module of the code that calculates Laplace inversions (heavily used for determining the electrostatic potential in the equations of the model). This is expected to improve the preconditioning of the solver, which is currently an important bottleneck, and thus allowing a significant speed up of the code. This, in turn, would make possible high resolution, massively parallel simulations of plasma turbulence in statistical steady state. This prototype work will also be preparatory for future modifications of BOUT++ aimed at implementing multigrid techniques more extensively. This will allow multigrid techniques to be extended to the inversion in the implicit time stepping and also make it possible to solve 3D potential inversion which are important in strongly non-orthogonal coordinates (e.g. near X-point).