Coarsening of Simplicial Meshes for Large ScaleParallel FEM Computations with DOLFIN HPC : A parallel implementation of the edge collapse algorithm

Abstract: Adaptive mesh refinement and coarsening methods are effective techniques to reduce the computation time of finite element based solvers. Parallel imple- mentations of such adaption routines, suitable for large scale computations on distributed memory machines, need additional care. In this thesis, a coarsening technique based on edge collapses is presented, its implementation and opti- mization for parallel computations explained and it is analyzed with respect to coarsening efficiency and performance. As a possible application the use of mesh coarsening in adaptive flow simulations is demonstrated