A Dynamically Adaptive Parallel Solver for Porous Media Flow and Tsunami Simulation

Zeit:	13. Juni 2013
Referent*in:	Oliver Meister Institut für Informatik, TU München
Veranstaltungsort:	Pfaffenwaldring 61, Raum U1.003 (MML), Universität Stuttgart
Download als iCal:

Many problems in geosciences require dynamically adaptive grids for reasons of precision and efficiency. Compared to regular grids, the precision of the solutions is higher while memory requirements of the data are lower. This is important especially for computationally inexpensive simulations, whose performance is generally memory-bound.In our approach, triangular structured grids based on space-filling curves are exploited for efficient traversal, refinement and parallelization schemes. Applications are implemented in a generic framework, similar to Dune-PDELab, and include Porous Media Flow, Tsunami Simulation and Inviscid Flow.