An adaptive domain-decomposition technique for parallelization of the fast marching method

Parallel re-initialization of level set functions on distributed unstructured tetrahedral grids

“…The parallel re-initialization algorithm is ap- To avoid the approximation error introduced by the usage of k-d trees, we chose m ¼ jF [ Lj as previously used for the error analysis in Section 7. In Table 2, the error and the rate of convergence of the re-initialization algorithm are shown if disturbing u sp by (17). The rate of convergence (ROC) is given by the quotient of the error on two consecutive triangulations.…”

“…This process is iteratively refined exchanging data across sub-domain boundaries. In contrast to a domain decomposition approach, the parallel FMM presented in [17] is based on distributing the interface and is discussed for Cartesian grids. To the best of our knowledge, there is no publication of any parallel FMM on unstructured grids.…”

A Fine-Grained Parallel Model for the Fast Iterative Method in Solving Eikonal Equations

“…An other method is proposed in [16] where the decomposition is made on the narrow-band itself. The initial front is partitioned at the initialization and then each processor solves a sub-problem independently ( fig.…”

“…Fair load balancing is a real concern in [16] where the authors proposed to choose the sets in such a way that the emerging wave fronts ideally cover nearly the same portions of the computational domain. However, this assumes that we know the behavior of the solution.…”

“…It is interesting to study works on parallel FMM [15], [16] since the FIM takes root in the FMM. We present in this subsection two different programming parallel models which can be used for the FMM.…”