The Use of the Dynamic Space Decomposition Algorithm of a Computational Area in Heterogeneous Cluster Computations

This article presents an algorithm of the dynamic space decomposition being used in parallel methods of numerical simulations. This algorithm decomposes a numerical model defined by a rectangular grid of material data into subareas. These subareas are then assigned to particular computational nodes of a heterogeneous cluster system during initialisation of the computational process. The present work contains an example use of the algorithm applied to decompose the computational area in parallel implementation of the FDTD (finite-difference time-domain) method. The final part of the article contains the analysis of the decomposing algorithm in the form of the achieved speedup results of the parallel programme for various densities of the finite differences grid