Two concurrent algorithms of discrete potential field construction

Abstract

Increasing demand for computational power in contemporary constructions has created the need to build faster CPUs and construct more efficient algorithms. In this context especially the concurrent algorithms seem to be very promising. Depending on the number of available CPUs they may offer significant reductions in computation time. In this article two concurrent algorithms of potential field generation are proposed. They present two different approaches to problem domain partitioning called by the authors respectively as tearing and nibbling. It is shown that depending on the problem topology either Tear algorithm or Nibble algorithm is faster. Conclusions are summed up in form of experimental results presenting how the algorithms work in practice. However algorithms construct a discrete potential field according to some specific scheme, there should be no major problems with generalization them to other potential field schemes. © 2010 Springer-Verlag Berlin Heidelberg.