Parallel implementations of the False Nearest Neighbors method for distributed memory architectures

The False Nearest Neighbors (FNN) method is particularly relevant in several fields of science and engineering (medicine, economics, oceanography, biological systems, etc.). In some of these applications, it is important to give results within a reasonable time scale; hence, the execution time of the FNN method has to be reduced. This paper describes two parallel implementations of the FNN method for distributed memory architectures. A ‘Single-Program, Multiple Data’ (SPMD) paradigm is employed using a simple data decomposition approach where each processor runs the same program but acts on a different subset of the data. The computationally intensive part of the method lies mainly in the neighbor search and therefore this task is parallelized and executed using 2 to 64 processors. The accuracy and the performance of the two parallel approaches are then assessed and compared with the best sequential implementation of the FNN method, which appears in the TISEAN project. The results indicate that the two parallel approaches, when the method is run using 64 processors on a SGI Origin 3800, are between 40 and 80 times faster than the sequential one. The efficiency is between 65 and 125%. Copyright © 2010 John Wiley & Sons, Ltd.