Title: A new fast parallel statistical measurement technique for computational cosmology

Authors: Robert J. Thacker, H.M.P. Couchman

Addresses: Department of Physics, Engineering Physics and Astronomy, Queen's University, Kingston, Ontario, Canada K7L 3N6. ' Department of Physics and Astronomy, McMaster University, 1280 Main St. West, Hamilton, Ontario, Canada L8S 4M1

Abstract: Higher order cumulants of point processes require significant computational effort to calculate, particularly when evaluated using standard methods such as counts-in-cells. While newer techniques based on tree algorithms are more efficient, they still suffer from shot noise problems in homogeneous systems. We present a new algorithm for calculating higher order moments using Fourier methods. A filtering technique is used to suppress noise, and this approach allows us to calculate skew and kurtosis even when the point process is highly homogeneous. The algorithm can be implemented efficiently in a shared memory parallel environment provided a data-local random sampling technique is used.

Keywords: computational cosmology; statistics; filtering; parallel programming; simulation; high performance computing; skew; kurtosis; shared memory; random sampling.

DOI: 10.1504/IJHPCN.2006.013485

International Journal of High Performance Computing and Networking, 2006 Vol.4 No.5/6, pp.303 - 310

Available online: 01 May 2007 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article