Three-Level Parallel J-Jacobi Algorithms for Hermitian Matrices - Computer Science > Numerical AnalysisReport as inadecuate




Three-Level Parallel J-Jacobi Algorithms for Hermitian Matrices - Computer Science > Numerical Analysis - Download this document for free, or read online. Document in PDF available to download.

Abstract: The paper describes several efficient parallel implementations of theone-sided hyperbolic Jacobi-type algorithm for computing eigenvalues andeigenvectors of Hermitian matrices. By appropriate blocking of the algorithmsan almost ideal load balancing between all available processors-cores isobtained. A similar blocking technique can be used to exploit local cachememory of each processor to further speed up the process. Due to diversity ofmodern computer architectures, each of the algorithms described here may be themethod of choice for a particular hardware and a given matrix size. Allproposed block algorithms compute the eigenvalues with relative accuracysimilar to the original non-blocked Jacobi algorithm.



Author: Sanja Singer, Sasa Singer, Vedran Novakovic, Davor Davidovic, Kresimir Bokulic, Aleksandar Uscumlic

Source: https://arxiv.org/







Related documents