ACM Transactions on Mathematical Software (TOMS)
Implementing a parallel matrix factorization library on the cell broadband engine
Scientific Programming - High Performance Computing with the Cell Broadband Engine
A divide-and-conquer approach for solving singular value decomposition on a heterogeneous system
Proceedings of the ACM International Conference on Computing Frontiers
Hi-index | 0.00 |
A divide-and-conquer method for solving symmetric tridiagonal eigenproblems has evolved from work by Cuppen, Dongarra, Sorensen, Tang, and most recently Gu and Eisenstadt. At the heart of their methods is the solution of a so-called {\em Secular Equation.} Proposed here is a more efficient organization of the equation--solving process, including some crucial implementation details.