STABILITY ANALYSIS AND IMPROVEMENT OF THE BLOCK GRAM-SCHMIDT ALGORITHM
暂无分享,去创建一个
The advent of supercomputers with hierarchical memory systems has imposed the use of block algorithms for the linear algebra algorithms. Although block algorithms may result in impressive improvements in performance, their numerical properties are quite different from their scalar counterpart and deserve an in-depth study. In this paper, the numerical stability ofblock Gram-Schmidt orthogonalization is studied and a variant is proposed which has numerical properties similar to the modified Gram-Schmidt procedure while retaining most of the performance advantages of the block formulation.
[1] Å. Björck. Solving linear least squares problems by Gram-Schmidt orthogonalization , 1967 .
[2] William Jalby,et al. The use of BLAS3 in linear algebra on a parallel processor with a hierarchical memory , 1987 .
[3] B. Philippe. An algorithm to improve nearly orthonormal sets of vectors on a vector processor , 1987 .
[4] William Jalby,et al. Impact of Hierarchical Memory Systems On Linear Algebra Algorithm Design , 1988 .