中文

Fast linear algebra is stable

数值分析 2011-11-09 v3 计算复杂性 数据结构与算法

摘要

In an earlier paper, we showed that a large class of fast recursive matrix multiplication algorithms is stable in a normwise sense, and that in fact if multiplication of nn-by-nn matrices can be done by any algorithm in O(nω+η)O(n^{\omega + \eta}) operations for any η>0\eta > 0, then it can be done stably in O(nω+η)O(n^{\omega + \eta}) operations for any η>0\eta > 0. Here we extend this result to show that essentially all standard linear algebra operations, including LU decomposition, QR decomposition, linear equation solving, matrix inversion, solving least squares problems, (generalized) eigenvalue problems and the singular value decomposition can also be done stably (in a normwise sense) in O(nω+η)O(n^{\omega + \eta}) operations.

关键词

引用

@article{arxiv.math/0612264,
  title  = {Fast linear algebra is stable},
  author = {James Demmel and Ioana Dumitriu and Olga Holtz},
  journal= {arXiv preprint arXiv:math/0612264},
  year   = {2011}
}

备注

26 pages; final version; to appear in Numerische Mathematik