arXivDaily arXiv每日学术速递 周一至周五更新
arXiv math/0612264math.NAcs.CCcs.DScs.NA
收藏

快速线性代数是稳定的

Fast linear algebra is stable

James Demmel, Ioana Dumitriu, Olga Holtz

AI总结:

本文证明,若n×n矩阵乘法可在O(n^{ω+η})次操作内完成,则几乎所有标准线性代数运算(如LU分解、QR分解、线性方程组求解、矩阵求逆、最小二乘问题、特征值问题及奇异值分解)均可在相同复杂度下稳定实现。

摘要:

在之前的一篇论文中,我们证明了大量快速递归矩阵乘法算法在范数意义下是稳定的,并且实际上,如果n×n矩阵的乘法可以通过任何算法在O(n^{ω+η})次操作内完成(对于任意η>0),那么它也可以在O(n^{ω+η})次操作内稳定地完成。这里我们将这一结果扩展,证明基本上所有标准线性代数运算,包括LU分解、QR分解、线性方程组求解、矩阵求逆、求解最小二乘问题、(广义)特征值问题和奇异值分解,也可以在O(n^{ω+η})次操作内稳定地(在范数意义下)完成。

英文摘要:

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 $n$-by-$n$ matrices can be done by any algorithm in $O(n^{ω+ η})$ operations for any $η> 0$, then it can be done stably in $O(n^{ω+ η})$ operations for any $η> 0$. 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 $O(n^{ω+ η})$ operations.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑