Testing Matrix Function Algorithms Using Identities

Testing Matrix Function Algorithms Using Identities
复制标题

使用恒等式测试矩阵函数算法

DOI:
10.1145/2723157
复制
发表时间:
2016
影响因子:
2.7
通讯作者:
Deadman E
Deadman E
中科院分区:
计算机科学3区
文献类型:
--
作者:
Deadman E

文献摘要

参考文献

被引文献

相似文献

用于计算矩阵函数的算法通常通过将前向误差与条件数和单位舍入的乘积进行比较来测试。前向误差是借助参考解来计算的,通常以高精度计算。另一种方法是使用功能恒等式,例如“往返测试”elogA=A 和 (A1/p)p=A,正如当前在 SciPy 测试模块中所采用的那样。我们展示了函数恒等式的线性化扰动分析如何确定与组成矩阵函数评估的后向稳定性一致的最大残差。最大残差与计算残差的比较为后向稳定性提供了必要的测试。我们还展示了如何计算这些关系的实际线性化后向误差。我们的方法利用 Fréchet 导数及其规范的估计。数值实验表明,所提出的方法既能够检测不稳定性,又能够确认稳定性。
Algorithms for computing matrix functions are typically tested by comparing the forward error with the product of the condition number and the unit roundoff. The forward error is computed with the aid of a reference solution, typically computed at high precision. An alternative approach is to use functional identities such as the “round-trip tests”elogA=Aand (A1/p)p=A, as are currently employed in a SciPy test module. We show how a linearized perturbation analysis for a functional identity allows the determination of a maximum residual consistent with backward stability of the constituent matrix function evaluations. Comparison of this maximum residual with a computed residual provides a necessary test for backward stability. We also show how the actual linearized backward error for these relations can be computed. Our approach makes use of Fréchet derivatives and estimates of their norms. Numerical experiments show that the proposed approaches are able both to detect instability and to confirm stability.
DOI: 10.1007/s11075-009-9357-1
发表时间: 2010-01
影响因子: 2.1
作者:
F. Greco;B. Iannazzo
通讯作者: F. Greco;B. Iannazzo
DOI: 10.1137/130920137
发表时间: 2014-01
期刊: SIAM J. Matrix Anal. Appl.
影响因子: --
作者:
M. Aprahamian;N. Higham
通讯作者: M. Aprahamian;N. Higham
DOI: 10.1137/s0895479894273614
发表时间: 1996
期刊: SIAM J. Matrix Anal. Appl.
影响因子: --
作者:
L. Dieci;B. Morini;A. Papini
通讯作者: A. Papini
DOI: 10.1137/080716426
发表时间: 2008
期刊: SIAM J. Matrix Anal. Appl.
影响因子: --
作者:
Awad H. Al;N. Higham
通讯作者: N. Higham
DOI: 10.1007/s11075-012-9665-8
发表时间: 2012-12
影响因子: 2.1
作者:
J. Baglama;L. Reichel;D. Richmond
通讯作者: J. Baglama;L. Reichel;D. Richmond