【最优化方法】对称矩阵的对角化

最新推荐文章于 2024-11-24 00:17:00 发布

撕得失败的标签

最新推荐文章于 2024-11-24 00:17:00 发布

阅读量1.6k

点赞数 24

分类专栏：最优化方法文章标签：矩阵线性代数正交化对角化

本文链接：https://blog.csdn.net/qq_61828116/article/details/135332150

版权

最优化方法专栏收录该内容

8 篇文章

订阅专栏

文章目录

正交化方法
- 示例
矩阵正交化

正交化方法

设 $R^n$ 中线性无关组 $a_1,a_2,a_3,\dots,a_n$ ，令
$\begin{aligned} & \beta_1 = \alpha_1 \\ & \beta_2 = \alpha_2 - {\frac{[\alpha_2\beta_1]}{||\beta_1||}} \beta_1 \\ & \beta_3 = \alpha_3 - {\frac{[\alpha_3\beta_1]}{||\beta_1||}} \beta_1 - {\frac{[\alpha_3\beta_2]}{||\beta_2||}} \beta_2 \\ & \beta_n = \alpha_3 - {\frac{[\alpha_n\beta_1]}{||\beta_1||}} \beta_1 - \cdots - {\frac{[\alpha_n\beta_{n-1}]}{||\beta_{n-1}||}} \beta_{n-1} \end{aligned}$

该方法称为施密特正交化（Gram–Schmidt process）

$[x, y]$ 为向量的内积， $∣∣ x ∣∣ = [x, x]$
$x_1y_1 + x_2y_2 + \cdots + x_ny_n$

示例

将向量组
$\begin{align} & \alpha_1=(1,1,0,0)^T,\alpha_2=(1,0,1,0)^T \\ & \alpha_3=(-1,0,0,1)^T,\alpha_4=(1,-1,-1,1)^T \\ \end{align}$

标准正交化

解：先正交化
$\begin{aligned} & \beta_1 =(1,1,0,0)^T \\ & \beta_2 = (1,0,1,0)^T-\frac{1}{2}(1,1,0,0)^T = \frac{1}{2}(1,-1,2,0)^T \\ & \beta_3 = (-1,0,0,1)^T + \frac{1}{2}(1,1,0,0)^T + \frac{1}{6}(1,-1,2,0)^T = \frac{1}{3}(-1,1,1,3)^T \\ & \beta_4 = (1,-1,-1,1)^T-0-0-0=(1,-1,-1,1)^T \end{aligned}$

再标准化
$\begin{aligned} & \beta_1 = \frac{1}{\sqrt2} (1,1,0,0)^T \\ & \beta_2 = \frac{1}{\sqrt6} (1,-1,2,0)^T \\ & \beta_3 = \frac{1}{2\sqrt3} (-1,1,1,3)^T \\ & \beta_4 = \frac{1}{2} (1,-1,-1,1)^T \end{aligned}$

矩阵正交化

$\begin{pmatrix} 0 & 1 & 1 & -1 \\ 1 & 0 & -1 & 1 \\ 1 & -1 & 0 & 1 \\ -1 & 1 & 1 & 0 \\ \end{pmatrix}$

求一正交矩阵 $P$ ，使 $P^{T}AP$ 成对角形。

解：
$\begin{aligned} & |A-\lambda E| ~=~ \begin{vmatrix}-\lambda & 1 & 1 & -1 \\ 1 & -\lambda & -1 & 1 \\ 1 & -1 & -\lambda & 1 \\ -1 & 1 & 1 & -\lambda \\ \end{vmatrix} ~=~ \begin{vmatrix} 1-\lambda & 1-\lambda & 1-\lambda & 1-\lambda \\ 1 & -\lambda & -1 & 1 \\ 1 & -1 & -\lambda & 1 \\ -1 & 1 & 1 & -\lambda \\ \end{vmatrix} \\\\\\ & =~ (1-\lambda) \begin{vmatrix} 1 & 1 & 1 & 1 \\ 1 & -\lambda & -1 & 1 \\ 1 & -1 & -\lambda & 1 \\ -1 & 1 & 1 & -\lambda \\ \end{vmatrix} ~=~ (1-\lambda) \begin{vmatrix} 1 & 1 & 1 & 1 \\ 0 & -\lambda-1 & -2 & 0 \\ 0 & -2 & -\lambda-1 & 0 \\ 0 & 2 & 2 & 1-\lambda \\ \end{vmatrix} \\\\\\\ & =~ (1-\lambda)^2(\lambda^2+2\lambda-3) = (\lambda-1)^3(\lambda+3) \end{aligned}$

求得 $\large \lambda_1=\lambda_2=\lambda_3=1, \lambda_4=-3$ ，

把 $\lambda_1=1$ （3重）带入齐次方程组，得
$\begin{pmatrix}-1 & 1 & 1 & -1 \\ 1 & -1 & -1 & 1 \\ 1 & -1 & -1 & 1 \\-1 & 1 & 1 & -1 \\ \end{pmatrix}= \begin{pmatrix} 1 & -1 & -1 & 1 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 \\ \end{pmatrix}$ $\begin{cases} x_1 = x_2 + x_3 - x_4 \\ x_2 = x_2 \\ x_3 = ~~~~~~~~~x_3 \\ x_4 = ~~~~~~~~~~~~~~~~~~x_4 \\ \end{cases} => x_2 \begin{pmatrix} 1 \\ 1 \\ 0 \\ 0 \end{pmatrix}+ x_3 \begin{pmatrix} 1 \\ 0 \\ 1 \\ 0 \end{pmatrix} + x_4 \begin{pmatrix} -1 \\ 0 \\ 0 \\ 1 \end{pmatrix}$

得出基础解系 $\zeta_1,\zeta_2,\zeta_3$ ，
$\zeta_1 = \begin{pmatrix} 1 \\ 1 \\ 0 \\ 0 \end{pmatrix}, \zeta_2 = \begin{pmatrix} 1 \\ 0 \\ 1 \\ 0 \end{pmatrix}, \zeta_3 = \begin{pmatrix} -1 \\ 0 \\ 0 \\ 1 \end{pmatrix}$

将 $\zeta_1,\zeta_2,\zeta_3$ 正交化 : 取 $\eta_1 = \zeta_1$ ，
$\begin{aligned} & \eta_2 = \zeta_2 - \frac{[\eta_1,\zeta_2]}{||\eta_1||}\eta_1 = \begin{pmatrix} 1 \\ 0 \\ 1 \\ 0 \end{pmatrix}- \frac{1}{2} \begin{pmatrix} 1 \\ 1 \\ 0 \\ 0 \end{pmatrix}= \frac{1}{2} \begin{pmatrix} 1 \\ -1 \\ 2 \\ 0 \end{pmatrix} \\ & \eta_3 = \zeta_3 - \frac{[\eta_3,\zeta_1]}{||\eta_1||}\eta_1- \frac{[\eta_3,\zeta_2]}{||\eta_2||}\eta_2 = \begin{pmatrix} -1 \\ 0 \\ 0 \\ 1 \end{pmatrix} + \frac{1}{2} \begin{pmatrix} 1 \\ 1 \\ 0 \\ 0 \end{pmatrix} + \frac{1}{6} \begin{pmatrix} 1 \\ -1 \\ 2 \\ 0 \end{pmatrix} = \frac{1}{3} \begin{pmatrix} -1 \\ 1 \\ 1 \\ 3 \end{pmatrix} \end{aligned}$

将 $\eta_1,\eta_2,\eta_3$ 单位化求得 $p_1,p_2,p_3$
$\begin{aligned} & p_1 = \frac{1}{\sqrt2} (1,1,0,0)^T \\ & p_2 = \frac{1}{\sqrt6} (1,-1,2,0)^T \\ & p_3 = \frac{1}{\sqrt{12}} (-1,1,1,3)^T \end{aligned}$

把 $\lambda_4=-3$ 带入齐次方程组，得
$\begin{pmatrix} 3 & 1 & 1 & -1 \\ 1 & 3 & -1 & 1 \\ 1 & -1 & 3 & 1 \\-1 & 1 & 1 & 3 \\ \end{pmatrix}= \begin{pmatrix} 1 & 0 & 0 & -1 \\ 0 & 1 & 0 & 1 \\ 0 & 0 & 1 & 1 \\ 0 & 0 & 0 & 0 \\ \end{pmatrix}$ $\begin{cases} x_1 = x_4 \\ x_2 = -x_4 \\ x_3 = -x_4 \\ x_4 = x_4 \\ \end{cases} => x_4 \begin{pmatrix} 1 \\ -1 \\ -1 \\ 1 \end{pmatrix}$

得出基础解系 $\zeta_4$ ，
$\zeta_4 = \begin{pmatrix} 1 \\ -1 \\ -1 \\ 1 \end{pmatrix}$

将 $\zeta_4$ 单位化，得 $p_4$ ，
$p_4 = \frac{1}{2} (1,-1,-1,1)^T$

将 $p_1,p_2,p_3,p_4$ 构成正交矩阵 $P$
$(p_1,p_2,p_3,p_4) = \begin{pmatrix} \frac{1}{\sqrt2} & \frac{1}{\sqrt6} & -\frac{1}{\sqrt{12}} & \frac{1}{2} \\ \frac{1}{\sqrt2} & -\frac{1}{\sqrt6} & \frac{1}{\sqrt{12}} & -\frac{1}{2} \\ 0 & \frac{2}{\sqrt6} & \frac{1}{\sqrt{12}} & -\frac{1}{2} \\ 0 & 0 & \frac{3}{\sqrt{12}} & \frac{1}{2} \\ \end{pmatrix}$