线性代数 05.01 向量的内积

$\color{blue}{第五章 相似矩阵及二次型}$

$\color{blue}{\S 第五章 第一节 向量的内积}$

$\color{blue}{一、向量的内积}$

$\color{blue}{1.内积的概念}$

$定义1.设有n维向量\\ x = \begin{pmatrix} x_1 \\ x_2 \\ \vdots \\ x_n \end{pmatrix}, y = \begin{pmatrix} y_1 \\ y_2 \\ \vdots \\ y_n \end{pmatrix} \\ 令[x, y] = x_1y_1 + x_2y_2 + \cdots + x_ny_n,\\ 称[x, y]为向量x与y的内积.$
$1)内积是一个数(或一个多项式)$
$2)内积是向量的一种运算,可用矩阵的运算.$
$列向量: [x, y] = x^T y$
$行向量: [x, y] = xy^T$

$\color{blue}{2.内积的性质}$

$1)对称性:[x, y] = [y, x];$
$2)齐次性:[\lambda x, y] = \lambda[x, y];$
$3)线性性:[x+y, z] = [x, z] + [y, z].$

$\color{blue}{二、向量的范数与夹角}$

$\color{blue}{1.向量的范数(长度)}$

$定义2.令 \Vert x \Vert = \sqrt{[x, x]} = \sqrt{x_1^2 + x_2^2 + \cdots + x_n^2}\\ 称\Vert x \Vert 为n维向量x的范数.$

$\color{blue}{2.范数的性质}$

$1) 非负性:当x \neq 0时,\Vert x \Vert > 0; 当x = 0时, \Vert x \Vert = 0;$
$2) 齐次性:\Vert \lambda x \Vert = \lambda \Vert x \Vert;$
$3) 三角不定式: \Vert x+y \Vert \leq \Vert x \Vert + \Vert y \Vert .$

$\color{blue}{3.单位向量}$

$称\Vert x \Vert = 1时的向量x为单位向量.任意\alpha \neq 0, e = \dfrac{\alpha}{\Vert \alpha \Vert}为单位向量.$

$\color{blue}{4.许瓦兹不定式}$

$[x, y]^2 \leq [x, x][y, y], \\ 即\left| \dfrac{[x, y]}{\Vert x \Vert \Vert y \Vert } \right| \leq 1 (当\Vert x \Vert \neq 0, \Vert y \Vert \neq 0 时)$

$\color{blue}{5.向量的夹角}$

$\theta = \arccos \dfrac{[x, y]}{\Vert x \Vert \Vert y \Vert } (当\Vert x \Vert \neq 0, \Vert y \Vert \neq 0 时)$

$\color{blue}{三.向量组的正交性}$

$\color{blue}{1.正交}$

$设x、y为n维向量,当[x, y] = 0时,称x与y正交的.$
$若x = 0,则x与任何向量都正交.$

$\color{blue}{2.正交向量组}$

$正交向量组：指一组两两正交的非零向量组.$

$定理1.若n维向量\alpha_1, \alpha_2, \cdots, \alpha_r是一组两两正交的非零向量,\\ 则\alpha_1, \alpha_2, \cdots, \alpha_r线性无关.$
$证:设有\lambda_1, \lambda_2, \cdots, \lambda_r,使\\ \lambda_1 \alpha_1 + \lambda_2 \alpha_2 + \cdots + \lambda_r \alpha_r = 0, \\ 取\alpha_i(i = 1, 2, \cdots, r)在上式两端作内积. \\ [\lambda_1 \alpha_1 + \lambda_2 \alpha_2 + \cdots + \lambda_r \alpha_r, a_i] = [0, \alpha_i] \\ [\lambda_i \alpha_i, \alpha_i] = 0 \\ 亦即 \lambda_i[\alpha_i, \alpha_i] = 0 \\ 因a_i \neq 0,故[\alpha_i, \alpha_i] = \Vert a_i \Vert ^2 \neq 0,\\ 从而必有\lambda_i = 0(i = 1, 2, \cdots, r),\\ 于是向量组\alpha_1, \alpha_2, \cdots, \alpha_a 线性无关.$

$\color{blue}{3.正交基}$

$用正交向量组作向量空间的基,称为向量空间的正交基.$
$例如:n个两两正交的n维非零向量,\\ 可构成向量空间R^n的一个正交基.$

$例1.已知3维向量空间R^3中的两个向量\\ \alpha_1 = \begin{pmatrix}1 \\ 1 \\ 1 \end{pmatrix}, \alpha_2 = \begin{pmatrix}1 \\ -2 \\ 1 \end{pmatrix} \\ 正交,式求一个非零向量\alpha_3,使\alpha_1, \alpha_2, \alpha_3两两正交.$
$解:设所求的向量\alpha_3 = (x_1, x_2, x_3)^T \\ 依题意可得:[\alpha_1, \alpha_3] = 0, [\alpha_2, \alpha_3] = 0,即$
$\left \{ \begin{array}{l} x_1 + x_2 + x_3 = 0 \\ x_1 -2x_2 + x_3 = 0 \end{array} \right.$
$由于A = \begin{pmatrix} 1 & 1 & 1 \\ 1 & -2 & 1 \end{pmatrix} \sim \begin{pmatrix} 1 & 1 & 1 \\ 0 & -3 & 0 \end{pmatrix} \\ \sim \begin{pmatrix} 1 & 0 & 1 \\ 0 & 1 & 0 \end{pmatrix}$
$得\left \{ \begin{array}{l} x_1 = -x_3 \\ x_2 = 0 \end{array} \right.$
$从而有基础解系\begin{pmatrix} -1 \\ 0 \\ 1 \end{pmatrix}, 取\alpha_3 = \begin{pmatrix} -1 \\ 0 \\ 1 \end{pmatrix} ,即为所求.$

$\color{blue}{四.规范正交基(标准正交基)}$

$\color{blue}{1.规范正交基的概念}$

$定义3.设有n维向量e_1, e_2, \cdots, e_r是向量空间V(V \subset R^n)的一个基,\\ 如果e_1, e_2, \cdots, e_r是两两正交的单位向量,\\ 则称e_1, e_2, \cdots, e_r是向量空间V的一个规范正交基. \\ 显然,若e_1, e_2, \cdots, e_r是V的一个规范正交基.则$
$[e_i, e_j] = \left \{ \begin{array}{l} 1 \quad i = j \\ 0 \quad i \neq j \end{array} \right.$
$例如:\\ e_1 = \begin{pmatrix} \dfrac{1}{\sqrt{2}} \\ \dfrac{1}{\sqrt{2}} \\ 0 \\ 0 \end{pmatrix}, e_2 = \begin{pmatrix} \dfrac{1}{\sqrt{2}} \\ -\dfrac{1}{\sqrt{2}} \\ 0 \\ 0 \end{pmatrix}, e_3 = \begin{pmatrix} 0 \\ 0 \\ \dfrac{1}{\sqrt{2}} \\ \dfrac{1}{\sqrt{2}} \end{pmatrix}, e_4 = \begin{pmatrix} 0 \\ 0 \\ \dfrac{1}{\sqrt{2}} \\ -\dfrac{1}{\sqrt{2}} \end{pmatrix},$
$由于 [e_i, e_j] = \left \{ \begin{array}{l} 1 \quad i = j \\ 0 \quad i \neq j \end{array} \right.（i, j = 1, 2, 3, 4)$
$所以e_1, e_2, e_3, e_4是一个规范正交基.$

$\color{blue}{2.向量的坐标}$

$设e_1, e_2, \cdots, e_n是V的一个规范正交基,\\ 那么V中任何一向量\alpha = (x_1, x_2, \cdots, x_n)\\ 应能由e_1, e_2, \cdots, e_n 线性表示,表示法为:\\ \alpha = x_1e_1 + x_2e_2 + \cdots + x_ne_n \\ 为求表示法中的系数x_i,可用e_i与\alpha作内积(i = 1, 2, \cdots, n), \\ [\alpha, e_i] = [x_1e_1 + x_2e_2 + \cdots + x_ne_n, e_i] \\ = x_i[e_i, e_i] = x_i \\ 称x_1, x_2, \cdots, x_n是\alpha在基e_1, e_2, \cdots, e_n下的坐标.$

$\color{blue}{3.施密特标准正交化}$

$设\alpha_1, \alpha_2, \cdots, \alpha_r是向量空间V的一个基, \\ 令b_1 = \alpha_1, \quad e_1 = \dfrac{b_1}{\Vert b_1 \Vert} \\ b_2 = \alpha_2 - [\alpha_2, e_1]e_1, \quad e_2 = \dfrac{b_2}{\Vert b_2 \Vert } \\ b_3= \alpha_3 - [\alpha_3, e_1]e_1 - [\alpha_3, e_2]e_2, \quad e_3= \dfrac{b_3}{\Vert b_3 \Vert } \\ \cdots \qquad \cdots \qquad \cdots \\ b_r = \alpha_r - [\alpha_r, e_1]e_1 - \cdots - [\alpha_r, e_{r-1}]e_{r-1}, \quad e_r = \dfrac{b_r}{\Vert b_r \Vert } \\ 可以证明,e_1, e_2, \cdots, e_r是两两正交的单位向量,\\ 故e_1, e_2, \cdots, e_r是V的一个规范正交基.$

$例2.设\\ \alpha_1 = \begin{pmatrix}1 \\ 2 \\ -1 \end{pmatrix}, \alpha_2 = \begin{pmatrix}-1 \\ 3 \\ 1 \end{pmatrix}, \alpha_3 = \begin{pmatrix}4 \\ -1 \\ 0 \end{pmatrix}, \\ 试用施密特正交化过程把这组向量规范正交化.$
$解:取 b_1 = \alpha_1, e_1 = \dfrac{b_1}{\Vert b_1 \Vert} = \dfrac{1}{\sqrt{6}}\begin{pmatrix}1 \\ 2 \\ -1 \end{pmatrix}, \\ b_2 = \alpha_2 - [\alpha_2, e_1]e_1 \\ = \alpha_2 - \left[ \alpha_2, \dfrac{b_1}{\Vert b_1 \Vert } \right] \cdot \dfrac{b_1}{\Vert b_1 \Vert } \\ = \alpha_2 - \dfrac{[\alpha_2, b_1] b_1}{\Vert b_1 \Vert ^2} \\ = \begin{pmatrix}-1 \\ 3 \\ 1 \end{pmatrix} - \dfrac{4}{6}\begin{pmatrix}1 \\ 2 \\ -1 \end{pmatrix} \\ = \dfrac{5}{3}\begin{pmatrix}-1 \\ 1 \\ 1 \end{pmatrix} \\ e_2 = \dfrac{b_2}{\Vert b_2 \Vert } = \dfrac{1}{\sqrt{3}} \begin{pmatrix}-1 \\ 1 \\ 1 \end{pmatrix} \\ b_3 = \alpha_3 - \dfrac{[\alpha_3, b_1]}{\Vert b_1 \Vert ^2} b_1 - \dfrac{[\alpha_3, b_2]}{\Vert b_2 \Vert ^2} b_2 \\ = \begin{pmatrix}4 \\ -1 \\ 0 \end{pmatrix} -\dfrac{1}{3}\begin{pmatrix}1 \\ 2 \\ -1 \end{pmatrix} + \dfrac{5}{3}\begin{pmatrix}-1 \\ 1 \\ 1 \end{pmatrix} \\ = 2\begin{pmatrix}1 \\ 0 \\ 1 \end{pmatrix} \\ e_3 = \dfrac{b_3}{\Vert b_3 \Vert } = \dfrac{1}{\sqrt{2}}\begin{pmatrix}1 \\ 0 \\ 1 \end{pmatrix} \\ 故e_1, e_2, e_3即合所求.$

$例3.已知\alpha_3 = (1, 1, 1)^T \\ 求非零向量\alpha_1, \alpha_2,使\alpha_3与\alpha_1, \alpha_2正交,\\ 并把它们化成R^3的规范正交基.$
$解:\alpha_1, \alpha_2应满足\alpha_3^Tx = 0的非零解,即\\ x_1 + x_2 + x_3 = 0 \\ 它的基础解系为\xi_1 = \begin{pmatrix}1 \\ 0 \\ -1 \end{pmatrix}, \xi_2 = \begin{pmatrix}0 \\ 1 \\ -1 \end{pmatrix}, \\ 令\alpha_1 = \xi_1, \alpha_2 = \xi_2, \\ 则\alpha_3 与 \alpha_1, \alpha_2正交,显然\alpha_1与\alpha_2线性无关,\\ 因此可用施密特标准正交化. \\ 取 b_1 = \alpha_1, 则e_1 = \dfrac{b_1}{\Vert b_1 \Vert } = \dfrac{1}{\sqrt{2}} \begin{pmatrix}1 \\ 0 \\ -1 \end{pmatrix}, \\ 取 b_2 = \alpha_2 - [\alpha_2, e_1]e_1 = \begin{pmatrix}-\dfrac{1}{2} \\ 0 \\ -\dfrac{1}{2} \end{pmatrix} \\ 则e_2 = \dfrac{b_2}{\Vert b_2 \Vert } = \dfrac{1}{\sqrt{2}}\begin{pmatrix}-1 \\ 2 \\ -1 \end{pmatrix} \\ 再把\alpha_3单位化,得e_3 = \dfrac{\alpha_3}{\Vert \alpha_3 \Vert } = \dfrac{1}{\sqrt{3}}\begin{pmatrix}1 \\ 1 \\ 1 \end{pmatrix}, \\ 故e_1, e_2, e_3即为R^3的一个规范正交基.$

$\color{blue}{五.正交矩阵与正交变换}$

$\color{blue}{1.正交矩阵}$

$定义4.如果n阶矩阵A满足\\ A^TA = E(即A^{-1} = A^T),那么称A为正交矩阵.$
$将矩阵A按列分块,即A = (\alpha_1, \alpha_2, \cdots, \alpha_n), \\ 则\begin{pmatrix} \alpha_1^T \\ \alpha_2^t \\ \vdots \\ \alpha_n^T \end{pmatrix}\begin{pmatrix}\alpha_1, \alpha_2, \cdots, \alpha_n \end{pmatrix} = E.$
$亦即(\alpha_i^T, \alpha_j) = \delta_{ij} = \left \{ \begin{array}{l} 1 \quad i = j \\ 0 \quad i \neq j \end{array} \right.$
$这说明方阵A为正交矩阵的充分必要条件是\\ A的列向量都是两两正交的单位向量. \\ 同理可得方阵A为正交矩阵的充分必要条件是\\ A的行向量都是两两正交的单位向量.$

$例4.验证\\ P = \begin{pmatrix} \dfrac{1}{2} & -\dfrac{1}{2} & \dfrac{1}{2} & -\dfrac{1}{2} \\ \dfrac{1}{2} & -\dfrac{1}{2} & -\dfrac{1}{2} & \dfrac{1}{2} \\ \dfrac{1}{\sqrt{2}} & \dfrac{1}{\sqrt{2}} & 0 & 0 \\ 0 & 0 & \dfrac{1}{\sqrt{2}} & \dfrac{1}{\sqrt{2}} \\ \end{pmatrix}是正交矩阵.$
$解:显然P的每个列向量是两两正交的单位向量.所以P为正交矩阵.$

$例5.设e_1, e_2, \cdots, e_n是R^n的一个规范正交基.A为正交矩阵. \\ 试证:Ae_1, Ae_2, \cdots, Ae_n也是R^n的一个规范正交基.$
$证:由于\\ [Ae_i, Ae_j] = (Ae_i)^TAe_j = e_i^TA^TAe_j \\ = e_i^Te_j = \left \lbrace \begin{array}{l} 1 \quad i = j \\ 0 \quad i \neq j \end{array} \right.$

$\color{blue}{2.正交变换}$

$定义5.设P为正交矩阵,则线性变换y = Px称为正交变换.$
$设y = Px为正交变换,则有 \\ \Vert y \Vert = \sqrt{y^T y} \\ = \sqrt{x^TP^TPx} \\ = \sqrt{x^Tx} \\ = \Vert x \Vert \\ 按\Vert x \Vert表示向量长度,\Vert x \Vert = \Vert y \Vert 说明经过正交变换, \\ 向量的长度保持不变,这是正交变换的优良特性.$