线性代数 04.03 线性相关性的判定

$\color{blue}{\S 第四章 第三节 线性相关性的判定}$

$\color{blue}{一、方程组、矩阵、向量组的关系}$

$\left \{ \begin{array}{l} a_{11}x_1 + a_{12}x_2 + \cdots + a_{1n}x_n = b_1 \\ a_{21}x_1 + a_{22}x_2 + \cdots + a_{2n}x_n = b_2 \\ \cdots \qquad \cdots \qquad \cdots \qquad \cdots \quad \cdots \\ a_{m1}x_1 + a_{m2}x_2 + \cdots + a_{mn}x_n = b_m \\ \end{array} \right. \qquad (1)$
$\begin{pmatrix} a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ \cdots & \cdots & \cdots & \cdots \\ a_{m1} & a_{m2} & \cdots & a_{mn} \\ \end{pmatrix}\begin{pmatrix}x_1 \\ x_2 \\ \vdots \\ x_n \end{pmatrix} = \begin{pmatrix}b_1 \\ b_2 \\ \vdots \\ b_n \end{pmatrix}$
$即 Ax = b \qquad (2)$
$将A按列分块,由(2)得\\ x_1 \alpha_1 + x_2 \alpha_2 + \cdots + x_n \alpha_n = b \qquad (3)$
$显然,由(3)式知,若b能由\alpha_1, \alpha_2, \cdots, \alpha_n线性表示,\\ 则线性方程组(1)有解,\\ 若不能由\alpha_1, \alpha_2, \cdots, \alpha_n线性表示,\\ 则线性方程组(1)无解;$
$当b = 0时,(3)式变为\\ x_1 \alpha_1 + x_2 \alpha_2 + \cdots + x_n \alpha_n = 0 \quad (4)$
$显然,由(4)知,若\alpha_1, \alpha_2, \cdots, \alpha_n线性相关,\\ 则它所对应的齐次线性方程组Ax = 0有非零解,\\ 若\alpha_1, \alpha_2, \cdots, \alpha_n线性无关,则Ax = 0仅有零解.$
$\color{green}{综上所述,\\ 向量b能不能由向量组\alpha_1, \alpha_2, \cdots, \alpha_n线性表示,\\ 则说明它所对应的非齐次线性方程组Ax = b有没有解;\\ 向量组\alpha_1, \alpha_2, \cdots, \alpha_n的线性相关性,\\ 则说明它所对应的齐次线性方程组Ax = 0有什么样的解.}$

$例如,向量组 \\ \alpha_1 = \begin{pmatrix} 1 \\ 2 \\ -3 \end{pmatrix}, \alpha_2 = \begin{pmatrix} 3 \\ 0 \\ 1 \end{pmatrix}, \alpha_3 = \begin{pmatrix} 5 \\ 1 \\ 0 \end{pmatrix}, \beta = \begin{pmatrix} 9 \\ 6 \\ -7 \end{pmatrix}$
$显然, \beta = 3\alpha_1 + 2\alpha_2 + 0 \alpha_3,所以\\\\线性方程组 x_1 \alpha_1 + x_2 \alpha_2 + x_3 \alpha_3 = \beta$
$即 \left \{ \begin{array}{l}x_1 + 3x_2 + 5x_3 = 9 \\ 2x_1 + x_3 = 6 \\ -3x_1 + x_2 = -7 \end{array} \right.有一个解是(3, 2, 0)$

$向量组 \\ \alpha_1 = \begin{pmatrix} 1 \\ 1 \\ -1 \end{pmatrix}, \alpha_2 = \begin{pmatrix} 3 \\ 0 \\ 1 \end{pmatrix}, \beta = \begin{pmatrix} 1 \\ 1 \\ 0 \end{pmatrix}$
$由于 \begin{vmatrix} 1 &3 & 1 \ 1 & 0 & 1 \ -1 & 1 & 0 \end{vmatrix} \neq 0, 即\alpha_1, \alpha_2, \beta 线性无关,\\ 所以\beta不能由\alpha_1, \alpha_2线性表示,即线性方程组x_1\alpha_1 + x_2 \alpha_2 = \beta无解.$
$即\left \{ \begin{array}{l} x_1 + 3x_2 = 1 \\ x_1 = 1 \\ -x_1 + x_2 = 0 \end{array} \right. 无解.$

$向量组 \\ \alpha_1 = \begin{pmatrix} 1 \\ 2 \\ -3 \end{pmatrix}, \alpha_2 = \begin{pmatrix} 3 \\ 0 \\ 1 \end{pmatrix}, \alpha_3 = \begin{pmatrix} 9 \\ 6 \\ -7 \end{pmatrix}$
$由于 \\ \begin{vmatrix} 1 & 2 & -3 \\ 3 & 0 & 1 \\ 9 & 6 & -7 \end{vmatrix} = 0, 即\alpha_1, \alpha_2, \alpha_3 线性相关,\\ 且\alpha_3 = 3\alpha_1 + 2\alpha_2$
$所以x_1\alpha_1 + x_2\alpha_2 + x_3\alpha_3 = 0有非零解$

$\color{blue}{二、线性相关性的判定}$

$定理4.\\ 向量组\alpha_1,\alpha_2,\cdots, \alpha_m线性相关的充分必要条件\\ 是它所构成的矩阵A = (\alpha_1, \alpha_2, \cdots, \alpha_m)的秩小于向量个数m;\\ 向量组线性无关的充分必要条件是R(A) = m.$

$例1.n维向量组$
$e_1 = \begin{pmatrix} 1 \\ 0 \\ \vdots \\ 0 \end{pmatrix},e_2 = \begin{pmatrix} 0 \\ 1 \\ \vdots \\ 0 \end{pmatrix}, \cdots, e_n = \begin{pmatrix} 0 \\ 0 \\ \vdots \\ 1 \end{pmatrix}$
$称为n维单位坐标向量组,试讨论它的线性相关性.$
$解:n维单位坐标向量组构成的矩阵\\ E = (e_1, e_2, \cdots, e_n) \\ 是n阶的单位矩阵.由|E| = 1 \neq 0,知R(E) = n,\\ 即R(E)等于向量组中向量的个数,故由定理4知向量组\\ 是线性无关的.$

$例2.已知\\ \alpha_1 = \begin{pmatrix} 1 \\ 1 \\ 1 \end{pmatrix}, \alpha_2 = \begin{pmatrix} 0 \\ 2 \\ 5 \end{pmatrix}, \alpha_3 = \begin{pmatrix} 2 \\ 4 \\ 7 \end{pmatrix}, \\ 试讨论向量组\alpha_1, \alpha_2, \alpha_3及向量组\alpha_1, \alpha_2的线性相关性.$
$解:\\ 对矩阵(\alpha_1, \alpha_2, \alpha_3)施以初等行变换,\\ 使之变成行梯形矩阵,即可同时看出\\ 矩阵(\alpha_1, \alpha_2, \alpha_3)及矩阵(\alpha_1, \alpha_2)的秩,\\ 由定理4即可得出结论. \\ \begin{pmatrix} \alpha_1, \alpha_2, \alpha_3 \end{pmatrix} = \begin{pmatrix} 1 & 0 & 2 \\ 1 & 2 & 4 \\ 1 & 5 & 7 \end{pmatrix} \\ \sim \begin{pmatrix} 1 & 0 & 2 \\ 0 & 2 & 2 \\ 0 & 5 & 5 \end{pmatrix} \sim \begin{pmatrix} 1 & 0 & 2 \\ 0 & 1 & 1 \\ 0 & 0 & 0 \end{pmatrix} \\ 矩阵(\alpha_1, \alpha_2, \alpha_3)的秩为2, 小于向量个数3,线性相关\\ 矩阵(\alpha_1, \alpha_2)的秩为2,等于向量个数2,线性无关.$

$例3.已知向量组\alpha_1, \alpha_2, \alpha_3线性无关,令\\ \beta_1 = \alpha_1 + \alpha_2, \beta_2 = \alpha_2 + \alpha_3, \beta_3 = \alpha_3 + \alpha_1, \\ 试证向量组\beta_1, \beta_2, \beta_3线性无关.$
$证:设有x_1, x_2, x_3使\\ x_1 \beta_1 + x_2 \beta_2 + x_3 \beta_3 = 0 \\ 即x_1(\alpha_1 + \alpha_2) + x_2(\alpha_2 + \alpha_3) + x_3(\alpha_3 + \alpha_1) = 0 \\ 亦即(x_1 + x_3) \alpha_1 + (x_1 + x_2)\alpha_2 + (x_2 + x_3)\alpha_3 = 0 \\ 因\alpha_1, \alpha_2, \alpha_3线性无关,故有:$
$\left \{ \begin{array}{l} x_1 + x_3 = 0 \\ x_1 + x_2 = 0 \\ x_2 + x_3 = 0 \end{array} \right.$
$由此方程组的系数行列式 \begin{vmatrix} 1 & 0 & 1 \\ 1 & 1 & 0 \\ 0 & 1 & 1 \end{vmatrix} = 2 \neq 0\\ 故方程组只有零解x_1 = x_2 = x_3 = 0$
$\therefore 向量组\beta_1, \beta_2, \beta_3线性无关$

$\color{blue}{定理5.(1)若向量组A:\alpha_1, \alpha_2, \cdots, \alpha_m 线性相关,\\ 则向量组B:\alpha_1, \alpha_2, \cdots, \alpha_m, \alpha_{m + 1}也线性相关.\\ 反之,若向量组B线性无关,则向量组A也线性无关.}$
$证:记A = \begin{pmatrix} \alpha_1, \alpha_2, \cdots, \alpha_m \end{pmatrix}, \\ B = \begin{pmatrix} \alpha_1, \alpha_2, \cdots, \alpha_m, \alpha_{m + 1} \end{pmatrix}\\ 有R(B) \leq R(A) + 1, 若向量组A线性相关,\\ 则由定理4有R(A) < m,\\ 从而R(B) \leq R(A) + 1 < m + 1,\\ 再由定理4知向量组B线性相关. \\ 用反证法证明线性无关部分:\\ 假设A线性相关,则R(A) < m,\\ 从而R(B) < m + 1,这与B线性无关矛盾.\\ 所以,若向量组B线性无关,则向量组A也线性无关$

$由上面证明知:\\ 一个向量组若有线性相关的部分组,则该向量组必线性相关.\\ 特别地,含有零向量的向量组一定线性相关.\\ 一个向量组线性无关,则它的任何部分组都线性无关.$

$定理5.(2)设\alpha_j = \begin{pmatrix} a_{1j} \\ \vdots \\ a_{rj} \end{pmatrix}, \beta_j = \begin{pmatrix} a_{1j} \\ \vdots \\ a_{rj} \\ a_{(r+1)j}\end{pmatrix}, (j = 1, 2, \cdots, m)\\ 即向量\alpha_j添上一个分量后得向量\beta_j,若向量组A:\alpha_1, \alpha_2, \cdots, \alpha_m线性无关, \\ 则向量组B:\beta_1, \beta_2, \cdots, \beta_m也线性无关,\\ 反之,若向量组B线性相关,则向量组A也线性相关.$
$证:记A_{r \times m} = \begin{pmatrix} \alpha_1, \alpha_2, \cdots, \alpha_m \end{pmatrix}, \\ B_{(r+1) \times m} = \begin{pmatrix} \beta_1, \beta_2, \cdots, \beta_m \end{pmatrix}, \\ 有R(A) \leq R(B).若向量组A线性无关,\\ 则R(A) = m,从而R(B) \geq m,\\ 但R(B) \leq m,故R(B) = m,\\ 因此向量组B线性无关. \\ 证线性相关用反证法, \\ 设:若向量组B线性相关时,向量组A线性无关 \\ 由前半部分可知，向量组A线性无关,则向量组B线性无关 \\ 这与假设矛盾,故假设不成立.\\ 即：若向量组B线性相关,则向量组A也线性相关.$

$推论:若r维的向量线性无关,在r维的向量组每个向量都添\\ 上n - r个分量,得n维的向量组,则n维的向量组线性无关.$

$定理5.(3)m个n维向量组成的向量组,当维数n小于向量的个数m时一定线性相关.$
$证:m个n维向量\alpha_1, \alpha_2, \cdots, \alpha_m构成的矩阵A_{n \times m} = (\alpha_1, \alpha_2, \cdots, \alpha_m),有R(A) \leq n. \\ 若n < m,则R(A) < m,故m个向量\alpha_1, \alpha_2, \cdots, \alpha_m线性相关.$

$例4.设有向量组\alpha_i^T = (a_i, a_i^2, \cdots, a_i^n)(i = 1, 2, \cdots, m. m \leq n),\\ 试证:向量组\alpha_1^T, \alpha_2^T, \cdots, \alpha_m^T,线性无关,\\ 其中a_1, a_2, \cdots, a_m为m个互不相等且不等于零的常数.$
$证:因为\\ \alpha_1^T = (a_1, a_1^2, \cdots, a_1^m, \cdots, a_1^n) \\ \alpha_2^T = (a_2, a_2^2, \cdots, a_2^m, \cdots, a_2^n) \\ \cdots \quad \cdots \quad \cdots \quad \cdots \\ \alpha_m^T = (a_m, a_m^2, \cdots, a_m^m, \cdots, a_m^n) \\ 前m个分量做成一个行列式\\ \begin{vmatrix} a_1 & a_2 & \cdots & a_m \\ a_1^2 & a_2^2 & \cdots & a_m^2 \\ \cdots & \cdots & \cdots & \cdots \\ a_1^m & a_2^m & \cdots & a_m^m \\ \end{vmatrix} = a_1a_2 \cdots a_m \begin{vmatrix} 1 & 1 & \cdots & 1 \\ a_1 & a_2 & \cdots & a_m \\ \cdots & \cdots & \cdots & \cdots \\ a_1^{m -1} & a_2^{m -1} & \cdots & a_m^{m -1} \\ \end{vmatrix} \\ = a_1a_2 \cdots a_m \prod_{i \leq j < i \leq m}(a_j - a_i) \neq 0 \\ 从而向量组\\ \beta_1^T = (a_1, a_1^2, \cdots, a_1^m) \\ \beta_2^T = (a_2, a_2^2, \cdots, a_2^m) \\ \cdots \quad \cdots \quad \cdots \quad cdots \\ \beta_m^T = (a_m, a_m^2, \cdots, a_m^m) \\ 线性无关,所以增加分量后所得到的向量组\alpha_1^T, \alpha_2^T, \cdots, \alpha_m^T线性无关.$

$例5.设A是n \times m矩阵,B是m \times n矩阵,其中n < m,若AB = E,证明B的列向量线性无关.$
$证:设B = (\beta_1, \beta_2, \cdots, \beta_n),其中\\ \beta_1, \beta_2, \cdots, \beta_n是B的列向量,若\\ x_1\beta_1 + x_2 \beta_2 + \cdots + x_n \beta_n = 0 \\ 即\begin{pmatrix} \beta_1, \beta_2, \cdots, \beta_n \end{pmatrix} \begin{pmatrix} x_1 \\ x_2 \\ \vdots \\ x_n \end{pmatrix} = BX = 0 \\ 两边左乘A得ABX = 0, 即EX = 0,从而X = 0,\\ 所以\beta_1, \beta_2, \cdots, \beta_n线性无关.$

$例6.设向量 \beta 可由向量组\alpha_1, \alpha_2, \cdots, \alpha_m线性表示,\\ 但不能由向量组(Ⅰ)\alpha_1, \alpha_2, \cdots, \alpha_{m -1}表示,\\ 记向量组(Ⅱ) \beta,\alpha_1, \alpha_2, \cdots, \alpha_{m -1},\\ 则\alpha_m能由(Ⅱ)线性表示,但不能由(Ⅰ)线性表示.$

$证:由于\beta可由\alpha_1, \alpha_2, \cdots, \alpha_m线性表示,即\\ \beta = \lambda_1 \alpha_1 + \lambda_2 \alpha_2 + \cdots + \lambda_m \alpha_m \\ 又因为\beta不能用向量组\alpha_1, \alpha_2, \cdots, \alpha_{m- 1} 线性表示,所以\lambda_m \neq 0, \\ 从而\\ \alpha_m = \dfrac{1}{\lambda_m}\beta - \dfrac{\lambda_1}{\lambda_m} \alpha_1 - \dfrac{\lambda_2}{\lambda_m}\alpha_2 - \cdots - \dfrac{\lambda_{m -1}}{\lambda_m}\alpha_{m - 1} \\ 故 \alpha_m能由(Ⅱ)线性表示. \\ 假设\alpha_m能由(Ⅰ)线性表示,则有\\ \alpha_m = k_1 \alpha_1 + k_2 \alpha_2 + \cdots + k_{m - 1} \alpha_{m - 1} \\ 所以\\ \beta = \lambda_1 \alpha_1 + \lambda_2 \alpha_2 + \cdots + \lambda_m \alpha_m \\ = \lambda_1 \alpha_1 + \lambda_2 \alpha_2 + \cdots + \lambda_m (k_1 \alpha_1 + k_2 \alpha_2 + \cdots + k_{m - 1} \alpha_{m - 1}) \\ = (\lambda_1 + \lambda_m k_1) \alpha_1 + (\lambda_2 + \lambda_m k_2)\alpha_2 + \cdots + (\lambda_{m - 1} + \lambda_m k_{m -1})\alpha_{m -1} \\ 这与\beta不能由(Ⅰ)线性表示矛盾,\\ 故\alpha_m不能由(Ⅰ)线性表示.$