线性代数 04.06 线性方程组解的结构

$\color{blue}{\S 第四章 第六节 线性方程组解的结构}$

复习:关于线性方程组的两个重要定理:
$1) n个未知数的齐次线性方程组Ax = 0有非零解的\\\\充分必要条件是系数矩阵的秩R(A) < n.$
$2) n个未知数的非齐次线性方程组Ax = b有解的\\\\充分必要条件是系数矩阵的秩R(A)等于增广矩阵的秩R(B).\\\\且当R(A) = R(B) = n时,Ax = b有唯一解.\\\\当R(A) = R(B) = r < n时,Ax = b有无穷多个解.$
$3) 线性相关性与齐次线性方程组解的关系.$
$4) 线性表示与非齐次线性方程组解的关系.$

$\color{blue}{一、齐次线性方程组}$

设有齐次线性方程组
$\left \{ \begin{array}{l} a_{11}x_1 + a_{12}x_2 + \cdots + a_{1n}x_n = 0 \\ a_{21}x_1 + a_{22}x_2 + \cdots + a_{2n}x_n = 0 \\ \cdots \quad \cdots \quad \cdots \quad \cdots \quad \cdots \quad \cdots \\ a_{m1}x_1 + a_{m2}x_2 + \cdots + a_{mn}x_n = 0 \\ \end{array} \right. \qquad (1)$
$记A = \begin{pmatrix} a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ \cdots & \cdots & \cdots & \cdots \\ a_{m1} & a_{m2} & \cdots & a_{mn} \\ \end{pmatrix}, x = \begin{pmatrix} x_1 \\ x_2 \\ \vdots \\ x_n \end{pmatrix}$
$则(1)式可写成向量方程: Ax = 0 \qquad (2)$
$称x = \begin{pmatrix} x_1 \\ x_2 \\ \vdots \\ x_n \end{pmatrix}为方程组(1)的解向量,它也就是向量方程(2)的解.$
$性质1.若x = \xi_1, x = \xi_2为(2)的解,则x = \xi_1 + \xi_2也是(2)的解.$
$证:只要验证x = \xi_1 + \xi_2满足(2): \\ A(\xi_1 + \xi_2) = A\xi_1 + A\xi_2 = 0 + 0 = 0.$
$性质2.若x = \xi_1为(2)的解,k为实数,则x = k\xi_1也是(2)的解.$
$证: 因为A(k\xi_1) = k(A\xi_1) = k0 = 0$

$若用S表示方程组(1)的全体解向量所组成的集合,\\ 则性质1、2即为\\ 1.若\xi_1 \in S, \xi_2 \in S,则\xi_1 + \xi_2 \in S; \\ 2.若\xi_1 \in S, k \in R,则k\xi_1 \in S.$
$这说明集合S对向量的线性运算是封闭的,所以集合S\\ 是一个向量空间,称为齐次线性方程组(1)的解空间.$

$下面我们来求解空间S的一个基.\\ 设系数矩阵A的秩为r,并不妨设A的前r个列向量线性无关,\\ 于是A的行最简形矩阵为:\\ B = \begin{pmatrix} 1 & \cdots & 0 & b_{11} & \cdots & b_{1n - r} \\ \vdots & & \vdots & \vdots & & \\ 0 & \cdots &1 & b_{r1} & \cdots & b_{rn-r} \\ 0 & & & \cdots & & 0 \\ \vdots & & & & & \\ 0 & & & \cdots & & 0 \\ \end{pmatrix}, \\ 与B对应,即有同解方程组$
$\left \{ \begin{array}{l} x_1 = -b_{11}x_{r+1} - \cdots - b_{1n -r} x_n \\ \cdots \quad \cdots \quad \cdots \quad \cdots \\ x_r = -b_{r1}x_{r + 1} - \cdots -b_{rn -r}x_n \end{array} \right. \qquad (3)$

$由于A与B的行向量组等价,故方程组(1)与(3)同解. \\ 在(3)中任给x_{r+1}, \cdots, x_n一组值,即唯一确定x_1, \\ \cdots, x_r的值,就得(3)的一个解,也就是(1)的解. \\ 现在令x_{r+1}, \cdots, x_n取下列n-r组数:$
$\begin{pmatrix} x_{r+1} \\ x_{r+2} \\ \vdots \\ x_n \\ \end{pmatrix} = \begin{pmatrix}1 \\ 0 \\ \vdots \\ 0 \\ \end{pmatrix},\begin{pmatrix}0 \\ 1 \\ \vdots \\ 0 \\ \end{pmatrix},\cdots,\begin{pmatrix}0 \\ 0 \\ \vdots \\ 1 \\ \end{pmatrix},$
$\begin{pmatrix} x_{r+1} \\ x_{r+2} \\ \vdots \\ x_r \\ \end{pmatrix} =\begin{pmatrix}-b_{11} \\ -b_{21} \\ \vdots \\ -b_{r1} \\ \end{pmatrix},\begin{pmatrix} -b_{12} \\ -b_{22} \\ \vdots \\ -b_{r2} \\ \end{pmatrix},\cdots,\begin{pmatrix}-b_{1n-r} \\ -b_{2n-r} \\ \vdots \\ -b_{rn-r} \\ \end{pmatrix},$
$从而求得(3)[也就是(1)]的n-r个解.$
$\xi_1 = \begin{pmatrix} -b_{11} \\ \vdots \\ -b_{r1} \\ 1 \\ 0 \\ \vdots \\ 0 \end{pmatrix}, \xi_2 = \begin{pmatrix} -b_{12} \\ \vdots \\ -b_{r2} \\ 0 \\ 1 \\ \vdots \\ 0 \end{pmatrix}, \cdots, \xi_{n-r} = \begin{pmatrix} -b_{1(n-r)} \\ \vdots \\ -b_{r(n-r)} \\ 0 \\ 0 \\ \vdots \\ 1 \end{pmatrix}.$
$下面证明\xi_1, \xi_2, \cdots, \xi_{n-r}就是解空间S的一个基.$
$首先,由于(x_{r+1}, x_{r+2}, \cdots, x_n)^T所取的n-r个n-r维的向量$
$\begin{pmatrix}1 \\ 0 \\ \vdots \\ 0 \end{pmatrix}, \begin{pmatrix}0 \\ 1 \\ \vdots \\ 0 \end{pmatrix}, \cdots, \begin{pmatrix}0 \\ 0 \\ \vdots \\ 1 \end{pmatrix} \\ 线性无关,所以在每个向量的前面添加r个分量而得到的n-r个\\ n维向量\xi_1, \xi_2, \cdots, \xi_{n-r}也线性无关.$
$其次, 证明(1)的任一解\\ x = \xi = \begin{pmatrix} \lambda_1 \\ \vdots \\ \lambda_r \\ \lambda_{r+1} \\ \vdots \\ \lambda_n \end{pmatrix} \\ 都可由\xi_1, \xi_2, \cdots, \xi_{n-r}线性表示.为此,做向量\\ \eta = \lambda_{r+1} \xi_1 + \lambda_{r+2} \xi_2 + \cdots + \lambda_n \xi_{n-r},\\ 由于\xi_1, \xi_2, \cdots, \xi_{n-r}是(1)的解,故\eta 也是(1)的解,\\ 比较\eta与\xi,知它们的后面n-r个分量对应相等,由于\\ 它们都满足方程组(3),从而知它们的前面r个分量亦必\\ 对应相等[方程组(3)表明任一解的前r个分量由后n-r\\ 个分量唯一决定],因此\eta = \xi, 即\\ \xi = \lambda_{r+1} \xi_1 + \lambda_{r+2} \xi_2 + \cdots + \lambda_n \xi_{n-r}, \\ 这样就证明了\xi_1, \xi_2, \cdots, \xi_{n-r}是解空间S的一个基,\\ 从而知解空间S的维数是n-r. \\ 根据以上证明,我们就得到如下定理.$

$定理8.n元齐次线性方程组A_{m \times n} x = 0 的全体解\\ 所构成的集合S是一个向量空间,当系数矩阵的秩\\ R(A_{m \times n}) = r时,解空间的维数为n-r.\\ 1)解空间S的基不是唯一的. \\ 2)解空间S的基又称为方程组(1)的基础解系. \\ 3)当R(A) = n时,方程组(1)只有零解,\\ 因而没有基础解系(此时解空间S只有一个零向量,\\ 为0维的向量空间).而当R(A)=r < n时,\\ 方程组(1)必含有n-r个向量的基础解系.\\ 设求得的基础解系为\\ \xi_1, \xi_2, \cdots, \xi_{n-r} \\ 则(1)的解可表示为\\ x = k_1 \xi_1 + k_2 \xi_2 + \cdots + k_{n-r}\xi_{n-r}, \\ 其中k_1, k_2, \cdots, k_{n-r}为任意实数.\\ 上式称为方程组(1)的通解. \\ 方程组(1)的解空间可表示为\\ S = \lbrace x = \ k_1 \xi_1 + k_2 \xi_2 + \cdots + k_{n-r} \xi_{n-r} | k_1, k_2, \cdots, k_{n-r} \in R \rbrace.$

$例1.求齐次线性方程组$
$\left \{ \begin{array}{l}x_1 + x_2 - x_3 - x_4 = 0 \\ 2x_1 - 5x_2 + 3x_3 + 2x_4 = 0 \\ 7x_1 - 7x_2 + 3x_3 + x_4 = 0 \end{array} \right.$
$的基础解系与通解.$
$解:对系数矩阵A作初等行变换,变为行最简形矩阵,有\\ A = \begin{pmatrix}1 & 1 & -1 & -1 \\ 2 & -5 & 3 & 2 \\ 7 & -7 & 3 & 1 \\ \end{pmatrix} \\ \sim \begin{pmatrix} 1 & 1 & -1 & -1 \\ 0 & -7 & 5 & 4 \\ 0 & -14 & 10 & 8 \\ \end{pmatrix} \sim \begin{pmatrix} 1 & 1 & -1 & -1 \\ 0 & -7 & 5 & 4 \\ 0 & 0 & 0 & 0 \\ \end{pmatrix} \\ \sim \begin{pmatrix} 1 & 0 & -\dfrac{2}{7} & -\dfrac{3}{7} \\ 0 & 1 & -\dfrac{5}{7} & -\dfrac{4}{7} \\ 0 & 0 & 0 & 0 \\ \end{pmatrix}$
$即\left \{ \begin{array}{l}x_1 = \dfrac{2}{7}x_3 + \dfrac{3}{7}x_4 \\ x_2 = \dfrac{5}{7}x_3 + \dfrac{4}{7}x_4 \\ x_3 = x_3 \\ x_4 = x_4 \end{array} \right.$
$亦即 \begin{pmatrix}x_1 \\ x_2 \\ x_3 \\ x_4 \end{pmatrix} = x_3\begin{pmatrix}\dfrac{2}{7} \\ \dfrac{5}{7} \\ 1 \\ 0 \end{pmatrix} + x_4\begin{pmatrix}\dfrac{3}{7} \\ \dfrac{4}{7} \\ 0 \\ 1 \end{pmatrix}$
$基础解系为:\\ \xi_1 = \begin{pmatrix}\dfrac{2}{7} \\ \dfrac{5}{7} \\ 1 \\ 0 \end{pmatrix}, \xi_2 = \begin{pmatrix}\dfrac{3}{7} \\ \dfrac{4}{7} \\ 0 \\ 1 \end{pmatrix}$
$方程组的通解为: \\ \begin{pmatrix}x_1 \\ x_2 \\ x_3 \\ x_4 \end{pmatrix} = c_1\begin{pmatrix}\dfrac{2}{7} \\ \dfrac{5}{7} \\ 1 \\ 0 \end{pmatrix} + c_2\begin{pmatrix}\dfrac{3}{7} \\ \dfrac{4}{7} \\ 0 \\ 1 \end{pmatrix}, (c_1, c_2 \in R).$

$例2.证明R(A^TA) = R(A).$
$证:设A为m \times n矩阵,x为n维列向量.\\ 若x满足Ax = 0,则有A^T(Ax) = 0,即\\ (A^TA)x = 0; \\ 若x满足(A^TA)x = 0,则x^T(A^TA)x = 0,即\\ (Ax)^T(Ax) = 0, \\ 从而推知\\ Ax = 0.\\ 综上可知方程组Ax = 0与(A^TA)x = 0同解,因此\\ R(A^TA) = R(A)$

$\color{blue}{二、非齐次线性方程组}$

$设有非齐次线性方程组$
$\left \{ \begin{array}{l} a_{11}x_1 + a_{12}x_2 + \cdots + a_{1n}x_n = b_1 \\ a_{21}x_1 + a_{22}x_2 + \cdots + a_{2n}x_n = b_2 \\ \cdots \quad \cdots \quad \cdots \quad \cdots \quad \cdots \quad \cdots \\ a_{m1}x_1 + a_{m2}x_2 + \cdots + a_{mn}x_n = b_m \\ \end{array} \right. \qquad (4)$
$记A = \begin{pmatrix} a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ \cdots & \cdots & \cdots & \cdots \\ a_{m1} & a_{m2} & \cdots & a_{mn} \\ \end{pmatrix}, x = \begin{pmatrix} x_1 \\ x_2 \\ \vdots \\ x_n \end{pmatrix}, b = \begin{pmatrix} b_1 \\ b_2 \\ \vdots \\ b_m \end{pmatrix}$
$则(4)式可写成向量方程: Ax = b \qquad (5)$

$性质3.设x = \eta_1, x = \eta_2为(5)的解,\\ 则x = \eta_1 - \eta_2为对应的齐次线性方程组\\ Ax = 0 \qquad (6)\\ 的解.$
$证:因为: \\ Ax = A(\eta_1 - \eta_2) = A\eta_1 - A\eta_2 = b - b = 0. \\ 故x = \eta_1 - \eta_2 是Ax = 0的解.$

$性质4.若x = \eta为方程(5)的解,x = k\xi是方程(6)的解,\\ 则x = \xi + \eta 仍是方程(5)的解.$
$证: 因为Ax = A(\xi + \eta) = A\xi + A\eta = 0 + b = b$
$由性质3可知,若求得(5)的一个特解\eta^{* },\\ 则(5)的任意解总可表示为x = \xi + \eta^{* }, \\ 其中x = \xi为方程Ax = 0的解,\\ 又若方程Ax = 0的通解为\\ x = k_1 \xi_1 + k_2 \xi_2 + \cdots + k_{n-r} \xi_{n-r} \\ 则方程Ax = b的任一解总可表示为\\ x = k_1 \xi_1 + k_2 \xi_2 + \cdots + k_{n-r}\xi_{n-r} + \eta^{*} \\ 而由性质4可知,对任何的实数k_1, k_2, \cdots, k_{n-r}\\ 上式总是方程Ax = b的解.于是方程Ax = b的通解为\\ x = k_1 \xi_1 + k_2 \xi_2 + \cdots + k_{n-r}\xi_{n-r} + \eta^{*}$

$例3.求解线性方程组$
$\left \{ \begin{array}{l}x_1 - x_2 - x_3 + x_4 = 0 \\ x_1 - x_2 + x_3 - 3x_4 = 1 \\ 2x_1 - 2x_2 - 4x_3 + 6x_4 = -1 \end{array} \right.$
$解:对增广矩阵B作初等行变换:\\ B = \begin{pmatrix} 1 & -1 & -1 & 1 & 0 \\ 1 & -1 & 1 & -3 & 1 \\ 2 & -2 & -4 & 6 & -1 \\ \end{pmatrix} \sim \begin{pmatrix} 1 & -1 & -1 & 1 & 0 \\ 0 & 0 & 2 & -4 & 1 \\ 0 & 0 & -2 & 4 & -1 \\ \end{pmatrix} \\ \sim \begin{pmatrix} 1 & -1 & -1 & 1 & 0 \\ 0 & 0 & 1 & -2 & \dfrac{1}{2} \\ 0 & 0 & 0 & 0 & 0 \\ \end{pmatrix} \sim \begin{pmatrix} 1 & -1 & 0 & -1 & \dfrac{1}{2} \\ 0 & 0 & 1 & -2 & \dfrac{1}{2} \\ 0 & 0 & 0 & 0 & 0 \\ \end{pmatrix}$
$可见R(A) = R(B) = 2 ,故方程组有解,等价方程组为:$
$\left \{ \begin{array}{l}x_1 = x_2 + x_4 + \dfrac{1}{2} \\ x_3 = 2x_4 + \dfrac{1}{2} \end{array} \right.$
$即\left \{ \begin{array}{l}x_1 = x_2 + x_4 + \dfrac{1}{2} \\ x_2 = x_2 \\ x_3 = 2x_4 + \dfrac{1}{2} \\ x_4 = x_4 \end{array} \right.$
$写成向量形式:$
$\begin{pmatrix}x_1 \\ x_2 \\ x_3 \\ x_4 \end{pmatrix} = c_1\begin{pmatrix}1 \\ 1 \\ 0 \\ 0 \end{pmatrix} + c_2\begin{pmatrix}1 \\ 0 \\ 2 \\ 1 \end{pmatrix} + \begin{pmatrix}\dfrac{1}{2} \\ 0 \\ \dfrac{1}{2} \\ 0\end{pmatrix}, (c_1, c_2 \in R)$
$于是对应的齐次线性方程组的基础解系为:$
$\xi_1 = \begin{pmatrix} 1 \\ 1 \\ 0 \\ 0 \end{pmatrix}, \xi_2 = \begin{pmatrix} 1 \\ 0 \\ 2 \\ 1 \end{pmatrix}.$
$原方程组的一个特解为:$
$\eta^{*} = \begin{pmatrix} \dfrac{1}{2} \\ 0 \\ \dfrac{1}{2} \\ 0 \end{pmatrix}$

$例4.设向量组\\ \alpha_1 = \begin{pmatrix} 1 \\ 1 \\ 1 \\ 3 \end{pmatrix}, \alpha_2 = \begin{pmatrix} -1 \\ -3 \\ 5 \\ 1 \end{pmatrix}, \alpha_3 = \begin{pmatrix} 3 \\ 2 \\ -1 \\ p+2 \end{pmatrix}, \alpha_4 = \begin{pmatrix} -2 \\ -6 \\ 10 \\ p \end{pmatrix}, \\ (1) p为和值时,该向量组线性无关?并在此时将向量\\ \alpha = \begin{pmatrix} 4, 1, 6, 10 \end{pmatrix} ^T用\alpha_1, \alpha_2, \alpha_3, \alpha_4线性表示; \\ (2) p为何值时，该向量组线性相关?\\ 并在此时求出它的秩和一个极大无关组.$
$解:\\ 对矩阵\begin{pmatrix} \alpha_1, \alpha_2, \alpha_3, \alpha_4, \alpha \end{pmatrix}作初等行变换\\ \begin{pmatrix} 1 & -1 & 3 & -2 & 4 \\ 1 & -3 & 2 & -6 & 1 \\ 1 & 5 & -1 & 10 & 6 \\ 3 & 1 & p + 2 & p & 10 \\ \end{pmatrix} \sim \begin{pmatrix} 1 & -1 & 3 & -2 & 4 \\ 0 & -2 & -1 & -4 & -3 \\ 0 & 6 & -4 & 12 & 2 \\ 0 & 4 & p -7 & p + 6 & -2 \\ \end{pmatrix} \sim \begin{pmatrix} 1 & -1 & 3 & -2 & 4 \\ 0 & -2 & -1 & -4 & -3 \\ 0 & 0 & -7 & 0 & -7 \\ 0 & 0 & p -9 & p -2 & -8 \\ \end{pmatrix} \sim \begin{pmatrix} 1 & -1 & 0 & -2 & 1 \\ 0 & 1 & 0 & 2 & 1 \\ 0 & 0 & 1 & 0 & 1 \\ 0 & 0 & p -9 & p -2 & -8 \\ \end{pmatrix} \sim \begin{pmatrix} 1 & 0 & 0 & 0 & 2 \\ 0 & 1 & 0 & 2 & 1 \\ 0 & 0 & 1 & 0 & 1 \\ 0 & 0 & 0 & p -2 & 1-p \\ \end{pmatrix}$
$显然,当p \neq 2时,向量组\alpha_1, \alpha_2, \alpha_3, \alpha_4线性无关,此时设\alpha = x_1 \alpha_1 + x_2 \alpha_2 + x_3 \alpha_3 + x_4 \alpha_4,于是该方程组有唯一解,将上面的阶梯形矩阵化成最简形矩阵.$
$\sim \begin{pmatrix} 1 & 0 & 0 & 0 & 2 \\ 0 & 1 & 0 & 0 & \dfrac{3p-4}{p -2} \\ 0 & 0 & 1 & 0 & 1 \\ 0 & 0 & 0 & 1 & \dfrac{1-p}{p -2} \\ \end{pmatrix}$
$所以, \alpha = 2\alpha_1 + \dfrac{3p-4}{p -2}\alpha_2 + \alpha_3 + \dfrac{1-p}{p -2}\alpha_4$
$(2)由阶梯形矩阵 \begin{pmatrix} 1 & 0 & 0 & 0 & 2 \\ 0 & 1 & 0 & 2 & 1 \\ 0 & 0 & 1 & 0 & 1 \\ 0 & 0 & 0 & p -2 & 1-p \\ \end{pmatrix} \\ 可以看出,当p = 2时,向量组\alpha_1, \alpha_2, \alpha_3, \alpha_4线性相关,此时向量组的秩为3,向量组\alpha_1, \alpha_2, \alpha_3或 \alpha_2, \alpha_3, \alpha_4为向量组\alpha_1, \alpha_2, \alpha_3, \alpha_4的极大无关组.$