不妨设\(A\)可经过行初等变换化为简化阶梯形矩阵(请思考为什么可以这样不妨设)
\begin{equation*}
\left( {\begin{array}{*{20}{r}}
1&0& \dots &0&{{c_{11}}}&{{c_{12}}}& \dots &{{c_{1,n - r}}}\\
0&1& \dots &0&{{c_{21}}}&{{c_{22}}}& \dots &{{c_{2,n - r}}}\\
\dots & \dots & \dots & \dots & \dots & \dots & \dots & \dots \\
0&0& \dots &1&{{c_{r1}}}&{{c_{r2}}}& \dots &{{c_{r,n - r}}}\\
0&0& \dots &0&0&0& \dots &0\\
\dots & \dots & \dots & \dots & \dots & \dots & \dots & \dots \\
0&0& \dots &0&0&0& \dots &0
\end{array}} \right)
\end{equation*}
对应的同解方程组为
\begin{equation*}
\left\{\begin{array}{ccccc}
x_1&&&&=-c_{11}x_{r+1}-c_{12}x_{r+2}-\dots -c_{1,n-r}x_n\\
&x_2&&&=-c_{21}x_{r+1}-c_{22}x_{r+2}-\dots -c_{2,n-r}x_n\\
\dots&\dots&\ddots&\dots&\dots\dots\dots\dots\dots\dots\dots\dots\dots\dots\dots\dots\dots\\
&&&x_r&=-c_{r1}x_{r+1}-c_{r2}x_{r+2}-\dots -c_{r,n-r}x_n
\end{array}\right.
\end{equation*}
用\(n-r\)组数 \((1,0,\dots,0)^T,(0,1,\dots,0)^T,\dots,(0,0,\dots,1)^T\) 代入自由未知量 \(({{x_{r + 1}}},{{x_{r + 2}}}, \dots ,{{x_n}})^T\),得到\(n-r\)个解向量
\begin{equation*}
\begin{array}{c}{\eta _1} = \left( {\begin{array}{*{20}{c}}
{ - {c_{11}}},&{ - {c_{21}}},& \dots ,&{ - {c_{r1}}},&1,&0,& \dots ,&0
\end{array}} \right)^T,\\{\eta _2} = \left( {\begin{array}{*{20}{c}}
{ - {c_{12}}},&{ - {c_{22}}},& \dots ,&{ - {c_{r2}}},&0,&1,& \dots ,&0
\end{array}} \right)^T, \\\vdots \\{\eta _{n - r}} = \left( {\begin{array}{*{20}{c}}
{ - {c_{1,n - r}}},&{ - {c_{2,n - r}}},& \dots ,&{ - {c_{r,n - r}}},&0,&0,& \dots ,&1
\end{array}} \right)^T.\end{array}
\end{equation*}
解向量 \(\eta_1,\dots ,\eta_{n-r}\) 满足:
\(\eta_1,\dots ,\eta_{n-r}\)线性无关;
事实上, \(\eta_1,\dots ,\eta_{n-r}\)的后\(n-r\)个分量构成的向量组
\begin{equation*}
(1,0,\dots,0)^T,(0,1,0,\dots,0)^T,\dots,(0,\dots,0,1)^T
\end{equation*}
线性无关,而 \(\eta_1,\dots ,\eta_{n-r}\) 是这\(n-r\)个向量的加长向量组, 所以, \(\eta_1,\dots ,\eta_{n-r}\) 线性无关。
任意解向量\(\eta=(c_1,\dots,c_n)^T\)都可由 \(\eta_1,\dots ,\eta_{n-r}\) 线性表出。
事实上,由 \(\eta_1,\eta_2,\dots ,\eta_{n-r}\) 是\(Ax=0\)的解,得
\begin{equation*}
c_{r+1}\eta_1+c_{r+2}\eta_2+\dots +c_n\eta_{n-r}=(*,\dots,*,c_{r+1},c_{r+2},\dots ,c_n)^T
\end{equation*}
也是\(Ax=0\)的解。它与\(\eta\)的最后\(n-r\)个分量相同,即自由未知量的值相同,所以它们是同一个解,即
\begin{equation*}
\eta=
c_{r+1}\eta_1+c_{r+2}\eta_2+\dots +c_n\eta_{n-r}.
\end{equation*}