文章目錄
  1. 1. 行列式按一行或一列的元素展开
  2. 2. 特殊矩阵的行列式
  3. 3. 习题

行列式按一行或一列的元素展开

有一种计算行列式的常用方法,基于逐次降低行列式的阶数.它需要用到子式$M_{ij}$和代数余子式$A_{ij}$的概念(见$\S 1$的定义).

定理1$\quad $设$A=(a_{ij} )\in M_n (\mathbb{R} )$.下述公式成立:

$$\det A=\sum_{i=1}^n (-1)^{i+j} a_{ij} M_{ij} =\sum_{i=1}^n a_{ij} A_{ij} \label{1} \tag{1} $$

(行列式按照第$j$列的元素展开);

$$\det A=\sum_{j=1}^n (-1)^{i+j} a_{ij} M_{ij} =\sum_{j=1}^n a_{ij} A_{ij} \label{2} \tag{2} $$

(行列式按照第$i$列的元素展开).

换言之,矩阵$A$的行列式等于某列(或某行)的一切元素与它们的代数余子式的乘积之和.

证明$\quad 1)$根据基本性质$\text{D1}$和$\text{D2}$,行列式(先对应于列,而后对应于行)满足一系列等式:

$$\begin{align}
\det A & =\begin{vmatrix} a_{11} & \cdots & a_{1j} & \cdots & a_{1n} \\ a_{21} & \cdots & a_{2j} & \cdots & a_{2n} \\ \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{n1} & \cdots & a_{nj} & \cdots & a_{nn} \end{vmatrix} \\
& =\begin{vmatrix} a_{11} & \cdots & a_{1j} & \cdots & a_{1n} \\ a_{21} & \cdots & 0 & \cdots & a_{2n} \\ \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{n1} & \cdots & 0 & \cdots & a_{nn} \end{vmatrix} + \begin{vmatrix} a_{11} & \cdots & 0 & \cdots & a_{1n} \\ a_{21} & \cdots & a_{2j} & \cdots & a_{2n} \\ \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{n1} & \cdots & 0 & \cdots & a_{nn} \end{vmatrix} \\
& +\cdots +\begin{vmatrix} a_{11} & \cdots & 0 & \cdots & a_{1n} \\ a_{21} & \cdots & 0 & \cdots & a_{2n} \\ \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{n1} & \cdots & a_{nj} & \cdots & a_{nn} \end{vmatrix} \\
& =\sum_{i=1}^n \begin{vmatrix} a_{11} & \cdots & a_{1,j-1} & 0 & a_{1,j+1} & \cdots & a_{1n} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{i1} & \cdots & a_{i,j-1} & a_{ij} & a_{i,j+1} & \cdots & a_{in} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{n1} & \cdots & a_{n,j-1} & 0 & a_{n,j+1} & \cdots & a_{nn} \end{vmatrix} \\
& = \sum_{i=1}^n (-1)^{j-1} \begin{vmatrix} 0 & a_{11} & \cdots & a_{1,j-1} & a_{1,j+1} & \cdots & a_{1n} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{ij} & a_{i1} & \cdots & a_{i,j-1} & a_{i,j+1} & \cdots & a_{in} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & a_{n1} & \cdots & a_{n,j-1} & a_{n,j+1} & \cdots & a_{nn} \end{vmatrix} \\
& =\sum_{i=1}^n (-1)^{(j-1)+(i-1)} \begin{vmatrix} a_{ij} & a_{i1} & \cdots & a_{i,j-1} & a_{i,j+1} & \cdots & a_{in} \\ 0 & a_{11} & \cdots & a_{1,j-1} & a_{1,j+1} & \cdots & a_{1n} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & a_{i-1,1} & \cdots & a_{i-1,j-1} & a_{i-1,j+1} & \cdots & a_{i-1,n} \\ 0 & a_{i+1,1} & \cdots & a_{i+1,j-1} & a_{i+1,j+1} & \cdots & a_{i+1,n} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & a_{n1} & \cdots & a_{n,j-1} & a_{n,j+1} & \cdots & a_{nn} \end{vmatrix} \\
& =\sum_{i=1}^n (-1)^{i+j} a_{ij} M_{ij} . \end{align}$$

最后一个等式基于对矩阵$A’$使用$\S 1$命题$2$,

$$A’ =\begin{pmatrix} a’_{11} & a’_{12} & \cdots & a’_{1n} \\ 0 & a’_{22} & \cdots & a’_{2n} \\ \cdots & \cdots & \cdots & \cdots \\ 0 & a’_{n2} & \cdots & a’_{nn} \end{pmatrix} ,$$

其中$a’_{11} =a_{ij} ,a’_{12} =a_{i1} ,\cdots ,a’_{1n} =a_{in} ,M’_{11} =M_{ij}$.回忆定义$A_{ij} =(-1)^{i+j} M_{ij}$.公式$\eqref{1}$得证.

$2)$令$\sideset{^t}{}A =(a’_{ji} ) ,a’_{ji} =a_{ij}$.我们再次指出,对应于$\det \sideset{^t}{}A $的元素$a’_{ji}$的子式为$M’_{ji} =M_{ij}$.用$1)$的结论,$\det A =\det \sideset{^t}{}A =\displaystyle \sum_{j=1}^n (-1)^{j+i} a’_{ji} M’_{ji} =\sum_{j=1}^n (-1)^{i+j} a_{ij} M_{ij}$,公式$\eqref{2}$得证.亦可直接引用$\S 1 $注记$3$简单地证明.

举两个例子用来说明行列式的上述性质.

例1$\quad $行列式

$$\Delta _n =\begin{vmatrix} 1 & 1 & \cdots & 1 \\ x_1 & x_2 & \cdots & x_n \\ x_1^2 & x_2^2 & \cdots & x_n^2 \\ \cdots & \cdots & \cdots & \cdots \\ x_1^{n-1} & x_2^{n-1} & \cdots & x_n^{n-1} \end{vmatrix} =\Delta (x_1 ,x_2 ,\cdots ,x_n )$$

叫作范德蒙德行列式,可按公式

$$\Delta _n =\prod_{1\leq i < j\leq n} (x_j -x_i ) \label{3} \tag{3} $$

计算,或更详细地写成

$$\Delta _n =(x_2 -x_1 )(x_3 -x_1 ) \cdots (x_n -x_1 )(x_3 -x_2 )\cdots (x_n -x_2 )\cdots (x_n -x_{n-1} )$$

(回顾$\S 1$习题$1$与这一公式的联系).特别地,当元素$x_1 ,\cdots ,x_n $两两不同时,范德蒙德行列式不等于零.它的这一性质经常被用到.根据$\S 1$定理$1$,我们也有

$$\Delta _n =\begin{vmatrix} 1 & x_1 & x_1^2 & \cdots & x_1^{n-1} \\ 1 & x_2 & x_2^2 & \cdots & x_2^{n-1} \\ \cdots & \cdots & \cdots & \cdots & \cdots \\ 1 & x_n & x_n^2 & \cdots & x_n^{n-1}\end{vmatrix} .$$

现在对$n$作归纳法来证明公式$\eqref{3}$.假设当$m < n$时,$\Delta _m $可由公式$\eqref{3}$计算,根据性质$\text{D7}$,对于每一个$i$,我们从行列式$\Delta _n $的第$i$行减去第$(i-1)$行乘以$x_1$:

$$\Delta _n =\begin{vmatrix} 1 & 1 & \cdots & 1 \\ 0 & x_2 -x_1 & \cdots & x_n -x_1 \\ 0 & x_2^2 -x_2 x_1 & \cdots & x_n^2 -x_n x_1 \\ \cdots & \cdots & \cdots & \cdots \\ 0 & x_2^{n-1} -x_2^{n-2} x_1 & \cdots & x_n^{n-1} -x_n^{n-2} x_1 \end{vmatrix} $$

现在将行列式按照第一列的元素展开,并在所得到的$(n-1)$阶行列式中将第$j$列$(j=1,2,\cdots ,n-1)$的公因子$x_{j+1} -x_1 $提到行列式符号的外面(行列式关于列的性质$\text{D1}$).我们得到了表达式

$$\begin{align}
\Delta _n & =(x_n -x_1 )(x_{n-1} -x_1 ) \cdots (x_2 -x_1 ) \begin{vmatrix} 1 & 1 & \cdots & 1 \\ x_2 & x_3 & \cdots & x_n \\ \cdots & \cdots & \cdots & \cdots \\ x_2^{n-2} & x_3^{n-2} & \cdots & x_n^{n-2} \end{vmatrix} \\
& =(x_n -x_1 )(x_{n-1} -x_1 )\cdots (x_2 -x_1 ) \cdot \Delta (x_2 ,x_3 ,\cdots ,x_n ),
\end{align}$$

由归纳条件对最后一个因子使用$\eqref{3}$

$$\Delta (x_2 ,\cdots ,x_n ) =\prod_{2\leq i < j\leq n } (x_j -x_i ).$$

例2$\quad $形如

$$A=\begin{pmatrix} 0 & a_{12} & a_{13} & \cdots & a_{1n} \\ -a_{12} & 0 & a_{23} & \cdots & a_{2n} \\ -a_{13} & -a_{23} & 0 & \cdots & a_{3n} \\ \cdots & \cdots & \cdots & \cdots & \cdots \\ -a_{1n} & -a_{2n} & -a_{3n} & \cdots & 0 \end{pmatrix} $$

的矩阵$A=(a_{ij} )$叫作斜对称的(其行列式亦称为斜对称的).换言之$\sideset{^t}{}A =-A$,考虑到$\S 1$的定理$1$,我们有

$$\det A=\det \sideset{^t}{}A =\det (-A) =(-1)^n \det A,$$

从而$[1+(-1)^{n-1}]\det A =0$.当$n$为奇数时,得到$\det A=0$,即任意奇数阶斜对称矩阵的行列式等于零.

特殊矩阵的行列式

如果矩阵$A$的元素中有较多的零,并且它们的位置“较好”,那么计算行列式$\det A$比较容易.在某些情况下这样的直觉可以导致准确的公式.例如我们知道(见$\S 1(6)$式)(上或下)三角矩阵的行列式等于主对角线上元素之积.另一个重要的特殊情况是

定理2$\quad $设$m+n$阶行列式$D$在前$n$列与后$m$行的交叉处为$0$,则有下述公式:

$$\begin{vmatrix} a_{11} & \cdots & a_{1n} & a_{1,n+1} & \cdots & a_{1,n+m} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{n1} & \cdots & a_{nn} & a_{n,n+1} & \cdots & a_{n,n+m} \\ 0 & \cdots & 0 & b_{11} & \cdots & b_{1m} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & \cdots & 0 & b_{m1} & \cdots & b_{mm} \end{vmatrix} = \begin{vmatrix} a_{11} & \cdots & a_{1n} \\ \cdots & \cdots & \cdots \\ a_{n1} & \cdots & a_{nn} \end{vmatrix} \cdot \begin{vmatrix} b_{11} & \cdots & b_{1m} \\ \cdots & \cdots & \cdots \\ b_{m1} & \cdots & b_{mm} \end{vmatrix}$$

(等式左边的行列式叫作准三角带有零角的行列式).

证明$\quad $首先固定$n(n+m)$个元素$a_{ij}$,并将行列式$D$看作元素$b_{kl}$的函数,$b_{kl}$组成一个$m$阶方阵$B$.因而所得函数可看作矩阵$B$的函数:$D=\mathcal{D} (B)$.

显然,行列式$D$关于后$m$行的多重线性性质和斜对称性,等价于$\mathcal{D} (B)$关于$B$的行有同样的性质.这就意味着将$\S 1$定理$3$应用于$\mathcal{D} (B)$是合理的,因此$\mathcal{D} (B)=\mathcal{D} (E)\det B$.根据函数$\mathcal{D} $的定义,我们有:

$$\mathcal{D} (E)=\begin{vmatrix} a_{11} & \cdots & a_{1n} & a_{1,n+1} & \cdots & a_{1,n+m} \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ a_{n1} & \cdots & a_{nn} & a_{n,n+1} & \cdots & a_{n,n+m} \\ 0 & \cdots & 0 & 1 & \cdots & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & \cdots & 0 & 0 & \cdots & 1 \end{vmatrix} .$$

按照最后一行展开$\mathcal{D} (E)$(见公式$\eqref{2}$),然后按倒数第二行展开等等.重复这一算法$m$次,我们得到$\mathcal{D} (E)=\det A$,其中

$$A=\begin{pmatrix} a_{11} & \cdots & a_{1n} \\ \cdots & \cdots & \cdots \\ a_{n1} & \cdots & a_{nn} \end{pmatrix} .$$

最后有$D=\mathcal{D} (B)=\det A\cdot \det B$.

使用新符号可将定理$2$的公式写成更紧凑的形式

$$\det \begin{pmatrix} A & C \\ 0 & B \end{pmatrix} =\det A \cdot \det B,\label{4} \tag{4} $$

其中$A,B$是方阵,而$0$和$C$是长方阵.

将$\S 1$的定理$1$和定理$2$结合起来,或使用定理$2$的证法直接论证,易得

$$\det \begin{pmatrix} A & 0 \\ C & B \end{pmatrix} =\det A \cdot \det B.$$

当我们试图写出行列式$\det \begin{pmatrix} C & A \\ B & 0 \end{pmatrix}$的表达式时,旋即得到了一个简单的反例$\begin{vmatrix} 0 & 1 \\ 1 & 0 \end{vmatrix} =-1$.问题在于符号.为了得到正确的结论,必须交换行或列,将矩阵$\begin{pmatrix} C & A \\ B & 0 \end{pmatrix}$化成$\begin{pmatrix} B & 0 \\ C & A \end{pmatrix}$或$\begin{pmatrix} A & C \\ 0 & B \end{pmatrix}$的形式.

更简单的方法基于已使用过若干次的$\S 1$定理$3$.事实上

$$\det \begin{pmatrix} C & A \\ B & 0 \end{pmatrix} =\det \begin{pmatrix} C & A \\ E & 0 \end{pmatrix} \cdot \det B.$$

其次运用公式$\eqref{2}$ $m$次,得到

$$\begin{align}
\det \begin{pmatrix} C & A \\ E_m & 0 \end{pmatrix} & =\begin{vmatrix} & & & a_{11} & \cdots & a_{1n} \\ & \ast & & \cdots & \cdots & \cdots \\ & & & a_{n1} & \cdots & a_{nn} \\ 1 & \cdots & 0 & 0 & \cdots & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & \cdots & 1 & 0 & \cdots & 0 \end{vmatrix} \\
& =(-1)^{(n+2)+(n+4)+\cdots +(n+2m)} \det A =(-1)^{mn} \det A.
\end{align}$$

最后,如果$A,B$分别是$n,m$阶方阵,则

$$\det \begin{pmatrix} C & A \\ B & 0 \end{pmatrix} =(-1)^{nm} \det A\cdot \det B.\label{5} \tag{5} $$

公式$\eqref{4}$和$\eqref{5}$包含在关于行列式展开的拉普拉斯一般定理中.但这个定理的使用面较窄,我们就不再讨论了,留给好学的读者作为下一节后面的练习.

关于矩阵的行列式在理论方面最重要的论断是

定理3$\quad $设$A$和$B$是$n$阶方阵,则

$$\det AB =\det A \cdot \det B.$$

证明$\quad $根据第$2$章$\S 3$公式$(7)$和$(9)$,可以将矩阵$(c_{ij} )=AB=(a_{ij} )(b_{ij} )$的系数用矩阵$A$和$B$的系数表示出来,第$i$行$(AB)_{(i)}$写成下式

$$(AB)_{(i)} =(A_{(i)} B^{(1)} ,A_{(i)} B^{(2)} ,\cdots ,A_{(i)} B^{(n)} ),$$

$$A_{(i)} B^{(j)} =\sum_{k=1}^n a_{ik} b^{kj} .$$

固定矩阵$B$,任取矩阵$A$,令

$${\mathcal{D}}_{B} (A) =\det AB.$$

我们证明,函数$\mathcal{D} ={\mathcal{D}}_{B}$满足$\S 1$定理$3$的条件$i),ii)$.事实上,将$A_{(s)}$与$A_{(t)}$交换位置.由于矩阵$AB$的第$s$行和第$t$行形如

$$(A_{(s)} B^{(1)} ,\cdots ,A_{(s)} B^{(n)} ),$$

$$(A_{(t)} B^{(1)} ,\cdots ,A_{(t)} B^{(n)} ),$$

那么$(AB)_{(s)}$与$(AB)_{(t)}$也交换了位置,于是根据定理$1$,

$$\begin{align}
& \mathcal{D} (\cdots ,A_{(s)} ,\cdots ,A_{(t)} ,\cdots ) \\
= & \mathcal{D} (A) \\
= & \det AB \\
= & \det[\cdots ,(AB)_{(s)} ,\cdots ,(AB)_{(t)} ,\cdots ] \\
= & -\det [\cdots ,(AB)_{(t)} ,\cdots ,(AB)_{(s)} ,\cdots ] \\
= & -\mathcal{D} (\cdots ,A_{(t)} ,\cdots ,A_{(s)} ,\cdots ).
\end{align}$$

进一步,$\det AB$是第$i$行$(AB)_{(i)}$的元素的线性函数:

$$\det AB =\lambda_1 A_{(i)} B^{(1)} +\lambda_2 A_{(i)} B^{(2)} +\cdots +\lambda_n A_{(i)} B^{(n)} .$$

所以$\mathcal{D} (A) =\displaystyle \sum_{j=1}^n \lambda_j \sum_{k=1}^n a_{ik} b_{kj} =\sum_{k=1}^n a_{ik} \sum_{j=1}^n \lambda_j b_{kj} =\sum_{k=1}^n \mu_k a_{ik}$,其中$\mu_k =\displaystyle \sum_{j=1}^n \lambda_j b_{kj}$是一个纯量,它不依赖于矩阵$A$的第$i$行的元素.

我们看到$\mathcal{D}$对矩阵$A$的第$i$行的元素是线性的.

这样,$\S 1$定理$3$的两个条件皆满足,故$\mathcal{D} (A) =\mathcal{D} (E)\cdot \det A$.但是根据定义$\mathcal{D} (E) =\det EB =\det B$.所求公式得证.

当$n=2$时,定理$3$容易直接验证,但当$n=3$时,直接计算已经非常困难.然而在一般情况下,可以采用迂回手段,直接运用性质$\text{D1} \sim \text{D2} $,或运用定理$2$(见习题$3$).

习题

$1$.整数$1798,2139,3255,4867$可以被$31$整除.不必计算,证明$4$阶行列式

$$\begin{vmatrix} 1 & 7 & 9 & 8 \\ 2 & 1 & 3 & 9 \\ 3 & 2 & 5 & 5 \\ 4 & 8 & 6 & 7 \\ \end{vmatrix}$$

也可以被$31$整除.

证明$\quad $第一列乘以$1000$,第二列乘以$100$,第三列乘以$10$,分别加到第四列,得

$$\begin{vmatrix} 1 & 7 & 9 & 8 \\ 2 & 1 & 3 & 9 \\ 3 & 2 & 5 & 5 \\ 4 & 8 & 6 & 7 \end{vmatrix} =\begin{vmatrix} 1 & 7 & 9 & 1798 \\ 2 & 1 & 3 & 2139 \\ 3 & 2 & 5 & 3255 \\ 4 & 8 & 6 & 4867 \end{vmatrix}$$

第四列的数都可以被$31$整除,于是第四列包含$31$的倍数,这样$4$阶行列式的值也包含$31$的倍数,故其也可以被$31$整除.

$2$.证明任意四阶斜对称行列式$\vert a_{ij} \vert$,其中$a_{ij} \in \mathbb{Z}$,是一个整数的平方.

注记$\quad $这对于任意阶的斜对称行列式都是对的.

证明$\quad $对于任意的四阶斜对称行列式$\vert a_{ij} \vert$来说,其中$a_{ij} \in \mathbb{Z}$,则

$$\begin{align}
& \begin{vmatrix} 0 & a_{12} & a_{13} & a_{14} \\ -a_{12} & 0 & a_{23} & a_{24} \\ -a_{13} & -a_{23} & 0 & a_{34} \\ -a_{14} & -a_{24} & -a_{34} & 0 \end{vmatrix} \\
= & -\begin{vmatrix} a_{12} & 0 & a_{13} & a_{14} \\ 0 & -a_{12} & a_{23} & a_{24} \\ -a_{23} & -a_{13} & 0 & a_{34} \\ -a_{24} & -a_{14} & -a_{34} & 0 \end{vmatrix} \\
= & -\begin{vmatrix} a_{12} & 0 & a_{13} & a_{14} \\ 0 & -a_{12} & a_{23} & a_{24} \\ 0 & -a_{13} & \dfrac{a_{23}}{a_{12}} a_{13} & \dfrac{a_{23}}{a_{12}} a_{14} +a_{34} \\ -a_{24} & -a_{14} & -a_{34} & 0 \end{vmatrix} \\
= & -\begin{vmatrix} a_{12} & 0 & a_{13} & a_{14} \\ 0 & -a_{12} & a_{23} & a_{24} \\ 0 & -a_{13} & \dfrac{a_{23}}{a_{12}} a_{13} & \dfrac{a_{23}}{a_{12}} a_{14} +a_{34} \\ 0 & -a_{14} & \dfrac{a_{24}}{a_{12}} a_{13} -a_{34} & \dfrac{a_{24}}{a_{12}} a_{14} \end{vmatrix} \\
= & -\begin{vmatrix} a_{12} & 0 & a_{13} & a_{14} \\ 0 & -a_{12} & a_{23} & a_{24} \\ 0 & 0 & 0 & \dfrac{a_{23}}{a_{12}} a_{14} +a_{34} -\dfrac{a_{13}}{a_{12}} a_{24} \\ 0 & -a_{14} & \dfrac{a_{24}}{a_{12}} a_{13} -a_{34} & \dfrac{a_{24}}{a_{12}} a_{14} \end{vmatrix} \\
= & -\begin{vmatrix} a_{12} & 0 & a_{13} & a_{14} \\ 0 & -a_{12} & a_{23} & a_{24} \\ 0 & 0 & 0 & \dfrac{a_{23}}{a_{12}} a_{14} +a_{34} -\dfrac{a_{13}}{a_{12}} a_{24} \\ 0 & 0 & \dfrac{a_{13}}{a_{12}} a_{24} -a_{34} -\dfrac{a_{23}}{a_{12}} a_{14} & 0 \end{vmatrix} \\
= & \begin{vmatrix} a_{12} & 0 & a_{13} & a_{14} \\ 0 & -a_{12} & a_{23} & a_{24} \\ 0 & 0 & \dfrac{a_{13}}{a_{12}} a_{24} -a_{34} -\dfrac{a_{23}}{a_{12}} a_{14} & 0 \\ 0 & 0 & 0 & \dfrac{a_{23}}{a_{12}} a_{14} +a_{34} -\dfrac{a_{13}}{a_{12}} a_{24} \end{vmatrix} \\
= & a_{12} \cdot (-a_{12} ) (\dfrac{a_{13}}{a_{12}} a_{24} -a_{34} -\dfrac{a_{23}}{a_{12}} a_{14} )(\dfrac{a_{23}}{a_{12}} a_{14} +a_{34} -\dfrac{a_{13}}{a_{12}} a_{24} ) \\
= & a_{12}^2 (\dfrac{a_{13}}{a_{12}} a_{24} -a_{34} -\dfrac{a_{23}}{a_{12}} a_{14} )^2
\end{align}$$

由上可知,任意四阶斜对称行列式$\vert a_{ij} \vert$,其中$a_{ij} \in \mathbb{Z}$,是一个整数的平方.

$3$.用下述方法证明$\det AB=\det A\cdot \det B$(定理$3$):令$2n\times 2n$阶辅助矩阵$C=\begin{pmatrix} E & B \\ -A & 0 \end{pmatrix}$运用Ⅱ型初等行变换将$C$化成

$$C’=\begin{pmatrix} E & B \\ 0 & AB \end{pmatrix} .$$

提示:利用等式$\det C=\det C’$和公式$\eqref{4}$,$\eqref{5}$.

同样地,证明也可以根据第$2$章$\S 3$习题$17,18$给出,注意到$\begin{pmatrix} E & A \\ 0 & E \end{pmatrix}$是上三角矩阵.

证明$\quad $令$2n\times 2n$阶辅助矩阵$C=\begin{pmatrix} E & B \\ -A & 0 \end{pmatrix}$,它的行列式等于

$$(-1)^{n^2} \det (-A) \cdot \det B=(-1)^{n^2+n} \det A \cdot \det B =\det A \cdot \det B.$$

为了与$\det AB $联系起来,我们把分块矩阵$C=\begin{pmatrix} E & B \\ -A & 0 \end{pmatrix}$的第一块行的$A$倍加到第二块行上,而这相当于在分块矩阵$C$的左边乘上一个相应的分块初等矩阵,即

$$\begin{pmatrix} E & 0 \\ A & E \end{pmatrix} \begin{pmatrix} E & B \\ -A & 0 \end{pmatrix} =\begin{pmatrix} E & B \\ 0 & AB \end{pmatrix} \label{6} \tag{6}$$

为了计算$\eqref{6}$式右端的分块矩阵的行列式,得用$\eqref{4}$式,得

$$\det \begin{pmatrix} E & B \\ 0 & AB \end{pmatrix} =\det E\cdot \det AB =\det AB \label{7} \tag{7} .$$

现在来计算$\eqref{6}$式左端的两个分块矩阵乘积的行列式.设$A=(a_{ij})$,因为

$$\begin{align}
\begin{pmatrix} E & 0 \\ 0 & E \end{pmatrix} & \xrightarrow[]{\begin{matrix} \displaystyle F_{n+1,1} (a_{11} ) \\ \cdots \\ \displaystyle F_{n+1,n} (a_{1n} ) \end{matrix} } \begin{pmatrix} E & 0 \\ \begin{matrix} a_{11} & a_{12} & \cdots & a_{1n} \\ 0 & 0 & \cdots & 0 \\ \vdots & \vdots & & \vdots \\ 0 & 0 & \cdots & 0 \end{matrix} & E \end{pmatrix} \\
& \xrightarrow[]{\begin{matrix} \displaystyle F_{n+2,1} (a_{21} ) \\ \cdots \\ \displaystyle F_{n+2,n} (a_{2n} ) \end{matrix} } \begin{pmatrix} E & 0 \\ \begin{matrix} a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ 0 & 0 & \cdots & 0 \\ \vdots & \vdots & & \vdots \\ 0 & 0 & \cdots & 0 \end{matrix} & E \end{pmatrix} \\
& \xrightarrow[]{Ⅱ型行变换} \cdots \xrightarrow[]{Ⅱ型行变换} \begin{pmatrix} E & 0 \\ A & E \end{pmatrix}\\
\end{align}$$

所以据初等行变换与初等矩阵的乘法的关系得

$$\begin{pmatrix} E & 0 \\ A & E \end{pmatrix} =P_t \cdots P_2 P_1 \begin{pmatrix} E & 0 \\ 0 & E \end{pmatrix} =P_t \cdots P_2 P_1 ,\\$$

其中$P_1 ,P_2 ,\cdots ,P_t $是与上述Ⅱ型初等行变换相应的初等矩阵,并且$t=n^2$.于是$\eqref{6}$式左端为

$$\begin{pmatrix} E & 0 \\ A & E \end{pmatrix} \begin{pmatrix} E & B \\ -A & 0 \end{pmatrix} =P_t \cdots P_2 P_1\begin{pmatrix} E & B \\ -A & 0 \end{pmatrix} .$$

因为Ⅱ型初等行变换不改变矩阵的行列式的值,所以

$$\det \begin{pmatrix} E & 0 \\ A & E \end{pmatrix} \begin{pmatrix} E & B \\ -A & 0 \end{pmatrix} =\det \begin{pmatrix} E & B \\ -A & 0 \end{pmatrix} =\det A\cdot \det B .\label{8} \tag{8} $$

由$\eqref{6}$,$\eqref{7}$,$\eqref{8}$式得

$$\det AB =\det A\cdot \det B.$$

$4$.(扎哈洛夫В.И.——图拉,$1984$)在平稳随机过程模型的研究中,出现了下述行列式:

$$\Delta _n (k_1 ,x_1 ;\cdots ;k_m ,x_m )=\begin{vmatrix} M_{k_1}^n (x_1 ) \\ M_{k_2}^n (x_2 ) \\ \cdots \\ M_{k_m}^n (x_m )\end{vmatrix} ,$$

其中$x_1 ,x_2 ,\cdots ,x_m$是未知量;$k_1 ,\cdots ,k_m$是自然数,$k_1 +k_2 +\cdots +k_m =n$;$M_{k}^n (x)$是$k\times n$阶矩阵,形如

$$M_k^n (x)=\begin{pmatrix} 1 & x & x^2 & \cdots & x^{n-1} \\ 0 & 1 & \begin{pmatrix} 2 \\ 1 \end{pmatrix} x & \cdots & \begin{pmatrix} n-1 \\ 1 \end{pmatrix} x^{n-2} \\ 0 & 0 & 1 & \cdots & \begin{pmatrix} n-1 \\ 2 \end{pmatrix} x^{n-3} \\ \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & \begin{pmatrix} n-1 \\ k-1 \end{pmatrix} x^{n-k} \end{pmatrix} .$$

证明

$$\Delta _n (k_1 ,x_1 ;\cdots ;k_m ,x_m ) =\prod_{1\leq j < i\leq m} (x_i -x_j )^{k_i k_j } .$$

提示:当$k_1 =\cdots =k_m =1$时,即当$m=n$时,得到范德蒙德行列式.

证明$\quad $将$\Delta _n (k_1 ,x_1 ;\cdots ;k_m ,x_m )$的第$n-1,n-2,\cdots ,2,1$列各乘以$-x_1 $,然后分别加到第$n,n-1,\cdots ,2$列,得

$$\begin{vmatrix} 1 & 0 & 0 & \cdots & 0 \\ 0 & 1 & (\displaystyle {2\choose 1} -{1\choose 1} )x_1 & \cdots & (\displaystyle {n-1\choose 1} -{n-2\choose 1} )x_1^{n-2} \\ 0 & 0 & 1 & \cdots & (\displaystyle {n-1\choose 2} -{n-2\choose 2} )x_1^{n-3} \\ \vdots & \vdots & \vdots & & \vdots \\ 0 & 0 & 0 & \cdots & (\displaystyle {n-1\choose k_1 -1 }-{n-2\choose k_1 -1 } )x_1^{n-k_1} \\ 1 & x_2 -x_1 & (x_2 -x_1 )x_2 & \cdots & (x_2 -x_1 ) x_2^{n-2} \\ 0 & 1 & \displaystyle {2\choose 1} x_2 -{1\choose 1} x_1 & \cdots & \displaystyle {n-1\choose 1} x_2^{n-2} -{n-2\choose 1} x_1 x_2^{n-3} \\ 0 & 0 & 1 & \cdots & \displaystyle {n-1\choose 2} x_2^{n-3} -{n-2\choose 2} x_1 x_2^{n-4} \\ \vdots & \vdots & \vdots & & \vdots \\ 0 & 0 & 0 & \cdots & \displaystyle {n-1\choose k_2 -1} x_2^{n-k_2}-{n-2\choose k_2 -1} x_1 x_2^{n-k_1 -1} \\ \vdots & \vdots & \vdots & & \vdots \\ 1 & x_m -x_1 & (x_m -x_1 )x_m & \cdots & (x_m -x_1 ) x_m^{n-2} \\ 0 & 1 & \displaystyle {2\choose 1} x_m -{1\choose 1} x_1 & \cdots & \displaystyle {n-1\choose 1} x_m^{n-2} -{n-2\choose 1} x_1 x_m^{n-3} \\ 0 & 0 & 1 & \cdots & \displaystyle {n-1\choose 2} x_m^{n-3} -{n-2\choose 2} x_1 x_m^{n-4} \\ \vdots & \vdots & & \vdots \\ 0 & 0 & 0 & \cdots & \displaystyle {n-1\choose k_m -1} x_m^{n-k_m}-{n-2\choose k_m -1} x_1 x_m^{n-k_m -1} \end{vmatrix} ,$$

并按第$1$行展开得到一个$n-1$阶行列式,设为$\Delta _{n-1}$,也即

$$\begin{align}
& \Delta _n (k_1 ,x_1 ;\cdots ;k_m ,x_m ) \\
= & \begin{vmatrix} 1 & (\displaystyle {2\choose 1} -{1\choose 1} )x_1 & \cdots & (\displaystyle {n-1\choose 1} -{n-2\choose 1} )x_1^{n-2} \\ 0 & 1 & \cdots & (\displaystyle {n-1\choose 2} -{n-2\choose 2} )x_1^{n-3} \\ \vdots & \vdots & & \vdots \\ 0 & 0 & \cdots & (\displaystyle {n-1\choose k_1 -1 }-{n-2\choose k_1 -1 } )x_1^{n-k_1} \\ x_2 -x_1 & (x_2 -x_1 )x_2 & \cdots & (x_2 -x_1 ) x_2^{n-2} \\ 1 & \displaystyle {2\choose 1} x_2 -{1\choose 1} x_1 & \cdots & \displaystyle {n-1\choose 1} x_2^{n-2} -{n-2\choose 1} x_1 x_2^{n-3} \\ 0 & 1 & \cdots & \displaystyle {n-1\choose 2} x_2^{n-3} -{n-2\choose 2} x_1 x_2^{n-4} \\ \vdots & \vdots & & \vdots \\ 0 & 0 & \cdots & \displaystyle {n-1\choose k_2 -1} x_2^{n-k_2}-{n-2\choose k_2 -1} x_1 x_2^{n-k_1 -1} \\ \vdots & \vdots & & \vdots \\ x_m -x_1 & (x_m -x_1 )x_m & \cdots & (x_m -x_1 ) x_m^{n-2} \\ 1 & \displaystyle {2\choose 1} x_m -{1\choose 1} x_1 & \cdots & \displaystyle {n-1\choose 1} x_m^{n-2} -{n-2\choose 1} x_1 x_m^{n-3} \\ 0 & 1 & \cdots & \displaystyle {n-1\choose 2} x_m^{n-3} -{n-2\choose 2} x_1 x_m^{n-4} \\ \vdots & \vdots & & \vdots \\ 0 & 0 & \cdots & \displaystyle {n-1\choose k_m -1} x_m^{n-k_m}-{n-2\choose k_m -1} x_1 x_m^{n-k_m -1} \end{vmatrix} \\
\overset{\triangle }{=} & \Delta _{n-1} (n-1阶)
\end{align}$$

显然,$\Delta _{n-1}$的第$1$行是$n-1$维向量$(1,x_1 ,\cdots ,x_1^{n-2} )$,$\Delta _{n-1}$的第$2$行是$(0,1,\cdots ,\displaystyle {n-2\choose 1} x_1^{n-3} )$,$\Delta _{n-1}$的第$3$行是$(0,0,\cdots ,\displaystyle {n-2\choose 2} x_1^{n-4} )$,$\cdots $,$\Delta _{n-1}$的第$k_1 -1$行是$(0,0,\cdots ,0,1,\displaystyle {n-2\choose k_1 -2 } x_1 ,\cdots ,{n-2\choose k_1 -2 } x_1^{n-k_1 } )$,从而知$\Delta _{n-1}$的前第$k_1 -1$行是$M_{k_1 -1}^{n-1} (x_1 )$,又易知$x_2 -x_1 $是$\Delta _{n-1}$的第$k_1 $行各元素的公因子,故第$k_1 $行可变成(将$x_2 -x_1 $提到行列式的外边相乘):$(1,x_2 ,x_2^2 ,\cdots ,x_2^{n-2} )$.

再把第$k_1 $行乘以$-1$加到第$k_1 +1$行上去,得第$k_1 +1$行为:

$$\begin{align}
& (0,(\displaystyle {2\choose 1} -{1\choose 0})x_2 -{1\choose 1} x_1 ,\cdots ,(\displaystyle {n-1\choose 1} -{n-2\choose 0})x_2^{n-2} -{n-2\choose 1} x_1 x_2^{n-3} ) \\
= & (0,x_2 -x_1 ,\cdots ,{n-2\choose 1} (x_2 -x_1 ) x_2^{n-3}),
\end{align}$$

它也有公因子$(x_2 -x_1 )$,也提到行列式外边相乘,这时$k_1 +1$行变为:$(0,1,\displaystyle {2\choose 1} x_2 ,\cdots ,{n-2\choose 1} x_2^{n-3})$.

再把第$k_1 +1$行乘以$-1$加到第$k_1 +2$行,于是第$k_1 +2$行变为:

$$\begin{align}
& (0,0,(\displaystyle {3\choose 2} -{2\choose 1})x_2 -{2\choose 2} x_1 ,\cdots ,(\displaystyle {n-1\choose 2} -{n-2\choose 1})x_2^{n-3} -{n-2\choose 2} x_1 x_2^{n-4} ) \\
= & (0,0,x_2 -x_1 ,\cdots ,{n-2\choose 2} (x_2 -x_1 ) x_2^{n-4}),
\end{align}$$

它也有公因子$(x_2 -x_1 )$,可提到行列式外边相乘,这时第$k_1 +2$行变为:$\displaystyle (0,0,1,\cdots ,{n-2\choose 2} x_2^{n-4})$.这样一直进行到第$k_1 +k_2 -1$行(共$k_2 $次)为:$\displaystyle (0,0,\cdots ,0,1,\cdots ,{n-2\choose k_2 -1} x_2^{n-k_2 -1})$,而提到行列式外面的因子为$(x_2 -x_1 )^{k_2}$.

同理,可依次得出$\Delta _{n-1}$的其余$n-k_1 -k_2 +1$行,最后得出

$$\Delta _n (k_1 ,x_1 ;\cdots ;k_m ,x_m ) =\prod_{i=2}^m (x_i -x_1 )^{k_i} \begin{vmatrix} M_{k_1 -1}^{n-1} (x_1 ) \\ M_{k_2}^{n-1} (x_2 ) \\ \cdots \\ M_{k_m}^{n-1} (x_m )\end{vmatrix} =\prod_{i=2}^m (x_i -x_1 )^{k_i} \Delta _{n-1} (k_1 -1,x_1 ;k_2 ,x_2;\cdots ;k_m ,x_m ),$$

即有

$$\Delta _n (k_1 ,x_1 ;\cdots ;k_m ,x_m ) =\prod_{i=2}^m (x_i -x_1 )^{k_i} \Delta _{n-1} (k_1 -1,x_1 ;k_2 ,x_2;\cdots ;k_m ,x_m ),$$

反复用上式即得

$$\begin{align}
& \Delta _n (k_1 ,x_1 ;\cdots ;k_m ,x_m ) \\
= & (x_2 -x_1 )^{k_2} \cdots (x_m -x_1 )^{k_m } \Delta _{n-1} (k_1 -1,x_1 ;k_2 ,x_2;\cdots ;k_m ,x_m ) \\
= & (x_2 -x_1 )^{2k_2} \cdots (x_m -x_1 )^{2k_m } \Delta _{n-2} (k_1 -2,x_1 ;k_2 ,x_2;\cdots ;k_m ,x_m ) \\
\cdots & =(x_2 -x_1 )^{k_1 k_2} \cdots (x_m -x_1 )^{k_1 k_m } \Delta _{n-k_1 } (k_2 ,x_2;\cdots ;k_m ,x_m ) \\
= & \prod_{2\leq i\leq m} (x_i -x_1 )^{k_1 k_i }\Delta _{n-k_1 } (k_2 ,x_2;\cdots ;k_m ,x_m ) \\
\cdots & = \prod_{2\leq i\leq m} (x_i -x_1 )^{k_1 k_i } \prod_{3\leq i\leq m} (x_i -x_2 )^{k_2 k_i } \Delta _{n-k_1 -k_2 } (k_3 ,x_3;\cdots ;k_m ,x_m ) \\
= & \prod_{2\leq i\leq m} (x_i -x_1 )^{k_1 k_i } \prod_{3\leq i\leq m} (x_i -x_2 )^{k_2 k_i } \prod_{4\leq i\leq m} (x_i -x_3 )^{k_3 k_i }\Delta _{n-k_1 -k_2 -k_3 } (k_4 ,x_4 ,\cdots ;k_m ,x_m ) \\
\cdots & =\prod_{2\leq i\leq m} (x_i -x_1 )^{k_1 k_i } \prod_{3\leq i\leq m} (x_i -x_2 )^{k_2 k_i } \cdots (x_m -x_{m-1} )^{k_m k_{m-1} } \\
= & \prod_{1\leq j < i\leq m} (x_i -x_j )^{k_i k_j } ,
\end{align}$$

得证.

参考:普丰山,陈军.广义范德蒙行列式及其应用.河南科学:第24卷第5期,2006年10月.

$5$.证明

$$\begin{align}
B_n (s,t) & =\begin{vmatrix} \displaystyle {s\choose t} & \displaystyle {s\choose t+1} & \cdots & \displaystyle {s\choose t+n-1} \\ \displaystyle {s+1\choose t} & \displaystyle {s+1\choose t+1} & \cdots & \displaystyle {s+1\choose t+n-1} \\ \cdots & \cdots & \cdots & \cdots \\ \displaystyle {s+n-1\choose t} & \displaystyle {s+n-1\choose t+1} & \cdots & \displaystyle {s+n-1\choose t+n-1} \end{vmatrix} \\
& =\dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t\choose n} }{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n\choose n} } .
\end{align}$$

提示:逐次从第$k$行中提出$(s+k-1),k=1,2,\cdots ,n$,然后从第$l$列中提出$\dfrac{1}{t+l-1}$,$l=1,2,\cdots ,n$.直到第一列中除了$1$之外不再含有其他元素.

证明$\quad $根据题意,

$$\begin{align}
B_n (s,t) & =\begin{vmatrix} \dfrac{s!}{(s-t)!t!} & \dfrac{s!}{(s-t-1)!(t+1)!} & \cdots & \dfrac{s!}{(s-t-n+1)!(t+n-1)!} \\ \dfrac{(s+1)!}{(s+1-t)!t!} & \dfrac{(s+1)!}{(s-t)!(t+1)!} & \cdots & \dfrac{(s+1)!}{(s-t-n+2)!(t+n-1)!} \\ \vdots & \vdots & & \vdots \\ \dfrac{(s+n-1)!}{(s+n-1-t)!t!} & \dfrac{(s+n-1)!}{(s+n-t-2)!(t+1)!} & \cdots & \dfrac{(s+n-1)!}{(s-t)!(t+n-1)!} \end{vmatrix}
\end{align}$$

当第$i=1,2,\cdots ,t$次时,从第$k$行中提出$(s+k-i),k=1,2,\cdots ,n$,然后从第$l$列中提出$\dfrac{1}{t+l-i}$,$l=1,2,\cdots ,n$,得到

$$\begin{align}
& B_n (s,t) \\
= & \dfrac{s\cdots (s+n-1)}{t\cdots (t+n-1)} \\
& \cdot \begin{vmatrix} \dfrac{(s-1)!}{(s-t)!(t-1)!} & \dfrac{(s-1)!}{(s-t-1)!t!} & \cdots & \dfrac{(s-1)!}{(s-t-n+1)!(t+n-2)!} \\ \dfrac{s!}{(s+1-t)!(t-1)!} & \dfrac{s!}{(s-t)!t!} & \cdots & \dfrac{s!}{(s-t-n+2)!(t+n-2)!} \\ \vdots & \vdots & & \vdots \\ \dfrac{(s+n-2)!}{(s+n-1-t)!(t-1)!} & \dfrac{(s+n-2)!}{(s+n-t-2)!t!} & \cdots & \dfrac{(s+n-2)!}{(s-t)!(t+n-2)!} \end{vmatrix} \\
= & \dfrac{s\cdots (s+n-1)}{t\cdots (t+n-1)} \dfrac{(s-1)\cdots (s+n-2)}{(t-1)\cdots (t+n-2)} \\
& \cdot \begin{vmatrix} \dfrac{(s-2)!}{(s-t)!(t-2)!} & \dfrac{(s-2)!}{(s-t-1)!(t-1)!} & \cdots & \dfrac{(s-2)!}{(s-t-n+1)!(t+n-3)!} \\ \dfrac{(s-1)!}{(s+1-t)!(t-2)!} & \dfrac{(s-1)!}{(s-t)!(t-1)!} & \cdots & \dfrac{(s-1)!}{(s-t-n+2)!(t+n-3)!} \\ \vdots & \vdots & & \vdots \\ \dfrac{(s+n-3)!}{(s+n-1-t)!(t-2)!} & \dfrac{(s+n-3)!}{(s+n-t-2)!(t-1)!} & \cdots & \dfrac{(s+n-3)!}{(s-t)!(t+n-3)!} \end{vmatrix} \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} }{\displaystyle {n+t-1\choose n} {n+t-2\choose n}} \\
& \cdot \begin{vmatrix} \dfrac{(s-2)!}{(s-t)!(t-2)!} & \dfrac{(s-2)!}{(s-t-1)!(t-1)!} & \cdots & \dfrac{(s-2)!}{(s-t-n+1)!(t+n-3)!} \\ \dfrac{(s-1)!}{(s+1-t)!(t-2)!} & \dfrac{(s-1)!}{(s-t)!(t-1)!} & \cdots & \dfrac{(s-1)!}{(s-t-n+2)!(t+n-3)!} \\ \vdots & \vdots & & \vdots \\ \dfrac{(s+n-3)!}{(s+n-1-t)!(t-2)!} & \dfrac{(s+n-3)!}{(s+n-t-2)!(t-1)!} & \cdots & \dfrac{(s+n-3)!}{(s-t)!(t+n-3)!} \end{vmatrix} \\
= & \cdots \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t+1\choose n} }{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n+1\choose n} } \\
& \cdot \begin{vmatrix} \dfrac{(s-t+1)!}{(s-t)!1!} & \dfrac{(s-t+1)!}{(s-t-1)!2!} & \cdots & \dfrac{(s-t+1)!}{(s-t-n+1)!n!} \\ \dfrac{(s-t+2)!}{(s+1-t)!1!} & \dfrac{(s-t+2)!}{(s-t)!2!} & \cdots & \dfrac{(s-t+2)!}{(s-t-n+2)!n!} \\ \vdots & \vdots & & \vdots \\ \dfrac{(s+n-t)!}{(s+n-1-t)!1!} & \dfrac{(s+n-t)!}{(s+n-t-2)!2!} & \cdots & \dfrac{(s+n-t)!}{(s-t)!n!} \end{vmatrix} \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t+1\choose n} {n+s-t\choose n}}{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n+1\choose n} {n\choose n} } \\
& \cdot \begin{vmatrix} \dfrac{(s-t)!}{(s-t)!} & \dfrac{(s-t)!}{(s-t-1)!1!} & \cdots & \dfrac{(s-t)!}{(s-t-n+1)!(n-1)!} \\ \dfrac{(s-t+1)!}{(s+1-t)!} & \dfrac{(s-t+1)!}{(s-t)!1!} & \cdots & \dfrac{(s-t+1)!}{(s-t-n+2)!(n-1)!} \\ \vdots & \vdots & & \vdots \\ \dfrac{(s+n-t-1)!}{(s+n-1-t)!} & \dfrac{(s+n-t-1)!}{(s+n-t-2)!1!} & \cdots & \dfrac{(s+n-t-1)!}{(s-t)!(n-1)!} \end{vmatrix} \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t+1\choose n} {n+s-t\choose n}}{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n+1\choose n} {n\choose n} } \\
& \cdot \begin{vmatrix} 1 & s-t & \displaystyle {s-t\choose 2} & \cdots & \displaystyle {s-t\choose n-1} \\ 1 & s-t+1 & \displaystyle {s-t+1\choose 2} & \cdots & \displaystyle {s-t+1\choose n-1} \\ \vdots & \vdots & \vdots & & \vdots \\ 1 & s-t+n-2 & \displaystyle {s-t+n-2\choose 2} & \cdots & \displaystyle {s-t+n-2\choose n-1} \\ 1 & s-t+n-1 & \displaystyle {s-t+n-1\choose 2} & \cdots & \displaystyle {s-t+n-1\choose n-1} \end{vmatrix} \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t+1\choose n} {n+s-t\choose n}}{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n+1\choose n} {n\choose n} } \\
& \cdot \begin{vmatrix} 1 & s-t & \displaystyle {s-t\choose 2} & \cdots & \displaystyle {s-t\choose n-1} \\ 0 & 1 & \displaystyle {s-t\choose 1} & \cdots & \displaystyle {s-t\choose n-2} \\ \vdots & \vdots & \vdots & & \vdots \\ 0 & 1 & \displaystyle {s-t+n-3\choose 1} & \cdots & \displaystyle {s-t+n-3\choose n-2} \\ 0 & 1 & \displaystyle {s-t+n-2\choose 1} & \cdots & \displaystyle {s-t+n-2\choose n-2} \end{vmatrix} \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t+1\choose n} {n+s-t\choose n}}{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n+1\choose n} {n\choose n} } \\
& \cdot \begin{vmatrix} 1 & \displaystyle {s-t\choose 1} & \cdots & \displaystyle {s-t\choose n-2} \\ \vdots & \vdots & & \vdots \\ 1 & \displaystyle {s-t+n-3\choose 1} & \cdots & \displaystyle {s-t+n-3\choose n-2} \\ 1 & \displaystyle {s-t+n-2\choose 1} & \cdots & \displaystyle {s-t+n-2\choose n-2} \end{vmatrix} \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t+1\choose n} {n+s-t\choose n}}{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n+1\choose n} {n\choose n} } \\
& \cdot \begin{vmatrix} 1 & \displaystyle {s-t\choose 1} & \cdots & \displaystyle {s-t\choose n-2} \\ 0 & 1 & \cdots & \displaystyle {s-t-1\choose n-3} \\ \vdots & \vdots & & \vdots \\ 0 & 1 & \cdots & \displaystyle {s-t+n-4\choose n-3} \\ 0 & 1 & \cdots & \displaystyle {s-t+n-3\choose n-3} \end{vmatrix} \\
= & \cdots \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t+1\choose n} {n+s-t\choose n}}{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n+1\choose n} {n\choose n} } \cdot \begin{vmatrix} 1 \end{vmatrix} \\
= & \dfrac{\displaystyle {n+s-1\choose n} {n+s-2\choose n} \cdots {n+s-t+1\choose n} {n+s-t\choose n}}{\displaystyle {n+t-1\choose n} {n+t-2\choose n} \cdots {n+1\choose n} {n\choose n} }
\end{align}$$

得证.

$6$.设

$$C_n (\lambda _1 ,\cdots ,\lambda _n )=\begin{pmatrix} \lambda _1 & 1 & 0 & \cdots & 0 & 0 & 0 \\ -1 & \lambda _2 & 1 & \cdots & 0 & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & \lambda _{n-2} & 1 & 0 \\ 0 & 0 & 0 & \cdots & -1 & \lambda _{n-1} & 1 \\ 0 & 0 & 0 & \cdots & 0 & -1 & \lambda _n \end{pmatrix} .$$

证明$\det C_n =\lambda _n C_{n-1} +\det C_{n-2}$.当$\lambda _1 =\lambda _2 =\cdots =\lambda _n =1$时,求出数值$\det C_n $.

提示:用第$2$章$\S 3$第$3$段的例$3$,并注意到事实$\det C_n (1,\cdots ,1)=(-1)^n \det C_n (-1,\cdots ,-1)$.

证明:$(1)\;$矩阵$C_n $的行列式从最后一行展开,得

$$\begin{align}
& \begin{vmatrix} \lambda _1 & 1 & 0 & \cdots & 0 & 0 & 0 \\ -1 & \lambda _2 & 1 & \cdots & 0 & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & \lambda _{n-2} & 1 & 0 \\ 0 & 0 & 0 & \cdots & -1 & \lambda _{n-1} & 1 \\ 0 & 0 & 0 & \cdots & 0 & -1 & \lambda _n \end{vmatrix} \\
= & \lambda _n \begin{vmatrix} \lambda _1 & 1 & 0 & \cdots & 0 & 0 \\ -1 & \lambda _2 & 1 & \cdots & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & \lambda _{n-2} & 1 \\ 0 & 0 & 0 & \cdots & -1 & \lambda _{n-1} \end{vmatrix} +(-1)^{n+(n-1)}\cdot (-1) \begin{vmatrix} \lambda _1 & 1 & 0 & \cdots & 0 & 0 \\ -1 & \lambda _2 & 1 & \cdots & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & \lambda _{n-2} & 0 \\ 0 & 0 & 0 & \cdots & -1 & 1 \end{vmatrix} \\
= & \lambda _n \begin{vmatrix} \lambda _1 & 1 & 0 & \cdots & 0 & 0 \\ -1 & \lambda _2 & 1 & \cdots & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & \lambda _{n-2} & 1 \\ 0 & 0 & 0 & \cdots & -1 & \lambda _{n-1} \end{vmatrix} +(-1)^{n+(n-1)}\cdot (-1) \begin{vmatrix} \lambda _1 & 1 & 0 & \cdots & 0 & 0 \\ -1 & \lambda _2 & 1 & \cdots & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & \lambda _{n-3} & 1 \\ 0 & 0 & 0 & \cdots & -1 & \lambda _{n-2} \end{vmatrix} ,\\
\end{align}$$

由上可知,$\det C_n =\lambda _n C_{n-1} +\det C_{n-2}$.

$(2)\;$当$\lambda _1 =\lambda _2 =\cdots =\lambda _n =1$时,令$\alpha +\beta =1$,$\alpha \beta =-1$,则$\alpha ,\beta $是方程

$$x^2-x-1 =0$$

的两个根:

$$\alpha =\dfrac{1+\sqrt{5} }{2} ,\beta =\dfrac{1-\sqrt{5} }{2} .$$

于是

$$\begin{align}
& \det C_n (1,\cdots ,1) \\
= & \det C_n \\
= & \begin{vmatrix} 1 & 1 & 0 & \cdots & 0 & 0 & 0 \\ -1 & 1 & 1 & \cdots & 0 & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & 1 & 1 & 0 \\ 0 & 0 & 0 & \cdots & -1 & 1 & 1 \\ 0 & 0 & 0 & \cdots & 0 & -1 & 1 \end{vmatrix} \\
= & \begin{vmatrix} \alpha +\beta & 1 & 0 & \cdots & 0 & 0 & 0 \\ \alpha \beta & \alpha +\beta & 1 & \cdots & 0 & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & \cdots & \alpha +\beta & 1 & 0 \\ 0 & 0 & 0 & \cdots & \alpha \beta & \alpha +\beta & 1 \\ 0 & 0 & 0 & \cdots & 0 & \alpha \beta & \alpha +\beta \end{vmatrix} .\\
\end{align}$$

当$n\geq 3$时,按第$1$行展开,得

$$\begin{align}
& \det C_n \\
= & (\alpha +\beta )\det C_{n-1} +(-1)^{1+2} 1 \cdot \alpha \beta \cdot \det C_{n-2} \\
= & (\alpha +\beta )\det C_{n-1} -\alpha \beta \det C_{n-2} .\label{9} \tag{9} \\
\end{align}$$

由$\eqref{9}$式得

$$\det C_n -\alpha \det C_{n-1} =\beta (\det C_{n-1} -\alpha \det C_{n-2} ).$$

于是$\det C_2 -\alpha \det C_1 ,\det C_3 -\alpha \det C_2 ,\cdots ,\det C_n -\alpha \det C_{n-1}$是公比为$\beta $的等比数列.从而

$$\det C_n -\alpha \det C_{n-1} =(\det C_2 -\alpha \det C_1 )\beta ^{n-2} .$$

由于$\det C_1 =\vert \alpha +\beta \vert =\alpha +\beta $,

$$\det C_2 =\begin{vmatrix} \alpha +\beta & 1 \\ \alpha \beta & \alpha +\beta \end{vmatrix} =(\alpha +\beta )^2 -\alpha \beta =\alpha ^2 +\alpha \beta +\beta ^2.$$

因此$\det C_2 -\alpha \det C_1 =\beta ^2$.从而

$$\det C_n -\alpha C_{n-1} =\beta ^n .\label{10} \tag{10} $$

由$\eqref{9}$式又可得出

$$\det C_n -\beta \det C_{n-1} =\alpha (\det C_{n-1} -\beta \det C_{n-2} ).$$

同理可得

$$\det C_n -\beta \det C_{n-1} =\alpha ^n .\label{11} \tag{11} $$

联立$\eqref{10}$,$\eqref{11} $式,解得

$$\det C_n =\dfrac{\alpha ^{n+1} -\beta ^{n+1}}{\alpha -\beta } .\\label{12} \tag{12} $$

当$n=1,2$时,公式$\eqref{12}$也成立.

$$\det C_n =\dfrac{\left( \dfrac{1+\sqrt{5}}{2} \right) ^{n+1} -\left( \dfrac{1-\sqrt{5}}{2} \right) ^{n+1}}{\sqrt{5}} .$$

$7$.证明$n\times n$矩阵

$$A_n =\begin{pmatrix} 2 & -1 & 0 & 0 & \cdots & 0 & 0 & 0 \\ -1 & 2 & -1 & 0 & \cdots & 0 & 0 & 0 \\ 0 & -1 & 2 & -1 & \cdots & 0 & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & 0 & \cdots & -1 & 2 & -1 \\ 0 & 0 & 0 & 0 & \cdots & 0 & -1 & 2 \end{pmatrix} $$

的行列式等于$n+1$.

证明$\quad n=1$时,$A_1 =\vert 2\vert =2$.下面设$n > 1$,把第$2,3,\cdots ,n$列都加到第$1$列上,然后按第$1$列展开:

$$\begin{align}
& \det A_n \\
= & \begin{vmatrix} 2 & -1 & 0 & 0 & \cdots & 0 & 0 & 0 \\ 0 & 2 & -1 & 0 & \cdots & 0 & 0 & 0 \\ 0 & -1 & 2 & -1 & \cdots & 0 & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ 0 & 0 & 0 & 0 & \cdots & -1 & 2 & -1 \\ 1 & 0 & 0 & 0 & \cdots & 0 & -1 & 2 \end{vmatrix} \\
= & 1\cdot \det A_{n-1} +(-1)^{n+1} 1\cdot (-1)^{n-1} \\
= & \det A_{n-1} +1.
\end{align}$$

由此看出,$\det A_1 ,\det A_2 ,\cdots ,\det A_n $是首项为$2$,公差为$1$的等差数列.

因此

$$\det A_n =2+(n-1)\cdot 1=n+1.$$

$8$.设$A,B$是任意$n$阶方阵.证明

$$\det \begin{pmatrix} A & B \\ B & A \end{pmatrix} =\det (A+B) \cdot \det (A-B) .$$

证明

$$\begin{align}
\begin{pmatrix} A & B \\ B & A \end{pmatrix} & \xrightarrow[]{F_{1,2} (1)} \begin{pmatrix} A+B & B+A \\ B & A \end{pmatrix} \\
& \xrightarrow[F_{2,1} (-1)]{} \begin{pmatrix} A+B & 0 \\ B & A-B \end{pmatrix} .
\end{align}$$

于是

$$\begin{pmatrix} E & E \\ 0 & E \end{pmatrix} \begin{pmatrix} A & B \\ B & A \end{pmatrix} \begin{pmatrix} E & -E \\ 0 & E \end{pmatrix} =\begin{pmatrix} A+B & 0 \\ B & A-B \end{pmatrix} .$$

两边取行列式,得

$$\det \begin{pmatrix} A & B \\ B & A \end{pmatrix} =\det (A+B) \cdot \det (A-B) .$$

$9$.设$X$是$n\times k$矩阵而$Y$是$k\times n$矩阵.证明

$$\det (E_n +XY)=\det (E_k +YX).$$

提示:利用关系式

$$\begin{pmatrix} E_k +YX & 0 \\ X & E_n \end{pmatrix} \begin{pmatrix} E_k & Y \\ 0 & E_n \end{pmatrix}=\begin{pmatrix} E_k & Y \\ 0 & E_n \end{pmatrix} \begin{pmatrix} E_k & 0 \\ X & E_n +XY \end{pmatrix}.$$

证明$\quad $引入矩阵$\begin{pmatrix} E_k +YX & Y+YXY \\ X & E_n +XY \end{pmatrix}$.

因为

$$\begin{pmatrix} E_k +YX & 0 \\ X & E_n \end{pmatrix} \begin{pmatrix} E_k & Y \\ 0 & E_n \end{pmatrix} =\begin{pmatrix} E_k +YX & Y+YXY \\ X & E_n +XY \end{pmatrix} ,$$

$$\begin{pmatrix} E_k & Y \\ 0 & E_n \end{pmatrix} \begin{pmatrix} E_k & 0 \\ X & E_n +XY \end{pmatrix} =\begin{pmatrix} E_k +YX & Y+YXY \\ X & E_n +XY \end{pmatrix} .$$

故有

$$\begin{pmatrix} E_k +YX & 0 \\ X & E_n \end{pmatrix} \begin{pmatrix} E_k & Y \\ 0 & E_n \end{pmatrix}=\begin{pmatrix} E_k & Y \\ 0 & E_n \end{pmatrix} \begin{pmatrix} E_k & 0 \\ X & E_n +XY \end{pmatrix}.$$

于是

$$\det \begin{pmatrix} E_k +YX & 0 \\ X & E_n \end{pmatrix} \det \begin{pmatrix} E_k & Y \\ 0 & E_n \end{pmatrix}=\det \begin{pmatrix} E_k & Y \\ 0 & E_n \end{pmatrix} \det \begin{pmatrix} E_k & 0 \\ X & E_n +XY \end{pmatrix} ,$$

$$\det (E_k +YX) \cdot \det E_n \cdot \det E_k \cdot \det E_n =\det E_k \cdot \det E_n \cdot \det E_k \cdot \det (E_n +XY) ,$$

$$\det (E_n +XY)=\det (E_k +YX).$$

文章目錄
  1. 1. 行列式按一行或一列的元素展开
  2. 2. 特殊矩阵的行列式
  3. 3. 习题