行列式

Determinant & Computations

我们将从行列式的第二公理化构造开始阐述行列式的定义和一些计算性质。最后我们讲解伴随矩阵法求逆矩阵。

行列式的第二公理化构造

我们想通过方程组的系数判断方程组是否有解,核心就是使用高斯消元法,由此,我们假设行列式是由方程组系数构成的某一个运算,该运算满足如下性质:

  • 将方程的某一行乘\(k\)倍,加到某一行上,该运算的值不变。
  • 将方程某一行同乘\(k\)倍,该运算的值也乘\(k\)倍。
  • 最终消元后的结果,对其执行运算的结果为1。

用数学的话说,就是,定义函数\(D:M_n(R) \to R\),它将一个\(n\)阶实数矩阵转化成实数。该函数满足三条性质:

  • \(D(\cdots, A_{(i)} + A_{(j)},\cdots, A_{(j)},\cdots) = D(\cdots,A_{(i)},\cdots,A_{(j)},\cdots)\)
  • \(D(\cdots,kA_{(i)},\cdots) = kD(\cdots,A_{(i)},\cdots)\)
  • \(D(E) = 1\),\(E\)是\(n\)阶单位矩阵。

接下来我们可以通过这三条性质证明\(D\)就是行列式运算。

证明

首先我们证明互换矩阵的两行,函数\(D\)变号。

\[\begin{align} D(\cdots, A_{(i)},\cdots, A_{(j)},\cdots) =& D(\cdots, A_{(i)}+A_{(j)},\cdots,A_{(j)},\cdots)\\ =& -D(\cdots, A_{(i)} + A_{(j)},\cdots, -A_{(j)},\cdots)\\ =& -D(\cdots, A_{(i)} + A_{(j)},\cdots, A_{(i)},\cdots)\\ =& -D(\cdots, A_{(j)}, \cdots, A_{(i)},\cdots) \end{align} \]

接下来证明\(D(\cdots, A_{(i)} + B_{(i)},\cdots) = D(\cdots, A_{(i)},\cdots) + D(\cdots, B_{(i)},\cdots)\)

1.假设\(A_{(i)}|_{i=1}^n\)线性无关

\[B_{(i)} = \sum_{j=1}^nk_jA_{(j)} \]

所以

\[\begin{align} D(\cdots, A_{(i)} + B_{(i)},\cdots) =& D(\cdots, A_{(i)} + \sum_{j=1}^nk_jA_{(j)},\cdots)\\ =& D(\cdots, A_{(i)} + k_iA_{(i)},\cdots)\\ =& (1+k_i)D(\cdots,A_{(i)},\cdots)\\ =& D(\cdots,A_{(i)},\cdots) + D(\cdots,k_iA_{(i)},\cdots)\\ =& D(\cdots,A_{(i)},\cdots) + D(\cdots,\sum_{j=1}^nk_jA_{(j)},\cdots)\\ =& D(\cdots,A_{(i)},\cdots) + D(\cdots,B_{(i)},\cdots) \end{align} \]

2.假设\(A_{(i)}|_{i=1}^n\)线性相关

那么存在一组不全为0的常数,使得

\[\sum_{i=1}^nk_iA_{(i)} = 0 \]

假如\(k_i \ne 0\),那么

\[A_{(i)} = \sum_{j=1,j\ne i}^nk_jA_{(j)} \]

所以

\[\begin{align} D(\cdots, A_{(i)}+B_{(j)},\cdots) =& D(\cdots, \sum_{j=1,j\ne i}^nA_{(j)} + B_{(i)},\cdots)\\ =& D(\cdots, B_{(i)},\cdots) \end{align} \]

那么

\[\begin{align} &D(\cdots,A_{(i)},\cdots) = D(\cdots, 0,\cdots) = 0\\ \therefore &D(\cdots,A_{(i)} + B_{(i)},\cdots) = D(\cdots,A_{(i)},\cdots) + D(\cdots,B_{(i)},\cdots) \end{align} \]

假如\(k_i = 0\),那么一定存在\(k_j \ne 0(j \ne i)\),使得

\[A_{(j)} = \sum_{i=1,i\ne j}^nk_iA_{(i)} \]

所以

\[\begin{align} D(\cdots,A_{(j)},\cdots,A_{(i)} + B_{(i)},\cdots) =& D(\cdots, 0 ,\cdots,A_{(i)}+B_{(i)},\cdots)\\ =& D(\cdots,0,\cdots,A_{(i)},\cdots) + D(\cdots,0,\cdots,B_{(i)},\cdots)\\ =& D(\cdots,A_{(i)},\cdots) + D(\cdots,B_{(i)},\cdots) \end{align} \]

证毕。

根据以上两个公式,我们建立\(n\)维坐标系\([\vec e_1,\vec e_2,\cdots,\vec e_n]\),其中,任意的\(A_{(i)} = \sum_{j=1}^nk_j\vec e_j\)

\[\begin{align} D(A_{(1)},A_{(2)},\cdots,A_{(n)}) =& \sum_{i_1=1}^nk_{i_1}D(\vec e_{i_1},A_{(2)},\cdots, A_{(n)})\\ =& \sum_{i_1=1}^n\sum_{i_2=1}^nk_{i_1}k_{i_2}D(\vec e_{i_1}, \vec e_{i_2},\cdots,A_{(n)})\\ =& \sum_{i_1,i_2,\cdots,i_n}\textstyle \prod_{j=i_1}^{i_n}k_jD(\vec e_{i_1},\vec e_{i_2},\cdots,\vec e_{i_n}) \end{align} \]

其中,假如任何两个\(i_n,i_m\)相同,那么\(D(\cdots, \vec e_{i_n},\cdots,\vec e_{i_m}) = 0\),所以

\[D(A_{(1)},A_{(2)},\cdots,A_{(n)}) = \sum_{1\leq i_1\leq i_2\leq \cdots\leq n,i_n \ne i_m}\textstyle \prod_{j=i_1}^{i_n}k_jD(\vec e_{i_1},\vec e_{i_2},\cdots, \vec e_{i_n}) \]

其中,任意两个\(i_n \ne i_m\),这代表从\(i_1\)到\(i_n\)构成了集合\(\left\lbrace 1,2,3,\cdots,n\right\rbrace\)的一个全排列。

因为

\[\begin{align} D(E) = D(\vec e_1,\vec e_2,\cdots,\vec e_n) = 1 \end{align} \]

所以当我们对\(i_1,i_2,\cdots,i_n\)进行排列时,如果交换了奇数次,那么根据推论一,有\(D(\vec e_{i_1},\vec e_{i_2},\cdots,\vec e_{i_n}) = -1\),如果交换了偶数次,那么\(D(\vec e_{i_1},\vec e_{i_2},\cdots, \vec e_{i_n}) = 1\)。

于是

\[D(A_{(1)},A_{(2)},\cdots,A_{(n)}) = \sum_{1\leq i_1\leq i_2 \leq \cdots\leq n, i_n \ne i_m}(-1)^{\tau(i_1i_2\cdots i_n)}\prod_{j=i_1}^{i_n}k_j \]

最后,由于我们建立的坐标系是与坐标轴对齐的,对于一个行向量\(A_{(i)} = [a_1,a_2,\cdots,a_n]\)来说,每一个分向量都唯一对应坐标轴的系数,即

\[A_{(i)} = \sum_{i=1}^nk_i\vec e_i = [k_1,k_2,\cdots,k_n] \]

所以\(D\)函数就是行列式运算的定义。


行列式的性质

有了行列式的定义,我们就能证明行列式的运算性质。

  • 行列式转置后值不变:此性质说明对于行列式来说,行和列是等价的,应用在行身上的性质同样也出现在列身上。

    根据行列式定义

    \[\begin{align} \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ a_{21} & a_{22} & \cdots & a_{2n}\\ \vdots & \vdots & \ddots & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix} = \sum_{i_1i_2\cdots i_n}(-1)^{\tau(i_1i_2\cdots i_n)}a_{1i_1}a_{2i_2}\cdots a_{ni_n} \end{align} \]

    ​ 我们知道数\(i_1i_2\cdots i_n\)是\(1,2,\cdots ,n\)的一个排列,这代表这行列式的每一项都是从各行各列中取一个,并且相互分离不能有交集。那么同样的,我们从列取和行取式等价的,转置后的矩阵依旧是从各行各列中取一个,即当\(i_1i_2\cdots i_n\)取遍所有排列时,表示形式\(a_{1i_1}a_{2i_2}\cdots a_{ni_n}\)等价于\(a_{i_11}a_{i_22}\cdots a_{i_nn}\)。

    ​ 考虑某一个排列\(a_{i_11}a_{i_22}\cdots a_{i_nn}\),在行列式中,它对应项的正负如何确定呢?当我们将其排列回到原始\(a_{1i_1}a_{2i_2}\cdots a_{ni_n}\)时,相当于我们将排列\(i_1i_2\cdots i_n\)回到\(123\cdots n\)去,因为从\(123\cdots n\)排到\(i_1i_2\cdots i_n\)至少经过\(k\)次调换,那么同样也需要至少\(k\)次调换回到\(123\cdots n\)当中,所以正负号依旧由\(\tau(i_1i_2\cdots i_n)\)确定,所以:

    \[\begin{align} \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ a_{21} & a_{22} & \cdots & a_{2n}\\ \vdots & \vdots & \ddots & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix} = \sum_{i_1i_2\cdots i_n}(-1)^{\tau(i_1i_2\cdots i_n)}a_{i_11}a_{i_22}\cdots a_{i_nn} \end{align} \]

    即\(|A^T| = |A|\)

  • 将行列式的某一行(列)乘\(k\)倍加到另一行(列)上去,行列式不变:这是由公理条件所确定的,不证自明。

  • 将行列式的某一行(列)乘\(k\)倍,行列式的值也变为\(k\)倍:同上。

  • 将行列式的某两行(列)对调,行列式异号:已由公理化构造证明。

  • 行列式的某一行元素都可拆成两个元素之和,那么行列式等于这两个元素个成一行的两个行列式之和:已由公理化构造证明。

  • 行列式可按行(列)展开(拉普拉斯展开)计算:

    我们将代数余子式表示为\(A\),将余子式表示为\(B\),那么有:

    \[\begin{align} \begin{vmatrix} a_{11} & 0 & \cdots & 0\\ a_{21} & a_{22} & \cdots & a_{2n}\\ \vdots & \vdots & \ddots & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix} = a_{11}A_{11} \end{align} \]

    证明:

    \[\begin{align} D = \begin{vmatrix} a_{11} & 0 & \cdots & 0\\ a_{21} & a_{22} & \cdots & a_{2n}\\ \vdots & \vdots & \ddots & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix} =& (-1)^{\tau(i_1i_2\cdots i_n)}a_{1i_1}a_{2i_2}\cdots a_{ni_n} \end{align} \]

    因为第一行只有\(a_{11}\)不为零,所以

    \[\begin{align} D =& (-1)^{\tau(1i_2\cdots i_n)}a_{11}a_{2i_2}\cdots a_{ni_n}\\ =& a_{11}(-1)^{\tau(i_2i_3\cdots i_n)}a_{2i_2}\cdots a_{ni_n}\\ =& a_{11}A_{11} \end{align} \]

    其次,有:

    \[\begin{align} \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1j} & \cdots & a_{1n}\\ a_{21} & a_{22} & \cdots & a_{2j} & \cdots & a_{2n}\\ \vdots & \vdots & & \vdots & & \vdots\\ 0 & 0 & \cdots & a_{ij} & \cdots & 0\\ \vdots & \vdots & & \vdots & & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nj} & \cdots & a_{nn} \end{vmatrix} = a_{ij}A_{ij} \end{align} \]

    ​ 我们将\(a_{ij}\)通过调换行列的方式将其变为上面第一行首元素不为0的行列式形式。注意,为了保证代数余子式的行列还是按照顺序排列,我们必须将\(a_{ij}\)依次一排排一列列的同前面的元素进行调换,换句话说,我们在调换时必须保证其他元素的相对顺序不变。于是,我们就经过了\(i +j -2\)次调换。

    所以

    \[\begin{align} \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1j} & \cdots & a_{1n}\\ a_{21} & a_{22} & \cdots & a_{2j} & \cdots & a_{2n}\\ \vdots & \vdots & & \vdots & & \vdots\\ 0 & 0 & \cdots & a_{ij} & \cdots & 0\\ \vdots & \vdots & & \vdots & & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nj} & \cdots & a_{nn} \end{vmatrix} =(-1)^{i+j-2} \begin{vmatrix} a_{ij} & 0 & \cdots & 0\\ a_{1j} & a_{11} & \cdots & a_{1n}\\ \vdots & \vdots & \ddots & \vdots \\ a_{nj} & a_{n1} & \cdots & a_{nn} \end{vmatrix} = a_{ij}A_{ij} \end{align} \]

    那么,可推导出拉普拉斯展开定理:

    \[\begin{align}D =& \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ a_{21} & a_{22} & \cdots & a_{2n}\\ \vdots & \vdots & & \vdots \\ a_{i1} & a_{i2} & \cdots & a_{in} \\ \vdots & \vdots & & \vdots \\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix}\\ \\ =& \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ \vdots & \vdots & & \vdots\\ a_{i1} & 0 & \cdots & 0\\ \vdots & \vdots & & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix} + \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ \vdots & \vdots & & \vdots\\ 0 & a_{i2} & \cdots & 0\\ \vdots & \vdots & & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix} + \cdots + \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ \vdots & \vdots & & \vdots\\ 0 & 0 & \cdots & a_{in}\\ \vdots & \vdots & & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix}\\ =& \sum_{j=1}^na_{ij}A_{ij} \end{align} \]

    可得推论:某一行的元素乘上另一行的代数余子式结果必为0,因为该结构可以看作是有两行元素完全相同的行列式的值,该值必为0。

矩阵的逆

对于任意一个矩阵\(A\),假如存在一个矩阵\(B\),满足\(AB=BA=E\),那么我们称矩阵\(A\)可逆,并且其逆矩阵为\(B\)。

我们利用拉普拉斯展开构造一个伴随矩阵,其结构为:它由矩阵\(A\)对应元素的代数余子式经过转置后构成,即:

\[A^* = \begin{bmatrix} A_{11} & A_{21} & \cdots &A_{n1}\\ A_{12} & A_{22} & \cdots & A_{n2}\\ \vdots & \vdots & \ddots & \vdots \\ A_{1n} & A_{2n} & \cdots & A_{nn} \end{bmatrix} \]

通过矩阵乘法,我们有\(A^*A = AA^* = |A|\),所以\(A^{-1} = \dfrac{A^*}{|A|}\)

这意味着如果一个矩阵可逆,那么它是行列式一定不为0,于是我们可以通过对矩阵进行初等变换将其转化为一个阶梯矩阵。

如果\(A\)是一个方阵,那么它一定可以仅靠将某一行\(k\)倍加到另一行上来将\(A\)转换成一个三角矩阵,而且他的行列式的值不变。

行列式的计算

  • 三角矩阵:三角矩阵的行列式的值是对角线元素相乘。

    我们将形式如下的矩阵称为三角行列式(主对角线行列式)

    \[\begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ 0 & a_{22} & \cdots & a_{2n}\\ \vdots & \vdots & \ddots & \vdots\\ 0 & 0 & \cdots & a_{nn} \end{vmatrix}= \begin{vmatrix} a_{11} &0 & \cdots & 0\\ a_{21} & a_{22} & \cdots & 0\\ \vdots & \vdots & \ddots & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix}= \begin{vmatrix} a_{11} & 0 & \cdots & 0\\ 0 & a_{22} & \cdots & 0\\ \vdots & \vdots & \ddots & \vdots\\ 0 & 0 & \cdots & a_{nn} \end{vmatrix}=\prod_{i=1}^na_{ii} \]

    证明:

    ​ 由行列式性质一可得,对行满足的计算对列具有同样的结果,所以我们只需证明第一种和最后一种情况。

    将两种行列式都按第一列使用拉普拉斯展开

    \[\begin{align} \begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ 0 & a_{22} & \cdots & a_{2n}\\ \vdots & \vdots & \ddots & \vdots\\ 0 & 0 & \cdots & a_{nn} \end{vmatrix} = \begin{vmatrix} a_{11} & 0 & \cdots & 0\\ 0 & a_{22} & \cdots & 0\\ \vdots & \vdots & \ddots & \vdots\\ 0 & 0 & \cdots & a_{nn} \end{vmatrix} =a_{11}A_{11} \end{align} \]

    同样的,对于\(A_{11}\),我们依旧按照列展开,之后不断重复这个操作:

    \[a_{11}A_{11} = a_{11}a_{22}A'_{11} = \cdots = \prod_{i=1}^na_{nn}\\ A'_{11}是a_{11}子式继续按照第一列展开后所得到属于a_{22}的代数余子式 \]

  • 副对角线行列式:我们将形式如下的行列式称为副对角线行列式。

    \[\begin{vmatrix} a_{11} & a_{12} & \cdots & a_{1n}\\ a_{21} & a_{22} & \cdots & 0\\ \vdots & \vdots & \ddots & \vdots\\ a_{nn} & 0 & \cdots & 0 \end{vmatrix}= \begin{vmatrix} 0 & 0 & \cdots & a_{1n}\\ 0 & 0 & \cdots & a_{2n}\\ \vdots & \vdots & \ddots & \vdots\\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{vmatrix}= \begin{vmatrix} 0 & 0 & \cdots & a_{1n}\\ 0 & 0 & \cdots & 0\\ \vdots & \vdots & \ddots & \vdots\\ a_{n1}& 0 & \cdots & 0 \end{vmatrix}=(-1)^{\frac{n(n-1)}{2}}\prod_{i=1}^na_{i(n+1-i)} \]

    证明:

    ​ 我们可以通过将副对角线行列式通过数次行列交换变成主对角线行列式。为了找出正负关系,我们可以尝试每次只与相邻的行(列)对调,那么第\(n\)行(列)需要对调\(n-1\)次。同理,第\(i\)行(列)需要对调\(i-1\)次,最后加起来,总共对调了\(\dfrac{n(n-1)}{2}\)次。

  • 拉普拉斯展开式:

    我们把形如下面的行列式称为拉普拉斯展开式。

    \[\begin{vmatrix} A & C\\ O & B \end{vmatrix} = \begin{vmatrix} A & O\\ C & B \end{vmatrix} = \begin{vmatrix} A & O\\ O & B \end{vmatrix} = |A||B| \]

    我们将行列式分块,其中分块矩阵\(A\)和\(B\)是\(n\)阶,\(m\)阶方阵。

    证明:

    我们记初等矩阵\(E_i(k)\)表示将矩阵的第\(i\)行(列)乘\(k\)被,\(E_{ij}\)为交换矩阵的\(i,j\)行(列),\(E_{ij}(k)\)为将第\(j\)行\(k\)倍加到第\(i\)行上(或者第\(i\)列\(k\)倍加到第\(j\)列上)。

    我们有三个显而易见的结论:

    \[\begin{align} \begin{vmatrix} \begin{bmatrix} E_i(k) & O \\ O & E \end{bmatrix} \begin{bmatrix} A & B \\ C & D \end{bmatrix} \end{vmatrix} =& \begin{vmatrix} \begin{bmatrix} E & O\\ O & E_i(k) \end{bmatrix} \begin{bmatrix} A & B\\ C & D \end{bmatrix} \end{vmatrix} = k \begin{vmatrix} A & B \\ C & D \end{vmatrix}\\ \\ \begin{vmatrix} \begin{bmatrix} E_{ij} & O \\ O & E \end{bmatrix} \begin{bmatrix} A & B \\ C & D \end{bmatrix} \end{vmatrix} =& \begin{vmatrix} \begin{bmatrix} E & O\\ O & E_{ij} \end{bmatrix} \begin{bmatrix} A & B\\ C & D \end{bmatrix} \end{vmatrix} = - \begin{vmatrix} A & B \\ C & D \end{vmatrix}\\ \\ \begin{vmatrix} \begin{bmatrix} E_{ij}(k) & O \\ O & E \end{bmatrix} \begin{bmatrix} A & B \\ C & D \end{bmatrix} \end{vmatrix} =& \begin{vmatrix} \begin{bmatrix} E & O\\ O & E_{ij}(k) \end{bmatrix} \begin{bmatrix} A & B\\ C & D \end{bmatrix} \end{vmatrix} = \begin{vmatrix} A & B \\ C & D \end{vmatrix}\\ \end{align} \]

    上面三个结论根据分块矩阵乘法和行列式性质简单易证。

    那么有(上述的初等矩阵在下面都用\(P_i\)表示):

    \[\begin{align} \begin{vmatrix} P_n\cdots P_2P_1 \begin{bmatrix} A & C \\ O & B \end{bmatrix} \end{vmatrix} =& (-1)^{a_1}\prod_{i=1}^{a_2}k_i \begin{vmatrix} A & C \\ O & B \end{vmatrix} \end{align} \]

    其中一共用了\(a_1\)次交换行列,\(a_2\)次\(k\)倍倍乘。

    此外,还有

    \[\begin{align} \begin{vmatrix} P_n\cdots P_2P_1 \begin{bmatrix} A & C \\ O & B \end{bmatrix} \end{vmatrix} =& \begin{vmatrix} P_n\cdots P_2P_1A & P_n\cdots P_2P_1C\\ O & B \end{vmatrix} \end{align} \]

    同理,对\(B\)进行同样的操作,其中包含\(b_1\)次行交换,\(b_2\)次倍乘。

    那么

    \[\begin{align} \begin{vmatrix} Q_n\cdots Q_1P_n\cdots P_1 \begin{bmatrix} A & C \\ O & B \end{bmatrix} \end{vmatrix} =& \begin{vmatrix} P_n\cdots P_1A & P_n\cdots P_1C\\ O & Q_n\cdots Q_1B \end{vmatrix}\\ =& (-1)^{a_1 + b_1}\prod_{i=1}^{a_2+b_2}k_i\begin{vmatrix} A & C \\ O & B \end{vmatrix} \end{align} \]

    由于\(A,B\)都是方阵,假如\(A,B\)其中任意一个不可逆,那么经过初等变换后其等价的标准型必含0列,不失一般性,我们假设\(A\)不可逆:

    \[\begin{vmatrix} P_n\cdots P_1A & P_n\cdots P_1C\\ O & Q_n\cdots Q_1B \end{vmatrix} = \begin{vmatrix} D_m & P_n\cdots P_1C \\ O & Q_n\cdots Q_1B \end{vmatrix} = |A||B| = 0 \]

    假如\(A,B\)均可逆,那么其等价标准型必是单位矩阵有:

    \[P_n\cdots P_1A = I\\ |P_n\cdots P_1A| = (-1)^{a_1}\prod_{i=1}^{a_2}k_i|A| = 1\\ |A| = (-1)^{-a_1}\dfrac{1}{\prod_{i=1}^{a_2}k_i} \]

    同理,\(|B| = (-1)^{-b_1}\dfrac{1}{\prod_{i=1}^{b_2}k_i}\),于是

    \[\begin{align} \therefore \begin{vmatrix} P_n\cdots P_1A & P_n\cdots P_1C\\ O & Q_n\cdots Q_1B \end{vmatrix} =& \begin{vmatrix} I & P_n\cdots P_1C \\ O & I \end{vmatrix} = 1\\ \therefore \begin{vmatrix} A & C \\ O & B \end{vmatrix} =& (-1)^{-a_1}\dfrac{1}{\prod_{i=1}^{a_1}k_i} (-1)^{b_1}\dfrac{1}{\prod_{i=1}^{b_1}k_i}\\ =& |A||B| \end{align} \]

  • 范德蒙德行列式:

    记

    \[\begin{vmatrix} 1 & 1 & 1 & \cdots & 1\\ x_1 & x_2 & x_3 & \cdots & x_n\\ x_1^2 & x_2^2 & x_3^2 & \cdots & x_n^2\\ \vdots & \vdots & \vdots & \ddots & \vdots\\ x_1^{n-1} &x_2^{n-1} & x_3^{n-1} & \cdots & x_n^{n-1} \end{vmatrix} = \prod_{1\leq i < j \leq n}(x_j - x_i) \]

    证明:

    使用数学归纳法,已知当\(n=2\)时成立:

    \[\begin{vmatrix} 1 & 1\\ x_1 & x_2 \end{vmatrix} = x_2 - x_1 \]

    假设当\(n>2\)时,结论对\(n-1\)的情况成立,那么

    \[\begin{align} \begin{vmatrix} 1 & 1 & 1 & \cdots & 1\\ x_1 & x_2 & x_3 & \cdots & x_n\\ x_1^2 & x_2^2 & x_3^2 & \cdots & x_n^2\\ \vdots & \vdots & \vdots & \ddots & \vdots\\ x_1^{n-1} &x_2^{n-1} & x_3^{n-1} & \cdots & x_n^{n-1} \end{vmatrix} =& \begin{vmatrix} 1 & 1 & 1 & \cdots & 1\\ 0 & x_2 - x_1 & x_3-x_1 & \cdots & x_n-x_1\\ 0 & x_2^2 - x_1x_2 & x_3^2-x_1x_3 & \cdots & x_n^2 - x_1x_n\\ \vdots & \vdots & \vdots & \ddots & \vdots\\ 0 &x_2^{n-1} - x_1x_2^{n-2} & x_3^{n-1}-x_1x_3^{n-2} & \cdots & x_n^{n-1} - x_1x_n^{n-2} \end{vmatrix}\\ \\ =& \begin{vmatrix} x_2 - x_1 & x_3-x_1 & \cdots & x_n-x_1\\ x_2^2 - x_1x_2 & x_3^2-x_1x_3 & \cdots & x_n^2 - x_1x_n\\ \vdots & \vdots & \ddots & \vdots\\ x_2^{n-1} - x_1x_2^{n-2} & x_3^{n-1}-x_1x_3^{n-2} & \cdots & x_n^{n-1} - x_1x_n^{n-2} \end{vmatrix}\\ \\ =&(x_2-x_1)(x_3-x_1)\cdots (x_n-x_1) \begin{vmatrix} 1 & 1 & \cdots & 1\\ x_2& x_3& \cdots & x_n\\ x_2^2& x_3^2& \cdots & x_n^2\\ \vdots & \vdots & \ddots & \vdots\\ x_2^{n-2} & x_3^{n-2} & \cdots & x_n^{n-2} \end{vmatrix}\\ =&(x_2-x_1)\cdots (x_n-x_1)\prod_{2\leq i < j\leq n}(x_j- x_i)\\ =& \prod_{1\leq i < j\leq n} (x_j -x_i) \end{align} \]

    于是,由归纳法得出,当\(n\ge2\)时,结论成立。

  • 行列式乘法

    如果\(A,B\)都是\(n\)阶矩阵,那么有\(|AB| = |A||B|\)

    证明:

    根据拉普拉斯展开式和行列式性质,有

    \[\begin{vmatrix} \begin{bmatrix} E & O \\ X & E \end{bmatrix} \begin{bmatrix} A & B \\ C & D \end{bmatrix} \end{vmatrix} = \begin{vmatrix} A & B\\ C & D \end{vmatrix} \]

    同理

    \[\begin{vmatrix} \begin{bmatrix} E & X \\ O & E \end{bmatrix} \begin{bmatrix} A & B \\ C & D \end{bmatrix} \end{vmatrix} = \begin{vmatrix} A & B\\ C & D \end{vmatrix} \]

    由此,我们可以构造矩阵:

    \[\begin{bmatrix} E & 0\\ -E & E \end{bmatrix} \begin{bmatrix} E & E\\ 0 & E \end{bmatrix} \begin{bmatrix} E & -A\\ 0 & E \end{bmatrix} \begin{bmatrix} A & 0\\ E & B \end{bmatrix} = \begin{bmatrix} E & B-AB\\ 0 & AB \end{bmatrix} \]

    那么,有

    \[\begin{vmatrix} \begin{bmatrix} E & B-AB\\ 0 & AB \end{bmatrix} \end{vmatrix} =|AB|= \begin{vmatrix} \begin{bmatrix} A & 0\\ E & B \end{bmatrix} \end{vmatrix} = |A||B| \]

posted @ 2023-07-29 19:14  ᴮᴱˢᵀ  阅读(149)  评论(0)    收藏  举报