2. Matrix Inversion and Determinants

这一节继续复习线性代数知识,回顾了初等矩阵、求逆、行列式、代数余子式、Cramer法则等相关内容,总体上也易于理解。

Exercises

Exercise 1. Consider the matrix
A = [ 1 2 1 − 1 0 1 ] . A=\begin{bmatrix}1&2\\1&-1\\0&1\end{bmatrix}. A=110211.
( a ) Find two different left inverses for A A A.
( b ) Show that A A A has no right inverse.
Solution:
(a) The two left inverses for A A A may be
B 1 = [ 1 0 − 2 0 0 1 ] , B 2 = [ 0 1 1 0 0 1 ] B_1=\begin{bmatrix}1&0&-2\\0&0&1\end{bmatrix},\quad B_2=\begin{bmatrix}0&1&1\\0&0&1\end{bmatrix} B1=[100021],B2=[001011]
(b) Assume A A A has a right inverse
B = [ a b c d e f ] , A B = I 3 B=\begin{bmatrix}a&b&c\\d&e&f\end{bmatrix},\quad AB=I_3 B=[adbecf],AB=I3
then
A B = [ 1 2 1 − 1 0 1 ] [ a b c d e f ] = [ a + 2 d b + 2 e c + 2 f a − d b − e e − f d e f ] = [ 1 0 0 0 1 0 0 0 1 ] AB=\begin{bmatrix}1&2\\1&-1\\0&1\end{bmatrix}\begin{bmatrix}a&b&c\\d&e&f\end{bmatrix}=\begin{bmatrix}a+2d&b+2e&c+2f\\a-d&b-e&e-f\\d&e&f\end{bmatrix}=\begin{bmatrix}1&0&0\\0&1&0\\0&0&1\end{bmatrix} AB=110211[adbecf]=a+2daddb+2ebeec+2feff=100010001
from the bottom row we have f = 1 , d = e = 0 f=1,d=e=0 f=1,d=e=0, so e − f = − 1 e-f=-1 ef=1, but from the second row we know e − f = 0 e-f=0 ef=0, a contradiction.

Exercise 2. Let A A A be an n n n by m m m matrix with n ≠ m n\neq m n=m.
( a ) If rank  A = m \text{rank }A=m rank A=m, show there exists a matrix D D D that is a product of elementary matrices such that
D ⋅ A = [ I m 0 ] . D\cdot A=\begin{bmatrix}I_m\\0\end{bmatrix}. DA=[Im0].
( b ) Show that A A A has a left inverse if and only if rank  A = m \text{rank }A=m rank A=m.
( c ) Show that A A A has a right inverse if and only if rank  A = n \text{rank }A=n rank A=n.
Solution:
( a ) If rank  A = m \text{rank }A=m rank A=m, then A A A has at least m m m rows, but n ≠ m n\neq m n=m, so n > m n>m n>m. Since we can use elementary row operations to reduce A A A to [ I m 0 ] \begin{bmatrix}I_m\\0\end{bmatrix} [Im0], D D D is the product of the corresponding elementary matrices.
( b ) If rank  A = m \text{rank }A=m rank A=m, then by (a) we can find D D D such that D ⋅ A = [ I m 0 ] D\cdot A= \begin{bmatrix}I_m\\0\end{bmatrix} DA=[Im0], let the m × n m\times n m×n matrix E E E be E = [ I m 0 ] E=\begin{bmatrix}I_m&0\end{bmatrix} E=[Im0], then
E D A = [ I m 0 ] [ I m 0 ] = I m EDA=\begin{bmatrix}I_m&0\end{bmatrix}\begin{bmatrix}I_m\\0\end{bmatrix}=I_m EDA=[Im0][Im0]=Im
thus E D ED ED is a left inverse of A A A. Conversely, if A A A has a left inverse, then there is B B B s.t. B A = I m BA=I_m BA=Im, from the proof of Theorem 2.3 we know rank  A = m \text{rank }A=m rank A=m.
( c ) From step 3 of the proof of Theorem 2.3 we know if A A A has a right inverse, then rank  A = n \text{rank }A=n rank A=n. Conversely if rank  A = n \text{rank }A=n rank A=n, then rank  A t r = n \text{rank }A^{tr}=n rank Atr=n, by (b), A t r A^{tr} Atr has a left inverse B B B, so B A t r = I n BA^{tr}=I_n BAtr=In, thus ( B A t r ) t r = A B t r = I n (BA^{tr} )^{tr}=AB^{tr}=I_n (BAtr)tr=ABtr=In.

Exercise 3. Verify that the functions defined in Example 1 satisfy the axioms for the determinant function.
Solution: For 1 by 1 matrices, we have det ⁡ ⁡ [ a ] = a \det⁡[a]=a det[a]=a, the verification is
(1) there is no way to exchange rows.
(2) we have det ⁡ [ a ] = a = a ⋅ 1 = a det ⁡ [ 1 ] \det[a]=a=a\cdot 1=a \det[1] det[a]=a=a1=adet[1].
(3) det ⁡ [ 1 ] = 1 \det[1]=1 det[1]=1.
For 2 by 2 matrices, we have det ⁡ ⁡ [ a b c d ] = a d − b c \det⁡\begin{bmatrix}a&b\\c&d\end{bmatrix}=ad-bc det[acbd]=adbc, the verification is
(1) det ⁡ ⁡ [ c d a b ] = c b − a d = − ( a d − b c ) = − det ⁡ ⁡ [ a b c d ] \det⁡\begin{bmatrix}c&d\\a&b\end{bmatrix}=cb-ad=-(ad-bc)=-\det⁡\begin{bmatrix}a&b\\c&d\end{bmatrix} det[cadb]=cbad=(adbc)=det[acbd]
(2) Let x = ( x 1 , x 2 ) , y = ( y 1 , y 2 ) \mathrm x=(x_1,x_2 ),\mathrm{y}=(y_1,y_2 ) x=(x1,x2),y=(y1,y2), then
A 1 ( x ) = [ x 1 x 2 c d ] , A 2 ( x ) = [ a b x 1 x 2 ] , A 1 ( y ) = [ y 1 y 2 c d ] , A 2 ( y ) = [ a b y 1 y 2 ] A_1 (x)=\begin{bmatrix}x_1&x_2\\c&d\end{bmatrix},A_2 (x)=\begin{bmatrix}a&b\\x_1&x_2 \end{bmatrix},A_1 (y)=\begin{bmatrix}y_1&y_2\\c&d\end{bmatrix},A_2 (y)=\begin{bmatrix}a&b\\y_1&y_2 \end{bmatrix} A1(x)=[x1cx2d],A2(x)=[ax1bx2],A1(y)=[y1cy2d],A2(y)=[ay1by2]
so
det ⁡ ⁡ A 1 ( m x + n y ) = det ⁡ ⁡ [ m x 1 + n y 1 m x 2 + n y 2 c d ] = ( m x 1 + n y 1 ) d − ( m x 2 + n y 2 ) c = m ( x 1 d − x 2 c ) + n ( y 1 d − y 2 c ) = m det ⁡ ⁡ A 1 ( x ) + n det ⁡ ⁡ A 1 ( y ) \begin{aligned}\det ⁡A_1 (m\mathrm{x}+n\mathrm{y})&=\det⁡\begin{bmatrix}mx_1+ny_1&mx_2+ny_2\\c&d\end{bmatrix}\\&=(mx_1+ny_1 )d-(mx_2+ny_2 )c\\&=m(x_1 d-x_2 c)+n(y_1 d-y_2 c)\\&=m \det⁡A_1 (x)+n \det⁡ A_1 (y)\end{aligned} detA1(mx+ny)=det[mx1+ny1cmx2+ny2d]=(mx1+ny1)d(mx2+ny2)c=m(x1dx2c)+n(y1dy2c)=mdetA1(x)+ndetA1(y)
Similarly det ⁡ ⁡ A 2 ( m x + n y ) = m det ⁡ ⁡ A 2 ( x ) + n det ⁡ ⁡ A 2 ( y ) \det⁡ A_2 (mx+ny)=m \det⁡ A_2 (x)+n \det ⁡A_2 (y) detA2(mx+ny)=mdetA2(x)+ndetA2(y).
(3) det ⁡ ⁡ [ 1 0 0 1 ] = 1 ⋅ 1 − 0 ⋅ 0 = 1 \det⁡\begin{bmatrix}1&0\\0&1\end{bmatrix}=1\cdot 1-0\cdot 0=1 det[1001]=1100=1
The verification for 3 by 3 matrices is of the same logic and omitted.

Exercise 4. ( a ) Let A A A be an n n n by n n n matrix of rank n n n. By applying elementary row operations to A A A, one can reduce A A A to the identity matrix. Show that by applying the same operations, in the same order, to I n I_n In, one obtains the matrix A − 1 A^{-1} A1.
( b ) Let
A = [ 1 2 3 0 1 2 1 2 1 ] . A=\begin{bmatrix}1&2&3\\0&1&2\\1&2&1\end{bmatrix}. A=101212321.
Calculate A − 1 A^{-1} A1 by using the algorithm suggested in (a).
( c ) Calculate A − 1 A^{-1} A1 using the formula involving determinants.
Solution:
( a ) We suppose, by Theorem 2.1, that there are elementary matrices E 1 , … , E k E_1,\dots,E_k E1,,Ek such that E k ⋯ E 1 A = I n E_k{\cdots}E_1 A=I_n EkE1A=In, let B = E k ⋯ E 1 B=E_k{\cdots}E_1 B=EkE1, then B B B is a left inverse for A A A, by Theorem 2.5, B B B is also a right inverse for A A A, thus B = A − 1 B=A^{-1} B=A1, further we have E k … E 1 I n = B I n = B = A − 1 E_k{\dots}E_1 I_n=BI_n=B=A^{-1} EkE1In=BIn=B=A1.
( b ) We have
[ A I 3 ] = [ 1 2 3 1 0 0 0 1 2 0 1 0 1 2 1 0 0 1 ] → [ 1 2 3 1 0 0 0 1 2 0 1 0 0 0 − 2 − 1 0 1 ] → [ 1 2 0 − 1 / 2 0 3 / 2 0 1 0 − 1 1 1 0 0 1 1 / 2 0 − 1 / 2 ] → [ 1 0 0 3 / 2 − 2 − 1 / 2 0 1 0 − 1 1 1 0 0 1 1 / 2 0 − 1 / 2 ] \begin{aligned}\begin{bmatrix}A&I_3 \end{bmatrix}&=\begin{bmatrix}1&2&3&1&0&0\\0&1&2&0&1&0\\1&2&1&0&0&1\end{bmatrix}{\rightarrow}\begin{bmatrix}1&2&3&1&0&0\\0&1&2&0&1&0\\0&0&-2&-1&0&1\end{bmatrix}\\&{\rightarrow}\begin{bmatrix}1&2&0&-1/2&0&3/2\\0&1&0&-1&1&1\\0&0&1&1/2&0&-1/2\end{bmatrix}{\rightarrow}\begin{bmatrix}1&0&0&3/2&-2&-1/2\\0&1&0&-1&1&1\\0&0&1&1/2&0&-1/2\end{bmatrix}\end{aligned} [AI3]=1012123211000100011002103221010100011002100011/211/20103/211/21000100013/211/22101/211/2
thus A − 1 = [ 3 / 2 − 2 − 1 / 2 − 1 1 1 1 / 2 0 − 1 / 2 ] A^{-1}=\begin{bmatrix}3/2&-2&-1/2\\-1&1&1\\1/2&0&-1/2\end{bmatrix} A1=3/211/22101/211/2.
( c ) The calculation is based on Theorem 2.14, and is omitted.

Exercise 5. Let A = [ a b c d ] A=\begin{bmatrix}a&b\\c&d\end{bmatrix} A=[acbd], where a d − b c ≠ 0 ad-bc\neq 0 adbc=0. Find A − 1 A^{-1} A1.
Solution: By Theorem 2.14,
b 11 = ( − 1 ) 1 + 1 d a d − b c , b 12 = ( − 1 ) 1 + 2 b a d − b c , b 21 = ( − 1 ) 2 + 1 c a d − b c , b 22 = ( − 1 ) 2 + 2 a a d − b c b_{11}=\frac{(-1)^{1+1} d}{ad-bc},b_{12}=\frac{(-1)^{1+2}b}{ad-bc},b_{21}=\frac{(-1)^{2+1}c}{ad-bc},b_{22}=\frac{(-1)^{2+2}a}{ad-bc} b11=adbc(1)1+1d,b12=adbc(1)1+2b,b21=adbc(1)2+1c,b22=adbc(1)2+2a
thus
A − 1 = 1 a d − b c [ d − b − c a ] A^{-1}=\frac{1}{ad-bc}\begin{bmatrix}d&-b\\-c&a\end{bmatrix} A1=adbc1[dcba]

Exercise 6. Prove the following theorem: Let A A A be a k k k by k k k matrix, let D D D have size n n n by n n n and let C C C have size n n n by k k k. Then
det ⁡ [ A 0 C D ] = ( det ⁡ A ) ⋅ ( det ⁡ D ) . \det\begin{bmatrix}A&0\\C&D\end{bmatrix}=(\det A)\cdot(\det D). det[AC0D]=(detA)(detD).
Solution: We have
[ A 0 0 I n ] ⋅ [ I k 0 C D ] = [ A 0 C D ] \begin{bmatrix}A&0\\0&I_n \end{bmatrix}⋅\begin{bmatrix}I_k&0\\C&D\end{bmatrix}=\begin{bmatrix}A&0\\C&D\end{bmatrix} [A00In][IkC0D]=[AC0D]
and by Lemma 2.12 we have
det ⁡ ⁡ [ A 0 0 I n ] = det ⁡ ⁡ A , det ⁡ ⁡ [ I k 0 C D ] = det ⁡ D \det⁡\begin{bmatrix}A&0\\0&I_n \end{bmatrix}=\det⁡ A,\det⁡\begin{bmatrix}I_k&0\\C&D\end{bmatrix}=\det D det[A00In]=detA,det[IkC0D]=detD
so by Theorem 2.10 we have
det ⁡ ⁡ [ A 0 C D ] = ( det ⁡ A ) ⋅ ( det ⁡ D ) \det⁡\begin{bmatrix}A&0\\C&D\end{bmatrix}=(\det A)\cdot(\det D) det[AC0D]=(detA)(detD)

  • 0
    点赞
  • 0
    收藏
    觉得还不错? 一键收藏
  • 0
    评论

“相关推荐”对你有帮助么?

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值