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
$$A=\left[\begin{array}{cc}1& 2\\ 1& -1\\ 0& 1\end{array}\right].$$
( a ) Find two different left inverses for $A$.
( b ) Show that $A$ has no right inverse.
Solution:
(a) The two left inverses for $A$ may be
$$B_1=\left[\begin{array}{ccc}1& 0& -2\\ 0& 0& 1\end{array}\right],B_2=\left[\begin{array}{ccc}0& 1& 1\\ 0& 0& 1\end{array}\right]$$
(b) Assume $A$ has a right inverse
$$B=\left[\begin{array}{ccc}a& b& c\\ d& e& f\end{array}\right],AB=I_3$$
then
$$AB=\left[\begin{array}{cc}1& 2\\ 1& -1\\ 0& 1\end{array}\right]\left[\begin{array}{ccc}a& b& c\\ d& e& f\end{array}\right]=\left[\begin{array}{ccc}a+2d& b+2e& c+2f\\ a-d& b-e& e-f\\ d& e& f\end{array}\right]=\left[\begin{array}{ccc}1& 0& 0\\ 0& 1& 0\\ 0& 0& 1\end{array}\right]$$
from the bottom row we have $f=1,d=e=0$, so $e-f=-1$, but from the second row we know $e-f=0$, a contradiction.

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

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 $\det [a]=a$, the verification is
(1) there is no way to exchange rows.
(2) we have $\det [a]=a=a\cdot 1=a\det [1]$.
(3) $\det [1]=1$.
For 2 by 2 matrices, we have $\det \left[\begin{array}{cc}a& b\\ c& d\end{array}\right]=ad-bc$, the verification is
(1) $\det \left[\begin{array}{cc}c& d\\ a& b\end{array}\right]=cb-ad=-(ad-bc)=-\det \left[\begin{array}{cc}a& b\\ c& d\end{array}\right]$
(2) Let $\mathrm{x}=(x_1,x_2),\mathrm{y}=(y_1,y_2)$, then
$$A_1(x)=\left[\begin{array}{cc}{x}_1& x_2\\ c& d\end{array}\right],A_2(x)=\left[\begin{array}{cc}a& b\\ x_1& x_2\end{array}\right],A_1(y)=\left[\begin{array}{cc}{y}_1& y_2\\ c& d\end{array}\right],A_2(y)=\left[\begin{array}{cc}a& b\\ y_1& y_2\end{array}\right]$$
so
$$\begin{array}{rl}\det A_1(m\mathrm{x}+n\mathrm{y})& =\det \left[\begin{array}{cc}m{x}_1+n{y}_1& m{x}_2+n{y}_2\\ c& d\end{array}\right]\\ & =(m{x}_1+n{y}_1)d-(m{x}_2+n{y}_2)c\\ & =m(x_{1}d-x_{2}c)+n(y_{1}d-y_{2}c)\\ & =m\det A_1(x)+n\det A_1(y)\end{array}$$
Similarly $\det A_2(mx+ny)=m\det A_2(x)+n\det A_2(y)$.
(3) $\det \left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right]=1\cdot 1-0\cdot 0=1$
The verification for 3 by 3 matrices is of the same logic and omitted.

Exercise 4. ( a ) Let $A$ be an $n$ by $n$ matrix of rank $n$. By applying elementary row operations to $A$, one can reduce $A$ to the identity matrix. Show that by applying the same operations, in the same order, to $I_n$, one obtains the matrix $A^{-1}$.
( b ) Let
$$A=\left[\begin{array}{ccc}1& 2& 3\\ 0& 1& 2\\ 1& 2& 1\end{array}\right].$$
Calculate $A^{-1}$ by using the algorithm suggested in (a).
( c ) Calculate $A^{-1}$ using the formula involving determinants.
Solution:
( a ) We suppose, by Theorem 2.1, that there are elementary matrices $E_1,\dots ,E_k$ such that $E_k\cdots E_{1}A=I_n$, let $B=E_k\cdots E_1$, then $B$ is a left inverse for $A$, by Theorem 2.5, $B$ is also a right inverse for $A$, thus $B=A^{-1}$, further we have $E_k\dots E_1{I}_n=B{I}_n=B=A^{-1}$.
( b ) We have
$$\begin{array}{rl}\left[\begin{array}{cc}A& I_3\end{array}\right]& =\left[\begin{array}{cccccc}1& 2& 3& 1& 0& 0\\ 0& 1& 2& 0& 1& 0\\ 1& 2& 1& 0& 0& 1\end{array}\right]\to \left[\begin{array}{cccccc}1& 2& 3& 1& 0& 0\\ 0& 1& 2& 0& 1& 0\\ 0& 0& -2& -1& 0& 1\end{array}\right]\\ & \to \left[\begin{array}{cccccc}1& 2& 0& -1/2& 0& 3/2\\ 0& 1& 0& -1& 1& 1\\ 0& 0& 1& 1/2& 0& -1/2\end{array}\right]\to \left[\begin{array}{cccccc}1& 0& 0& 3/2& -2& -1/2\\ 0& 1& 0& -1& 1& 1\\ 0& 0& 1& 1/2& 0& -1/2\end{array}\right]\end{array}$$
thus $A^{-1}=\left[\begin{array}{ccc}3/2& -2& -1/2\\ -1& 1& 1\\ 1/2& 0& -1/2\end{array}\right]$.
( c ) The calculation is based on Theorem 2.14, and is omitted.

Exercise 5. Let $A=\left[\begin{array}{cc}a& b\\ c& d\end{array}\right]$, where $ad-bc\ne 0$. Find $A^{-1}$.
Solution: By Theorem 2.14,
$$b_{11}=\frac{(-1{)}^{1+1}d}{ad-bc},b_{12}=\frac{(-1{)}^{1+2}b}{ad-bc},b_{21}=\frac{(-1{)}^{2+1}c}{ad-bc},b_{22}=\frac{(-1{)}^{2+2}a}{ad-bc}$$
thus
$$A^{-1}=\frac{1}{ad-bc}\left[\begin{array}{cc}d& -b\\ -c& a\end{array}\right]$$

Exercise 6. Prove the following theorem: Let $A$ be a $k$ by $k$ matrix, let $D$ have size $n$ by $n$ and let $C$ have size $n$ by $k$. Then
$$\det \left[\begin{array}{cc}A& 0\\ C& D\end{array}\right]=(\det A)\cdot (\det D).$$
Solution: We have
$$\left[\begin{array}{cc}A& 0\\ 0& I_n\end{array}\right]\cdot \left[\begin{array}{cc}{I}_k& 0\\ C& D\end{array}\right]=\left[\begin{array}{cc}A& 0\\ C& D\end{array}\right]$$
and by Lemma 2.12 we have
$$\det \left[\begin{array}{cc}A& 0\\ 0& I_n\end{array}\right]=\det A,\det \left[\begin{array}{cc}{I}_k& 0\\ C& D\end{array}\right]=\det D$$
so by Theorem 2.10 we have
$$\det \left[\begin{array}{cc}A& 0\\ C& D\end{array}\right]=(\det A)\cdot (\det D)$$