Chapter 15
By the end of this section, you should be able to understand:
A great deal of this section should be familiar to you. We start by recalling some results on vector spaces associated with matrices.
For $n\times n$ square matrix $A$, we have several conditions for the existence of $A^{-1}$. Then the following statements are equivalent:
1. | $A$ is non-singular. |
2. | $A{\mathbf x}={\mathbf 0}$ has only the trivial solution ${\mathbf x}={\mathbf 0}$. |
3. | If $U$ is a reduced echelon form for $A$, then $U$ has no row of all zeros. |
4. | $A{\mathbf x}={\mathbf b}$ has a solution for every $n$-dimensional column vector ${\mathbf b}$. |
Then the following statements are equivalent:
1. | $A$ is non-singular. |
2. | $A{\mathbf x}={\mathbf 0}$ has only the trivial solution ${\mathbf x}={\mathbf 0}$. |
3. | If $U$ is a reduced echelon form for $A$, then $U$ has no row of all zeros. |
4. | $A{\mathbf x}={\mathbf b}$ has a solution for every $n$-dimensional column vector ${\mathbf b}$. |
5. | det$(A)\ne 0$. |
6. | The columns of $A$ are linearly independent. |
7. | The rows of $A$ are linearly independent. |
8. | nullity($A$) = 0. |
9. | rank($A$) = $n$. |
Let $A$ be a square matrix. Then an eigenvector of $A$ is a vector ${\mathbf v}\ne {\mathbf 0}$ such that $$ A{\mathbf v} = \lambda {\mathbf v}, $$ for some scalar $\lambda$. We call $\lambda$ the eigenvalue corresponding to ${\mathbf v}$. If ${\mathbf v}$ is an eigenvector of $A$, then so is $t{ \v}$ for any scalar $t\ne 0$.
We have $$ A{\mathbf v}=\lambda{\mathbf v}=\lambda I{\mathbf v}\ \ \Rightarrow \ \ (A-\lambda I){\mathbf v}={\mathbf 0}. $$
We have $$ A{\mathbf v}=\lambda{\mathbf v}=\lambda I{\mathbf v}\ \ \Rightarrow \ \ (A-\lambda I){\mathbf v}={\mathbf 0}. $$
Hence ${\mathbf x}={\mathbf v}$ is a non-trivial solution to the homogeneous system of equations $(A-\lambda I){\mathbf x}={\mathbf 0}$, and conversely, if there is a non-trivial solution then $\lambda$ is an eigenvalue of $A$.
$\lambda \mbox{ is an eigenvalue of $A$}$ | $\Leftrightarrow$ | $(A-\lambda I){\mathbf x}={\mathbf 0} \mbox{ has a non-trivial solution}$ |
$\Leftrightarrow$ | $A-\lambda I \mbox{ is singular}$ | |
$\Leftrightarrow$ | $\mbox{det}(A-\lambda I)=0. $ |
$\lambda \mbox{ is an eigenvalue of $A$}$ | $\Leftrightarrow$ | $(A-\lambda I){\mathbf x}={\mathbf 0} \mbox{ has a non-trivial solution}$ |
$\Leftrightarrow$ | $A-\lambda I \mbox{ is singular}$ | |
$\Leftrightarrow$ | $\mbox{det}(A-\lambda I)=0. $ |
For an $n\times n$ matrix $A$, det$(A-\lambda I)$ is a polynomial of degree $n$ in $\lambda$, called the characteristic polynomial of $A$. The equation det$(A-\lambda I)=0$ is the characteristic equation of $A$.
Note that eigenvalues $\lambda$ may be complex numbers, and the eigenvectors ${\mathbf v}$ may have complex components, even for real matrices $A$.
Note that eigenvalues $\lambda$ may be complex numbers, and the eigenvectors ${\mathbf v}$ may have complex components, even for real matrices $A$.
For example, consider \[ W = \left( \begin{array}{cc} 0 & 1 \\ -1 & 0 \end{array} \right) \]
The charateristic equation is
$\text{det}\left(W - \lambda I\right)$ $=\left| \begin{array}{cc} -\lambda & 1 \\ -1 & -\lambda \end{array} \right|$ $=\lambda^2+1$ $=0$ $\;\Ra \;\lambda = \pm i.$
To find the eigenvalues and eigenvectors, do the following:
Find the eigenvalues and eigenvectors of $A= \left(\begin{array}{rrr}-3&1&0\\1&-2&1\\0&1&-3\end{array}\right).$
Find the eigenvalues and eigenvectors of $A= \left(\begin{array}{rrr}-3&1&0\\1&-2&1\\0&1&-3\end{array}\right).$
Here we have that
$\ds
\text{det}\left(A-\lambda I\right)=
\left|
\begin{array}{ccc}
-3-\lambda & 1 & 0 \\
1 & -2-\lambda & 1 \\
0 & 1 & -3-\lambda \\
\end{array}
\right|
$
$\qquad \qquad \quad =-\left(\lambda +3\right)
\big[
(\lambda +3 )(\lambda +2)
-2
\big]
$
$\qquad \qquad \; =-(\lambda +3 )(\lambda +4)(\lambda +1)$
$= 0.$
Thus the eigenvalues are $\lambda_1 = -1, \lambda_2=-3,$ and $\lambda_3=-4.$
Find the eigenvalues and eigenvectors of $A= \left(\begin{array}{rrr}-3&1&0\\1&-2&1\\0&1&-3\end{array}\right).$
For $\lambda_1 = -1:$ $\ds \left(A-\lambda_1 I\right) \v=\mathbf 0 $ $ \;\Ra \; \left( \begin{array}{ccc} -2 & 1 & 0 \\ 1 & -1 & 1 \\ 0 & 1 & -2 \\ \end{array} \right) \left( \begin{array}{c} v_1 \\ v_2 \\ v_3 \\ \end{array} \right) =\left( \begin{array}{c} 0 \\ 0 \\ 0 \\ \end{array} \right) $
$
\Ra \;
\left\{
\begin{array}{rrrrrcc}
- 2v_1 & + & v_2 & & & = & 0 \qquad (1)\\
v_1 & - & v_2 & + & v_3 & = & 0 \qquad (2)\\
& & v_2 & - & 2v_3 & = & 0 \qquad (3)\\
\end{array}
\right. .
$
Equation (3) $\;\Ra\; $ $v_2 = 2v_3.\qquad$
Substitute into (1) and (2) $\; \Ra \; \left\{ \begin{array}{ccccc} -2v_1 & + & 2v_3 & = & 0 \\ v_1 & - & v_3 & = & 0 \\ \end{array} \right. . $
Find the eigenvalues and eigenvectors of $A= \left(\begin{array}{rrr}-3&1&0\\1&-2&1\\0&1&-3\end{array}\right).$
$
\Ra \;
\left\{
\begin{array}{rrrrrcc}
- 2v_1 & + & v_2 & & & = & 0 \qquad (1)\\
v_1 & - & v_2 & + & v_3 & = & 0 \qquad (2)\\
& & v_2 & - & 2v_3 & = & 0 \qquad (3)\\
\end{array}
\right. .
$
Equation (3) $\;\Ra\; $ $v_2 = 2v_3.\qquad$
Substitute into (1) and (2)
$\;
\Ra \;
\left\{
\begin{array}{ccccc}
-2v_1 & + & 2v_3 & = & 0 \\
v_1 & - & v_3 & = & 0 \\
\end{array}
\right. .
$
Then $v_1 = v_3,$
and $ \left( \begin{array}{c} v_1 \\ v_2 \\ v_3 \\ \end{array} \right) = \left( \begin{array}{c} v_3 \\ 2v_3 \\ v_3 \\ \end{array} \right) $ $ = v_3 \left( \begin{array}{c} 1 \\ 2 \\ 1 \\ \end{array} \right). $ The eigenvector is $ \left( \begin{array}{c} 1 \\ 2 \\ 1 \\ \end{array} \right). $
Find the eigenvalues and eigenvectors of $A= \left(\begin{array}{rrr}-3&1&0\\1&-2&1\\0&1&-3\end{array}\right).$
For $\lambda_2 = -3:$ $\ds \left(A-\lambda_2 I\right) \v=\mathbf 0 $ $ \;\Ra \; \left( \begin{array}{ccc} 0 & 1 & 0 \\ 1 & 1 & 1 \\ 0 & 1 & 0 \\ \end{array} \right) \left( \begin{array}{c} v_1 \\ v_2 \\ v_3 \\ \end{array} \right) =\left( \begin{array}{c} 0 \\ 0 \\ 0 \\ \end{array} \right) $
$
\Ra \;
\left\{
\begin{array}{ccccccc}
& & v_2 & & & = & 0 \qquad (1)\\
v_1 & + & v_2 & + & v_3 & = & 0 \qquad (2)\\
& & v_2 & & & = & 0 \qquad (3)\\
\end{array}
\right. .
$
$\;\Ra\; $ $v_2 = 0$ and $v_3 = -v_1.$
Thus $ \;\left( \begin{array}{c} v_1 \\ v_2 \\ v_3 \\ \end{array} \right) = \left( \begin{array}{c} v_1 \\ 0 \\ -v_1 \\ \end{array} \right) $ $ = v_1 \left( \begin{array}{r} 1 \\ 0 \\ -1 \\ \end{array} \right). $ The eigenvector is $ \left( \begin{array}{r} 1 \\ 0 \\ -1 \\ \end{array} \right). $
Find the eigenvalues and eigenvectors of $A= \left(\begin{array}{rrr}-3&1&0\\1&-2&1\\0&1&-3\end{array}\right).$
For $\lambda_3 = -4:$ $\ds \left(A-\lambda_3 I\right) \v=\mathbf 0 $ $ \;\Ra \; \left( \begin{array}{ccc} 1 & 1 & 0 \\ 1 & 2 & 1 \\ 0 & 1 & 1 \\ \end{array} \right) \left( \begin{array}{c} v_1 \\ v_2 \\ v_3 \\ \end{array} \right) =\left( \begin{array}{c} 0 \\ 0 \\ 0 \\ \end{array} \right) $
$
\Ra \;
\left\{
\begin{array}{ccccccc}
v_1 & + & v_2 & & & = & 0 \qquad (1)\\
v_1 & + & 2v_2 & + & v_3 & = & 0 \qquad (2)\\
& & v_2 & + & v_3 & = & 0 \qquad (3)\\
\end{array}
\right. .
$
From equation (1) $\,\Ra\, $ $v_1 = -v_2.\qquad$
Substitute into (2) $ \;\Ra v_2 + v_3 = 0, $ which is also equation (3).
Find the eigenvalues and eigenvectors of $A= \left(\begin{array}{rrr}-3&1&0\\1&-2&1\\0&1&-3\end{array}\right).$
$
\Ra \;
\left\{
\begin{array}{ccccccc}
v_1 & + & v_2 & & & = & 0 \qquad (1)\\
v_1 & + & 2v_2 & + & v_3 & = & 0 \qquad (2)\\
& & v_2 & + & v_3 & = & 0 \qquad (3)\\
\end{array}
\right. .
$
From equation (1) $\,\Ra\, $ $v_1 = -v_2.\qquad$
Substitute into (2)
$
\;\Ra
v_2 + v_3 = 0,
$
which is also equation (3).
Then $v_3 = -v_2,$
and $\; \left( \begin{array}{c} v_1 \\ v_2 \\ v_3 \\ \end{array} \right) = \left( \begin{array}{r} -v_2 \\ v_2 \\ -v_2 \\ \end{array} \right) $ $ = -v_2 \left( \begin{array}{r} 1 \\ -1 \\ 1 \\ \end{array} \right). $ The eigenvector is $ \left( \begin{array}{r} 1 \\ -1 \\ 1 \\ \end{array} \right). $
Find the eigenvalues and eigenvectors of $A= \left(\begin{array}{rrr}-3&1&0\\1&-2&1\\0&1&-3\end{array}\right).$
In summary: |
|
For a square matrix $A$:
1. $A$ and $A^T$ have the same eigenvalues.
$\text{det}\left(A- \lambda I\right)$
$= \text{det}\left(\left(A- \lambda I\right)^T\right)$
$\qquad \qquad = \text{det}\left(A^T- \lambda I^T\right)$
$= \text{det}\left(A^T- \lambda I\right).$
2. $A$ is singular if and only if $\lambda=0$ is an eigenvalue of $A.$
$A$ is singular if and only if $A\mathbf x = \mathbf 0$ has a non-trivial solution $\iff \lambda =0 $ is an eigenvalue of $A.$
3. If $\lambda$ is an eigenvalue of $A,$ then $\lambda^2$ is an eigenvalue of $A^2,$ and $1/{\lambda}$ is an eigenvalue of $A^{-1}$ when $A$ is non-singular.
Exercise: 📝
4. If $\lambda$ is an eigenvalue of $A,$ then $\lambda-m$ is an eigenvalue of $A-mI,$ for any scalar $m.$
Exercise: 📝
If $\lambda_1, \lambda_2,\ldots,\lambda_k$ are distinct eigenvalues of $A$, with corresponding eigenvectors $\v_1, \v_2,\ldots,\v_k$ (such that $\v_i$ corresponds to $\lambda_i$), then the set of eigenvectors $\{ \v_1,\v_2,\ldots,\v_k \}$ is linearly independent.
Proof: For $k=1,$ the set $\{\v_1\}$ is linearly independent. Easy! 😃
Assume that $\left\{\v_1, \v_2, \ldots, \v_n\right\}$ is linearly independent. This is our induction hypothesis! Easy as well! 😃
Now we have to prove the case for $n+1$. Set $$a_1\v_1+ a_2\v_2+ \cdots + a_{n+1}\v_{n+1}= \mathbf 0 \qquad (\star)$$
Proof:
Now we have to prove the case for $n+1$. Set $$a_1\v_1+ a_2\v_2+ \cdots + a_{n+1}\v_{n+1}= \mathbf 0 \qquad (\star)$$
Multiply ($\star$) by the matrix $A$ from the left to obtain \[ \begin{multline*} A\left(a_1\v_1+ \cdots + a_{n+1}\v_{n+1}\right) \\ = a_1\lambda_1\v_1+ \cdots + a_{n}\lambda_{n}\v_{n} + a_{n+1}\lambda_{n+1}\v_{n+1} = \mathbf 0\quad (1) \end{multline*} \] Now multiply ($\star$) by $\lambda_{n+1}\neq 0$ to get \[ a_1\lambda_{n+1}\v_1+ a_2\lambda_{n+1}\v_2+ \cdots + a_{n}\lambda_{n+1}\v_{n} + a_{n+1}\lambda_{n+1}\v_{n+1} = \mathbf 0\qquad (2) \]
Proof:
\[ a_1\lambda_1\v_1+ \cdots + a_{n}\lambda_{n}\v_{n} + a_{n+1}\lambda_{n+1}\v_{n+1} = \mathbf 0\qquad (1) \] \[ a_1\lambda_{n+1}\v_1+ a_2\lambda_{n+1}\v_2+ \cdots + a_{n}\lambda_{n+1}\v_{n} + a_{n+1}\lambda_{n+1}\v_{n+1} = \mathbf 0\qquad (2) \]
Substracting (2) - (1) we obtain \[ a_1\left(\lambda_{n+1} -\lambda_1\right)\v_1+ a_2\left(\lambda_{n+1} -\lambda_2\right)\v_2+\cdots + a_n\left(\lambda_{n+1} -\lambda_n\right)\v_n=\mathbf 0. \]
Proof:
Substracting (2) - (1) we obtain \[ a_1\left(\lambda_{n+1} -\lambda_1\right)\v_1+ a_2\left(\lambda_{n+1} -\lambda_2\right)\v_2+\cdots + a_n\left(\lambda_{n+1} -\lambda_n\right)\v_n=\mathbf 0. \]
Recall that $\left\{\v_1, \v_2, \ldots, \v_n\right\}$ is linearly independent by induction hypotesis.
Then $a_1\left(\lambda_{n+1} -\lambda_1\right) = a_2\left(\lambda_{n+1} -\lambda_2\right) = \cdots = a_n\left(\lambda_{n+1} -\lambda_n\right)=0.$ Since $\{\lambda_i\}$ are distinct, then $$a_1=a_2=\cdots = a_n=0.$$
Proof:
So we have that $a_i=0$ for $i=1,\ldots,n$. Substituting in $$a_1\v_1+ a_2\v_2+ \cdots + a_{n+1}\v_{n+1}= \mathbf 0,\quad (\star)$$ we have that $a_{n+1}\v_{n+1}= \mathbf 0\;$ $\;\Ra\; a_{n+1} = 0.$
Thus $\left\{\v_1, \v_2, \ldots, \v_{n+1}\right\}$ is linearly independent. Hence, the set of eigenvectors $$\left\{\v_1, \v_2, \ldots, \v_{k}\right\}$$ is linearly independent. $\quad\blacksquare$
Applet in GeoGebra site: Click here