Math 2270 Lecture 30: Eigenvalues

Dylan Zwick

Fall 2012

This lecture covers section 6.1 of the textbook.

When a square matrix A acts upon a vector x, it generally outputs a
new vector Ax. Usually this new vector will be stretched and rotated. For
example, if we take the matrix

A=(
)
and apply it to the vector
() we get:

(11 oN (i
1

1
If we apply A to the vector
( ) we get:

1 (1(2
1

However, if we apply A to the vector

() we get:

1 1’

‘4
7

2
 The matrix A leaves the vector
( ) unchanged! In particular, it does
not rotate the vector. When a matrix A acts upon a vector and does not
rotate it, we have Ax = Ax, where A is a scaling factor. In our example A =
1. We call such a vector an eigenvector for the matrix A, and the associated
scaling factor A an eigenvnlue. This section, and in fact this chapter, explore
eigenvectors and eigenvalues.
Note we’ll assume throughout this lecture, and in fact throughout all
lectures about chapter 6, that all matrices are square.
The assigned problems for this section are:

Section 6.1 2, 3, 5, 16, 17

-

1 Eigenvectors and Eigenvalues

The basic equation for eigenvectors and eigenvalues is

Ax=Ax

where A is a matrix and A is a number. One property we see right away

is

x
2
A = AAx = x
2
A

and in general

Ax = x.
7
A

A are the same as the eigenvectors of A, while

So, the eigenvectors of 1
7 are the eigenvalues for A raised to the nth power.
the eigenvalues for A
OK, now let’s see how we actually calculate what these eigenvalues
and eigenvectors are. The first thing I want to stress is that we calculate
the eigenvectors from the eigenvalues, so the first thing we want to do is
calculate the eigenvalues. How do we do this? Well, we note that if

3
 Ax = Ax

then

(A — AI)x = 0

So, the matrix (A — Al) has a nontrivial nulispace, and therefore must
be singular. So,

det(A — Al) = 0.

So, if A is an eigenvalue of A then det(A Al) = 0. It turns out that this

—

equation can be used to calculate every eigenvalue. The equation det(A —

Al) 0 will be a polynomial equation in A, and its roots will give us all the
eigenvalues. We call this polynomial equation the characteristic equation of
A.
For example, the characteristic equation of the matrix

A=(
)
is

1—A =(1_A)(4_A)_4=A2_5A.
4—A

2 5A = A(A 5) are A = 0 and A = 5,

The roots of the polynomial A — —

and these will be the eigenvalues of A. From these we can compute the
eigenvectors

A=O.(A-OI)x=(
)(‘)()
yields the solution
( ) ( ),=
which is an eigenvector for

4
 A 5. (A — 51)x
= ( 2 (y (0
-1)y2)O

(1
yields the solution (
Y2J
‘ =

j
2
, which is an eigenvector for A =

5.

Note that these eigenvectors are not unique. In fact, any non-zero mul
tiple cx (c 0) of an eigenvector is another eigenvector.
Oh, and we should probably mention right now that if x = 0 then
Ax = Ax for any A. So, we have to put in a qualifier that the zero vector is

never an eigenvector. Please keep in mind that the number 0 can certainly
be an eigenvalue.
Example Find the eigenvalues and associated eigenvectors for the ma
-

trix

/2 —2 3
A=( 0 3 —2
\o —1 2

L-A -

0 (-)(5-) (z-)
O -I z-
-A
(/
-A
7
3 A IHA - (-i) (A’+)
- (A-I)(-)(A-L) iy VC{
I/ f

Z3)f;i)(

( O
6-i
5
(
i) E (JLJ)) () ) -? -
Jyfr

7)
f7

iJ () IeDflP
2 Some Facts About Eigenvectors
First, some bad news. We cannot use elimination to calculate eigenvalues.
Sorry. If we use elimination to convert a a matrix A into an upper triangu
lar matrix U, the eigenvalues of U could be different than the eigenvalues
of A. However, the two will not be completely unrelated.
What relates them is the amazing fact that the determinant of a matrix
is equal to the product of its eigenvalues. That is to say, if ),
.. are
the eigenvalues of a matrix A, then

det(A) .•
2
A
1

So, because A and U have the same determinant, the product of their
eigenvalues will be the same.
Example Calculate the determinant for the matrix A from the previous
-

example, and verify that the determinant is, in fact, equal to the product
of the eigenvalues.

/2 —2 3
A=f 0 3 —2
—1 2

6
 Finally, we note that the trace of a matrix is defined as being the sum
of the diagonal elements.

trace(A) = 1 +a
a 22 + +a

The trace of a matrix will be equal to the sum of the eigenvalues of the
matrix

trace(A) = 1 + 2 + + .

Example Verify this for the matrix

-

/2 —2 3
A=f 0 3 —2
0 —1 2