48 MYOUNGJEAN BAE

10. Lecture 10
10.1. Application 3: Eigenvalues and Eigenvectors (an easy introduction §4.4) Cont’d.

10.1.1. A quick introduction.

Definition (Eigenvalues and Eigenvectors). Given a matrix A ∈ Rn×n , if a nonzero vector
v ∈ Cn×1 satisfies
Av = λv
for some λ ∈ C, the scalar λ is called en eigenvalue, and the vector v is called an eigenvector
corresponding to λ of the matrix A.

Note. There are numerous applications of eigenvalues and eigenvectors. Solving a linear ordinary
differential equation of higher order, or a system of first order ordinary differential equations is
one of them. We shall discuss about this in more details later.

Example 10.1. Find all the eigenvalues and corresponding eigenvectors of

 
1 3
A=
4 2

Example 10.2. Find all the eigenvalues and corresponding eigenvectors of

⎡ ⎤
0 −1 0
B=⎣ 0 0 1⎦
−4 −17 8
 INTRODUCTION TO LINEAR ALGEBRA 49

Definition. For A ∈ Rn×n , the equation

det(A − λI) = 0
is called the characteristic equation of A.

Suppose that λ1 , λ2 , · · · , λk are all distinct roots of the characteristic equation det(A − λI) = 0.
Then, we can write as
det(λI − A) = (λ − λ1 )m1 (λ − λ2 )m2 · · · (λ − λk )mk
for positive integers m1 , m2 , · · · , mk with
m1 + m2 + · · · + mk = n.
Definition. For each i = 1, 2, · · · , k, mi is called the multiplicity of the eigenvalue λi .

Example. Find all the eigenvalues of

⎡ ⎤
1 0 0 0 0
⎢ −1 −2 0 0 0⎥
⎢ 1 ⎥
A=⎢
⎢ 31 −2 1 0 0⎥⎥,
⎣ −2 1 −2 0⎦
100
1 1 1 1 1
and determine their multiplicities.
50 MYOUNGJEAN BAE

10.1.2. Further discussion on eigenvalues of A ∈ R2×2 .

 
a a
For A = 11 12 ,
a21 a22
a11 − λ a12
det(A − λI) =
a21 a22 − λ
= λ2 − (a11 + a22 )λ + (a11 a22 − a12 a21 )
= λ2 − λTrA + det A
Put
T := TrA, D := det A.
Equation for eigenvalue λ:
λ2 − T λ + D = 0. (Chracteristic equation of A)
Let λ1 and λ2 be two eigenvalues of A. (λ1 = λ2 is possible.) Then,
T = λ 1 + λ2 , D = λ1 λ2 .

More generally, the following theorem holds:

Theorem 10.3. For A ∈ Rn×n , let λ1 , λ2 , · · · , λn be all the eigenvalues of A. Then,
TrA = λ1 + λ2 + · · · + λn , det A = λ1 λ2 · · · λn .

Theorem 10.4. Let A ∈ R2×2 . Define

μ := T 2 − 4D
(a) If μ > 0, then A has two distinct real eigenvalues;
(b) If μ = 0, then A has one repeated real eigenvalues;
(c) If μ < 0, then A has two complex eigenvalues.

Example 10.5.      
2 2 0 −1 2 3
A= , B= , C=
−1 5 1 2 −3 2
 INTRODUCTION TO LINEAR ALGEBRA 51

Lemma 10.6. Given A ∈ Rn×n , let λ ∈ R be an eigenvalue of A. Define

Eλ := {v ∈ Rn×1 : Av = λv}.
Then E is a subspace of Rn×1 .
Definition. The set Eλ is called the eigenspace corresponding to λ.
Example 10.7. Find all the eigenvalues, and the corresponding eigenspaces.
 
3 2
A=
2 3
Example 10.8. Find all the eigenvalues, and the corresponding eigenspaces.
 
1 0
I=
0 1
52 MYOUNGJEAN BAE

10.2. Matrices as linear transformation (§6.1).

Quick review of Lecture 1

Definition. Let V and W be two real vector spaces. A mapping
L:V→W
is called a linear mapping if
L(c1 v1 + c2 v2 ) = c1 L(v1 ) + c2 L(v2 )
for any v1 , v2 ∈ V, and any c1 , c2 ∈ R.
Theorem*. Fix n, m ∈ N. If L : Rn → Rm is a linear mapping, then it has a m × n matrix
representation that is,
⎡ ⎤
a11 · · · a1n
⎢ .. ⎥ (=: [a ]) with a ∈ R ∀i = 1, · · · m, j = 1, · · · , n.
L = ⎣ ... ..
. . ⎦ ij ij
am1 · · · amn
Question. How to find such [aij ] explicitly?
 INTRODUCTION TO LINEAR ALGEBRA 53

Definition. (i) Given n, m ∈ N, a linear mapping T : Rn → Rm is called a linear transfor-

mation.
(ii) A linear transformation T : Rn → Rn is called a linear operator on Rn .

Definition (Standard unit vector in Rn ).

Kronecker delta:
1 if i = j
δij =
0 if i =
 j

For each k = 1, 2, · · · , n, define

⎡ ⎤ ⎡ ⎤
δ1k 0
⎢ δ2k ⎥ ⎢0⎥
⎢ . ⎥ ⎢.⎥
⎢ . ⎥ ⎢.⎥
⎢ . ⎥ ⎢.⎥
ek = ⎢ ⎥ = ⎢ ⎥
⎢δkk ⎥ ⎢1⎥
⎢ . ⎥ ⎢.⎥
⎣ .. ⎦ ⎣ .. ⎦
δnk 0

Observation:
(a) The set {e1 , e2 , · · · , en } is linearly independent, that is,

c1 e1 + c2 e2 + · · · + cn en = 0 ⇔ c1 = c2 = · · · = cn = 0.
⎡ ⎤
v1
⎢ v2 ⎥
⎢ ⎥
(b) ∀v = ⎢ .. ⎥ ∈ Rn (= Rn×1 ),
⎣.⎦
vn

v = v 1 e1 + v 2 e2 + · · · + v n en ,

that is,
v ∈ span{e1 , e2 , · · · , en }.

Definition. The set {e1 , e2 , · · · , en } is called the standard basis of Rn .

Observation: Let T : Rn → Rm be a linear transformation.

(i) For each k = 1, 2, · · · , n, put
zk := T (ek ).
Let us write as
⎡ ⎤
ζ1k
⎢ ζ2k ⎥
⎢ ⎥
zk = ⎢ .. ⎥ .
⎣ . ⎦
ζmk
54 MYOUNGJEAN BAE
⎡ ⎤
v1
⎢ v2 ⎥
⎢ ⎥
(ii) ∀v = ⎢ .. ⎥,
⎣.⎦
vn
T (v) = T (v1 e1 + v2 e2 + · · · + vn en )
= v1 T (e1 ) + v2 T (e2 ) + · · · + vn T (en )
= v 1 z 1 + v 2 z2 + · · · + v n z n
⎡ ⎤
⎡ ⎤ v1
··· ⎢ v2 ⎥
⎢ ⎥
= ⎣z1 z2 · · · zn ⎦ ⎢ .. ⎥
··· ⎣.⎦
vn
⎡ ⎤
ζ11 ζ12 · · · ζ1n
⎢ ζ21 ζ22 · · · ζ2n ⎥
⎢ ⎥ m×1
= ⎢ .. .. .. ⎥v ∈ R (= Rm )
⎣ . . . ⎦
ζm1 ζm2 · · · ζmn
 
∈Rm×n

Theorem 10.9. Any linear transformation T : Rn → Rm can be expressed a m by n matrix.