Lecture 7

Eigenvalues and Eigenvectors

7.1 Eigenvalues and Eigenvectors

7.2 Diagonalization
7.3 Symmetric Matrices and Orthogonal Diagonalization

7.1 Eigenvalues and Eigenvectors

n Eigenvalue problem:
If A is an n×n matrix, do there exist nonzero vectors x in Rn
such that Ax is a scalar multiple of x？

n Eigenvalue and eigenvector: n Geometrical Interpretation

A：an n×n matrix
λ：a scalar
x： a nonzero vector in Rn
Eigenvalue

Ax = lx
Eigenvector

1
 n Ex 1: (Verifying eigenvalues and eigenvectors)
é2 0 ù é1ù é0 ù
A=ê ú x1 = ê ú x2 = ê ú
ë0 - 1û ë0 û ë1û
Eigenvalue
é2 0 ù é1ù é2ù é1ù
Ax1 = ê = = 2 êë0úû = 2 x1
ë0 - 1úû êë0úû êë0úû
Eigenvector

Eigenvalue
é 2 0 ù é0 ù é 0 ù é0 ù
Ax2 = ê ú ê ú = ê ú = -1ê ú = (-1) x2
ë 0 - 1 1
ûë û ë û - 1 1
ë û

Eigenvector

n Thm 7.1: (The eigenspace of A corresponding to λ)

If A is an n× n matrix with an eigenvalue λ, then the set of all
eigenvectors of λ together with the zero vector is a subspace of
Rn. This subspace is called the eigenspace of λ .
Pf:
x1 and x2 are eigenvectors corresponding to λ
(i.e. Ax1 = lx1 , Ax2 = lx2 )
(1) A( x1 + x2 ) = Ax1 + Ax2 = lx1 + lx2 = l ( x1 + x2 )
(i.e. x1 + x2 is an eigenvector corresponding to λ)
(2) A(cx1 ) = c( Ax1 ) = c(lx1 ) = l (cx1 )
(i.e. cx1 is an eigenvector corresponding to l )

3
 n Ex 3: (An example of eigenspaces in the plane)
Find the eigenvalues and corresponding eigenspaces of
é - 1 0ù
A=ê ú
ë 0 1û
Sol:
If v = ( x, y )
é - 1 0ù é x ù é - x ù
Av = ê ú ê yú = ê y ú
ë 0 1 ûë û ë û
For a vector on the x-axis Eigenvalue l1 = -1
é - 1 0ù é x ù é - x ù é xù
=
ê 0 1ú ê0 ú ê 0 ú = -1ê0 ú
ë ûë û ë û ë û

For a vector on the y-axis Eigenvalue l2 = 1

é - 1 0ù é 0 ù é 0 ù é 0 ù
ê 0 1ú ê y ú = ê y ú = 1ê y ú
ë ûë û ë û ë û

Geometrically, multiplying a vector (x, y)

in R2 by the matrix A corresponds to a
reflection in the y-axis.

The eigenspace corresponding to l1 = -1 is the x-axis.

The eigenspace corresponding to l2 = 1 is the y-axis.

5
 n Thm 7.2: (Finding eigenvalues and eigenvectors of an n × n matrix
A ) Let A is an n×n matrix.

(1) An eigenvalue of A is a scalar λ such that det(lI - A) = 0.

(2) The eigenvectors of A corresponding to λ are the nonzero
solutions of (lI - A) x = 0 .
n Note:
Ax = lx Þ (lI - A) x = 0 (homogeneous system)
If (lI - A) x = 0 has nonzero solutions iff det(lI - A) = 0.
n Characteristic polynomial of A:

det(lI - A) = (lI - A) = ln + cn -1ln -1 + ! + c1l + c0

n Characteristic equation of A:
det(lI - A) = 0

n Ex 4: (Finding eigenvalues and eigenvectors)

é2 - 12ù
A=ê ú
ë1 - 5 û

Sol: Characteristic equation:

l -2 12
det(lI - A) =
-1 l +5
= l2 + 3l + 2 = (l + 1)(l + 2) = 0
Þ l = -1, - 2

Eigenvalues : l 1 = -1, l2 = -2

7
 é- 3 12ù é x1 ù é0ù
(1)l1 = -1 Þ (l1I - A) x = ê ú ê x ú = ê0 ú
ë - 1 4 ûë 2 û ë û
é- 3 12ù é1 - 4ù é x1 ù é4t ù é4ù
!ê ~
ú ê0 0 ú Þ ê x ú = ê t ú = t ê1 ú , t ¹ 0
ë - 1 4 û ë û ë 2û ë û ë û

é- 4 12ù é x1 ù é0ù
(2)l2 = -2 Þ (l2 I - A) x = ê ú ê x ú = ê0 ú
ë - 1 3 ûë 2 û ë û
é- 4 12ù é1 - 3ù é x1 ù é3t ù é3ù
!ê ~
ú ê0 0 ú Þ ê x ú = ê t ú = t ê1ú, t ¹ 0
ë - 1 3 û ë û ë 2û ë û ë û

Check : Ax = l x i

n Ex 5: (Finding eigenvalues and eigenvectors)

Find the eigenvalues and corresponding eigenvectors for
the matrix A. What is the dimension of the eigenspace of
each eigenvalue?

é2 1 0ù
A = ê0 2 0 ú
ê ú
ë0 0 2û
Sol: Characteristic equation:
l - 2 -1 0
lI - A = 0 l -2 0 = (l - 2) 3 = 0
0 0 l -2
Eigenvalue: l = 2

9
 The eigenspace of A corresponding to l = 2 :
é0 - 1 0ù é x1 ù é0ù
(lI - A) x = êê0 0 0úú êê x2 úú = êê0úú
êë0 0 0úû êë x3 úû êë0úû
é0 - 1 0 ù é0 1 0 ù é x1 ù é s ù é1ù é0ù
! êê0 0 0úú ~ êê0 0 0úú Þ êê x2 úú = êê0úú = s êê0úú + t êê0úú, s, t ¹ 0
êë0 0 0úû êë0 0 0úû êë x3 úû êë t úû êë0úû êë1úû
ì é1ù é0ù ü
ï ê ú ê ú ï
ís ê0ú + t ê0ú s, t Î R ý : the eigenspace of A corresponding to l = 2
ï ê0ú ê1ú ï
î ë û ë û þ
Thus, the dimension of its eigenspace is 2.

10

n Notes:
(1) If an eigenvalue λ1 occurs as a multiple root (k times) for
the characteristic polynominal, then we say λ1 has
multiplicity k.
(2) The multiplicity of an eigenvalue is greater than or equal
to the dimension of its eigenspace.

11
 n Ex 6：Find the eigenvalues of the matrix A and find a basis
for each of the corresponding eigenspaces.
é1 0 0 0 ù
ê0 1 5 - 10ú
A=ê ú
ê 1 0 2 0 ú
ëê1 0 0 3 ûú
Sol: Characteristic equation:
l -1 0 0 0
0 l -1 - 5 10
lI - A =
-1 0 l -2 0
-1 0 0 l -3
= (l - 1) 2 (l - 2)(l - 3) = 0
Eigenvalues : l1 = 1, l2 = 2, l3 = 3

12

(1)l1 = 1 é0 0 0 0 ù é x1 ù é0ù
ê0 0 - 5 10 úú êê x2 úú êê0úú
Þ (l1I - A) x = ê =
ê- 1 0 - 1 0 ú ê x3 ú ê0ú
ê úê ú ê ú
ë- 1 0 0 - 2 û ë x4 û ë0 û
é0 0 0 0 ù é1 0 0 2 ù é x1 ù é- 2t ù é0 ù é - 2 ù
ê0
ê 0 - 5 10 úú êê0 0 1 - 2úú êê x2 úú êê s úú ê1ú ê 0 ú
~ Þ = = s ê ú + t ê ú, s, t ¹ 0
ê- 1 0 - 1 0 ú ê0 0 0 0 ú ê x3 ú ê 2t ú ê0 ú ê 2 ú
ê ú ê ú ê ú ê ú ê ú ê ú
ë- 1 0 0 - 2 û ë0 0 0 0 û ë x4 û ë t û ë0 û ë 1 û
ì é0 ù é - 2 ù ü
ïê ú ê ú ï
ï 1 0 ï
Þ íê ú, ê ú ý is a basis for the eigenspace
ïê0ú ê 2 ú ï of A corresponding to l = 1
ïîêë0úû êë 1 úû ïþ

13
 (2)l2 = 2 é1 0 0 0 ù é x1 ù é0ù
ê0 1 - 5 10 úú êê x2 úú êê0úú
Þ (l2 I - A) x = ê =
ê- 1 0 0 0 ú ê x3 ú ê0ú
ê úê ú ê ú
ë- 1 0 0 - 1û ë x4 û ë0û
é1 0 0 0 ù é1 0 0 0ù é x1 ù é 0 ù é0ù
ê 0 1 - 5 10 ú ê0 1
ê ú~ê - 5 0úú êê x2 úú êê5t úú êê5úú
Þ = =t , t ¹0
ê- 1 0 0 0 ú ê0 0 0 1ú ê x3 ú ê t ú ê1ú
ê ú ê ú ê ú ê ú ê ú
ë- 1 0 0 - 1û ë0 0 0 0 û ë x4 û ë 0 û ë0 û
ì é0 ù ü
ïê ú ï
ï 5 ï
Þ íê ú ý is a basis for the eigenspace
ïê1ú ï of A corresponding to l = 2
ïîêë0úû ïþ

14

(3)l3 = 3 é2 0 0 0 ù é x1 ù é0ù
ê0 2 - 5 10úú êê x2 úú êê0úú
Þ (l3 I - A) x = ê =
ê- 1 0 1 0 ú ê x3 ú ê0ú
ê úê ú ê ú
ë- 1 0 0 0 û ë x4 û ë0 û
é2 0 0 0 ù é1 0 0 0ù é x1 ù é 0 ù é 0 ù
ê0
ê 2 - 5 10úú êê0 1 0 5úú êê x2 úú êê- 5t úú êê- 5úú
~ Þ = =t , t¹0
ê- 1 0 1 0 ú ê0 0 1 0ú ê x3 ú ê 0 ú ê 0 ú
ê ú ê ú ê ú ê ú ê ú
ë- 1 0 0 0 û ë0 0 0 0 û ë x4 û ë t û ë 1 û
ìé 0 ù ü
ïê ú ï
ï -5 ï
Þ íê ú ý is a basis for the eigenspace
ïê 0 ú ï of A corresponding to l = 3
ïîêë 1 úû ïþ

15
 n Thm 7.3: (Eigenvalues of triangular matrices)
If A is an n× n triangular matrix, then its eigenvalues are
the entries on its main diagonal.
n Ex 7: (Finding eigenvalues for diagonal and triangular matrices)
é - 1 0 0 0 0ù
é2 0 0ù ê 0 2 0 0 0ú
ê ú
(a ) A = ê- 1 1 0 ú (b) A = ê 0 0 0 0 0ú
ê 0 0 0 - 4 0ú
êë 5 3 - 3úû êë 0 0 0 0 3úû
Sol: l -2 0 0
( a ) lI - A = 1 l -1 0 = (l - 2)(l - 1)(l + 3)
-5 -3 l +3
l1 = 2, l2 = 1, l3 = -3
(b) l1 = -1, l2 = 2, l3 = 0, l4 = -4, l5 = 3

16

n Eigenvalues and eigenvectors of linear transformations:

A linear transformation T : V ® V can always be represented by

a matrix A . So, we can imitate the notion of eigenvectors l and
eigenvector x of a matrix A for the linear transformation T .

A number λ is called an eigenvalue of a linear transformation

T :V → V if there is a nonzero vector x such that T (x) = λ x.
The vector x is called an eigenvector of T corresponding to λ,
and the set of all eigenvectors of λ (with the zero vector) is
called the eigenspace of λ.

17
 n Ex 8: (Eigenvalues and eigenspaces of a transformation)
Consider the linear transformation T from R3 into R3 given by
T(x1,x2,x3)=(x1 + 3x2, 3x1 + x2, -2x3).

Find its eigenvalues and eigenvectors.

Call A the standard matrix of the transformation T. This is the

matrix having columns being coordinate columns of T(e1),
T(e2) and T(e3). Here B={e1 = (1,0,0), e2 = (0,1,0), e3 =
(0,0,1)} is the standard basis of R3. This is computed as
follows.

18

n Ex 8: (Eigenvalues and eigenspaces of a transformation)

Find the eigenvalues and corresponding eigenspaces
é1 3 0 ù
A = ê3 1 0 ú.
ê ú
êë0 0 - 2úû
Sol: l -1 -3 0
lI - A = - 3 l -1 0 = (l + 2) 2 (l - 4)
0 0 l+2
eigenvalues : l1 = 4, l2 = -2
The eigenspaces for these two eigenvalues are as follows.
B1 = {(1, 1, 0)} Basis for l1 = 4
B2 = {(1, - 1, 0), (0, 0, 1)} Basis for l2 = -2

19
 n Notes:
(1) Let T:R 3 ® R 3 be the linear transformation whose standard matrix
is A in Ex. 8, and let B' be the basis of R 3 made up of three linear
independent eigenvectors found in Ex. 8. Then A' , the matrix of T
relative to the basis B' , is diagonal.
é4 0 0ù
A' = ê0 - 2 0 ú
B ' = {(1, 1, 0), (1, - 1, 0), (0, 0, 1)} ê ú
ë0 0 - 2û
Eigenvectors of A Eigenvalues of A

(2) The main diagonal entries of the matrix A' are the eigenvalues of A.

20

7.2 Diagonalization
n Diagonalization problem:
For a square matrix A, does there exist an invertible matrix P
such that P-1AP is diagonal?
n Diagonalizable matrix:
A square matrix A is called diagonalizable if there exists an
invertible matrix P such that P-1AP is a diagonal matrix.
(P diagonalizes A)
n Notes:
(1) If there exists an invertible matrix P such that B = P -1 AP,
then two square matrices A and B are called similar.
(2) The eigenvalue problem is related closely to the
diagonalization problem.

21
 n Thm 7.4: (Similar matrices have the same eigenvalues)
If A and B are similar n× n matrices, then they have the
same eigenvalues.
Pf:
A and B are similar Þ B = P -1 AP

lI - B = lI - P -1 AP = P -1lIP - P -1 AP = P -1 (lI - A) P
= P -1 lI - A P = P -1 P lI - A = P -1 P lI - A
= lI - A
A and B have the same characteristic polynomial.
Thus A and B have the same eigenvalues.

22

n Ex 1: (A diagonalizable matrix)
é1 3 0 ù
A = ê3 1 0 ú
ê ú
ë0 0 - 2 û
Sol: Characteristic equation:
l -1 - 3 0
lI - A = - 3 l - 1 0 = (l - 4)(l + 2) 2 = 0
0 0 l+2
Eigenvalues : l1 = 4, l2 = -2, l3 = -2
⎡ 1 ⎤
⎢ ⎥
(1)λ = 4 ⇒ Eigenvector:p1 = ⎢ 1 ⎥
⎢⎣ 0 ⎥⎦

23
 ⎡ 1 ⎤ ⎡ 0 ⎤
⎢ ⎥ ⎢ ⎥
(2)λ = −2 ⇒ Eigenvector : p2 = ⎢ −1 ⎥ , p3 = ⎢ 0 ⎥ (See Ex.5)
⎢⎣ 0 ⎥⎦ ⎢⎣ 1 ⎥⎦
é1 1 0ù é4 0 0ù
P = [ p1 p2 p3 ] = êê1 - 1 0úú Þ P -1 AP = êê0 - 2 0 úú
êë0 0 1úû êë0 0 - 2úû
n Notes: é1 1 0ù é- 2 0 0 ù
(1) P = [ p2 p1 p3 ] = êê- 1 1 0úú Þ P -1 AP = êê 0 4 0 úú
êë 0 0 1úû êë 0 0 - 2úû
é1 0 1ù é- 2 0 0ù
(2) P = [ p2 p3 p1 ] = êê- 1 0 1ú Þ P -1 AP = êê 0 - 2 0úú
ú
êë 0 1 0úû êë 0 0 4úû

24

n Thm 7.5: (Condition for diagonalization)

An n× n matrix A is diagonalizable if and only if
it has n linearly independent eigenvectors.
Pf:
(Þ) A is diagonalizable
there exists an invertible P s.t. D = P -1 AP is diagonal
Let P = [ p1 p2 ! pn ] and D = diag (l1 , l2 , ! , ln )
él1 0 ! 0ù
ê 0 l2 ! 0ú
PD = [ p1 p2 ! pn ]ê ú
ê" " # "ú
êë 0 0 ! ln úû
= [l1 p1 l2 p2 ! ln pn ]

25
 AP = A[ p1 p2 ! pn ] = [ Ap1 Ap2 ! Apn ]

" AP = PD
\ Api = li pi , i = 1, 2, ! , n
(i.e. the column vector pi of P are eigenvectors of A)
" P is invertible Þ p1 , p2 , ! , pn are linearly independent.
\ A has n linearly independent eigenvectors.

(Ü) A has n linearly independent eigenvectors p1 , p2 , ! pn

with corresponding eigenvalues l1 , l2 , ! ln
i.e. Api = li pi , i = 1, 2, !, n
Let P = [ p1 p2 ! pn ]

26

AP = A[ p1 p2 ! pn ]
= [ Ap1 Ap2 ! Apn ]
= [l1 p1 l2 p2 ! ln pn ]
él1 0 ! 0 ù
ê0 l ! 0 ú
= [ p1 p2 ! pn ]ê 2 ú = PD
ê" " # " ú
ê ú
ë 0 0 ! l nû

" p1 , p1 , ! , pn are linearly independent Þ P is invertible

\ P -1 AP = D
Þ A is diagonalizable
Note: If n linearly independent eigen vectors do not exist, then the
n×n matrix A is not diagonalizable.

27
 n Ex 4: (A matrix that is not diagonalizable)
Show that the following matrix is not diagonalizable.
é1 2 ù
A=ê ú
ë0 1 û
Sol: Characteristic equation:
l -1 - 2
lI - A = = (l - 1) 2 = 0
0 l -1
Eigenvalue : l1 = 1

é0 - 2 ù é0 1 ù é1ù
lI - A = I - A = ê ~
ú ê ú Þ Eigenvector : p1 = ê0 ú
ë0 0 û ë0 0 û ë û
A does not have two (n=2) linearly independent eigenvectors,
so A is not diagonalizable.

28

n Steps for diagonalizing an n×n square matrix:

Step 1: Find n linearly independent eigenvectors p1 , p2 , ! , pn
for A with corresponding eigenvalues l1 , l2 , ! , ln
Step 2: Let P = [ p1 p2 ! pn ]
Step 3:
él1 0 " 0 ù
ê0 l " 0 ú
P AP = D = ê
-1 2 ú, where Ap = l p , i = 1, 2, ! , n
ê# # $ # ú i i i

ê ú
ë 0 0 " l nû
Note:
The order of the eigenvalues used to form P will determine the order
in which the eigenvalues appear on the main diagonal of D.

29
 n Ex 5: (Diagonalizing a matrix)
é 1 - 1 - 1ù
A=ê 1 3 1ú
ê ú
ëê - 3 1 - 1úû
Find a matrix P such that P -1 AP is diagonal.

Sol: Characteristic equation:

l -1 1 1
lI - A = - 1 l - 3 - 1 = (l - 2)(l + 2)(l - 3) = 0
3 -1 l +1

Eigenvalues : l1 = 2, l2 = -2, l3 = 3

30

l1 = 2 é 1 1 1 ù é1 0 1ù
Þ l1I - A = ê- 1 - 1 - 1ú ~ ê0 1 0ú
ê ú ê ú
ë 3 - 1 3 û ë0 0 0 û
é x1 ù é- t ù é- 1ù é- 1ù
Þ êê x2 úú = êê 0 úú = t êê 0 úú Þ Eigenvector : p1 = êê 0 úú
êë x3 úû êë t úû êë 1 úû êë 1 úû
l2 = -2
é- 3 1 1 ù é1 0 - 14 ù
Þ l2 I - A = ê - 1 - 5 - 1ú ~ ê0 1 14 ú
ê ú ê ú
ë 3 - 1 - 1û ë0 0 0 û
é x1 ù é 14 t ù é1ù é1ù
Þ êê x2 úú = êê- 14 t úú = 14 t êê- 1úú Þ Eigenvector : p2 = êê- 1úú
êë x3 úû êë t úû êë 4 úû êë 4 úû

31
 l3 = 3 é 2 1 1 ù é1 0 1 ù
Þ l3 I - A = ê- 1 0 - 1ú ~ ê0 1 - 1ú
ê ú ê ú
ë 3 - 1 4 û ë0 0 0 û
é x1 ù é- t ù é- 1ù é- 1ù
Þ êê x2 úú = êê t úú = t êê 1 úú Þ Eigenvector : p3 = êê 1 úú
êë x3 úû êë t úû êë 1 úû êë 1 úû
é- 1 1 - 1ù
Let P = [ p1 p2 p3 ] = êê 0 - 1 1 úú
êë 1 4 1 úû
é 2 0 0ù
Þ P -1 AP = êê0 - 2 0úú
êë0 0 3úû

32

n Notes: k is a positive integer

éd1 0 ! 0ù éd1k 0 ! 0ù
ê0 d2 ! 0ú ê0 d 2k ! 0ú
(1) D = ê ú Þ D k
= ê ú
ê" " # "ú ê" " # "ú
ëê 0 0 ! d n úû êë 0 0 ! d nk úû
(2) D = P -1 AP
Þ D k = ( P -1 AP) k
= ( P -1 AP)( P -1 AP) × × × ( P -1 AP)
= P -1 A( PP -1 ) A( PP -1 ) × × × ( PP -1 ) AP
= P -1 AA × × × AP
= P -1 Ak P
\ Ak = PD k P -1

33
 n Thm 7.6: (Sufficient conditions for diagonalization)
If an n× n matrix A has n distinct eigenvalues, then the
corresponding eigenvectors are linearly independent and
A is diagonalizable.

34

n Ex 7: (Determining whether a matrix is diagonalizable)

é1 - 2 1 ù
A = ê0 0 1ú
ê ú
êë0 0 - 3úû

Sol: Because A is a triangular matrix,

its eigenvalues are the main diagonal entries.
l1 = 1, l2 = 0, l3 = -3
These three values are distinct, so A is diagonalizable. (Thm.7.6)

35
 Diagonalize a linear transformation: Suppose that T: V → V is
a LT. Find a basis B of V such that the matrix of T relative to
B is a diagonal matrix.

n Ex 8: (Finding a diagonalizing matrix for a linear transformation)

Let T:R 3 ® R 3 be the linear transformation given by
T ( x1,x2 ,x3 ) = ( x1 - x2 - x3 , x1 + 3 x2 + x3 , - 3 x1 + x2 - x3 )
Find a basis B for R 3 such that the matrix for T
relative to B is diagonal.

Sol: We need to find a basis of V=R3, consisting of all

eigenvectors

36

The standard matrix for T is given by

é 1 - 1 - 1ù
A = [T (e1 ) T (e2 ) T (e3 )] = êê 1 3 1 úú
êë- 3 1 - 1úû

It is the matrix in the Ex. 5. Recall from Ex. 5, characteristic

equation of A is
l -1 1 1
lI - A = - 1 l - 3 - 1 = (l - 2)(l + 2)(l - 3) = 0
3 -1 l +1

Eigenvalues : l1 = 2, l2 = -2, l3 = 3

37
 l1 = 2 é 1 1 1 ù é1 0 1ù
Þ l1I - A = ê- 1 - 1 - 1ú ~ ê0 1 0ú
ê ú ê ú
ë 3 - 1 3 û ë0 0 0 û
é x1 ù é- t ù é- 1ù é- 1ù
Þ êê x2 úú = êê 0 úú = t êê 0 úú Þ Eigenvector : p1 = êê 0 úú
êë x3 úû êë t úû êë 1 úû êë 1 úû
l2 = -2
é- 3 1 1 ù é1 0 - 14 ù
Þ l2 I - A = ê - 1 - 5 - 1ú ~ ê0 1 14 ú
ê ú ê ú
ë 3 - 1 - 1û ë0 0 0 û
é x1 ù é 14 t ù é1ù é1ù
Þ êê x2 úú = êê- 14 t úú = 14 t êê- 1úú Þ Eigenvector : p2 = êê- 1úú
êë x3 úû êë t úû êë 4 úû êë 4 úû

38

l3 = 3 é 2 1 1 ù é1 0 1 ù
Þ l3 I - A = ê- 1 0 - 1ú ~ ê0 1 - 1ú
ê ú ê ú
ë 3 - 1 4 û ë0 0 0 û
é x1 ù é- t ù é- 1ù é- 1ù
Þ êê x2 úú = êê t úú = t êê 1 úú Þ Eigenvector : p3 = êê 1 úú
êë x3 úû êë t úû êë 1 úû êë 1 úû

Eigenvectors B= { p1 , p2 , p3 } are linearly independent and hence

is a basis for R3.
Since
T(p1)=2.p1; T(p2)=(-2)p2; T(p3)=3.p3
we have

39
 ⎛ 2 0 0 ⎞
The matrix of T relative to the basis B is ⎜ ⎟
⎜ 0 −2 0 ⎟
⎝ 0 0 3 ⎠

⎛ −1 1 −1 ⎞
⎜ ⎟
Moreover, matrix P = [p1 p2 p3] = ⎜ 0 −1 1 ⎟ is the
⎝ 1 4 1 ⎠

transition matrix from the basis B to the standard basis. So, we

get
⎛ 2 0 0 ⎞
⎜ ⎟
P-1AP = ⎜ 0 −2 0 ⎟
⎝ 0 0 3 ⎠

40

7.3 Symmetric Matrices and Orthogonal Diagonalization

n Symmetric matrix:
A square matrix A is symmetric if it is equal to its transpose:
A = AT

n Ex 1: (Symmetric matrices and nonsymetric matrices)

é 0 1 - 2ù
A=ê 1 3 0 ú (symmetric)
ê ú
ë- 2 0 5 û
é4 3ù
B=ê (symmetric)
ë3 1úû
é3 2 1ù
C = ê1 - 4 0ú (nonsymmetric)
ê ú
ë1 0 5 û

41
 n Thm 7.7: (Eigenvalues of symmetric matrices)
If A is an n× n symmetric matrix, then the following properties
are true.
(1) A is diagonalizable.
(2) All eigenvalues of A are real.
(3) If λ is an eigenvalue of A with multiplicity k, then λ has k
linearly independent eigenvectors. That is, the eigenspace
of λ has dimension k.

42

n Ex 2:
Prove that a symmetric matrix is diagonalizable.
éa c ù
A=ê
ë c b úû

Pf: Characteristic equation:

l -a -c
lI - A = = l2 - (a + b)l + ab - c 2 = 0
-c l -b
As a quadratic in λ, this polynomial has a discriminant of
(a + b) 2 - 4(ab - c 2 ) = a 2 + 2ab + b 2 - 4ab + 4c 2
= a 2 - 2ab + b 2 + 4c 2
= (a - b) 2 + 4c 2 ³ 0

43
 (1) (a - b) 2 + 4c 2 = 0

Þ a = b, c = 0
éa 0 ù
A=ê ú is a matrix of diagonal.
ë 0 a û

(2) (a - b) 2 + 4c 2 > 0

The characteristic polynomial of A has two distinct real roots,

which implies that A has two distinct real eigenvalues. Thus,
A is diagonalizable.

44

n Orthogonal matrix:
A square matrix P is called orthogonal if it is invertible and

P -1 = P T

n Ex 4: (Orthogonal matrices)
é 0 1ù é0 - 1ù
(a) P = ê ú is orthogonal because P -1
= P T
= ê1 0 ú .
ë - 1 0 û ë û
é3 - 4ù é 3 4ù
ê5 0 5 ú ê 5 0 5ú
(b) P = ê 0 1 0 ú is orthogonal because P -1 = P T = ê 0 1 0ú .
ê4 3 ú ê- 4 3ú
ê 0 ú ê 0 ú
êë 5 5 úû êë 5 5 úû

45
 n Thm 7.8: (Properties of orthogonal matrices)

An n×n matrix P is orthogonal if and only if its column vectors

form an orthonormal set.

46

n Ex 5: (An orthogonal matrix)

é 13 2
3
2
3
ù
ê ú
P = ê - 25 1
5
0ú
ê -2 -4 5 ú
ë3 5 3 5 3 5û

Sol: If P is a orthogonal matrix, then P -1 = P T Þ PPT = I

é 1 2 2 ù é 13 -2 -2 ù é1 0 0ù
ê 3 3 3
úê 5 3 5
ú ê ú
PPT = ê - 25 1
5
0 ú ê 23 1
5
-4
3 5 ú = ê0 1 0 ú = I
ê 3- 25 -4 5 úê 2
0 5 ú ë0 0 1 û
ë 3 5 3 5 ûë 3 3 5 û
é 13 ù é 23 ù é 23 ù
Let p1 = êê -25 úú, p2 = êê 15 úú, p3 = êê 0 úú
êë 3-25 úû êë 3-45 úû êë 3 5 5 úû

47
 produces
p1 × p2 = p1 × p3 = p2 × p3 = 0
p1 = p2 = p3 = 1
{ p1 , p2 , p3 } is an orthonormal set.

48

n Thm 7.9: (Properties of symmetric matrices)

Let A be an n× n symmetric matrix. If λ1 and λ2 are distinct

eigenvalues of A, then their corresponding eigenvectors x1
and x2 are orthogonal.

n Ex 6: (Eigenvectors of a symmetric matrix)

⎡ 3 1 ⎤
Show that any two eigenvectors of A= ⎢ ⎥
⎣ 1 3 ⎦
corresponding to distinct eigenvalues are orthogonal

49
 n Sol: Characteristic function
l - 3 -1
lI - A = = l2 - 6l + 8 = (l - 2)(l - 4) = 0
-1 l - 3
Þ Eigenvalues : l 1 = 2, l2 = 4
é- 1 - 1ù é1 1ù é- 1ù
(1) l1 = 2 Þ l1I - A = ê ~
ú ê0 0 ú Þ x = s ê 1 ú, s ¹ 0
- 1 - 1
1
ë û ë û ë û
é 1 - 1ù é1 - 1ù é1ù
(2) l2 = 4 Þ l2 I - A = ê ~
ú ê0 0 ú Þ x = t ê1ú, t ¹ 0
- 1 1
2
ë û ë û ëû
é- s ù ét ù
x 1 × x 2 = ê ú × ê ú = st - st = 0 Þ x 1 and x 2 are orthogonal.
ë s û ët û

50

n Orthogonal Diagonalization:
A matrix A is orthogonally diagonalizable if there exists an
orthogonal matrix P such that
P-1.A.P=D
is a diagonal matrix.

n Thm 7.10: (Fundamental theorem of symmetric matrices)

Let A be an n×n matrix. Then A is orthogonally diagonalizable
and has real eigenvalue if and only if A is symmetric.

51
 nOrthogonal diagonalization of a symmetric matrix:
Let A be an n× n symmetric matrix.
(1) Find all eigenvalues of A and determine the multiplicity of each.
(2) For each eigenvalue of multiplicity 1, choose a unit eigenvector.
(3) For each eigenvalue of multiplicity k ≥ 2, find a set of k linearly
independent eigenvectors. If this set is not orthonormal, apply
Gram-Schmidt orthonormalization process.
(4) The composite of steps 2 and 3 produces an orthonormal set of
n eigenvectors. Use these eigenvectors to form the columns of
P. The matrix P -1 AP = P T AP = D will be diagonal.

52

n Ex 7: (Determining whether a matrix is orthogonally diagonalizable)

Symmetric Orthogonally
matrix diagonalizable
é1 1 1ù
A1 = ê1 0 1ú
ê ú
ë1 1 1û
é 5 2 1ù
A2 = ê 2 1 8ú
ê ú
ë - 1 8 0û
é 3 2 0ù
A3 = ê
ë2 0 1úû
é0 0 ù
A4 = ê
ë0 - 2úû

53
 n Ex 9: (Orthogonal diagonalization)
Find an orthogonal matrix P that diagonalizes A.
é2 2 - 2ù
A = ê 2 -1 4 ú
ê ú
êë- 2 4 - 1 úû
Sol:
(1) lI - A = (l - 3) 2 (l + 6) = 0

l1 = -6, l2 = 3 (has a multiplicity of 2)

v1
(2) l1 = -6, v1 = (1, - 2, 2) Þ u1 = = ( 13 , -32 , 23 )
v1
(3) l2 = 3, v2 = (2, 1, 0), v3 = (-2, 0, 1)

Linear Independent

54

Gram-Schmidt Process:
v3 × w2
w2 = v2 = (2, 1, 0), w3 = v3 - w2 = ( -52 , 54 , 1)
w2 × w2
w w
u2 = 2 = ( 25 , 15 , 0), u3 = 3 = ( 3- 25 , 3 4 5 , 3 5 5 )
w2 w3

é 13 2
5
-2
3 5 ù
(4) P = [ p1 p2 p3 ] = êê -32 1
5
4 ú
3 5
ú
êë 23 0 3 5û
5
ú

é - 6 0 0ù
Þ P -1 AP = P T AP = êê 0 3 0úú
êë 0 0 3úû

55