CHAPTER 8

Matrices
Outline

8.1 Matrix Algebra

8.2 Systems of Linear Equations
8.3 Rank of a Matrix
8.4 Determinants
8.5 Properties of Determinants
8.6 Inverse of a Matrix
8.7 Cramer’s Rule
8.8 Eigenvalue Problem
8.9 Powers of Matrices
Outline (cont’d.)

8.10 Orthogonal Matrices

8.11 Approximation of Eigenvalues
8.12 Diagonalization
8.13 LU-Factorization
8.14 Cryptography
8.15 Error-Correcting Code
8.16 Method of Least Squares
8.17 Discrete Compartmental Models
Inverse of a Matrix

• If A is an n × n matrix and there exists an n × n

matrix B such that AB = BA = I , then A is said to
be nonsingular or invertible and B is the inverse
of A
Revisiting The Property of Cofactors

▪ Suppose A is an n×n matrix. If ai,0, ai,1,...,ai,n-1 (i=0, ..., n-1) are the entries in
the ith row and Ck,0, Ck,1,..., Ck,n-1 (k=0, ..., n-1) are cofactors of the entries
in the kth row, then
ai,0Ck,0 + ai,1Ck,1 + … + ai,n−1Ck,n−1 = 0 for i ≠ k

▪ If a0,j, a1,j,...,an-1,j are the entries in the jth column and D0,k, D1,k,..., Dn-1,k are
cofactors of the entries in the kth column, then
a0,j D0,k + a1,j D1,k + … + an−1,j Dn−1,k = 0 for i ≠ k
Example : Cofactors

6 2 7 a0
A3×3 = −4 −3 2 = a1
2 4 8 a2
3×3

−3 2 −4 2 −4 −3 = a0,0(−32) + a0,1(36) + a0,2(−10) = a0 ⋅ c0

det A3×3 = 6 −2 +7
4 8 2 8 2 4

2 7 6 7 6 2 = a1,0(12) + a1,1(34) + a1,2(−20) = a1 ⋅ c1

= − (−4) + (−3) − (2)
4 8 2 8 2 4

2 7 6 7 6 2 = a2,0(25) + a2,1(−40) + a2,2(−10) = a2 ⋅ c2

= + (2) − (4) + (8)
−3 2 −4 2 −4 −3

a0 a0 −32 12 25 a0c0T a0c1T a0c2T det A 0 0

ACT = a1 (c0T c1T c2T) = a1
3×3
36 34 −40 = a1c0T a1c1T a1c2T = 0 det A 0
a2 a2 −10 −20 −10 a2c0T a2c1T a2c2T 0 0 det A
3×3 3×3
Property of Cofactors

ai,0Ck,0 + ai,1Ck,1 + … + ai,n−1Ck,n−1 = 0 for i ≠ k

(ai,0 ai,1 … ai,n−1) ⋅ (Ck,0 Ck,1 … Ck,n−1) = 0

ck = (Ck,0 Ck,1 … Ck,n−1) ai ⋅ ck = 0
1×n
ai ⋅ ci = det A
a0,0 a0,1 … a0,n−1 a0 c0
a1,0 a1,1 … a1,n−1 a1 c1
An×n = = C=
⋮ ⋮ … ⋮ ⋮ ⋮
an−1,0 an−1,1 … an−1,n−1 an−1 cn−1
n×n n×n n×n

a0 a 0 c 0T a0c1T … a0cn−1T det A 0 … 0

a1 a1c0T a1c1T … a1cn−1T 0 det A … 0
(c0 c1 … cn−1 ) =
T
AC = T T T =
⋮ ⋮ ⋮
an−1 an−1c0T an−1c1T … an−1cn−1T 0 0 … det A
Elimination Matrix (Row-Column Operations)

6 2 7
A0 = −4 −3 2 A3 = E2A2 = E2E1A1 = E2E1E0A0
2 4 8 = (E2E1E0) A0
1 0 0 1 0 0 0 0 1
(−3 0 1) (0 0 1) (1 0 0)
row 0 row 2
E = E2E1E0 = 0 1 0 2 1 0 0 1 0
0 0 1 6 2 7 2 4 8
(1 0 0) 2
A1 = E0A0 = 0 1 0 −4 −3 2 = −4 −3 2 0 0 1
(1 0 −3)
4 8 6 2 7 = 0 1 2
row 1 = row 1 + 2 × row 0
1 0 0 2 4 8 2 4 8 0 0 1 6 2 7 2 4 8
(0 0 1) 6 (1 0 −3) 2
A2 = E1A1 = 2 1 0 −4 −3 2 = 0 5 18 EA0 = 0 1 2 −4 −3 2 = 0 5 18
2 7 6 2 7 4 8 0 −10 −17

row 2 = row 2 + -3 × row 0

1 0 0 2 4 8 2 4 8
(−3 0 1)
A3 = E2A2 = 0 1 0 0 5 18 = 0 5 18
6 2 7 0 −10 −17
Gauss Jordan Elimination

(−4 −3)
2 4 AUG4 = E3AUG3 = E3E2E1E0AUG0
A0 =
E = E3E2E1E0

(−4 −3 0 1) (0 1 ) (0 0.2) (4 1) ( 0 1)
2 4 1 0 1 −2 1 0 1 0 0.5 0
AUG0 = =

( 0 1) (−4 −3 0 1) (−4 −3 0 1) ( 0.4 0.2 )

0.5 0 2 4 1 0 1 2 0.5 0 −0.3 −0.4
AUG1 = E0AUG0 = = =

(4 1) (−4 −3 0 1) (0 5 2 1) ( 0.4 0.2 ) (−4 −3)

1 0 1 2 0.5 0 1 2 0.5 0 −0.3 −0.4 2 4
AUG2 = E1AUG1 = = EA0 =

(0 0.2) (0 5 2 1) (0 1 0.4 0.2) (0 1)

1 0 1 2 0.5 0 1 2 0.5 0 1 0
AUG3 = E2AUG2 = = =

(0 1 ) (0 1 0.4 0.2) (0 1 0.4 0.2 )

1 −2 1 2 0.5 0 1 0 −0.3 −0.4
AUG4 = E3AUG3 = =
Cramer’s Rule

• For a system of n linear equations in n variables

a0,0 x0+ a0,1x1+ …+ a0,n−1xn−1 = b0
a1,0 x0+ a1,1x1+ …+ a1,n−1xn−1 = b1
⋮
an−1,0 x0+ an−1,1x1+ …+ an−1,n−1xn−1 = bn−1

it is convenient to define a special matrix

a0,0 a0,1 … a0,k−1 b0 a0,k+1 … a0,n−1

a1,0 a1,1 … a1,k−1 b1 a1,k+1 … a1,n−1
Ak =
⋮
an−1,0 an−1,1 … an−1,k−1 bn−1 an1,k+1 … an−1,n−1
n×n
Cramer’s Rule (cont’d.)

• If det A ≠ 0 , the solution of the system is given by

det A0 det A1 det An−1
x0 = , x1 = , …, xn−1 =
det A det A det A
where Ak, k = 0,1,…, n − 1 is given by the special
matrix previously defined
Example : Cramer’s Rule

3x0+ 2x1+ x2 = 7
x0− x1+ 3x2 = 3
5x0+ 4x1− 2x2 = 1
3 2 1
det A = 1 −1 3 = 13
5 4 −2

7 2 1 3 7 1 3 2 7
det A0 = 3 −1 3 = − 39 det A1 = 1 3 3 = 78 det A2 = 1 −1 3 = 52
1 4 −2 5 1 −2 5 4 1

1 −39 −3 3 2 1 −3 −9 + 12 + 4 7
( ) ( ) (1)
x= 78 = 6 1 −1 3 6 = −3 − 6 + 12 = 3
13
52 4 5 4 −2 4 −15 + 24 + −8
The Eigenvalue Problem

An×nxn×1 = λxn×1
An×nxn×1 − λ1×1xn×1 = 0n×1
An×nxn×1 − λ1×1In×nxn×1 = 0n×1
(An×n − λ1×1In×n) xn×1 = 0n×1
The Eigenvalue Problem

• It is often important to determine if there exist

nonzero n × 1 matrices K such that the product
vector AK is a constant multiple λ of K
• The problem of solving AK = λ K for nonzero
vectors K is the eigenvalue problem for A
• The solution vector K is said to be an eigenvector
corresponding to the eigenvalue λ
• We must have det (A − λ I ) = 0 to find a nonzero
solution for K
Example : Multiplying a Matrix with one of its Eigenvectors

0 −1 −3 1
(−2 1 1) (1)
A= 2 3 3 K = −1

0 −1 −3 1 0+1−3 −2 1
(−2 1 ) ( 1) (−2) (1)
2 3 3 −1 = 2 − 3 + 3 = 2 = (−2) −1 = (−2)K = λK
1 −2 − 1 + 1

0 −1 −3 1 0 0 1
(−2 1 ) ( ) ( 1)
(A − λI) K = 2 3 3 − (−2) 0 1 0 −1
1 0 0 1
2 −1 −3 1 2+1−3 0
( 1) (0)
= 2 5 3 −1 = 2 − 5 + 3 = 0
−2 1 3 −2 − 1 + 3
The Eigenvalue Problem (cont’d.)

• We must have det (A − λ I ) = 0 to find a nonzero

solution for K
• det (A − λ I ) = 0 is the characteristic equation of
A
• To find an eigenvector corresponding to an
eigenvalue, we solve the system of equations (A − λ I )K = 0
by applying Gauss–Jordan elimination to (A − λ I 0 )
Example : Finding the Eigenvalues (3×3)

1 2 1
(−1 −2 −1)
A = 6 −1 0 det A = 0

1 2 1 1 0 0 1−λ 2 1
(−1 −2 −1) (0 0 1)
A − λI = 6 −1 0 − λ 0 1 0 = 6 −1 − λ 0
−1 −2 −1 − λ
1−λ 2 1
2 1 1−λ 1
det (A − λI) = 6 −1 − λ 0 = (−6) + (−1 − λ)
−2 −1 − λ −1 −1 − λ
−1 −2 −1 − λ
= (−6)(−2 − 2λ + 2) + (−1 − λ)((1 − λ)(−1 − λ) + 1)
= 6(2 + 2λ − 2) + (−1 − λ)λ 2
= 12λ + (−1 − λ)λ 2 λ0 0
(3)
= λ (12 + (−λ − λ 2)) λ1 = −4
= − λ (λ 2 + λ − 12) = − λ(λ + 4)(λ − 3) = 0 λ2
Example : Finding the Eigenvectors (3×3)
λ0 = 0 λ1 = − 4 λ2 = 3
1 2 1 0 5 2 1 0 −2 2 1 0
(A − λ0I | 0) = 6 −1 0 0 (A − λ1I | 0) = 6 3 0 0 (A − λ2I | 0) = 6 −4 0 0
−1 −2 −1 0 −1 −2 3 0 −1 −2 −4 0
R1 += (-6) R0 R2 R0 R2 R0
R2 += R0 1 2 1 0 −1 −2 3 0 −1 −2 −4 0
0 −13 −6 0 6 3 0 0 6 −4 0 0
R1 += 6 R0 5 2 1 0 R1 += 6 R0 −2 2 1 0
0 0 0 0
R1 *= (-1/13) R2 += 5 R0 R2 += (-2) R0
1 2 1 0 R0 *= -1 1 2 −3 0 R0 *= -1 1 2 4 0
6
0 1 13 0 0 −9 18 0 0 −16 −24 0
0 −8 16 0 0 6 9 0
0 0 0 0 R1 *= -1/9 1 2 −3 0 R1 *= -1/16 1 2 4 0

(0 1 −2 0)
R0 += (-2)R1
1 0 1
0 R2 *= -1/8 0 1 −2 0 R2 *= 1/6
0 1 32 0
13
6 3
0 1 0 R0 += (-2)R1 0 1 0
13 1 0 1 0
(0 0 0 0)
2
R2 += (-1)R1
1 0 0 0 0 0 1 −2 0 R0 += (-2)R1 1 0 1 0
k0 + k =0 R2 += (-1)R1 0 1 3 0
13 2 2
6 k0 + k2 = 0 0 0 0 0 k0 + k2 = 0
k1 + k =0
13 2 k1 − 2k2 = 0 3
k1 + k2 = 0
x0 = (− 13
1 6
k2 − 13 k 2 k 2) x1 = (−k2 2k2 k2) x2 = (−k2 − 32 k2 k2) 2
Example : Verifying Eigenvectors (3×3)
λ0 = 0 λ1 = − 4 λ2 = 3
1
− 13 k2 −k2 −k2
1 2 1 1 2 1 1 2 1
(−1 −2 −1) (−1 −2 −1) (−1 −2 −1)
Ax0 = 6 −1 0 6
− 13 k2 Ax1 = 6 −1 0 2k2 Ax2 = 6 −1 0 − 32 k2
k2 k2
k2
k2 = 13 k2 = 1 k2 = − 2
1 2 1 −1 1 2 1 −1 1 2 1 2
(−1 −2 −1) ( 13 ) (−1 −2 −1) ( 1 ) (−1 −2 −1) (−2)
Ax0 = 6 −1 0 −6 Ax1 = 6 −1 0 2 Ax2 = 6 −1 0 3

−1 + 4 + 1 4
(−4)
−1 − 12 + 13 0 2+6−2 6
(0)
Ax0 = −6 + 6 + 0 = 0 Ax1 = −6 − 2 + 0 = −8 Ax2 = 12 − 3 + 0 = 9
1 + 12 + −13 1−4−1 −2 − 6 + 2 −6
Example : Finding the Eigenvalues (2×2)

(−1 7) (0 1) ( −1 7 − 5)
3 4 1 0 3−5 4
(−1 7)
3 4 A − λI = −5 =
A= det A = 21 + 4 = 25

(−1 2)
−2 4
=
(−1 7) (0 1) ( −1 7 − λ)
3 4 1 0 3−λ 4 R1 R0
A − λI = −λ =

(−2 4 0)
R0 *= -1 1 −2 0

3−λ 4 R1 += 2 R0
det (A − λI) = = (3 − λ) (7 − λ) + 4
(0 0 0)
−1 7 − λ 1 −2 0
= (3 − λ) (7 − λ) + 4
= 21 − 7λ − 3λ + λ 2 + 4 k0 − 2k1 = 0
= λ 2 − 10λ + 25
= (λ − 5)
2
x0 = x1 = (2k1 k1)

(λ1) (5)
λ0 5
=
Example : Finding Complex Eigenvalues (2×2)

( 4 − (5 ± 2j)) ( 5 −1 ∓ 2j)
6 − (5 ± 2j) −1 1 ∓ 2j −1
(5 4 )
6 −1 A − λI = =
A= det A = 24 + 5 = 29 5
R1 R0

(5 4 ) (0 1) ( 5 4 − λ)
6 −1 1 0 6 − λ −1
(1 ∓ 2j 0)
A − λI = −λ = R0 *= 0.2 1 −0.2 ∓ 0.4j 0
−1

6 − λ −1 R1 += (-1±2j) R0
det (A − λI) = = (6 − λ) (4 − λ) + 5
(0 0)
5 4−λ 1 −0.2 ∓ 0.4j 0
0
= (6 − λ) (4 − λ) + 5
= 24 − 4λ − 6λ + λ 2 + 5 k0 + (−0.2 ∓ 0.4j)k1 = 0
= λ 2 − 10λ + 29
x0 = ((0.2 + 0.4j)k1 k1)
= (λ − 5 − 2i) (λ − 5 + 2i)
x1 = ((0.2 − 0.4j)k1 k1)
(λ1) (5 − 2j)
λ0 5 + 2j
=
The Eigenvalue Problem (cont’d.)

• λ = 0 is an eigenvalue of A if and only if A is

singular
• If λ is an eigenvalue of nonsingular matrix A with
eigenvector K, the 1/λ is an eigenvalue of A–1 with
the same K
• The eigenvalues of an upper triangular, lower
triangular, and diagonal matrix are the main
diagonal entries
Example : Eigenvalues of the Inverse Matrix
3

(2 3)
4 0 12
0 1
−1 −1
A= det A = 12 A = 2 4
det A =
− 12 12
12

(2 3) (0 1) ( 2 3 − λ)
4 0 1 0 4−λ 0
A − λI = −λ = 3 3

(0 1)
12
0 1 0 12
−λ 0
A − λI = 2 4
−λ = 2 4
− 12 12
− 12 12
−λ
4−λ 0
det (A − λI) = = (4 − λ) (3 − λ) + 0
2 3−λ 3
−λ 0 7 1
det (A − λI) =
12
= λ2 − λ+ =0
2
− 12 4
−λ 12 12
12

12λ 2 − 7λ + 1 = 0
(3λ − 1)(4λ − 1) = 0

(λ1) (3) (λ1)

λ0 4 λ0 4
= = 1
3
Example : Eigenvectors of the Inverse Matrix
3 1

(2 3) (λ1) (3) (λ1)

4 0 λ0 4 12
0 1 λ0 4
A= det A = 12 = A−1 = det A−1 = = 1
2
− 12 4 12
12 3

1 1
− 0 0
) ( 2 3 − 4 0) (2 −1 0)
0 0 0
( 0)
4−4 0 0 0 0 0
( (A − λ0I | 0) =
4 4
A − λ 0 I | 0 = = = 1 1
− 16 1
− 1
0 − 6 12
3 4

2k0 − k1 = 0 −2k0 + k1 = 0
k1 = 2k0 k1 = 2k0
x0 = (k0 2k0) x0 = (k0 2k0)
1 1 1
− 0 0 − 12 0 0
(A − λ1I | 0) = ( 2 3 − 3 0) (2 0 0)
4−3 0 0 1 0 0
(A − λ1I | 0) =
4 3
= =
− 16 1
3
− 1
3
0 − 16 0 0
k0 = 0 k0 = 0
x1 = (0 k1) x1 = (0 k1)
Powers of Matrices : Cayley-Hamilton Theorem

• An n × n matrix A satisfies its own characteristic

equation
• If (−1)n λ n + cn−1λ n−1 + … + c1λ + c0 = 0 is the characteristic
equation of A, then
(−1)n An + cn−1An−1 + … + c1A + c0I = 0
• Any power Am can be written in terms of I, A, and
A2, with the coefficients depending on the value of
m
Am = c0I + c1A + c2A2 + … + cn−1An−1