Copyright: To balance the rights of copyright owners with the needs of the University,

we will restrict access to material only to the DUT community: academic staff, staff
related to teaching & learning and students enrolled in courses. Students are advised
to limit their use of material provided to them only for the purpose of their own private
study, and refrain from sharing with persons not enrolled for the class. We will maximize
the full extent of compliance in accordance with Copyright law.

MATRICES

1 Definition of a matrix and Order of a matrix

A Matrix is a rectangular array of elements enclosed between brackets (either round or square).
Examples
 
    −2
2 −3 5 1 −2  −1 
B=  −1 0 4 , C=  −1 0  D= 
 −4 
6 −1 −1 −5 −1
0
By convention, capital letters are used to represent matrices symbolically
The general representation of a matrix A is as follows;
 
a11 a12 a13 ··· a1n
 a21 a22 a23
 ··· a2n 

A = (aij ) =  a31 a32 a33
 ··· a3n 

 .. .. .. .. 
 . . . . 
am1 am2 am3 ··· amn
where each aij is the element of the matrix at the intersection of row i and column j.
So, in the examples B, C and D above, elements b23 = 4, c31 = −5 and d41 = 0.
The Order of a matrix indicates the number of rows and columns in a matrix. A matrix with
m rows and n columns has order (m × n) (pronounced “m by n”).
The matrices B, C and D above have orders (3 × 3), (3 × 2) and (4 × 1) respectively

2 Equal Matrices
Two matrices A and B are equal if corresponding elements ineach of the matrices
  are equal  i.e
2 x −3 a −1 b
A=B if and only if aij = bij for all i and j. For example, If = ,
y 0 z 5 c 4
then a = 2, b = −3, x = −1, y = 5 and z = 4

© 2019 Durban University of Technology: D Day Page 1

3 Addition and Subtraction of Matrices
Addition and/or subtraction of matrices is only possible if the matrices being added or sub-
tracted have the same order. Otherwise, the matrices cannot be added or subtracted.
     
a11 a12 b11 b12 a11 + b11 a12 + b12
A + B =  a21 a22  +  b21 b22  =  a21 + b21 a22 + b22 
a31 a32 b31 b32 a31 + b31 a32 + b32
     
a11 a12 b11 b12 a11 − b11 a12 − b12
A − B =  a21 a22  −  b21 b22  =  a21 − b21 a22 − b22 
a31 a32 b31 b32 a31 − b31 a32 − b32
Thus, addition and/or subtraction is executed by adding or subtracting corresponding elements
in the respective matrices. Note also, that the matrix representing the sum or difference of
matrices has the same order as the matrices being added or subtracted.
Example
     
2 −3 5 1 −2 1 4 −2
B =  −1 0 4 , C =  −1 0 , D =  1 −2 −1 
6 −1 −1 −5 −1 5 −2 1
Find B + D, C + D and D − B

     
2 −3 5 1 4 −2 3 1 3
B+D =  −1 0 4  +  1 −2 −1  =  0 −2 3 
6 −1 −1 5 −2 1 11 −3 0

C + D is not possible because the orders of the two matrices are not the same. C has order
3 × 2 while D has order 3 × 3
     
1 4 −2 2 −3 5 −1 7 −7
D−B = 1 −2 −1  −  −1 0 4 = 2 −2 −5 
5 −2 1 6 −1 −1 −1 −1 2

4 Multiplication
4.1 Scalar Multiplication
Multiplication of a matrix by a scalar (constant). When a matrix is multiplied by a scalar,
each element in the matrix is multiplied by the scalar quantity.

© 2019 Durban University of Technology: D Day Page 2

Examples:

   
1 −2 3 −6
1) 3·  −1 0  =  −3 0 
−5 −1 −15 −3

1
1 4 −2 2 −1
   
2
1
2)  1 −2 −1  = 2 ·  −1 − 21 
   
2
5 1
5 −2 1 2
−1 2

4.2 Matrix Multiplication

This refers to the multiplication of matrices with matrices. There is a condition under which
matrix multiplication is possible which implies that it is not always possible to multiply two
matrices.

4.2.1 Necessary Condition

Two matrices can be multiplied only if the number of columns in the first matrix is equal to
the number of rows in the second matrix.
Suppose that a matrix A with order (m × n) is multiplied to matrix B with order (n × p)
to produce matrix C, i.e (A · B = C). Matrix multiplication requires the multiplication of
elements, in order from left to right, from rows in matrix A respectively, with elements, in
order from top to bottom, from columns in matrix B. Thus, the number of elements in a row
in matrix A must correspond to the number of elements in a column in B. This is equivalent to
requiring the number of columns in matrix A being the same as the number of rows in matrix
B.

A · B = C
(m × n) (n × p) (m × p)
=

In practice, it is useful to write down the orders of the two matrices to be multiplied as done
for A · B above.

© 2019 Durban University of Technology: D Day Page 3

Note:

ˆ If the two inner values in the two orders are equal, then the multiplication is possible.

ˆ The two outer values in the two orders of A and B, give the values of the order of the
product C

Note that matrix multiplication is not Commutative for all matrices A and B. For the
products AB and BA to both exist, one matrix must have order (m × n) while the other
matrix must have order (n × n). Then, even if both products exist, it does not automatically
follow that AB = BA

4.2.2 Method
Each element cij in the product C = A · B is calculated as follows for each i and j:

cij = the sum of the products of the corresponding elements of row i

in matrix A and column j in matrix B

In other words, to find the value of cij , choose row i from the first matrix (A) and column j
from the second matrix (B). Multiply the first element of row i to the first element of column
j, then multiply the second element from rowi to the second element from column j and so on
until you have multiplied the last element from row i to the last element from column j. Add
these products together to get the value of cij .

Example:
   
1 −2 3 2 4
If A =  0 4 −5  and B =  −1 −3 , find (i) B · A and (ii) A · B
−1 1 0 1 2

(i) B · A is not possible (number of columns in B (i.e 2) does not equal the number of rows in
A (i.e 3) )

B · A
(3 × 2) (3 × 3)
̸=

(ii) A · B is possible (number of columns in A (i.e 3) equals the number of rows in B (i.e 3) )

© 2019 Durban University of Technology: D Day Page 4

 A · B = C
(3 × 3) (3 × 2) (3 × 2)
=

     
1 −2 3  2 4  c11 c12
   

 0 4 −5 


 −1 −3 
 = 
 c21 c22 

−1 1 0 1 2 c31 c32

Finding the value of c21

To find the value of c21 . Choose row 2 from matrix A and column
1 from matrix B. Note that the position of c21 in C i.e row 2,
column 1, indicates which row to choose from matrix A (row 2) and
which column to choose from matrix B (column 1).
     
1 −2 3  2 4  c11 c12
   
0 4 −5   −1 −3  =  c21 c22 
   
  
−1 1 0 1 2 c31 c32

Now multiply element 1 from the chosen row, i.e (0) to element 1 from the chosen column, i.e
(2), then multiply element 2 from the chosen row, i.e (4) to element 2 from the chosen column,
i.e (-1), then multiply element 3 from the chosen row, i.e (-5) to element 3 from the chosen
column, i.e (1). Adding these products together gives the value of c21 (see below)

0×2=0
−4
   
=
4 × −1 c11 c12

 2   
 0 4 −5  
 −1 
 = 
 −9 c22 

1 c31 c32
−5 ×
1 = −5

−9
(− 5) =
−4) +
0+(

Finding the value of c32

To find the value of c32 . Choose row 3 from matrix A and column
2 from matrix B.

© 2019 Durban University of Technology: D Day Page 5

     
1 −2 3  2 4  c11 c12
   

 0 4 −5 


 −1 −3 
 = 
 c21 c22 

−1 1 0 1 2 c31 c32

Now multiply element 1 from the chosen row, i.e (-1) to element 1 from the chosen column, i.e
(4), then multiply element 2 from the chosen row, i.e (1) to element 2 from the chosen column,
i.e (-3), then multiply element 3 from the chosen row, i.e (0) to element 3 from the chosen
column, i.e (2). Adding these products together gives the value of c21 (see below)

= −4
−1 × 4
−3
   
= c11 c12
−3

1 ×
 4   
   −3  =  −9 c22 
−1 1 0    
2 c31 −7
0×2=0

−7
) + (0) =
−4 + (−3

Continuing in this way for each of the cij gives;

     
1 −2 3  2 4  7 16
   

 0 4 −5  
 −1 −3 
  = 
 −9 −22 

−1 1 0 1 2 −3 −7

Note that because matrix multiplication is not commutative, the order in which matrices are
multiplied together is important. Suppose for example that we wish to multiply the matrix C
to both sides of the matrix equation
A·X =B
If C is multiplied on the left of the left side of the equation then it must also be multiplied on
the left of the right hand side as well i.e

C ·A·X =C ·B

Another important aspect of matrix algebra is that there is no such thing as Matrix Division.
However, multiplication by the inverse of a matrix is possible. If A is a square matrix then
it’s inverse, if it exists, is denoted by A−1 and A · A−1 = A−1 · A = I. The inverse of a matrix
will be discussed in a later section of these notes.

© 2019 Durban University of Technology: D Day Page 6

5 Types of Matrices
5.1 Transpose of a matrix
The Transpose of a matrix A is denoted by AT
If A = (aij ) then the transpose AT = (aji ). In other words row i of matrix A becomes column
i of AT which implies that column j of matrix A becomes row j of AT . This also implies that
if A has order (m × n) matrix, then AT will have order (n × m) matrix
Example
 
  2 8
2 1 3
If A = , then AT =  1 7 . Note here that A has order (2 × 3) while AT has
8 7 6
3 6
order (3 × 2)

5.2 Row matrix

A Row Matrix is a matrix of order (1 × n), i.e a matrix with only one row. A row matrix is
also called a Row Vector
Example

B= 1 5 −2 −3 3 has order (1 × 5)

5.3 Column matrix

A Column Matrix is a matrix of order (m × 1), i.e a matrix with only one column. A column
matrix is also called a column Vector
Example  
1
 5 
 −2  has order (4 × 1)
C= 

−3

5.4 Square Matrix

A Square Matrix is a matrix of order (n × n), i.e a matrix with the same number of rows as
columns
Example
 
1 2 3
A =  −2 4 −1  has order (3 × 3)
3 −1 0
Note: When describing a square matrix it can either be described as “a matrix or order (n×n)”
or as “a square matrix of order n”

© 2019 Durban University of Technology: D Day Page 7

5.5 Symmetrc Matrix
A Symmetric matrix A = (aij ) has the following properties;

ˆ The matrix is a square matrix

ˆ aij = aji (an element in row i and column j will be the same as the element in row j and
column i).

ˆ A = AT

Example
   
1 −2 3 −1 1 −2 3 −1
 −2 0 −2 8   −2 0 −2 8 
A=
 3 −2 −1
 and AT = 
 3 −2 −1

5  5 
−1 8 5 −4 −1 8 5 −4
Notice that the matrix is symmetric about the main diagonal. The Main Diagonal consists
of elements aii . In the example above the elements in the main diagonal are a11 = 1, a22 =
0, a33 = −1, a44 = −4

5.6 Skew Symmetric Matrix

A Skew Symmetric matrix A = (aij ) has the following properties;

ˆ The matrix is a square matrix

ˆ aij = −aji (an element in row i and column j will be the negative of the element in row
j and column i).

ˆ A = −AT

Example
   
1 2 −3 1 1 −2 3 −1
 −2 0 2 −8   2 0 −2 8 
A= 
 3 −2 −1 −5  and AT = 
 −3

2 −1 5 
−1 8 5 −4 1 −8 −5 −4
Notice that the matrix is symmetric about the main diagonal in terms of numerical value of
the elements but opposite in sign.

© 2019 Durban University of Technology: D Day Page 8

5.7 Diagonal Matrix
A Diagonal Matrix is a square matrix in which the elements aij = 0, for i ̸= j. In other
words, all elements which are not on the main diagonal are equal to zero. The elements on the
main diagonal may or may not be zero.
Example
 
1 0 0 0
 0 0 0 0 
A= 
 0 0 −1 0 
0 0 0 −4

5.8 Unit (or Identity) Matrix

A Unit Matrix is a Diagonal matrix in which all elements on the main diagonal are equal to
1.
Examples
I1 = (1) The unit matrix or order 1
 
1 0
I2 = The unit matrix of order 2
0 1
 
1 0 0
I3 =  0 1 0  The unit matrix of order 3
0 0 1
 
1 0 0 0
 0 1 0 0 
I4 = 
 0 0 1 0  The unit matrix of order 4


0 0 0 1

The unit matrix behaves in the same way in matrix multiplication as the number 1 does in the
multiplication of numbers, hence the name Identity Matrix.
Note

1. If any matrix A has order (m × n) then Im · A = A and A · In = A

Example
 
1 −2
If A =  5 2  then
−3 1
     
1 0 0 1 −2 1 −2
I3 · A =  0 1 0 · 5 2 = 5 2 =A
0 0 1 −3 1 −3 1
and

© 2019 Durban University of Technology: D Day Page 9

    
1 −2   1 −2
1 0
A · I2 =  5 2 · = 5 2 =A
0 1
−3 1 −3 1

2. If A is a square matrix of order n, then In · A = A · In = A

Example
 
1 −2 0
If A =  5 2 −1 , then
−3 1 4
     
1 0 0 1 −2 0 1 −2 0
I3 · A =  0 1 0 · 5 2 −1  =  5 2 −1  = A,
0 0 1 −3 1 4 −3 1 4
and
     
1 −2 0 1 0 0 1 −2 0
A · I3 =  5 2 −1  ·  0 1 0  =  5 2 −1  = A
−3 1 4 0 0 1 −3 1 4

5.9 Null matrix

A Null matrix is a matrix in which every element is zero. It can have any order and needn’t
be square. For matrix addition, subtraction and multiplication, the null matrix of the relevant
order, acts like the number zero does in the addition, subtraction and multiplication of numbers.

6 Adjoint of a square matrix

Recall that if A is a square matrix of order n, then the Minor of an element aij of A is the
determinant of the square matrix of order n − 1 whose elements are the elements of matrix A
excluding row i and column j.
Example
 
−2 3 −1  
−2 3
If A =  0 1 −5 , then the minor of element a23 is det = −7 and the minor
1 2
1 2 −1 
−2 −1
of a22 is det =3
1 −1
The Cofactor of an element aij of A is,

Cofactor(aij ) = (−1)i+j · minor(aij )

The Matrix of cofactors of matrix A is a square matrix C, of the same order as A in which
each element
cij = cofactor(aij ) = (−1)i+j · minor(aij )

© 2019 Durban University of Technology: D Day Page 10

Example
 
−2 3 −1
If A =  0 1 −5  then
1 2 −1
 
1+1 1 −5 1+2 0 −5 1+3 0 1
 (−1) 2 −1
(−1)
1 −1
(−1)
1 2 
 
 
 
2+1 3 −1 −2 −1 −2 3
 
C=  (−1) (−1)2+2 (−1)2+3 
 2 −1 1 −1 1 2 

 
 
 3 −1 −2 −1 −2 3 
(−1)3+1 (−1)3+2 (−1)3+3
1 −5 0 −5 0 1

 
9 −5 −1
i.e C =  1 3 7 
−14 −10 −2

Now, the Adjoint of A is the transpose C T of the matrix of cofactors C of matrix A i.e

adj(A) = C T

Example
 
9 1 −14
The adjoint of matrix A above is adj(A) = C T =  −5 3 −10 
−1 7 −2

7 Inverse of a square matrix

A square matrix A is said to be Singular if it’s determinant is equal to zero i.e if |A| =
det (A) = 0, otherwise it is Non-singular in which case |A| = det (A) ̸= 0.
A square matrix A has an inverse A−1 if and only if A is non-singular (i.e if and only if
det (A) ̸= 0. If this condition is met we say that matrix A is Invertible.

7.1 Methods for finding the inverse of a square matrix

7.1.1 Adjoint/Cofactor Method
Given a square matrix A.

1. Find det(A) = |A| to test if it is non-singular and hence invertible. If det(A) ̸= 0 then
matrix A is non-singular and A is invertible. If so, then;

2. Form the matrix of cofactors C of matrix A.

© 2019 Durban University of Technology: D Day Page 11

 3. Form the adjoint of A i.e adj(A) = C T

4. The inverse A−1 of an invertible square matrix A is defined to be

1
A−1 = · adj(A)
det(A)

Example
 
1 1 −1
Find the inverse matrix of A =  2 0 −2 
1 1 1
1 −1 1 1
det(A) = |A| = (−1)2+1 (2) + 0 + (−1)2+3 (−2)
1 1 1 1
= −(2)(2) − (−2)(0) = −4
   
2 −4 2 2 −2 −2
C =  −2 2 0  and adj(A) = cT =  −4 2 0 
−2 0 −2 2 0 −2

− 21 1 1
   
2 −2 −2 2 2

∴ A−1 = 1
· adj(A) = 1
·  −4 2 0  or A−1 =  1 − 12 0 
 
|A| −4
2 0 −2 − 21 0 1
2

You can check the correctness of this answer by showing that A · A−1 = I3 or A−1 · A = I3

7.1.2 Guass-Jordan Method

It is possible to use elementary row operations to find the inverse of a square matrix.
Row Operations are as follows:

ˆ Interchange any two rows

ˆ Multiply any row throughout by a non-zero constant.

ˆ Add(or subtract) a constant multiple of one row to another row

Method
Given a square matrix A of order n.

1. Find det(A) = |A| to test if it is non-singular and hence invertible. If det(A) ̸= 0 then
matrix A is non-singular and A is invertible. If so, then;

2. form the extended matrix (A | I) of order n × 2n, where I is the identity matrix of order
n

© 2019 Durban University of Technology: D Day Page 12

 3. Apply the Guass-Jordan method, i.e use row operations to transform the extended matrix
(A | I) into the extended matrix (I | B)
4. If performed correctly, A−1 = B

Examples
Find the inverse matrices of;

 
3 −2
1) A =
9 5
 
1 −3 −1
2) A =  2 3 −3 
−3 1 2

Solution
1)
 
3 −2 1 0
(A|I) =  
9 5 0 1

− 23 1
 
1 3
0
=  [ (row 1) ÷ 3]
9 5 0 1

− 23 1
 
1 3
0
=  [ (−9) · (row 1) + (row 2)]
0 11 −3 1

− 23 1
 
1 3
0
=  [(row 2) ÷ 11]
3 1
0 1 − 11 11

5 2
 
1 0 33 33
=  [ 32 · (row 2) + (row 1)]
3 1
0 1 − 11 11

= (I|A−1 )

5 2
 
33 33
∴ A−1 =  
3 1
− 11 11

© 2019 Durban University of Technology: D Day Page 13

2)
1 −3 −1 1 0 0
 

(A|I) =  2 3 −3 0 1 0 
 

−3 1 2 0 0 1

1 −3 −1 1 0 0
 

= 0 9 −1 −2 1 0  [ (row 2) − 2 · (row 1)] & [(row 3) + 3 · (row 1)]

 

0 −8 −1 3 0 1

1 −3 −1 1 0 0
 

= 0 1 − 19 − 29 1
0  [ (row 2) ÷ 9]
 
9

0 −8 −1 3 0 1

1 0 − 34 1 1
0
 
3 3

= 0 1 − 91 − 29 1
0  [ (row 1) + 3 · (row 2)] & [(row 3) + 8 · (row 2)]
 
9

0 0 − 17
9
11
9
8
9
1

1 0 − 43 1 1
0
 
3 3

=  0 1 − 19 − 29 1 9
0  [(− 17 ) · (row 3)]
 
9

0 0 1 − 11
17
8
− 17 9
− 17
9 5
1 0 0 − 17 − 17 − 12
 
17
5 1 1 
=  0 1 0 − 17 − 17  [ (row 1) + 43 · (row 3)] & [(row 2) + 91 · (row 2)]

17

0 0 1 − 11
17
8
− 17 9
− 17

= (I|A−1 )

9 5
− 17 − 17 − 12
 
17

∴ A−1 =  − 17
 5 1 1 
17
− 17 
11 8 9
− 17 − 17 − 17

You should check both of the answers by showing that A · A−1 = A−1 · A = I in both cases.

8 Solution of systems of equations

Two methods will be considered;
1. Inverse matrix method
2. Guassian Elimination method

© 2019 Durban University of Technology: D Day Page 14

8.1 Inverse matrix method
Solve the system of equations;

x + y − z = −2
2x − 2z = −2
x + y+z =2

Such a system of equations can be written as an equivalent matrix equation A · X = B as

follows;
     
1 1 −1 x −2
 2 0 −2  ·  y  =  −2 
1 1 1 z 2
| {z } | {z } | {z }
A X B

Solving the system of equations is achieved by solving the matrix equation as follows;

1. Find A−1

2. Multiply A−1 on the left of both sides of the equation. By doing so, the matrix equation
is solved for the column matrix X as follows;

A·X =B
−1 −1
| {z· A} ·X = A · B
A
I
· X} = A−1 · B
∴ I| {z
X

∴ X = A−1 · B

So for the above

 example, 
2 −2 −2
A−1 = −4 1
·  −4 2 0 .
2 0 −2
(The matrix A and A−1 are those found in the example in section 7.1.1).
         
x 2 −2 −2 −2 −4 1
1
Thus X =  y  = −4 ·  −4 2 0  ·  −2  = − 14 ·  4  =  −1 
z 2 0 −2 2 −8 2
   
x 1
∴  y  =  −1 
z 2

Since these are equal matrices the corresponding elements in the matrices are equal.
Thus x = 1, y = −1 and z = 2

© 2019 Durban University of Technology: D Day Page 15

8.2 Gaussian Elimination
Consider again, the system of equations which were solved in the previous section.

x + y − z = −2
2x − 2z = −2
x + y+z =2

Another way of expresing this system of equations is by means of an Augmented Matrix.The

augmented matrix representing this system is given below. The columns in this matrix consist
of the coefficients of x, y, z and the right hand side constants respectively.
   
1 1 −1 −2 1 1 −1 −2
 2 0 −2 −2  or  2 0 −2 −2 
1 1 1 2 1 1 1 2

A vertical line can also be inserted just before the last column to separate the coefficients of
variables on the left hand side of the equations from the right hand side constants. Either of
these representations is valid.
Another concept we need to consider before solving the system by Gaussion Elimination is the
concept of Echelon form of a matrix.

A matrix is in Echelon Form if;

ˆ each non-zero row in the matrix (i.e a row containing at least one non-zero value) begins
with more leading zero’s than any rows above it in the matrix
ˆ all zero rows (i.e rows containing only zero’s) are grouped at the bottom of the matrix
Example  
1 −2 3 2 −4

 0 3 −1 0 2 


 0 0 1 −3 2 


 0 0 0 0 −5 

 0 0 0 0 0 
0 0 0 0 0

The solutions of a system of equations are not affected in any way by the application of any of
the following row operations to the system
ˆ Interchange any two rows

ˆ Multiply any row throughout by a non-zero constant.

ˆ Add(or subtract) a constant multiple of one row to another row

© 2019 Durban University of Technology: D Day Page 16

The Gaussian Elimination method is carried out as follows;

1. Form the Augmented Matrix of the system of equations

2. Apply row operations until the augmented matrix is in echelon form

3. Beginning with the last row of the matrix, form the equation it represents and solve for
the unknown. Then by substituting this value into the equation of the next row up in the
matrix, solve for the next unknown. Continue in this manner until all unknown values
are found

During the process of solving the system by this method, it is useful to make a note of each row
operation which is applied. It is convenient to make these notes on the right hand side of the
matrices. In the example below, R1 , R2 and R3 refer to row 1, row 2 and row 3 respectively.

x + y − z = −2
2x − 2z = −2
x + y+z =2

 
1 1 −1 −2 R1
 2 0 −2 −2  R2
1 1 1 2 R3
 
1 1 −1 −2 R1 use R1 and element a11 to
 0 −2 0 2  R2′ = −2 · R1 + R2 get zero’s in position a21 and
0 0 2 4 R3′ = −1 · R1 + R3 a31

The matrix is now in echelon form

∴ R3′ ⇒ 2z = 4 ⇒ z = 2

R2′ ⇒ −2y = 2 ⇒ y = −1

R1 ⇒ x + y − z = −2
⇒ x + (−1) − 2 = −2
⇒x=1

∴ x = 1, y = −1, z=2

© 2019 Durban University of Technology: D Day Page 17

Another example
Solve

2x − y + 3z = 9
− 3x + 2y + z = 4
5x − 3y + 4z = 11

 
2 −1 3 9 R1
 −3 2 1 4  R2
5 −3 4 11 R3
 
2 −1 3 9 R1 use R1 and element a11 to
 0 1 11 35  R2′ = 3 · R1 + 2 · R2 get zero’s in position a21 and
0 −1 −7 −23 R3′ = −5 · R1 + 2 · R3 a31
 
2 −1 3 9 R1
use R2 and element a22 to
 0 1 11 35  R2′
get a zero in position a32
0 0 4 12 R3′′ = −R2′ + R3′

The matrix is now in echelon form

∴ R3′′ ⇒ 4z = 12 ⇒ z = 3

R2′ ⇒ y + 11z = 35
⇒ y + 11(3) = 35
⇒y=2

R1 ⇒ 2x − y + 3z = 9
⇒ 2x − 2 + 3(3) = 9
⇒ 2x = 2
⇒x=1

∴ x = 1, y = 2, z=3

8.2.1 Infinite Solutions and No solution

The previous two examples resulted in solutions which were unique. Hoever, there are two
other possible types of solution
ˆ No solution
Suppose that Gaussian elimination is applied to an augmented matrix and the resulting
echelon form is as follows;  
1 −1 3 2
 0 1 2 −1 
0 0 0 3

© 2019 Durban University of Technology: D Day Page 18

 The last row implies that the system has no solution. The last row represents the equation
0x + 0y + 0z = 3 ⇒ 0 = 3, which is impossible. Therefore such a system of equations
would have no solution.

ˆ Infinite number of solutions

Suppose the Gaussian elimination is applied to an augmented matrix and the resulting
echelon form is as follows;  
1 −1 3 2
 0 1 2 −1 
0 0 0 0
The last row implies that the system has an infinite number of solutions. The last row
represents the equation 0x + 0y + 0z = 0, which is true for any values of x, y and z we
choose. To express this infinite solution mathematically we proceed as follows;
Let z = α where α ∈ R
Then row 2 ⇒ y + 2z = −1 ⇒ y = −1 − 2z = −1 − 2α
Row 1 ⇒ x − y + 3z = 2
⇒ x = 2 − 3z + y
⇒ x = 2 − 3(α) + (−1 − 2α)
⇒ x = 1 − 5α
∴ x = 1 − 5α, y = −1 − 2α and z = α, α∈R
So, choosing any value for α will produce a solution to the system of equations.
For example, if α = 0, then x = 1, y = −1 and z = 0 will be a solution to the original
system of equations
Choosing a different value for α, suppose α = −2, then x = 11, y = 3 and z = −2
would be another solution of the original system.

Example
Solve

2x + 3y − z = 1
2x + 5y + 9z = −1
x + 2y + 2z = 0

 
2 3 −1 1 R1
 2 5 9 −1  R2
1 2 2 0 R3
 
2 3 −1 1 R1 use R1 and element a11 to
 0 2 10 −2  R2′ = −1 · R1 + R2 get zero’s in position a21 and
0 −1 −5 1 R3′ = R1 − 2 · R3 a31

© 2019 Durban University of Technology: D Day Page 19

  
2 3 −1 1 R1
use R2 and element a22 to
 0 2 10 −2  R2′
get a zero in position a32
0 0 0 0 R3′′ = −R2′ + 2 · R3′

The matrix is now in echelon form and row 3 implies an infinite number
of solutions
∴ R3′′ ⇒ z = α where α ∈ R

R2′ ⇒ 2y + 10z = −2
⇒ 2y = −2 − 10α
⇒ y = −1 − 5α

R1 ⇒ 2x + 3y − z = 1
⇒ 2x = 1 − 3y + z
⇒ 2x = 1 − 3(−1 − 5α) + α = 4 + 16α
⇒ x = 2 + 8α

∴ x = 2 + 8α, y = −1 − 5α, z = α, α∈R

9 Equivalent matrices
Two matrices are Row Equivalent if one of the matrices can be obtained by performing
elementary row operations on the other.
Example
 
1 −1 0
A =  −2 1 1 
0 2 3
   
1 −1 0 R1 1 −1 0 R1
 −2 1 1  R2 ∼  0 −1 1  R2′ = 2R1 + R2
0 2 3 R3 0 2 3 R3
| {z } | {z }
A B

 
1 −1 0 R1
∼  0 −1 1  R2′
0 0 5 R′ = 2R2′ + R3
| {z } 3
C

Matrices A, B and C are Row Eqiuivalent.

If similar operations to row operations are performed on the columns of a matrix we term them
Column Operations. It is also true that two matrices are column Equivalent if one of the
matrices can be obtained by performing elementary column operations on the other.

© 2019 Durban University of Technology: D Day Page 20

Matrices A, B and C in the above example are equivalent matrices.

10 Rank of a matrix
The Rank of an n × m matrix is defined to be the number of linearly independent rows (or
columns) in the matrix.
It follows from this definition that the null(zero) matrix has rank equal to zero.
There are two equivalent methods for finding the Rank of a matrix;

- Using Equivalent Matrices (using row (column) operations)

- Using Determinants of sub-matrices

10.1 Rank by Equivalent Matrices

Note: Equivalent matrices all have the same Rank.
It can be shown that the number of linearly independent rows (columns) of a matrix is equal
to the number of non-zero rows (columns) in the equivalent matrix which is in echelon form.
Thus, the Rank of matrix A can be found as follows;

1. Apply row (column) operations successively to A until the equivalent matrix in ehelon
form is found.

2. Count the number of non-zero rows (columns) in this equivalent matrix. This number is
the Rank of A.

The example in section 9 above, illustrates this method. Matrix C is equivalent to matrix A
and is in echelon form. The number of non-zero rows in C is 3, hence Rank(A)=3
Example  
1 −1 0
Find the rank of A =  −2 1 1  using column operations. Note that this is the same
0 2 3
matrix as in section 9 above.

 
1 −1 0
A=  −2 1 1 
0 2 3
 
1 0 0
=  −2 −1 1  col. 2 = col. 1 + col. 2
0 2 3

© 2019 Durban University of Technology: D Day Page 21

  
1 0 0
=  −2 −1 0  col. 3 = col. 2 + col. 3
0 2 5

The final matrix is in column-echelon form and has 3 non-zero columns. Therefore Rank(A) = 3
Example  
2 −1 −4 1 R1
Find the rank of B =  3 0 1 −2  R2
12 −3 −10 −1 R3
   
2 −1 −4 1 R1 2 −1 −4 1 R1
B=  3 0 1 −2  R2 ∼  0 −3 −14 7  R2′ = 3R1 − 2R2
12 −3 −10 −1 R3 0 3 14 −7 R3′ = R3 − 6R1
 
2 −1 −4 1 R1
∼  0 −3 −14 7  R2′
0 0 0 0 R3′ = R2′ + R3′
The final matrix is in row-echelon form and has 2 non-zero rows.
Therefore, Rank(B) = 2

10.2 Rank by determinants of sub-matrices

The number of linearly independent rows (columns) in a matrix (i.e the rank of the matrix),
can be shown to be equivalent to the order of the largest square, non-singular sub-matrix i.e
the order of the largest square sub-matrix which has a non-zero determinant. (Recall that a
square matrix is singular if the value of its determinant is zero, otherwise it is said to be
non-singular)
It follows that a square matrix of order n which is non-singular has rank n. Also, the maximum
possible value of the rank of an n × m matrix is n if n ≤ m and is m if m ≤ n.
The Rank of a matrix A with order n × m can be found as follows;
Let s = min(n, m)

1. Successively find the values of the determinants of sub-matrices of order s until a non-zero
determinant is found. If one of these sub-matrices is non-singular, then Rank(A) = s. If
all sub-matrices of order s are singular, then Rank(A) < s.

2. Let s = min(n, m) − 1. Repeat the process in step 1 for sub-matrices of order s =

min(n, m) − 1. If no non-singular sub-matrix or order s is found, then reduce the value
os s by 1 and successively repeat this process until a non-singular sub-matrix is found.
The order of this sub-matrix will be the Rank of A.

© 2019 Durban University of Technology: D Day Page 22

Examples Find the rank of each of the following matrices
 
1 −1 0 −1 0
1) A =  −2 1 1 2 −1 
0 2 3 1 2
 
1 −1 0
det  −2 1 1  = 1[3 − 2] + 1[−6 − 0] + 0[−4 − 0] = −5 ̸= 0
0 2 3
therefore, Rank(A) = 3
 
2 3 −1 1
2) B =  2 5 9 −1 
1 2 2 0
 
2 3 −1
det 2 5
 9  = 2(−8) − 3(−5) − (−1) = 0 (columns 1,2,3)
1 2 2
 
2 3 1
det 2 5
 −1  = 2(2) − 3(1) + (−1) = 0 (columns 1,2,4)
1 2 0
 
2 −1 1
det 2
 9 −1  = 2(2) + (1) + (−5) = 0 (columns 1,3,4)
1 2 0
 
3 −1 1
det 5
 9 −1  = 3(2) + (2) + (−8) = 0 (columns 2,3,4)
2 2 0

Since all sub-matrices of order 3 are singular, Rank(B) < 3

Now attempt to find a non-singular matrix of order 2

 
2 3
det = 2(5) − 2(3) = 4 ̸= 0 (non-singular)
2 5

∴ Rank(B) = 2

© 2019 Durban University of Technology: D Day Page 23

  
3 −1 −1
3) C =  −3 1 1 
0 2 3
 
3 −1 −1
det(C) = det  −3 1 1 
0 2 3
= 3(3 − 2) − (−3)(−3 + 2) + 0(−1 + 1) = 0

∴ Rank(C) < 3

Consider determinants of sub-matrices of order 2

 
3 −1
det = (3)(1) − (−3)(−1) = 0
−3 1
 
−1 −1
det = (−1)(1) − (1)(−1) = 0
1 1
 
−3 1
det = (−3)(2) − (0)(1) = −6 ̸= 0
0 2

∴ Rank(C) = 2
Exercises:
Find the rank for each of the following matrices;
 
1 −2
(a) Answer: Rank = 1
−3 6
 
1 −3 2
(b)  0 5 −4  Answer: Rank = 3
−1 1 0
 
1 −2 3 5
(c)  2 6 −1 4  Answer: Rank = 2
−3 −14 5 −3

10.3 Rank and Consistency of a system of equations

A system of simultaneous equations is said to be Consistent if solutions to the system exist (i.e
if there is a unique solution or an infinite number of solutions to the system), and Inconsistent
if no solutions exist.

The system of equations;

a11 x1 + a12 x2 + a13 x3 = b1
a21 x1 + a22 x2 + a23 x3 = b2
a31 x1 + a32 x2 + a33 x3 = b3

© 2019 Durban University of Technology: D Day Page 24

can be expressed as a matrix equation Ax = b as follows;
     
a11 a12 a13 x1 b1
 a21 a22 a23  ·  x2  =  b2 
a31 a32 a33 x3 b3
| {z } | {z } | {z }
A x b

or as an Augmented matrix;
 
a11 a12 a13 b1
Ab =  a21 a22 a23 b2 
a31 a32 a33 b3

10.3.1 Rank as a test for consistency

In general, for a system of n equations;

ˆ Rank(A)=Rank(Ab ) ⇒ equations are consistent (solutions exist)

• Rank(A) = Rank(Ab ) = n ⇒ unique set of solutions

• Rank(A) = Rank(Ab ) < n ⇒ infinite number of solutions

ˆ Rank(A) < Rank(Ab ) ⇒ equations are inconsistent (solutions do not exist)

Example
Determine the rank of A and Ab for the following system and determine the nature of the
solution
x−y−z =1
−3x − y + z = 2
2y + z = 1

   
1 −1 −1 1 −1 −1 1
A =  −3 −1 1  and Ab =  −3 −1 1 2 
0 2 1 0 2 1 1

First, consider the rank of A

det(A) = 1(−1 − 2) − (−3)(−1 + 2) = −3 + 3 = 0

⇒ Rank(A) < 3
 
1 −1
det = 1(−1) − (−3)(−1) = −4 ̸= 0
−3 −1

⇒ Rank(A) = 2

© 2019 Durban University of Technology: D Day Page 25

 Now considering the rank of Ab
 
1 −1 1
det −3 −1
 2  = 1(−1 − 4) − (−3)(−1 − 2) + 0 = −14 ̸= 0
0 2 1

⇒ Rank(Ab ) = 3

Therefore Rank(A) <Rank(Ab ) which implies that there is no

solution to the system

Now verify this conclusion using Guass Elimination

   
1 −1 −1 1 R1 1 −1 −1 1 R1
 −3 −1 1 2  R2 ∼  0 −4 −2 5  R2′ = 3R1 + R2
0 2 1 1 R3 0 2 1 1 R3
 
1 −1 −1 1 R1
∼  −3 −1 1 2  R2′
0 0 0 7 R3′ = R2′ + 2R3
The last row indicates that there is no solution

Exercises:
Determine the rank of A and Ab for the following systems of equations, and determine the
nature of the solution. Then, use Gauss Elimination to find the solution, if it exists

1.

2x + 3y − z = −2
x − y + 3z = 12
x+z =5

Answer:
Rank(A)=Rank(Ab )=3, Unique solution, x = 2, y = −1 z = 3

2.

3x − y + 3z = −4
2x − 2z = 3
8x − 2y + 4z = −5

Answer:
3+2α
Rank(A)=Rank(Ab )=2. Infinite solutions, x = 2
,
12α+17
y= 2
, z = α, for α ∈ R

© 2019 Durban University of Technology: D Day Page 26

11 Eigenvalues and Eigenvectors
There are Engineering applications which require the solution of the matrix equation

Ax = λx

where A is a given square matrix of order n, x is a column vector of order (n × 1) and λ is a

scalar quantity.
A trivial solution of the equation occurs when x is the zero vector (i.e a column vector in
which all elements are zero)
Non-trivial solutions of the equation are found when x is not the zero vector. In such cases,
the scalar quantities λ satisfying the equation are called Eigenvalues and the corresponding
column vectors x are called Eigenvectors.
Now, the equation above can be rewritten in the following form;

Ax = λx ⇒ Ax = λIn X ⇒ Ax − λIn x = 0 ⇒ (A − λIn )x = 0

where In is the identity matrix of order n and 0 is the zero column vector.
In order to obtain non-trivial solutions of the equation, the Characteristic determinant
|A − λIn | must equal 0, i.e
det(A − λIn ) = 0
This equation is called the Characteristic Equation.
Solving this equation for λ gives the values of all the Eigenvalues of A. Substituting each
eigenvalue into the equation
(A − λIn )x = 0
gives the corresponding eigenvector for each eigenvalue.

Examples
Find the Eigenvalues and Eigenvectors of the following matrices
 
2 1
1) A =
2 3
 
2−λ 1
det(A − λI) = det =0
2 3−λ
⇒ (2 − λ)(3 − λ) − (2)(1) = λ2 − 5λ + 4 = 0

⇒ (λ − 1)(λ − 4) = 0

⇒ λ = 1 or λ = 4

© 2019 Durban University of Technology: D Day Page 27

 For λ = 1
   
1 1 0 R1 1 1 0 R1
∼
2 2 0 R2 0 0 0 R2′ = 2R1 − R2
Let x2 = α; α ∈ R ⇒ x1 = −x2 = −α
   
−α 1
∴ x= = −α
α −1

For λ = 4
   
−2 1 0 R1 −2 1 0 R1
∼
2 −1 0 R2 0 0 0 R2′ = R1 + R2
Let x2 = α; α ∈ R ⇒ x1 = 21 x2 = 12 α
 1   
2
α 1 1
∴ x= = 2α
α 2

The Eigenvalues and their corresponding simple specific

Eigenvectors are respectively;
   
1 1
λ = 1, x = and λ = 4, x =
−1 2
Note, we will use the convention that the first element in a simple specific
eigenvector is always positive.
 
2 −2 −1
2) B =  −2 2 1 
−1 1 5

 
2−λ −2 −1
det(A − λI) = det  −2 2−λ 1 =0
−1 1 5−λ
⇒ (2 − λ)[(2 − λ)(5 − λ) − 1] − (−2)[−2(5 − λ) − (−1)]
+ (−1)[−2 − (−1)(2 − λ)] = 0

⇒ (2 − λ)[λ2 − 7λ + 9] + 2[2λ − 9] − [−λ] = 0

⇒ −λ3 + 9λ2 − 23λ + 18 + 4λ − 18 + λ = 0

⇒ λ3 − 9λ2 + 18λ = 0

⇒ λ(λ − 3)(λ − 6) = 0

⇒ λ = 0 or λ = 3 or λ = 6

© 2019 Durban University of Technology: D Day Page 28

 For λ = 0
   
2 −2 −1 0 R1 2 −2 −1 0 R1
 −2 2 1 0  R2 ∼  0 0 0 0  R2′ = R1 + R2
−1 1 5 0 R3 0 0 9 0 R3′ = R1 + 2R3

9x3 = 0 ⇒ x3 = 0.

Let x2 = α; α ∈ R, then 2x1 − 2α = 0 ⇒ x1 = α ∈ R

 
α
∴ x=  α 
0

For λ = 3
   
−1 −2 −1 0 R1 −1 −2 −1 0 R1
 −2 −1 1 0  R2 ∼  0 3 3 0  R2′ = −2R1 + R2
−1 1 2 0 R3 0 −3 −3 0 R3′ = R1 − R3
 
−1 −2 −1 0 R1
∼  0 3 3 0  R2′
0 0 0 0 R3′′ = R2′ + R3′
Let x3 = α; α ∈ R, then 3x2 + 3x3 = 0 ⇒ x2 = −x3 = −α
and − x1 − 2x2 − x3 = 0 ⇒ x1 = −2x2 − x3 = α
 
α
∴ x =  −α 
α
For λ = 6
   
−4 −2 −1 0 R1 −4 −2 −1 0 R1
 −2 −4 1 0  R2 ∼  0 6 −3 0  R2′ = R1 − 2R2
−1 1 −1 0 R3 0 −6 3 0 R3′ = R1 − 4R3
 
−4 −2 −1 0 R1
∼  0 6 −3 0  R2′
0 0 0 0 R3′′ = R2′ + R3′
Let x3 = α; α ∈ R, then 6x2 − 3x3 = 0 ⇒ x2 = 12 x3 = 12 α
and − 4x1 − 2x2 − x3 = 0 ⇒ 4x1 = −2x2 − x3 = −2α ⇒ x1 = − 12 α
 1 
−2α
∴ x =  12 α 
α

© 2019 Durban University of Technology: D Day Page 29

 The Eigenvalues and their corresponding simple specific Eigenvectors are;
     
1 1 1
λ = 0, x =  1  , λ = 3, x =  −1  and λ = 6, x =  −1 
0 1 −2

Exercises
Find Eigenvalues and Eigenvectors for the following matrices

 
−2 −9
1.
−6 1    
1 3
Answer: λ1 = 7, x1 = . λ2 = −8, x2 =
−1 2
 
3 1
2.
5 −1    
1 1
Answer: λ = 4, x = and λ = −2, x =
1 −5
 
0 −1 0
3.  3 4 0 
−1 1 2
     
1 0 1
Answer: λ = 1, x =  −1  , λ = 2, x =  0  and λ = 3, x =  −3 
2 1 −4

 
1 0 1
4.  0 2 0 
3 −1 −1
   
1 1
Answer: λ = −2, x =  0  , and λ = 2, x =  0 
−3 1

12 Cayley-Hamilton Theorem
The Cayley-Hamilton Theorem is stated below without proof.

Every square matrix satisfies it’s Characteristic Equation

 
2 1
A= has Characteristic Equation λ2 −5λ+4 = 0 (example 1 from Eigenvalue section)
2 3
     
2 2 1 2 1 6 5
A = · =
2 3 2 3 10 11

© 2019 Durban University of Technology: D Day Page 30

Substituting A for λ in the characteristic equation gives;
A2 − 5A + 4I = 0
       
6 5 2 1 1 0 0 0
i.e −5 +4 =
10 11 2 3 0 1 0 0

13 Systems of First-Order Homogeneous Ordinary Dif-

ferential Equations
Before considering the solution of systems of first order homogeneous differential equations,
recall that the differential equation y ′ (x) = ay(x) has the solution y(x) = ceax as follows;

y ′ (x) = ay(x) · · · (i)

dy
= ay
dx
Z Z
1
dy = a dx
y
ln y = ax + C ⇒ y = eax+k = eax · ek
y(x) = y = ceax · · · (ii)
where c = ek

Now, consider the solution (solving for y1 (x) and y2 (x)) of the folllowing system of ordinary
differential equations;

y1′ (x) = a11 y1 (x) + a12 y2 (x)

y2′ (x) = a21 y1 (x) + a22 y2 (x)
y1 (0) = m1 and y2 (0) = m2

The system can be expressed in matrix form as follows;

y1′
     
a11 a12 y1
= ·
y2′ a21 a22 y2
i.e
Y ′ = Y ′ (x) = AY (x) = AY ··· (iii)
where
 ′       
′ y1 a11 a12 y1 y1 (x)
Y = , A= and Y = =
y2′ a21 a22 y2 y2 (x)
subject to initial conditions    
y1 (0) m1
Y (0) = =
y2 (0) m2

© 2019 Durban University of Technology: D Day Page 31

Since (iii) has the same form as (i), we try a solution similar to (ii), of the form
 
λx c1
Y = Ce = eλx · · · (iv)
c2

Substituting (iv) into (iii) gives

λCeλx = ACeλx
⇒ AC = λC
or
(A − λI)C = 0 ··· (v)
From what we know now about Eigenvalues and Eigenvectors, this equation (v) implies that
the values of λ which satisfy (v) are Eigenvalues of A and the vectors C are the corresponding
Eigenvectors.
To solve the system of differential equations we need to find the Eigenvalues (λ1 and λ2 )
and the corresponding Eigenvectors (C1 and C2 ) of A .
Then, Y1 = C1 eλ1 x and Y2 = C2 eλ2 x are both solutions of the system of equations and so, by
superposition, the complete general solution can be expressed as

Y = aC1 eλ1 x + bC2 eλ2 x ··· (v)

Where a and b are constants found after applying the initial conditions
   
y1 (0) m1
Y (0) = =
y2 (0) m2

Example
Solve

y1′ = 2y1 + 4y2

y2′ = y1 − y2
y1 (0) = 1, y2 (0) = 0

y1′

   
′ 2 4 y1
Y (x) = AY (x) ⇒ = subject to
y2′ 1 −1 y2
   
y1 (0) 1
Y (0) = =
y2 (0) 0

© 2019 Durban University of Technology: D Day Page 32

  
2 4
Solve AC = λC or (A − λI)C = 0 where A =
1 −1
 
2−λ 4
det(A − λI) = det =0
1 −1 − λ
⇒ (2 − λ)(−1 − λ) − 4 = 0
⇒ λ2 − λ − 6 = 0 ⇒ (λ + 2)(λ − 3) = 0
⇒ λ = −2 or λ = 3

λ1 = −2
   
4 4 0 R1 4 4 0 R1
∼
1 1 0 R2 0 0 0 R2′ = R1 − 4R2
Let c2 = α ⇒ 4c1 + c2 = 0 ⇒ c1 = −α
   
−α 1
C1 = , and the simple specific Eigenvector is C1 =
α −1
∴ Y = C1 eλ1 x is a solution to the system of differential equations
e−2x
     
y1 1 −2x
i.e Y = = e =
y2 −1 −e−2x
So, y1 = e−2x and y2 = −e−2x satisfy the system of differential
equations.
λ2 = 3
   
−1 4 0 R1 −1 4 0 R1
∼
1 −4 0 R2 0 0 0 R2′ = R1 + R2
Let c2 = α ⇒ −c1 + 4c2 = 0 ⇒ c1 = 4c2 = 4α
   
4α 4
C2 = , and the simple specific Eigenvector is C2 =
α 1
∴ Y = C2 eλ2 x is a solution to the system of differential equations
     3x 
y1 4 3x 4e
i.e Y = = e =
y2 1 e3x
So, y1 = 4e3x and y2 = e3x satisfy the system of differential
equations.
The general solution to the system is:
   
λ1 x λ2 x 1 −2x 4
Y = aC1 e + bC2 e = a e +b e3x
−1 1
   
y1 (0) 1
Apply initial conditions Y (0) = =
y2 (0) 0
       
1 1 −2(0) 4 3(0) a + 4b
Y (0) = =a e +b e =
0 −1 1 −a + b

© 2019 Durban University of Technology: D Day Page 33

This yields the equations

a + 4b = 1
−a + b = 0
1 1
which have solutions, a = 5
and b = 5

Therefore, the specific solution of the system is;

    1
! 4
!
1 4 5 5
Y = 15 e−2x + 51 e3x = e−2x + e3x
−1 1 − 15 1
5

1 −2x 4 3x
!
e + e
 
y1 5 5
i.e Y = =
y2 − e1 −2x
+ 1 e3x
5 5

Therefore, y1 = 51 e−2x + 45 e3x and y2 = − 51 e−2x + 15 e3x are the specific solutions to the system.
Exercises

1. Find the general solution to the system

y1′ = 3y1 + 2y2

y2′ = y1 + 4y2

Solution:
       2x 5x

y1 2 1 2ae + be
Y = =a e2x + b e5x =
y2 −1 1 −ae2x + be5x
2. Solve the system

y1′ =y1 + 2y2

y2′ =2y1 − 2y2
y1 (0) =1, y2 (0) = −1

Solution:
       2 2x 2 −3x 
y1 1 2 2x 2 1 −3x 5
e + 5e
Y = =5 e +5 e =
y2 1 −2 5
e − 45 e−3x
1 2x

3. Solve the system

y1′ =3y1 + 11y2

y2′ =2y1 + 5y2
y1 (0) = − 1, y2 (0) = 2

© 2019 Durban University of Technology: D Day Page 34

 Solution:
! !
11 11
  √ √ √ √
y1 23−1 (4+ 23)x 23+1
Y = = 22 √ e − 22 √ e(4− 23)x
y2 1 + 23 1 − 23
√ √ √ √ !
23−1 23+1 (4− 23)x
e(4+23)x
− e
 
y1 2 2
∴ Y = = √ √
y2 e(4+ 23)x
+e (4− 23)x

14 Diagonalisation of a Matrix
If λ1 , λ2 , λ3 · · · λn are Eigenvalues and x1 , x2 , x3 · · · xn the corresponding Eigenvectors of
a matrix A of order n, then;

 
x11 x12 x13 · · · x1n

 x21 x22 x23 · · · x2n 

M = (x1 , x2 , x3 , · · · , xn ) = 
 x31 x32 x33 · · · x3n 

 .. .. .. .. .. 
 . . . . . 
xn1 xn2 xn3 · · · xnn
where the elements in column j are the elements of the eigenvector xj , is defined as the Modal
Matrix (M ) of A
The Spectral Matrix (S) of A is defined to be the diagonal matrix with the eigenvalues of A
on the main diagonal, i.e  
λ1 0 0 · · · 0
 0 λ2 0 · · · 0 
 
S =  0 0 λ3 · · · 0 
 
 .. .. .. .. .. 
 . . . . . 
0 0 0 · · · λn
Theorem
If M and S are the modal and spectral matrices respectively of A, then

M −1 AM = S

Proof:

AM = A(x1 , x2 , x3 , · · · ; xn )
= (Ax1 , Ax2 , Ax3 , · · · , Axn )
= (λ1 x1 , λ2 x2 , λ3 x3 , · · · , λn xn ), (since Ax = λx)
Also, M S = (x1 , x2 , x3 , · · · , xn )S = (λ1 x1 , λ2 x2 , λ3 x3 , · · · , λn xn )
∴ AM = M S
⇒ M −1 AM = M −1 M S = IS = S

© 2019 Durban University of Technology: D Day Page 35

Example  
2 1
The matrix A = has the following eigenvalues and eigenvectors respectively;
2 3    
1 1
λ1 = 1, λ2 = 4 and x1 = , x2 =
  −1  2
1 1 1 0
∴ M= and S = .
−1 2 0 4
det(M ) = (1)(2) − (−1)(1) = 3
 
2 1
The matrix of co-factors of M is C =
−1 1
 
T 2 −1
adj(M ) = C =
1 1
 
−1 1 1 2 −1
therefore, M = det(M )
· adj(M ) = 3 1 1
     
−1 1 2 −1 2 1 1 1
Now M AM = · ·
3 1 1 2 3 −1 2
     
1 2 −1 1 4 1 0
= · = =S
3 1 1 −1 8 0 4
Exercises
Show that M −1 AM = S for the following matrices
 
−2 −9
1. A = where
−6 1   
1 3
λ1 = 7, x1 = . and λ2 = −8, x2 =
−1 2
   
2 −2 −1 1
2. A =  −2 2 1  , where λ1 = 0; x1 =  1 ,
−1 1 5 0
   
1 1
λ2 = 3; x2 =  −1 , and λ3 = 6; x3 =
  −1 
1 −2

Before considering the solution of second order ordinary differential equations, we note the
following;
1. eax = cosh ax + sinh ax
eax +e−ax eax −e−ax 2eax
cosh ax + sinh ax = 2
+ 2
= 2
= eax

2. e−ax = cosh ax − sinh ax

© 2019 Durban University of Technology: D Day Page 36

 eax +e−ax eax −e−ax
cosh ax − sinh ax = 2
− 2
= 2eax
2
= e−ax

3. If y ′′ = −a2 y, then y = A cos ax + B sin ax

since, the auxiliary equation is m2 + a2 = 0 ⇒ m = ±ja

4. If y ′′ = a2 y, then y = A cosh ax + B sinh ax

since, the auxiliary equation is m2 − a2 = 0 ⇒ m = ±a
Thus, y = c1 eax + c2 e−ax and by 1 and 2 above,
y = c1 (cosh ax + sinh ax) + c2 (cosh ax − sinh ax
= (c1 + c2 ) cosh ax + (c1 − c2 ) sinh ax
= A cosh ax + B sinh ax

15 Systems of Second-Order Homogeneous Ordinary Dif-

ferential Equations
Solve for y1 (x) and y2 (x) in the folllowing system of ordinary differential equations;

y1′′ (x) = a11 y1 (x) + a12 y2 (x)

y2′′ (x) = a21 y1 (x) + a22 y2 (x)
subject to
y1 (0) = m1 and y2 (0) = m2 and
y1′ (0) = n1 and y2′ (0) = n2

The system can be expressed in matrix form as follows;

y1′′
     
a11 a12 y1
= ·
y2′′ a21 a22 y2
i.e as
Y ′′ = Y ′′ (x) = AY (x) = AY ··· (iii)
where
y1′′
     
′′ a11 a12 y1 (x)
Y = , A= and Y =
y2′′ a21 a22 y2 (x)
subject to boundary conditions
   
y1 (0) m1
Y (0) = = and
y2 (0) m2
 ′   
′ y1 (0) n1
Y (0) = =
y2′ (0) n2

© 2019 Durban University of Technology: D Day Page 37

Given the system Y ′′ = AY , let us assume that the solution can be expressed in terms of the
modal matrix, i.e  
g1 (x)
Assume that Y = M G, where M is the modal matrix of A, and G = is a vector
g2 (x)
whose elements are functions of x, i.e g1 (x) and g2 (x)
Thus, Y ′′ = AY where Y = M G. Thus

Y ′′ = AM G · · · (i)

Now, by differentiating Y = M G twice with respect to x yields

Y ′′ = M G′′ · · · (ii)

Equating (i) and (ii) yields

M G′′ = AM G · · · (iii)
Then, pre- multiplying (iii) by M −1 yields

M −1 M G′′ = M −1 AM G

i.e G′′ = SG
which in matrix form is
g1′′ (x)
      
λ1 0 g1 (x) λ1 g1 (x)
= =
g2′′ (x) 0 λ2 g2 (x) λ2 g2 (x)

We can now find expressions for g1 and g2 by solving the two equations g1′′ = λ1 g and g2′′ = λ2 g2 .
The general solution for the system can now be derived by substituting these functions into
Y = M G. The specific solution can then be established by applying the initial conditions y(0)
and Y ′ (0)
Example
Solve the system

y1′′ = 3y1 + 10y2

y2′′ = 2y1 + 4y2 , subject to

y1 (0) = 0, y2 (0) = 1 and y1′ (0) = 1, y2′ (0) = 0

 
3 10
A=
2 4  
3−λ 10
det(A − λI) = det =0
2 4−λ
⇒ (3 − λ)(4 − λ) − 20 = 0 ⇒ λ2 − 7λ − 8 = 0
⇒ (λ − 8)(λ + 1) = 0 ⇒ λ = 8 or λ = −1
For λ = 8

© 2019 Durban University of Technology: D Day Page 38

   
−5 10 0 R1 −5 10 0 R1
∼
2 −4 0 R2 0 0 0 R2′ = 2R1 + 5R2
Let c2 = α. R1 ⇒ −5c1 + 10c2 = 0 ⇒ c1 = 2c2 = 2α
 
2
∴ λ1 = 8, and C1 =
1

For λ = −1
   
4 10 0 R1 4 10 0 R1
∼
2 5 0 R2 0 0 0 R2′ = R1 − 2R2
Let c2 = α. R1 ⇒ 4c1 + 10c2 = 0 ⇒ c1 = − 25 c2 = − 52 α
 
5
∴ λ2 = −1, and C1 =
−2
   
2 5 8 0
M= and S =
1 −2 0 −1
 ′′      
′′ g1 8 0 g1 8g1
G = = =
g2′′ 0 −1 g2 −g2
√ √ √
g1′′ = 8g1 ⇒ m = ±2 2 ⇒ g1 = A cosh 2 2x + B sinh 2 2x
g2′′ = −g2 ⇒ m = ±j ⇒ g2 = C cos x + D sin x
∴ Y = MG ⇒
    √ √ 
y1 2 5 A cosh 2 2x + B sinh 2 2x
Y = =
y2 1 −2 C cos x + D sin x
   √ √ 
y1 2A cosh 2√2x + 2B sinh √
2 2x + 5C cos x + 5D sin x
=
y2 A cosh 2 2x + B sinh 2 2x − 2C cos x − 2D sin x
This is the general solution, and
 ′   √ √ √ √ 
y1 4√2A sinh 2√2x + 4√2B cosh 2√2x − 5C sin x + 5D cos x
=
y2′ 2 2A sinh 2 2x + 2 2B cosh 2 2x + 2C sin x − 2D cos x
Now, applying initial conditions
     
y1 (0) 0 2A + 5C
Y (0) = = =
y2 (0) 1 A − 2C
Solving simultaneously gives C = − 92 and A = 59
 ′     √ 
′ y1 (0) 1 4√2B + 5D
Y (0) = = =
y2′ (0) 0 2 2B − 2D
1 1
Solving simultaneously gives B = √
9 2
and D = 9

© 2019 Durban University of Technology: D Day Page 39

 √ √ √ !
10
cosh 2 2x + 92 sinh 2 2x − 10 cos x + 59 sin x
 
y1 9 9
∴ = √ √
y2 5
cosh 2 2x + 1
√ sinh 2 2x + 49 cos x − 92 sin x
9 9 2

Exercises:

1. Solve the system;

ÿ(t) = 2x(t) − y(t)

ẍ(t) = −2x(t) + y(t)

Answer: √ √
x(t) = A + Bt − D cosh 3 t − E sinh 3 t
√ √
y(t) = 2A + 2Bt + D cosh 3 t + E sinh 3 t

2. Solve the system;

y1′′ = −y1 − 4y2

y2′′ = −y1 + 2y2
y1 (0) = 0, y2 (0) = 1, y1′ (0) = −1, y2′ (0) = 0

Answer: √ √ √ √
y1 = − 54 cos 3t + 3√1 3 sin 3 t + 54 cosh 2 t − 3√4 2 sinh 2 t
√ √ √ √
y2 = 54 cos 3t − 3√1 3 sin 3 t + 15 cosh 2 t − 3√1 2 sinh 2 t

© 2019 Durban University of Technology: D Day Page 40