Introduction to Matrices

November 2014
• Augmented matrix of the system: A matrix
derived from a system of linear equations
(each written in standard form with the
constant term on the right)

• Coefficient matrix of the system: the matrix

derived from the coefficients of the system
(but not including the constant terms)
 System:
x − 4 y + 3z = 5
−x + 3y − z = −3
2x − 4z = 6
Augmented Matrix: Coefficient Matrix:
 1 −4 3  5   1 −4 3 
 
− 1 3 − 1  − 3 −1 3 −1
 
 2 0 − 4  6   2 0 −4 
 Elementary Row Operations
1. Interchange two equations.
Interchange two rows

2. Multiply an equation by a nonzero constant.

Multiply a row by a nonzero constant

3. Add a multiple of an equation to another

equation.
Add a multiple of a row to another row
 Row-Echelon and Reduced Row-Echelon
Forms
• These are Row-equivalent forms of a matrix.
• One can easily solve a system of linear equations
when matrices are in one of these forms.
• A matrix is in reduced row-echelon form if it
satisfies the following conditions:
1. If a row does not consist entirely of zeros,
then the first nonzero number in the row is a 1
( called a leading 1).
 2. If there are any rows that consist entirely of
zeros, then they are grouped together at the
bottom of the matrix.
3. In any two successive rows that do not consist
entirely of zeros, the leading 1 in the lower
row occurs farther to the right than the
leading 1 in the higher row.
4. Each column that contains a leading 1 has
zeros everywhere else in that column.

• A matrix that has the first three properties is said

to be in row-echelon form.
 Row-Echelon and Reduced Row-Echelon
Forms…
• Thus, with any real numbers substituted for the
*'s, all matrices of the following types are in row-
echelon form:
 Row-Echelon and Reduced Row-Echelon
Forms…
• Moreover, all matrices of the following types are
in reduced row-echelon form:
 Elimination Methods
Example
Solve the system
 y+z =3

 2x + 4 y − 2z = 8

 4 x + 8y + 2z = 22

Solution: Write augmented matrix.

0 1 1  3 
 2 4 − 2  8 
 
4 8 2  22
Step 1. Locate the leftmost column that does not
consist entirely of zeros.
0 1 1  3 
2 4 − 2  8 
 
4 8 2  22
Leftmost nonzero column
• Step 2. Interchange the top row with another, if
necessary, to bring a nonzero entry to the top of the
column found in Step 1.
R2 2 4 − 2  8  Interchange R1 and R2 so
R1 0 1 1  3  first column has leading 1
 
4 8 2  22 in upper left corner.
• Step 3. If the entry that is now at the top of the
column found in Step 1 is a, multiply the first row by
1/a in order to introduce a leading 1.
1 2 − 1  4  R1 → 1 2 R1
0 1 1  3 
 
4 8 2  22
• Step 4. Add suitable multiples of the top row to the
rows below so that all entries below the leading 1
become zeros.
1 2 −1 M 4 
 
0 1 1 M 3
0 0 6 M 6  R3 →R3 − 4R1
• Step 5. Now cover the top row in the matrix and
begin again with Step 1 applied to the submatrix
that remains. Continue in this way until the entire
matrix is in row-echelon form.
1 2 − 1  4
0 1 1  3
 
0 0 1  1 R3 → 1 6 R3
• The entire matrix is now in row-echelon form.

• To find the reduced row-echelon form we need the

following additional step.
•
• Step 6. Beginning with the last nonzero row and
working upward, add suitable multiples of each row
to the rows above to introduce zeros above the
leading 1’s.
1 2 0  5 R1 → R1 + R3
0 1 0  2 R → R − R
  2 2 3

0 0 1  1

1 0 0  1 R1 → R1 − 2 R2
0 1 0  2
 
0 0 1  1
• The last matrix is in reduced row-echelon form.
• If we use only the first five steps, the above
procedure produces a row-echelon form and is
called Gaussian elimination.
• Carrying the procedure through to the sixth step
and producing a matrix in reduced row-echelon
form is called Gauss-Jordan elimination.
• Remark: Every matrix has a unique reduced row-
echelon form but a row-echelon form of a given
matrix is not unique. That is, one will arrive at the
same reduced row-echelon form for a given matrix
no matter how the row operations are varied. In
contrast, different sequences of row operations can
produce different row-echelon forms.
• Example (Using Elementary Row Operations)
• In the left column we solve a system of linear
equations by operating on the equations in
the system, and in the right column se solve
the same system by operating on the rows of
the augmented matrix.

x1 + x 2 + 2x 3 = 9 1 1 2 9
 
2x1 + 4 x 2 − 3x 3 = 1  2 4 −3 1
3x1 + 6x 2 − 5x 3 = 0 3 6 −5 0
Add – 2 times the first equation Add – 2 times the first row to
to the second to obtain the second to obtain
x1 + x 2 + 2x 3 = 9
1 1 2 9 
2x 2 − 7x 3 = −17  
 0 2 −7 −17 
3x1 + 6x 2 − 5x 3 = 0
3 6 −5 0 
Add – 3 times the first equation Add – 3 times the first row to
to the third to obtain the third to obtain
x1 + x 2 + 2x 3 = 9 1 1 2 9 
 
2x 2 − 7x
 3 = −17 0 2 −7 −17 
3x 2 −11x 3 = −27 0 3 −11 −27 
Multiply the second equation Multiply the second row by ½
by ½ to obtain to obtain
x1 + x 2 + 2x 3 = 9 1 1 2 9 
7 17  7 17 
x2 − x3 = − 0 1 − − 
2 2 0 2 2
 3 −11 −27 
3x 2 −11x 3 = −27
Add – 3 times the second Add – 3 times the second row
equation to the third to obtain to the third to obtain
x1 + x 2 + 2x 3 = 9  
 1 1 2 9 
x 2 − 7 2 x 3 = −17 2 0 1 −7 2 −17 2 
 
− 12 x 3 = − 3 2 0 0 − 12 3
− 2
Multiply the third equation by Multiply the third row by – 2
– 2 to obtain to obtain
x1 + x 2 + 2x 3 = 9 1 1 2 9 
 
x 2 − 7 2 x 3 = −17 2 0 1 −7 2
17
− 2
0 0 1 3 
x3 = 3
Add – 1 times the second Add – 1 times the second row
equation to the first to obtain to the first to obtain
11  35
x1 + 0x 2 + 2 x 3 = 2 1 11 35 
0 2 2

x 2 − 7 2 x 3 = −17 2 0 1 −72 −17 2 
0 
x3 = 3  0 1 3 
Add −11 2 times the third Add −11 2 times the
equation to the first and 7
third row to the first and2
7 times the third times the third row to the
2
equation to the second to second to obtain
obtain 
1 0 0 1
x1 =1  
0 1 0 2
x2 =2 0 0 1 3
x3 =3
The solution
x1 =1, x 2 = 2, x 3 = 3 is
now evident.
 Example:
Let A be the matrix
1 −1 −1 2 
 
2 1 −1 2 
1 −3 2 7 
Reduce A to row-echelon form.
Solution:

1.Get a 1 as the first entry of Row 1. Done.
2.Use this first leading 1 to clear out column 1
as follows :
 1 −1 −1 2 
 
0 3 1 −2 R2 →R2 − 2R1
0 −2 3 5  R3 →R3 − R1

3. Get a leading 1 as the second entry of Row 2,

for example as follows :
1 −1 −1 2 
 
0 1 4 3 R2 →R2 + R3
0 −2 3 5 
4. Use this leading 1 to clear out whatever
appears below it in Column 2 :
1 −1 −1 2 
 
0 1 4 3 
0 0 11 11 R3 →R3 + 2R2
5. Get a leading 1 in Row 3:
1 −1 −1 2 
 
0 1 4 3
0 0 1 1  R → 1 R
3 3
11
This matrix is now in row-echelon form.
 Rank of a Matrix
• Definition: Let A be a matrix. The rank of A,
denoted rank (A) or r(A) is the number of
leading 1’s in a row-echelon form obtained
from A by Gaussian elimination as above.

Rank of a matrix by triangular form:

• The rank of a matrix is equal to the number of
non-zero rows in upper triangular matrix.
• Example: The rank of the matrix
1 −1 −1 2 
 
A = 2 1 −1 2 
1 −3 2 7 
in Example above is 3, since the row-echelon
form obtained had 3 leading 1’s (one in each
row). That is  
 1 −1 −1 2
 
A ~ 0 1 4 3
0 0 1 1 
 r(A) = 3
 Example : Find the rank of the matrix
1 2 3 2 
 
A = 2 3 5 1 
1 3 4 5 
Solution: We have
1 2 3 2 
  
A = 2 3 5 1 
1 3 4 5 
• Applying elementary row operation to reduce
the given matrix in row-echelon form, we
have
1 2 3 2 
 
A ~ 0 −1 −1 −3 R2 →R2 − 2R1
0 1 1 3  R3 →R3 − R1

1 2 3 2 
 
A ~ 0 −1 −1 −3
0 0 0 0  R3 →R3 + R2
Or

1 2 3 2 
 
A ~ 0 1 1 3 R2 →(−1)  R2
0 0 0 0 

which is in row-echelon form and has 2 leading

1’s. Thus, the rank of the matrix is 2.
 Inverse of Matrices
• Let A be an n × n matrix. We know that
multiplication by the identity matrix leaves a
matrix unchanged, that is, AI = IA = A, where I is
the n × n identity matrix.
• Definition
Let A and B be n × n matrices. Suppose that
AB = BA = I.
Then B is called the inverse of A and is written as
A-1. Thus we have
AA-1 = A-1 A= I.
• Now A-1 may not exist, i.e., all square matrices do not
necessarily have an inverse.
• If A does have an inverse, then A is said to be
nonsingular or invertible.
• If A does not have an inverse, then A is said to be
singular or noninvertible.
• Theorem: If a matrix has an inverse, then the
inverse is unique
Proof
Let B and C be inverses of A. Then BA=AC=I.
Therefore
B = BI = B(AC) = (BA)C = IC = C
which completes the proof
 Inverse of a Square Matrix
Method for computing the inverse of a square
matrix
1.Form the matrix [A:I ] by adjoining A and I.
Note it will be an n×2n matrix.
2.If possible, row reduce A to I using Gauss-
Jordan elimination on the entire matrix [A :
I]. The result will be the matrix [ I : A-1].
That is, [A | I] ⎯row operations
⎯⎯⎯ −1
⎯→ [I | A ]
If this is not possible, then A does not have
an inverse.
 Inverse of a Square Matrix
Method for computing the inverse of a square matrix
• Step1. Form the n ×2n matrix [A|In], obtained by
adjoining the identity matrix In to the given matrix A.
• Step 2. Transform the matrix obtained in step 1 to
reduced row echelon form by using elementary row
operations. Remember that whatever we do to a row of
A we also do to the corresponding row of In.
• Step 3. Suppose that Step 2 has produced the matrix
[C |D] in reduced echelon form.
a) If C = In, then D = A-1
b) If C ≠In, then C has a row of zeros. In this case A is
singular and A-1 does not exist.
 Inverse of a Square Matrix
• Example: Find the inverse of the matrix
1 1 1
 
A = 0 2 3

5 5 1

• Solution: Step 1. The 3×6 matrix [A|I] is
1 1 11 0 0
  
A | I3  = 0 2 30 1 0

5 5 10 0 1

• Step 2. We transform the matrix obtained in Step 1
to reduced echelon form:
1 1 1 1 0 0
 3 1  1
0 1 0 0  R2 := R2
0 2 2  2
 0 −4 −5 0 1  R3 := R3 − 5R1
 1 1 
1 0 − 1 − 0 
 3
2
1
2  R1 := R1 − R2
0 1 0 0 
 2 2  1
0 0 1 5
0 −
1  R3 := − R3
 4 4
 4 
 13 1 1 1
 − −  R1 := R1 + R3
1 0 0 8 2 8 2
15 1 3 3
0 1 0−  R2 := R2 − R3
 8 2 8  2
0 0 1 5
0 −
1

 4 4

• Step 3. Since C=I, we conclude that D=A-1. Hence

 13 1 1
 8 − − 
2 8
 15 1 3 
A −1 = − 
 8 2 8 
 5 0 −
1
 4 4
• It is easy to verify that AA-1=A-1A=I
 Inverse of a Square Matrix
• Example: Calculate A−1, if it exists, where
2 4 6 
 
A = 4 5 6 

3 1 −2 

• Solution:
2 4 6 1 0 0
 
A | I  = 4 5 6 0 1 0

3 1 −2 0 0 1

  1  1
1 2 3 0 0  R1 := R1
 2  2
~ 4 5 6 0 1 0
 
3 1 −2 0 0 1
 

 1 
1 2 3 0 0
 2 
~ 0 −3 −6 −2 1 0 R2 := R2 − 4R1
 3 
0 −5 −11 − 0 1  R3 := R3 − 3R1
 2 
  1 
 0 0
1 2 3 2 
2 1 1
~ 0 1 2 − 0  R2 := − R2
 3 3  3
0 −5 −11 3
− 0 1

 2 


 5 2 
 − 0
1 0 −1 6 3  R1 := R1 − 2R2
2 1
~ 0 1 2 − 0
 3 3 
0 0 −1 11 5  R3 := R3 + 5R2
− 1

 6 3 

  5 2 
 − 0
1 0 −1 6 3 
2 1
~ 0 1 2 − 0
 3 3 
0 0 1 11 5  R3 := −R3
− −1

 6 3 


 8 7 
 − −1
1 0 0
13
3 3
11  R1 := R1 + R3
~ 0 1 0 − 2  R2 := R2 − 2R3
 3 3 
0 0 1 11
−
5
−1

 6 3 

• Hence  8 7 
−3 3
−1
 13 11 
A −1 = − 2 
 3 3 
 − 11 5
−1
 6 3 
and it can be checked that AA-1=A-1A=I.
• Example: Calculate A−1, if it exists, where

1 −3 4
 
A = 2 −5 7 .

0 −1 1
Solution:
1 −3 41 0 0
 
A | I  = 2 −5 70 1 0

0 −1 10 0 1


1 −3 4 1 0 0 
 
~ 0 1 −1 −2 1 0  R2 := R2 − 2R1
0 −1 1 0 0 1 
Solution:
1 −3 41 0 0
 
A | I  = 2 −5 70 1 0

0 −1 10 0 1


1 −3 4 1 0 0 
 
~ 0 1 −1 −2 1 0  R2 := R2 − 2R1
0 −1 1 0 0 1 
 1 0 1 −5 3 0 R1 := R1 + 3R2
 
~ 0 1 −1 −2 1 0
0 0 0 −2 1 1 R3 := R3 + R2
• We cannot go any further since the left hand matrix
is now in reduced row echelon form, so we cannot
reduce A to the identity matrix.

• Thus A is not invertible.

 Theorem: (Properties of the Inverse)
a) If A is a nonsingular matrix, then A-1 is a
nonsingular and (A-1)-1= A.
b) If A and B be nonsingular matrices, then AB is
nonsingular and (AB)-1= B-1A-1 .
c) If A is a nonsingular matrix, then (AT)-1= (A-1)T
d) If A and B be nonsingular matrices and k a
−1 1 −1
nonzero scalar, then ( kA) = A
k

e) (A2)-1= (A-1)2 and in general, (Am)-1= (A-1)m , for

nonnegative integer m.
• Proof
a) A-1 is nonsingular if we can find a matrix B such
that A-1B = BA-1 = I
Now A-1 A= AA-1 = I.
Thus B = A is an inverse of A-1, and since the
inverse is unique, we conclude that (A-1)-1= A.

b)We have
(AB)(B-1A-1) = A(BB-1)A-1 )= AIA-1 = AA-1 = I
and
(B-1A-1)(AB) = B-1 (A-1A)B = B-1 I B = B-1 B = I
• Therefore, AB is nonsingular. Since the inverse of
a matrix is unique, we conclude that
(AB)-1= B-1A-1 .

c)We have AA-1 = I and A-1 A= I.

Taking transposes, we obtain
(AA-1 )T = IT = I and (A-1 A )T = IT = I
Then
(A-1)T AT = I and AT (A-1)T = I
These equations imply that
(AT)-1= (A-1)T