lOMoARcPSD|51865458

SMA 2201 Linear Algebra I Notes 4

Bsc. mathematics and computer science (Jomo Kenyatta University of Agriculture and
Technology)

Scan to open on Studocu

Studocu is not sponsored or endorsed by any college or university

Downloaded by Steve Lewis ([email protected])
 lOMoARcPSD|51865458

MATRIX ALGEBRA
Definition: A matrix A is a rectangular array of scalars (numbers), i.e., a collection of
numbers ordered by rows and columns. The ij th component of the matrix A, denoted by
aij , is the number in the intersection of the ith row and j th column of A. The components
(elements/entrie)s of a matrix are usually enclosed in parentheses, brackets or braces. The
matrix A is sometimes written as A = [aij ]. A matrix with all components equal to zero
is called the zero matrix. For each matrix A = [aij ], there corresponds a unique matrix
−A = [−aij ], called the additive inverse of A, obtained by negating each entry of A.
Definition: A matrix A is said to have order (shape/size) m × n, read “m by n” if A has m
rows and n columns. In this case, if m = n (i.e. A has same number of rows and columns),
A is called a square matrix; otherwise, A is said to be rectangular. A matrix consisting of
a single row is called a row vector, while a matrix consisting of a single column is called a
column vector. Conventionally, a 1 × 1 matrix is a scalar and vice versa.
Definition: Two matrices A = [aij ] and B = [bij ] are equal if they have the same shape and
corresponding entries are equal, i.e., aij = bij for each i = 1, 2, · · · , m and j = 1, 2, · · · , n.

Matrix Addition
Let A = [aij ] and B = [bij ] be two m × n matrices. Then the sum of A and B is the m × n
matrix A + B obtained by adding corresponding entries. Matrix subtraction is done in a
similar manner except that entries are subtracted instead of being added. For two matrices
of the same shape, the difference A − B is the matrix A − B = A + (−B).
Example:    
1 −2 3 −1 0 2
Let A = and B = . Then
4 5 −6 −5 −4 5
     
1 −2 3 −1 0 2 0 −2 5
A+B = + =
4 5 −6 −5 −4 5 −1 1 −1

and      
1 −2 3 −1 0 2 2 −2 1
A−B = − = .
4 5 −6 −5 −4 5 9 9 −11

Properties of Matrix Addition

Let A, B, and C be m × n matrices. Then:
(i ) A + B is an m × n matrix (closure property).
(ii ) (A + B) + C = A + (B + C) (associative law for matrix addition).
(iii ) A + B = B + A (commutative law for matrix addition).
(iv ) There is an m × n matrix denoted by 0 consisting of all zeros (the zero matrix) such
that A + 0 = 0 + A = A (existence of additive identity).
(v ) There is an m × n matrix (−A) such that A + (−A) = (−A) + A = 0 where 0 is the
zero matrix (existence of additive inverses).

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

Scalar Multiplication
Let A be an m × n matrix. The product of a scalar α times a matrix A, denoted by αA, is
the matrix obtained by multiplying each entry of A by α.
Example: 
−5 0 2
Let A =  3 4 −3  and let α = −1/2. Then
1 −1 1
   
−5 0 2 5/2 0 −1
1
αA = −  3 4 −3  =  −3/2 −2 3/2  .
2
1 −1 1 −1/2 1/2 −1/2

Properties of Scalar Multiplication

Let A and B be a m × n matrices and let α and β be scalars. Then:
(i ) αA is an m × n matrix (closure property).
(ii ) (αβ) A = α (βA) (associative law for scalar multiplication).
(iii ) α (A + B) = αA + αB (scalar multiplication is distributive over matrix addition).
(iv ) (α + β) A = αA + βA (scalar multiplication is distributive over scalar addition).
(v ) 1A = A (identity property - the number 1 is an identity element under scalar
multiplication).

Transpose
Let A be an m × n matrix. The transpose of A, denoted by AT , is the n × m matrix obtained
by interchanging rows and columns of A, i.e., the ith row of A is the ith column of AT and
the j th column of A is the j th row of AT .
Example:  
  −1 2 3
−1 0 2 5 
 . Then AT =  0 4 −1  
Let A = 2 4 −3 2  2 −3 0 .
3 −1 0 −1
5 2 −1

Properties of the Transpose

Let A and B
be a m × n matrices and let α be a scalar. Then:
T
(i ) AT = A.
(ii ) (A + B)T = AT + B T .


(iii ) (αA)T = α AT .

Symmetry
Let A = [aij ] be an n × n matrix. Then A is said to be a symmetric matrix if A = AT , i.e.,
aij = aji . In other words, A is symmetric if the entries of A are symmetrically located about

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

the main diagonal (the line from the upper-left-hand corner to the lower-right-hand corner).
Example: 
1 −2 0
Let A =  −2 4 −3 . Then A is symmetric since A = AT .
0 −3 0

Trace
Let A = [aij ] be an n × n matrix. The trace of A, denoted by trace(A), is the sum of the
entries on the main diagonal. That is,
n
X
trace(A) = a11 + a22 + · · · + ann = aii .
i=1

Example:
 
−3 2 1
Let A =  4 −1 −2 . Then trace(A) = (−3) + (−1) + 0 = −4.
5 0 0

Linear Combinations
Let A1 , A2 , · · · , An be m × n matrices and let α1 , α2 , · · · , αn be scalars. Then, the expression
n
X
α1 A1 + α2 A2 + · · · + αn An = αi Ai
i=1

is called a linear combination of the A1 , A2 , · · · , An .

Example:     
4 −1 −3 −6 0 6
Let A = , B = , and C = . Let α = 2, β = −1, and
5 0 1 2 −2 −4
λ = 21 . Then
     
4 −1 −3 −6 1 0 6
αA + βB + λC = 2 + (−1) +
5 0 1 2 2 −2 −4
 
11 7
=
8 −4
=D

So, D is a linear combination of A, B, and C.

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

Matrix Multiplication
 
b1

 b2 
 
Let a = a1 a2 · · · an and b =  ..  be two n-vectors. Then the scalar product
e e  . 
bn

Xn
a · b = a1 b1 + a2 b2 + · · · + an bn = ai bi
e e i=1

forms the basis for matrix multiplication.

Definition: Let A = [aij ] be an m × n matrix and let and B = [bij ] be an s × t matrix.
Then the product AB exists if and only if n = s, i.e., if and only if A has exactly as many
columns as B has rows, and we say that A and B are compatible for multiplication in the
order AB. In this case, AB is the m × t matrix whose ij th entry is the dot product of the ith
row of A and the j th column of B.
Example:    
0 1 −1 0
1. Let A = , and B = . Then, the product AB exists and has shape
0 1 −1 0
    
0 1 −1 0 −1 0
2 × 2. In this case, AB = = .
0 1 −1 0 −1 0
The product BA is also defined and is the 2 × 2 matrix
    
−1 0 0 1 0 −1
BA = = .
−1 0 0 1 0 −1
 
  2 −3 1 0
1 2 5
2. Let A = , and B =  8 −1 2 −3 . Then,
−3 −4 0
2 0 −1 −1
 
  2 −3 1 0  
1 2 −5   8 −5 10 −1
AB = 8 −1 2 −3 = .
−3 −4 0 −38 13 −11 12
2 0 −1 −1
 
2 −3 1 0  
1 2 −5
However, BA =  8 −1 2 −3  is not defined.
−3 −4 0
2 0 −1 −1
Remark : In either example above AB 6= BA. This shows that matrix multiplication is not
a commutative operation.

Properties of MatrixMultiplication
For conformable matrices, we have the following:
(i ) A (B + C) = AB + AC and (D + E) F = DF + EF (distributivity of matrix
multiplication over matrix addition).
(ii ) A (BC) = (AB) C (associative law).

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

(iii ) (AB)T = B T AT .

Identity Matrix
 
1 0 ··· 0
 0 1 ··· 0 
 
The n × n matrix In =  .... . . ..  with 1’s on the main diagonal and 0’s elsewhere is
 . . . . 
0 0 ··· 1
called the identity matrix of order n.
In the notation In , the subscript is suppressed if the size is obvious from the context. For
every m × n matrix, AIn = A and Im A = A.

ROW ECHELON FORM

Definition: A matrix E is in row echelon form if the following conditions hold:
(i ) all rows (if any) consisting entirely of zeros appear below non-zero rows (i.e., they
appear at the bottom of the matrix).
(ii ) each leading entry (pivot), i.e., the first (left-most) non-zero entry, of a row is in a
column to the right of the leading entry of the row above it.
(iii ) all entries in a column, below a leading entry, are zero.
If a matrix E in echelon form satisfies the following additional conditions, then it is said to
be in reduced row echelon form.
(iv ) the leading entry in each non-zero row is 1.
(v ) each leading 1 is the only non-zero entry in its column.
Definition: Let A be an m × n matrix matrix with an echelon form E. Then a position
corresponding to a pivot of a nonzero row is called a pivot (pivotal) position. The pivot
columns, also called basic columns, of A are the columns of A that contain the pivot positions.
Note: In the preceeding definition:
(i ) The basic columns are columns of A and not the echelon form E.
(ii ) Each nonbasic column in A is a linear combination of the basic columns in A.
(iii ) Because of the flexibility in choosing row operations to reduce A to a row echelon
form E, the “form” E is not unique. However, the reduced row echelon form of A is
unique.
(iv ) The pivot positions in E are unique. Accordingly, the number of pivots in E, which
is equal to the number of basic columns in A, or the number of nonzero rows in E,
is a constant.
Definition: Let A be an m×n matrix with an echelon form E. Then the rank of A, denoted
by R(A), is the number of pivots in E.
Examples:  
    −5 −1 0
1 2 3 2 −3 2 1  0 3 −2 
Matrices  0 1 5 ,  0 1 −4 8 , and   0
 are in echelon form while
5 0 0 
0 0 0 0 0 0 2
0 0 0

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

 
   1  0 −1
1 0 0 0 1 0 2 5  0
 0 1 0 0 ,  0 1 3 6 , and  1 2  are in reduced echelon form.
 0 0 0 
0 0 0 0 0 0 0 0
0 0 0

Elementary Row Operations used in Reducing a Matrix to Echelon Form

(i ) Multiply rows by non-zero constants.
(ii ) Add a scalar multiple of one row to another.
(iii ) Permute (interchange) rows.
The process of applying these operations to simplify a matrix is called row reduction.
These elementary operations are used to determine the number of linearly independent vec-
tors in a given set of vectors. To do this, form a matrix using the vectors and then reduce
the matrix to echelon form. If one of the rows has all entries zero, then the set is linearly
dependent. The number of linearly independent vectors in the set is the number of non-zero
rows in the echelon form matrix; this just the rank of the matrix.
Theorem: The number of linearly independent columns of a matrix equals the number of
linearly independent rows.
Examples:
1. Determine the dimension of the space spanned by the vectors (1, 1, 2), (1, 0, 3), and
(1, 0, 1).
Solution:    
1 1 2 R1 − R2 1 1 2
 1 0 3  −→  0 1 −1 
1 0 1 R2 − R3 0 0 2
From the echelon form, R(A) = 3. So, the three rows (vectors) are linearly independent
and the dimension of the space spanned by the three vectors is 3. Therefore, the three
vectors form a basis for the vector space.
To obtain the reduced echelon form of the original matrix:
     
1 1 2 R1 − R2 1 0 3 R2 + R3 1 0 0
 0 1 −1  −→  0 1 −1  −→  0 1 0 .
1
0 0 2 R
2 3
0 0 1 R1 − 3R3 0 0 1
2. How many of the (1, 2, 1, 4), (−1, 0, −1, 0), (3, 1, 3, 2), and (−1, −1, −1, −2) vectors are
linearly independent?
Solution:
     
1 2 1 4 R1 + R2 1 2 1 4 1 1 2 1 4
 −1 0 −1 0  −→  0 2 0 4  2 R2 − R3  0 2 0 4 
    −→  
 3 1 3 2  3R2 + R3  0 1 0 2   0 0 0 0 .
R3 − R4
−1 −1 −1 −2 R2 − R4 0 1 0 2 0 0 0 0
In this case, two rows have not been reduced to all entries zero. So, only two rows are
linearly independent (however, the four, or any three of the vectors, are dependent).

The dimension of the space spanned by the four vectors is two. So, the space spanned

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

by the four vectors can also be spanned by only two of them. A basis for this space is
{(1, 2, 1, 4) , (−1, 0, −1, 0)} or even {(1, 2, 1, 4) , (0, 2, 0, 4)}.
To obtain the reduced echelon form of the original matrix:
     
1 2 1 4 1 2 1 4 1 0 1 0
 0 2 0 4  1
R  0 1 0 2  R1 − 2R2  0 1 0 2 
  2 2    .
 0 0 0 0  −→  0 0 0 0  −→  0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0

THEORY OF DETERMINANTS
The determinant function is only defined for n × n (square) matrices. If A is an n × n matrix,
the determinant of A is denoted det A or |A|.
Definition (2 × 2 Determinant):
 
a11 a12
Let A = be a 2 × 2 matrix. Then
a21 a22

a11 a12
det A =
a21 a22
= a11 a22 − a12 a21 . . . . . (1)

Definition (3 × 3 Determinant):
 
a11 a12 a13
Let A =  a21 a22 a23 . Then,
a31 a32 a33

a11 a12 a13

det A = a21 a22 a23
a31 a32 a33
a22 a23 a21 a23 a21 a22
= a11 − a12 + a13 . . . . . (2)
a32 a33 a31 a33 a31 a32

Take note of the minus sign before the second term on the right of (2).
Examples:
 
3 5 2
1. Let A =  4 2 3 . Calculate |A|.
−1 2 4
Solution:

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

3 5 2
|A| = 4 2 3
−1 2 4
2 3 4 3 4 2
=3 −5 +2
2 4 −1 4 −1 2
= 3 (2) − 5 (19) + 2 (10)
= −69.
2 −3 5
2. Calculate 1 0 4 .
3 −3 9
Solution:

2 −3 5
0 4 1 4 1 0
1 0 4 =2 − (−3) +5
−3 9 3 9 3 −3
3 −3 9
= 2 (12) + 3 (−3) + 5 (−3)
= 0.
There exists another method of calculating 3 × 3 determinants. From equation (2) above we
have
det A = a11 (a22 a33 − a23 a32 ) − a12 (a21 a33 − a23 a31 ) + a13 (a21 a32 − a22 a31 )
= a11 a22 a33 + a12 a23 a31 + a13 a21 a32 − a11 a23 a32 − a12 a21 a33 − a13 a22 a31 . . . . . (3)
We write A and adjoin its first two columns to it:
a11 a12 a13 a11 a12
a21 a22 a23 a21 a22
a31 a32 a33 a31 a32
We then calculate the six products, put minus signs before the products with arrows pointing
upwards, and add. This gives the sum in equation (3).
Example:
3 5 2
Calculate 4 2 3 .
−1 2 4
Solution:
3 5 2 3 5
Writing 4 2 3 4 2 and multiplying as indicated, we obtain
−1 2 4 −1 2
|A| = 3 (2) (4) + 5 (3) (−1) + 2 (4) (2) − (−1) (2) (2) − (2) (3) (3) − (4) (4) (5)
= 24 − 15 + 16 + 4 − 18 − 80
= −69.
Note: This method will not work for n × n determinants if n > 3.

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

n × n Determinants
 
a11 a12 a13  
  a22 a23
Let A = a21 a22 a23 . Then M11 = is the matrix obtained by deleting the
a32 a33
a31 a32 a33  
a21 a23
first row and first column of A; M12 = is the matrix obtained by deleting the
a31 a33

a21 a22
first row and second column of A; M13 = is the matrix obtained by deleting the
a31 a32
first row and third column of A. Let A11 = det M11 , A12 = − det M12 and A13 = det M13 .
Then equation (2) above becomes
det A = a11 A11 + a12 A12 + a13 A13 .
Definition (Minor ): Let A be an n × n matrix and let Mij be the (n − 1) × (n − 1) matrix
obtained from A by deleting the ith row and j th column of A. Then Mij is called the ij th
minor of A.
Examples:
 
2 −1 4
1. Let A =  0 1 5 . Find M13 and M32 .
6 3 −4
Solution:  
0 1
Deleting the first row and third column of A, we obtain M13 = , and deleting
6 3
 
2 4
the third row and second column of A, we obtain M32 = .
0 5
 
1 −3 5 6
 2 4 0 3 
2. Let A =  
 1 5 9 2 . Find M24 and M32 .
4 0 2 7
Solution:  
1 −3 5
Deleting the second row and fourth column of A, we obtain M24 =  1 5 9 , and
4 0 2
 
1 5 6
deleting the third row and second column of A, we obtain M32 =  2 0 3 .
4 2 7
Definition (Cofactor ): Let A be an n × n matrix. The ij th cofactor of A, denoted by Aij
is given by
Aij = (−1)i+j |Mij | .
i.e. the ij th cofactor of A is obtained by taking the determinant of the ij th minor and
multiplying it by (−1)i+j .
Note: (
1 if i + j is even
(−1)i+j = .
−1 if i + j is odd

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

Examples:
1. In Example 1 above we have
A13 = (−1)1+3 |M24 |
0 1
= = −6
6 3
and
A32 = (−1)3+2 |M32 |
2 4
=− = −10.
0 5
2. In Example 2 above we have
A24 = (−1)2+4 |M24 |
1 −3 5
= 1 5 9 = −192
4 0 2
and
A32 = (−1)3+2 |M32 |
1 5 6
=− 2 0 3 = −8.
4 2 7
Definition (n × n Determinant):  
a11 a12 · · · a1n
 a21 a22 · · · a2n 
 
Consider the general n × n matrix A =  .. .. .. .. . Then the determinant of A
 . . . . 
an1 an2 · · · ann
is given by
det A = a11 A11 + a12 A12 + · · · + a1n A1n
n
X
= a1k A1k . . . . . . . . (4)
k=1

The expression on the right-hand-side of equation (4) is called an expansion of det A by

cofactors.
Remark : In equation (4) we defined the determinant by expanding by cofactors in the first
row of A. We obtain the same result if we expand by cofactors in any other row or column.
Examples:
 
1 3 5 2
 0 −1 3 4 
1. Calculate |A| if A = 
 2 1 9 6 .


3 2 4 8

10

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

Solution:

|A| = a11 A11 + a12 A12 + a13 A13 + a14 A14

−1 3 4 0 3 4 0 −1 4 0 −1 3
=1 1 9 6 −3 2 9 6 +5 2 1 6 −2 2 1 9
2 4 8 3 4 8 3 2 8 3 2 4
= 1 [−1 (48) − 3 (−4) + 4 (−14)] − 3 [0 − 3 (−2) + 4 (−19)]
+ 5 [0 + 1 (−2) + 4 (1)] − 2 [0 + 1 (−19) + 3 (1)]
= 1 (−92) − 3 (−70) + 5 (2) − 2 (−16)
= 160.
 
3 3 0 5
 2 2 0 −2 
2. Find the determinant of A =  
 4 1 −3 0 .
2 10 3 2
Solution:

3 3 5 3 3 5
det A = 0 + 0 − 3 2 2 −2 − 3 2 2 −2
2 10 2 4 1 0
= −3 [3 (24) − 3 (8) + 5 (16)] − 3 [5 (−6) − (−2) (−9) + 0]
= −3 (128) − 3 (−48)
= −240.

Exercise:
Evaluate each of the following determinants:
1 3 1 1
2 5 2 2
(a) ; Ans: 1
1 3 8 9
1 3 2 2
2 −2 −9 −3
5 4 −1 0
(b) ; Ans: 383
2 3 2 8
−3 1 6 0
2 3 9 3
5 4 −1 4
(c) ; Ans: −268.
1 1 6 3
2 3 2 8
Definition (Triangular Matrix ): A square matrix is called upper triangular if all its
components below the diagonal are zero. It is lower triangular if all its components above

11

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

the diagonal are zero. A matrix is called diagonal if all its elements not on the diagonal are
zero. In other words, A = [aij ] is upper triangular if aij = 0 for i > j, lower triangular if
aij = 0 for i < j, and diagonal if aij = 0 for i 6= j.
Note: A diagonal matrix is both upper and lower triangular.
Examples:
 
  −2 3 0 1
2 1 7  0 0 2 4 
1. The matrices A = 0 2 −5  and B = 
 
 0 0 1 3  are upper triangular;
0 0 1
0 0 0 −2
 
5 0 0  
  0 0
C= 2 3 0 and D = are lower triangular; while the identity matrix
1 0
−1 2 4
 
2 0 0
I, and E =  0 −7 0  are diagonal.
0 0 −4
 
a11 0 0 0
 a21 a22 0 0 
2. Compute the determinant of the lower triangular matrix A =   a31 a32 a33 0 .


a41 a42 a43 a44

Solution:

det A = a11 A11 + 0A12 + 0A13 + 0A14

= a11 A11
a22 0 0
= a11 a32 a33 0
a42 a43 a44
a33 0
= a11 a22
a43 a44
= a11 a22 a33 a44 .

Theorem: Let A = [aij ] be an upper or lower triangular n × n matrix. Then

det A = a11 a22 a33 · · · ann

i.e., the determinant of a triangular matrix equals the product of its diagonal entries.

Examples:
The determinants of the six matrices in Example 1 above are
|A| = 2 (2) (1) = 4,

12

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

|B| = (−2) (0) (1) (−2) = 0,

|C| = 5 (3) (4) = 60,
|D| = 0 (0) = 0,
|I| = 1 (1) (1) · · · (1) = 1,
|E| = 2 (−7) (−4) = 56.

Properties of Determinants
Theorem: Let A and B be n×n matrices. Then det AB = det A det B, i.e. the determinant
of the product is the product of the determinants.
Example:
   
1 −1 2 1 −2 3
Let A =  3 1 4  and B =  0 −1 4 . Then det A = 16 and det B = −8.
0 −2 5 2 0 −2
    
1 −1 2 1 −2 3 5 −1 −5
Now, AB =  3 1 4   0 −1 4  =  11 −7 5  and
0 −2 5 2 0 −2 10 2 −18
det AB = −128 = 16 (−8) = det A det B.
Note: The determinant of the sum is not always equal to the sum of the determinants, i.e. if
A and B are n × n matrices, then det (A + B) 6= det A + det B in general. So, the determinat
function is not linear.
Example:
     
1 2 3 0 4 2
Let A = and B = . Then A + B = . In this case det A = −2,
3 4 −2 2 1 6
det B = 6, and det (A + B) = 22 6= 4 = det A + det B.
Theorem: Let A be an n × n matrix. Then det AT = det A.
Example:
   
1 −1 2 1 3 0
Let A =  3 1 4 . Then AT =  −1 1 −2 . By direct calculation, it can be verified
0 −2 5 2 4 5
T
that det A = det A = 16.

Property 1 :
If any row or column of A is a zero vector, then det A = 0.
Proof :
Suppose the ith row of A contains all zeros, i.e., aij = 0 for j = 1, 2, · · · , n. Then expanding

13

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

det A by cofactors in this row, we obtain

det A = ai1 Ai1 + ai2 Ai2 + · · · + ain Ain

= 0Ai1 + 0Ai2 + · · · + 0Ain
= 0+0+···+0
= 0.

A similar proof works if the j th column is the zero vector.

Example:
1 −3 0 1
4 2 0 5
It is easy to verify that = 0.
−1 6 0 4
2 1 0 1

Property 2 :
If the ith row or j th column of A is multiplied by a scalar c, then det A is multiplied by c.
Proof :    
a11 a12 · · · a1n a11 a12 ··· a1n
 a21 a22 · · · a2n   a21 a22 ··· a2n 
 . .. .. ..   .. .. .. .. 
 .   
 . . . .   . . . . 
Let A =  . If B =  , then expanding in the
 ai1 ai2 · · · ain   cai1 cai2 · · · cain 
 . .. .. ..   . .. .. .. 
 .. . . .   .. . . . 
an1 an2 · · · ann an1 an2 · · · ann
ith row of B, we have

det B = cai1 Ai1 + cai2 Ai2 + · · · + cain Ain

= c (ai1 Ai1 + ai2 Ai2 + · · · + ain Ain )
= c det A.

A similar proof works for columns.

Example:
 
1 −1 2
Let A =  3 1 4 . Then, det A = 16. Multiplying the second column by −2, we obtain
0 −2 5
 
1 2 2
B =  3 −2 4  and det B = −32 = −2 det A.
0 4 5

Remark : For any scalar α and n × n matrix A, det (αA) = αn det A.

14

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

Property 3 :
   
a11 a12 · · · a1j · · · a1n a11 a12 · · · b1j · · · a1n
 a21 a22 · · · a2j · · · a2n   a21 a22 · · · b2j · · · a2n 
   
Let A =  .. .. .. .. .. .. , B =  .. .. .. .. .. .. , and
 . . . . . .   . . . . . . 
an1 an2 · · · anj · · · ann an1 an2 · · · bnj · · · ann
 
a11 a12 · · · a1j + b1j · · · a1n
 a21 a22 · · · a2j + b2j · · · a2n 
 
C =  .. .. .. .. .. .. . Then det C = det A + det B. In other words,
 . . . . . . 
an1 an2 · · · anj + bnj · · · ann
suppose A, B, and C are identical except for the j th column and that the j th column of C is
the sum of the j th columns of A and B. Then det C = det A + det B. The same statement
is true for rows.
Proof :
Expand det C in the j th column to obtain

det C = (a1j + b1j ) A1j + (a2j + b2j ) A2j + · · · + (anj + bnj ) Anj
= (a1j A1j + a2j A2j + · · · + anj Anj ) + (b1j A1j + b2j A2j + · · · + bnj Anj )
= det A + det B.

Example:
       
1 −1 2 1 −6 2 1 (−1 − 6) 2 1 −7 2
Let A =  3 1 4 , B =  3 2 4 , and C =  3 (1 + 2) 4  =  3 3 4 .
0 −2 5 0 4 5 0 (−2 + 4) 5 0 2 5
Then, det A = 16, det B = 108, and det C = 124 = det A + det B.

Property 4 :
Interchanging any two rows (or columns) of A has the effect of multiplying det A by −1.
Proof :
We prove the statement for rows and assume first that two adjacent rows are interchanged,
i.e., we assume first that the ith and (i + 1)st rows are interchanged.
   
a11 a12 · · · a1n a11 a12 · · · a1n
 a21 a22 · · · a2n   a21 a22 · · · a2n 
 . .. .. ..   . .. .. .. 
 .   . 
 . . . .   . . . . 
   
Let A =  ai1 ai2 · · · ain  and B =  ai+1, 1 ai+1, 2 · · · ai+1, n .
   
 ai+1,1 ai+1,2 · · · ai+1,n   ai1 ai2 · · · ain 
 . .. .. ..   . .. .. .. 
 .. . . .   .. . . . 
an1 an2 · · · ann an1 an2 · · · ann
Then expanding det A in its ith row and det B in its (i + 1)st row, we obtain
det A = ai1 Ai1 + ai2 Ai2 + · · · + ain Ain
and

15

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

det B = ai1 Bi+1, 1 + ai2 Bi+1, 2 + · · · + ain Bi+1, n .

i+j
Here Aij = (−1) |Mij | where Mij is obtained by crossing off the ith row and j th column of
A. If we cross off the (i + 1)st row and j th column of B, we obtain the same Mij . Thus,

Bi+1, j = (−1)i+j+1 |Mij |

= (−1) (−1)i+j |Mij |
= −Aij

so that, from the expansions of det A and det B above, det B = − det A.
Now, suppose that i < j and that the ith and j th rows are to be interchanged. We can
do this by interchanging adjacent rows several times. It will take j − i interchanges to
move row j into the ith row. Then row i will be in the (i + 1)st row and it will take an
additional j − i − 1 interchanges to move row i into the j th row. Finally, the total number
of interchanges of adjacent rows is (j − i) + (j − i − 1) = 2 (j − i) − 1, which is odd. Thus,
det A is multiplied by −1 an odd number of times, which is what we needed to show, i.e.,
det B = (−1)2(j−i)−1 det A = − det A.
Example:
   
1 −1 2 0 −2 5
Let A =  3 1 4 . By interchanging the first and third rows, we obtain B =  3 1 4 .
0 −2 5 1 −1 2
 
−1 1 2
By interchanging the first and and second columns of A, we obtain C =  1 3 4 . By
−2 0 5
direct calculation we find that det A = 16, and det B = det C = −16.

Property 5 :
If A has two equal rows (or columns), then det A = 0.
Proof :
Suppose the ith and j th rows of A are equal. By interchanging these rows we obtain a matrix
B with the property that det B = − det A (from Property 4 ). But since row i = row j,
interchanging them gives the same matrix. Thus, A = B and det A = det B = − det A. So,
2 det A = 0 which is the case only if det A = 0.
Example:
   
1 −1 2 5 2 2
The matrix A =  5 7 3  has two equal rows while matrix B =  3 −1 −1  has
1 −1 2 −2 4 4
two equal columns. By direct calculation we can verify that det A = det B = 0.

Property 6 :
If one row (column) of A is a scalar multiple of another row (column), then det A = 0.

16

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

Proof :
Let (aj1 , aj2 , · · · , ajn ) = c (ai1 , ai2 , · · · , ain ). Then from Property 2,

a11 a12 · · · a1n

a21 a22 · · · a2n
.. .. .. ..
. . . .
ai1 ai2 · · · ain
det A = c .. .. .. .. = 0,
. . . .
ai1 ai2 · · · ain
.. .. .. ..
. . . .
an1 an2 · · · ann

from Property 5.
Example:
2 −3 5
It can be verified that 1 7 2 = 0.
−4 6 −10
From Properties 5 and 6 above we have the following theorem:
Theorem: Let A be an n × n matrix. Then det A = 0 if and only if the rows or columns of
A are linearly independent.

Property 7 :
If a scalar multiple of one row (column) of A is added to another row (column), then the
determinant of A is unchanged.

17

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

Proof :
Let B be the matrix obtained by adding c times the ith row of A to the j th row of A. Then

a11 a12 ··· a1n

a21 a22 ··· a2n
.. .. .. ..
. . . .
ai1 ai2 ··· ain
det B = .. .. .. ..
. . . .
aj1 + cai1 aj2 + cai2 · · · ajn + cain
.. .. .. ..
. . . .
an1 an2 ··· ann

a11 a12 · · · a1n a11 a12 ··· a1n

a21 a22 · · · a2n a21 a22 ··· a2n
.. .. .. .. .. .. .. ..
. . . . . . . .
ai1 ai2 · · · ain ai1 ai2 ··· ain
= .. .. .. .. + .. .. .. ..
. . . . . . . .
aj1 aj2 · · · ajn cai1 cai2 · · · cain
.. .. .. .. .. .. .. ..
. . . . . . . .
an1 an2 · · · ann an1 an2 · · · ann
= det A + 0
= det A.

Example:
 
1 −1 2
Let A =  3 1 4 . Multiply the first row by −3 and adding the result to the second row
0 −2 5  
1 −1 2
to obtain a new matrix B =  0 4 −2 . Now, multiply the second row of B by 12 and
0 −2 5
 
1 −1 2
add the result to the third row of B to obtain C =  0 4 −2 . By direct calculation it
0 0 4
can be shown that det A = det B = det C = 16.

Examples Using Properties of Determinants

1 3 5 2
0 −1 3 4
1. Calculate |A| = .
2 1 9 6
3 2 4 8
Solution: We do row reduction aiming for a triangular matrix

18

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

1 3 5 2 1 3 5 2
−2R1 + R3
0 −1 3 4 0 −1 3 4
|A| = −→ =
2 1 9 6 0 −5 −1 2
−3R1 + R4
3 2 4 8 0 −7 −11 2

1 3 5 2 1 3 5 2
5R2 − R3
0 −1 3 4 R3 − 12 R4 0 −1 3 4
−→ = (−1) (−1) = (−2)
0 0 16 18 −→ 0 0 16 18
7R2 − R4
0 0 32 26 0 0 0 5
= (−2) (1) (−1) (16) (5) = 160
(since the final determinant is that of an upper triangular matrix).
−2 1 0 4
3 −1 5 2
2. Calculate |A| = .
−2 7 3 1
3 −7 2 5
Solution:
−2 1 0 4 3R1 + 2R2 −2 1 0 4
3 −1 5 2 R1 − R3 0 1 10 16
|A| = = 12 (−1) (−1) = 1
2
(0) = 0.
−2 7 3 1 −→ 0 −6 −3 3
3 −7 2 5 R2 − R4 0 6 3 −3

Alternatively,
−2 1 0 4 −2 0 0 0
C1 + 2C2
3 −1 5 2 1 3 1 5 8
|A| = −→ = 2
−2 7 3 1 −2 12 3 −3
2C1 + C4
3 −7 2 5 3 −11 2 11
−2 0 0 0
5C2 − C3
3 1 0 0
−→ = 21 (−1) = − 21 (0) = 0.
−2 12 57 −99
−8C2 + C4
3 −11 −57 99

1 −2 3 −5 7
2 0 −1 −5 6
3. Calculate |A| = 4 7 3 −9 4 .
3 1 −2 −2 3
−5 −1 3 7 −9
Solution:
1 −2 3 −5 7 1 −2 3 −5 7
2 0 −1 −5 6 2 0 −1 −5 6
R2 + R5
|A| = 4 7 3 −9 4 = 4 7 3 −9 4 = 0,
−→
3 1 −2 −2 3 3 1 −2 −2 3
−5 −1 3 7 −9 −3 −1 2 2 −3

19

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

by Property 6 (since row 5 = −row 4).

Exercise:
1. Use properties of determinants or otherwise to evaluate:
2 −1 0 4 1 a 0 0 0 0
3 1 −1 2 0 0 0 b 0 0
(a) 3 2 −2 5 1 ; Ans: 138 (b) 0 0 0 0 c ; Ans: abcde
0 0 4 −1 6 0 0 0 d 0
3 2 1 −1 1 0 e 0 0 0
1 −1 2 0 0 2 5 −6 8 0
3 1 4 0 0 0 1 −7 6 0
(c) 2 −1 5 0 0 (d ) 0 0 0 4 0
0 0 0 2 3 0 2 1 5 1
0 0 0 −1 4 4 −1 5 3 0
a11 a12 a13
2. Suppose a21 a22 a23 = 8. Compute:
a31 a32 a33

a31 a32 a33

(a) a21 a22 a23 ; Ans: −8
a11 a12 a13
a11 a12 a13
(b) 2a21 2a22 2a23 ; Ans: 16
a31 a32 a33
a11 −a13 a12
(c) a21 −a23 a22 ; Ans: 8
a31 −a33 a32

MATRIX INVERSION
For an n × n matrix A, the n × n matrix B such that AB = BA = In is called the inverse
of A and is denoted by B = A−1 .
Note: Matrix inversion is defined for square matrices only; the condition AA−1 = A−1 A
rules out inversion of nonsquare matrices.
Not all square matrices are invertible. An invertible matrix is said to be nonsingular while
a square matrix with no inverse is called a singular matrix.
Theorem: Let A be an n × n matrix. Then A is invertible if and only if det A 6= 0.
Theorem: If A is invertible, then det A 6= 0 and det A−1 = det1 A .
Proof : Suppose A is invertible. Then from the preceeding theorem, det A 6= 0. Applying a
previous theorem,
1 = det I = det AA−1 = det A det A−1

20

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

so that
1
det A−1 = .
det A

Computing an Inverse Using Determinant and Adjoint

 
A11 A12 · · · A1n
 A21 A22 · · · A2n 
 
Definition (Adjoint): Let A be an n × n matrix and let B =  .. .. .. .. 
 . . . . 
An1 An2 · · · Ann
be the matrix of its cofactors. Then the adjoint (adjugate) of A, denoted adj A, is the
 
A11 A21 · · · An1
 A12 A22 · · · An2 
T  
transpose of the n × n matrix B, i.e., adjA = B =  .. .. .. .. .
 . . . . 
A1n A2n · · · Ann
Theorem: Let A be an n × n matrix. Then
 
det A 0 0 ··· 0
 0 det A 0 ··· 0 
 
 0 0 det A · · · 0 
A (adjA) =   = (det A) I
 .. .. .. . . .. 
 . . . . . 
0 0 0 · · · det A

Proof :
Let C = [cij ] = A (adjA). Then
  
a11 a12 · · · a1n A11 A21 · · · An1
 a21 a22 · · · a2n   A12 A22 · · · An2 
  
C =  .. .. .. ..   .. .. .. ..  . . . . . . . . (5)
 . . . .  . . . . 
an1 an2 · · · ann A1n A2n · · · Ann

We have cij = (ith row of A)·(j th column of adjA), i.e.,

 
Aj1
 Aj2 
 
cij = (ai1 ai2 · · · ain )  .. 
 . 
Ajn
= ai1 Aj1 + ai2 Aj2 + · · · + ain Ajn . . . . . . . . (6)

Now, if i = j, the sum in (6) equals

ai1 Ai1 + ai2 Ai2 + · · · + ain Ain

21

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

which is the expansion of det A in the ith row of A. On the other hand, if i 6= j, then the
sum in (6) equals zero. Thus,
(
det A if i = j
cij = .
0 if i 6= j

This proves the theorem.

Theorem: Let A be an n × n matrix. Then A is invertible if and only if det A 6= 0. If
det A 6= 0, then A−1 = det1 A (adjA).
Proof :
The first part of the theorem is just a previous theorem. Now, suppose det A 6= 0. We show
that det1 A (adjA) is the inverse of A by multiplying it by A and getting the identity matrix:
 
1 1
A (adjA) = [A (adjA)]
det A det A
1
= [(det A) I]
det A
= I.

Examples:
Determine the inverse of each of the following matrices:
 
−1 2
(a) A =
2 −4
 
3 1
(b) B =
2 4
 
2 4 3
(c) C =  0 1 −1 
3 5 7
 
3 1 0
(d ) D =  −2 −4 3 
5 4 −2
Solution:
(a) det A = (−1) (−4) − (2) (2) = 0. So, A−1 does not exist.
(b) det B = 3 (4) − (2) (1) = 10 6= 0. So, B −1 exists. In this case,
   
1 4 −1 2/5 −1/10
B =−1
= −1 .
10 −2 3 /5 3/10

(c) Since det C = 2 (12) − 0 + 3 (−7) = 3 6= 0, then C is invertible. The matrix of

cofactors of C is
   
C11 C12 C13 12 −3 −3
 C21 C22 C23  =  −13 5 2 
C31 C32 C33 −7 2 2

22

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

and    
C11 C21 C31 12 −13 −7
adjC =  C12 C22 C32  =  −3 5 2 .
C13 C23 C33 −3 2 2
So,    
12 −13 −7 4 −13/3 −7/3
1   .
C −1 = −3 5 2 = −1 5/3 2/3
3
−3 2 2 −1 2/3 2/3

(d ) det D = 3 (−4) − 1 (−11) = −1 6= 0. So, D −1 exists. The matrix of cofactors of D

   
−4 11 12 −4 2 3
is  2 −6 −7  and adjD =  11 −6 −9 . So,
3 −9 −10 12 −7 −10
   
−4 2 3 4 −2 −3
1 
D −1 = 11 −6 −9  =  −11 6 9 .
−1
12 −7 −10 −12 7 10

Computing an Inverse Using Gauss-Jordan Elimination

Gauss-Jordan elimination can also be used to find the inverse of a nonsingular matrix A by
the reduction
Gauss − Jordan  
[A | I] I | A−1 .
−→
Note: For a singular matrix A cannot be reduced to I by Gauss-Jordan elimination because
a zero row will be obtained on the left-hand side of the augmented array.
Examples:  
−1 2 −3
1. Find the inverse of A =  0 1 −1  if it exists.
3 −1 7
Solution:
In this case,
   
−1 2 −3 | 1 0 0 −R1 1 −2 3 | −1 0 0
[A | I] =  0 1 −1 | 0 1 0  −→  0 1 −1 | 0 1 0 
3 −1 7 | 0 0 1 3R1 + R3 0 5 −2 | 3 0 1
   
1 −2 3 | −1 0 0 1 1 −2 3 | −1 0 0
5R1 − R3  − 3 R3 
0 1 −1 | 0 1 0  0 1 −1 | 0 1 0 
−→ −→
0 0 −3 | −3 5 −1 0 0 1 | 1 − /3 /3
5 1

 
R2 + R3 1 0 0 | −2 11/3 −1/3
−→  0 1 0 | 1 −2/3 1/3  = [I | A−1 ].
R1 + 2R2 − R3 0 0 1 | 1 −5/3 1/3
 
−2 11/3 −1/3
Since A has been reduced to I, A is nonsingular, and A−1 =  1 −2/3 1/3 .
1 −5/3 1/3

23

Downloaded by Steve Lewis ([email protected])

 lOMoARcPSD|51865458

    
−1 2 −3 −2 11/3 −1/3 1 0 0
As a check, AA−1 =  0 1 −1   1 −2/3 1/3  =  0 1 0  = I.
3 −1 7 1 −5/3 1/3 0 0 1
 
−1 2 −1
2. Use Gauss-Jordan elimination to show that the inverse of A =  1 −1 1  does
2 −4 2
not exist.
Solution:
  −R1  
−1 2 −1 | 1 0 0 1 −2 1 | −1 0 0
R1 + R2 
[A | I] =  1 −1 1 | 0 1 0  0 1 0 | 1 1 0 
−→
2 −4 2 | 0 0 1 0 0 0 | 2 0 1
2R1 + R3
 
1 0 1 | 1 2 0
R1 + 2R2  0 1 0 | 1 1 0 .
−→
0 0 0 | 2 0 1
The process has not reduced A to I. So, A−1 does not exist.

Properties of Matrix Inversion

Let A and B be nonsingular n × n matrices. Then:
(i ) (A−1 ) = A.
−1

(ii ) The product AB is also nonsingular.

(iii ) (AB)−1 = B −1 A−1 .
T
−1
(iv ) (A−1 ) = AT .
Exercise:  
1 0 1
1. Let A =  0 1 1 . Determine whether A is invertible and calculate A−1 if it is.
1 1 0
2. Prove that if A and B are invertible n × n matrices, then (AB)−1 = B −1 A−1 .

24

Downloaded by Steve Lewis ([email protected])