LN1-Matrix Algebra

advertisement
MATRIX ALGEBRA
1. WHAT IS A MATRIX?
A matrix is defined as rectangular array of numbers, parameters, or variables. Matrix algebra provides a method to
solve a system of simultaneous equations:
1. It provides a compact way of writing an equation system.
2. It provides a procedure to test for the existence of a solution.
3. If the solution exists, it provides a method of finding that solution.
General format of a system of ๐‘š equations with ๐‘› variables:
๐‘Ž11 ๐‘ฅ1 + ๐‘Ž12 ๐‘ฅ2 + โ‹ฏ + ๐‘Ž1๐‘› ๐‘ฅ๐‘› = ๐‘‘1
๐‘Ž21 ๐‘ฅ1 + ๐‘Ž22 ๐‘ฅ2 + โ‹ฏ + ๐‘Ž2๐‘› ๐‘ฅ๐‘› = ๐‘‘2
โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™โˆ™
๐‘Ž๐‘š1 ๐‘ฅ1 + ๐‘Ž๐‘š2 ๐‘ฅ2 + โ‹ฏ + ๐‘Ž๐‘š๐‘› ๐‘ฅ๐‘› = ๐‘‘๐‘š
The matrix format of writing a system of equations is as follows.
๐‘Ž11
๐‘Ž21
๐ด=[ โ‹ฎ
๐‘Ž๐‘š1
๐‘Ž12
๐‘Ž22
โ‹ฎ
๐‘Ž๐‘š2
โ‹ฏ
โ‹ฏ
โ‹ฎ
โ‹ฏ
๐‘Ž1๐‘›
๐‘Ž2๐‘›
โ‹ฏ ]
๐‘Ž๐‘š๐‘›
๐‘ฅ1
๐‘ฅ2
๐‘ฅ=[ โ‹ฎ]
๐‘ฅ๐‘›
๐‘‘1
๐‘‘
๐‘‘ = [ 2]
โ‹ฎ
๐‘‘๐‘š
The shorthand way of representing matrix ๐ด is: ๐ด = [๐‘Ž๐‘–๐‘— ]
Example 1:
6๐‘ฅ1 + 3๐‘ฅ2 + 1๐‘ฅ3 = 22
6๐‘ฅ1 + 4๐‘ฅ2 − 2๐‘ฅ3 = 12
4๐‘ฅ1 − 3๐‘ฅ2 + 5๐‘ฅ3 = 10
In matrix format the equation is presented as:
6
๐ด = [1
4
3
4
−1
1
−2]
5
๐‘ฅ1
๐‘ฅ = [ ๐‘ฅ2 ]
๐‘ฅ๐‘›
22
๐‘‘ = [12]
10
The number of rows and number of columns of a matrix define the dimensions of the matrix. The dimension is shown
as ๐‘š × ๐‘›. A matrix with 4 rows and 3 columns is a "4 by 3" matrix: 4 × 3.
A matrix with m rows and 1 column is called a column vector: ๐‘š × 1. Similarly, a matrix with 1 row and n columns is
called a row vector: 1 × ๐‘›.
A system of simultaneous equations can be represented as: ๐‘จ๐’™ = ๐’…
The shorthand representation of the system of equations implies that the left-hand side of the equation is obtained as
the product of the coefficient matrix ๐ด times the variable matrix ๐‘ฅ. The product of these two matrices produces a new
matrix. This new matrix, in turn, is equal to the constant matrix ๐‘‘.
To explain the product of two matrices and what makes two matrices equal we need to understand matrix operations.
LN1—Matrix Algebra
Page 1 of 12
2. MATRIX OPERATIONS
2.1. Equality of Matrices
Two matrices ๐ด = [๐‘Ž๐‘–๐‘— ] and ๐ต = [๐‘๐‘–๐‘— ] are said to be equal if and only if ๐‘Ž๐‘–๐‘— = ๐‘๐‘–๐‘— . For example,
๐ด=[
4
2
3
]
0
๐ต=[
4
2
3
]
0
๐ด=๐ต
2.2. Addition and Subtraction of Matrices
Two matrices can be added (subtracted) if and only if they are conformable for addition (subtraction). That is, they
have the same dimensions.
๐‘Ž11
๐‘Ž21
๐ด=[ โ‹ฎ
๐‘Ž๐‘š1
๐‘Ž12
๐‘Ž22
โ‹ฎ
๐‘Ž๐‘š2
๐‘Ž11 ± ๐‘11
๐‘Ž21 ± ๐‘21
๐ด=[
โ‹ฎ
๐‘Ž๐‘š1 ± ๐‘๐‘š1
โ‹ฏ
โ‹ฏ
โ‹ฎ
โ‹ฏ
๐‘Ž1๐‘›
๐‘Ž2๐‘›
โ‹ฏ ]
๐‘Ž๐‘š๐‘›
๐‘Ž12 ± ๐‘12
๐‘Ž22 ± ๐‘22
โ‹ฎ
๐‘Ž๐‘š2 ± ๐‘๐‘š2
๐‘11
๐‘21
๐ต=[
โ‹ฎ
๐‘๐‘š1
โ‹ฏ
โ‹ฏ
โ‹ฎ
โ‹ฏ
๐‘12
๐‘22
โ‹ฎ
๐‘๐‘š2
โ‹ฏ
โ‹ฏ
โ‹ฎ
โ‹ฏ
๐‘1๐‘›
๐‘2๐‘›
]
โ‹ฏ
๐‘๐‘š๐‘›
๐‘Ž1๐‘› ± ๐‘1๐‘›
๐‘Ž2๐‘› ± ๐‘2๐‘›
]
โ‹ฏ
๐‘Ž๐‘š๐‘› ± ๐‘๐‘š๐‘›
Example 2
4
๐ด = [2
6
8
3
1 −2]
5
7
6
๐ต=[ 3
−6
10
2
7
10
๐ด+๐ต =[ 5
0
5
4]
8
18
3
18
8
2]
15
(See the Excel file E375 CH4 Outline for the numeric example)
2.2.1. Scalar Multiplication
a scalar is a number. The scalar multiplication of a matrix involves multiplying each element of that matrix by that
number.
Example 3
๐‘ =2
4
๐ด = [2
6
8
1
5
3
−2]
7
8
๐‘ ๐ด = [ 4
12
16
2
10
6
−4]
14
2.2.2. Multiplication of Matrices
To multiply two matrices ๐ด and ๐ต, the matrices must by conformable for multiplication. In multiplying matrix ๐ด by
matrix ๐ต, ๐ด is the lead matrix and ๐ต is the ๐‘™๐‘Ž๐‘” ๐‘š๐‘Ž๐‘ก๐‘Ÿ๐‘–๐‘ฅ. ๐ด and ๐ต are conformable for multiplication if the number of
columns (column dimension) of the lead matrix ๐ด is equal to the number of rows (row dimension) of the lag matrix ๐ต.
Consider the matrix A with dimensions (๐‘š × ๐‘›) and matrix B with dimensions (๐‘› × ๐‘). A and ๐ต are conformable for
multiplication because the column dimension of ๐ด is ๐‘› and row dimension of ๐ต is also ๐‘›. The product matrix is ๐ด๐ต =
๐ถ has dimensions (๐‘š × ๐‘).
A ๏ƒ— B ๏€ฝ C
( m๏‚ดn ) ( n๏‚ด p )
( m๏‚ด p )
Note that if ๐ต were the lead matrix and ๐ด the lag matrix, then ๐ต and ๐ด would not be conformable for multiplication.
Thus, the commutative property of multiplication of numbers does not apply to matrices.
LN1—Matrix Algebra
Page 2 of 12
2.2.2.1.
How to multiply two matrices
Take two matrices A and B , which are conformable for multiplication. The product matrix is, then
( 3๏‚ด 2 )
( 2๏‚ด 2 )
A๏ƒ— B ๏€ฝ C
(3๏‚ด2) ( 2๏‚ด3)
(3๏‚ด3)
๐‘Ž11
๐‘Ž
๐ด = [ 21
๐‘Ž31
๐‘Ž12
๐‘Ž22 ]
๐‘Ž32
๐ต=[
๐‘11
๐‘21
๐‘12
๐‘22
๐‘13
]
๐‘23
To determine C you must perform the following operation:
๐‘Ž11 ๐‘11 + ๐‘Ž12 ๐‘21
๐ด๐ต = ๐ถ = [๐‘Ž21 ๐‘11 + ๐‘Ž22 ๐‘21
๐‘Ž31 ๐‘11 + ๐‘Ž32 ๐‘21
๐‘Ž11 ๐‘12 + ๐‘Ž12 ๐‘22
๐‘Ž21 ๐‘12 + ๐‘Ž12 ๐‘22
๐‘Ž31 ๐‘12 + ๐‘Ž12 ๐‘22
Generally, we can express two matrices
๐‘Ž11 ๐‘13 + ๐‘Ž12 ๐‘23
๐‘Ž21 ๐‘13 + ๐‘Ž22 ๐‘23 ]
๐‘Ž31 ๐‘13 + ๐‘Ž32 ๐‘23
A and B conformable for multiplication as:
( m๏‚ดn )
( n๏‚ด p )
๐ด = [๐‘Ž๐‘–๐‘˜ ] and ๐ต = [๐‘๐‘˜๐‘— ]
where
๐‘– = 1,2, โ‹ฏ ๐‘š
๐‘— = 1,2, โ‹ฏ ๐‘
๐‘˜ = 1,2, โ‹ฏ , ๐‘›
Then the product matrix is
๐‘›
๐ถ = [๐ถ๐‘–๐‘— ] = [∑ ๐‘Ž๐‘–๐‘˜ ๐‘๐‘˜๐‘— ]
๐‘˜=1
In the above specific case,
๐‘– = 1,2,3
๐‘— = 1,2,3
๐‘˜ = 1,2
For example, the element in the second row and third column of the product matrix is:
2
๐ถ23 = ∑ ๐‘Ž2๐‘˜ ๐‘๐‘˜3 = ๐‘Ž21 ๐‘13 + ๐‘Ž22 ๐‘23
๐‘˜=1
Example 4
4
๐ด = [6
2
9
3]
5
๐ต=[
8
4
4×8+9×4
๐ถ = ๐ด๐ต = [6 × 8 + 3 × 4
2×8+5×4
1
7
3
]
6
4×1+9×7
6×1+3×7
2×1+5×7
4×3+9×6
68
6 × 3 + 3 × 6] = [60
2×3+5×6
36
67
27
37
66
36]
36
In Excel to find the product ๐ด๐ต use the =๐‘€๐‘€๐‘ˆ๐ฟ๐‘‡ command. =๐‘€๐‘€๐‘ˆ๐ฟ๐‘‡ is an “array” formula, requiring the following
steps:
1.
2.
3.
Highlight a 3-by-3 block of cells.
Call up =MMULT(array1, array2). “array 1” is the block of cells containing the elements of ๐ด and “array2” is
the block containing elements of ๐ต.
Hold Ctrl and Shift keys and press Enter.
LN1—Matrix Algebra
Page 3 of 12
Example 5
0
6
๐ด=[
9
5
1
1
3
2
3
7
]
8
4
7
๐ต = [5
7
4
8]
6
A๏ƒ— B ๏€ฝ C
( 4๏‚ด3) (3๏‚ด2)
( 4๏‚ด2)
26
96
๐ถ=[
134
73
26
74
]
108
60
2.3. Back to the simultaneous equation system
Recall the system of equations,
6๐‘ฅ1 + 3๐‘ฅ2 + 1๐‘ฅ3 = 22
6๐‘ฅ1 + 4๐‘ฅ2 − 2๐‘ฅ3 = 12
4๐‘ฅ1 − 3๐‘ฅ2 + 5๐‘ฅ3 = 10
In matrix format the equation is presented as:
6
3
A = [1
4
(3๏‚ด3)
4 −1
๐‘ฅ1
๐‘ฅ
x = [ 2]
( 3๏‚ด1)
๐‘ฅ๐‘›
1
−2]
5
22
d = [12]
( 3๏‚ด1)
10
Using the matrix multiplication procedure, we have
6
A x = [1
(3๏‚ด3) (3๏‚ด1)
4
3
4
−1
6๐‘ฅ1 + 3๐‘ฅ2 + 2๐‘ฅ3
1 ๐‘ฅ1
−2] [ ๐‘ฅ2 ] = [6๐‘ฅ1 + 4๐‘ฅ2 − 2๐‘ฅ3 ]
4๐‘ฅ1 − 4๐‘ฅ2 + 5๐‘ฅ3
5 ๐‘ฅ๐‘›
The product matrix ๐ด๐‘ฅ has the dimensions (3 × 1) and is equal to the constant matrix d .
( 3๏‚ด1)
6๐‘ฅ1 + 3๐‘ฅ2 + 2๐‘ฅ3
22
[6๐‘ฅ1 + 4๐‘ฅ2 − 2๐‘ฅ3 ] = [12]
4๐‘ฅ1 − 4๐‘ฅ2 + 5๐‘ฅ3
10
3. IDENTITY MATRICES
An identity matrix is a square matrix with 1s in its principal diagonal and 0s everywhere else.
๐ผ2 = [
1
0
0
]
1
1
๐ผ3 = [0
0
0
1
0
0
0]
1
The identity matrix plays a role similar to 1 in scalar algebra (regular number system). You can pre or post multiply a
matrix ๐ด by an identity matrix and obtain the same result—the original matrix A.
๐ผ๐ด = ๐ด๐ผ = ๐ด
Example 6
LN1—Matrix Algebra
Page 4 of 12
8
A =[
9
๏€จ2๏‚ด3๏€ฉ
3
5
1
I =[
0
0
]
1
๏€จ2๏‚ด2 ๏€ฉ
0
]
7
1 0
I = [0 1
๏€จ3๏‚ด3๏€ฉ
0 0
8
A =[
9
0
0]
1
3
5
๏€จ2๏‚ด3๏€ฉ
8
AI = [
9
3
5
0
]
7
8
IA = [
9
3
5
0
]
7
๏€จ2๏‚ด3๏€ฉ
0
]
7
๏€จ2๏‚ด3๏€ฉ
4. TRANSPOSE OF A MATRIX
The transpose of matrix ๐ด, denoted by ๐ด′, is obtained by interchanging the rows with columns.
Example 7
The following are several examples of matrix transposes.
๐ด=[
8
9
3
5
0
]
7
8
๐ด′ = [3
0
9
5]
7
7
๐ต = [5
7
4
8]
6
7
๐ต′ = [
4
5
8
7
]
6
6
๐ถ = [9
5
1 7
3 8]
2 4
6
๐ถ ′ = [1
7
9
3
8
5
2]
4
5. INVERSE OF A MATRIX
The inverse of a square matrix ๐ด, denoted by ๐ด−1 , is a matrix that satisfies the following condition:
๐ด๐ด−1 = ๐ด−1 ๐ด = ๐ผ
Example 8
6
๐ด = [9
5
1 7
3 8]
2 4
1 0
๐ด๐ด−1 = [0 1
0 0
0
0]
1
−1
๐ด−1 = [ 4
3
10
−11
−7
1
๐ด−1 ๐ด = [0
0
0
1
0
−13
15]
9
0
0]
1
To find the inverse of ๐ด above I have used the Excel array function =๐‘€๐ผ๐‘๐‘‰๐ธ๐‘…๐‘†๐ธ. Finding the inverse of a matrix, if it
exists, without a computer is an involved process. The process is explained below.
If the inverse of a matrix A exists, then it is called a nonsingular matrix. If the inverse does not exist, then A is called a
singular matrix.
5.1. Properties of Inverse Matrices
1.
2.
3.
(๐ด−1 )−1 = ๐ด
The inverse of an inverse matrix is the original matrix:
The inverse of product of the lead matrix ๐ด and the lag matrix ๐ต is equal to the product of the inverse of the
(๐ด๐ต)−1 = ๐ต −1 ๐ด−1
lead matrix ๐ต −1 and the inverse of lag matrix ๐ด−1 :
(๐ด′)−1 = (๐ด−1 )′
Inverse of the transpose is equal to the transpose of the inverse:
5.2. Inverse Matrix and Solution of Linear Equation System
LN1—Matrix Algebra
Page 5 of 12
It was shown that a linear equation system can be presented in the following matrix format:
x ๏€ฝ d
A
( m๏‚ดm) ( m๏‚ด1)
( m๏‚ด1)
Now pre-multiply both sides by ๐ด−1 :
๐ด−1 ๐ด๐‘ฅ = ๐ด−1 ๐‘‘
Given that ๐ด−1 ๐ด = ๐ผ, then
๐ด−1 ๐ด๐‘ฅ = ๐ผ๐‘ฅ = ๐‘ฅ
Thus,
๐‘ฅ = ๐ด−1 ๐‘‘
The matrix product
A ๏€ญ1 d
( m๏‚ดm) ( m๏‚ด1)
yields an (๐‘š × 1) matrix whose elements are the solutions for the equation system.
Example 9
Using the equation system from Example 1 we have
6
[1
4
3
4
−1
1 ๐‘ฅ1
22
−2] [ ๐‘ฅ2 ] = [12]
5 ๐‘ฅ๐‘›
10
Using the =๐‘€๐ผ๐‘๐‘‰๐ธ๐‘…๐‘†๐ธ command in Excel we can find the ๐ด−1 .
0.3462
๐ด−1 = [−0.2500
−0.3269
−0.3077
0.5000
0.3462
−0.1923
0.2500]
0.4038
Thus, the solution matrix is obtained by finding the product of ๐ด−1 and d on the right-hand-side (using the =๐‘€๐‘€๐‘ˆ๐ฟ๐‘‡
command):
๐‘ฅ1
0.3462
[ ๐‘ฅ2 ] = [−0.2500
๐‘ฅ๐‘›
−0.3269
−0.3077
0.5000
0.3462
−0.1923 22
2
0.2500] [12] = [3]
0.4038 10
1
Finding the inverse matrix is yet to be explained. This is what comes next.
5.3. Finding the Inverse Matrix
Clearly, for a system of linear equations to have a set of unique solutions, the coefficient matrix ๐ด must have an
inverse. In other words, ๐ด must be a nonsingular matrix.
5.3.1. The Requirements for Non-singularity of a Matrix
In order for matrix ๐ด to be nonsingular all of its rows must be linearly independent. This means that none of the rows
can be a linear combination of other rows. Consider, for example the following equation system.
6๐‘ฅ1 + 13๐‘ฅ2 + 1๐‘ฅ3 = 22
6๐‘ฅ1 + 14๐‘ฅ2 − 2๐‘ฅ3 = 12
LN1—Matrix Algebra
Page 6 of 12
8๐‘ฅ1 + 11๐‘ฅ2 − 3๐‘ฅ3 = 10
The coefficient matrix is
6
[1
8
3
4
11
1
−2]
−3
Note that, although not obvious, the third row is a linear combination of the first two rows:
[8
−3] = [6 + 2 × 1
11
3+2×4
1 − 2 × 2]
This matrix does not have an inverse. If you try to find the inverse in Excel, you will receive an error message.
5.3.2. Using the Determinant of a Matrix to Test for Singularity
The determinant of a square Matrix, denoted by |๐ด| is a uniquely defined number or numeric value associated with
that matrix. Let’s start first with a (2 × 2) matrix and show how to find the determinant:
๐‘Ž11
๐ด = [๐‘Ž
21
๐‘Ž12
๐‘Ž22 ]
๐‘Ž
|๐ด| = |๐‘Ž11
๐‘Ž12
๐‘Ž22 | = ๐‘Ž11 ๐‘Ž22 − ๐‘Ž12 ๐‘Ž21
21
Example 10
๐ด=[
10
4
|๐ด| = |
8
]
5
10 8
| = 10 × 5 − 8 × 4 = 18
4 5
The determinant of a (3 × 3) matrix A
๐‘Ž11
๐‘Ž
๐ด = [ 21
๐‘Ž31
๐‘Ž12
๐‘Ž22
๐‘Ž32
๐‘Ž13
๐‘Ž23 ]
๐‘Ž33
is determined as follows:
๐‘Ž11
|๐ด| = |๐‘Ž21
๐‘Ž31
๐‘Ž12
๐‘Ž22
๐‘Ž32
๐‘Ž13
๐‘Ž
๐‘Ž23 | = ๐‘Ž11 | 22
๐‘Ž32
๐‘Ž33
๐‘Ž23
๐‘Ž21
๐‘Ž33 | − ๐‘Ž12 |๐‘Ž31
๐‘Ž23
๐‘Ž21
๐‘Ž33 | + ๐‘Ž13 |๐‘Ž31
๐‘Ž22
๐‘Ž32 |
|๐ด| = ๐‘Ž11 (๐‘Ž22 ๐‘Ž33 − ๐‘Ž23 ๐‘Ž32 ) − ๐‘Ž12 (๐‘Ž21 ๐‘Ž33 − ๐‘Ž23 ๐‘Ž31 ) + ๐‘Ž13 (๐‘Ž21 ๐‘Ž32 − ๐‘Ž22 ๐‘Ž31 )
Example 11
6
|๐ด| = |9
5
1 7
3
3 8| = 6 |
2
2 4
8
9
|−|
4
5
8
9
| + 7|
4
5
3
|
2
|๐ด| = 6(3 × 4 − 8 × 2) − (9 × 4 − 8 × 5) + 7(9 × 2 − 3 × 5) = 1
Note that the above “3rd order” determinant is expanded into an expression containing three “2 nd order”
๐‘Ž22 ๐‘Ž23
๐‘ ๐‘ข๐‘determinants. For example, the subdeterminant |๐‘Ž
๐‘Ž33 | is obtained by deleting the third row and third column
32
LN1—Matrix Algebra
Page 7 of 12
of |๐ด|. This subdeterminant is called the minor of the element ๐‘Ž11 , which is the element located in first row and first
column of |๐ด|. The minor of ๐‘Ž11 is denoted by |๐‘€11 |.
๐‘Ž11
|๐‘€11 | = |๐‘Ž21
๐‘Ž31
๐‘Ž12
๐‘Ž22
๐‘Ž32
๐‘Ž13
๐‘Ž
๐‘Ž23 | = | 22
๐‘Ž32
๐‘Ž33
๐‘Ž23
๐‘Ž33 |
In general, |๐‘€๐‘–๐‘— | represents the minor of the element ๐‘Ž๐‘–๐‘— , which is obtained by deleting the ๐‘–th row and the ๐‘—th column.
Thus, we can write the 3rd order determinant |๐ด| above as:
|๐ด| = ๐‘Ž11 |๐‘€11 | − ๐‘Ž12 |๐‘€12 | + ๐‘Ž13 |๐‘€13 |
A minor with an algebraic sign attached to it is called the cofactor of a given element ๐‘Ž๐‘–๐‘— . The “+” or “−“ sign depends
whether the sum ๐‘– + ๐‘— is even or odd:
|๐ถ๐‘–๐‘— | ≡ (−1)๐‘–+๐‘— |๐‘€๐‘–๐‘— |
Thus, the determinant |๐ด|, obtained through the expansion by the first row can be represented as,
3
|๐ด| = ∑ ๐‘Ž1๐‘— |๐ถ1๐‘— |
๐‘—=1
Using the cofactor notation the determinant is expressed as:
|๐ด| = ๐‘Ž11 |๐ถ11 | + ๐‘Ž12 |๐ถ12 | + ๐‘Ž13 |๐ถ13 |
You can expand any nth order determinant by any row or any column.
determinant |B| by the second row we have:
For example, expanding a 4th order
4
|๐ต| = ∑ ๐‘2๐‘— |๐ถ2๐‘— |
๐‘—=1
In general,
๐‘›
|๐ด| = ∑ ๐‘Ž๐‘–๐‘— |๐ถ๐‘–๐‘— |
(expansion by ๐‘–th row)
๐‘—=1
๐‘›
|๐ด| = ∑ ๐‘Ž๐‘–๐‘— |๐ถ๐‘–๐‘— |
(expansion by ๐‘—th column)
๐‘–=1
1.
5.3.3. Basic Properties of Determinants
The determinant of the transpose A′ is the same as the determinant of A
|๐ด| = |5
4
2.
3
| = 28
8
|๐ด′| = |5
3
4
| = 28
8
Interchange of any two rows (or any two) columns will change the algebraic sign of the determinant,
|๐ด| = |5
4
LN1—Matrix Algebra
3
| = 28
8
|๐ต| = |
4
5
8
| = −28
3
Page 8 of 12
3.
Multiplication of any one row (or one column) by a scalar k will change the value of the determinant k-fold.
|๐ด| = |5
4
4.
3
| = 28
8
2×3
| = 2 × 28 = 56
8
The addition of a multiple of any row (column) to another row (column) will leave the value of the
determinant unchanged.
|๐ด| = |5
4
5.
|๐ต| = |2 × 5
4
3
| = 28
8
|๐ต| = |
5
4+2×5
3
| = 5 × 14 − 3 × 14 = 28
8+2×3
If one row (or column) is a multiple of another row (or column) the determinant vanishes—it is zero.
|๐ด| = | 5
2×5
3
| = 2(5 × 3 − 5 × 3) = 0
2×3
By extension, if one row (column) is a linear combination of another row (column) or a linear combination
any two rows (columns) the determinant vanishes.
|๐ด| = |
5
5+5×๐‘˜
6
|๐ต| = | 1
6+๐‘˜
|๐ต| = 6 |
3
| = 15 + 15๐‘˜ − 15 − 15๐‘˜ = 0
3+3×๐‘˜
3
4
3 + 4๐‘˜
4
3 + 4๐‘˜
1
−2 | = 6|๐ถ11 | + 3|๐ถ12 | + |๐ถ13 |
1 − 2๐‘˜
−2
1
| + 3(−1) |
1 − 2๐‘˜
6+๐‘˜
−2
1
|+|
1 − 2๐‘˜
6+๐‘˜
4
|
3 + 4๐‘˜
|๐ต| = 6(4 − 8๐‘˜ + 6 + 8๐‘˜) − 3(1 − 2๐‘˜ + 12 + 2๐‘˜) + (3 + 4๐‘˜ − 24 − 4๐‘˜)
|๐ต| = 6(10) − 3(13) + (−21) = 0
6.
The expansion of a determinant by alien cofactors yields a value of zero.
6
|๐ด| = |9
5
1
3
2
7
8|
4
Multiply the elements of the first row by the cofactors of the elements of second row.
3
∑ ๐‘Ž1๐‘— |๐ถ2๐‘— | = ๐‘Ž11 |๐ถ21 | + ๐‘Ž12 |๐ถ22 | + ๐‘Ž13 |๐ถ23 |
๐‘—=1
3
∑ ๐‘Ž1๐‘— |๐ถ2๐‘— | = 6(−1) |
๐‘—=1
LN1—Matrix Algebra
1
2
7
6
|+|
4
5
7
6
| + 7(−1) |
4
5
1
| = −6(−10) − (11) − 7(7) = 0
2
Page 9 of 12
5.3.4. Criterion for Non-singularity of a Matrix—Non-vanishing Determinant
As mentioned in Section 5.3.1.: In order for matrix ๐ด to be nonsingular all of its rows must be linearly independent.
This means that none of the rows can be a linear combination of other rows. There you have it! If any row of the
determinant of the coefficient matrix of system of linear equations is a linear combination of any two rows, or a
multiple of any other row, the determinant vanishes, the matrix is singular and the inverse of the matrix does not
exist.
5.4. How to Find the Inverse of Matrix A
Let’s start with a (3 × 3) matrix:
๐‘Ž11
๐ด = [๐‘Ž21
๐‘Ž31
1.
๐‘Ž12
๐‘Ž22
๐‘Ž32
๐‘Ž13
๐‘Ž23 ]
๐‘Ž33
Form the cofactor matrix by replacing each element by its cofactor.
|๐ถ11 |
๐ถ = [|๐ถ21 |
|๐ถ31 |
2.
|๐ถ12 |
|๐ถ22 |
|๐ถ32 |
|๐ถ13 |
|๐ถ23 |]
|๐ถ33 |
Form the transpose of the cofactor matrix: ๐ถ′. This called the adjoint of matrix A and is donate by adj ๐ด.
|๐ถ11 |
๐ถ = adj ๐ด = [|๐ถ12 |
|๐ถ13 |
′
3.
|๐ถ21 |
|๐ถ22 |
|๐ถ23 |
|๐ถ31 |
|๐ถ32 |]
|๐ถ33 |
Pre multiply adj ๐ด by matrix ๐ด
๐‘Ž11
๐ด๐ถ′ = [๐‘Ž21
๐‘Ž31
๐‘Ž12
๐‘Ž22
๐‘Ž32
๐‘Ž13 |๐ถ11 |
๐‘Ž23 ] [|๐ถ12 |
๐‘Ž33 |๐ถ13 |
3
|๐ถ21 |
|๐ถ22 |
|๐ถ23 |
3
|๐ถ31 |
|๐ถ32 |]
|๐ถ33 |
3
∑ ๐‘Ž1๐‘— |๐ถ1๐‘— |
∑ ๐‘Ž1๐‘— |๐ถ2๐‘— |
∑ ๐‘Ž1๐‘— |๐ถ3๐‘— |
๐‘—=1
3
๐‘—=1
3
๐‘—=1
3
∑ ๐‘Ž2๐‘— |๐ถ2๐‘— |
∑ ๐‘Ž2๐‘— |๐ถ3๐‘— |
๐‘—=1
3
๐‘—=1
3
∑ ๐‘Ž3๐‘— |๐ถ2๐‘— |
∑ ๐‘Ž3๐‘— |๐ถ3๐‘— |
]
๐‘—=1
๐ด๐ถ′ = ∑ ๐‘Ž2๐‘— |๐ถ1๐‘— |
๐‘—=1
3
∑ ๐‘Ž3๐‘— |๐ถ1๐‘— |
[๐‘—=1
๐‘—=1
Note that in the three elements of the principal diagonal of AC′ all are determinants of matrix ๐ด, and the other
elements are expansions by alien cofactors and thus are all equal to zero.
|๐ด| 0
๐ด๐ถ′ = [ 0 |๐ด|
0
0
0
0]
|๐ด|
Now you can view ๐ด๐ถ′ as the identity matrix multiplied by the scalar |๐ด|.
1
๐ด๐ถ ′ = |๐ด| [0
0
LN1—Matrix Algebra
0
1
0
0
0] = |๐ด|๐ผ
1
Page 10 of 12
๐ด๐ถ ′ = |๐ด|๐ผ
4.
Multiply both sides by the scalar 1⁄|๐ด|.
๐ด๐ถ′
=๐ผ
|๐ด|
or
๐ด
๐ถ′
=๐ผ
|๐ด|
Then pre multiply both sides by ๐ด−1 :
๐ด−1 ๐ด
๐ถ′
= ๐ด−1 ๐ผ
|๐ด|
๐ถ′
= ๐ด−1
|๐ด|
Now we have obtained the inverse of ๐ด
๐ด−1 =
๐ถ′
1
=
adj ๐ด
|๐ด| |๐ด|
Example 12
Find the inverse of the coefficient matrix from Example 1. Show each step.
6
๐ด = [1
4
3
4
−1
1
−2]
5
The cofactor matrix is:
4 −2
|
−1
5
3 1
๐ถ = −|
|
−1 5
3
1
[ |4 −2|
−2
|
5
6 1
|
|
4 5
6
1
−|
|
1 −2
|
−|
1
4
1
4
6
−|
4
|
4
|
−1
18
3
| = [−16
−1
−10
6 3
|
|]
1 4
−13
26
13
−17
18]
21
Adjoint of A is:
18
๐ถ ′ = [−13
−17
−16
26
18
−10
13]
21
Now pre multiply ๐ถ′ by ๐ด.
6
3
๐ด๐ถ′ = [1
4
4 −1
1
18
−2] [−13
5 −17
52 0
๐ด๐ถ′ = [ 0 52
0
0
−16
26
18
−10
13]
21
0
0]
52
The determinant of A is the element in the principal diagonal of ๐ด๐ถ′.
|๐ด| = 52
LN1—Matrix Algebra
Page 11 of 12
Thus,
๐ด−1 =
18
๐ถ′
1
= ( ) [−13
|๐ด|
52
−17
0.3462
๐ด−1 = [−0.2500
−0.3269
−16
26
18
−0.3077
0.5000
0.3462
−10
13]
21
−0.1923
0.2500]
0.4038
You may check the result by applying the =MINVERSE command in Excel.
.
LN1—Matrix Algebra
Page 12 of 12
Download