Академический Документы
Профессиональный Документы
Культура Документы
Chapter 4
Eigenvalues
Let A be an n n matrix. A number (scalar) is called an eigenvalue of A if there exists a nonzero vector called eigenvector v such that Av = v.
(4.1)
Eigenvalues and eigenvectors are widely used in various applications such as to determine the stability of a finite-difference scheme to solve a partial differential equation and finding the solution for the system of differential equations. Rewriting Eq. (4.1) as Av = Iv, where I is an identity matrix, yields ( A I )v = 0 . (4.2)
The linear homogeneous system (4.2) has a nontrivial solution v 0 if and only if the matrix A I is singular that is | A I |= 0 . Solving | A I | = 0 leads to solve the characteristics equation which will yields n eigenvalues of matrix A. Substituting each eigenvalue into Eq. (4.2) will get its corresponding eigenvector.
4.1
Power Method
For an n n matrix A, the most dominant eigenvalue, 1 , where | 1 | > | 2 | > | n | and its corresponding eigenvector, , can be obtained by 1 the power method
( k +1) =
1 A ( k ) , k = 0, 1, 2, mk +1
65
Eigenvalues
The iteration is iterated with a given initial eigenvector, v ( 0 ) until | mk +1 mk | < . Then, the most dominant eigenvalue is
= m1 + , 1 k
2. 3. 4. 5. 6. 7.
Input the matrix A into the calculator. Input v ( 0 ) as matrix B into the calculator. Calculate Av ( 0 ) as matrix multiplication of AB using the calculator. Find m1 (the maximum absolute value of Av ( 0 ) ).
Av ( o) by updating matrix B in the calculator. m Repeat Step 4 to Step 6 until | m k +1 m k | < . The most dominant
Calculate v (1) =
1 = mk +1 ,
and its corresponding eigenvector is v1 = ( k +1) . EXAMPLE 4.1 Given the matrix 4 3 0 A = 3 4 1 . 0 1 4
66
Eigenvalues
Use power method to find the most dominant eigenvalue, and its 1 (0) T corresponding eigenvector, of the matrix A using = (1 1 0) . Calculate 1 until | mk +1 mk | 0.005. SOLUTION Step 1: Put in an initial eigenvector, ( 0) = (1 1 0) T into the table below: k 0
v (k ) 1 Av (k )
mk + 1
67
Eigenvalues
z1p
Screen
Eigenvalues
Screen
Av ( k ) 7
mk + 1
Screen
69
Eigenvalues
k 0 1 2 3 4
v (k ) 1 1 1 1 1
7 7 6.920 6.866
mk + 1
Since | m4 m3 |= 0.002 < 0.005, so the most dominant eigenvalue, 1 m4 = 7.162, (4) T and its corresponding eigenvector, v1 v = (0.959 1 0.286) .
4.2
For an n n matrix A, the smallest eigenvalue, n , where | 1 | > | 2 | > | n | and its corresponding eigenvector, n , is given by the shifted power method
( k +1) =
1 Ashifted ( k ) , k = 0, 1, 2, mk +1
The iteration is iterated with a given initial eigenvector, v ( 0 ) until | mk +1 mk | < . Then, the most dominant eigenvalue of the matrix Ashifted is
70
Eigenvalues
sit d he f
= k+ m 1,
mk + 1
Input the matrix A into the calculator. Input matrix I as matrix B into the calculator. Calculate A shifted = A 1 I using the calculator. Input matrix Ashifted as matrix A into the calculator. Input vector v (k ) as matrix C into the calculator. (k ) Calculate A shifted v as matrix multiplication of AC into the calculator.
(0) Find m1 (the maximum absolute value of Ashifted v ).
(1) Calculate v =
Av ( o ) by updating matrix C in the calculator. m1 10. Repeat Step 7 to Step 9 until | m k +1 m k | < . The smallest eigenvalue of
matrix A is given by
3 = 1 + mk +1 ,
and its corresponding eigenvector is v3 = ( k +1) . EXAMPLE 4.2 Given the matrix
71
Eigenvalues
4 3 0 A = 3 4 1 . 0 1 4 Use the power method to find the smallest eigenvalue and its corresponding eigenvector, of the matrix A by the shifted power method using 1 (0) T = (1 1 0) . Calculate until | mk +1 mk | 0.005. SOLUTION Note: Using the answer from Example 4.1, the dominant eigenvalue is 1 m4 = 7.162 . Step 1: Put in an initial eigenvector, v ( 0 ) into the table below: k 0
v (k ) 1
Ashifted v (k )
mk + 1
Step 2: To get A shifted = A 1 I , we must first input matrix A into the calculator. Refer Step 2 in Example 4.1 to input matrix A into the calculator. Step 3: Input matrix I as matrix B into the calculator. Press q4 2 2 1 1p0p0p 0p1p0p 0p0p1p Screen
72
Eigenvalues
C Step 4: Calculate A shifted = A l I using 1 = 7.162 from Example 4.1. Press q43-7.162q44 Screen
p Step 5: Input matrix Ashifted Press q(STO) (A) to copy matrix answer to matrix A Step 6: Input vector v (k ) as matrix C into the calculator. Press q4 2 3 3 1p1p0p Screen as matrix A into the calculator. Screen
73
Eigenvalues
C
(k ) Step 7: Calculate A shifted v as matrix multiplication of AC into the calculator.
Press q4 3Oq45 p
Ashifted v (k )
Screen
k 0
v (k ) 1
mk + 1
-0.162
-0.162
-1
mk + 1
-0.162
-0.162
-1
-1
Screen
74
Eigenvalues
k 0 1
1 0.162
v (k ) 1 0.162
Ashifted v (k )
mk + 1
0 1
-0.162
-0.162
-1
-1
Step 10: Repeat Step 7 to Step 9 until | mk +1 mk | 0.005 . The complete result is shown below. k v (k ) 0 1 1 0 1 0.162 0.162 1 2 0.008 0.309 1 3 -0.260 0.563 1 4 -0.674 0.955 1 5 -0.827 1 0.681 6 -0.888 1 0.499 | m6 m5 |= 0.000 < 0.005 , Since 3 1 + m6 = 7.162 6.324 = 0.838,
vs = ( 0.888 1 0.499 )T .
Ashifted v (k )
mk + 1
-0.162 -1 -1 -1.026 -3.324 -3.324 -1.953 -3.471 -3.471 -3.558 -3.724 -3.724 -6.042 -4.118 -6.042 -6.324 -3.155 -6.324 -6.324 -2.577 -6.324 the smallest eigenvalue, corresponding eigenvector, is
4.3
For an n n matrix A, the smallest eigenvalue, n , where | 1 | > | 2 | > | n | and its corresponding eigenvector, n , is given by the inverse power method
( k +1) =
1 1 ( k ) A , k = 0, 1, 2, mk +1
where m k + is the maximum absolute value of A1v ( k ) . 1 The iteration is iterated with a given initial eigenvector, v ( 0 ) until | mk +1 mk | < . Then, the smallest eigenvalue is
n =
1 , mk +1
75
Eigenvalues
2. 3. 4. 5. 6. 7.
Input the matrix A into the calculator. Input v ( 0 ) as matrix B into the calculator. Calculate A 1v ( 0 ) as matrix multiplication of A1 B using the calculator. Find m1 (the maximum absolute value of A1v (0) ).
Av ( o) by updating matrix B in the calculator. m Repeat Step 4 to Step 6 until | m k +1 m k | < . The smallest eigenvalue of
Calculate v (1) =
matrix A is given by
3 =
and its corresponding eigenvector is
1 , mk +1
v3 = ( k +1) . EXAMPLE 4.3 Given the matrix 4 3 0 A = 3 4 1 . 0 1 4 Use the inverse power method to find the smallest eigenvalue, and its 3 3 of the matrix A using (0) = (1 1 0)T . Calculate corresponding eigenvector, until | mk +1 mk | 0.005. SOLUTION
76
Eigenvalues
A1v ( k ) 0
mk + 1
Step 2: Input matrix A into the calculator. Refer to Step 2 in Example 4.1. Step 3: Input vector as matrix B into the calculator. Refer to Step 3 in Example 4.1. Step 4: Calculate A 1 ( k ) as matrix multiplication of A 1 B into the calculator.
77
Eigenvalues
Press q4 3 u O q4 4 p k 0
v (k ) 1 A 1v ( k ) 0.167
Screen
mk + 1
0.125
0.042
0.125
A 1v ( k ) 0.167
mk + 1
0.042
0.167
Screen
78
Eigenvalues
k 0 1
1 0.749
v (k ) 1 1
0 0.251
0.125
0.042
0.167
Step 7: Repeat Step 4 to Step 6 until | mk +1 mk | 0.005 . The complete result is shown below. k 0 1 2 3 4 5 6 7
v (k ) 1 1 1 1 1 1 1 1 Av (k ) 0.167 0.334 0.833 1.133 1.186 1.193 1.194
mk + 1
Since | m6 m5 |= 0.001 < 0.005 , so the smallest eigenvalue, s m = 0.838 , and 7 its corresponding eigenvector, v1 v ( 7 ) = ( 0.949 1 0.316 )T .
4.4
EXAMPLE 4.4 4 3 0 Find the determinant of A = 3 4 1 . 0 1 4 SOLUTION Step 1: Input matrix A into the calculator. Refer to Step 2 in Example 4.1.
79
Eigenvalues
Step 2: Press q4 7 q4 3 p
4 So the determinant of A = 3 0 3 4 1 0 1 is 24. 4
Screen
Screen
Eigenvalues
q4 3 p 1 4 7 So, A = 2 5 8 . 3 6 9
T
4.6
EXAMPLE 4.6 4 3 0 Find the inverse of A = 3 4 1 . 0 1 4 SOLUTION Step 1: Input matrix A into the calculator. Refer to Step 2 in Example 4.1. Step 2: Press q4 3 u Screen
81
Eigenvalues
p 0.625 0.5 0.125 So, A = 0.5 0.667 0.167 . 0.125 0.167 0.292
1
EXERCISE 4.1 2 1 0 1. Given the matrix A = 1 2 1 . Find the most dominant eigenvalue, 1 , 0 1 2 and its corresponding eigenvector, v1 , of the matrix A using the power method. Use v ( 0 ) = (0 1 0) T and calculate until | mk +1 mk |< 0.005.
2. Given the matrix
Find the most dominant eigenvalue, and its corresponding eigenvector, of the matrix A using the power method. Use and calculate until Find the most dominant eigenvalue, 1 , and its corresponding eigenvector, of the matrix A using the power method. Use and calculate until Find the most dominant eigenvalue, 1 , and its corresponding eigenvector, of the matrix A using the power method. Use and calculate until
82
Eigenvalues
EXERCISE 4.2 1. Given the matrix Find the smallest eigenvalue, 3 , and its corresponding eigenvector, of the matrix A by using the shifted power method. Use and calculate until
2. Given the matrix
Find the smallest eigenvalue, 3 , and its corresponding eigenvector, of the matrix A by using the shifted power method. Use and calculate until Find the smallest eigenvalue, 3 , and its corresponding
eigenvector, of the matrix A by using the shifted power method. Use and calculate until
4. Given the matrix Find the smallest eigenvalue, 3 , and its corresponding
eigenvector, v3 , of the matrix A by using the shifted power method. Use and calculate until EXERCISE 4.3 Repeat Exercise 4.2 by using the inverse power method.
83
Eigenvalues
ANSWER 4.1 1. 1 = 3.414, v1 = ( 0.707 1 0.707 ) 2. v1 = ( 1 0.43 0.785 ) ANSWER 4.2 1. 3 = 0.586 , v1 = ( 0.290 1 0.580 )
T T
2. 3 = 0.238 , v3 = ( 1 0.762 0.432 ) 4. 3 =1.354 , v3 = ( 0.323 1 0.823) ANSWER 4.3 1. 3 = 0.586 , v3 = ( 0.707 1 0.707 )
v3 = ( 0.707 1 0.707 )
T T T
T T T
84