Scribd
Upload
4 views10 pages

Lay Linalg5 05 08

The document discusses iterative methods for estimating eigenvalues, focusing on the Power Method and the Inverse Power Method. The Power Method is applicable to matrices with a strictly dominant eigenvalue and involves iteratively computing matrix products to converge on the dominant eigenvalue and corresponding eigenvector. The Inverse Power Method is used to find the smallest eigenvalue and requires an initial estimate close to the desired eigenvalue, ultimately providing an accurate estimate through iterative calculations.

Uploaded by

pou
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views10 pages

Lay Linalg5 05 08

The document discusses iterative methods for estimating eigenvalues, focusing on the Power Method and the Inverse Power Method. The Power Method is applicable to matrices with a strictly dominant eigenvalue and involves iteratively computing matrix products to converge on the dominant eigenvalue and corresponding eigenvector. The Inverse Power Method is used to find the smallest eigenvalue and requires an initial estimate close to the desired eigenvalue, ultimately providing an accurate estimate through iterative calculations.

Uploaded by

pou
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

5 Eigenvalues and Eigenvectors

5.8
ITERATIVE ESTIMATES FOR
EIGENVALUES

© 2016 Pearson Education, Ltd.


THE POWER METHOD
 The power method applies to an matrix A with a
strictly dominant eigenvalue 1, which means that 1
must be larger in absolute value than all the other
eigenvalues.

© 2016 Pearson Education, Ltd. Slide 5.8- 2


THE POWER METHOD
 The power method for estimating a strictly dominant
eigenvalue
1. Select an initial vector x0, whose largest entry is 1.
2. For k = 0, 1, . . . ,
a) Compute Axk.
b) Let k be an entry in Axk whose absolute value is as large as
possible.
c) Compute xk+1 = (1/ k)Axk.
3. For almost all choices of x0, the sequence { k}
approaches the dominant eigenvalue, and the sequence
{xk} approaches a corresponding eigenvector.

© 2016 Pearson Education, Ltd. Slide 5.8- 3


THE POWER METHOD

 Example 1 Let
 A has eigenvalues 2 and 1, and the eigenspace for 1=

2 is the line through 0 and 1 .

 Start from .
 For k = 0, . . . , 8, compute Akx and construct the line
through 0 and Akx. What happens as k increases?

© 2016 Pearson Education, Ltd. Slide 5.8- 4


THE POWER METHOD
 Solution The first three calculations are

2 =

3 =
 Analogous calculations complete Table 1.

© 2016 Pearson Education, Ltd. Slide 5.8- 5


THE POWER METHOD
 The vectors x, Ax, . . . , A4x are shown in Fig. 1 below:

 The angle between the line (subspace) determined by


Akx and the line (eigenspace) determined by v1 goes to
zero as .
© 2016 Pearson Education, Ltd. Slide 5.8- 6
THE INVERSE POWER METHOD
 The inverse power method for estimating an eigenvalue of A
1. Select an initial estimate sufficiently close to .
2. Select an initial vector x0 whose largest entry is 1.
3. For k = 0, 1, . . . ,
a) Solve (A - I yk xk for yk.
b) Let k be an entry in yk whose absolute value is as large as
possible.
c) Compute vk = 1/ k)
d) Compute xk+1 = (1/ k)yk.
4. For almost all choices of x0, the sequence { k} approaches the
eigenvalue of A, and the sequence {xk} approaches a
corresponding eigenvector.
© 2016 Pearson Education, Ltd. Slide 5.8- 7
THE INVERSE POWER METHOD
 Example 3 It is not uncommon in some applications to
need to know the smallest eigenvalue of a matrix A and to
have at hand rough estimates of the eigenvalues.
 Suppose 21, 3.3, and 1.9 are estimates for the eigenvalues
of the matrix A below.
 Find the smallest eigenvalue, accurate to size decimal
places.

© 2016 Pearson Education, Ltd. Slide 5.8- 8


THE INVERSE POWER METHOD
 Solution The smallest eigenvalues seem close together so
we use the inverse power method for A – 1.9I. Results of a
MATLAB calculation are shown in Table 3 below:

© 2016 Pearson Education, Ltd. Slide 5.8- 9


THE INVERSE POWER METHOD
 Here:
 x0 was chosen arbitrarily,
 y = (A – 1.9I )-1xk,
 k is the largest entry in yk,
 vk = 1.9 + 1/ k, and
 xk+1 = (1/ k) yk .

 The smallest eigenvalue is 2.


 The estimated value is 2.0000002.

© 2016 Pearson Education, Ltd. Slide 5.8- 10

You might also like