Numerical Analysis

EE, NCKU
Tien-Hao Chang (Darby Chang)

1
In the previous slide
Special matrices
strictly diagonally dominant matrix
symmetric positive definite matrix
Cholesky decomposition
tridiagonal matrix
Iterative techniques
Jacobi, Gauss-Seidel and SOR methods
conjugate gradient method
Nonlinear systems of equations
(Exercise 3)

2
In this slide
Eigenvalues and eigenvectors
The power method
locate the dominant eigenvalue
Inverse power method
Deflation

3
 Chapter 4
Eigenvalues and eigenvectors

4
Eigenvalues and eigenvectors
Eigenvalue

= = 0
det = 0
characteristic polynomial
Eigenvector
the nonzero vector for which =
associated with the eigenvalue

5
In Chapter 4
Determine the dominant eigenvalue
Determine a specific eigenvalue
Remove a eigenvalue
Determine all eigenvalues

6
 4.1
The power method

7
The power method
Different problems have different requirements
a single, several or all of the eigenvalues
the corresponding eigenvectors may or may not
also be required
To handle each of these situations efficiently,
different strategies are required
The power method
an iterative technique
locate the dominant eigenvalue
also computes an associated eigenvector
can be extended to compute eigenvalues

8
The power method

Basics

9
10
11
The power method

Approximated eigenvalue

12
Any Questions?

13
The power method

A common practice
Make the vector () have a unit
length

Why we question
need this step?

14
The power method

A common practice
Make the vector () have a unit
length

Why we need this step?

15
 Make the vector () have a unit
length
to avoid overflow and underflow

16
The power method

Complete procedure

17
Any Questions?

18
http://thomashawk.com/hello/209/1017/1024/Jackson%20Running.jpg

In action

19
what is the first estimate?

20
21
22
23
Any Questions?
The power method for generic matrices

24
The power method for symmetric matrices

25
 When is symmetric
more rapid convergence
still linear, but smaller asymptotic error
different scaling scheme (norm)
based on the theorem

26
The power method variation

27
28
 The power method

Approximated eigenvalue

Recall that 29
http://www.dianadepasquale.com/ThinkingMonkey.jpg
Any Questions?
The power method for symmetric
matrices

30
Why to
Require the matrix to be symmetric?

31
Any Questions?
4.1 The power method

32
An application of eigenvalue

33
34
35
Undirected graph

Relation to eigenvalue
Proper coloring
how to color the geographic regions on
a map regions that share a common
border receive different colors
Chromatic number
the minimum number of colors that can
be used in a proper coloring of a graph

36
Undirected graph

The dominant eigenvalue

37
Undirected graph

The corresponding eigenvector

38
Any Questions?

39
 4.2
The inverse power method

40
The inverse power method
An eigenvalue other than the dominant one
To derive the inverse power method, we will
need
the relationship between the eigenvalues of a
matrix to a class of matrices constructed
from
With that, we can
transform an eigenvalue of the dominant
eigenvalue of later
= 1

41
 is a polynomial of

42
43
44
The inverse power method
An eigenvalue other than the dominant one
To derive the inverse power method, we will
need
the relationship between the eigenvalues of a
matrix to a class of matrices constructed
from
With that, we can
transform an eigenvalue of the dominant
eigenvalue of
= 1

45
46
Any Questions?

47
How to
Find the eigenvalue smallest in
magnitude

48
Any Questions?
4.2 The inverse power method

49
 4.3
Deflation

50
Deflation
So far, we can approximate
the dominant eigenvalue of a matrix
the one smallest in magnitude
the one closest to a specific value
What if we need several of the
largest/smallest eigenvalues?
Deflation
to remove an already determined solution,
while leaving the remainder solutions
unchanged

51
 Within the context of polynomial rootfinding
remove each root by dividing out the
monomial
3 6 2 + 11 6 = 1 2 5 + 6 = (
1)( 2)( 3)
2 5 + 6 = ( 2)( 3) is a deflation of
3 6 2 + 11 6
For the matrix eigenvalue problem
shift the previously determined eigenvalue to
zero (while leaving the remainder eigenvalues
unchanged)
to do this, we need the relationship among the
eigenvalues of a matrix and

52
53
 Recall that 54
http://www.dianadepasquale.com/ThinkingMonkey.jpg
Deflation

Shift an eigenvalue to zero

55
While leaving the remaining eigenvalues unchanged

56
57
58
Deflation

Summary

59
Any Questions?

60
Do we
Miss something?

61
 Recall that 62
http://www.dianadepasquale.com/ThinkingMonkey.jpg
How to choose for the formula = 1 1 ?

63
Wielandt deflation

64
65
Wielandt deflation

Bonus

66
http://thomashawk.com/hello/209/1017/1024/Jackson%20Running.jpg

In action

67
68
69
Hotelling deflation
Recall that we choose 1, based on
infinity norm
Like the power method, there is
another deflation variation for
symmetric matrices

70
71
Any Questions?
4.3 Deflation

72