Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

86
Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Transcript of Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Page 1: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Numerical Analysis

1

EE, NCKUTien-Hao Chang (Darby Chang)

Page 2: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

In the previous slide Rootfinding

– multiplicity

Bisection method– Intermediate Value Theorem

– convergence measures

False position– yet another simple enclosure method

– advantage and disadvantage in comparison with bisection method

2

Page 3: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

In this slide Fixed point iteration scheme

– what is a fixed point?

– iteration function

– convergence

Newton’s method– tangent line approximation

– convergence

Secant method3

Page 4: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Rootfinding Simple enclosure

– Intermediate Value Theorem

– guarantee to converge• convergence rate is slow

– bisection and false position

Fixed point iteration– Mean Value Theorem

– rapid convergence• loss of guaranteed convergence

4

Page 5: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

2.3

5

Fixed Point Iteration Schemes

Page 6: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

6

Page 7: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

7

There is at least one point on the graph at which the tangent lines is parallel to the secant line

Page 8: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Mean Value Theorem

We use a slightly different formulation

An example of using this theorem– proof the inequality

8

Page 9: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

9

Page 10: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Fixed points Consider the function

– thought of as moving the input value of to the output value

– the sine function maps to • the sine function fixes the location of

– is said to be a fixed point of the function

10

Page 11: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

11

Page 12: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Number of fixed points According to the previous figure, a

trivial question is– how many fixed points of a given

function?

12

Page 13: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

13

Page 14: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

14

𝑔 ′ (𝑥 )≤𝑘<1

Page 15: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Only sufficient conditions

Namely, not necessary conditions– it is possible for a function to violate one or more of the

hypotheses, yet still have a (possibly unique) fixed point

15

Page 16: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Fixed point iteration

16

Page 17: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Fixed point iteration If it is known that a function has a

fixed point, one way to approximate the value of that fixed point is fixed point iteration scheme

These can be defined as follows:

17

Page 18: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

18

In action

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

Page 19: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

19

Page 20: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

20

Page 21: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Any Questions?

21

About fixed point iteration

Page 22: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Relation to rootfinding Now we know what fixed point

iteration is, but how to apply it on rootfinding?

More precisely, given a rootfinding equation, f(x)=x3+x2-3x-3=0, what is its iteration function g(x)?

22

hint

Page 23: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Iteration function Algebraically transform to the form

– …

Every rootfinding problem can be transformed into any number of fixed point problems– (fortunately or unfortunately?)

23

Page 24: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

24

Page 25: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

25

In action

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

Page 26: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

26

Page 27: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Analysis #1 iteration function converges

– but to a fixed point outside the interval

#2 fails to converge– despite attaining values quite close to #1

#3 and #5 converge rapidly– #3 add one correct decimal every iteration

– #5 doubles correct decimals every iteration

#4 converges, but very slow

27

Page 28: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Convergence This analysis suggests a trivial question

– the fixed point of is justified in our previous theorem

28

Page 29: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

29

Page 30: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

30

Page 31: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

31

Page 32: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

𝑘 demonstrates the importance of the

parameter – when , rapid

– when , dramatically slow

– when , roughly the same as the bisection method

32

Page 33: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Order of convergence of fixed point iteration schemes

33

All about the derivatives,

Page 34: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

34

Page 35: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

35

Page 36: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

36

Page 37: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

37

Page 38: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

38

Page 39: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Stopping condition

39

Page 40: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

40

Page 41: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Two steps

41

Page 42: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

The first step

42

Page 43: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

The second step

43

Page 44: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Any Questions?

44

2.3 Fixed Point Iteration Schemes

Page 45: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

2.4

45

Newton’s Method

Page 46: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

46

Page 47: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

47

Page 48: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Newton’s Method

Definition

48

Page 49: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

49

In action

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

Page 50: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

50

Page 51: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

In the previous example Newton’s method used 8 function

evaluations Bisection method requires 36

evaluations starting from False position requires 31 evaluations

starting from

51

Page 52: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

52

Page 53: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Any Questions?

53

Page 54: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Initial guess Are these comparisons fair?

– , converges to after 5 iterations

– , fails to converges after 5000 iterations

– , converges to after 42 iterations

54

example

answer

Page 55: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Initial guess Are these comparisons fair?

– , converges to after 5 iterations

– , fails to converges after 5000 iterations

– , converges to after 42 iterations

55

answer

Page 56: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Initial guess Are these comparisons fair?

– , converges to after 5 iterations

– , fails to converges after 5000 iterations

– , converges to after 42 iterations

56

Page 57: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

in Newton’s method Not guaranteed to converge

– , fails to converge

May converge to a value very far from – , converges to

Heavily dependent on the choice of

57

Page 58: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Convergence analysis for Newton’s method

58

Page 59: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

59

The simplest plan is to apply the general fixed point iteration convergence theorem

Page 60: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Analysis strategy To do this, it is must be shown that

there exists such an interval, , which contains the root , for which

60

Page 61: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

61

Page 62: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

62

Page 63: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

63

Page 64: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

64

Page 65: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Any Questions?

65

Page 66: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Newton’s Method

Guaranteed to Converge?

Why sometimes Newton’s method does not converge?

This theorem guarantees that exists But it may be very small

66

hint

answer

Page 67: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Newton’s Method

Guaranteed to Converge?

Why sometimes Newton’s method does not converge?

This theorem guarantees that exists But it may be very small

67

answer

Page 68: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Newton’s Method

Guaranteed to Converge?

Why sometimes Newton’s method does not converge?

This theorem guarantees that exists But it may be very small

68

Page 69: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

69

Oh no! After these annoying analyses, the Newton’s method is still not guaranteed to converge!?

http://img2.timeinc.net/people/i/2007/startracks/071008/brad_pitt300.jpg

Page 70: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Don’t worry Actually, there is an intuitive method Combine Newton’s method and bisection

method– Newton’s method first

– if an approximation falls outside current interval, then apply bisection method to obtain a better guess

(Can you write an algorithm for this method?)

70

Page 71: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Newton’s Method

Convergence analysis At least quadratic

– , since

Stopping condition–

71

Page 72: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

72http://www.dianadepasquale.com/ThinkingMonkey.jpg

Recall that

Page 73: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

73

Is Newton’s method always faster?

Page 74: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

74

Page 75: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

75

In action

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

Page 76: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

76

Page 77: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Any Questions?

77

2.4 Newton’s Method

Page 78: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

2.5

78

Secant Method

Page 79: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Secant method Because that Newton’s method

– 2 function evaluations per iteration

– requires the derivative

Secant method is a variation on either false position or Newton’s method– 1 additional function evaluation per iteration

– does not require the derivative

Let’s see the figure first

79

answer

Page 80: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

80

Page 81: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Secant method Secant method is a variation on

either false position or Newton’s method– 1 additional function evaluation per

iteration

– does not require the derivative

– does not maintain an interval

– is calculated with and

81

Page 82: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

82

Page 83: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

83

Page 84: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

84

Page 85: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

85

Page 86: Numerical Analysis 1 EE, NCKU Tien-Hao Chang (Darby Chang)

Any Questions?

86

2.5 Secant Method