METODE ITERASI TITIK TETAP

 2.2 Fixed-Point Iteration

equivalent
f (x) = 0 x = g (x)

Root of f (x) Fixed-point of g (x)

Start from an initial approximation p0 and generate the sequence

 pn  n0 by letting pn = g( pn – 1 ), for each n  1. If the sequence
converges to p and g is continuous, then
Idea p  lim pn  lim g  pn 1   g ( lim pn 1 )  g ( p )
n  n  n 

So basically we are
Oh yeah? Who tells
done! I can’t believe
you that the method
it’s so simple!
is convergent?
What’s the problem?
 y y=x y y=x
t1 t0
y=g(x)
t0

 t1
y=g(x)

x x
p0 p1 p p0 p p1

y y=x y y=x
y=g(x)
y=g(x)
t0

t0
 t1

t1

x x
p1 p0 p p0 p p1
6/10
 Theorem: (Fixed-Point Theorem)
Let g  C[a, b] be such that g(x)  [a, b], for all x in [a, b]. Suppose, in
addition, that g’ exists on (a, b) and that a constant 0 < k < 1 exists with
|g’(x)|  k for all x  (a, b). Then, for any number p0 in [a, b], the
  
sequence defined by pn = g( pn – 1 ), n  1, converges to the unique fixed
point p in [a, b].
Proof: a. Let f(x) = g(x) – x.
a  g( x )  b f (a )  g(a )  a  0 and f (b )  g (b)  b  0
The Intermediate Value Theorem implies that f has a root,
and hence g has a fixed point.
b. Prove by contradiction:
Suppose p and q are both fixed points of g in [a, b] and p  q.
The Mean Value Theorem implies that a number  exists
between p and q with g(p) – g(q) = g’() (p – q).
(1 – g’()) (p – q) == 0p = q Contradiction!
Hence the fixed point is unique.
7/10
 Proof (continued):
c. lim | pn  p | 0 ?
n

g(x)  [a, b], for all x in [a, b] pn is defined for all n  0

| pn  p || g( pn 1 )  g ( p ) || g' ( ) || pn1  p | k | pn1  p |
 k 2 | pn 2  p | ...  k n | p0  p |
0

Corollary: If g satisfies the hypotheses of the Fixed-Point Theorem,

then bounds for the error involved in using pn to approximate p are
given by (for all n  1)
1 kn
| pn  p |  | pn  1  p n | and | pn  p |  | p1  p0 |
1 k 1 k
Proof: | pn 1  pn |  | pn  p |  | pn  1  p |  | pn  p |  k | pn  p |

| pnCan be used  Thesmaller

1  pn |  | g ( x n )  g ( x n  1 ) |  | g ( ξ n )( pn  pn 1 ) |
to control the the value of k, the faster
 k | pn  pn1 |  ......  k n | p1  p0 |
accuracy the convergence.

8/10
Example 3.8

6
7
 Algorithm: Fixed-Point Iteration

Find a solution to p = g(p) given an initial approximation p0.

Input: initial approximation p0; tolerance TOL; maximum number of

iterations Nmax.
Output: approximate solution p or message of failure.
Step 1 Set i = 1;
Step 2 While ( i  Nmax) do steps 3-6
Step 3 Set p = g(p0); /* compute pi */
Step 4 If | p  p0 | < TOL then Output (p); /* successful */
STOP;
Step 5 Set i ++;
Step 6 Set p0 = p ; /* update p0 */

Step 7 Output (The method failed after Nmax iterations); /* unsuccessful */

STOP.

9/10
Flow Chart – Fixed Point

Start

Input: xo , s, maxi

i=0
a=1.1s

1
 1

while False
a< s &
i >maxi

xn  g  x0 Print: xo, f(xo) ,a , i

i  i 1

i=1
or Stop
xn=0

True xn  xo
a  100%
xn

x0=xn
g (x)

11
Examples
Example: fixed point problems
Examples: FPI
Example: FPI
Convergence of FPI
 Example: Find the unique root of the equation x3 + 4x2 – 10 = 0 in [1, 2].
Using the following equivalent fixed-point forms with p0 = 1.5, which one is
the best? (The root is approximately 1.365230013.)

a) 
x  g1 ( x )  x  x 3  4 x 2  10 ;
b ) x  g ( x )  10 / x  4 x ;
2

c ) x  g ( x )  10  x / 2 ;3
d ) x  g ( x )  10 /( 4  x ) ;
 3

x  4 x  10
3  2
4

e) x  g ( x)  x 

5
3 x  8 x
;2 OK in [1,But why? k  0.15
1.5].
k  0.66

HW:
p.64 #3, p.65 #19

10/1