A.1 Iterative Methods

Embed Size (px)

DESCRIPTION

Numerical Analysis : Iterative Method

Citation preview

  • Solution to

    Algebraic &Transcendental

    Equations

    A

  • Algebraic functions

    0011

    1

    fyfyfyfn

    n

    n

    n

    The general form of an Algebraic function:

    fi = an i-th order polynomial.

    Example : 073362 233 yxyxxf3 f2 f0

    Polynomials are a simple class of algebraic function

    011

    1 axaxaxaxfn

    n

    n

    nn

    ais are constants.

  • A transcendental function is non-algebraic.

    May include trigonometric, exponential, logarithmic

    functions

    Examples:

    Transcendental functions

    12 xxf ln

    ).sin(. 50320 xexf x

  • Equation Solving

    Given an approximate location (initial value)

    find a single real root

    Root

    Finding

    Open

    Methods

    Brackting

    Methods

    Iterative

    Newton-

    Rapson

    Secant

    False-

    position

    Bisection

    non-linear

    Single variable

    A

  • Iterative method

    A.1

  • 6

    Simple Fixed-point Iteration

    ... 2, 1,k ,given )(

    )(0)(

    1

    okk xxgx

    xxgxf

    Now progressively estimate the value of x.

    Rearrange the function so that x is on the

    left side of the equation:

  • Problem

    Find the root of

    f(x) = e-x x

    There is no exact or

    analytic solution

    Numerical solution:

    0

    0)(

    21

    1

    2

    xfxfxf

    xxf

    exf

    xexf

    x

    x

  • Iterative Solution

    1. Start with a guess say x1=1,

    2. Generate

    a) x2=e-x1 = e-1= 0.368

    b) x3=e-x2= e-0.368 = 0.692

    c) x4=e-x3= e-0.692=0.500

    In general:

    After a few more iteration we will get

    nx

    nex

    156705670 .. e

  • Iteration

  • Convergence Examples

    Convergent staircase pattern Convergent spiral pattern

  • Divergence Example

    Divergent staircase pattern Divergent spiral pattern

  • Existence of Root

    There exists one and only one root if

    L is Lipschitz constant,

    10

    Liii

    baxxxxLxgxgii

    bxgabxai

    .,,.

    .

    Lxg

    xx

    xgxg

    xx

    )(lim 1

  • Convergence?

    ])( aagnn

    nn

    n

    nn

    aga

    agagag

    ag

    xgx

    and 0 [

    221

    1

    If x=a is a solution then,

    11

    111

    agif

    ag

    nn

    anxnnn ] [

    error reduces at each step

    i.e. iteration will converge

    If magnitude of 1st derivative

    at x=a is less than 1

    i

    ni

    i

    axi

    afafxf

    !

  • Problem

    Find a root near x=1.0 and x=2.0

    Solution:

    Starting at x=1, x=0.292893 at 15th iteration

    Starting at x=2, it will not converge

    Why? Relate to g'(x)=x. for convergence g'(x) < 1

    Starting at x=1, x=1.707 at iteration 19

    Starting at x=2, x=1.707 at iteration 12

    Why? Relate to

    142 2 xxxf

    412

    21 xxgx

    212 xxgx

    21

    212

    xxg

  • Aitkens Process

    A.2

  • kth Order Convergence

    Pervious iterative method has linear (1st order)

    convergence, since:

    For kth order convergence we have:

    Now consider a 2nd order method.

    Aitkens 2 process

    ag nn 1

    A knn 1

  • Aitkens process

    If is a root of the equation i.e., =g() then,

    Now if we use

    10

    2

    1

    1

    n

    n

    nn

    g

    g

    g

    1

    1

    0

    n

    n A

    Ag and

  • Aitkens process

    11

    2

    11

    1

    1

    1

    1

    1

    1

    1

    1

    2

    nnn

    nnn

    n

    n

    n

    n

    n

    n

    n

    n

    n

    n

    n

    n

    xxx

    xxx

    x

    x

    x

    x

    Ax

    Ax

    Ax

    A

  • Algorithm

    guess_value;

    while (! g()) {

    }

    11

    2

    11

    1

    1

    1

    2

    nnn

    nnn

    nn

    nn

    n

    xxx

    xxx

    xgx

    xgx

    x

  • Why 2?

    11

    11

    1

    1

    11

    2

    1

    1

    2

    2

    1

    11

    2

    11

    11

    2

    11

    2

    2

    2

    nnn

    nnnn

    nn

    nn

    nn

    nnn

    n

    nn

    nnn

    nnn

    nnn

    nnn

    xxx

    xxxx

    xx

    xx

    xx

    xxx

    where

    x

    xx

    xxx

    xxx

    xxx

    xxx