6 OneD Unconstrained Opt

download 6 OneD Unconstrained Opt

of 29

Transcript of 6 OneD Unconstrained Opt

  • 8/14/2019 6 OneD Unconstrained Opt

    1/29

    CBB 4333Process Optimisation

    ONE-DIMENSIONALUNCONSTRAINED OPTIMISATION

    Dr Abd Halim Shah Maulud

    Universiti Teknologi PETRONAS

    Sept 2013

  • 8/14/2019 6 OneD Unconstrained Opt

    2/29

    TOPIC OUTCOMES

    By the end of the topic, students are able to:

    discuss the concept of one-dimensionalunconstrained optimisation

    solve one-dimensional unconstrained optimisation analytically

    via solving f= 0 & determine x*

    analytically or numerically function value-based methods

    derivative-based methods

  • 8/14/2019 6 OneD Unconstrained Opt

    3/29

    OPTIMALITY

    necessary conditions

    x* is stationary point

    sufficient conditions

    max

    min

    saddle point

  • 8/14/2019 6 OneD Unconstrained Opt

    4/29

    ANALYTICAL METHOD

    necessary conditionoptimal solution

    problems?

    non-linear equations difficult to solve

    need more convenient methods

    nix

    f

    i

    ,,1;0

  • 8/14/2019 6 OneD Unconstrained Opt

    5/29

    METHODS FOR OPTIMAL SOLUTIONS SEARCH

    Function value-based methods

    search by compare function values at a sequence of

    trial points

    Derivative-based methods involve derivative at each iteration and determine

    potential optimum by necessary condition

  • 8/14/2019 6 OneD Unconstrained Opt

    6/29

    FUNCTION VALUE-BASED METHODS

    general procedure for minimisation

    1. Start with an initial value, x0

    2. Calculatef(x0)

    3. Change x0for next stage x1

    4. Calculatef(x1)

    5. Make suref(xk+1)

  • 8/14/2019 6 OneD Unconstrained Opt

    7/29

    Example 1

    Solve minf(x) = x2xusing function values

    Solution:

    Assume x0= 3;

    f(x) = 32

    3 = 6

    -1

    0

    1

    2

    3

    4

    5

    6

    7

    8

    9

    10

    -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5

    x

    f(x)=

    x

    2-

    x

    x*= 0.5

    x f(x) = x 2- x

    -1 2

    -0.5 0.75

    0 0

    0.5 -0.25

    1 0

    1.5 0.75

    2 2

    2.5 3.75

    3 6

    3.5 8.75

  • 8/14/2019 6 OneD Unconstrained Opt

    8/29

    Narrow down the solution region

    issues in one-dimensional problems

    exhaustive, wide range

    bracketing method

    to narrow down the solution region

    to avoid excessive search within a wide range

    how?

    assume an initial bracket, b0(contain optimum)

    determine reduced bracket, bkat stage k

  • 8/14/2019 6 OneD Unconstrained Opt

    9/29

    Example

    Use bracketing method to solve:

    minf(x) = ( x100 )2

    Solution:

    Consider a sequence of xgiven by

    xk+1= xk+ b 2

    (k-1)

    Assume b= 1 and x1= 0,f(x1)= 10000

    x2= x1+ (1) 2(1-1)= 0 + 1 = 1,f(x2)=9801

    x3= x2+ (1) 2(2-1)= 1 + 2 = 3,f(x3)=9409

  • 8/14/2019 6 OneD Unconstrained Opt

    10/29

    x 0 1 3 7 15 31 63 127 255

    f(x) 104 9801 9409 8649 7225 4761 1369 729 2325

    New bracket is 63 < x< 255

    Continue search using bracketing method

    Bracket bound decreases, bneeds to be reduced

    Assume b= 0.5 and repeat calculations

  • 8/14/2019 6 OneD Unconstrained Opt

    11/29

    Derivative-based method

    Newtons method

    (Quasi-Newton) Secant method

  • 8/14/2019 6 OneD Unconstrained Opt

    12/29

    DERIVATIVE-BASED METHODS

    general procedure for minimisation

    1. Start with an initial value, x0

    2. Calculatef(x0) and derivatives off(x0)

    3. Change x0for next stage x1using the derivatives4. Calculatef(x1) and derivatives off(x1)

    5. Make suref(xk+1)

  • 8/14/2019 6 OneD Unconstrained Opt

    13/29

  • 8/14/2019 6 OneD Unconstrained Opt

    14/29

    NEWTONS METHOD

    necessary condition forf(x)to have an optimum is

    that f(x)= 0

    apply Newtons method to solve f(x)= 0

    )(

    )(

    21xf

    xf

    xx k

    kk

    f(x)f(x)

    f(x)f(x)

  • 8/14/2019 6 OneD Unconstrained Opt

    15/29

    NEWTONS METHOD

    approximatef(x)by quadratic function at xk

    using

    Taylors expansion

    f(x) = f(xk) + f(x

    k) (x-x

    k) + 2f(x

    k) (x-x

    k)2

    stationary point can found by

    df(x)/dx = f(xk) + 2 2f(x

    k) (x-x

    k) = 0

    x = xk( f(x

    k) / 2f(x

    k) )

    Newtons method is equivalent to using quadratic

    approximation for a function & applying necessary

    condition

  • 8/14/2019 6 OneD Unconstrained Opt

    16/29

  • 8/14/2019 6 OneD Unconstrained Opt

    17/29

    Example 2

    Find minimum off(x) =x4x +1, tol =10-7

    Solution:

    f(x) = 4x31, 2f(x)= 12x2

    Assumex0= 3;

    x1= x0(4(3)3-1)/12(32)

    = 3 107/108

    = 2.00926

    x2 = x1(4(2.00926)3-1)/12(2.009262)

    = 2.00926 31.4465/48.4454

    = 1.36015

  • 8/14/2019 6 OneD Unconstrained Opt

    18/29

    Solution

    | x

    k+1

    x

    k

    |

    x* = 0.6299605

  • 8/14/2019 6 OneD Unconstrained Opt

    19/29

    NEWTONS METHOD

    this method approximates the function by a

    quadratic function

    for quadratic functions, the optimum is obtained

    in one iteration both f(x) and 2f(x) have to be calculated

    if f(x)< 0, the method converges slowly

  • 8/14/2019 6 OneD Unconstrained Opt

    20/29

    Derivative-based method

    (Quasi-Newton) Secant method

  • 8/14/2019 6 OneD Unconstrained Opt

    21/29

    SECANT METHOD

    starts out by using two points xp

    and xq

    spanning the interval

    first derivatives should have opposite signs at xp

    andx

    q

    f (xp) = f(xq)

    f (x)is approximated by a straight line

    interval bounds are updated and narrow down at

    each iteration

    the procedure terminates when f(xkp) < tol

  • 8/14/2019 6 OneD Unconstrained Opt

    22/29

    QUASI-NEWTON (SECANT) METHOD

    mxfxx

    xx

    xfxfm

    qqp

    pq

    pq

    )(~

    )()(

    in next iteration replaces xp

    Remarks:

    this method only uses 1storder derivatives

    px~

    f(x)

  • 8/14/2019 6 OneD Unconstrained Opt

    23/29

    Example

    Find minimum off(x) =x4x +1, tol =10-7

    Solution:

    f(x) = 4x3

    1

    Assume xp= 3, xq= 3;

    f(xp) =

    109f(xq) = 107opposite signsacceptable choices

  • 8/14/2019 6 OneD Unconstrained Opt

    24/29

    Example

    f(x) =x4x +1

    The shape off(x)implies

    that a large number of

    iterations are needed.

    f(x)

    f(x) = 4x3-1

  • 8/14/2019 6 OneD Unconstrained Opt

    25/29

    Solution (Cont.)

    ( )p

    f x

  • 8/14/2019 6 OneD Unconstrained Opt

    26/29

    RATE OF CONVERGENCE

    three rates of convergence are used to compare the

    effectiveness of different search methods

    Linear

    Slow in practice

    Secant method has a linear convergence

    Order P

    Fastest in practiceNewton method has order p = 2 convergence

    Superlinear

    fast in practice

    large,10*

    *

    1kcc

    xx

    xx

    k

    k

    large1p0, kcc

    xx

    xx

    p

    k

    k,

    *

    *1

    kcc

    xx

    xx

    kk

    k

    k

    k asandor0 0lim*

    *1

  • 8/14/2019 6 OneD Unconstrained Opt

    27/29

  • 8/14/2019 6 OneD Unconstrained Opt

    28/29

    RECAP

    discuss the concept of one-dimensionalunconstrained optimisation

    solve one-dimensional unconstrained optimisation

    analytically via solving f= 0 & determine x*

    analytically or numerically

    function value-based methods derivative-based methods

    Assignment 3

  • 8/14/2019 6 OneD Unconstrained Opt

    29/29