Scribd
Upload
196 views8 pages

Golden Search and Quadratic Estimation

This document summarizes optimization techniques for civil engineering problems, including the golden section search method, point estimation method, and successive quadratic estimation method. The golden section search method iteratively eliminates portions of the search space using the golden ratio. The point estimation method approximates the minimum using function values at selected points. The successive quadratic estimation method fits a quadratic function through three points to estimate the minimum at each iteration. An example applies these methods to minimize a function.

Uploaded by

Himanshu Bohra
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
196 views8 pages

Golden Search and Quadratic Estimation

This document summarizes optimization techniques for civil engineering problems, including the golden section search method, point estimation method, and successive quadratic estimation method. The golden section search method iteratively eliminates portions of the search space using the golden ratio. The point estimation method approximates the minimum using function values at selected points. The successive quadratic estimation method fits a quadratic function through three points to estimate the minimum at each iteration. An example applies these methods to minimize a function.

Uploaded by

Himanshu Bohra
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

CE 771: Optimization in Civil

Engineering
Avijit Maji
Department of Civil Engineering
IIT Bombay
Golden section search method
• Calculating the Fibonacci is additional burden
• At every iteration the proportion of elimination is not same
• In Golden section search method the search space (a, b) is linearly
mapped to unit interval (0, 1)
• Two points at a distance of ! from either end of the search space are
chosen
• In every iteration, (1 − !) portion of the search space is eliminated
• The golden number for ! is 0.618

23rd August 2019 CE 771


Algorithm
• Step 1: For a given lower and upper bounds (a, b), choose a small
number (!) representing the precision of the result. Normalize the
variable x by using the equation w = (x-a)/(b-a). Hence, aw = 0, bw = 1
and Lw = 1. Set k = 1.
• Step 2: Set w1 = aw + 0.618Lw and w2 = bw – 0.618Lw. Compute f(w1) or
f(w2) not evaluated before. Use fundamental region-elimination rule
to eliminate a region. Set new aw and bw and estimate Lw = bw – aw
• Step 3: If |Lw| = !, end the process; Else set k = k + 1 and go to Step 2
• The interval reduces to (0.618)n-1L after nth function evaluation
• It can be used to estimate the accuracy (!)

23rd August 2019 CE 771


Example
• Minimize f(x) = x2 + 54/x in the interval of 0 and 4.
• You may consider ! = 0.2 for this problem.

23rd August 2019 CE 771


Point Estimation Method
• The magnitude of the function values at the chosen points may also
provide some information about the location of the minimum point
• However, derivation of certain functions could be cumbersome
• If a surrogate function with known derivates is used to represent the
objective function, the optimal value can be obtained faster
• Successive quadratic estimation method is one such method

23rd August 2019 CE 771


Successive quadratic estimation method
• Three points can define a quadratic equation
• Function values at three different points (!", !# and !$) can be used to develop
the quadratic equation that approximates the actual objective function
• Quadratic function passing through two points (!" and !#) is presented as:
% ! = '( + '" ! − !" + '# ! − !" ! − !#
, -. /, -0
Where, '( = + !" '" = -. /-0
" , -1 /, -0
'# = -1 /-. -1 /-0
− '"
The minimum point on the quadratic function
!" + !# '"
!̅ = −
2 2'#

23rd August 2019 CE 771


Algorithm
• Step 1: Let !" be an initial point and ∆ be the step size. !$ = !" + ∆
• Step 2: Evaluate ' !" and ' !$
• Step 3: If ' !" > ' !$ , then !) = !" + 2∆ Else, !) = !" − ∆
• Step 4: Determine ,-./ = min ' !" , ' !$ , ' !) and 3-./
representing the ,-./
• Step 5: Calculate !̅ and 5$ . For –ve 5$ replace one of the points with a
random point
• Step 6: Stop the process if ,-./ − ' !̅ and 3-./ − !̅ are sufficiently
small Else save the best point among !" , !$ , !) , !̅ along with two bracketing
points and relabel them as !" < !$ < !) . Then go to Step 4
23rd August 2019 CE 771
Assignment Problem
• Minimize f(x) = x2 + 54/x for an initial value of x1 = 2.7 and ∆ = 0. 5
• Stop the process when "#$% − ' )̅ < 0.1 and *#$% − )̅ < 0.01

23rd August 2019 CE 771

You might also like