Scribd
Upload
31 views17 pages

#8 Univariate - Pattern - Search

Uploaded by

Robert Dalton
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
31 views17 pages

#8 Univariate - Pattern - Search

Uploaded by

Robert Dalton
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/ 17

ME 782 Design Optimization

Univariate Pattern Search


Univariate and Pattern search
x2
(-0.72,0.81)
x6
x 4 (-0.67,0.78)
(-0.72, 0.78) x5
x 2 (-0.46,0.64)
x3
(-0.67,0.64)

Powell’s pattern search proposes


a direction for search, after one
cycle of univariate search

x1 x0
x1
(-0.46,0) (0,0)
Univariate and Pattern search
s2
x6
x 4 (-0.67,0.78)
x2
(-0.72, 0.78) x5 s1
x 2 (-0.46,0.64)
(-0.67,0.64) x3

x1 x0
x1
(-0.46,0) (0,0)
Univariate and Pattern search
Minimize
f ( x1 , x2 ) = 5.5 x1 − 2 x2 + 6 x12 + 3x22 + 4 x1 x2

Using univariate and pattern search


Assume x1 varies along the horizontal direction and
x2 varies along the vertical direction

Let the initial guess be given by

x T
0 =  0 0

8/12/2024 D.N. Manik 4


Univariate and Pattern search
Let us first vary the value of x1 along the horizontal direction
ε=0.01
x1T =  x1 +  x2 
x2 f + (x1T ) = f  x1 +  x2  = f  0 = 0.0556  0
f − (x1T ) = f  x1 −  x2  = f  − 0 = −0.0554  0
f − (x1T )  f + (x1T )
 −1
Direction s=  Search direction
of minimum -ε +ε 0
(0,0)
𝐱0
x1
0  −1  − 
x1 = x 0 +  s =   +  0= 0  Univariate search
0    
8/12/2024 D.N. Manik 5
Univariate and Pattern search
 −  Univariate search
x1 =  
0

f ( − , 0 ) = 6 2 − 5.5
 −   −0.4583
df x1 =   =  
= 12 − 5.5 = 0 0
   0 
d
5.5
= = −0.4583
12

8/12/2024 D.N. Manik 6


x2

Univariate search

x1 x0
x1
(-0.46,0) (0,0)
Univariate and Pattern search
Vary the value of x2 along the vertical direction
ε=0.01
xT2 =  x1 x2   
x2
f + (xT2 ) = f  −0.4583   = −1.2984
f − (xT2 ) = f  −0.4583 −  = −1.2218
Direction f + (xT2 )  f − (xT2 )
of minimum
+ε
(-0.4583,0)

𝐱0
𝐱1 x1 Univariate search
-ε

8/12/2024 D.N. Manik 8


Univariate and Pattern search

 −0.4583 0   −0.4583
x 2 = x1 +  s =   + 1  =   
 0     

f ( −0.4583,  ) = 3 2 − 3.83 + 1.2604


df Univariate search
= 6 − 3.83 = 0
d  −0.4583
3.83 x2 =  
= = 0.6383  0.6383 
6
8/12/2024 D.N. Manik 9
x3 = x 2   s
s = x2 − x0
 −0.4583  −0.4583
 −0.46 
s= 
 0.6389   0.01  0.6389 
     0.64  x2
+  −0.4583  −0.4583  −0.4629  ε x 2 (-0.46,0.64)
x =
3  + 0.01   = 
 0.6389   0.6389   0.6453 
-ε
−  −0.4629   −0.4629   −0.4537 
x =
3  − 0.01   = 
 0.6453   0.6453   0.6325 
Direction of minimum

f + (x3+ ) = −2.4965
f − (x3− ) = −2.4731
x1 x0
x1
(-0.46,0) (0,0)
Pattern search
Consider the following equation

f ( x1 , x2 ) = 20 + 3x1 + 7 x2 + 10 x12 + 5x22 + 3x1 x2

Due to a univariate search in the horizontal and vertical directions for searching the
minimum, the coordinates have moved from (-0.15,-0.65) to (-0.0517, -0.6550).
Determine the direction of pattern search and the next coordinate that can be reached
from the pattern search in search of minimum value.
 −0.15  −0.0517 
x0 =  ; x =
 2  −0.6550 
 −0.65   
The direction of the vector joining the above points is given by
 −0.0517   −0.15 0.0983 
s = x2 − x0 =   −  = 
 −0.6550   −0.65  −0.0050
The normalized direction is given by

1 0.0983   0.9987 
s= =
 
0.0984  −0.0050  −0.0508

Taking a small step in the positive and negative directions of

+  −0.0517   0.9987   −0.0507 


x =
2  + 0.001   = 
 −0.6550   −0.0508  −0.6550 

−  −0.0517   0.9987   −0.0527 


x =
2  − 0.001   = 
 −0.6550   −0.0508  −0.6549 
f + (xT2 ) = f  −0.0507 −0.6550 = 17.5333
f − (xT2 ) = f  −0.0527 −0.6549 = 17.5334
f + (xT2 )  f − (xT2 )

 −0.0517   0.9987 
x3 = x 2 +  s =   +  −0.0508
 −0.6550   
 −0.0517 + 0.9987 
= 
 −0.6550 − 0.0508  
The given objective function can be written as follows

f ( x1 +  s1 , x2 +  s2 ) = 20 + 3 ( x1 +  s1 )
+7 ( x2 +  s2 ) + 10 ( x1 +  s1 )
2

+ 5 ( x2 +  s2 ) + 3 ( x1 +  s1 )( x2 +  s2 )
2

Where  is the extent to which the current search can be pursued

from points x1 , x2 in the direction indicated in the coordinate directions s1 , s2


Derivative of the above equation is given by

df ( x1 +  s1 , x2 +  s2 )
= 3s1 + 7 s2 + 20s1 ( x1 +  s1 )
d
+10 s2 ( x2 +  s2 ) + 3s2 ( x1 +  s1 ) + 3s1 ( x2 +  s2 ) = 0
−3s1 − 7 s2 − 20 x1s1 − 10 x2 s2 − 3 x1s2 − 3x2 s1
=
20 s12 + 10 s22 + 6 s1s2

x1 = −0.0517
x2 = −0.6550
s1 = 0.9987
s2 = −0.0508
−3  0.9987 − 7  −0.0508 − 20  −0.0517  0.9987
−10  −0.6550  −0.0508 − 3  −0.0517  −0.0508
−3  −0.6550  0.9987
 =
20  0.9987 2 + 10 ( −0.0508 ) + 6  0.9987  −0.0508
2

= 7.1083  10−4
 −0.0517   0.9987 
x3 = x 2 +  s =   +  
 −0.6550   −0.0508 
 −0.0517 + 0.9987   −0.0517 + 0.9987  7.1083 10 −4 
=  = −4 
 −0.6550 − 0.0508   −0.6550 − 0.0508  7.1083 10 
 −0.0510 
= 
 −0.6550 
Conclusion

8/13/2024 D.N. Manik 17

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy