27.06.2013 Views

Evolution and Optimum Seeking

Evolution and Optimum Seeking

Evolution and Optimum Seeking

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Test Problems for the Second Part of the Strategy Comparison 341<br />

Figure A.14: Graphical representation of Problem 2.21 for n =2<br />

F (x) ==2 4 6 8 10=<br />

Powell also ended the search without making any signi cant improvement on the initial<br />

condition. Both the simplex method of Nelder <strong>and</strong> Mead <strong>and</strong> the complex method of Box<br />

also had trouble in the minimum search in their cases the initially constructed simplex<br />

or complex collapsed long before reaching the minimum, again near one of the corners.<br />

Problem 2.22<br />

Objective function:<br />

Minimum:<br />

Start:<br />

F (x) =<br />

nX<br />

i=1<br />

jxij +<br />

nY<br />

i=1<br />

jxij for n =5<br />

x i =0 for i = 1(1)n F (x )=0<br />

x (0)<br />

i =10 for i = 1(1)n F (x (0) ) = 100050<br />

The simplex <strong>and</strong> complex methods did not nd the minimum. As in the previous Problem<br />

2.21, this is due to the sharply pointed corners of the contours. The variable metric<br />

strategy also nally got stuck at one of these corners <strong>and</strong> converged no further. In this<br />

case the discontinuity in the partial derivatives of the objective function at the corners is<br />

to blame for its failure.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!