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.

46 Hill climbing Strategies<br />

Step 10: (Iteration loop)<br />

Increase k k +1, set i = 1, <strong>and</strong> go to step 1.<br />

Figure 3.6, together with the following table, presents a possible sequence of iteration<br />

points. From the starting point (0), a successful step (1) <strong>and</strong> (3) is taken in each coordinate<br />

direction. Since the end point of this exploratory move is better than the starting point,<br />

it serves as a basis for the rst extrapolation. This leads to (4). It is not checked here<br />

whether or not any improvement over (3) has occurred. At the next exploratory move,<br />

from (4) to (5), the objective function value can only be improved in one coordinate<br />

direction. It is now checked whether the condition (5) is better than that of point (3).<br />

This is the case. The next extrapolation step, to (8), has a changed direction because<br />

of the partial failure of the exploration, but maintains its increased length. Now it will<br />

be assumed that, starting from (8) with the hitherto constant exploratory step length,<br />

no success will be scored in any coordinate direction compared to (8). The comparison<br />

with (5) shows that a reduction in the value of the objective function has nevertheless<br />

occurred. Thus the next extrapolation to (13) remains the same as the previous one with<br />

respect to direction <strong>and</strong> step length. The next exploratory move leads to a point (15),<br />

which although better than (13) is worse than (8). Now there is a return to (8). Only<br />

after the exploration again has no success here, are the step lengths halved in order to<br />

make further progress possible. The fact that at some points in this case the objective<br />

function was tested several times is not typical for n>2.<br />

(0)<br />

(2)<br />

(3)<br />

(1)<br />

(4)<br />

(7)<br />

(5)<br />

(21)<br />

(12)<br />

(22)<br />

(18)<br />

(17)<br />

(9)<br />

(6)<br />

(10)<br />

(19)<br />

(23)<br />

(24)<br />

(8)<br />

(25) (11)<br />

(20)<br />

Starting point<br />

Success<br />

Failure<br />

Extrapolation<br />

Final point<br />

(16)<br />

(15)<br />

(13) (14)<br />

Figure 3.6: Strategy of Hooke <strong>and</strong> Jeeves

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

Saved successfully!

Ooh no, something went wrong!