Prev: fmincon function
Next: how to randomly select pixels from an image
From: hailiang shen on 5 Apr 2010 10:24
Dear All,
I am working with a optimization problem with the number of 193 decision variables, and am using ga to solve it.
My GA parameter setting:
pop size = 193*3
generation=300
other parameters are default

I found:
1 GA premature: after generation 100, the objective value does not reduce significantly,

2 GA convergence: after generation 100, in addition to the prematrure, GA cannot locate the optimal solution.

Any suggestion is greatly appreaciated,

Hailiang
From: Alan Weiss on 7 Apr 2010 13:19
hailiang shen wrote:
> Dear All,
> I am working with a optimization problem with the number of 193 decision
> variables, and am using ga to solve it. My GA parameter setting:
> pop size = 193*3
> generation=300
> other parameters are default
>
> I found:
> 1 GA premature: after generation 100, the objective value does not
> reduce significantly,
> 2 GA convergence: after generation 100, in addition to the prematrure,
> GA cannot locate the optimal solution.
>
> Any suggestion is greatly appreaciated,
> Hailiang

You will probably have better luck with the patternsearch solver.
It is more robust and faster than ga.
See
http://www.mathworks.com/access/helpdesk/help/toolbox/gads/bsa_e70.html#bsbjg7m

Alan Weiss
MATLAB mathematical toolbox documentation
 | 
Pages: 1
Prev: fmincon function
Next: how to randomly select pixels from an image