Prev: simulation for wimax
Next: Alzare i tergicristalli MONDEO?!
From: Ajay on 16 Mar 2010 03:52
Hello James,

Many thanks for the reply and sending me the link on my quarry of "Non Linear Constrained Optimization for Mixed Variable Problem". However,

I am using the GA-Tool, which is a graphical interface, of MATLAB (2009b). I have following questions:

1) How MATLAB treat the integer variable different from Continuous Variable when there is no way to specify the integer variable and continuous variable separately? Please advice...

2) Does MATLAB treats the integer variable as binary string for performing the cross over and mutation? or Can MATLAB handles it in this way? Please advice...

3) Does MATLAB considers the 'normalized' value of the constraints for refining the population for next generation? or Which method does it use? Please advice...

4) As suggested, customs functions for the crossover and mutation should be written for the specific problem and integer variables should be rounded-off. However, what if I want to apply different types of cross-over to continuous variables and integer variables?
For example, let's say there are 5 variable (3 continuous and 2 integer) in a Single Objective Constrained problem. I would like to apply single point cross-over for integer variable and two point cross over or a crossover by custom function for continuous variables. How can we handle this by a custom function? Please advice...

In anticipation of an early reply from you, I remain...

Best Regards,
Ajay






James Allison <james.allison(a)mathworks.com> wrote in message <hn32au$r0e$1(a)fred.mathworks.com>...
> The genetic algorithm (ga) in the GADS toolbox can handle MINLPs, but
> you will need to create custom crossover and mutation functions for your
> particular problem:
>
> http://www.mathworks.com/products/global-optimization/index.html
>
> Here is a solution with a demo:
>
> http://www.mathworks.com/support/solutions/en/data/1-10PDHC/?solution=1-10PDHC
>
> -James
>
> Ajay wrote:
> > Is there any code available for "Mixed Variable (Mixed Integer) Non
> > Linear Constrained Optimization" as I am dealing with a optimization
> > problem which has both the continuous and discrete variables in design
> > vector?? Moreover, a single objective function is subjected to multiple
> > constraints.
> >
> > Thanks....
> > Ajay
 | 
Pages: 1
Prev: simulation for wimax
Next: Alzare i tergicristalli MONDEO?!