Genetic Algorithm and Direct Search Toolbox

If i want to optimize a problem using genetic algorithm ,what is the appropriate,

is it use : the type of matlab code in this link <http://codem.group.shef.ac.uk/index.php/ga-toolbox >

or using the optimization tool ( Genetic Algorithm and Direct Search Toolbox) ,or this is related to the nature of the problem ,

any information is helpful for me,I have read lots of article but i am still confused about it.

Answer 1

[Alan Weiss]

I think that it depends on WHY you want to optimize a problem using the genetic algorithm. If it is to get a good answer to a tough optimization problem that has a nonsmooth objective function, then I think that the best thing to do in most cases is to use patternsearch instead.

However, if your goal is to investigate genetic algorithms, not to solve a particular problem the best way, then I am not sure that I can give you good advice, because I only know about MATLAB products. Our ga genetic algorithm solver is flexible, and allows you to customize it in a wide variety of ways; see Genetic Algorithm Options. But perhaps other genetic algorithm solvers have similar flexibility.

Good luck,

Alan Weiss

MATLAB mathematical toolbox documentation

[Johan Johan]

Thank you for your answer the difference has become clear for me, can you help me in my main problem because all my search about this: I have a vector of complex coefficient and i want to optimize it by ga , in ga must be have an objective function ,and i have a vector of initialization (is vector of complex coefficient) ,how can i link between this vector and the objective function ,this case in many other type of optimization ,do you have any idea?

[Alan Weiss]

Generally speaking, it is not clear what it means to optimize a complex-valued function, because complex numbers are not ordered.

If, however, you mean that the control variables are complex, but the objective function is real, then it is easy to split the variables into twice as many real variables, two real variables for each complex number. For more discussion of this (in another context), see Fit a Model to Complex-Valued Data, particularly the later parts of the example.

Alan Weiss

MATLAB mathematical toolbox documentation

[Johan Johan]

The real genetic algorithm is extended to complex cases ,do you have any idea what this means?

[Salah Djerouni]

Alan Weiss ,

Dear docteur ,

I gonna ask a question about the GA optimization ,once the optimization is completed the optimal values selected only the min or max vale from the boundary .what's the probalem in this case .

Please I need help.