How to define the model in a genetic algorithm when some model parameters are to be searched by the algorithm?

How to define the model in a genetic algorithm when some model parameters are to be searched by the algorithm?

I have an optimization problem, it has some formulas and parameters and 2 of them should be found by the algorithm.

but these two parameters are needed for the definition of the model, too.

How should I define them at first?

[Sam Chak]

I'm just guessing, the "model" that you refer in your problem is about searching for a proper cost function (a.k.a. fitness in G.A.) where it should contain the two parameters to be optimized.

https://en.wikipedia.org/wiki/Mathematical_optimization

Obviously, you cannot simply use the same Quadrative function (which is a convex function) like @Star Strider did.

[Star Strider]

The purpose of my answer is to demonstrate the approach.

There is also nothing wrong with the function I chose to do so. Note that it converges on the correct values, demonstrating the robust nature of the ga function.

[Sam Chak]

Yes true, @Star Strider is very helpful with the example. Hope @Fatemeh Sharafi is able to find the fitness function. If she needs help, she can consider showing all governing equations and describe what she is trying to optimze.

Answer 1

[Star Strider]

To pass extra parameters to the fitness function, and optimise only the parameters-of-interest (here ‘p’), do something like this —

fitness = @(p,a,b) (p(1) - a)^2 + (p(2) - b)^2;         % Parameters To Optimise: 'p';  Extra Parameters: 'a', 'b'
a = 3;
b = 5;
nvars = 2;                                              % Two Elements in 'p'
P = ga(@(p)fitness(p,a,b), nvars)

Optimization terminated: average change in the fitness value less than options.FunctionTolerance.

P = 1×2

2.9935 5.0021

.