4 variables problem optimization with Genetic Algorithm

I have a 4 variables nonlinear problem optimization that want to solve it with Genetic Algorithm, my lower bounds are [-100 -100 -3000 -3000] and upper bounds are

[100 100 700 700]. But every time I run the program it gives me the different results for variables and the variables don't converge. How can I fix this problem to get a specific result?

Answer 1

[Alan Weiss]

ga is a stochastic algorithm, and is expected to give different results on different runs. You can get reproducible behavior by setting the random seed. For example, run

rng default

before running ga.

A different question: is ga the right solver for you? See Choose a Solver.

Alan Weiss

MATLAB mathematical toolbox documentation

[Ali Ghalavand]

Thanks Alan,

I used the optimization toolbox and I didn't write the GA code, so where I can use the rng function?

Also,I tried patternsearch too but it needs a good starting point but I don't have any clue of them, cause there is no work on this field before.

I'm low experienced in Matlab.Thanks for helping me.

[Alan Weiss]

You can try patternsearch from several initial points randomly chosen within bounds, as described here:

x0 = lb + rand(size(lb)).*(ub - lb);

Write a loop and see what you get.

I don't understand your question "I used the optimization toolbox and I didn't write the GA code, so where I can use the rng function?" Call the rng function before your ga call. If you canot because someone gave you precompiled code somehow, then I don't know how to help you.

Alan Weiss

MATLAB mathematical toolbox documentation

[Ali Ghalavand]

Thank you Alan,

I wrote GA code yesterday and I used rng function then I saw the variables converged and my problem was solved.