GA best fitness plot

Hello

I am using MATLAB Global Optimization Toolbox to optimize operation in water systems. GA is the solver. I have 24 variables, lower bounds, upper bounds, and integer variable indices (the 24 variables are integer).

The number of generations is 250, and function tolerance is 1e-1000. Other option are default (including the Elite count in Reproduction).

However, when activate the best fitness plot function and then start the optimization, the best penalty value (not the mean value) in the plot keeps going up and down. As I know, the best penalty value should go down or stay constant as some Elite solutions will be kept for next generations.

I have another question, how can I change the seed (to change the starting point) for every run?

Thanks

Answer 1

[Shashank Prasanna]

If you are referring to the seed of the random generator, you can reset it or provide a different seed as follows, for reproduction:

    >> rng(0);

For your first questions, which plotFcns are you using? Due to the stochastic nature of the solver, mutations are random to come out of local minimas, and may be the reason you see the ups and downs.

[Mohammed]

Thank you for your reply.

The plotFcns is best fitness. I have Elite count in Reproduction, so I should not ups in the plot. I should have only downs.

Regards