About termination of genetic algorithm

I have set the options of the ga as shown:

options=gaoptimset(options,'Generations',100);

The algorithm terminates after 6 generations. I would like to ask how i will make the algorithm terminate at 100th generation. Thank you in advance.

Answer 1

[Alan Weiss]

ga would normally not exit until at least 50 generations. Therefore, either you have nonlinear constraints, or your options are not what you say they are.

Can you please show us your real options settings (the line you gave assumes that some options have already been created), your real function call, and your real output?

Alan Weiss

MATLAB mathematical toolbox documentation

[dimitris]

The options are:

      options=gaoptimset('PopulationSize',40,'SelectionFcn',@selectionroulette,'CrossoverFraction',0.75,'CrossoverFcn',@crossovertwopoint,'MutationFcn',@mutationadaptfeasible,'Generations',1000,'PlotFcns', {@gaplotbestf,@gaplotscores,@gaplotbestindiv},'Display','iter','UseParallel', 'always', 'Vectorized', 'off');

function [c ceq] = nlinconst(x) 
c=[];
ceq=[];
end

matlabpool open 4

[x,y,exitflag,output]=ga(@(x) obj_opt(x),n,A,b,Aeq,beq,LB,UB,@nlinconst,options)

matlabpool close

And the output is

y=10

exitflag = 1 

output= 

      problemtype: 'nonlinearconstr'

         rngstate: [1x1 struct]

      generations: 5

        funccount: 10440

          message: [1x140 char]
    maxconstraint: 0

[Sean de Wolski]

Whya re you giving it non-linear constraints if you don't use them?

[dimitris]

Because the parameters are passed to the ga function in this particular way. The following line produces error

[x,y,exitflag,output]=ga(@(x) obj_opt(x),n,A,b,Aeq,beq,LB,UB,options)

[Sean de Wolski]

Yes, you need to pass in the empty set:

[x,y,exitflag,output]=ga(@(x) obj_opt(x),n,A,b,Aeq,beq,LB,UB,[],options)

[dimitris]

That was the problem. Now it works fine. Thank you very much sir.

Answer 2

[Sean de Wolski]

What is the exitflag returnmed from ga()?

This will tell us a lot about what is causing it to exit...

doc ga see exitflag

More From doc:

Without nonlinear constraints — Average cumulative change in value of the fitness function over StallGenLimit generations is less than TolFun, and the constraint violation is less than TolCon.

With nonlinear constraints — Magnitude of the complementarity measure (see Definitions) is less than sqrt(TolCon), the subproblem is solved using a tolerance less than TolFun, and the constraint violation is less than TolCon.

So I would recommend starting with those settings.

[dimitris]

Of course, i should have posted it before, exitflag=1. I don't have any nonlinear constraints.

[Sean de Wolski]

see more