Minimise a fitness function using GA toolbox having variable number of variables

I have an error function which I wish to minimise using GA. The error function can be a function of 1,2, 3 or many variables depending on the input given by the user. How should I write a fitness function for the GA toolbox? For example, if we have to minimise y which is a function of x, we create a separate matlab file with the command function y = abc(x) and use it in the GA toolbox. But in my case, the number of variables itself is not fixed. So how should I write such a fitness function ?

Answer 1

[Walter Roberson]

ga() always minimizes a function of one variable -- but the variable is permitted to be a vector.

For example,

F = @(x) (sum( (-x).^(0:length(x)-1) ) - 10000).^2

nvar = input('Number of variables?');

opts = optimoptions('ga', 'PlotFcn', @gaplotbestf, 'MaxGenerations', 1800);
[best_x, fval] = ga(F, nvar, [], [], [], [], [], [], [], opts)

(2 is the most interesting number of variables for the above)

[Rakesh Jain]

Getting following error on running the above code Error using optimoptions (line 114) Invalid solver specified. Provide a solver name or handle (such as 'fmincon' or @fminunc). Type DOC OPTIMOPTIONS for a list of solvers.

Error in abcd (line 3) opts = optimoptions('ga', 'PlotFcn', @gaplotbestf, 'MaxGenerations', 1800);

>>

[Walter Roberson]

You could try

opts = optimoptions(@ga, 'PlotFcn', @gaplotbestf, 'MaxGenerations', 1800);

but I suspect that you might not have the Global Optimization Toolbox installed (or licensed perhaps) and so do not have ga. Try

which ga

to see.