global optimization ga with anonymous function

Hello, I'm practicing using global optimization toolbox. I have matrix of "Observed" and "Time", and function to calculate the least sum of squared, "my_fun". Using ga produced some errors... Could you tell me what's wrong? Note, error messages were translated to English from Japanese, so it might not quite correct.

Thank you!

Observed =

    6.0111
    5.9374
    3.4616
    1.1857
    1.3133
    3.2228
    4.8029
    5.9841
    4.8722
    2.3978

>> Time

Time =

     1
     2
     3
     4
     5
     6
     7
     8
     9
    10

function z = my_fun(x,Time,Observed) y=x(:,1).*sin(Time)+x(:,2); z=sum(abs(Observed-y));

f = @(x)my_fun(x,Time,Observed) ;

options = gaoptimset('Vectorized','on'); [x fval] = ga(f,2,[],[],[],[],[],[],[],options);

error: .* The dimension of matrix has to match.

error: my_fun (line 5) y=x(:,1).*sin(Time)+x(:,2);

error: @(x)my_fun(x,Time,Observed)

error: createAnonymousFcn>@(x)fcn(x,FcnArgs{:}) (line 11) fcn_handle = @(x) fcn(x,FcnArgs{:});

error: makeState (line 64) Score = FitnessFcn(state.Population(initScoreProvided+1:end,:));

error: gaunc (line 40) state = makeState(GenomeLength,FitnessFcn,Iterate,output.problemtype,options);

error: ga (line 356) [x,fval,exitFlag,output,population,scores] = gaunc(FitnessFcn,nvars, ...

Cause: Failure in initial user-supplied fitness function evaluation. GA cannot continue.

Answer 1

[Alan Weiss]

The error is clear:

 y=x(:,1).*sin(Time)+x(:,2);

The sizes of the matrices x(:,1) and Time are not the same. You can use the debugger, set a break point in my_fun before line 5, and manually check the sizes of these matrices while the program is running.

Alan Weiss

MATLAB mathematical toolbox documentation