MATLAB Genetic Algorithm “Subscripted assignment dimension mismatch” Error

Question

When trying to use the Genetic Algorithm solver in MATLAB, I'm getting the following "Subscripted assignment dimension mismatch" error: Error Message Pastebin

Now, it says the error has to do with the fitness function at the end, but when I test my fitness function separately, it works without errors. I can also link the code for my fitness and constraint functions if that would help.

Thank you very much!

Set a breakpoint in your fitness function and examine the arguments passed in by the solver. This should tell you what's different about the way you're testing it and the way it's actually used. — wakjah
Thanks @wakjah. I did that, but found nothing wrong with the fitness function. The arguments were all what I was expecting. On the other hand, the optimization worked when I commented out lines between 15-41 of my constrint function here: (gist.github.com/shasdemir/18eaeda95833b6618113) It is weird, because the constraint function works when I call it by itself outside of ga. I'm puzzled... — Şükrü Hasdemir
The error sounds like it's coming from something that's being done internally, after your constraint function is called (i.e., c is assigned to some position in a matrix that is not the right size for it). Does the output c not have to be the same length as your set of variables (i.e., 24 elements)? Looks to me like you're returning a 47 element vector... (should 2:timeLength be 2:2:timeLength?) — wakjah
As far as I can tell, the length of c doesn't have to be the same as the number of variables: mathworks.com/help/gads/examples/… -Each element is simply the value of a nonlinear contraint that needs to be <=0... — Şükrü Hasdemir
Besides, when I comment out the part I mentioned, c is still a 46 element vector, but it works... — Şükrü Hasdemir

wakjah wakjah · Accepted Answer · 2013-05-06T11:14:18

I think I see what is happening... Because one of the appendages to CDraft is inside the if, you don't always return the same length vector - i.e., you return a constraint vector the first time, it preallocates a matrix for your constraint output, then the next time round you give it back something that doesn't fit in this matrix, so you get the error.

The clue is in the error stack: at the top of the stack we have

Subscripted assignment dimension mismatch.

Error in C:\Program Files\MATLAB\R2012b\toolbox\globaloptim\globaloptim\private\gaminlppenaltyfcn.p>i_convectorizer (line 135)

Clearly this is not a function you've written, and inspecting your function there's nothing that should cause this error. The end of the error gives another clue

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

The ga function is reporting an error in the bit of the code that handles user-supplied constraints. It is likely that this whole bit of code (the call to the user-supplied function and other associated lines) are all within one try...catch statement that returns this error. Something like this:

try
    c = userConFun(x);
    if isempty(cHistory)
        cHistory = zeros(length(c), 1000); % or whatever
    end
    cHistory(:, currentIterationIndex) = c;
catch err
    error('Failure in user-supplied fitness function blah blah blah');
end

Consequently, it looks like the error came from your function, even though it occurred after your function returned.

MATLAB Genetic Algorithm “Subscripted assignment dimension mismatch” Error

1 Answers