Issue with PopulationSize in Genetic Algorithm toolbox

There is a project I wanted to use the genetic algorithm tools on and attempted to rig a basic script to test it out. However, I could not run the code because it kept giving me the following error.

Error using gaoptimset>checkfield (line 436)

Invalid value for OPTIONS parameter PopulationSize: must be a positive numeric (not a character vector).

I didn't understand where I had specified PopulationSize as a character vector, so I used optimoptions to change PopulationSize to 50 manually. However it kept throwing me the same error.

When stepped through the program, I entered ga() and found that the default value for PopulationSize is the character vector '50 when numberOfVariables <= 5, else 200' which obviously isn't a numeric value. In any case, this value is supposed to be overridden when optimoptions is used to specify a value other than the default, but it didn't.

Am I doing something wrong or is the genetic algorithm solver broken? Code is as follows.

Further details is that I am doing this on my laptop with a copy of MATLAB R2024b downloaded on it. I have the Global Optimization toolbox installed.

% Testing Genetic Algorithm
clc, clear, close all
rng default % For reproducibility
numberOfVariables = 2;
lb = [-3,-3];
ub = [3,3];

a = 0.1;
b = 0.05; % define constant values
FitnessFunction = @(x) test_fitness(x,a,b);

options = optimoptions("ga",PopulationSize=50);

[x,fval] = ga(FitnessFunction,2,[],[],[],[],lb,ub,[],options);

[Torsten]

Works for me:

% Testing Genetic Algorithm
clc, clear, close all
rng default % For reproducibility
numberOfVariables = 2;
lb = [-3,-3];
ub = [3,3];

a = 0.1;
b = 0.05; % define constant values
FitnessFunction = @(x) test_fitness(x,a,b);

options = optimoptions("ga",PopulationSize=50);

[x,fval] = ga(FitnessFunction,2,[],[],[],[],lb,ub,[],options);

ga stopped because the average change in the fitness value is less than options.FunctionTolerance.

x

x = 1×2

1.0e-04 * 0.0389 -0.1267

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

function y = test_fitness(x,a,b)
y = sum(x.^2);
end

What MATLAB release do you have in use ?

[Victor]

Currently R2024b, but I wouldn't have thought that to cause this level of dysfunction in my code.

[Torsten]

According to this page

https://uk.mathworks.com/help/gads/ga.html

"PopulationSize" is to be prescribed as a positive integer, not as a character.

But if the above code doesn't work on your computer, I cannot tell you what the problem is.

Maybe you have a second function "ga.m" on your computer ? What do you get if you type

which -all ga.m

/MATLAB/toolbox/globaloptim/globaloptim/ga.m

Maybe there is confusion between MATLAB's "ga" and the Genetic Algorithm Toolbox under

https://uk.mathworks.com/matlabcentral/fileexchange/37998-open-genetic-algorithm-toolbox

?

[Victor]

Only the one included with MATLAB appears. This issue mystifies me.

Answer 1

[Star Strider]

The problem is that the arguments to ga are defined by thier positions, and there have to be 10 arguments before the 'options' argument at the end. Adding one more '[],' before hte 'options' argument solves that problem.

All it needs now is the 'test_fitness' function to be defined, and it should work.

Try something like this --

% Testing Genetic Algorithm
clc, clear, close all
rng default % For reproducibility
numberOfVariables = 2;
lb = [-3,-3];
ub = [3,3];

a = 0.1;
b = 0.05; % define constant values
FitnessFunction = @(x) test_fitness(x,a,b);

options = optimoptions("ga",PopulationSize=50);

[x,fval] = ga(FitnessFunction,2,[],[],[],[],lb,ub,[],[],options);

Unrecognized function or variable 'test_fitness'.

Error in solution>@(x)test_fitness(x,a,b) (line 10)
FitnessFunction = @(x) test_fitness(x,a,b);
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in createAnonymousFcn>@(x)fcn(x,FcnArgs{:}) (line 11)
fcn_handle = @(x) fcn(x,FcnArgs{:});
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in makeState (line 58)
firstMemberScore = FitnessFcn(state.Population(initScoreProvided+1,:));
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in galincon (line 24)
state = makeState(GenomeLength,FitnessFcn,Iterate,output.problemtype,options);
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Error in ga (line 420)
[x,fval,exitFlag,output,population,scores] = galincon(FitnessFcn,nvars, ...
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

.

EDIT -- (26 Jan 2026 at 18:01)

Minor text tweaks.

.

[Victor]

I adjusted the code accordingly but I have the exact same error. It still tells me that PopulationSize is a character vector. This issue baffles me. I'm thinking I should just upload my version of Matlab to the lastest just in case that's responsible. It certainly wouldn't hurt.

[Torsten]

The OP's call is possible:

x = ga(fun,nvars,A,b,Aeq,beq,lb,ub,nonlcon,options)

[Star Strider]

It works in R2025b with the latest update (although the note says R2025a), however I've been running ga code for a few years across various versions/releases, and your code should run in them as well (although the position problem may remain in some versions).

I would have to see what 'test_fitness' is in order to determine what the problem may be.

In the interim --

% Testing Genetic Algorithm
clc, clear, close all
rng default % For reproducibility
numberOfVariables = 2;
lb = [-3,-3];
ub = [3,3];

a = 0.1;
b = 0.05; % define constant values
FitnessFunction = @(x) test_fitness(x,a,b);

options = optimoptions("ga",PopulationSize=50);

[x,fval] = ga(FitnessFunction,2,[],[],[],[],lb,ub,[],[],options);

ga stopped because the average change in the fitness value is less than options.FunctionTolerance.

x

x = 1×2

-0.5000 -0.5000

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

fval

fval = 5.7982e-07

function z = test_fitness(x,a,b)
z = norm(a*x+b);
end

.

[Victor]

I've ran my code in MATLAB online and it works there. I still have not determined the cause of the initial problem but this is the best answer this far. i'll be doing this project in MATLAB Online for the foreseeable future.

[Star Strider]

Thank you!