Do 2 matrices in PSO Cost function must have the same amount of column?

I am trying to run a cost function for my PSO, which met the error: "Failure in initial objective function evaluation. PARTICLESWARM cannot continue" and "Error using * .Incorrect dimensions for matrix multiplication. Check that the number of columns in the first matrix matches the number of rows in the second matrix. To operate on each element of the matrix individually, use TIMES (.*) for elementwise multiplication."

I found out this section is causing trouble:

filepath_for_T_matrix = append("C:\Users\Admin\ur3+cobotops\dataset\experimental_setup\IFFCM\IF_FCM_Learning\taus.csv");
taus = readtable(filepath_for_T_matrix); %Weight Matrix
taus = table2array(taus);

[row,col] = find(taus);%Indexes(rows and colums) for non-zeros %line5

candidate_weights = x(1,2:width(x));

W = zeros(height(taus),width(taus)); %line 6
for i=1:height(row)
   W(row(i),col(i)) = candidate_weights(1,i);
end

W=transpose(W);%The weight matrix is the transpose of taus %line 11

filepath_for_training_X = append("C:\Users\Admin\ur3+cobotops\dataset\k-fold cross validation datasets\gl\",num2str(fold),"\Training Dataset","\Scaled_X_train_iter_",num2str(fold),".csv");
filepath_for_training_y = append("C:\Users\Admin\ur3+cobotops\dataset\k-fold cross validation datasets\gl\",num2str(fold),"\Training Dataset","\y_train_iter_",num2str(fold),".csv");

X_train = readtable(filepath_for_training_X);
X_train = table2array(X_train);
    
y_train = readtable(filepath_for_training_y);
y_train = table2array(y_train);
    
Simulations_for_all_observations = cell(height(X_train),1);

number_of_input_concepts = width(X_train);
number_of_Activation_Decision_Concepts = width(y_train); % After this calculation will start

Taus is a 7x7 matrix so W is a 7x7 zero matrix, but the combined width of X_train and y_train is 20. This is why we have error using *, and leads to PSO can't go on. I know this is the problem because after making W a 20x20 zero matrix no more error occurs, my training is faster but the optimized values stay still. I have just tried 11 iterations as my laptop is weak. I would like to thank you all in advance for clearing my question or giving me advices. The code snippet was taken from Marios Tyrovolas Github: https://github.com/marios-tyrovolas

[Torsten]

You found the reason for the error message - so what is the question you think we can answer with the above piece of code ?

[Star Strider]

You did nothing wrong. The spam fiilter is quite good although not perfect. I removed the ‘spam’ flag.

The error message is clear.

If you are multiplying two matrices using mtimes, *. the matrices have to be conformable under matrix multiplication, such that the column size (number of columns) of the first matrix has to be equal to the row size (number of rows) of the second matrix. The product has the row size of the first matrix and the column size of the second matrix.

If you are using times, .*, (known as ‘element-wise multiplication’), the row and column sizes have to be respectively equal.

See the documenatation that I linked to here on each function for relevant details.

.

[Nghi]

Hello, thank you. I just wanted clarifications that we always need the same matrices column number to do PSO for all applications

[Stephen23]

"I just wanted clarifications that we always need the same matrices column number to do PSO for all applications"

The particleswarm documentation states that "...the objective function to accept a row vector of length nvars and return a scalar value". What matrices your objective function defines inside and how many columns they might have is your business.

The error you get has nothing (directly) to do with the particleswarm function, but is due to an attempt to MTIMES arrays which have incompatible sizes.

[Star Strider]

@Nghi — With respect to your earlier Comment, if you are comparing a data matrix (or vector) with a calculated matrix (or vector) in order to compare them (probably subttracting them and then calculating the norm to determine how close they are), those matrices must have the same row and column dimensions.