Replace specific/diagonal elements in Matrix using editor, scripting

Hello,

Looking a simple way to replace diagonal values in vector to 0. The matrix is a variance-covariance matrix, however in system will be assuming covariance are

zero, and are independent to each. Have the following matrix:

A = [ 2.5 0.25; 0.25 2.4]
B = flip(A) %Flipped to get in correct orientation to select diagonals 0.25, 0.25
%Not sure if it is possible to select mirrored diagonal in original matrix A

Covariance values are 0.25, which would like to set to zero. However these value could change, therefore if had method to replace these diagonal values with respect to m-row, n-column position, or other method?

If can be selected from original matrix A, would be ideal to avoid flipping.

Thanks in advance

kind rgds

[Max Murphy]

Will your system always be a bivariate distribution or will A ever be larger than 2x2?

[Vincent Hodnett]

Yes, for this instance. Shall only be 2x2 matrix.Thks

Answer 1

[Max Murphy]

% If this is to be called iteratively
% Example while loop
iteration = 0;
while (iteration < numel(data)) % Some conditional to stay in your loop
    if ~iteration
        % Initially
        % A = [2.5 0; 0 2.4]; % Can you just initialize it with zeros?
        
        % Alternatively, suppose it was imported elsewhere:
        A = [ 2.5 0.25; ... % Top right = A(1,2);
              0.25 2.4];    % Bottom left = A(2,1);
        A = diag(diag(A)); % Keep only diagonal elements
    else
        A(1,2) = cov_ab; % Assume you get this from elsewhere
        A(2,1) = cov_ba;
    end 
    ...
    % Get covariance values 
    doc cov % May be helpful
    ...
    iteration = iteration + 1;
end

[Vincent Hodnett]

Hello Max,

Thanks, for advice. Yes, the A = diag(diag(A)) will solve issue. To keep only diagonal element

A = [ 2.5 0.25; 0.25 0.78];

A = diag(diag(A))

% This will give the matrix that is needed, 0.25 and 0.25 are equal to 0 and we have kept only %the variance, diagonal. :) :)

Thanks,

kind regards Vincent