Concatenation of 3D Matrices

Greetings!

I would like to find a solution to a bottleneck that I've encountered. So I have a 3D array of the 3x3x644 size, consisting of 644 uniques 3x3 sub-matrices. The whole array represents 644 seconds and state information for each timestep is represented by those 3x3 sub-matrices. My goal is to create such rouitine that would allow me create an artificial delay for a consideration, meaning that I'd like to test how system system behaves with a state information delay of 2, 4, etc seconds. This implies that even though I have current system state information for every second, I would need to create a dataset where system state information is updated every 2,4,etc seconds. For instance, if to consider a delay of 2 seconds, I need every two matrices of that 3x3x644 array be the same, or in a case of 4 seconds - every four matrices shall be duplicated.

Here is partial routine I've came up with, but I am struggling with the last step. I am pretty sure that there is a way easier and elegant solution to my problem, but I cannot figure it out. Thank you for any suggestions!

r=2; %'update rate'
n=644; %total number of entries
A=rand(3,3,n); %creating 3x3x644 3D-matrix
A1=A(:,:,1:r:end); %getting rid of every n-th matrix (where every r-th matrix is the one to be updated) with 'previous' matrix, resulting in a 3x3x322 matrix
A2=repmat(A1,1,2);  %duplicating 3x3 matrix in all of 322 entries, resulting in 3x6x322 matrix

for i=1:size(A2,3) %i=1:332
        B=A2(:,:,i); %localizing 3x6 i-th matrix
        B1=B(:,1:3); %splitting 3x6 i-th matrix into two separate 3x3 clone matrices
        B2=B(:,4:end);
        %C1(:,:,i)=cat(3,B1,B2); %concatinating splitted 3x3 clone matrices into 3x3x2 3d matrix
        
        %Here I need some routine that would stack 3x3x2 matrices C1
        %onto each other resulting in matrix of original dimentionality of
        %3x3x644
end

Answer 1

[Arthur Roué]

This should do the trick

r = 2; %'update rate'
n = 644; %total number of entries
A = rand(3,3,n); %creating 3x3x644 3D-matrix
for idx = 1:r:n
    A(:,:,idx+(0:r-1)) = repmat(A(:,:,idx), 1, 1, r);
end

[h_chep]

Thank you, Arthur!

Very elegant and quick solution!

[Arthur Roué]

You're welcome ! ;)