How do I implement a sliding window technique to initialize a large matrix?

I need to initialize a large 5 Dimensional Matrix. I am unable to initialize the whole matrix at once as I run out of memory. How can I implement a sliding window technique to initialize a small part of the matrix and then slide along one direction to gradually add the remaining elements?

Answer 1

[MathWorks Support Team]

The following example shows how to implement a sliding window approach to initialize a large matrix.

The required matrix is a 5-D matrix which is 1000x6x4x3x6. We can initially initialize a 50x6x4x3x6 matrix and then slide along the 1st dimension to add the remaining 9950x6x4x3x6 elements

function sliding_matrix

clear

countMatrix = zeros(50, 6, 4, 3, 6);  %initialize a small matrix

for t = 1 : 50

      for bm =1 : 6                       % 5 bm

          for bw =1 : 4                  % 4 bw

              for lv = 1 : 3              % 5 lv

                  for ta=1 : 6            % 200 ta

                      countMatrix(t,bm,bw,lv,ta) = ceil(101*rand(1)); %initialze the values

                  end

              end

          end

      end

end

%

% perform necessary operations on the matrix

%

for i = 1:9950    %slde the window 10 times

      %call function to add new data 

      countMatrix = SlideMatrix(countMatrix, i);

      %

      % perform necessary operations on the matrix

      %

end

end

function A = SlideMatrix(A, i)

A(1,:,:,:,:) = [];  %Reduce the the first dimension by 1

for bm = 1:6

      for bw = 1:4

          for lv = 1:3

              for ta = 1:6

                  A(50, bm, bw, lv, ta) = i;  %add a new elements along the first dimension

                  i = i + 1;

              end

          end

      end

end

end