How to save "for" loop outputs in cell array?

I generated a code that gives all the 2x2 submatrices from a 10x7 matrix while keeping the order. I want to save each output (submatrix) as a matrix in a cell array to view it later. Here's my code:

Any help is appreciated!

clear 
clc
matrix=rand(10,7)
[r, c]= size(matrix);                                              
for s=1:r-1
    for n=s+1:r-s
for j=1:c
for i=j+1:c
submatrix=matrix([s n],[j i])
end
end
    end
  
end 

Answer 1

[Walter Roberson]

matrix=rand(10,7)

matrix = 10×7

0.6318 0.0444 0.8269 0.0640 0.5899 0.3461 0.0173 0.2847 0.9609 0.5289 0.7458 0.2970 0.9897 0.3715 0.5628 0.4570 0.3468 0.2856 0.6580 0.6526 0.8978 0.0288 0.0271 0.0540 0.7092 0.2769 0.9067 0.6632 0.2362 0.1842 0.1355 0.2507 0.8704 0.9809 0.1556 0.8881 0.2483 0.9374 0.5056 0.3340 0.8535 0.9362 0.1724 0.3315 0.5370 0.7434 0.6314 0.9615 0.2852 0.8577 0.5271 0.5347 0.2187 0.0493 0.2417 0.7776 0.1535 0.7549 0.9788 0.8407 0.5979 0.6664 0.4310 0.5730 0.7781 0.1726 0.3665 0.3435 0.5795 0.9217

submatrix = {};
[r, c]= size(matrix);                                              
for s=1:r-1
    for n=s+1:r-s
        for j=1:c
            for i=j+1:c
                submatrix{end+1}=matrix([s n],[j i]);
            end
        end
    end
end

whos submatrix

  Name           Size             Bytes  Class    Attributes

  submatrix      1x420            57120  cell               

celldisp(submatrix(1:5))

 
ans{1} =
 
    0.6318    0.0444
    0.2847    0.9609

 
 
ans{2} =
 
    0.6318    0.8269
    0.2847    0.5289

 
 
ans{3} =
 
    0.6318    0.0640
    0.2847    0.7458

 
 
ans{4} =
 
    0.6318    0.5899
    0.2847    0.2970

 
 
ans{5} =
 
    0.6318    0.3461
    0.2847    0.9897

 

[HN]

This works! thank you. I was just wondering why did you use "end+1". Could you please explain how it works?

[Walter Roberson]

Indexing at end+1 in an assignment is a common MATLAB idiom for appending data to the end of an array. end is a keyword in this context that converts to the index of the last element of the index. In the case of a single index being used (linear indexing) end converts to the number of elements currently in the array. So assigning to submatrix{end+1} is a more compact way of assigning to submatrix{numel(submatrix)+1} which is assigning to one past the current end of the array.

It saves having to write something like

submatrix = [submatrix, {matrix([s n],[j i])}];

Although the language model is such that assigning one past the end of an array "should" be the same as overwriting the variable with the output of explicitly creating an appended list like the above, it is not necessarily the same internally. The MATLAB memory allocation routines use fixed sized blocks for small arrays, and those fixed sized blocks might have room to extend the array in-place without allocating more memory, so for efficiency purposes assigning to {end+1} has the potential to be faster.

[HN]

This makes sense! Thank you for the explanation.