Finding all possible combinations of a matrix

1 view (last 30 days)
Hello Everyone,
I have the following matrix:
mat=[0 0 0 0;1 1 1 1;2 2 2 2;3 3 3 3]
and i want to create a new matrix from it that would have all possible combinations: [ 0 0 0 0; 0 0 0 1; 0 0 0 2; 0 0 0 3; 0 0 1 0; 0 0 2 0; . . . ;3 3 3 3]
How can I do that?
Thanks a lot in advance
Best Regards, Sameh

Accepted Answer

sameh eldessoki
sameh eldessoki on 17 Nov 2015
Hello All,
for "U=4" --> the inputs of the input matrix (0,1,2,3) and "depth=4" --> (number of columns of the output matrix)
I have found a simple answer, which is also generic for other values of "U" and "depth":
combs = dec2base(0:power(U,depth)-1,U) - '0';
Thank you all for your replies and I hope that my answer or yours would be of help to others as well.
Best Regards,
Sameh
  1 Comment
Guillaume
Guillaume on 17 Nov 2015
This is basically the first part of my answer. As stated in my answer, this only works for U < 10.

Sign in to comment.

More Answers (2)

Thorsten
Thorsten on 17 Nov 2015
Edited: Thorsten on 17 Nov 2015
val = 0:3;
X = [];
for i = val, for j= val, for k = val, for l = val, X = [X; [i,j,k,l]]; end, end, end, end
or
val = 0:3;
[w x y z] = ndgrid(val);
X = [z(:) y(:) x(:) w(:)];
  2 Comments
Guillaume
Guillaume on 17 Nov 2015
This is hardly generic to any input mat of an arbitrary size
sameh eldessoki
sameh eldessoki on 17 Nov 2015
@Thorsten the second part of your answer is very useful, but won't have control over the number of columns of the output matrix.

Sign in to comment.


Guillaume
Guillaume on 17 Nov 2015
Edited: Guillaume on 17 Nov 2015
This will work as long as mat has less than 10 rows:
assert(size(mat, 1) < 10, 'mat has too many rows')
rowidx = dec2base(0 : size(mat, 1).^size(mat, 2)-1, size(mat, 1)) - '0' + 1;
colidx = repmat(1:size(mat, 2), size(mat, 1).^size(mat, 2), 1);
result = mat(sub2ind(size(mat), rowidx, colidx))
There's no limit on the number of columns (other than memory / computation time)
edit:
if more than 10 rows, then you can use ndgrid as per Thorsten's answer to generate the rowidx:
rowvals = cell(1, size(mat, 2)); %to receive output of ndgrid
[rowvals{:}] = ndgrid(1:size(mat, 1));
rowidx = reshape(cat(size(mat, 2)+1, rowvals{:}), size(mat, 1).^size(mat, 2), size(mat, 2));
%continue as before
  1 Comment
sameh eldessoki
sameh eldessoki on 17 Nov 2015
@Guillaume I guess your edit does what i want with freedom for the choice of the number of rows and columns. However, it might be a bit complicated. Anyhow, thanks a lot for your effort :)

Sign in to comment.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!