Reputation: 3
i am struggling with the following problem.
I have 2 matrices, let's say A and B which dimension can change in time and I want to create a new matrix C as follows:
I want to subtract all columns in matrix B from all columns in matrix A. So if for example matrix A was [1 2;5 6;8 5] and matrix B [2 2;2 1;3 5], then matrix C would be [1-2 1-2 2-2 2-2;5-2 5-1 6-2 6-1;8-3 8-5 5-3 5-5].
I have tried this code, but there's a mistake somewhere and I can't find it. Could you please correct the code so it works or provide a better solution to get the right result?
matrixA=[4 5; 4 7;9 6];
matrixB=[3 6 4;5 8 9;4 1 2];
[m,n]=size(matrixA);
[o,p]=size(matrixB);
C = zeros(m,n*p);
for p = 1:n
for q = 1:p
C(:,3*p+q-3) = matrixA(:,p)-matrixB(:,q);
end
end
Upvotes: 0
Views: 98
Reputation: 23675
Your example with column-wise equally sized matrices is pretty easy to deal with:
matrixA = [1 2; 5 6; 8 5];
matrixB = [2 2; 2 1; 3 5];
A = repelem(matrixA,1,2);
B = repmat(matrixB,1,2);
C = A - B
C =
-1 -1 0 0
3 4 4 5
5 3 2 0
But for handling mismatching column sizes too, things get a little bit more complex:
matrixA = [1 2; 5 6; 8 5];
matrixB = [2 2; 2 1; 3 5];
A = repelem(matrixA,1,size(matrixB,2));
B = repmat(matrixB,1,size(matrixA,2));
C = (B - A) .* -1
C =
-1 -1 0 0
3 4 4 5
5 3 2 0
matrixA = [4 5; 4 7; 9 6];
matrixB = [3 6 4; 5 8 9; 4 1 2];
A = repelem(matrixA,1,size(matrixB,2));
B = repmat(matrixB,1,size(matrixA,2));
C = (B - A) .* -1
C =
1 -2 0 2 -1 1
-1 -4 -5 2 -1 -2
5 8 7 2 5 4
Upvotes: 0
Reputation: 112679
I haven't looked into your code, but you can do it simpler and vectorized with permute
, bsxfun
, reshape
:
A = [1 2;5 6;8 5];
B = [2 2;2 1;3 5];
C = reshape(bsxfun(@minus, permute(A, [1 3 2]), B), size(A,1), []);
From R2016b onwards bsxfun
is not necessary:
C = reshape(permute(A, [1 3 2]) - B, size(A,1), []);
Upvotes: 3