Construct a 3D matrix with linear index

Question

I want to construct a 3D matrix of size = 80 * 80 * 2 based on a set of data:

 1        4532   1257.0
 1        4556   1257.0
 1        4622   257.0
 1        4633   257.0
 2        7723   31.0
 2        8024   31.0
 2        8099   31.0
 2        9800   31.0
 2        8524   34.0
 2        8525   34.0
 2        8700   734.0
 2        8701   734.0

The first column denotes the slice of matrix.
The second column denotes the linear index of the matrix.
The third column denotes the values of the elements.

What I'm doing now is: I first obtain two 80 * 80 2D matrices A and B and then concatenate them using cat(3, A, B):

Denote the above data be M.

for i = 1 : size(M,1)
    if (M(:,1)==1)
        [r c]=ind2sub(M(:,2));
        A = accumarray([r c], M(:,3));
    elseif (M(:,1)==2)
        [r c]=ind2sub(M(:,2));
        B = accumarray([r c], M(:,3));
    end
end

cat(3, A, B)

I am curious if there is any solutions that can build the 80*80*2 matrix merely by the linear index (the second column of my data) or any other simpler solution works for the purpose.

I appreciate for your help.

subtracting 6400 for the second column if 2 appears at the first column seems to be what you mean? — nam

Stewie Griffin Stewie Griffin · Accepted Answer · 2015-08-05T10:22:29

So, I'm assuming your example data is incorrect, and that all values in column 2 are less than n*n, where nxn is the size of the matrix (80x80 in your case).

If that's the case, the following two lines should do the trick.

out = zeros(n,n,2);
out((M(:,1)-1).*n^2+M(:,2)) = M(:,3)

If the second column contains values up to 2*n*n, and thus are the linear indices, then:

out = zeros(n,n,2);
out(M(:,2)) = M(:,3)

Construct a 3D matrix with linear index

1 Answers