Try using randperm instead of randn. What randperm does is that it generates a random permutation from 1 to n. This can be useful for randomizing matrix indices. Please refer to the documentation of randperm for more details. Here is an example:
B = eye(5)
beta = 0.6;
[rows,cols]=find(B);
B =
1 0 0 0 0
0 1 0 0 0
0 0 1 0 0
0 0 0 1 0
0 0 0 0 1
Use randperm to generate a random permutation from 1 to length(rows), which is in this case 5:
randRowIndices=randperm(length(rows))
randRowIndices =
5 3 2 1 4
It has been assumed that beta is the percentage of rows changed, 60%. Thus in this case, 3 rows in this case will be changed. Note that the function round is applied to make sure that numRowsChanged is an integer. One way to do this is to use vector indexing:
numRowsChanged=round(beta*length(rows));
rowChangeIndices=randRowIndices(1:numRowsChanged)
rowChangeIndices =
5 3 2
Thus, the 5th, 3rd, and 2nd rows will be changed. Use matrix indexing to change the B matrix:
B(rows(rowChangeIndices),cols(rowChangeIndices))=0
B =
1 0 0 0 0
0 0 0 0 0
0 0 0 0 0
0 0 0 1 0
0 0 0 0 0
