I have a matrix a(16x3) and a vector b(16x1). b shows how many of the values in the matrix a are valid per row.
a = magic(3)
a =
8 1 6
3 5 7
4 9 2
b = [1;3;2]
b =
1
3
2
What I'm trying to do is setting the invalid values to NaN:
a(:,b+1:end)=NaN
Result is:
a =
8 NaN NaN
3 NaN NaN
4 NaN NaN
But what I would have expected is:
a =
8 NaN NaN
3 5 7
4 9 NaN
What is wrong here?
Perfect setup for bsxfun with @gt to create a logical mask of those elements and then logically index into a to set them as NaNs -
a(bsxfun(@gt,1:size(a,2),b(:))) = NaN
Sample run for a generic m x n case -
a =
2 9 7 2 9
5 7 2 9 5
7 5 1 3 1
8 1 6 2 2
b =
1
4
3
2
a =
2 NaN NaN NaN NaN
5 7 2 9 NaN
7 5 1 NaN NaN
8 1 NaN NaN NaN
Here, the logical mask was -
>> bsxfun(@gt,1:size(a,2),b(:))
ans =
0 1 1 1 1
0 0 0 0 1
0 0 0 1 1
0 0 1 1 1
So, the 1s were used to select the elements that were to be set as NaNs and rest of the elements were not to be touched or changed.
You can also use a(~bsxfun(@le,1:size(a,2),b(:))) = NaN for the same effect.
I could come up with an idea,
a = magic(3);
b = [1;3;2];
the code:
a (repmat(b,[1 , 3]) < repmat(1 : 3 , [3 , 1])) = NaN
gives;
a =
8 NaN NaN
3 5 7
4 9 NaN
For your matrix sizes it would be,
a = randi(9,[16 3]);
b = randi([0 3],[16 1]);
the code:
a (repmat(b,[1 , 3]) < repmat(1 : 3 , [16 , 1])) = NaN
