Question

Is there a more robust way for the following problem:

d = 1+(20-1).*rand(365,5);
bthD = 1:5;
I = repmat(bthD',73,1);

for i = 1:length(d);
    q(i) = d(i,I(i));
end  

Here, I are the indices that indicate which column of the matrix are required. q is the outcome where we have each row with the corresponding column number as specified by I. Is there an alternative way to find q?

modified question:

I have data:

d = 1+(20-1).*rand(365,5);

I would like to find the value in each row corresponding to the indices in I:

I = floor(1+(5-1).*rand(365,1));

This can easily be achieved by using a for loop:

for i = 1:length(d);
    q(i) = d(i,I(i));
end    

However, I would like to find an alternative way i.e. one that does not involve loops.

Answer 1

Use linear indexing:

ii = 1:length(d);
linear_index = sub2ind(size(d),ii,I(ii));
q = d(linear_index);

sub2ind converts row (ii) and column (I(ii)) indices to linear indices, which you can use to grab all the desired elements at once.