It really doesn't help if you invent your own syntax for your examples instead of using valid matlab syntax. It also doesn't help if you write .5 in you example then say that the value is 5 (10 times more).
Assuming I understood correctly:
A = [1, 4, 2, 5, 4] %note that this is a ROW vector not a column, but that doesn't matter anyway
B = [3 5 6 3 1
4 2 7 9 2
3 3 4 8 9
5 1 4 5 6
2 3 6 8 3]; %I'm assuming all these . in your example are not meant to be there. The values don't matter anyway
%option 1:
result = B(sub2ind(size(B), A, A))
%option 2:
diagB = diag(B).';
result = diagB(A)
