Section: Array Generation and Manipulations
rank function is
y = rank(A,tol)
where tol is the tolerance to use when computing the
rank.  The second form is
y = rank(A)
in which case the tolerance tol is chosen as
tol = max(size(A))*max(s)*eps,
where s is the vector of singular values of A.  The
rank is computed using the singular value decomposition svd.
--> rank([1,3,2;4,5,6]) ans = 2 --> rank([1,2,3;2,4,6]) ans = 1
Here we construct an ill-conditioned matrix, and show the use 
of the tol argument.
--> A = [1,0;0,eps/2]
A = 
    1.0000         0 
         0    0.0000 
--> rank(A)
ans = 
 1 
--> rank(A,eps/8)
ans = 
 2