Bidiagonalization question, Golub-Kahan (lanczos)

For the following attempt at bidiagonalization during a test of A = magic(5), I get a matrix of NaN for B, and an entry in U lower right corner as NaN. Please help me understand why, I have followed the algorithm:

function [U,B,V]=bidiagonal_reduction(A)

[m n] = size(A);

U = zeros(m);

V = zeros(n);

for i = 1:n

   x_col = A(i:m,i);

   y_col = norm(x_col);

   sizexc = size(x_col);

   y_c = zeros(sizexc(1,1),1);

   y_c(1) = y_col;

   u = x_col-y_c;

   U(i:m,i) = u;

   usize = size(u);

   A(i:m,i:n) = (eye(usize(1,1))-2*((u*u')/(u'*u)))*A(i:m,i:n);

   x_row = A(i,i+1:n);

   y_row = norm(x_row);

   sizexr = size(x_row);

   y_r = zeros(1,sizexr(1,2));

   y_r(1) = y_row;

   v = x_row-y_r;

   V(i,i+1:n) = v;

   vsize = size(v);

   A(i:m,i+1:n) = A(i:m,i+1:n)*(eye(vsize(1,1)) - 2*((v'*v)/(v*v')));

end

B = U'*A*V;

end