large dense matrix eigen decomposition using simultaneous iteration

U = eye(n, r);

for i = 1:100

    U = orth(A * U);

end

n is the dimension of A. r is the rank to approximate. 

simple, fast, and can be used for very large dense matrix