A note on the Newton iteration for the algebraic eigenvalue problem