Direct numerical computation and its application to the higher-order radiative corrections