Hi Guibin, As far as I understand, the imaginary part of epsilon is closely related to the joint density of the states of the valence and conduction band electrons. If LAPW and pseudo really give the same band structure, they should give the same jdos, even the actual wave functions are different.
It is a bit unthinkable to me that the previous pseudopotential can give the correct band structure, while its jdos is so off the mark. Something must be wrong. Have you compared the DOS calculated by LAPW and pseudo, or a dispersion along a less symmetric direction? Are they really the same? These questions may not really help you much... I am just curious. Best wishes. Tao
