normalization electronic density of states

[ahromero]

I am a little puzzle on how the electronic density of states is obtained.
As far I know, the integral over the electronic density of states should
be the total number of valence electrons in your system, for Fe3O4 we should
obtain 96 electrons (for the cubic structure) but we are getting something of the order or 81...the value changes if the wigner-seitz radius is changed or if the range of the electronic density of staes (EMIN and EMAX) is changed... is there anything I am missing? am I wrong?

thanks

[admin]

the TOTAL DOS is normalized as to :
∫ dos(E) dE = the number of electrons (integration is from the bottom of the valence band to the Fermi level).
the local dos is obtained by weighting the DOS with the local charges which are not uniquely defined in all mixed and PW basis set calculations. Moreover, all charge which is distributed over the 'interstitial volume' (volume that is not covered by the spheres around the atoms) is not included in the partial DOS.
--) changes in RWIGS only affect the local DOS
--) changes in EMIN and EMAX should not change the normalization itself, except if the interval was chosen too small (i.e. it has not captured the energy range of the valence states). This is not the case if EMIN and EMAX are not explicitely set , but may of course be the case if given explicitely in INCAR.
--) errors in the integrated total dos may occur if eg the tetrahedron method is used and the BZ volume is insufficiently filled by tetrahedra (eg, if the Gamma point and points at the BZ boundaries are missing in the k-mesh)