Accurate DOS/ISMEAR=-5

[sf]

Dear all,

I experienced some troubles when computing DOS with ISMEAR=-5.

Here is an extract of a DOSCAR :

1.598 0.0000E+00 0.0000E+00 0.2000E+03 0.2000E+03
1.744 0.0000E+00 0.0000E+00 0.2000E+03 0.2000E+03
1.890 0.0000E+00 0.0000E+00 0.2020E+03 0.2000E+03
2.036 0.0000E+00 0.0000E+00 0.2020E+03 0.2000E+03
2.182 0.0000E+00 0.0000E+00 0.2040E+03 0.2000E+03
2.328 0.0000E+00 0.0000E+00 0.2040E+03 0.2000E+03

As you see, the number of alpha electrons increases (4th column) while the corresponding DOS (2nd column) remains null...

I carried out the same calculation with ISMEAR=1 and I didn't have this problem. The integration is correct, the error is in the DOS column.

It is a spin polarized calculation with USPP pseudos and vasp 4.6.26

Thank you for your help...
<span class='smallblacktext'>[ Edited ]</span>

[admin]

your DOS energy grid (NEDOS) seems so coarse that a peak (or 2
peaks) including 2 spin-up electrons is missing. Please increase NEDOS

[sf]

You are right. I increased the default value of NEDOS and now it works.

Thank you.