Page 1 of 1

integrated DOS at Fermi level does not match NELECT

Posted: Mon Aug 07, 2017 11:29 am
by byin
Dear admin,

I just found that the integrated DOS at Fermi level does not match NELECT.

For example, in the calculation of hcp Mg, 2 atoms, 4 electrons.
In vasp.4.6, the integrated DOS at Fermi level is 4.
In vasp.5.3.5, the integrated DOS at Fermi level is 0.08, although the shape of DOS is similar in both results, indicating this may be just an error in the DOSCAR output.

And the missing scaling parameter 50 (0.08*50=4) seems to be universal in different materials. I found this 50 again in bcc Li.

Please explain this in details. Or fix the bug ASAP. Thanks!

Best,
Binglun Yin

Re: integrated DOS at Fermi level does not match NELECT

Posted: Fri Sep 08, 2017 1:53 pm
by byin
Sorry for the misleading info. This problem seems not to be related to the version, 5.3 or 4.6.

This happens when I use the DOSCAR directly generated from an ionic relaxation calculation. When I use the DOSCAR from a static calculation (NSW=0), the total number of electrons is correct.

So why this problem happens in ionic relaxation? Thanks in advance!

Binglun

Re: integrated DOS at Fermi level does not match NELECT

Posted: Mon Mar 05, 2018 2:55 pm
by admin
It seems you have figured it out by yourself:
The resulting WAVECAR, DOSCAR and CHGCAR files after an ionic relaxtion are not reliable as described in the VASP manual.
The purpose of an ionic relaxtion is to minimize the forces acting on the ions in the unit cell, not to produce high quality DOS plots.
Please, read the VASP manual for more details on how to produce high quality density of states.