I am trying to figure out why the DOSCAR file during structure relation is useless, for the last step of structure, the atoms' positions are almost fixed, is that the same as a static calculation? But according to the VASP wiki, after structure relaxation, I have to do the static calculation one more time. Why is that?

For relaxations, the DOSCAR is usually useless. If you want to get an accurate DOS for the final configuration, first copy CONTCAR to POSCAR and continue with one static (ISTART=1; NSW=0) calculation.


2 Answers 2


The DOSCAR requires another run after a relaxation to be useful because the DOSCAR gives an average of the DOS across the ionic steps.

In the first paragraph of the DOSCAR VASP wiki page:

For dynamic simulations and relaxations, an averaged DOS and an averaged integrated DOS is written to the file. For a description of how the averaging is done see the tags NBLOCK, KBLOCK, EMIN, EMAX and NEDOS

The average is contributed to every NBLOCK steps and written every NBLOCK*KBLOCK steps. By default, the average is contributed to by each step (NBLOCK=1) and written at the last ionic step (KBLOCK=NSW).

Maybe the DOS could be useful if NBLOCK=KBLOCK=1, where the average would be written and reset after each ionic step, assuming your Brillouin zone integration method gave an acceptable DOS and acceptable forces, as mentioned in the answer by Chi Kou.


As far as my knowledge, you need to run SCF with ISMEAR = -5 to just get very accurate energies, while for relaxations you need to use ISMEAR = 0 (for semiconductors and insulators). Also, the DOSCAR file is used for the density of states calculations, and the tetrahedron smearing (ISMEAR = -5) must be toggled on to get the correct Fermi energy.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .