Page 1 of 1

about NBANDS problem

Posted: Sat Jan 17, 2009 3:29 pm
by chengjun
hi everyone,
In the VASP manual, it said you should increase the NBANDS in the INCAR file when you deal with transition metal.

so, when i calculate the band stucture of Sc doped SrTiO3 (2x2x2supercell with one doping atom at Ti position), I increased the NBANDS=240 value like manual said.

But after calculating, I find the number of bands each k (also NBANDS=256 in OUTCAR) is a little bit larger than NBANDS I set.

Can anyone explain it for me? thanks in advance.

about NBANDS problem

Posted: Sun Jan 18, 2009 5:33 pm
by Danny
The NBANDS has to be a multiple of your CPU-cores(in your case you ran 32 CPU cores (if I calculated correctly))

This, you also have to think of when you want to use the WAVECAR file in the future,
1) set the NBANDS to that of the calculation that generated the WAVECAR
2) only run on a number of cpu-cores such that NBANDS is a multiple of this.

Danny