My Community

I am wondering the ISMEAR setting of -5 or 0 in DOS calculation. Which one is a better choice?

I also find there is a warning when I set both ISMEAR=-5 and ALGO=all. The warning is attached.

From the warning, should I set IALGO=53 and ISMEAR=0 when I use ALGO=all?

Dear xiaolei_liu,

As explained on our wiki page about ISMEAR, the "best" choice for calculating DOS is ISMEAR=-5.
However, as the error message states, the tetrahedron method is not variational; one should not use it in relaxtions for example. Here, ISMEAR=0 is a safe choice for most applications and often yields sufficiently accurate DOS. However, this method cannot capture phenomena, like van-Hove singularities. Thus, to obtain even more accurate plots, one can calculate the groundstate with ISMEAR=0 and write WAVECAR and CHGCAR to file. In a subsequent step you can switch to ISMEAR=-5 in combination with ALGO=None (and ICHARG=11) to calculate high quality density of states.