[Wien] Q:How many k-points are appropriate in a calculation on supercell

[Stefaan Cottenier]

>Usually, I chose 1000-3000 k-points in a unit cell with RminKmax=7.00.
>
>  
>
Such a statement is dangerous. k-mesh size and basis set size must
always be TESTED. There are no safe default values. With a choice as you
state it, you might sometimes be using unneccessary computer resources,
or (worse) sometimes be producing uncorrect results.

>It takes too much time,if I still keep 3000 k-points in the SCF.
>
>How many k-points are appropriate when dealing with a 2*2*2 supercell?
>Is it feasible to choose 300 k-points or decrease the value of RminKmax to 
>6.00?
>  
>
First test the k-mesh (and RKmax) for the small cell. A 2x2x2 cell is 8
times larger, hence the IBZ is 8 times smaller. A k-mesh of the same
quality as for you small cell contains 8 times *less* points in the IBZ
for your supercell.

Stefaan


Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm