A problem about how to set the K-point set for optimization

[pengdou]

Dear everyone,

I want to relax the two-probe system like (Au wire)-(long DTB system)-(Au wire) by siesta, I don't know how to set the parameter k-sampling grid,I have a try of

 %block kgrid_Monkhorst_Pack
   1 0 0 0.0
   0 1 0 0.0
   0 0 60 0.5
%endblock kgrid_Monkhorst_Pack
but it is too slow to get the result ,so I want to know which number is enough for the k-sampling grid along z-direction?I think 60 is too big.

Thanks a lot in advance!

[zh]

For this kind of calculation setup, the more reliable solution is to do the test calculation for the convergence of fixed structure with respect to the k-point mesh, and then do the structural optimization.  

You should also show us the lattice vectors of your system. Assuming the lattice vectors of your system look like the followings:
A11   A12   0
B12  B22   0
0      0       C
If  the lengths of first (A11, A12, 0) and second (B12, B22, 0) lattice vectors are much shorter than the third one (0, 0, C), your setup of k-point mesh will be not suitable.

For an equivalent bulk system of two-probe system,  the choice of the division number for the z direction in the k-point sampling is not same as the one for a two-probe system.


By the way, no support for the SIESTA calculations,  you may post your question to here: http://www.icmab.es/siesta/Mailing-List.

[pengdou]

thanks for your reply.