[Wien] [SPAM?] How to select the appropriate nband in hybrid functionals calculations

[Peter Blaha]

First of all:  Setting   nband to n_occupied + 1   works tecnically, but 
the results will NOT be converged.
This was discussed numerous times. In principle nband should be tested, 
until eg. the gap is converged. Typically you need 10-50% more bands 
than n_occupied.

Second: You get the DOS only for the bands you calculate. with a certain 
nband you get the DOS up to the minimum of the last calculated band 
range of the nband' band.If youdon't calculate high energy states, you 
cant get a DOS. The emin/emax in case.in1 are for lapw1, but not for hf.

You noticed correctly, that the cpu time scales crucially with nband.

You did not mention anything about the k-mesh.
a) The 3x3 supercell should be done by a 3x3 smaller k-mesh !
b) We have the option of a reduced k-mesh for the hf-potential. This 
saves a lot of time and can still be reasonably accurate. Test on the 
small system what reduction still gives you converged results.


Am 08.05.2021 um 22:01 schrieb Yifan Ding:
> Dear Wien2k developers，
> 
> When YS-PBE0 is used to calculate the DOS of pristine h-BN, the problem 
> I encounter is that I don't know how to select the appropriate nband to 
> obtain the DOS of energy window -20 eV ~ 20 eV.
> 
> At first, I calculated the DOS of pristine h-BN using PBE96. The emin 
> and emax in case.in1 file are -9.0 and 2.5, respectively. I can easily 
> get the DOS of -20 eV ~ 20 eV that I need. Then I executed init_hf_lapw, 
> and I set nband in case.inhf file to 14 ( the number of occupied bands 
> plus one) according to the following contents in case.scf
> 
> /:BAN00001:   1   -1.024214   -0.744172  2.00000000/
> /:BAN00002:   2   -1.013037   -0.743874  2.00000000/
> /:BAN00003:   3   -0.370466   -0.032709  2.00000000/
> /:BAN00004:   4   -0.368369    0.033611  2.00000000/
> /:BAN00005:   5   -0.240838    0.187128  2.00000000/
> /:BAN00006:   6   -0.240784    0.187128  2.00000000/
> /:BAN00007:   7   -0.006555    0.283584  2.00000000/
> /:BAN00008:   8    0.032524    0.287635  2.00000000/
> /:BAN00009:   9    0.598642    0.890756  0.00000000/
> /:BAN00010:  10    0.625260    1.041091  0.00000000/
> /:BAN00011:  11    0.874243    1.203608  0.00000000/
> /:BAN00012:  12    1.014159    1.265192  0.00000000/
> /:BAN00013:  13    1.122810    1.303539  0.00000000/
> /        Energy to separate low and high energystates:   -0.42047/
> /
> /
> /:NOE  : NUMBER OF ELECTRONS          =  16.000/
> /
> /
> /:FER  : F E R M I - ENERGY(TETRAH.M.)=   0.2876346466/
> /:GMA  : POTENTIAL AND CHARGE CUT-OFF  12.00 Ry**.5/
> 
> Meanwhile, other contents of the case.inhf file are set to YS-PBE0 for 
> calculation. When I executed run_lapw -hf and finished the DOS 
> calculation, I found that the energy window maximum of DOS can only 
> reach 13.42366 eV, but emin and emax in the case.in1 file are still -9.0 
> and 2.5. I noticed that first three lines of case.dos1ev in PBE96 
> calculations show:
> 
> /#  BAND   28 /
> /#EF=   0.00000     NDOS= 3     NENRG= 1750    Gaussian bradening: 0.01600/
> /# ENERGY  total-DOS        1:total        2:total /
> 
> Therefore, I changed the nband of case.inhf file in YS-PBE0 calculation 
> to 28, and finally successfully calculated the density of States of -20 
> eV ~ 20 eV that I needed. Because the structure of h-BN is very simple, 
> when I increase nband from 14 to 28, the increase of calculation time is 
> not particularly obvious.
> 
> When I need to calculate the DOS of h-BN with defects, I set 3*3*1 
> supercell. At this time, nband obtained from case.scf and nband obtained 
> from case.dos1ev are 79 and 910, respectively. I guess that when 79 is 
> used, the complete density of States of -20 eV ~ 20 eV cannot be 
> obtained, and when 910 is used, the calculation time cannot be 
> tolerated. Is there a better solution to this situation?
> 
> Any comments would be highly appreciated. Thanks in advance!
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
> 

-- 
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300             FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at    WIEN2k: http://www.wien2k.at
WWW:   http://www.imc.tuwien.ac.at
-------------------------------------------------------------------------