Dear Admin,
1) please tell me if I increase the value of NBANDS from the normal settings then other than affecting the speed of convergence will it have any other effect on the results in terms of accuracy?
2) Is it OK to use LMAXMIX = 6 for simple LDA/GGA calculations for POTCAR files treating F electrons as valence?. As far as I understand from the manual it is to be used when doing L(S)DA+U
NBANDS & LMAXMIX
Moderators: Global Moderator, Moderator
-
TAT
- Newbie
- Posts: 39
- Joined: Thu Aug 02, 2007 7:28 pm
- License Nr.: 691 5-1
NBANDS & LMAXMIX
Last edited by TAT on Sun Nov 23, 2008 12:32 am, edited 1 time in total.
-
admin
- Administrator
- Posts: 2921
- Joined: Tue Aug 03, 2004 8:18 am
- License Nr.: 458
NBANDS & LMAXMIX
usually the default value of NBANDS is high enough to give sufficient variational degrees of freedom for the iterative matrix diagonalization, it is a reasonable trade-off beween accuracy and computing time/memory requirements. For very accurate results and for comparing results of parallel runs done on different numbers of nodes, please however make sure that NBANDS are the same (for equal NVAL) for the runs you want to compare, the total energies slightly depend on NBANDS.
2) yes. the l-components for charge density mixing and storing the charge densities are increased from the default value (2, or 3 for f-systems) then, which may increase the accuracy and the speed of convergence.
2) yes. the l-components for charge density mixing and storing the charge densities are increased from the default value (2, or 3 for f-systems) then, which may increase the accuracy and the speed of convergence.
Last edited by admin on Wed Nov 26, 2008 11:06 am, edited 1 time in total.