various questions on VASP

[luke419]

I have questions on VASP.

1. If total number of unpair spin is 6 (hence, spin is 1/2*6), do I only have to set "nspinupdown=6" for running of it, remaining other keywords the same?

2. Can I see electrostatic potential around a molecule using VASP?
If it is available, what should I do for it in detail?

3. For geometric optimization job (not single point energy), what values should I use for ENCUT, ENAUG generally?
Do the default values of them work well?
If not, should they change according to the system for the reliable geometric optimization?
Then, how can I determine them?

4. In principle, we have to check for the convergence of energy, varying ENCUT and ENAUG and varying K-points.
However, is there any empirical recommendable values of them for single point energy such as K-points of 3X3X3 for cubic system etc.

5. Even if we'd like to perform geometric optimization and need not to investigate DOS, should DOS analysis (keywords of DOS are ISMEAR and SIGMA) be done automatically during geometric optimazation job?

6. Incidentally asking, for the analysis of DOS or band structure, what should I do in detail?
Is there a freeware for it?


Thanks a lot for the reading and I hope for your good answers.
Cheers,


Luke

[admin]

1) yes, please set ISPIN=2 and the difference between spin up and down electrons as NUPDOWN (not nspinupdown) (6 in your case) if you want to keep the spin difference fixed to 6. otherwise just set ISPIN=2 and the system will converge to its magnetic ground state.
2) pelase set LVTOT=.True. in INCAR, you will obtain a file LOCPOT showing the potential (default: without xc-contributions) distribution in the cell (the output format of this file is the same as in CHGCAR). YOu can then eg use the tool vtotav.f to draw line scans of the planar averages of the potential (along x,y, and z) or use any other graphics tool which you use to visualize the charge densities.
3) in general, this is not necessary if you do not optimize the cell parameters using force tensors (please check the ISIF tag in the manual). If you use the force tensors for optimization, please set
PREC=Accurate and ENCUT to approximately 1.3*max(ENMAX) of the atoms you use to put up your POTCAR file. ENAUG is adjusted automatically if you chooese PREC=accurate.
4) for ENCUT and ENAUG the settings in (3) are in general sufficient. The k-points have to be tested however, because the set you will need depends on the electronic structure of your system (especially at the Fermi edge). For metals, you will typically need more than 3x3x3 points in the simple cubic cell (Cu eg requires about 11x11x11), for insulators and wide band semiconductors 3x3x3 might be a sufficient choice. Please do spend the time to converge the k-set for the smallest supercell possible for your system, and rescale to smaller k-sets later if you use supercells for your investigations.
5) the choice of ISMEAR and SIGMA again depends on the electronic structure, in particular please do not use ISMEAR=-5 for the geometry optimization of metallic systems, because the occupancies of the levels are not variational in the Blöchl method. The final, accurate DOS run (denser k-mesh, ISMEAR=-5) only has to be done as a single-point calculation after you have achieved geometry convergence.
6) I would recommend to use p4vasp (please have a look at Orest Dubay's p4vasp forum, you can access it via our homepage), especially for plotting the band structures. the DOS can of course be plotted with any graphics tool that can do simple x-y plots (xmgrace, xfig, gnuplot,... all of them are free).

[ Edited Wed Nov 26 2008, 12:36PM ]