My Community

Dear friends
My setting in the INCAR file is
IBRION=2; ENCUT=500; ISMEAR=0; SIGMA=0.1; ISIF=3; POTIM=0.1; EDIFFG=-0.01;
After relaxaton, to apply the strain in the longitudinal direction, I change the following parameters in the INCAR file ISIF=2; EDIFFG=0.001;

Any suggestion to increase the accuracy of the results?

There are many things you can do to increase the accuracy for various aspects (which also are connected to what you mean with accuracy). However, it depends on what you want to do, so you will need to be more specific. The general requirement is that you make sure that the relevant quantities you're after are converged with respect to the cutoff energy, k-points, semearing width, supercell size etc. If you e.g. want very accurate total energy evaluations then improving EDIFF and setting PREC=Accurate is strongly recommended.

Further, for carbon nano tubes, do not use ISIF = 3 for the relaxation. The spurious interactions through the vacuum regions are very likely to cause slow convergence as well as unwanted reduction of the vacuum region.

If you want more specific comments, you'll need to be more specific as well. There are many considerations to account for. Reading the pdf-talks , the manual as well as published articles in order to get a better picture about accuracy aspects are therefore strongly recommended.

Best regards,
/Dan
<span class='smallblacktext'>[ Edited Wed Sep 08 2010, 10:17AM ]</span>

Your notes are very helpful for me.
I think that with setting ISIF=3, every thing is relaxed and this is appropriate to find the optimum cell shape with minimum internal forces! Please explain me what is the proper setting of ISIF for relaxation?

The isssue is that ISIF=3 will allow for a change of the vacuum layer between the periodic images of the nano tube, which you want to avoid. The best option would instead be to use ISIF=2 and perform calculations for various unit cell lengths along the nano tube and in that way map out the minimum.

Cheers,
/Dan
<span class='smallblacktext'>[ Edited Wed Sep 08 2010, 10:22AM ]</span>