The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hi,

First, let me wish you (and every Abinit's forum users) an happy new year. 
I'm not sure to understand your question very clearly. Yet let me try to answer. I Hope it may help :
1- for any property you want to calculate, the best is to always to double check convergency. So for optical calculation as well as anything else you want to get from Abinit, it is always better, safer, and utterly recommended to check the convergency of your simulation parameters on your final result. Although I can't tell you how much k-point you should use in your particular case, I can give you the following advice: add as much k-point as needed to get a kptrlen that is sensibly larger than the one you have with your initial mech (for instance 20 or  40 bohrs largers); then check if the result remains unchanged (within your tolerance).
2- If the new k-point grid results in noticeable changes of lattice properties the answer is definitely yes since it means that previous calculation were unconverged. Otherwise I don't think it is necessary yet you'd better check.
3- The tutorial says you don't need more bands than those that covers your target grequency/energy domain; therefore I'm not sure to understand this part of your question.

regards

PMA

On Sun, Dec 28, 2008 at 4:42 PM, tokoyama takeshi <tokoyama.takeshi@gmail.com> wrote:

Hello everyone,

Happy 2009 ^_^

I have always followed the complete tutorial of abinit to calculate different parameters for my systems GaN. Now I want to follow toptic_3.in.

1. In the usual rf calculations I have used 8x8x4 grid of k-points, I want to know how much I should increase the number of k-points and what is the reason?
2. Shall I go for optimization for the new k-point grid as well? otherwise what should I put for the lattice parameters?
3. regarding the nband, I usually use nband 20 for the system, then for the optical calculations what parameters I should consider inorder to make sure that the increament is logical.

I hope I can have my answer soon.

Best wishes for the new year.
Takeshi

-- 
Pierre-Matthieu Anglade