[Wien] large system calculation

[Oleg Rubel]

As far as I know, P. Blaha and his group did a calculation for several 
hundreds of atoms, which is closer to your aim. I will make a comment 
based on my experience with ~100 atoms supercell.

*k-points: 1-kpt should be sufficient for relaxation of atomic 
positions; for optics you probably need more kpts.
*RMTs: http://www.wien2k.at/reg_user/faq/rmt.html (aim for lowest 
possible disparity between atoms).
*diagonalization: iterative for sure (on 100 atoms calculation the speed 
up is x10).
*RKMAX: depends on RMTs, but maybe start with 5; think about an 
effective RKMAX for individual atoms if RMTs are too different.
*computing: you will probably need about 0.5-1 TB of RAM for such an 
ambitious project, probably ~300 cores and InfiniBand interconnect 
between compute nodes; do not send *vector files over the network, 
rather use local storage.

I am not sure how the 'optics' package will handle the momentum 
conservation in the case of zine folding, which is inevitable here.

Regards
Oleg

On 20/06/2013 5:02 AM, abdel Mar.. wrote:
> Dear Wien2k users,
>
> i never worked on a large system (~ 350 Atoms), I appeal to your
> *experience* to give me some recommnadations
> to save calculations time, the aim is to simulate the optical properties...:
> method (LAPW / APW ...), Rkmax, Gmax, k points , diagonalization...?
>
>   thank you
>
> Regards
>
>
> _______________________________________________
> 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
>