[Wien] Large scale problem

[Peter Blaha]

Divide-conquer methods need a "localized" basis set (like a short range 
LCAO). Plane wave methods cannot do this.

One can probably do 3000 atoms/cell on a machine with 1000 cores.
It depends on the type of atoms/systems.

However, you should think twice about your specific system and which 
method to use.

WIEN2k is a very general program, which can treat all atoms from H - U, 
insulators/metals; non-magnetic/magnetic, ..... with highest precision.

If your system contains only C,H,O,N,Si, ..., is an insulator and 
non-magnetic, other methods which are tailored to such systems and 
sacrify a little the accuracy, can be way more efficient (orders of 
magnitude !).

On 03/13/2013 10:08 AM, César de la Fuente wrote:
> Dear all,
>
> I was wondering if anybody knows a method to implement a large scale problem
> (about 3000-10000 atoms or more) within wien2k (for being executed in a
> reasonable of time)? It is known that the executing time scales roughly as
> N^3, where N is the number of atoms.
>
> To be more specific...
>
> Can Wien2k implement methods as for example LS3DF (Linear scale
> three-dimensional fragment) method which it is based on the
> "divide-and-conquer" scheme?
>
> Thanks in advance for the comments...
>
> Dr. César de la Fuente.
> Depto. de Física de la Materia Condensada.
> Edificio Torres-Quevedo
> EINA-Universidad de Zaragoza.
> C/María de Luna 3, 50018-Zaragoza (SPAIN).
>
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>

-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300             FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at    WWW: 
http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------