[Wien] nHOw to accelerate the caculation?

[Stefaan Cottenier]

> > You may also rely on other program packages, like vasp, to do the
> > relaxation. Of course you then need to test whether this relaxed
> > structure is also relaxed in Wien.
>
> Dear Users,
>
> I guess that it is worth waiting for the results of relaxation with WIEN
> on case of surfaces and unit cells of many materials. This is because
> WIEN is 'all electron code' that typically includes all these eg. Zn and
> Ga 3d or In 4d electrons. E.g. pseudopotential calculations on ZnSe with
> Zn+12 would not be much faster then WIEN calculations.
>
> I found out that e.g. in case of GaSe layered crystals relaxation with
> WIEN gives positions very similar to the experiment while relaxation with
> pseudopotential code using Ga+3 (no 3d included) pseudopotential gives
> very different result.
>
> Comments are welcome!

Dear Lukasz,

For some time, we are using the strategy of relaxing large unit cells with
vasp, and then doing one static calculation with wien afterwards (to obtain
hyperfine field information). While this is definitely much faster than
relaxing with wien (often an order of magnitude), it is not clear how
reliable the results are. In many cases the wien-forces in the vasp-relaxed
structure are quite small (< 5 mRy/au), while in others they can be up to 20
or 30 mRy/au. That seems to be related to the elements involved (and
therefore the properties of the pseudopotential): Ni surfaces give very
satisfying wien-forces, for Pd they are much larger (although Ni and Pd are
in the same group).

It would be very useful to know criteria to decide when such a vasp+wien
strategy is likely to work, and when not. Has somebody on this mailing list
looked at this in a more systematic way? It might be that there are no
physical criteria (if there were, I would expect Ni and Pd to perform
similarly), but that it just depends on details of the pseudopotential on
which nobody cared so far. Maybe it would even be possible to generate
pseudopotentials that are optimized with the only goal of delivering
structures that have small forces in a subsequent wien calculation ?

Stefaan