[Wien] wien.struct <--> abinit.in

[Peter Blaha]

> Is anyone on this list experienced with using the pseudopotential (psp)
> based abinit package?
>
> I found abinit very good for geometry optimisation (unit cell and atom
> positions). In that respect, the results are very close (~0.2% for
> distances / cell dimensions)  to those obtained with full-electron wien2k
> (as long as one uses the same XC scheme; note: I only checked properly
> Al-Cu phases, gga xc).
>
> So, geometry is fine with abinit (using wien2k as my yardstick), BUT the
> energies are sometimes very different from wien results (e.g., difference
> in energy between substitutional Cu in fcc Al and pure fcc Al/Cu).

It is clear that any PP method is faster than WIEN.
It is also clear that one of the big disadvantages of WIEN is that unitcell
optimizations are almost impossible once the symmetry get's below
tetragonal/orthorhombic. If you need this, you have to use a different code.

However, as indicated above, you can never be sure that your PP results are
correct (in particular when transition metals are present). Thus, if you are
serious, you always have to check your PP results.

For "delicate" problems I would never trust a PP calculation alone.


                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------