After long and careful consideration we have decided to release a new set of POTCAR files covering the periodic table. This includes GW potentials for almost all elements.
You will find the new potentials on the download-portal and on the VASP.5.X user account of the ftp server.
The potentials have been tested using standard DFT-"benchmark" runs (see the data_base file in the tar files). In most cases the potentials are literally identical to the previous releases, but ALL have been recalculated using a new version of the PAW generation code, to include additional information allowing for calculations with meta-GGA functionals. The present potentials can be used in VASP.4.6, but we strongly recommend to use them only in VASP.5.X, since some compatibility issues might emerge (e.g. LDA+U). Notable updates: Many of the 4d and 5d potentials have been updated to improve f-scattering properties. This decreases the lattice constants slightly for metals, and it might make a substantial difference for strongly ionic TM compounds (including oxides). As you will realize, we now have a large number of potentials for some elements. The rules which potential to use are identical to previous releases (nothing new here). Of course the GW potentials are all "new". As you will see (database) they give virtually identical results to the standard potentials, and it is safe to assume that one can use the GW potentials instead of standard LDA/GGA potential for groundstate calculations, without deteriorating the results. In fact, we believe the GW potentials are generally at least as good as the old standard potentials, but might be much better for some "groundstate" properties including EFG and NMR (well, matter of fact, these are properties related to excitations). However, the GW potentials yield much better scattering properties at high energies well above the Fermi-level (typically up to 10-20 Ry above the vacuum level). We believe that this is important for GW and RPA calculations, although firm proof is still missing. For GW calculations, we also recommend to treat as many electrons as computationally feasible as valence electrons (e.g. use Zn_sv_GW for GW, whereas the Zn or Zn_GW potentials are sufficiently accurate for groundstate calculations). The LDA and PBE sets are identical, except for the lanthanides, for which only PBE potentials are available. Cheers, The VASP team