
Adding a new dimension to DFT calculations of solids ...
Known limitations and bugs of WIEN2k
©2001 by P. Blaha, K. Schwarz and J. Luitz
This section will be updated regularly as problems get fixed (or new appear)
 Centered monoclinic lattices: Only settings with a monoclinic angle
gamma (Bsetting, or in WIENterms a CXZ lattice) are allowed. You must
transform your setting to this one. You can use the Bilbao crystallographic
server and its SETSTRU (structure utilities) program. For instance for SG#15
we need one of the "B" settings (often B2/b11, but this depends on the initial setting
of your structure, but you should be able to find the right one by noting
which yields as monoclinic angle gamma);
for SG#12 we need (B112/m) or (B2/m11), .... After transformation always
check if the multiplicities are still correct and after "x nn" if the
distances are correct with the ones given in the experimental papers.
 In spinorbit calculations forces are not yet implemented . However,
it is possible to switchoff SO for light elements (eg. oxygens) and keep SO
only for the heavy ones. In this case, the forces on the atoms without SO are
still ok and can be used for structural optimization.
 In spinorbit calculations with relativistic local orbitals (RLO's) the
EFG (and also the dipolar Hyperfinefields) are wrong. Use RLO's only for
energies. Also optics does not work with RLO's.
 The optic package does not work with relativistic local orbitals, but multiple LOs (hight Energy LOs) can be used.
 Plotting wavefunctions using lapw7 does work with (L)APW+lo, but only with ONE lo.
It does not work with relativistic local orbitals.
 The usage of HDLOs (highderivative local orbital) is only possible for
programs of the scf cycle, but not for hybridDFT (hf) or DFT+U (orb).
Also the optic package does not work with HDLOs. Use a HELO (high energy
local orbital) instead.
 The stress tensor works only in a nonrelativistic setting (switched
automatically when using str 0.1). Spinpolarization only with LDA,
nonspinpolarized calculations with LDA and any GGA, but not mGGAs.
 The irrep package does not work for kpoints at the surface of a BZ for nonsymmorphic spacegroups, since the corresponding (more complicated) group theory is not implemented. Eventually you may shift the kpoints at the surface slightly away to break this special symmetry.
 Possible problems with stack size on a Mac and W2kutils: A rather
unclear situation happens sometimes ?? on Mac systems. In case of problems
you may try to comment the call to W2kutils. Make sure you set stacksizes to
unlimeted yourself.
 Case sensitive file system on a Mac: Since WIEN2k uses files like
case.inm and case.inM, you should convert your file system to be case
sensitive.
©2001 by P. Blaha and K. Schwarz
