From: John Kay Dewhurst <dewhurst@mp...> - 2011-12-01 20:09:12
An additional feature in version 1.4.5 is the variable 'vgtrim'. When
this is .true. then the high G components of the interstitial effective
potential are set to zero. This fixes the instability problem which
occurs for large 'rgkmax'
Also to note for meta-GGA is that the spin-averaged kinetic energy
density, tau, is always used, even for magnetic systems. The
spin-resolved tau should be used in principle, but we cannot make it
consistent with non-collinear magnetism in this case.
Finally please note that the Tran-Blaha functional [Phys. Rev. Lett.
102, 226401 (2009)] does not produce exactly the same gaps as in the
article with Wien2k. We're not sure why this is and are still investigating.
Kay Dewhurst, Sangeeta Sharma, Lars Nordstrom, Francesco Cricchio,
Fredrik Bultmark and Hardy Gross