[Wien] force difference between LAPW and PAW method

[Pavel Ondračka]

Laurence Marks píše v Pá 29. 01. 2016 v 04:22 -0600:
> Did you check for lattice parameter changes? A relatively small
> change for a single unit cell can translate to something larger.
> Unfortunately you need a reference compound, which is not so simple
> in your case. Perhaps test SiO2, TiO2 & a few simple similar
> compounds with the PAW & Wien2k.

Yeah, some difference in cell size or positions was one of my suspects.
I usually use VESTA to read the structure file from PAW software, then
output to cif and use cif2struct to convert to struct and I was
suspecting that maybe somewhere in this process I lose precision. 
So I've actually written a small python script to convert it directly
with full precision allowed by the struct format, however the result
were almost identical.
Regarding the test cases, we tested eg. with rutile (with one atom
slightly displaced to induce some forces) and the result were
comparable (few percent difference).

Best regards
Pavel Ondračka