[Wien] hyper fine field on different platforms

[Björn Hülsen]

Dear Wien users,

recently I have compiled Wien2k 6.4 for a IBM Power5 system and a 
Pentium 4 cluster. Both versions work fine. When I set up a calculation 
of Co2MnSi on both platforms with exactly the same input and convergence 
criteria almost all relevant data from the scf files (e.g. eigenvalues, 
mag. moments,  total energy) agree in reasonable tolerances as expected. 
But the values for the hyper fine fields show large differences as you 
can see below:

                         valence         semi-core        
core                  total
Pentium 4 
:HFF001: HFF: 230.220           0.000        -320.734         -90.514
:HFF002: HFF: 446.428           0.000        -655.243        -208.816
:HFF003: HFF:   26.940           0.000             1.443            28.383

Power5
:HFF001: HFF: 369.412           0.000        -318.441           50.971
:HFF002: HFF: 528.449           0.000        -653.136        -124.687
:HFF003: HFF:   27.626           0.000              1.421          29.048

Here HFF001 belongs to Co, HFF002 to Mn, and HFF003 to Si. Why are the 
values for Co and Mn so different (and even change sign for Co)?

Best regards,
Bjoern
-- 

Björn Hülsen
Fritz-Haber-Institut
Faradayweg 4-6
14195 Berlin

phone:  +49 30 8413-4863
email:  huelsen at fhi-berlin.mpg.de
www:    www.fhi-berlin.mpg.de