[Wien] scf error

[Peter Blaha]

Compilation with ifort9 does not produce a "static" executable (including
all library routines), but a "dynamic", so called "shared-object" executable.
Such a program requires, that at runtime the identical (or at least compatible)
libraries (compiler, mkl, but also all "Unix"-libraries) are present.

I don't know the reason, but "big" static Intel-executables do not work 
anymore since Intel changed from ifc7 to ifort8 and later.

Thus you cannot just move your WIEN2k installation to a different computer,
but need to have identical (compatible) compiler/libraries there.

Thus usually, I NEVER move a compiled WIEN2k version, but always compile it
on the corresponding system.

(PS: You can use the binaries from our website. They are statically linked
(using an old (suboptimal) compiler.

> I used ifort9, but still error was not solved. 
> Yesterday it worked without error, and today it does not.
> Yesterday I successfully complied wien2k for test. And after that the computer administrator moved it to supercom for evrybody using it. Wien does not work in this supercom.
> I reactivated the test wien2k and now it also does no work well. It can calculate the simply sturcture such like TiC and TaN. But could not for the complex stucture, such like HfO2.
> We can not guess the reason. I
> What's NFS problem? Anyway on the 32bit computer(our server computer), it works very well.
> Clmval file only has NaN's. 
> It is very difficult to use that. 
> 
> as like
>     VALENCE CHARGE DENSITY   IN  MT SPHERES       1 ITERATION
>     NORM OF CLM(R) =      
>     
>    ATOMNUMBER     1
>    NUMBER OF LM 49
>     
>     
>    CLM(R) FOR L  0   M= 0
>     
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
>     NaN                NaN                NaN                NaN
> 
> 
> > I used ifort8.1 and MKL7.1 with gcc compiler and during installation no error was found. The example of TiC introduced in manual was calculated correctely. 
> 
> We highly recommend upgrade to ifort 9, since ifort8 is known to be very
> buggy.
> 
> Provided you have set your "limits" to unlimited (there are many discussions
> about that in the mailinglist), the TiC222 error in lapw0 (segmentation 
> fault) could be due to your compiler.
> 
> The problem in HfO2 is in the file HfO60902.clmval
> It seems, it has not been written correctly. Inspect the file in an editor.
> any ***** or other unexpected characters ?
> Hard to say where it comes from. Could be a NFS problem ? ???
> 
> Regards
> 
> > 
> > But if the structure become complex, scf does not finish properely. The command "init_lapw" of the structure files did not show any error massage.  I also adjust mix factor in case.inm file to 0.01.  But finally there was any progress.
> > 
> > Here I attached two structure files.  One is for HfO2.  And the other is supercell of TiC. One atomic position of C was substituted as an Oxygen.  I checked these struct file using nn, sgroup, and symmetry command.  No error was there. 
> > 
> > Here I copied the error massage and the struct file.  Becaus , I cannot attached any file through the mail server, I copied them below. 
> > 
> > The error massage 
> > 1. for HfO2
> > The structure file
> 


                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------