[Wien] system- or usererror?

Peter Blaha pblaha at zeus.theochem.tuwien.ac.at
Mon Jul 21 21:42:00 CEST 2003


> I try to run the same sample with the same parameters and get different
> results all the time. When I check, the settings seem correct, i.e. the
> space-group is shown correctly, the number and kind of nearest neighbors
> corresponds to a face-centered cubic structure and so on. When I run SCF,
> most of the time I get either a "address error" or a segfault.
> Once the calculation went through, I even could plot an Electron Density Plot,
> which sadly enough did not correspond to the sample.
>
> Shouldn't WIEN give reproducable results with the same settings and input
> data?

It's hard to guess what's happening just from your general complains, but
little specific info. At least now you
say that an scf cycle sometimes works, sometimes not.

Do you always "save_lapw" a calculation before repeating it ?

The type of errors you mention points to limited memory / swap space or
some other system limits.

Do you have enough memory ?
How big is your swapspace ? Use  "free" or top !!
Is your system used by other jobs ? This would explain why it works
sometimes (on an empty machine), sometimes not.

What gives   "limit" (under a tcsh, I don't know the corresponding bash
command). set datasize and most other parameters to "unlimited".

Be systematic:
init_lapw
  Any error ? Check the dayfile.
run_lapw -i 1
  Just one iteration. Any error, check the dayfile and/or stdout.
As soon an error occurs stop and analyze or send as much info as possible.

Did lapw0,1 or 2 ever run through ?

About 1000 groups !!! (WIEN97+WIEN2k) have successfully ran the TiC
example.

                                      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/
--------------------------------------------------------------------------




More information about the Wien mailing list