[Wien] Orbital polarization error

Roberto Iglesias roberto.iglesias at psi.ch
Fri Oct 13 16:22:45 CEST 2006


I've just tried to do a calculation with TEMP 0.003 instead of TETRA

and the new error messages in STDOUT read:

LAPW0 END
  LAPW1 END
  LAPW1 END
LAPWSO END
  LAPW2 END
  LAPW2 END
LAPWDM END
  CORE  END
  CORE  END
  MIXER END
in cycle 2    ETEST: .0000020000000000   CTEST: .0023619
  LAPW0 END
  ORB   END
  ORB   END
  LAPW1 END
  LAPW1 END
LAPWSO END
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC        Routine            Line        Source
lapw2c             08247252  Unknown               Unknown  Unknown
lapw2c             08242F3F  Unknown               Unknown  Unknown
lapw2c             0822A35C  Unknown               Unknown  Unknown
lapw2c             0808022B  fermi5_                   111  fermi5.f
lapw2c             080791A7  fermi_                    105  fermi.f
lapw2c             0809B03C  MAIN__                    216  lapw2_tmp_.F
lapw2c             0804BDC5  Unknown               Unknown  Unknown
libc.so.6          00B1EE23  Unknown               Unknown  Unknown
lapw2c             0804BD01  Unknown               Unknown  Unknown

Could this help in some way?

Roberto


Roberto Iglesias wrote:
> 
> Peter Blaha wrote:
>> If your emin/emax is the same, you should find about twice as many 
>> eigenvalues in energyso as you have in energy !
>> You energyso files are again tiny, so they cannot contain all the 
>> required eigenvalues (at all specified k-points).
>> The energyso files must be incomplete/corrupted,...
>>
>> Is case.inso still ok (small EMAX) ?
> 
> Yes
> 
>> Is the list of eigenvalues in output1up and outputso "comparable" (2x 
>> more eigenvalues in lapwso) and is this list the same in energyso and 
>> outputso ?
> 
> You question made wonder about my initialization procedure. After a normal initialization, I executed 
> initso_lapw, modifying appropriately the case.inso file. After this, I went again to re-initialize. Before, I 
> executed again from x dstart, x dstart -up and x dstart -dn. But I've done another try looking at case.in1, 
> .in2, inm and excuting kgen again and from there to dstart. Should i go all the way from x nn or is it really 
> not necessary?
> 
> After the spin orbit calculation, I checked the sizes of the files you asked about:
> 
> [l_iglesias at pc6096 k8000_completeinit]$ ll k8000_completeinit.output1up
> -rw-rw-r--  1 l_iglesias l_iglesias 1011014 Oct 13 15:11 k8000_completeinit.output1up
> [l_iglesias at pc6096 k8000_completeinit]$ ll k8000_completeinit.outputso
> -rw-rw-r--  1 l_iglesias l_iglesias 1152549 Oct 13 15:01 k8000_completeinit.outputso
> [l_iglesias at pc6096 k8000_completeinit]$ ll k8000_completeinit.energy*
> -rw-rw-r--  1 l_iglesias l_iglesias  986882 Oct 13 15:11 k8000_completeinit.energydn
> -rw-rw-r--  1 l_iglesias l_iglesias 1908425 Oct 13 15:11 k8000_completeinit.energydum
> -rw-rw-r--  1 l_iglesias l_iglesias 1908425 Oct 13 15:11 k8000_completeinit.energysodn
> -rw-rw-r--  1 l_iglesias l_iglesias 1908425 Oct 13 15:11 k8000_completeinit.energysoup
> -rw-rw-r--  1 l_iglesias l_iglesias  991840 Oct 13 15:11 k8000_completeinit.energyup
> 
> And after the well known crash in the orbital polarization run:
> 
> [l_iglesias at pc6096 k8000_completeinit]$ ll k8000_completeinit.output1up
> -rw-rw-r--  1 l_iglesias l_iglesias 1011474 Oct 13 15:13 k8000_completeinit.output1up
> [l_iglesias at pc6096 k8000_completeinit]$ ll k8000_completeinit.outputso
> -rw-rw-r--  1 l_iglesias l_iglesias 478025 Oct 13 15:14 k8000_completeinit.outputso
> [l_iglesias at pc6096 k8000_completeinit]$ ll k8000_completeinit.energy*
> -rw-rw-r--  1 l_iglesias l_iglesias 986882 Oct 13 15:13 k8000_completeinit.energydn
> -rw-rw-r--  1 l_iglesias l_iglesias  65677 Oct 13 15:14 k8000_completeinit.energydum
> -rw-rw-r--  1 l_iglesias l_iglesias  65677 Oct 13 15:14 k8000_completeinit.energysodn
> -rw-rw-r--  1 l_iglesias l_iglesias  65677 Oct 13 15:14 k8000_completeinit.energysoup
> -rw-rw-r--  1 l_iglesias l_iglesias 991840 Oct 13 15:13 k8000_completeinit.energyup
> 
> So it is evident that again I don't have enough eigenvalues in outputso and energyso. But this is because 
> their size dramatically decreased after trying the -orb calculation. Below are the sizes of the 001 
> calculation that finished OK, for comparison
> 
> [l_iglesias at pc6096 k8000]$ ll k8000.output1up
> -rw-rw-r--  1 l_iglesias l_iglesias 335968 Oct 12 11:49 k8000.output1up
> [l_iglesias at pc6096 k8000]$ ll k8000.outputso
> -rw-rw-r--  1 l_iglesias l_iglesias 380242 Oct 12 11:50 k8000.outputso
> [l_iglesias at pc6096 k8000]$ ll k8000.energy*
> -rw-rw-r--  1 l_iglesias l_iglesias 320780 Oct 12 11:50 k8000.energydn
> -rw-rw-r--  1 l_iglesias l_iglesias 620285 Oct 12 11:50 k8000.energydum
> -rw-rw-r--  1 l_iglesias l_iglesias 620285 Oct 12 11:50 k8000.energysodn
> -rw-rw-r--  1 l_iglesias l_iglesias 620285 Oct 12 11:50 k8000.energysoup
> -rw-rw-r--  1 l_iglesias l_iglesias 322482 Oct 12 11:49 k8000.energyup
> 
> And I'm wondering now if there is some step I missed before starting the -orb calculation, apart from writing 
> down the proper case.inorb and cas.indmc files.
> 
> Again now the error printed to the screen is:
> 
> LAPW0 END
>   LAPW1 END
>   LAPW1 END
> LAPWSO END
>   LAPW2 END
>   LAPW2 END
> LAPWDM END
>   CORE  END
>   CORE  END
>   MIXER END
> in cycle 2    ETEST: 0   CTEST: 0
>   LAPW0 END
>   ORB   END
>   ORB   END
>   LAPW1 END
>   LAPW1 END
> LAPWSO END
> forrtl: severe (174): SIGSEGV, segmentation fault occurred
> Image              PC        Routine            Line        Source
> lapw2c             08079CA3  fermi_tetra_              488  fermi.f
> lapw2c             0807909D  fermi_                    109  fermi.f
> lapw2c             0809B03C  MAIN__                    216  lapw2_tmp_.F
> lapw2c             0804BDC5  Unknown               Unknown  Unknown
> libc.so.6          00B1EE23  Unknown               Unknown  Unknown
> lapw2c             0804BD01  Unknown               Unknown  Unknown
> 
> 
> Regards
> 
> Roberto
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien


More information about the Wien mailing list