[Wien] troubles with mixer in scf

Peter Blaha pblaha at theochem.tuwien.ac.at
Wed Mar 13 09:12:24 CET 2013 

It looks like an error in some input file and most likely it is the 
source of the segmentation fault occuring later.
Best would be if you find the line number, where the READ-problem 
occurs. (compiled with -traceback and ifort, not aix !)

Since you mentioned core-holes: maybe case.inc got wrong when you 
introduced the hole ???

Anyway, it seems to work now.

PS: The knowledge of spin-polarization in w2web does not depend on the 
directory, but the "Session" of w2web. So deleting all files in a 
directory, but "reusing" the old project in w2web does not change the 
settings of w2web (but you can modify this manually in the "Session 
management" section. But in any case, -up is ignored for nn, sgroup,...


On 03/13/2013 08:56 AM, David Holec wrote:
> Dear Laurence,
>
> Many thanks for your prompt response. To your seggestions:
>
> --> cat *.error:
> Error in MIXER
> Error in MIXER
>
> --> mixer mixer.def:
> 1525-097 A READ statement using decimal base input found the invalid
> digit '.' in the input file.  The program will recover by assuming a
> zero in its place.
> Segmentation fault (core dumped)
>
> --> deleting *broyd* files didn't help.
>
> But, I have started the calculation from fresh (used only the struct
> file, deleted everything else in the directory), and after
> initialisation everything seems to be running fine now. I have no idea
> where and when I did something wrong, but apparently it was again the
> human factor.
>
> Hmm, just checking the difference between the \:log file from
> yesterday and today, I realise that yesterday all initialisation
> routines (nn, sgroup, etc.) were invoked with -up switch. I have no
> idea why (I used w2web in a fresh directory both yesterday and today).
> Could thins be the trouble maker?
>
> Thanks for your help!
> David
> --
> Dr. David Holec
> Dept. of Physical Metallurgy and Materials Testing
> Montanuniversität Leoben
> Franz-Josef-Strasse 18
> A-8700 Leoben
> Austria
> tel. +43-(0)3842-4024211
> fax. +43-(0)3842-4024202
>
>
> On 12 March 2013 18:37, Laurence Marks <L-marks at northwestern.edu> wrote:
>> Hard to know exactly. Normally this is because something went wrong
>> earlier (e.g. NaN in files) and did not crash the scf cycle.
>>
>> Is there anything in any of the other error files (cat *.error) ?
>>
>> Did you have a previous unpolarized calculation in the same directory
>> and did not delete the *.broyd* files?
>>
>> Did you use -traceback in the compilation? If you did what does the
>> command "mixer mixer.def" give you? (Sometimes this is more
>> informative.)
>>
>> A segmentation fault like this is normally due to some incorrect
>> arguments being passed to the Intel mkl routines which are hard to get
>> useful information from.
>>
>> Let me know please.
>>
>> On Tue, Mar 12, 2013 at 12:23 PM, David Holec
>> <david.holec at unileoben.ac.at> wrote:
>>> Dear Wien2k users,
>>>
>>> I am facing an error in the first scf cycle. Unfortunately, I am
>>> unable to find the information that would guide me down to the core of
>>> the problem. I would appreciate any guidance from you!
>>>
>>> It is a spin-polarized calculation of CrN. The error appear at the end
>>> of the first cycle:
>>>
>>> ...
>>>     w03   user=271.571    wallclock=17287.2
>>> 3.0u 0.6s 0:53 6% 2+129k 0+0io 30pf+0w
>>>>    lcore -up   (17:59:52) 0.2u 0.0s 0:00 51% 1+9k 0+0io 3pf+0w
>>>>    lcore -dn   (17:59:52) 0.2u 0.0s 0:00 75% 1+8k 0+0io 1pf+0w
>>>>    mixer       (17:59:53) Segmentation fault (core dumped)
>>> 0.1u 0.3s 0:01 35% 1+43k 0+0io 1pf+0w
>>> error: command   /home/d.holec/codes/WIEN2k_11.1/mixer mixer.def   failed
>>>
>>>>    stop error
>>>
>>> mixer.error is not really helpful: "Error in MIXER"
>>>
>>> Perhaps some hint is in the case.outputm file:
>>> ...
>>>   JATOM =  11   J =  2
>>>    ESX  =  -5.7200202706E+02 OCC =   2.0000000000E+00  2.0000000000E+00
>>>     ESUM,EKIN =  -94314.3061676110083 0.000000000000000000E+00
>>>   JATOM =  11   J =  3
>>>    ESX  =  -4.8320954414E+02 OCC =   2.0000000000E+00  2.0000000000E+00
>>>     ESUM,EKIN =  -94886.3081946670136 0.000000000000000000E+00
>>>   JATOM =  11   J =  4
>>>    ESX  =  -9.5116518730E+02 OCC =   4.0000000000E+00  4.0000000000E+00
>>>     ESUM,EKIN =  -95369.5177388110169 0.000000000000000000E+00
>>>   JATOM =  12   J =  1
>>>    ESX  =  -5.
>>>
>>> The struct file contains 12 in-equivalent atoms (6 N, 6 Cr), atom 12
>>> is the last Cr.
>>>
>>> I am not sure whether you need more information to track the problem -
>>> if so, I am happy to share it, of course.
>>>
>>> Perhaps last comment is: I am trying to calculate the N-K energy edge
>>> onset for various magnetic configurations as a difference between the
>>> ground state and excited state with core hole (extra electron in
>>> case.in2 file). Exactly the same settings in non-magnetic case (only
>>> different initial spin given in instgen_lapw for Cr atom (up vs. nm)
>>> and scf invoked by run_lapw instead of runsp_lapw) runs without any
>>> trouble.
>>>
>>> Thanks in advance for you help!
>>> David
>>> --
>>> Dr. David Holec
>>> Dept. of Physical Metallurgy and Materials Testing
>>> Montanuniversität Leoben
>>> Franz-Josef-Strasse 18
>>> A-8700 Leoben
>>> Austria
>>> tel. +43-(0)3842-4024211
>>> fax. +43-(0)3842-4024202
>>> _______________________________________________
>>> Wien mailing list
>>> Wien at zeus.theochem.tuwien.ac.at
>>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>>
>>
>>
>> --
>> Professor Laurence Marks
>> Department of Materials Science and Engineering
>> Northwestern University
>> www.numis.northwestern.edu 1-847-491-3996
>> "Research is to see what everybody else has seen, and to think what
>> nobody else has thought"
>> Albert Szent-Gyorgi
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at
>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>

-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300             FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at    WWW: 
http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------

More information about the Wien mailing list