[Wien] wrong with rkmax

[Gavin Abo]

Bringing the discussion back to the mailing list.

The Wien2k website does seem to be down at the moment.

Thanks for the additional information:

For RKM = 9.0 session, error happened in lapw1 in the first cycle
:RKM  : MATRIX SIZE 160LOs: 13 RKM=8.41
Fortran compiler l_cprof_p_11.1.073_ia32

 > We are using a 32 bit Linux cause it is a relatively old computer. I 
tried another computer system with 64 bit Linux, then there is no 
problem comes out. So the system problems may be the source of the error.

Glad to hear that it works on a 64 bit system.

 > We are still use the Wien2k 12.1 version. Because it takes a lot time 
to install the package last time, so we are hesitated to make the 
installation again.

Settings to compile 12.1 should work to compile 13.1 on the same computer.

I still recommend that you use 13.1, because developers fixed problems 
that 12.1 had in the fft routines (SRC_lapw0/fft_modules.F), which 
routines are a very important part of the SCF calculations.

On 4/16/2014 11:15 PM, Peter Blaha wrote:
> We need more details, nobody can help with such an error message.
>
> Where does it happen:  lapw2 ?? or which program ...
>
> First iteration, or after a couple of scf cycles ..??
>
> Any info on how far the failing program came (how long did it run ?
> case.dayfile); is there anything in the outputXX file ? ...
>
> What is your struct file ?? "A single Ti atom" could mean a big 
> calculation if
> you are using a too large supercell and small RMTs
>
>
> Am 17.04.2014 03:19, schrieb eagle ying:
>> Dear oleg:
>>      Thanks for your response. The memory usage is neglectable cause 
>> the calculation is only about one atom and my memory is 3G. As I do 
>> the command line:ulimit -s
>> unlimited as you suggested, the stack size is unlimited as ulimit -a 
>> suggested. But the same error still occurs.
>>
>> 2014-4-16 下午8:08于 "Oleg Rubel" <orubel at lakeheadu.ca 
>> <mailto:orubel at lakeheadu.ca>>写道：
>>
>>     The problem could be related to the stacksize. Consider setting 
>> it to unlimited. The command line looks different under different 
>> environments. Here is one possibility:
>>
>>     $ ulimit -s
>>     $ ulimit -s unlimited
>>
>>     You can also check the memory usage during the run via
>>
>>     $ top
>>
>>
>>     Oleg
>>
>>
>>     On Wed, Apr 16, 2014 at 3:57 AM, leithron at gmail.com 
>> <mailto:leithron at gmail.com> <leithron at gmail.com 
>> <mailto:leithron at gmail.com>> wrote:
>>
>>         Dear users:
>>              When I  am  trying to do a test of total energy and EFG 
>> as a function of RmtKmax for a single Ti  atom, errors happened after 
>> seting rkmax as 9.0 and any value
>>         bigger: forrtl:severe(174):SIGSEGV,segmentation fault 
>> occured; stack trace terminated abnormally.
>>             Any help will be appreciated. Thanks in advance.