[Wien] -NI switch in mini

Kakhaber Jandieri Kakhaber.Jandieri at physik.uni-marburg.de
Wed Oct 27 19:50:37 CEST 2010


Dear Prof. Blaha,

Using "min -NI"  I managed to continue the sequence of scf cycles from
the point where it was stopped.
Thank you very much.



Peter Blaha wrote:
> I doubt your analysis.   min does not even know if you run mpi
> parallel or not.
>
> min   NEVER runs    x dstart,
> but eventually      x dstart -super (without -NI). This is NOT a new
> initialization for the scf cycle.
>
> the   -NI switch  also avoids removing case.tmpM  (the
> minimization-history - similar to the broyd* files of run_lapw)
>
> PS: For a structural optimization I would NOT use 100 geometry steps
> and save only every 10 steps. Usually in much fewer steps you should
> get a decent geometry.
>
> PPS: check the :log file. You can see which programs are called with
> which options.
>
> Am 22.10.2010 18:43, schrieb Kakhaber Jandieri:
>> Dear wien2k users,
>>
>> I cannot re-run structural optimization without initialization.
>> When I use "mini.position" from w2web graphical interface, the "-NI"
>> switch works properly, but when I run MPI-paralleized optimization task
>> on the cluster using *.csh script, "-NI" has no effect.
>>
>> So the results of the different commands in my *.csh script:
>>
>> min -i 100 -s 10 -j 'run_lapw -p -I -i 40 -ec 0.001 -fc 1.0 -cc 0.001'
>>
>> and
>>
>> min -NI -i 100 -s 10 -j 'run_lapw -p -I -i 40 -ec 0.001 -fc 1.0 -cc
>> 0.001'
>>
>> are exactly the same - DSTART is performed prior scf cycle.
>> Ofcourse this is not what I want. The maximal time limit on our cluster
>> is 240h and for large surface calculations I am forced to interrupt and
>> continue the same task several times.
>>
>> I will be extremely thankfull for any advices and suggestions.
>>
>>
>> Dr. Kakhaber Jandieri
>> Department of Physics
>> Philipps University Marburg
>> Tel:+49 6421 2824159 (2825704)
>>
>>
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at
>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>

Dr. Kakhaber Jandieri
Department of Physics
Philipps University Marburg
Tel:+49 6421 2824159 (2825704)


More information about the Wien mailing list