[Wien] Finally the report about optimizing the structure with MSR1a and mBJ

[Luis Carlos Ogando Dacal]

Dear Prof. Marks and Wien2k community,

   I would like to report that even using the options you suggested in
the previous e-mail ( run_lapw -ec 0.0001 -cc 0.00025 -vec2pratt
-in1orig -i 200 -it ), the optimization with MSR1a-mBJ does not get
convergence.
   If you want to read any of the generated files, just ask me .
   All the best,
                       Luis



2012/9/11 Luis Carlos Ogando Dacal <ogando at ieav.cta.br>:
> Dear Prof. Marks,
>
>    Thank you for the comments.
>    As I said before, I took the convergence criteria from literature
> (pseudopotential, not lapw) and as they worked fine for MSCE1 with
> PBE, I supposed that they should also work with MSR1a-mBJ.
>
>    Anyway, I will try again the optimization, but using the options
> you suggested. As soon as possible, I will report the results.
>    By the way, I have fixed atoms because I am trying to simulate a
> system composed by a substrate with a cap layer, so only the cap layer
> atoms are relaxed.
>    All the best,
>                    Luis
>
>
>
> 2012/9/11 Laurence Marks <L-marks at northwestern.edu>:
>> The convergence criteria are about 100 times too small to be
>> realistic. Use -ec 0.0001 at most, and -fc has no bearing for MSR1a; I
>> suggest adding -cc 0.00025. With what you have used you are limited by
>> the noise of the iterative method. If you look in the documentation,
>> you will see that -vec2pratt is recommended (by me) as it is better
>> with MSR1a.
>>
>> When you say "some atoms were fixed" what do you mean? For a
>> non-centro symmetric structure it is OK to fix one atom, but if you
>> fix more then your results may well be unphysical.
>>
>> What matters is not whether MSR1a works with mBJ (it does), rather
>> whether the atom positions are better in some sense.
>>
>> On Tue, Sep 11, 2012 at 8:36 AM, Luis Carlos Ogando Dacal
>> <ogando at ieav.cta.br> wrote:
>>> Dear Prof. Blaha, Marks and WIEN2k community,
>>>
>>>    I finally tried the structure optimization using MSR1a and mBJ
>>> (without SOC).
>>>    As I have a previous successful optimization of the structure using
>>> MSEC1 and PBE (min_lapw), I repeated the same parameters whenever
>>> possible.
>>>    The result is that after 200 iterations I did not get the
>>> convergence with remarkable abrupt changes in ETEST during the
>>> calculation.
>>>    Please, let me show the steps I followed for the optimization and
>>> take a look to see if there is any obvious mistake in the procedure:
>>>
>>> 1) init_lapw (using the same parameter as in the previous
>>> min_lapw-MSEC1-PBE optimization)
>>>
>>> 2) create case.inm_vresp (from template)
>>>
>>> 3) change NR2V to R2V in case.in0
>>>
>>> 4) a short SCF cycle to create case.r2v
>>>
>>> 5) change indxc to 28 in case.in0
>>>
>>> 6) change indxc again to 50 and save the file as case.in0_grr
>>>
>>> 7) a short SCF cycle (6 iterations ) with MSR1
>>>
>>> 8) remove case.in0_grr
>>>
>>> 9) change to MSR1a in case.inm
>>>
>>> 10) finally  run_lapw -ec 0.000001 -fc 0.01 -in1orig -i 200 -it
>>>
>>>    Probably you will say that the convergence parameters are too
>>> strict, but I found them in a paper where a similar calculation where
>>> performed (using pseudopotentials, not lapw).
>>>    If you want, I can send you the files generated during the optimization.
>>>    All the best,
>>>                   Luis
>>> PS: some atoms where fixed through case.inM
>>>
>>>
>>>
>>>
>>>
>>
>> --
>> 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