[Pw_forum] Is maxirr=N working in phonon calculations?

[Stefano Baroni]

On Nov 6, 2008, at 2:03 AM, Serge Nakhmanson wrote:

> Dear Paolo,
>
> Thank you very much for exhaustively answering my questions, even
> the one that I did not ask, i.e., if irreps are picked at random,
> then what is the purpose of using "maxirr=N"? :))
>
> Your comment on randomness of irrep generation makes sense to me.
> I only know a general answer to how it is done -- by induction
> from an (Abelian) group of crystal translations -- but am
> rusty on particular numerical algorithms. However, I got an
> excellent new book by El-Batanouny and Wooten (ISBN 9780521828451),
> all the details should be in there.

I do not know about the book, but I am afraid that what is implemented  
in the code is not anything worth reporting in a book. The idea is as  
follows. The code already contains all the weaponery for simmetrizing  
all kind of stuff with respect to the small group of the point in the  
Brillouin zone we are working at. In particular, you fill the  
dynamical matrix with random numbers and then symmetrize it. The  
resulting matrix will commute with all the operations of the small  
group, and its eigenvectors will therefore realize a number of  
irreducible representation of the group. As simple as that. Do not ask  
me about the details of the implementation, though. The documentation  
(as poor as it may be) is the code itself ...

Cheers - SB

>
>
> THX again,
>
> S.
>
> Paolo Giannozzi wrote:
>> Serge Nakhmanson wrote:
>>
>>> Still the code blasted through all 10 irreps.
>>
>> not sure whether the "maxirr" option really works. In the CVS
>> version it was replaced by other options that are useful for
>> Grid execution (i.e. splitting a long phonon calculation in
>> many smaller pieces rto be run on different machines). Its main
>> usage was to prevent the code from crashing for cpu time limit
>> in a ungraceful way, but option "max_seconds" can do the same.
>>
>>> Also, is there any system to which irreps are computed first and
>>> which later? Or is this process completely random, i.e. it would
>>> be futile to cover a certain frequency window by computing only
>>> some, but not all, irreps?
>>
>> I don't think this is going to work. Not sure how exactly
>> irreps are selected. If no symmetry is present, taht's easy:
>> atom 1 component x, y,z, atom 2 component x, y, z and so on.
>> In presence of symmetry, the irreps are obtained via a procedure
>> in which random matrices are symmetrized, so the order in which
>> they are produced is quite unpredictable (also the accuracy of
>> this answer is unpredictable, since it is based on vague
>> recollections)
>>
>> Paolo
>
>
> -- 
> *********************************************************
>  Serge M. Nakhmanson               phone: (630) 252-5205
>  Assistant Scientist                 fax: (630) 252-4798
>  MSD-212, Rm. C-224
>  Argonne National Laboratory
>  9700 S. Cass Ave.
>  Argonne, IL 60439
> *********************************************************
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum

---
Stefano Baroni - SISSA  &  DEMOCRITOS National Simulation Center -  
Trieste
http://www.sissa.it/~baroni / [+39] 040 3787 406 (tel) -528 (fax) /  
stefanobaroni (skype)

La morale est une logique de l'action comme la logique est une morale  
de la pensée - Jean Piaget

Please, if possible, don't  send me MS Word or PowerPoint attachments
Why? See:  http://www.gnu.org/philosophy/no-word-attachments.html






-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.democritos.it/pipermail/pw_forum/attachments/20081106/8116fec9/attachment.htm