[Pw_forum] error on phonon calculation (root not converged)

[Liping YU]

Eyvaz Isaev wrote:

>Dear Paolo,
>
>--- Paolo Giannozzi <giannozz at nest.sns.it> wrote:
>  
>
>>this may actually work, but the problem that shows
>>up in  Liping's output is rather strange and nasty. 
>>    
>>
>
>  
>
>>The self-consistent  procedure seems to converge 
>>quite nicely: |ddv_scf|^2 goes  down towards the
>>threshold, then it starts to go up to crazy
>>values. 
>>    
>>
>Yes, I paid attention to this behaviour. Actually,
>recently I had very close situation: for some modes
>everything was OK, and next mode also started well,
>and then turned out to be crazy. At least in my case,
>using small mixing parameter fixed the problem.
>
>I suggest, Liping should keep us informed about the
>problem.
>
The same phenomenon still campe up in the self consistent
calculation of representation # 6 if I only reduce aplpha_mix(1)
to 0.1 and keep tr2_ph = 1.0e-18. So only using smaller mixing
parameter doesn't work in this case.

However it works as we expected if I set alpha_mix(1) = 0.1
and tr2_ph = 1.e-10. But the results (here 20-atom 2*2*1 supercell used)
don't agree well with those of calculation where 1*1*1 cubic perovskite
unit cell (5 atoms inside) are used. For example, the dielectric
constant of
Srtio3, is for 2*2*1 supercell

( 222.139158023 0.000000000 0.000000000 )
( 0.000000000 222.139158023 0.000000000 )
( 0.000000000 0.000000000 6.262166130 )
and for 1*1*1 unit cell
( 6.296436141 0.000000000 0.000000000 )
( 0.000000000 6.296436141 0.000000000 )
( 0.000000000 0.000000000 6.296436141 ).

In direction of xx and yy, dielectric constants in two cases should be
pretty
close to each other like the one in zz direction. I am not sure if the
problem
comes from something related to the supercell I used. There is no such
problem
for 1*1*1 unit cell when setting alpha_mix(1) = 0.7 and tr2_ph = 1.e-18.
On the other hand, I am also wondering if it's also possible due to
pseudopotential. Because there is no such problem for PbTiO3 in 2*2*1
supercell.
Thanks for your comments!
liping

>Bests,
>Eyvaz.
>
>  
>
>>Why, I don't know, and I am afraid there is
>>very little  one can do when this kind of behavior 
>>occurs in the  middle  of a long job without any 
>>apparent reason
>>    
>>
>
>
> 
>
>__________________________________________________
>Do You Yahoo!?
>Tired of spam?  Yahoo! Mail has the best spam protection around 
>http://mail.yahoo.com 
>_______________________________________________
>Pw_forum mailing list
>Pw_forum at pwscf.org
>http://www.democritos.it/mailman/listinfo/pw_forum
>  
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/attachments/20060409/4a50b802/attachment.htm