[QE-developers] Fwd: Different output information in phonon calculation of 'GRID' and'SPLIT'

Thomas Brumme thomas.brumme at uni-leipzig.de
Tue Apr 3 15:15:40 CEST 2018


Dear Zhishuo Huang,

I remember having a similar problem in the past and the only solution I 
found
was to NOT split the calculation on representations. I think that the 
code assumes
that the nscf calculation (which needs to be done in order to have the wave
functions for a certain q point) was already done in the first 
representation for
the q point - thus, for every other representation except the first the nscf
calculation is not done but the code reads the wave functions from file. 
And if
there aren't enough wave functions you get problems.
But I'm only guessing - I'm not a PHonon developer.

Another thing: is your system an insulator or a metal? If it's an 
insulator, does it
have a large gap? If the gap is small, certain phonon modes could lead to an
insulator-metal transition and in this case you should probably try 
treating the
system as a metal...

Regards

Thomas


On 29.03.2018 17:45, Paolo Giannozzi wrote:
>
> ---------- Forwarded message ----------
> From: *黄志硕* <zhishuohuang at gmail.com <mailto:zhishuohuang at gmail.com>>
> Date: Wed, Mar 28, 2018 at 2:40 PM
> Subject: RE: Different output information in phonon calculation of 
> 'GRID' and'SPLIT'
> To: Paolo Giannozzi <p.giannozzi at gmail.com <mailto:p.giannozzi at gmail.com>>
>
>
> Dear Prof. Paolo Giannozzi,
>
> Thank you all the same for your response.
>
> By the way, could you please tell me anyone who is following the 
> development or who could give me any hint on how to make out what’s 
> wrong with my calculation or the Phon module of the code?
>
> Best regards
>
> Zhishuo Huang
>
> *From: *Paolo Giannozzi <mailto:p.giannozzi at gmail.com>
> *Sent: *Tuesday, March 27, 2018 09:02 PM
> *To: *黄志硕 <mailto:zhishuohuang at gmail.com>
> *Subject: *Re: Different output information in phonon calculation of 
> 'GRID' and'SPLIT'
>
> It's not "a few minutes", it may take hours. I am no longer following 
> the development of the phonon code, by the way
>
> Paolo
>
> On Tue, Mar 27, 2018 at 12:36 PM, 黄志硕 <zhishuohuang at gmail.com 
> <mailto:zhishuohuang at gmail.com>> wrote:
>
>     Dear Prof. Paolo Giannozzi,
>
>     I am sorry to bother you through private e-mail, but I have one
>     problem which I posted on the PWscf-mail list and there is no
>     reply about it. I don’t know if I express my problem properly or I
>     disobeyed some rules of posting email on the PWscf-mail list, so I
>     forward to you the e-mail. May I ask you a few minutes to look
>     into my problem? I would appreciate it a lot.
>
>     I am calculating phonon dispersion using two modes of PH module of
>     quantum espresso. One mode is to split total calculation into
>     calculations of each irreducible q point (SPLIT), while the other
>     is the GRID calculation as the demonstrated in the GRID_example,
>     which is to take all the irreducible representations into
>     consideration. However, the output files shows different
>     information when I use these two method to calculate phonon
>     frequency with the same structure and scf input. The main
>     difference is that in the output of GRID method there is a warning
>     shown below:
>
>     ##########################################################
>
>          Starting wfc are  267 atomic wfcs
>
>     Possibly too few bands at point    1  -0.12500   0.12500 0.12500
>
>     Possibly too few bands at point    2   0.37500  -0.37500 0.62500
>
>     Possibly too few bands at point    3  -0.37500   0.37500 -0.12500
>
>     Possibly too few bands at point    4   0.12500  -0.12500 0.37500
>
>     ?
>
>     ?
>
>     ?
>
>     Possibly too few bands at point   87  -1.12500  -0.37500 -0.37500
>
>     Possibly too few bands at point   88  -0.62500  -0.87500 0.12500
>
>     ##########################################################
>
>     While there is no such information in the output file of SPLIT
>     method.
>
>     The input of GRID method for each irreducible representation is
>     shown below
>
>     ##########################################################
>
>     Phonon of K3C60
>
>     &inputph
>
>     tr2_ph=1.0d-17,
>
>     prefix='K3C60',
>
>       ldisp=.true.
>
>       nq1=2, nq2=2, nq3=2
>
>       start_q=2
>
>       last_q=2
>
>     start_irr=?irr_num?,
>
>     last_irr=?irr_num?,
>
>       amass(1)=12
>
>       amass(2)=39.1
>
>       outdir='./',
>
>     fildyn='K3C60.dynG',
>
>       reduce_io = .true.
>
>     /
>
>     ##########################################################
>
>     And the input of GRID method after collecting all the information
>     of each irreducible representation is
>
>     ##########################################################
>
>     Phonon of K3
>
>     &inputph
>
>     tr2_ph=1.0d-17,
>
>     prefix='K3C60',
>
>       ldisp=.true.
>
>       nq1=2, nq2=2, nq3=2
>
>     recover=.true.,
>
>       start_q=2
>
>       last_q=2
>
>       amass(1)=12
>
>       amass(2)=39.1
>
>       outdir='./',
>
>     fildyn='K3C60.dynG',
>
>       reduce_io = .true.
>
>     /
>
>     ##########################################################
>
>     And the input of SPLIT method is
>
>     ##########################################################
>
>     Phonon of K3C60
>
>     &inputph
>
>     tr2_ph=1.0d-17,
>
>     prefix='K3C60',
>
>       ldisp=.true.
>
>       nq1=2, nq2=2, nq3=2
>
>       start_q=2
>
>       last_q=2
>
>       amass(1)=12
>
>       amass(2)=39.1
>
>       outdir='./',
>
>     fildyn='K3C60.dynG',
>
>       reduce_io = .true.
>
>     /
>
>     ##########################################################
>
>     Why is there the difference information in the output file for the
>     same structure and scf input? I will appreciate it if anyone could
>     give me any information about that.
>
>     Best regards
>
>     Zhishuo Huang
>
>
>
>
> -- 
>
> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy 
> <https://maps.google.com/?q=Udine,+via+delle+Scienze+208,+33100+Udine,+Italy&entry=gmail&source=g>
> Phone +39-0432-558216 <tel:+39%200432%20558216>, fax +39-0432-558222 
> <tel:+39%200432%20558222>
>
>
>
>
> -- 
> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
> Phone +39-0432-558216, fax +39-0432-558222
>
>
>
> _______________________________________________
> developers mailing list
> developers at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/developers

-- 
Dr. rer. nat. Thomas Brumme
Wilhelm-Ostwald-Institute for Physical and Theoretical Chemistry
Leipzig University
Phillipp-Rosenthal-Strasse 31
04103 Leipzig

Tel:  +49 (0)341 97 36456

email: thomas.brumme at uni-leipzig.de

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20180403/c1de8efb/attachment-0001.html>


More information about the developers mailing list