[QE-users] Running PHonon w/ images

Thu Jan 31 22:30:16 CET 2019

Hi Everyone,
 I suppose I should write to at a bit of humble pie.  I was thoroughly
confused by the concept underlying the example, but now understand and my
calculations work fine.  Essentially, to use image parallelization , one
must run two PHonon calculations.  The first one is a parallel one over
multiple images, and the second one uses the same inputs but with
recover=.true.  This collects the results from the previously run images
calculation with the same prefix name automatically and then gives the
expected output.  This was probably obvious to other uses, but since I was
the one confused I thought I'd spell it out (again) in case someone else
out there was confused.
Sincerely,
James Johns

On Thu, Jan 31, 2019 at 2:44 PM James Johns <jjohns at umn.edu> wrote:

> Dear Paolo,
>   Thanks for pointing that out.  When I attempt to run that example,
> however, PHonon crashes with an error message as follows in the output file:
>
>
>  iter #   1 total cpu time :    19.1 secs   av.it.:   6.0
>       thresh= 1.000E-02 alpha_mix =  0.700 |ddv_scf|^2 =  1.326E-06
>
>       iter #   2 total cpu time :    21.1 secs   av.it.:   4.7
>       thresh= 1.151E-04 alpha_mix =  0.700 |ddv_scf|^2 =  6.508E-08
>      kpoint   1 ibnd   5 solve_e: root not converged        NaN
>      kpoint   1 ibnd   5 solve_e: root not converged        NaN
>      kpoint   1 ibnd   5 solve_e: root not converged        NaN
>      kpoint   2 ibnd   5 solve_e: root not converged        NaN
>      kpoint   2 ibnd   5 solve_e: root not converged        NaN
>      kpoint   2 ibnd   5 solve_e: root not converged        NaN
>
>
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>      Error in routine broyden (1):
>      factorization
>
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
>
> If you have any suggestions as to how I can get the example to work I'm
> all ears. I really appreciate all of the help this forum provides.
> Best,
> James
>
> On Thu, Jan 31, 2019 at 2:28 AM Paolo Giannozzi <p.giannozzi at gmail.com>
> wrote:
>
>> In PHonon/examples/Image_example/ there is an example showing how to
>> proceed in case of "image" parallelization
>>
>> Paolo
>>
>> On Wed, Jan 30, 2019 at 5:00 PM James Johns <jjohns at umn.edu> wrote:
>>
>>> Hi everyone,
>>>
>>>   I wanted to run a phonon calculation on a material with 24 heavy
>>> atoms, and so I needed to parallelize it.  I ran it with 4 images and
>>> nk=12  (there are 12 total k points in my irreducible BZ).  The calculation
>>> seemed to go okay (ie all 72 modes converged) but nothing was written to
>>> the dynG file.
>>> The output was written to 4 files (1 per image: fname.out, out.1_0,
>>> out.2_0, out.3_0) with the following representations / modes for each of
>>> the images
>>> Image 0
>>> modes 1-17
>>> Image 1
>>> modes 18-35
>>> Image 2
>>> modes 36-53
>>> Image 3
>>> modes 54-72
>>>
>>> The 4 images finished at 4 different times, no big surprise, with image
>>> #3 finishing last
>>>
>>> When Image 3 finished it output the following:
>>>
>>>    iter #  19 total cpu time : 60255.8 secs   av.it.:  12.3
>>>       thresh= 2.925E-07 alpha_mix =  0.400 |ddv_scf|^2 =  8.390E-13
>>>
>>>      End of self-consistent calculation
>>>
>>>      Convergence has been achieved
>>>
>>>      Stopping because representation    0 is not done
>>>
>>>      PHONON       :    16h18m CPU       18h16m WALL
>>>
>>> out.1_0 and out.2_0 said something similar (Stopping because
>>> representation 0 is not done)
>>> fname.out (the image 0 file) said stopping because representation 18 is
>>> not done, but it did say self consistency achieved.
>>>
>>> My question is if there is enough information somewhere to get the info
>>> to build the dynamical matrix, diagonalize it, and get the mode frequencies
>>> and atomic displacements, or if I am doing something wrong in my
>>> calculations.  Thanks for any help/insight.
>>> James Johns
>>>
>>>
>>> PHonon input file::
>>> phonons of Mo6Te6 at Gamma as Bulk Metal
>>>  &inputph
>>>   tr2_ph=1.0d-12,
>>>   prefix='Mo6Te6_Xtal_fixed_tight2_snug',
>>>   epsil=.false.,
>>>   outdir='/home/johnsj/jjohns/Mo6Te6',
>>>   fildyn='Mo6Te6.fixed_symm_snug.dynG',
>>>   verbosity='high',
>>>   alpha_mix(1)=0.4,
>>>   max_seconds=86300,
>>>   recover=.true.
>>>  /
>>> 0.0 0.0 0.0
>>>
>>> --
>>> James E. Johns
>>> Assistant Professor
>>> Department of Chemistry
>>> University of Minnesota
>>> 207 Pleasant St.
>>> Minneapolis, MN 55455
>>> http://www.chem.umn.edu/groups/johns/index.html
>>> _______________________________________________
>>> users mailing list
>>> users at lists.quantum-espresso.org
>>> https://lists.quantum-espresso.org/mailman/listinfo/users
>>
>>
>>
>> --
>> 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
>>
>> _______________________________________________
>> users mailing list
>> users at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/users
>
>
>
> --
> James E. Johns
> Assistant Professor
> Department of Chemistry
> University of Minnesota
> 207 Pleasant St.
> Minneapolis, MN 55455
> http://www.chem.umn.edu/groups/johns/index.html
>

-- 
James E. Johns
Assistant Professor
Department of Chemistry
University of Minnesota
207 Pleasant St.
Minneapolis, MN 55455
http://www.chem.umn.edu/groups/johns/index.html
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20190131/a30a1387/attachment.html>