[Pw_forum] phonon calculation not converge

Fri Oct 30 14:30:46 CET 2015

Hi, Xiaolin, I think you may want to support your colleagues' observations
in Raman, actually you can use the phonon spectrum to explain what you have
onserved given by phonon calculations in pwscf.
However, you must use more grid to produce quite accurate frequencies.
Please also take care of the convergence of the grid to which some
materials are very sensitive.

Bests
Fang

2015-10-30 20:38 GMT+08:00 Jiqiang Li <jqli14 at fudan.edu.cn>:

> Hi, Xiaolin,
>         Actually, you can use the gamma-only specific algorithm. N.B., you
> should use phcg.x instead of ph.x for the phonon calculation.
>
> Jiqiang
>
> ------------------------------
>
>
> *From:* Xiaolin Xu <xiaolin0314 at gmail.com>
> *Date:* 2015-10-30 13:37
> *To:* Ari Paavo Seitsonen <Ari.P.Seitsonen at iki.fi>; PWSCF Forum
> <pw_forum at pwscf.org>
> *Subject:* Re: [Pw_forum] phonon calculation not converge
> Dear Ari and Jiqiang,
>
> Thank you so much for all the instructions! Yes I ran pw.x using K_POINTS
> automatic // 1 1 1  0 0 0 before running ph.x. I will increase the
> cut-off energy and hope it will work!
>
> Lin
>
> On Thu, Oct 29, 2015 at 3:34 PM, Ari P Seitsonen <Ari.P.Seitsonen at iki.fi>
> wrote:
>
>>
>> Dear Xiaolin, Jiqiang et co,
>>
>>   Did you re-run 'pw.x' using the 'K_POINTS automatic // 1 1 1  0 0 0'
>> before trying to run 'ph.x'?
>>
>>   On the other hand the number of k points depends on your system, the
>> amount of band dispersion it etc. So indeed, if the electronic structure is
>> not yet converged well with only Gamma point, it is better to apply more k
>> points. This you can do by testing for the convergence.
>>
>>   By the way, the cut-off energy of 80 Ry is probably far too low for the
>> HGH/GTH pseudo potentials that you have used: We used 125 Ry for the oxygen
>> - I would guess that the one for iron, maybe also lithium - and that was in
>> molecular dynamics; for phonons I would expect even higher a value
>> necessary.
>>
>>     Greetings,
>>
>>        apsi
>>
>>
>> -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
>>   Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
>>   Ecole Normale Supérieure (ENS), Département de Chimie, Paris
>>   Mobile (F) : +33 789 37 24 25    (CH) : +41 79 71 90 935
>>
>>
>>
>>
>>> 在 2015-10-30 00:26:09，"Xiaolin Xu" <xiaolin0314 at gmail.com> 写道：
>>>
>>> Hi Jiqiang,
>>>>
>>>>
>>>> Thank you very much for your help!
>>>> I tried using only gamma point, but I got an error message when I
>>>> calculated phonon saying "can't start from gamma tricks...". Besides, I
>>>> need to calculate the raman frequencies after phonon calculation, so I
>>>> think it's better if I use more k points.
>>>>
>>>> Have you tried Raman calculation? Could you give me more advice?
>>>>
>>>> Best,
>>>>
>>>> Xiaolin
>>>>
>>>>
>>>> On Wed, Oct 28, 2015 at 10:21 PM, Xiaolin Xu <xxu2 at caltech.edu> wrote:
>>>>
>>>> Hi Jiqiang,
>>>>
>>>> Thank you very much for your help!
>>>> I tried using only gamma point, but I got an error message when I
>>>> calculated phonon saying "can't start from gamma tricks...". Besides, I
>>>> need to calculate the raman frequencies after phonon calculation, so I
>>>> think it's better if I use more k points.
>>>>
>>>> Have you tried Raman calculation? Could you give me more advice?
>>>>
>>>> Best,
>>>>
>>>> Xiaolin
>>>>
>>>>
>>>> On Wed, Oct 28, 2015 at 6:20 PM, jqli14 <jqli14 at fudan.edu.cn> wrote:
>>>> (1)employ k_points gamma instead of k_pionts automatic 1 1 1 0 0 0,
>>>> although both give the same result, but the later does not call the gamma
>>>> point specific algorithm to speed the calculation.
>>>> (2)if it can not converge within handreds of electronic steps, reduce
>>>> the value of mixing_beta or change the mixing_mode.
>>>>
>>>> Jiqiang Li
>>>>
>>>>
>>>> 在 2015-10-29 07:57:58，"Xiaolin Xu" <xiaolin0314 at gmail.com> 写道：
>>>>
>>>> Dear all,
>>>>>
>>>>> I was doing phonon dispersion calculations for lithium iron phosphate.
>>>>> However, the calculation speed is really slow and it does not converge. Is
>>>>> this because my system is too big? There are 28 atoms in the unit cell. My
>>>>> input file is like this:
>>>>> (1)
>>>>>
>>>>>
>>>>> &CONTROL
>>>>> calculation = "scf",
>>>>> pseudo_dir = "/home/xiaolin/pseudo",
>>>>> outdir = "/home/xiaolin/test2",
>>>>> tstress=.true.,
>>>>> tprnfor=.true.,
>>>>> nstep=100,
>>>>> /
>>>>>
>>>>> &SYSTEM
>>>>> ibrav = 8,
>>>>> celldm(1) = 19.749526
>>>>> celldm(2) = 0.582432,
>>>>> celldm(3) = 0.454693,
>>>>> nat = 28,
>>>>> ntyp = 4,
>>>>> ecutwfc = 80.D0,
>>>>> /
>>>>>
>>>>> &ELECTRONS
>>>>> mixing_beta = 0.7,
>>>>> /
>>>>>
>>>>> &IONS
>>>>> /
>>>>>
>>>>> ATOMIC_SPECIES
>>>>>
>>>>> Li 6.941 Li.pbe-s-hgh.UPF
>>>>> Fe 55.845 Fe.pbe-sp-hgh.UPF
>>>>> P 30.97376 P.pbe-hgh.UPF
>>>>> O 15.9994 O.pbe-hgh.UPF
>>>>> ATOMIC_POSITIONS (alat)
>>>>> Li 0 0 0
>>>>> ...
>>>>> K_POINTS (automatic)
>>>>> 1 1 1 0 0 0
>>>>>
>>>>> (2)
>>>>>
>>>>> Normal modes for lifepo4
>>>>> &inputph
>>>>> tr2_ph=1.5d-5,
>>>>> amass(1)=6.941,
>>>>> amass(2)=55.845,
>>>>> amass(3)=30.97376,
>>>>> amass(4)=15.9994,
>>>>> outdir = '/home/xiaolin/test2',
>>>>> trans=.true.,
>>>>> asr=.true.,
>>>>> nogg=.true.,
>>>>> fildyn='dmat.lifepo4'
>>>>> /
>>>>> 0.0 0.0 0.0
>>>>>
>>>>>
>>>>>
>>>>> Thank you in advance!
>>>>>
>>>>>
>>>>> Lin
>>>>>
>>>>
>>>>
>>>>
>>>>
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://pwscf.org/mailman/listinfo/pw_forum
>>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>

-- 
------------------------------------------------------------------------------------------------------------
Yue-Wen FANG, PhD candidate
East China Normal University  <http://english.ecnu.edu.cn/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20151030/7dd728d6/attachment.html>