[Pw_forum] "q mesh breaks symmetry" error even with q mesh being the same as k mesh

Thomas Brumme thomas.brumme at mpsd.mpg.de
Tue Aug 9 13:04:32 CEST 2016


OK, got it... That's why it worked when I chose nr1=nr2=nr3=128 - the 
code uses the symmetries including translations in this case as nr1=nr3. 
Similarly I should choose at least nk1=nk3, but as the 1st and the 2nd 
lattice vectors are nearly the same length I should probably use 
nk1=nk2=nk3.

Thanks

Thomas


On 08/09/2016 12:53 PM, stefano de gironcoli wrote:
> it does not look so much longer than the other to me ...
> Anyhow the point is that a grid in order to respect the symmetry must
> transform into itself under symmetry operations and yours does not
> satisfy this condition for reflexion x -> -x, z -> -z and rotations of
> pi around z ...
>
> unless nk1=nk3
>
> for the fft grid ... I would try to keep nr1=nr3 for the same reason.
> actually my suggestion was based on a real-space argument but I'm pretty
> sure it is valid in reciprocal space too.
>
> stefano
>
> On 09/08/2016 11:34, Thomas Brumme wrote:
>> Thanks for the suggestion Stefano!
>>
>> Yet, lattice vector 3 is in my cell the largest and correspondingly nk3
>> must be larger than nk1=nk2, or? That's why I choose 10 10 6 (even if I
>> might have to increase nk3 slightly).
>>
>> Or do I miss something? I meanwhile also played with the cutoff of the
>> wave functions (i.e. finding also an fft mesh which can be used with the
>> symmetries which include a translation) but even if the fft mesh is
>> 96x96x120 the symmetries which include the translation are neglected.
>> I'll try with 96x96x128...
>>
>> Thomas
>>
>> On 08/09/2016 02:59 AM, stefano de gironcoli wrote:
>>> I think that given your Bravais lattice nk3 must be equal to nk1
>>> so
>>> 10 nk2 10 0 0 0
>>> or
>>> 6 nk2  6 0 0 0
>>> should be fine
>>>
>>> 10 nk2 6 0 0 0
>>> is not
>>>
>>> the automatic unfolding of a grid generate additional points but this
>>> does not mean they form a regular grid. think about what happens for an
>>> hexagonal lattice with a shifted grid.
>>>
>>> stefano
>>>
>>> On 08/08/2016 17:51, Thomas Brumme wrote:
>>>> Dear all,
>>>>
>>>> I want to calculate the phonon frequencies on a regular q mesh specified
>>>> by using
>>>> ldisp=.true. and the nq1, nq2, nq3 variables. Yet, I always get the error
>>>>
>>>> q-mesh breaks symmetry
>>>>
>>>> even if the grid has the same dimensions as the k mesh in the PWscf
>>>> calculation.
>>>> This is weird as the q mesh is exactly the same as the k mesh, which was
>>>> created
>>>> by applying the symmetries to a regular grid. The system is LCO and the
>>>> input for
>>>> PWscf (in the reduced cell) is:
>>>>
>>>> &control
>>>>       calculation   = 'scf',
>>>>       restart_mode  = 'from_scratch',
>>>>       prefix        = 'LCO',
>>>>       pseudo_dir    = './',
>>>>       outdir        = './tmp/',
>>>>       nstep         = 300,
>>>>       wf_collect    = .true.,
>>>> /
>>>> &system
>>>>       ibrav       = 0,
>>>>       nat         = 14,
>>>>       ntyp        = 3,
>>>>       ecutwfc     = 200,
>>>>       occupations = 'smearing',
>>>>       smearing    = 'gauss',
>>>>       degauss     = 0.01,
>>>>       nspin       = 2,
>>>>       starting_magnetization(3) = 0.02,
>>>>       nr1 = 128,
>>>>       nr2 = 128,
>>>>       nr3 = 128,
>>>> /
>>>> &electrons
>>>>       electron_maxstep = 250,
>>>>       diagonalization  = 'cg',
>>>>       conv_thr         = 1.0d-10,
>>>> /
>>>> CELL_PARAMETERS (angstrom)
>>>>         5.261112503   0.000000000   0.000000000
>>>>         0.000000000   5.330947322   0.000000000
>>>>        -2.630556252  -0.000000000   6.548827231
>>>> ATOMIC_SPECIES
>>>>       La   138.90547   La1.UPF
>>>>       O    15.999   O.pz-n-mt.UPF
>>>>       Cu  63.546   Cu1.UPF
>>>> K_POINTS automatic
>>>>       10 10 6 0 0 0
>>>> ATOMIC_POSITIONS (crystal)
>>>> ...
>>>>
>>>> I defined the FFT grid by hand since then the code does not drop the
>>>> symmetries
>>>> including fractional translations... Could this be the problem? Or that
>>>> I did choose
>>>> a FFT grid which has the same number of points in x/y/z directions? Or
>>>> could this
>>>> be due to a similar problem" as in hexagonal crystals where shifting the
>>>> k mesh
>>>> away from Gamma is a bad idea? Or is there an obvious error in my input
>>>> for ph.x:
>>>>
>>>>       &inputph
>>>>        tr2_ph   = 1.0d-18,
>>>>        prefix   = 'LCO',
>>>>        amass(1) = 138.90547,
>>>>        amass(2) = 15.999,
>>>>        amass(3) = 63.546,
>>>>        outdir   = './tmp/',
>>>>        fildyn   = 'LCO.dynG',
>>>>        ldisp=.true.,
>>>>        nq1=10, nq2=10, nq3=6,
>>>>        fildvscf = 'LCO.dvscf',
>>>>        verbosity = 'high',
>>>>       /
>>>>
>>>> I searched the archive however I couldn't find a solution to my problem.
>>>> Thus, any
>>>> help would be very much appreciated :)
>>>>
>>>> Regards
>>>>
>>>> Thomas
>>>>
>>> _______________________________________________
>>> 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

-- 
Dr. rer. nat. Thomas Brumme
Max Planck Institute for the Structure and Dynamics of Matter
Luruper Chaussee 149
22761 Hamburg

Tel:  +49 (0)40 8998 6557

email: Thomas.Brumme at mpsd.mpg.de




More information about the users mailing list