[Pw_forum] imaginary frequency at extended Q point

Shuai zhao-shuai at edu.life.kyutech.ac.jp
Tue Oct 20 09:41:10 CEST 2015


Dear Tanlin,

I cannot give you suggestions but I have a confusion that you relaxed 
the structure under 120 Ry of ecutwfc but for the SCF calculation you 
used 50 Ry.

I guess the total energy for 50 Ry SCF is NOT lowest (to some extent) 
with the structure relaxed by 120 Ry. I am not sure whether it will 
finally impact the result. Why you reduce the Ecut so significantly?

Best wishes,

Shuai

On 10/20/2015 04:19 PM, Lorenzo Paulatto wrote:
> Dear 潭影空人心,
> the imaginary frequencies could just be an artifact of the Fourier
> interpolation procedure, which means that you need to do the ph.x calculation
> on a finer grid.. If that is not possible, your best bet is to try to push
> convergence a bit and try to change the method of acoustic sum rule used.
>
> hth
>
> On Tuesday, October 20, 2015 09:36:53 AM 潭影空人心 wrote:
>> Dear all,   I used nq1 = 2, nq2 = 2, nq3 = 2 and nq1 = 4, nq2 = 4, nq3 = 4
>> ,respectively, to do a phonon calculation. There is no imaginary frequency
>> at the calculated Q points, but after calculations of Q2R and MATDYN,
>> imaginary frequencies appear near GAMMA point(not GAMMA point).
>   
>> My settings are as follows:
>> 1.lattice relaxation under pressure:
>> &control
>>      calculation               = 'vc-relax'
>>         
>   restart_mode              = 'from_scratch'
>>      prefix                    = 'cras'
>>             
>   pseudo_dir                =
>> '/data1/tan1/software/espresso-5.1/pseudo/' outdir                    =
>> './tempdir/'
>>      etot_conv_thr             = 1.0D-6
>>      forc_conv_thr             = 1.0D-4
>>      tstress                   = .true.
>>      tprnfor                   = .true.
>>           
>   / &system ibrav                     = 8
>>      A                         = 5.58
>>      B                         = 3.36
>>      C                         = 6.17
>>      nat                       = 8
>>      ntyp                      = 4
>>      ecutwfc                   = 120
>>      occupations               = 'smearing'
>>      smearing                  = 'methfessel-paxton'
>>      degauss                   = 0.04
>>      nspin                     = 2
>>      starting_magnetization(1) = -1
>>      starting_magnetization(2) = 1
>>      starting_magnetization(3) = 0
>>      starting_magnetization(4) = 0
>> /
>> &electrons
>>      electron_maxstep          = 100
>>      conv_thr                  = 1.0d-6
>>      mixing_beta               = 0.7
>>      
>   /
>> &ions
>>      ion_dynamics             = 'bfgs'
>> /
>> &cell
>>     press                     = 1500
>>     cell_dynamics             = 'bfgs'
>>     press_conv_thr            = 0.1
>>     cell_factor               = 2.5d0
>> /
>> ATOMIC_SPECIES
>>     Cr1   51.9961       Cr.pz-hgh.UPF
>>     Cr2   51.9961       Cr.pz-hgh.UPF
>>     As1   74.9216       As.pz-hgh.UPF
>>     As2   74.9216       As.pz-hgh.UPF
>> ATOMIC_POSITIONS (crystal)
>> Cr1   0.012000      0.250000      0.201000  1 0 1
>> Cr2   0.988000      0.750000      0.799000  1 0 1
>> Cr2   0.512000      0.250000      0.299000  1 0 1
>> Cr1   0.488000      0.750000      0.701000  1 0 1
>> As1   0.195000      0.250000      0.582000  1 0 1
>> As2   0.805000      0.750000      0.418000  1 0 1
>> As2   0.695000      0.250000      0.918000  1 0 1
>> As1   0.305000      0.750000      0.082000  1 0 1
>> K_POINTS {automatic}
>>            
>   6 8 6 0 0 0
>>
>>
>>
>>
>> 2.interior coordination relaxation:
>> All is the same as above, but "ecutwfc = 50"
>> 3.self-consistent iteration for phonon calculation:
>> &control
>> calculation               = 'scf'
>>      restart_mode              = 'from_scratch'     prefix
>>   = 'cras'                                                       pseudo_dir
>>                = '/data1/tan1/software/espresso-5.1/pseudo/'     outdir
>>                = './tempdir/'      tstress                   = .true.
>> tprnfor                   = .true.
>>         /
>>       &system
>>         ibrav                     = 0     celldm(1)                 =
>> 10.54467182     nat                       = 8     ntyp
>> = 4     ecutwfc                   = 50     occupations               =
>> 'smearing'     smearing                  = 'methfessel-paxton'     degauss
>>                   = 0.05     nspin                     = 2
>> starting_magnetization(1) = -1     starting_magnetization(2) = 1
>> starting_magnetization(3) = 0     starting_magnetization(4) = 0 /
>>                                                          &electrons
>>                                                     electron_maxstep
>>   = 200     conv_thr                  = 1.0d-10     mixing_beta
>>   = 0.7                                               /
>> ATOMIC_SPECIES                                                         Cr1
>>   51.9961       Cr.pz-hgh.UPF    Cr2   51.9961       Cr.pz-hgh.UPF    As1
>> 74.9216       As.pz-hgh.UPF    As2   74.9216       As.pz-hgh.UPF
>> CELL_PARAMETERS (alat= 10.54467182)               0.9146 0.0000 0.0000
>> 0.0000 0.4765 0.0000 0.0000 0.0000 0.9628   ATOMIC_POSITIONS (crystal)
>>                    Cr1 0.0104 0.2500 0.1827  Cr2 0.9896 0.7500 0.8173  Cr2
>> 0.5104 0.2500 0.3173  Cr1 0.4896 0.7500 0.6827  As1 0.1833 0.2500 0.5616
>> As2 0.8167 0.7500 0.4384  As2 0.6833 0.2500 0.9384  As1 0.3167 0.7500
>> 0.0616  K_POINTS {automatic}
>>                      6 8 6 0 0 0
>> 4.self-consistent iteration for electron-phonon coupling:
>> All is the same as phonon calculation but k-mesh 12 16 12,and in addition to
>> "la2F = .true.".
>   5.phonon calculation:
>> &inputph   tr2_ph       = 1.0d-11   prefix       = 'cras'   fildvscf     =
>> 'crasdv'   amass(1)     = 51.9961   amass(2)     = 51.9961   amass(3)     =
>> 74.9216   amass(4)     = 74.9216   outdir       = './tempdir/',   fildyn
>>     = 'cras.dyn',   electron_phonon='interpolated'   el_ph_sigma  = 0.005,
>> el_ph_nsigma = 10,   trans        = .true.   ldisp        = .true.
>> alpha_mix(1) = 0.5   recover      =.true.   nq1          = 2   nq2
>> = 2   nq3          = 2  /
>   6.q2r:
>>   &input   zasr='crystal',  fildyn='cras.dyn', flfrc='cras222.fc',
>> la2F=.true.  /
>   7.&input     asr='crystal',
>> amass(1)=51.9961,amass(2)=51.9961,amass(3)=74.9216,amass(4)=74.9216,
>> flfrc='cras222.fc', flfrq='cras.freq', q_in_band_form=.true.,
>> la2F=.true.,dos=.false.  / 8 0.000000 	0.000000 	0.000000 	50 0.500000
>> 	0.000000 	0.000000 	50 0.500000 	0.785906 	0.000000 	50 0.000000 	
> 0.000000
>> 	0.000000 	50 0.000000 	0.000000 	0.456275 	50 0.500000 	0.000000 	
> 0.456275
>> 	50 0.500000 	0.785906 	0.456275 	50 0.000000 	0.000000 	0.456275 	1
>>
>>
>> I am eagerly hoping someone can help me. Thank you.
>> tanlin2015.10.20

-- 
PhD candidate
Graduate School of Life Science and Systems Engineering
Kyushu Institute of Technology, Japan




More information about the users mailing list