[Pw_forum] Convergence Problem

[孙阳]

Hi, Dr.A.Zhako Titarenko and other users,
You mean the mixing betta. My value for mixing betta is 0.05, is it still
too high?

Thank you in advance.

Yang

2009/4/15 AZGlobal <azgre at azgre.com>:
> Usually oscillating  SCF process is result of some flat band  (lots of
> low-dispersion bands in Pd near Ef) flipping below/above Ef from
> iteration to iteration, when you iterate. The cure is smaller admixing.
> Takes much more processor time, but you can iterate for weeks with too
> big admix and no result. Hope this helps.
>> Dear All,
>>
>> When I do the calculation of the electronic structure of slab
>> Palladium under the external electric field,
>> convergence problems appear. The typical problem is that the estimated
>> scf accuracy value drop down
>> down and bounce back. It seems that the calculation does not know
>> where the system would go for
>> minimizing the total energy. The electric field that I applied is
>> saw-tooth like one. I tried many things like
>> increasing the mixing_ndim, decreasing the mixing_betta to 0.1, adding
>> empty bands changing the mixing
>> mode and even increase the degauss. But, none of them would work for
>> the scf accuracy oscillating problem.
>> Below is the input file. Any suggestion is highly appreciated.
>>
>>
>>
>>
>> &control
>>     calculation='scf'
>>     restart_mode='from_scratch',
>>     pseudo_dir = '$PSEUDO_DIR/',
>>     outdir='$TMP_DIR/'
>>     prefix='Pd'
>>     tefield = .true.
>>     dipfield = .true.
>>     nstep = 300
>>
>>  /
>>  &system
>>     ibrav = 6
>>     celldm(1) = 7.3331
>>     nspin = 2
>>     celldm(3) = 4.5
>>     edir = 3
>>     emaxpos=0.72
>>     eopreg=0.28
>>     eamp = -0.001944
>>     starting_magnetization=0.05
>>     ntyp = 1
>>     nat= 14
>>     ecutwfc = 25.0,
>>     ecutrho = 500.0,
>>     occupations='smearing',
>>     degauss=0.2
>>     nbnd=160
>>  /
>>
>>  &electrons
>>     conv_thr = 1.0e-4
>>     mixing_mode='local-TF'
>>     mixing_beta = 0.05
>>     mixing_ndim = 12
>>
>>  /
>> ATOMIC_SPECIES
>>  Pd  100  Pd.pz-nd-rrkjus.UPF
>> ATOMIC_POSITIONS
>> Pd       0.500000000   0.000000000  -1.496126804
>> Pd       0.000000000   0.500000000  -1.496126804
>> Pd       0.000000000   0.000000000  -1.002542052
>> Pd       0.500000000   0.500000000  -1.002542051
>> Pd       0.500000000   0.000000000  -0.501470158
>> Pd       0.000000000   0.500000000  -0.501470158
>> Pd       0.000000000   0.000000000   0.000000000
>> Pd       0.500000000   0.500000000   0.000000000
>> Pd       0.500000000   0.000000000   0.501470158
>> Pd       0.000000000   0.500000000   0.501470158
>> Pd       0.000000000   0.000000000   1.002542052
>> Pd       0.500000000   0.500000000   1.002542051
>> Pd       0.500000000   0.000000000   1.496126804
>> Pd       0.000000000   0.500000000   1.496126804
>>
>>
>>
>>
>> K_POINTS (automatic)
>>  8 8 1 0 0 0
>>
>>
>>
>>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>



-- 
University of Nebraska at Lincoln