[QE-users] unconverged issue in spin-polarized SCAN calculations

[Giuseppe Mattioli]

Dear Yi-min Ding
Supposing that the pristine silicon nitride cell converges without  
problems, try first to use smearing (uncomment the lines) and to  
reduce mixing_beta to a low or very low value (0.1-0.01). I don't know  
silicon nitride very well, so excuse me for the possibly stupid  
question. Have you tried to use a larger supercell? 33% Al doping  
might be not only unphysical, but also the source of wild instability  
in scf iterations.
HTH
Giuseppe

Quoting 丁以民 via users <users at lists.quantum-espresso.org>:

> Dear all,
> I am doing a scf calculation for a FM system using SCAN functional.  
> But, the energy in output file is positive, and the run shown no  
> converged tendency.
>
>
> The input and out files are shown below:.
> input:
> &system
>    ibrav =4
>    a=2.9063917063813576
>    c=25
>    nat = 7
>    ntyp = 3
>    ecutwfc = 100.0
>    nbnd=48
>    nspin=2
>    ! starting_magnetization(1)= 0,
>    ! starting_magnetization(2)=0,
>    ! starting_magnetization(3)=1,
>      tot_magnetization =1
>    ! occupations='smearing',
>    ! smearing='gauss',
>    ! degauss=1.0d-2,
>     input_dft='scan'
> /
> &electrons
>    electron_maxstep = 100
>    conv_thr = 1.0d-8
>    mixing_mode = 'plain'
>    mixing_beta = 0.7
>    mixing_ndim = 8
>    diagonalization = 'david'
>    diago_david_ndim = 4
>    diago_full_acc = .true.
> /
> ATOMIC_SPECIES
>    Al  26.98  Al.upf
>    Si  28.09   Si.upf
>    N   14.01   N.upf
> ATOMIC_POSITIONS (crystal)
> Al 0.6666666666666643  0.3333333333333357   
> 0.5000000000000000   
> Si  0.0000000000000000  0.0000000000000000  0.3860058177698406
> Si  0.3333333429999996  0.6666666870000029   
> 0.6139941822301592 
> N  0.0000000000000000  0.0000000000000000   
> 0.4563348732468989 
> N  0.3333333429999996  0.6666666870000029   
> 0.5436651267531011 
> N  0.3333333429999996  0.6666666870000029   
> 0.3651769401128088 
> N 0.0000000000000000 -0.0000000000000000  0.6348230598871913 
> K_POINTS automatic
>  11 11 1 0 0 0
>
> .........................................................................................................
> output:
>      ...............
>      iteration # 99     ecut=   
>  100.00 Ry     beta= 0.70
>      Davidson diagonalization with overlap
>      c_bands:  3 eigenvalues not converged
>      c_bands:  5 eigenvalues not converged
>      c_bands:  5 eigenvalues not converged
>      c_bands:  5 eigenvalues not converged
>      c_bands:  5 eigenvalues not converged
>      c_bands:  5 eigenvalues not converged
>      ethr =  1.00E-02,  avg # of iterations = 24.8
>
>
>      negative rho (up,down):  5.052E+00 2.651E+00
>
>
>      total cpu time spent up to now is     
>  1964.0 secs
>
>
>      total energy           
>     =    3309.58902363 Ry
>      estimated scf accuracy    <   
> 304153.85545920 Ry
>
>
>      total magnetization       
>  =     1.00 Bohr mag/cell
>      absolute magnetization    =   
>   28.84 Bohr mag/cell
>
>
>      iteration #100     ecut=   
>  100.00 Ry     beta= 0.70
>      Davidson diagonalization with overlap
>      ethr =  1.00E-02,  avg # of iterations = 12.3
>
>
>      negative rho (up,down):  3.516E+00 2.027E+00
>
>
>      total cpu time spent up to now is     
>  1974.9 secs
>
>
>      total energy           
>     =    2760.95716210 Ry
>      estimated scf accuracy    <   
> 488812.03105752 Ry
>
>
>      total magnetization       
>  =     1.00 Bohr mag/cell
>      absolute magnetization    =   
>   30.68 Bohr mag/cell
>
>
>      End of self-consistent calculation
>
>
>      convergence NOT achieved after 100 iterations: stopping
>      Writing meta-gga kinetic term
>
> ...............................................................................................................
> Can you give me some help?
> Thanks in advance.
> Dr. Yi-min Ding



GIUSEPPE MATTIOLI
CNR - ISTITUTO DI STRUTTURA DELLA MATERIA
Via Salaria Km 29,300 - C.P. 10
I-00015 - Monterotondo Scalo (RM)
Mob (*preferred*) +39 373 7305625
Tel + 39 06 90672342 - Fax +39 06 90672316
E-mail: <giuseppe.mattioli at ism.cnr.it>