[Pw_forum] problem with vc-relax and spin-polarization

Stefano de Gironcoli degironc at sissa.it
Tue Jun 22 17:51:51 CEST 2010


I repeated your calculation on my computer and vc-relax was maintaining 
a finite value of
absolute magnetization till the end of the relaxation.
double check your results and in case you can confirm it for a recent 
version include your output and more information on the  run (parallel 
or serial,  machine, compiler, libraries...)

stefano

Bryan M. Wong wrote:
> Hi,
>
>      I seem to have a problem doing "vc-relax" with the
> "starting_magnetization" initialization turned on.
>
>      I can always do a regular "relax" with the starting_magnetization
> guess, and the structure will relax with a magnetic state constrained to my
> *fixed* unit cell dimensions. However, when I do a *vc-relax* with
> starting_magnetization set to any non-zero value, the 2nd step in the cell
> relaxation always reverts to the spin-unpolarized case (and the optimization
> will continue with the spin-unpolarized case).
>
>      I have attached an example of this behavior for a simple graphene
> nanoribbon.  After the first step, the nanoribbon is magnetic and nonzero,
> but the 2nd step reverts back to a non-magnetic initial condition.
>
>      Is there any way around this (without constraining the magnetization to
> a fixed value for all steps)?
>
> Thanks,
> Bryan
>
>
>
> &CONTROL
>                  calculation = 'vc-relax' ,
>                etot_conv_thr = 1.0e-4 ,
>                forc_conv_thr = 1.0e-3 ,
>                      tstress = .true. ,
>                      tprnfor = .true. ,
>  /
>  &SYSTEM
>                        ibrav = 8,
>                    celldm(1) = 4.644592221,
>                    celldm(2) = 6.70720764,
>                    celldm(3) = 4.393200375,
>                          nat = 10,
>                         ntyp = 4,
>                      ecutwfc = 60.D0 ,
>                        nosym = .false. ,
>                         nbnd = 28,
>                        nelec = 34,
>                   tot_charge = 0.000000,
>                  occupations = 'smearing' ,
>                      degauss = 0.02 ,
>                     smearing = 'gaussian' ,
>                        nspin = 2 ,
>    starting_magnetization(1) = 1.0,
>    starting_magnetization(2) = -1.0,
>    starting_magnetization(3) = 1.0,
>    starting_magnetization(4) = -1.0,
>  /
>  &ELECTRONS
>                     conv_thr = 1.D-6 ,
>                  mixing_beta = 0.7D0 ,
>  /
>  &IONS
>                 ion_dynamics = 'bfgs' ,
>            pot_extrapolation = 'second_order' ,
>            wfc_extrapolation = 'second_order' ,
>             trust_radius_ini = 0.2D0 ,
>  /
> ATOMIC_SPECIES
>    H1    1.00000  H.pbe-rrkjus.UPF
>    C1   12.00000  C.pbe-rrkjus.UPF
>    C2   12.00000  C.pbe-rrkjus.UPF
>    H2    1.00000  H.pbe-rrkjus.UPF
> ATOMIC_POSITIONS crystal
>    H1      0.000000000    0.218692155    0.000000000    1  1  0
>    C1      0.000000000    0.284883518    0.000000000    1  1  0
>    C2      0.500000000    0.326062773    0.000000000    1  1  0
>    C1      0.500000000    0.413734370    0.000000000    1  1  0
>    C2      0.000000000    0.456468822    0.000000000    1  1  0
>    C1      0.000000000    0.543530576    0.000000000    1  1  0
>    C2      0.500000000    0.586265119    0.000000000    1  1  0
>    C1      0.500000000    0.673937250    0.000000000    1  1  0
>    C2      0.000000000    0.715116935    0.000000000    1  1  0
>    H2      0.000000000    0.781308481    0.000000000    1  1  0
> K_POINTS automatic
>   13 1 1   0 0 0
>
>   
> ------------------------------------------------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>   




More information about the users mailing list