[Pw_forum] Fwd: pwscf: energy convergence of noncollinear with spin-orbit calculations, withdraw question

Gabriele Sclauzero sclauzer at sissa.it
Tue Jun 14 09:41:10 CEST 2011


In the first message you were speaking about fixing the total magnetization, here you do something different. Anyway I'm happy for you that you could converge the calculations. 
It's a bit curious that you get converge only with nosym=true. What is the symmetry found by the code at the beginning of the run?

Il giorno 13/giu/2011, alle ore 21.39, German Samolyuk ha scritto:

> Hello Gabriele,
> 
> Thank you for your response and advices.
> 
> I used version 4.2.1
> 
> the actual (final) input contains instructions:
> 
> for magnetization along z-direction
> 
> starting_magnetization(1) = 0.1,
> starting_magnetization(2) = 0.4,
>          noncolin = .true. ,
>         angle1(1) = 0,
>         angle1(2) = 0,
>         angle2(1) = 0,
>         angle2(2) = 0,
>          lspinorb = .true. ,
>   constrained_magnetization = 'atomic direction' ,
>            lambda = 0.3,
>            nosym = .true.,
> 
> and magnetization perpendicular to z-direction
> 
> starting_magnetization(1) = 0.1,
> starting_magnetization(2) = 0.4,
>          noncolin = .true. ,
>         angle1(1) = 90,
>         angle1(2) = 90,
>         angle2(1) = 0,
>         angle2(2) = 0,
>          lspinorb = .true. ,
>   constrained_magnetization = 'atomic direction' ,
>            lambda = 0.3,
>            nosym = .true.,
> 
> the resulting energies  looks reasonable and convergence is really fast.

The constraint (penalty) energy should be almost zero at convergence. Could you obtain that with such a small lambda?


Regards,

Gabriele


> 
> Thank you for advice with degauss value. I'll check dependence from it.
> 
> Best,
> 
> German
> 
> 
> On Wed, Jun 8, 2011 at 3:55 AM, Gabriele Sclauzero <sclauzer at sissa.it> wrote:
>> Hello German,
>> Il giorno 01/giu/2011, alle ore 23.02, German Samolyuk ha scritto:
>> 
>> I withdraw my question. It works with nosym=true.
>> German
>> 
>> 
>> ---------- Forwarded message ----------
>> From: German Samolyuk <samolyuk at gmail.com>
>> Date: Wed, May 18, 2011 at 10:25 AM
>> Subject: pwscf: energy convergence of noncollinear with spin-orbit
>> calculations
>> To: PWSCF Forum <pw_forum at pwscf.org>
>> 
>> 
>> Dear qe users,
>> 
>> I'm trying to calculate total energy of artificial FCC PtFe compound
>> with magnetic moment (MM) along z- and x-directionsi (MAE).
>> 
>> In order to supress oscillation of moment direction I used for MM||z
>> calculation
>> input instructions
>> 
>>  constrained_magnetization = 'total direction' ,
>>   fixed_magnetization(1) = 0.0,
>>   fixed_magnetization(2) = 0.0,
>>   fixed_magnetization(3) = 1.0,
>> 
>> 
>> If it is v.>4.2, then a
>> Are you sure that this is what you want?
>> From the Doc/INPUT_PW:
>>                    'total direction':
>>                              the angle theta of the total magnetization
>>                              with the z axis (theta =
>> fixed_magnetization(3))
>>                              is constrained:
>> 
>>                              LAMBDA * ( arccos(magnetization(3)/mag_tot) -
>> theta )**2
>> 
>>                              where mag_tot is the modulus of the total
>> magnetization.
>> Then fixed_magnetization(3) = 1.0 does not correspond to the magnetization
>> along the z axis (but you should reach the corresponding constrained
>> solution anyway, even if different from what you actually want).
>> Can you please specify which version of the code you've been using?
>> Could you try to compare v.4.1.1 with v.4.3.1 (using the correct value for
>> theta=fixed_magnetization(3))
>> The constraining potential implemented is different: in the most recent
>> version it should work also for theta=90° (magnetization along x), while in
>> the previous you could get into troubles, since it was defined as:
>>                              LAMBDA * ( magnetization(1) -
>> magnetization(3)*tan(theta) )**2
>> 
>> Also convergency should be better in the 4.3.1, though I'm not sure of that.
>> You should try to start with local magnetizations on Pt and Fe close to the
>> desired target direction. It would be nice if you perform this test.
>> Also, reporting only the total energy oscillations is not that useful. You
>> should at least report the magnitude and angles of the total magnetization
>> while the calculation is going.
>> Further advice: from your input, perhaps degauss=0.02 is not enough to get
>> accurate magnetizations. I would use at least 0.01, better 0.005.
>> 
>> HTH
>> 
>> GS
>> 
>> However I can not reach total energy convergence and it looks like I
>> still have direction oscillation of MM.
>> 
>> What can be done to resolve this problem?
>> 
>> Thank you in advance.
>> 
>> The complete input file for MM||z is and energy accuracy output file
>> are attached.
>> 
>> German Samolyuk,
>> ORNL, USA
>> <ptfe.pw.in><accu.out>_______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://www.democritos.it/mailman/listinfo/pw_forum
>> 
>> 
>> § Gabriele Sclauzero, EPFL SB ITP CSEA
>>    PH H2 462, Station 3, CH-1015 Lausanne
>> 
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://www.democritos.it/mailman/listinfo/pw_forum
>> 
>> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum


§ Gabriele Sclauzero, EPFL SB ITP CSEA
   PH H2 462, Station 3, CH-1015 Lausanne

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20110614/a3a56780/attachment.html>


More information about the users mailing list