[Pw_forum] Noncollinear magnetism

Andrea Dal Corso dalcorso at sissa.it
Fri Jul 1 09:48:07 CEST 2005


On Thu, 2005-06-30 at 19:48 -0700, W. YU wrote:
> Dear Pw users,
> 
> I have several questions about non collinear magnetism
> calculation as shown in example13.
> 
> The following is quoted from the calculation for Fe
> with penalty functional: 
> -----------------------------------------------------
> # self-consistent calculation with penalty functional
>  &system
>     ibrav = 3, celldm(1) =5.217, nat= 1, ntyp= 1,
>     ecutwfc = 25.0,ecutrho = 200.0,
>     report=1,
>     occupations='smearing', smearing='gaussian',
> degauss=0.05
>     noncolin = .true.
>     starting_magnetization(1) = 0.5
>     angle1(1) = 90.0
>     angle2(1) =  0.0
>     mcons(1,1) = 0.0
>     mcons(2,1) = 0.5
>     mcons(3,1) = 0.0
>     i_cons = 1
>     lambda = 1
>  /
> =================================================
> I have these questions:
> 
> 1.From the input menu, I know the mcons(i,j) is the
> i'th component of j'th atom type where i runs from 1
> to 3. But if angle1(1)=90.0 and angle2(1)=0.0, the
> magnetic moment should be along X axis, but the
> constraint seems to leave the Y axis unchanged is 2
> represents Y axis. It seems that 1,2 and 3 represent
> Z, X and Y axses, am I right?

You are right: these examples should be updated. 
Some problems have been already noticed and updated in the cvs 
version (where however the input variables for the constraint 
have been changed).
For the present 2.1.3, or 2.1.4 version of PWscf:

1,2,3 are X,Y,Z. angle1(1) and angle2(1) are the theta 
and phi angles of the
starting magnetization, which can be different from those
of the magnetization specified by the constraint.
In this particular example, starting with a magnetization
along x, the magnetization remains along x for symmetry reasons
The example can be corrected starting with a magnetization which
has a component along y to break the initial the symmetry,
for instance angle2(1)=10...

> ---------------------------------------------------
> 2. I hope to know how to choose lambda. why it is 1
> here in the example.

...and it is better to increase lambda.

> ===================================================
> The following is quoted from the calcualtion fo Ni:
> 
>  # self-consistent calculation
> cat > ni.scf.in << EOF
>  &control
>     calculation='scf'
>     restart_mode='from_scratch',
>     pseudo_dir = '$PSEUDO_DIR/',
>     outdir='$TMP_DIR/'
>     prefix='ni'
>  /
>  &system
>     ibrav=2, celldm(1) =6.48, nat=1, ntyp=1,
>     starting_magnetization(1)=0.7,
>     ecutwfc = 24.0, ecutrho = 288.0,
>     occupations='smearing',
> smearing='methfessel-paxton', degauss=0.02
>     noncolin = .true.
>     starting_magnetization(1) = 0.1
>     angle1(1) = 90.0
>     angle2(1) =  0.0
>  /
> ====================================================
> 3. Now the question is: Why there are two starting
> manetisation when there is only one type of Ni atom?

This will be updated, thank you for noticing it.


> 4. On what basis do we choose smearing scheme such as
> "methfessel-paxton" or "Gaussian"?

Both should work with the noncollinear version.
There is no particular reason to use Gaussian as in
the example.


> 5. Last but most important question is: can we do
> calculations on spiral? If the answer is positive, do
> we have to use a very large supper cell? 

Presently you need a supercell.
 
> ======================================================
> 
> Your answers and comments would be much a preciated.
> 
> W. YU
> 
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around 
> http://mail.yahoo.com 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
-- 
Andrea Dal Corso                    Tel. 0039-040-3787428
SISSA, Via Beirut 2/4               Fax. 0039-040-3787528
34014 Trieste (Italy)               e-mail: dalcorso at sissa.it




More information about the users mailing list