[Wien] Fixed spin moment calculation combined with spin scaling xc

[mazin]

Dear Peter,

It's been some time and I need to refresh my memory. The problem with 
implementing scale Exc was that the straightforward scaling of Exc was 
not variational with the scaling of vxc; I mean, taking variation of Exc 
was giving the scaled vxc.

at some point I reversed-engineered Exc that would yield, upon 
variation, the scaled vxc, and at that time I sent you my notes, but i 
need to gig them out now.

More recently, Kay Dewhurst showed that you van construct a much simpler 
variationally consistent scaling recipe for Exc, and they implemented 
that in ELK; I can find his notes, too, I think.

Best
Igor

On 5/16/2017 9:10 AM, Peter Blaha wrote:
> It was quite some time ago when I implemented the scaled Vxc.
>
> I looked again into the paper by Mazin and I'm not completely sure about
> his nomenclature about E_xc^P. If I assume that this term is just our
> (E_x(up)+E_x(dn)) and my "E_c" is his f(chi)*delta epsilon_xc(n), then
> spin scaling for the energy is trivial by multiplying the E_c term with
> the scaling parameter s.
>
> I've done this in the attached subroutine vxclm2.f in lapw0.
>
> You may test it and see if it gives reasonable results. However, no
> tests from my side yet. (Test eg. a non-spin-polarized calc with or
> without spin-scaling must give identical E-tot)
>
> Peter Blaha
>
> On 05/16/2017 02:25 PM, Peter Blaha wrote:
>> Yes and no.
>>
>> Technically I guess you can do it.
>>
>> However, the spin scaling is implemented only in the potential, i.e. the
>> spin-scaling will reduce the spin-polarized Vxc and thus the magnetic
>> moments. But with FSM you increase the EF-shifts, until the moment is
>> what you like.
>> Exc, which you usually need in an FSM calculation has not been
>> implemented with spin.scaling in wien2k.
>>
>> One should check the paper of Mazin and see how this should be
>> implemented. It should not be so difficult, but ...
>>
>> On 05/16/2017 12:30 PM, "Niedermayr, Arthur" wrote:
>>> Hello,
>>>
>>> I would really appreciate it if somebody could tell me if*_a fixed spin
>>> moment calculation combined with spin scaling_* is possible in WIEN2k.
>>>
>>> Best regards,
>>> Arthur Niedermayr
>>>
>>>
>>>
>>>> -------- Original Message --------
>>>> Subject: Fixed spin moment calculation combined with spin scaling xc
>>>> Local Time: 5. Mai 2017 6:59 PM
>>>> UTC Time: 5. Mai 2017 16:59
>>>> From: a.niedermayr at protonmail.com
>>>> To: A Mailing list for WIEN2k users <wien at zeus.theochem.tuwien.ac.at>
>>>>
>>>> Hello wien2k-users,
>>>>
>>>> I performed a LSDA calculation with spin scaling and got a reasonable
>>>> value for my magnetization (the scf cycle converged, no errors). [1]
>>>>
>>>> But if I try to do it the other way around, i.e. if I perform a fixed
>>>> spin moment calculation (with spin scaling parameter xc) for different
>>>> magnetization values, then I don't get a minimum in the total energy
>>>> at the received value of the magnetization in [1].
>>>>
>>>> My questions:
>>>> *_Can I combine a fixed spin moment calculation with spin scaling?_*
>>>> _If not, are there other possibilities to check the magnetization
>>>> dependence of the total energy?_
>>>>
>>>>
>>>>
>>>>
>>>> Best regards and thank you in advance!
>>>> Arthur Niedermayr
>>>
>>>
>>>
>>> _______________________________________________
>>> Wien mailing list
>>> Wien at zeus.theochem.tuwien.ac.at
>>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>>> SEARCH the MAILING-LIST at:
>>> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>>>
>>
>

-- 
**********************************************************************
Igor Mazin, NRL, code 6390, 4555 Overlook Ave SW, Washington, DC 20375
Phone: (202) 767-6990; Fax: (202) 404-7546; e-mail mazin at nrl.navy.mil
Home page (private) http://igormazin.us/
**********************************************************************