[gpaw-users] Fwd: Charged oxygen vacancy in periodic systems

[Ask Hjorth Larsen]

Hello

There is no fundamental incompatibility between using one
pseudopotential for some atoms and another for other atoms, so you
should be able to mix them.  However the HGH ones require much more
fine grid spacing.

The c_n, v_l and so on are lists of parameters for the HGH setups.
They are described here:
http://journals.aps.org/prb/abstract/10.1103/PhysRevB.58.3641

The parameters are described in the beginning.  In the code, c_n are
the coefficients of the local parts, while v_l is a list of
"VNonLocal" objects that each correspond to a value of l, and some
parameters that determine the non-local atomic part of the
Hamiltonian.

Unfortunately not a lot of tests have been done to compare HGH with
PAW, as far as I know.  HGH is only used when people need to do
"funny" stuff with the pseudopotentials, particularly testing things
in a simpler way than with full PAW.  You are welcome to do a few
comparisons.

Note that if there's a semicore ('sc') HGH pseudopotential available,
it's probably much better than the non-semicore one.

Best regards
Ask

2015-03-25 9:04 GMT+01:00 Teemu Parviainen <teemu.t.parviainen at student.jyu.fi>:
>
> Hi,
> Thank you for this information!
>
> I don't have any experience of generating and manipulating the setups. Also
> I have used only paw setups in my earlier calculations so a couple of
> questions arises:
>
> -How do you usually fiddle with all the parameters (c_n, v_l, rloc in this
> case). I mean is there any basic scheme to change the parameters? What is
> the criteria to accept the parameter set?
>
> -Is it OK to mix conventional paw setups with hgh setups (for example
> defining hgh setups for Mg and paw for O)?
>
> -How do the results calculated with hgh and paw setups compare with each
> other? For example, how do the calculated adsorption energies or vacancy
> formation energies compare in general? Has anyone even studied this?
>
> Greetings,
> -Teemu Parviainen, PhD student, Jyväskylä Finland
>
>
>
>
> On 13 March 2015 at 12:43, Ask Hjorth Larsen <asklarsen at gmail.com> wrote:
>>
>> Hello
>>
>> Setups are extremely complicated, so if you fiddle with them it'll
>> probably just result in garbage.  The HGH pseudopotentials are simple;
>> you can create your own, providing/hacking a humane number of
>> parameters.
>>
>> Grab the objects that encode the minimal set of parameters to specify
>> the setup, and then improvise from there changing number of valence
>> electrons, etc.:
>>
>> >>> from gpaw.hgh import setups
>> >>> setups['Si']
>> <gpaw.hgh.HGHParameterSet instance at 0x8cf798c>
>> >>> dir(setups['Si'])
>> ['Nv', 'Z', '__doc__', '__init__', '__module__', '__str__', 'c_n',
>> 'configuration', 'copy', 'get_occupation_numbers', 'nl_iter',
>> 'print_info', 'rloc', 'serialize', 'symbol', 'v_l', 'zeropad']
>>
>> Nv is valence electron number, c_n are coefficients, and v_l is a list
>> of VNonLocal objects with info on the atomic Hamiltonian
>> contributions.  Mix them however you like them and create a
>> HGHParameterSet with that.  Then instantiate a HGHSetupData from that,
>> and run your calculations with GPAW(setups={17: my_hghsetupdata}) to
>> specify a custom setup for atom 17.
>>
>> Best regards
>> Ask
>>
>>
>> 2015-03-13 7:53 GMT+01:00 Ioannis Remediakis <remed at materials.uoc.gr>:
>> > On 11/03/2015 09:15, Teemu Parviainen wrote:
>> >>
>> >> Hi all,
>> >> I want to do calculations for charged oxygen vacancies in periodic
>> >> systems (MgO surface). Does anyone have any experience what is the best
>> >> way to do this with gpaw? I came across to a method called Virtual
>> >> Crystal Approximation and if I understood correctly I have to change
>> >> the
>> >> nuclear number of all Mg atoms so that the overall change corresponds
>> >> to
>> >> the charge of the vacancy. Is this correct? How is this done in gpaw?
>> >> Is
>> >> it sufficient to just change the Z number in the PAW setup file or do I
>> >> need to generate completely new one?
>> >>
>> >> Thanks in advance,
>> >>
>> >> -Teemu Parviainen, PhD student, Jyväskylä Finland
>> >>
>> >> --
>> >
>> >
>> >
>> > Dear all,
>> >
>> > I would like to ask a more general question: is it possible to use VCA
>> > (virtual crystal approximation) in GPAW?
>> >
>> > This method is very efficient in alloys of atoms with similar sizes and
>> > same
>> > number of valence electrons, such as SiGe. Instead of using a large cell
>> > with Si and Ge atoms, you use a small cell with virtual "SiGe" atoms.
>> >
>> > VCA is implemented in some codes, such as Espresso. If I understand
>> > correctly, the pseudopotential for the virtual atom is the average of
>> > the
>> > two pseudopotentials. But I do not know how one could do this in GPAW.
>> >
>> > Any help would be appreciated. Thank you!
>> >
>> > best regards,
>> > Ioannis
>> >
>> >
>> > --
>> > ----------------------------------------------
>> > Ioannis N. Remediakis
>> > Department of Materials Science and Technology
>> > University Of Crete,  710 03 Heraklion, Greece
>> > Tel: +30 2810 394250      Fax: +30 2810 394201
>> >                 http://theory.materials.uoc.gr
>> > ----------------------------------------------
>> >
>> > --
>> > This message has been scanned for viruses and
>> > dangerous content by MailScanner, and is
>> > believed to be clean.
>> >
>> > _______________________________________________
>> > gpaw-users mailing list
>> > gpaw-users at listserv.fysik.dtu.dk
>> > https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>>
>> _______________________________________________
>> gpaw-users mailing list
>> gpaw-users at listserv.fysik.dtu.dk
>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
>
>
>
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users