[QE-users] On-site Hubbard U for background orbitals

[William Hewett]

Hi Iurii,

Thanks so much for your help. My code is working now, I think there was an
error somewhere in

Modules/set_hubbard_l.f90
Modules/set_hubbard_n.f90
PW/src/tabd.f90

As after redownloading / compiling with my new Hubbard channels everything
seemed to worked fine.

Thanks also for pointing me to the example with hubbard_back, as this also
helped a lot too.

One question, you said:

'This modification must be done for both DFT+U and DFT+U+V machineries'

You are talking about editing 'set_hubbard_l' and 'set_hubbard_l_back' to
my desired values? and this is done in 'Modules/set_hubbard_l.f90'

Cheers

Will

On Tue, Aug 18, 2020 at 8:57 PM Timrov Iurii <iurii.timrov at epfl.ch> wrote:

> > Note that by default the second Hubbard channel per atom is of the "s
> character".
>
>
> Sorry, this is true for most elements. But just for a few elements the
> second Hubbard channel is not of the s character. One should always check "
> set_hubbard_l_back".
>
>
> In the (near) future we would like to make it possible to specify the
> Hubbard channels (i.e. orbitals to which the Hubbard correction is applied)
> from the input, and not to have it hard-coded in  set_hubbard_l and
> set_hubbard_l_back.
>
>
> Greetings,
>
> Iurii
>
>
> --
> Dr. Iurii TIMROV
> Postdoctoral Researcher
> STI - IMX - THEOS and NCCR - MARVEL
> Swiss Federal Institute of Technology Lausanne (EPFL)
> CH-1015 Lausanne, Switzerland
> +41 21 69 34 881
> http://people.epfl.ch/265334
> ------------------------------
> *From:* users <users-bounces at lists.quantum-espresso.org> on behalf of
> Timrov Iurii <iurii.timrov at epfl.ch>
> *Sent:* Tuesday, August 18, 2020 10:50:28 AM
> *To:* users at lists.quantum-espresso.org
> *Subject:* Re: [QE-users] On-site Hubbard U for background orbitals
>
>
> Dear Will,
>
>
> The implementation of the multi-channel DFT+U and DFT+U+V codes is in the
> experimental stage (please use it with caution), we are still working on
> it to make it more user-friendly and robust. This is why there is no much
> documentation about it yet.
>
>
> In principle, yes you can use the DFT+U+V machinery to do what you want:
> to apply two different U values to 4f and 5d states of Gd and set V to
> zero. But actually there is a multi-channel implementation also in the
> DFT+U machinery, but this was not advertised yet for the reasons
> mentioned above. But you can see some examples in q-e-6.6/
> test-suite/pw_lda+U (in you case you should check "
> lda+U_background_one_channel.in").
>
>
> > Error in routine init_lda_plus_u (1):
> >      lda_plus_u calculation but Hubbard_l not set
>
>
> I presume that you get this problem because there is something wrong in
> your input. If you share your input it would be useful to find what is
> wrong.
>
>
> Note that by default the second Hubbard channel per atom is of the "s
> character". If you want to change it to something else, then you need to
> modify the routine "set_hubbard_l_back". The main Hubbard channel is
> specified in "set_hubbard_l" (in your case for Gd it is hard-coded and it
> is 4f), and the secondary ("background" [1]) Hubbard channel is specified
> in "set_hubbard_l_back" (in your case for Gd you need to specify that it
> is the d channel, i.e. hubbard_l_back = 2). This modification must be
> done for both DFT+U and DFT+U+V machineries.
>
>
> Greetings,
>
> Iurii
>
>
> [1] Note, the terminology "background" is a bit misleading. It should not
> be called like that (in general), because this is just a second Hubbard
> channel per atom. It is called "background" for historical reasons: this
> secondary (background) channel was originally implemented to do something
> else and hence it has such a name. But we want to change this naming, and
> only then advertise these features to the QE community.
>
> --
> Dr. Iurii TIMROV
> Postdoctoral Researcher
> STI - IMX - THEOS and NCCR - MARVEL
> Swiss Federal Institute of Technology Lausanne (EPFL)
> CH-1015 Lausanne, Switzerland
> +41 21 69 34 881
> http://people.epfl.ch/265334
> ------------------------------
> *From:* users <users-bounces at lists.quantum-espresso.org> on behalf of
> William Hewett <williamholmeshewett1 at gmail.com>
> *Sent:* Tuesday, August 18, 2020 4:35:59 AM
> *To:* users at lists.quantum-espresso.org
> *Subject:* [QE-users] On-site Hubbard U for background orbitals
>
> Hi all,
>
> I'm interested in doing DFT+U calculations on the rare earth mono-nitride
> GdN. The most successful calculations in the past have used multiple
> Hubbard U terms (U_f and U_d) on the 4f and 5d electron channels on the Gd
> ion. This has been done before using QE (Comput. Mater. Sci 95 (2014)
> 263-270).
>
> I suspect that in v6.6 this may now be possible in the stable branch via
> an onsite Hubbard V term on both the standard and background orbitals?
>
> The pw.x input guide gives that Hubbard_V(na,na,k) will be the onsite
> Hubbard interaction. After setting the appropriate values for the
> background n, l and occupations of Gd in:
>
> Modules/set_hubbard_l.f90
> Modules/set_hubbard_n.f90
> PW/src/tabd.f90
>
> and increasing sc_size = 3 in Modules/parameters.f90
>
> and compiling the code, I've used the following in the &SYSTEM block:
>
> lda_plus_u=.true.
> lda_plus_u_kind=2
> Hubbard_V(1,1,1)=a
> Hubbard_V(1,1,3)=b
>
> I guess there will now be two Hubbard terms on that atom at side na, on
> the standard and background orbitals.
>
> However, I find the following error:
>
> Error in routine init_lda_plus_u (1):
>      lda_plus_u calculation but Hubbard_l not set
>
> Any help is appreciated, I'm running QE 6.6 on Ubuntu 20.04.
>
> Thanks
>
> Will Hewett
> Postdoctoral Researcher
> Victoria University of Wellington
> New Zealand
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200824/2e51f4b3/attachment.html>