[Wannier] Projections

[Arash Mostofi]

Dear Soham,

There's no means by which to explicitly get the code to ask for 
projections that are linear combinations of different d-orbital 
projections. However, given that your system is effectively only a 
two-band problem, you may be able to perform the necessary linear 
combinations yourself. Eg, the electronic structure code will return to 
you projections such as

<\psi_nk | d_1> and <\psi_nk | d_2>

whereas you would like, for example,

<\psi_nk | (d_1 \pm d_2)> = <\psi_nk | d_1> + <\psi_nk | d_2>.

You may be able to achieve the result you want by taking appropriate 
linear combinations of the columns of the A_{mn} matrix that is returned 
by the electronic structure code.

Regarding your second question, whilst the code does not support 
Gaussian projections, it should not make too much of a difference to use 
"s" orbital (l=0) projections at the locations that you want. You can 
then control the narrowness using the "zona" parameter in the definition 
of the projection (Wannier90 User Guide, Ch. 3 for more information).

Best wishes,

Arash

--
Dr Arash A Mostofi
Departments of Materials and Physics
Deputy Director, CDT on Theory and Simulation of Materials
Imperial College London, London SW7 2AZ, UK
+44 (0)207 594 8154 | www.cmth.ph.ic.ac.uk/people/a.mostofi

On 28/01/2014 00:57, Ghosh, Soham Subhra wrote:
> Dear Wannier90 users,
>
> I am a new user of Wannier90. I have two questions, and I will be very
> thankful if you can help me out.
>
> I am trying to find MLWFs for the cuprate La4Ba2Cu2O10. A DFT LDA
> calculation shows (using the Vienna-Ab-Initio Simulation package) that
> it has 2 narrow  bands around the Fermi level, with a dispersion of only
> a few meV. They are well separated from all the other bands. My axes are
> along the lattice vectors of the SC cell and the Cu-O planes are at 45
> degrees to the walls of the cell.
>
> What I understand from the site-and-band projected DOS (in VASP) is that
> each of the two bands  is the sum of antibonding Cu-dxy and Cu-dz2-r2.
> There is one band per copper atom. Both Cu-dxy and Cu-dz2-r2 have two
> peaks, one at Fermi energy and one significantly below. Is there a way
> to incorporate that in my initial projections? For my 2 wannier
> functions, wannier90 allows only specific values of 'l' and 'm' as
> initial projection. The only projection that converges is Cu-dz2.
>
>
> My second question is about the use of narrow gaussian states centered
> at one of the main lobes of the Cu-d contribution as projection. Is that
> possible in Wannier90? As a first step of my calculation, I am trying to
> replicate the MLWFs in existing literature (PRL 89,16(2002)) where the
> use of this specific projection is quoted.
>
>
> I am copying below my .win file (without the k-point list). Delta spread
> becomes ~10E-11 only after 180 odd steps. The program does not converge
> below Delta-spread ~10E-02 for most other values of trial_step, and for
> any other projection. Energy from VASP itself is well-converged.
> ------------------------------------------------------------------------------
> exclude_bands = 1-72,75-160
> num_wann = 2
>
> begin projections
> c=0.00,3.42,2.93:dz2
> c=3.42,0.00,2.93:dz2
> end projections
>
> num_iter = 200
> guiding_centres = TRUE
> trial_step = 1.9
>
> begin unit_cell_cart
>       6.8447000     0.0000000     0.0000000
>       0.0000000     6.8447000     0.0000000
>       0.0000000     0.0000000     5.8637000
> end unit_cell_cart
>
> begin atoms_cart
> La       1.1964536     4.6188036     0.0000000
> La       2.2258964     1.1964536     0.0000000
> La       5.6482464     2.2258964     0.0000000
> La       4.6188036     5.6482464     0.0000000
> Ba       0.0000000     0.0000000     2.9318500
> Ba       3.4223500     3.4223500     2.9318500
> Cu       0.0000000     3.4223500     2.9318500
> Cu       3.4223500     0.0000000     2.9318500
> O        0.0000000     0.0000000     0.0000000
> O        3.4223500     3.4223500     0.0000000
> O        2.4818882     5.9042382     1.5509486
> O        2.4818882     5.9042382     4.3127514
> O        0.9404618     2.4818882     1.5509486
> O        0.9404618     2.4818882     4.3127514
> O        4.3628118     0.9404618     1.5509486
> O        4.3628118     0.9404618     4.3127514
> O        5.9042382     4.3628118     1.5509486
> O        5.9042382     4.3628118     4.3127514
> end atoms_cart
>
> ---------------------------------------------------------------------------------
>
> Thank a lot you for your time.
>
> Regards,
> Soham.
>
> Soham S. Ghosh
> Graduate Student
> Department of Physics,
> Florida State University
>
>
>
> _______________________________________________
> Wannier mailing list
> Wannier at quantum-espresso.org
> http://www.democritos.it/mailman/listinfo/wannier
>