[Pw_forum] About maximally localized wannier function

[Nicola Marzari]

Dear Silviu,

this is an interesting suggestion. The good thing about the
Gamma-sampling CP code algorithm is that it will always lead to
the true minimum of the localization functional, as long as a
cubic/tetragonal/orthorombic cells are used (i.e. derivatives are
calculated with 3 orthogonal G vectors). Depending on your application,
this could be very important or irrelevant - the more the system is 
large and disordered, the more it becomes useful/essential.
Another advantage is that the CP/Wannier subroutines work with ultrasoft
pseudopotentials, although this capability is forthcoming in 
wannier-transport.org and wannier.org .

The regular k-point wannier codes use instead trial projection orbitals
(very straightforward  ones - usually one specifies the center of a
gaussian, and the l,m of a spherical harmonics) to start the 
minimization already in the right basin of attraction. This
approach works again very well in most cases - to the point that some 
groups just use these projections, without localization, to get
localized WFs (I think, e.g., Pickett's group).

As a last note - if you use CP, your sampling can be "poor" if you have
a smallish unit cell. On the other hand, if you use PWSCF to feed CP,
you can make your charge density as good as you want it, sampling-wise.
Then, the only finite-size errors remaining will be on the calculation
of the MLWFs themselves, that, using CP, will only combine states
obtained at Gamma, instead of using all the Brillouin zone.

				nicola


Silviu Zilberman wrote:

> Dear Xianghjun,
> 
> MLWF are implemented in the CP code. I believe that the latest cvs 
> version has compatible restart files, which means you could use any 
> PWSCF calculation result as an input to a CP Wannier calculation. This 
> compatibility has only been introduced recently, and it may still be 
> unstable. You could also take the geometry from your PW calculation and 
> start a CP calculation from scratch and minimize the wavefunction using 
> Wannier dynamics. Look at the INPUT_CP file for more details, as well as 
> example #23.
> 
> Silviu.
> 
> xianghjun wrote:
> 
>> Dear all,
>>  I am trying to get MLWF for some systems and want to know if MLWF has 
>> been implemented in PWSCF.
>> Thanks in advance.
>>
>> Best regards,
>> xianghjun
>>
>> =============================================================
>> H. J. Xiang
>> Ph.D. Candidate,
>> Hefei National Laboratory For Physical Sciences at Microscale,
>> University of Science and Technology of China,
>> Hefei, Anhui, 230026,
>> People's Republic of China
>> Tel.: 86-551-3606428
>> Fax.: 86-551-3602969
>> E-mail: xhongjun at mail.ustc.edu.cn
>> =============================================================
>>
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://www.democritos.it/mailman/listinfo/pw_forum
> 
> 
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum

-- 
---------------------------------------------------------------------
Prof Nicola Marzari   Department of Materials Science and Engineering
13-5066   MIT   77 Massachusetts Avenue   Cambridge MA 02139-4307 USA
tel 617.4522758  fax 617.2586534  marzari at mit.edu  http://nnn.mit.edu