[Wannier] What is a good value for final spread?

[Cuauhtemoc Salazar]

Dear All,

I wonder if someone could if tell me if having MLWF’s with spreads between 5 and 8  Angs^2 is considered insufficiently localized for further computation of  physical properties.

I have let run the wannierization for num_iter in the order of few thousands and the MLWF's don’t  get more localized (indeed, the spreads reach convergence with a few hundred iterations). I think my initial guess for the projections  is reasonable, since it is based on the orbital character of the bands I included in the Wannierization. As a check, I got the Wannier-interpolated bandstructure and it is in good agreement with the ab-initio bands. 

DETAILS: I have computed MLWF for stanene, a monolayer of Tin atoms (Z=50), including spin-orbit coupling and d-electrons in the valence (Z_ion=14), using NCPP -Vanderbilt pseudo potentials. The lattice and plane wave parameters were carefully relaxed / converged, using the expresso suite. The disentanglement convergence criteria is satisfied after about 500 iterations, with a
      Final Omega_I    45.08612336 (Ang^2)

The Wannierization ends with a
 Final State
  WF centre and spread    1  (  2.664805, -0.000003,  0.419178 )     4.67225286
  WF centre and spread    2  (  2.664807, -0.000002,  0.419224 )     4.67237620
  WF centre and spread    3  (  1.748739,  1.587314, -0.120711 )     7.50851491
  WF centre and spread    4  (  1.748760,  1.587282, -0.120730 )     7.50791412
  WF centre and spread    5  (  1.748742,  3.028131, -0.120725 )     7.50859427
  WF centre and spread    6  (  1.748763,  3.028160, -0.120744 )     7.50800753
  WF centre and spread    7  (  0.499489,  2.307780, -0.120057 )     7.51866504
  WF centre and spread    8  (  0.499439,  2.307777, -0.120072 )     7.51823175
  Sum of centres and spreads ( 13.323545, 13.846439,  0.115362 )    54.41455668

         Spreads (Ang^2)       Omega I      =    45.086123364
 ================       Omega D      =     2.035781370
                                        Omega OD     =     7.292651945
Final Spread (Ang^2)       Omega Total  =    54.414556679

##########

The wannier90 input file includes

spinors = true

dis_num_iter      = 900   ! needs about 500
num_iter          = 5000   !   5,000

guiding_centres = true

Begin Projections     
Sn1: pz     ! one orbital
Sn2: sp2   ! three orbitals
End Projections       
! since we have spin, the number of projections is num_wann/2 = 4

I am using  PWSCF v.5.3.0 and wannier90 2.0.1.

With regards,
Cuauhtemoc Salazar

PhD Candidate
University of Toronto