[Wannier] Convergence of MLWFs

[Nicola Marzari]

Ciao Carlo,


the educated guess here is that your projections are just not in the 
right place. Your system should actually be fairly easy to deal with -
if you want the occupied space alone, you could start with a s orbital
in the middle of *each* bond (both carbon-carbon and carbon-H) and
a p_z orbital on every other carbon of the ring.

Maybe starting with benzene as an example could help?

I'm not sure the 2 upper valence bands (HOMO and HOMO-1) would
be isolated, so I'd first try to wannierize all bands, and then
think at what to disentangle. Posting a band structure of your
system could also help us give some advice.

I'll be at TCD next week (Wed and Thu) so we could also talk.

				nicola


On 09/05/2013 18:22, Carlo Motta wrote:
> Dear wannier90 users,
>
> I am a new user and I am trying to calculate the MLWFs for the (HOMO, HOMO-1) bands of a a molecular crystal with 2 molecules per unit cell.
> My problem is that I see instabilities in my results, in particular in the matrix elements in the durene_hr.dat.
> I did convergence tests wrt the k-mesh and cutoff energy. The durene_hr.dat values are crazy (no regularity with respect to the convergence parameter).
> I have checked that the wannierization has converged, since the Delta values are below 1E-7.
>
> My beginning projections are random, since I have no clue on how to improve it for this large system.
> Having 2 molecules in the unit cell, whose centers in crystal coordinates are 0.0,0.0,0.0 and 0.0,0.5,0.5, I tried to do this:
>      begin projections
>      c=0.0,0.0,0.0:l=0
>      c=0.0,0.0,0.0:l=0
>      c=0.0,0.5,0.5:l=0
>      c=0.0,0.5,0.5:l=0
>      end projections
> but the results are even worse (not converged).
> However, I am conscious that putting an s orbital (l=0) in the center of mass of the molecules as a starting guess does not make much sense.
>
> Probably I have still some parameter in my calculations that is still not at convergence. Which are the ones most influencing the wannierization?
> Is there a possibility to run a wannierization by starting from the solution found in a previous calculation? I am asking this also because in different calculations, the labels of the different WFs (1, 2, 3,...) mix up.
>
> If you need more informations I will provide them!
>
> Thank you in advance,
>
> Carlo
>
> --
>
> Carlo Motta,
>
> PostDoc Researcher
> School of Physics and CRANN,          IE Mob.: +353 (0) 83440 8930
> Trinity College Dublin,                          IT Mob: +39 3286845055
> Dublin 2, IRELAND
> _______________________________________________
> Wannier mailing list
> Wannier at quantum-espresso.org
> http://www.democritos.it/mailman/listinfo/wannier
>


-- 

----------------------------------------------------------------------
Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL