The ABINIT Users Mailing List ( CLOSED )

["Allan, Douglas C Dr" <AllanDC@corning.com>]

Dear Alex,

Bands can cross so if you are comparing bands of the same band number as
determined by Hamiltonian eigenvalues, you may find that a different
ordering is needed in order to get nearby k-points to agree.  I refer to
band crossing e.g. at high symmetry points.  I am not sure if this is
just a curiosity for you or if some real implication for calculations is
involved.  I do not expect any physical quantity to be affected by the
phase you are discussing.  The iterative solver could be involved
because if you let the phase wander then the Fourier coefficients can't
be integrated as efficiently if you want to use past values to project
forward.

Regards,
Doug

-----Original Message-----
From: Alex Kutana [mailto:akut@yahoo.com] 
Sent: Tuesday, August 12, 2008 4:19 AM
To: forum@abinit.org
Subject: Re: [abinit-forum] Smoothness of the wavefunction in abinit




--- On Mon, 8/11/08, Nicola Marzari <marzari@mit.edu> wrote:

> From: Nicola Marzari <marzari@mit.edu>
> Subject: Re: [abinit-forum] Smoothness of the wavefunction in abinit
> To: forum@abinit.org
> Date: Monday, August 11, 2008, 11:19 PM Dear Alex,
> 
> 
> well spotted. I think that the reason is that the solution of the 
> Kohn-Sham equations at any k-point is well-defined modulo an arbitrary

> phase.
> 

Dear Nicola,

Thank you for your prompt reply, and you were right, it /was/ the global
phase indeterminacy. By mistake, I ended up comparing the phases of c_nk
at different values of the band number n, as if they belonged to the
same wavefunction. Now, by looking at c_nk for the two k points
belonging to the same band number n, everything seems to work out. I
apologize for not checking my data thoroughly before posting the
question. 

Thanks again.
Alex Kutana