[Pw_forum] dynamic matrix and symmetry

[Giovanni Cangiani]

Yous should look into the phonon (PH) source directory.
The DM is symmetrized in symdynph_gq.
Symdynph_gq is called the first time from set_irr (for full DM
calculation) where the DM is filled with random numbers, 
symmetrized, and diagonalized in order to find the displacement
patterns. 
When all modes have been computed, Symdynph_gq is called again
on the true DM before the final diagonalization.

The symmetrization consists essentially averaging over all
rotated copies of the input DM. Where the rotations are those
of the small group of q.
The tricky thing is how to apply a rotation of the small group
of q to the dynamical matrix. The way this is done in symdynph_gq
is quite complicated but you can anyway understand the procedure
by carefully inspecting the code. 

I am not the author of the routine so it might be based on a reference
which I am not aware of but I found quite interesting the 
following ones:
Maradudin, Rev. Mod. Phys. 40 p.1
Hendrikse et. al. Computer Phys. Comm. 86 p.297

	giovanni


On Tue, 2003-04-29 at 05:55, xianghjun wrote:
> Dear all,
> I have been reading the source code of pwscf 1.2.0.I don't know how
> pwscf symmetrizes the
> dynamic matrix using symmetry.
> You are welcome to give me some hints or some reference.
> Any suggestion would be appreciated.
> 
> Best Regards.
> xianghjun
> 
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum