The ABINIT Users Mailing List ( CLOSED )

[Xavier Gonze <gonze@pcpm.ucl.ac.be>]

Dear Yunhee Chang,

On 22 Jun 2006, at 08:46, yunhee.chang@kriss.re.kr wrote:

> Dear Abinit users,
>
> From the reference of phonon dispersion relation using Ab inito  
> calculation,
> the dynamical matrix is calculted from interatomic force constant
> D(q)=C(q)/(MiMu)**(1/2).

Making references to PRB 55, 10355 (1997) :
(1) the main abinit compute by DFPT the quantities tilde{C(q)} - see  
Eq.(13) and section III A, and III B -
 that are the Fourier Transform of the interatomic force constants  
(see Eq.(10))
(2) the Dynamical matrix straighforwardly comes from tilde{C(q)} ,  
see Eq.(11),
  and gives phonon eigenvalues and eigenvectors, Eq.(12).
So, the main abinit does NOT compute interatomic force constants.

By contrast, ANADDB computes the interatomic force constants, as sum  
of two terms (in the case of
 insulators), see Eq. (78) : the short range one, see Eq. (77) and  
the dipole-dipole
contribution see Eq.(71). The short range contribution is computed  
from a set of tilde{C(q)}.

When the interatomic force constants are known, it is easy to compute  
the Fourier transform of interatomic
force constant everywhere in the BZ, see Eq.(79), and then dynamical  
matrices.

> Then in tutorial, we calculate the dynamical matrix first (tutorial  
> trf2_1.in),
> and that using ifcflag, we calculate the interatomic force constant  
> again. (tutorial trf2_4.in)
> I'm confusing these algorithms.
>
> Would you give me any comments?
>
> Best regards,
> Yunhee Chang