The ABINIT Users Mailing List ( CLOSED )

["Horacio W. Leite Alves" <hwlalves@funrei.br>]

Dear Arnaud,

About the migration from unit cells to supercells, one must take care, 
in the ifc input, about the new symmetry, which is usually orthorhombic, 
sometimes tetragonal or hexagonal and, we should inform it on "brav" 
input. It makes some difference. Also, to get the LO-TO splitting you 
should include a second "deck" of data,i.e.,
nphl2  0
qphl2  0.0  0.0  1.0  0.0

It is sure that you need to make a good converged electric field 
perturbation to get the right DDB file and, to
drive ifc calculations you must check the following variables:
rfmeth, asr, chneut, slectz, ifcflag=1, and so on .... I did not have 
used ifcflag=0 - I simply run abinis to get the right
DDB and then I use the ifc code. Anyway, you have the opportunity to 
check this.

greetings,


Horacio W. Leite Alves




arnaud marmier wrote:

> Hello all,
>
>  I'm a new user of Abinit, interested in latice dynamics.
>  As I'm migrating from supercell to get the phonons, I'm currently 
> trying to do very simple things, and to get a good feeling of ifc.
>
> I'll use the DDB generated in the tutorial 5, i.e. t54 (phonons at 
> Gamma, no Born charges) and t55 (at Gamma, with Born charges).
>
>   The ifc.in I use is very very simple :
>
>  nph1l     1     ! Number of phonons in list 1
>  qph1l     0.0 0.0 0.0    1.0
>
> When I run it for t45, it crashes (under Windows).
> For t55, it gives the frequency, but (by comparison with the abinit 
> output), it hasn't taken into account the born charges (no LO-TO 
> splitting).
>
> Is it true that IFC needs a DDB wich has electric field pertubation ?
>
> How do you enforce that this information is used by IFC ? Do you need 
> to  use ifcflag ?
>
> Best wishes
> Arnaud