ABINIT Discussion Forums

Hi;

I am very new at ABINIT. I decided to use it because it looks like it computes response functions using DFPT; which is what I want. I am specifically interested in elastic constants Cij using DFPT (PAW).

My question is, can ABINIT computes only the rigid-ion (or clammed-ion) using DFPT. I have a method that requires 2nd total energy derivatives (2DTE ) w.r.t. homogeneous strains. The relaxed-ion contribution is of no relevance to the method I am using. So, given any configuration, can ABINIT print the 2DTE w.r.t. the 6 strains using DFPT (PAW)?

Thank You;
Sabry

Dear Sabry,
You can get the rigid or clamped ion contribution, see the elastic tutorial:
http://www.abinit.org/doc/helpfiles/for-v8.0/tutorial/lesson_elastic.html
As you now know, PAW+DFPT works only with LDA...
Best wishes,
Eric

Hi Eric;

Thanks for the information; yes GGA+NC/US should be fine for my application. I went already through the "elastic" lesson and my only concern is whether ABINIT can compute only the rigid-atom elastic constant --- this is the only thing I need for a snapshot structure from MD simulation (i.e. non equilibrium configuration). As you know, computing the relaxed-ion term requires phonon calculations (to get the force constants matrix), which I do not need. However, if ABINIT must compute both together that's fine, but it would be better to have this option in the future? Do you have estimation of how expensive computing rigid-ion vs. relaxed-ion terms.

Looks like you guys did a great job with linear responses DFPT; I like it!

Again, thanks!
Sabry

Hi Sabry,
You can get the rigid-ion by running the rfstrs perturbation without the atomic ones (rfatpol). Saving the atomic perturbation might save you a lot of cpu time since rfstrs will require always the same number of strain perturbations while rfatpol depend on the number of atoms in the cell.
Good you like the DFPT implementation in Abinit, thanks
All the best,
Eric