[Pw_forum] some basic and general questions on Q.E calculations...

[Nguyen Doan Sau]

Hi 
I am learning DFT  calculation and I have some basic and general questions on Q.E calculations and I hope anyone could clarify them.
For short introduction, I synthesize a new oxide compound. I solve its crystal structure using single crystal diffractometer to get everything, atomic coordinates, unit cell...
After that, I measure physical properties like Second harmonic generation (SHG), dielectric, conduction, piezoelectric, ....
Finally to explain all these above, I want to do electronic structure calculation on the compound. I did it using TB-LMTO-ASA and every thing is good so far.
Now I switch to Quantum Espresso. My question are:
Firstly,
In TB-LMTO-ASA, I don't have to optimize crystal structure, I just import the crystal data from .cif or .ins or .res file. However, in quantum espresso when I did the same way, the convergence seems very slow or may be not converged. 
I checked the Q.E forum previous post
(http://www.democritos.it/pipermail/pw_forum/2004-December/001836.html)
(http://www.democritos.it/pipermail/pw_forum/2004-February/000877.html
and I realized that when doing relax calculation to optimize the crystal structure I also doing scf calculation (right?). 
So, my first question is why somebody do scf calculation while relax calculation is the better choice.
My second question is that when doing relax calculation we get an optimized structure. When we use this optimized structure for scf calculation, do we get anything difference to relax calculation one?
My third question is what kind of relax calculation, "relax" or "vc-relax" is better for electronic structure calculation of solid state material?
(http://www.democritos.it/pipermail/pw_forum/2008-August/009827.html)
Please correct me if I got something wrong in my understanding of the  way I should use for my own calculation.
Secondly, 
I knew it is very important to get a right pseudopotentials which should reflect the atomic wavefunctions of atoms state in materials. For examples,
in BaTeMo2O9 structure, the oxidation number Barium is Ba2+, Te is Te4+, Mo is Mo6+, O is O2-. Therefore, I think it shoud be better to create pseodopotentials with the electron configuration reference  of Ba2+, Te4+, Mo6+ and O2-. Though the pseodopotentials test is quite good if I start with the neutral atomic electron configuration reference, ie, Ba0+, Te0+, Mo0+, O0-. The question is which ways is better for calculation if I I assume that both have good transferability.
I am appreciated at any comment to my questions.
Sau Nguyen,
U of Houston.




      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20100820/707b1d26/attachment.html>