[Pw_forum] High Ecut for Ultrasoft PPs

[Nicola Marzari]

On 16/04/2014 11:29, Zarah Baiyee wrote:
> Dear all,
>
> I am studying a BaFeO3 perovskite structure, and I am struggling with
> the convergence, primarily the Ecut WFC convergence.
>
> So far I have used a variety of PPs for this structure, yet cannot get a
> sensible convergence for the Ecut Wfc, please see my convergence data
> below:



Dear Zarah,


1) you do not check cutoff convergence on the total energy,
but on its derivtives: forces for a distorted configuration,
lattice parameters, bulk moduli... E.g. you want an eq. lattice
parameter converged to better than .1 a.u, a bulk modulus converged
to better than 3-4%, and phonons converged to better than 1%.

2) for pseudopotentials, the defaults you should probably use are the
http://qe-forge.org/gf/project/pslibrary/ (we keep a compiled version
of all the pseudos here http://theossrv1.epfl.ch/Main/Pseudopotentials)
or the GBRV high-throghput library http://www.physics.rutgers.edu/gbrv/
(just be careful that this latter might have elements with very large
numbers of semicore electrons - listed here in a very cool manner:
http://theospc12.epfl.ch/GBRV_map/css3d_periodictable_GBRV.html )


			nicola


----------------------------------------------------------------------
Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL