[Q-e-developers] ALL electron and PSEUDOPOTENTIAL data base

[Nicola Marzari]

Dear All,

here is my summary of what is needed for the specific case of
d- and f-elements - I personally think it's the most urgent, since
the other elements tend to be easier to pseudize, and we have
a reliable set for them in the ps_library of Andrea.

For d- and f-elements, the general gist is that for each element
we want to have a set of standard structures that span all relevant
oxidation states (e.g. bcc metal, antifluorite, rocksalt,  zirconia,
and "ReO3" (i.e. XO3)), and, for a given spin-polarized exchange
correlation functional (starting from LSDA) we want to store,
with the appropriate units specified, the

1) equilibrium lattice parameter
2) bulk modulus, and eventually elastic constants
3) optical phonons at the gamma point (i.e. all phonons at gamma
that are not zero)
4) electronic bandstructure (spin up and down) along pre-defined paths
in the BZ (attached - paths to be stored also in the database)
5) total and absolute magnetization

Ideally there should be ways to take the all-electron
data, and the pseudo data, and calculate a figure of merit
on the agreement between the two (or between different all-electron
calculations), defined according to mean sqaure displacements between
AE and PS results for 1) to 5), with weights we'll need to figure out.

I attach here a first set of definitions for these standards,
prepared by Cheol Hwan Park (a postdoc of mine, at MIT - he will
move to a faculty position at Seoul National U. in September).

Cheol Hwan and I would be delighted to interact on the framing of the
database structure. Equally or even more urgent for us is *having* some
all electron calculations to compare with, so that we can start 
optimizing the pseudos. We need La and Eu, in the rocksalt (LaO and
EuO) structure, and ideally also in the zirconia (so it's 4 calculations
that we are asking for). Attached are the computational set up that we 
have used, the k-points used for band structure calculation, examples of 
pwscf input files for convenience, and an example result file (EuO
rock salt structure).

The following are part of the key information in the attached file -
typically for a pseudo calculation we would need to specify these 
computational parameters - we just need to discuss if for the AE 
calculations Gaussian smearing is the best choice, or Fermi-Dirac
would be more commonly used.

			nicola


# Computational setup
#
# spin-polarized (nspin=2)
# PZ LDA exchange-correlation
# scalar relativistic
#
# wavefunction cutoff: 80 Ry
# charge density cutoff: 320 Ry
# Troullier Martins norm-conserving pseudopotential
# 12 x 12 x 12 grid with shift (1/2, 1/2, 1/2)
# Gaussian smearing by 0.02 Ry
#

----------------------------------------------------------------------
Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pseudo.tar.gz
Type: application/gzip
Size: 70883 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20120208/dad7817a/attachment.gz>