[gpaw-users] gpaw setup generation

[Ville Mäkinen]

Dear GPAW users,

we are interested in performing GPAW calculations with Hafnium. Currently
there are no ready setups for this element so we decided to try to generate
it ourselves. There are some questions that we could not find answers from
the gpaw wiki page.

1. The electronic configuration of Hafnium is [Xe],4f14,5d2,6s2. Is there
any way to say whether the 4f-electrons should be included in the valence
states or not?

2. We generated the setup with both 4 and 18 valence electrons. Trying to
perform an actual GPAW calculation with 18-valence -version raises an
exception

...
    T_Lqp = self.calculate_T_Lqp(lcut, nq, _np, nj, jlL_i)
File "/fs/local/linux26_x86_64/appl/nano/gpaw/0.6/lib/python/gpaw/setup.py",
line 804, in
 calculate_T_Lqp
    T_Lqp[:, q, p] = G_LLL[L1, L2, :Lcut]
ValueError: shape mismatch: objects cannot be broadcast to a single shape

This happens at least with gpaw version 0.6.5150 and svn-version 0.8.0.7944.
Do you think this is a bug or is there some upper limit for the number of
valence electrons?

3. Choosing the cutting radii for the valence states is a bit unclear. On
some elements the different orbitals have very different radii whereas on
some others they are all the same. Is there any rule of thumb how to choose
them, or simply by trial and error?

4. One question concerning the convergence tests. The spin-paired isolated
atom energy approaches zero as h -> 0, but why does the dimer energy behave
in the same way? My first thought was that it should converge to the "real"
dimer bonding energy instead.

Best regards
----------------------------------------------
Ville Mäkinen
ville.p.makinen at jyu.fi
PhD student,
University of Jyväskylä, Department of Physics
----------------------------------------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20110414/ff14ac2b/attachment.html