[gpaw-users] Problem with LCAO/dzp and Indium

[Ask Hjorth Larsen]

Hi

On Fri, 3 Dec 2010, Kristen Kaasbjerg wrote:

> Hi
>
> I have bumped into problem when using the LCAO mode with
> a dzp basis for InAs. I get the following error message when trying to
> run a calculation:
>
>  File "lattice_constant.py", line 34, in ?
>    e0 = atoms.get_potential_energy()
>  File "/home/niflheim/kkaa/ase/ase/atoms.py", line 516, in
> get_potential_energy
>    return self._calc.get_potential_energy(self)
>  File "/home/niflheim/kkaa/gpaw/gpaw/aseinterface.py", line 33, in
> get_potential_energy
>    self.calculate(atoms, converge=True)
>  File "/home/niflheim/kkaa/gpaw/gpaw/paw.py", line 218, in calculate
>    self.initialize(atoms)
>  File "/home/niflheim/kkaa/gpaw/gpaw/paw.py", line 509, in initialize
>    self.wfs = LCAOWaveFunctions(lcaoksl, *args)
>  File "/home/niflheim/kkaa/gpaw/gpaw/wavefunctions/lcao.py", line 23,
> in __init__
>    self.tci = NewTCI(gd.cell_cv, gd.pbc_c, setups, kd.ibzk_qc, kd.gamma)
>  File "/home/niflheim/kkaa/gpaw/gpaw/lcao/overlap.py", line 656, in
> __init__
>    self.calculate_expansions()
>  File "/home/niflheim/kkaa/gpaw/gpaw/lcao/overlap.py", line 684, in
> calculate_expansions
>    l_Ij, phit_Ijq)
>  File "/home/niflheim/kkaa/gpaw/gpaw/lcao/overlap.py", line 516, in
> calculate_expansions
>    tsoe = calc.calculate_expansions(l1_j, f1_jq, l2_j, f2_jq)
>  File "/home/niflheim/kkaa/gpaw/gpaw/lcao/overlap.py", line 479, in
> calculate_expansions
>    oe = self.transformer.calculate_overlap_expansion(a_q, la, lb)
>  File "/home/niflheim/kkaa/gpaw/gpaw/lcao/overlap.py", line 447, in
> calculate_overlap_expansion
>    a_g = (8 * fbt(l, a_q * k1**(-2 - lmax - l), self.k_q, R) /
>  File "/home/niflheim/kkaa/gpaw/gpaw/lcao/overlap.py", line 96, in fbt
>    g += (dr * 2 * m * k**(l - n) *
> IndexError: list index out of range

As I recall, we solved this off-list, and the solution was 
--gpaw=fprojectors=True .  (Correct me if this didn't actually solve the 
problem)

In general we should probably support f-type polarization functions, but 
presently that is not the case.

Regards
Ask