[gpaw-users] Ti DZP LCAO basis

[Ask Hjorth Larsen]

Hi Duncan

On Fri, 20 Apr 2012, Duncan Mowbray wrote:

> Hi Ask,
> 
> I'm still not able to generate and run with a Ti dzp basis.  I'm generating it with gpaw-basis compiled from GPAW version
> 0.9.0.8839
> The single zeta (sz) basis works fine, but both the double zeta polarized and single zeta polarized basis functions yield
> the following error:
> 
>   File "/home/dmowbray/Software/gpaw/gpaw/lcao/overlap.py", line 457, in calculate_overlap_expansion
>     a_g = (8 * fbt(l, a_q * k1**(-2 - lmax - l), self.k_q, R) /
>   File "/home/dmowbray/Software/gpaw/gpaw/lcao/overlap.py", line 99, in fbt
>     ifft(C[l][n] * f * r**(1 + l - n), 2 * m)[:m].real)
> IndexError: list index out of range
> 
> I'm just doing a simple test of an isolated Ti atom with 4 Ang. of vacuum. 
> 
> Thanks,
> 
> Duncan

I'm taking the liberty to CC this to the gpaw-users list as it's a common 
problem, or to put it another way, Ti is very popular.  In other words I 
should probably add this to the documentation.

The problem is that it thinks it should use an f-type polarization 
function.  It thinks this because there are already valence s, p and d 
wavefunctions.  In fact the p function is a semicore state and doesn't 
really "count", so it should have a p-type polarization function; but 
there is no metadata to tell it this, so it assumes f.  (And f basis 
functions are not supported unless giving --gpaw=projectors=True or using 
trunk, but that's irrelevant now as we are not really interested in the f 
function.)

The solution is to specify the angular momentum quantum number of the 
polarization function explicitly by doing:

   gpaw-basis -t dzp Ti --lpol=1

(As I am running trunk, v0.10.0.9110, but this should also work with 0.9 
stable)

Let me know if there is any further trouble.

Best regards
Ask