[Molpro-user] DFT oddity

[Kirk Peterson]

I'm sure this is something very simple, but I was just trying to get a  
B3LYP energy for the N atom as a function of basis set and ran into a  
strange problem. The energies look perfectly reasonable until I hit cc- 
pV5Z and then the energy decreases by nearly 0.1 Eh.  Definitely wrong  
and I get the same answer with 2008.1 and 2006.1.  If I carry out a df- 
rks,b3lyp calculation the energy is more reasonable.  I thought  
perhaps this was a symmetry issue since this is turned off in the  
density fitting calc. but this doesn't help the conventional  
calculation.  Any suggestions out there?  The simple input follows  
below.

thanks,

-Kirk

energies:

cc-pVTZ:  -54.57679786
cc-pVQZ:  -54.58035208
cc-pV5Z:  -54.66764391

the latter with df-rks:  -54.57458878 (better but obviously not great)

input:

memory,30,m

basis=v5z

geom={N}

int
{hf;wf,7,8,3}
rks,b3lyp