[molpro-user] Simple unreasonable norm problem...

[Gerald Knizia]

On Thursday 14 October 2010 19:38, Charles Wood wrote:
> My problem is this:
> I am trying to run an extremely simple H - Ca2+ - H calculation (one
> calcium 2+ ion centred between 2 hydrogen atoms). I'm using UHF and
> UCCSD(T) programs in molpro2009.1. If the geometry is specified so each
> of the H atoms are less than 1.5 Angstroms then the calculation works fine.

At some bond stretching far beyond the equilibrium geometry, the molecule will 
become multi-referency (due to the near-degeneracy of the molecular state and 
the dissociated states). So starting at some geometry CCSD(T) will give bogus 
results.

For your 1.5 Angstrom geometry this is already the case. If you run mcscf and 
ci, you will see that there are two reference configurations with comparable 
weight.

> I saw some people suggest 'shift' but this looks like something
> connected to the geometry....any idea??

If you want to see the CCSD(T) blowing up instead of giving this message, you 
can include
	_CC_NORM_MAX=2.0
in the input. It will then happily converge without giving this message.
-- 
Gerald Knizia