[molpro-user] LQCISD geometry optimisation

[Gerald Knizia]

On Wednesday 24 November 2010 06:03, Matt Addicoat wrote:
> I am trying to undertake a LQCISD geometry optimisation of NO3- using
> molpro 2010.1
>
> The input below works for water, but for any larger molecule, where
> there are close and weak pairs, it fails:
>
>
>   CPU TIME FOR ITERATIVE CP-LMP2:    4.78 SEC
>
>
>   Time needed for computing effective density matrices:     10.9 sec.
>   Pair conflict in ccys
>
>   ERROR EXIT
>   CURRENT STACK:      MAIN
> -----
>
> I can't find mention of this error anywhere. Any pointers would be
> appreciated.

This might be the force program being confused. I don't think there are 
analytic energy gradients for LQCISD, but there are for QCISD and LMP2, and 
the force program appears to erroneously invoke some combination of both (or 
the simulation of LQCISD with the regular QCISD program or something like 
that). Anyway, you could try to explicitly specify "numerical" as parameter 
for the optimization, or switch down to LMP2 level.
-- 
Gerald Knizia