[molpro-user] Different CASSCF energies for XYZ and Z-matrix inputs

[Lorenzo Lodi]

I am using molpro 2006.1 to calculate energies of the water molecule.

For Cs geometries I get different CASSCF energies if I specify the 
geometry in the Z-matrix format (letting molpro determine the symmetry 
group) or in the XYZ format (specifying the symmetry group by hand).
I attach a sample input (input.Cs) that reproduces this behaviour.
I chose a Cs geometry very close to a C2v one for later cross-checks.

In the first case (XYZ format for the geometry specification) I get a 
CASSCF energy of -76.12413261 hartree. In the second run I use the 
Z-matrix format and I get -76.14722726 hartree. The RHF energy is the 
same in both cases, -76.02696984 hartree.

On the other hand, if I do the calculation in C2v symmetry (input.C2v) I 
get -76.14722726 in both cases, so the problem (if it is a problem…) 
seems to be in the combination Cs+XYZ.

Can anyone tell me what the origin of this behaviour is?

Lorenzo Lodi