[molpro-user] strange CCSD(T) results

[hinde at ion.chem.utk.edu]

Dear MOLPRO community:

I am a novice MOLPRO user trying to compute three-body terms in
weakly bound van der Waals trimers at the CCSD(T) level of theory.
I have come across some strange results that suggest that I'm doing
something wrong or that there is a bug in the MOLPRO version that
is installed on the machine I'm using (2002.6).

Short form of the problem:

I'm computing total CCSD(T) energies for the systems

   A-B-(A) and (A)-B-A

where (A) indicates molecule A replaced by ghost orbitals in order
to perform a counterpoise correction.  The underlying symmetry of
the full A-B-A trimer is such that the two systems listed above have
identical geometries.  I get the same RHF and MP2 energies for both
systems but different CCSD(T) energies.

Long form of the problem:

Here is the input file in question.

---------- start -----
***,I
memory,45,M;

gprint,orbitals,civector,basis;
geometry={
Z
H1,0, -2.77550006,  0.00000000,  0.00000000;
H2,0, -4.22450018,  0.00000000,  0.00000000;
H3,0,  0.00000000,  3.72097754,  0.00000000;
Cl,0,  0.00000000,  6.12957764,  0.00000000;
H5,0,  4.22450018,  0.00000000,  0.00000000;
H6,0,  2.77550006,  0.00000000,  0.00000000;
}

basis=avtz

dummy,H1,H2
hf;
ccsd(t);
dummy,H5,H6
hf;
ccsd(t);
---------- end -----

This file gives the following results:

        CCSD(T)         HF-SCF          CCSD(T)         HF-SCF
   -461.48298818   -461.23925352   -461.51582064   -461.23925352

When I replace the last six lines of the above input file with the
following:

dummy,H5,H6
hf;
ccsd(t);
dummy,H1,H2
hf;
ccsd(t);

which simply switches the order of the calculations, I get these
results instead:

        CCSD(T)         HF-SCF          CCSD(T)         HF-SCF
   -461.51582064   -461.23925352   -461.51582064   -461.23925352

To add to the puzzle, if I replace the basis=avtz line with an
explicit specification of the basis sets (downloaded from the EMSL
library), the first file gives these results:

        CCSD(T)         HF-SCF          CCSD(T)         HF-SCF
   -461.51582064   -461.23925352   -461.51413322   -461.23925352

and the second file gives these results:

        CCSD(T)         HF-SCF          CCSD(T)         HF-SCF
   -461.51582064   -461.23925352   -461.51106931   -461.23925352

So you can see that I've gotten four different CCSD(T) energies for
the same molecule, depending on how I do the calculation.  Every
calculation gives the same HF (and MP2) energy.

Finally, if I set basis=avdz instead, the two files each give the
same CCSD(T) energy for both species:  -461.43754496 hartree.

Any ideas what I might be doing wrong?

Thanks,
Robert Hinde