[molpro-user] Trouble with FCI calculation of O1+ doublet

[Bastiaan J. Braams]

I'm finding odd behavior trying to calculate an FCI energy for the O1+
doublet state; when "nosym" is used it seems to be giving me something
in the vicinity of the quartet true ground state result.  Perhaps I
misunderstand the Molpro symmetry specification or the input syntax;
otherwise it looks like an error to me.

Here is an attempt to do the calculation without explicit
specification of the symmetry.  I was originally using the aug-cc-pvtz
basis, but for present purposes I report results on a vdz basis, where
even the FCI calculations take just a few seconds.

***, ENERGY FOR O1+ doublet
memory,84,m
punch,mpsp.pun,new
geomtyp=xyz
geometry
nosym
  1
  null
O     0.0  0.0  0.0
end
basis=vdz
rhf;wf,charge=1,spin=1
rccsd(t)
fci
---

The result is:

RHF STATE 1.1 ENERGY       -74.16555358
RCCSD STATE  1.1 ENERGY    -74.29000674
RCCSD(T) STATE  1.1 ENER   -74.29987023
FCI STATE 1.1 ENERGY       -74.43769484

If we replace rhf/rccsd(t) by uhf/uccsd(t) then we obtain:

UHF STATE 1.1 ENERGY       -74.16935621
UCCSD STATE  1.1 ENERGY    -74.29038230
UCCSD STATE  1.1 ENERGY    -74.29038230
FCI STATE 1.1 ENERGY       -74.43818888

Observe the very large difference, in both cases, betweem the coupled
cluster and the fci results.  I then tried to do the calculation with
explicit specification of symmetry; so:

***, ENERGY FOR O1+ doublet
memory,84,m
punch,mpsp.pun,new
geomtyp=xyz
geometry
x,y,z;mass
  1
  null
O     0.0  0.0  0.0
end
basis=vdz
rhf;wf,charge=1,sym=1,spin=1
rccsd(t)
fci
---

and likewise for all symmetry values 1..8 and for both rhf/rccsd(t)
and for uhf/uccsd(t).  The ground state is obtained if one uses
symmetry specification sym=2 or sym=3 or sym=5 and then the
Hartree-Fock and Coupled Cluster results agree with those obtained
without explicit symmetry specification.  However, when the symmetry
is explicitly specified the FCI result comes out much more natural
looking, just a tad below the coupled cluster result.  For example,
with specification sym=2:

RHF STATE 1.2 ENERGY       -74.16555358
RCCSD STATE  1.2 ENERGY    -74.29000674
RCCSD(T) STATE  1.2 ENER   -74.29987023
FCI STATE 1.2 ENERGY       -74.30431432

UHF STATE 1.2 ENERGY       -74.16935621
UCCSD STATE  1.2 ENERGY    -74.29038022
UCCSD(T) STATE  1.2 ENER   -74.30034927
FCI STATE 1.2 ENERGY       -74.30471007

There is no symmetry specification, after the "x,y,z;mass" card, for
which the fci result is anywhere near the one obtained using nosym.

It seems that the "nosym" fci calculation is actually finding the
quartet (spin=3) ground state, below the doublet ground state.

Bas
--
Bastiaan J. Braams
braams at mathcs.emory.edu
Emory University, Atlanta, GA