Experiences With the Full and Virtuals-only Counterpoise Schemes

A large numer of studies are now available in which one or both CP schemes were used to correct for the BSSE. We here present some of the conclusions 

Full CP, virtuals-only CP and uncorrected interaction energies. 

A feature not apparent from Table VII is that for larger basis sets the Cl BSSE may be larger than the SCF BSSE, and it is difBcult to remove it by enlarging the basis Its value may easily reach 50% of the 

We stress that the purpose of a good BSSE correction scheme has to be to yield the true interaction energy corresponding to the quality of the basis set and the method. One should not expect all basis dependencies to be removed nor the experimental result to be reproduced. Comparison with, e.g., the experimental dissociation energy may be misleading. Since the main defect of most basis sets employed in practical calculations is a lack of sufficient polarization functions (cf Section IV) to reproduce the dispersion energy adequately, the true interaction for those basis sets will be above the experimental result and any scheme that underestimates the BSSE will be favoured in such a comparison . 

These considerations are frequently overlooked in studies that claim, on numerical grounds, that 5 is an overcorrection, or simply unhelpful. For example, in an SCF study on FH. OH2 Kocjan et al observed that correction of the STO-2G result at the calculated removed all attraction. Their conclusion that overcorrects is debatable, however, since was probably so short that a repulsive corrected A should have been anticipated. 

---

D. Experiences With the Full and Virtuals-only Counterpoise Schemes... 

---